text
stringlengths 87
880k
| pmid
stringlengths 1
8
| accession_id
stringlengths 9
10
| license
stringclasses 2
values | last_updated
stringlengths 19
19
| retracted
stringclasses 2
values | citation
stringlengths 22
94
| decoded_as
stringclasses 2
values | journal
stringlengths 3
48
| year
int32 1.95k
2.02k
| doi
stringlengths 3
61
| oa_subset
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-451611531310.1186/1471-2156-6-45SoftwareG2D: a tool for mining genes associated with disease Perez-Iratxeta Carolina [email protected] Matthias [email protected] Peer [email protected] Miguel A [email protected] Ontario Genomics Innovation Centre, Ottawa Health Research Institute. ON K1H 8L6, Ottawa, Canada2 Institut für Epidemiologie, GSF Forschungszentrum für Umwelt undGesundheit Ingolstadter, Landstrasse 1 D-85758, Neuherberg/Munich, Germany3 European Molecular Biology Laboratory, D-69118 Heidelberg, Germany4 Department of Bioinformatics, Max Delbrück Center for Molecular Medicine, Robert-Rössle-Strasse 10, 13092 Berlin, Germany2005 22 8 2005 6 45 45 31 3 2005 22 8 2005 Copyright © 2005 Perez-Iratxeta et al; licensee BioMed Central Ltd.2005Perez-Iratxeta et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Human inherited diseases can be associated by genetic linkage with one or more genomic regions. The availability of the complete sequence of the human genome allows examining those locations for an associated gene. We previously developed an algorithm to prioritize genes on a chromosomal region according to their possible relation to an inherited disease using a combination of data mining on biomedical databases and gene sequence analysis.
Results
We have implemented this method as a web application in our site G2D (Genes to Diseases). It allows users to inspect any region of the human genome to find candidate genes related to a genetic disease of their interest. In addition, the G2D server includes pre-computed analyses of candidate genes for 552 linked monogenic diseases without an associated gene, and the analysis of 18 asthma loci.
Conclusion
G2D can be publicly accessed at .
==== Body
Background
Mutations in genes are responsible for inherited diseases. The discovery of the association of a gene with a disease is essential for potential diagnosis and treatment and often helps understanding the mechanisms involved. The detection is usually a several step "gene hunt" where the gene is first located within a genomic region by linkage to anonymous markers followed by actual sequencing to find all genetic variation and finally to test for association of gene variants typical in diseased subjects [1].
Because the region eventually linked to a disease might contain hundreds of genes, and genotyping or directly sequencing genes of patients and controls is costly, it is important to use available information such as the complete sequence of the human genome plus a set of annotated genes and their functions (either known or predicted) to target the sequencing effort on those genes that appear to have more chances of being associated with the disease. The key to this prioritization is the expectation of the relation of a gene function to a disease (for example, a defect in a neural receptor could produce a neurological disease).
Following these ideas, we developed an algorithm to relate genes to human inherited diseases that combines the extraction of relations between phenotypes and gene functions in sequence, disease, and literature databases, with sequence similarity searches [2] (Figure 1). The main assumption of this method is that for a given disease with an undiscovered associated gene X, and a phenotypically similar disease with a known associated gene Y, some functions of the X and Y genes will be related and relevant to those phenotypes.
Figure 1 The G2D algorithm. The cylinders represent public databases. MEDLINE contains references to scientific literature annotated at the National Library of Medicine with terms from the MeSH ontology. For each disease being studied we take the MeSH C terms ('Diseases Category') from the publications associated in OMIM [3] as its keywords. For each gene we take the Gene Ontology (GO) terms [8] associated to its product in the RefSeq protein database [34] as its keywords. MEDLINE does not contain enough clinical literature to allow us to directly relate every symptom, represented by a MeSH C term, to every gene feature, represented by a GO term. Taking into account that genes relate to phenotypes by means of molecules, we can increase the robustness of the gene/phenotype relations using an intermediate association step through the MeSH D category of 'Chemicals & Drugs' (top). Accordingly, we first compute associations between MeSH C terms ('Diseases') and MeSH D terms ('Chemicals & Drugs') by their co-annotation on the same record, more specifically looking for dependences of MeSH D terms on MeSH C terms. For example, we would deduce a relation between "Alzheimer's disease" (MeSH C) and "Amyloid protein" (MeSH D) if the presence of the C term in a MEDLINE entry always implies the presence of the D term. Records in the RefSeq database contain annotations from GO that describe the protein function, and will often include a link to MEDLINE, mostly dealing with the experimental characterization of the protein. We use these links to relate MeSH D terms from the MEDLINE reference to GO terms from the sequence, again looking for GO term dependence on a MeSH D term. In this case we could deduce an association between the MeSH D term "Amyloid Protein" and the GO term "Amyloid Protein". Finally, we combine both sets of relations to obtain associations between MeSH C terms and GO terms (for example, the relation of Alzheimer's disease to the amyloid protein). To evaluate the genes associated with a particular disease we follow two directions. First, we deduce the gene functions (GO terms) related to the disease using the associations from phenotypes (MeSH C terms) describing the disease. For this, we collect the MeSH C terms found in the MEDLINE references from its corresponding OMIM entry (left), score all GO terms according to their relation to the terms in the MeSH C list (top), and finally, score all the proteins in RefSeq with the average of scores of their GO terms (right). For example, the analysis of late-onset familial Alzheimer disease (LOFAD) [9] would start by characterizing the disease with the MeSH C term "Alzheimer's Disease" among others. This would point to a series of GO terms including "Amyloid Protein" as a likely related function. One of the most related sequences in RefSeq (according to its GO annotations) would be the human amyloid beta A4 precursor protein-binding, which is annotated with the GO-term "amyloid protein". The other component of the analysis is a BLAST homology search [35] of the human genome region where the disease is mapped against the sequences stored in the RefSeq database (bottom). All hits in the region (red block) below a cut-off of E-value of 10e-10 are registered and sorted according to the score of the RefSeq protein they hit. Following our example, the analysis of the region where the LOFAD was mapped would show a gene similar to the human amyloid beta A4 precursor protein-binding annotated with the GO-term "amyloid protein": the APBA3 gene, which interacts with the Alzheimer's beta-amyloid precursor protein [12]. The analysis of LOFAD is extensively described in the Results section. Further details of the method are given in [2] and in the G2D web site.
We now implemented this method in the G2D web site allowing users to analyse diseases and genetic regions of their interest. The web site includes a collection of precomputed analyses of 552 inherited monogenic diseases stored in the OMIM database [3] that were linked to a genomic region but not yet associated with a gene. Here we describe the latest update of the method and illustrate its use via the G2D web server to propose original target genes for one monogenic disease and for asthma, a complex disease.
Implementation
The algorithm needs basically two inputs to work with: a phenotypical definition of a disease as a list of weighted MeSH terms of the C category ('Diseases' category) [4], and the definition of a genomic region in the human genome where it has to search for genes potentially associated with the disease.
In the current web implementation of G2D, we free the user from the production of a list of MeSH C terms by requiring instead the identifier of the disease in the OMIM database of human inherited diseases [3] or of a phenotypically equivalent one in that database. For example, a researcher investigating a particular variant of Alzheimer's disease not yet present in OMIM might search the database at the NCBI web server using "Alzheimer" as query term, and use one of the identifiers of the closest variant according to their phenotypes and the user's knowledge. Then, the system compiles automatically a list of MeSH C terms from those present in the MEDLINE references linked to the OMIM entry, weighting them by the fraction of linked MEDLINE references containing them. That is, a MeSH C present in all linked references will be taken more into account than one linked only to one of the references. Currently, a total of 1,663 different OMIM entries that contain enough linked MeSH C terms can be used to query the system.
The chromosomal location is a range that can be defined in three ways: two chromosomal markers (if one is given, a band of 5 Mb is taken around it), two base positions in the build hg17 of the human genome, or two cytogenetic bands (if one is given the band is taken as the range).
The current G2D web server implements new features to further guide the user in the analysis of the results, following strategies that take advantage of the combination of positional data with gene expression data to guide disease-gene data mining [5]. On the one hand, we indicate the overlap of candidate genes with predicted pseudo-genes [6], which suggests that they will not be likely expressed genes. On the other hand, we indicate the overlap of candidate genes with expressed sequence tags (ESTs), which suggests expression of the gene.
Results
Benchmark of the method
G2D is a method for the prioritization of genes according to their relation to a disease. The benchmark of such a method must be done testing diseases for which the disease-related gene is known but without providing the system with the obvious link between the target gene and the disease. Otherwise the real capacity of the system in evaluating relations between genes and diseases cannot be assessed. It is also important that the diseases are chosen in a completely unbiased manner.
Therefore, in our previous evaluation of G2D [2], we first compiled a set of 100 monogenic diseases randomly chosen among those with known associated gene according to their entries in the LocusLink database [3]. For the sake of comparison we will use the same set and benchmark procedure in this work, noting that any differences between the two benchmark results will be due exclusively to the newer versions of the databases used, since the method was not changed since then.
In brief, we attempted to remove all experimental and genetic information on the diseases used for the benchmark by producing a version of MEDLINE devoid of references matching the disease names. In this way, we wanted to assure that our algorithm would be discovering the associations of genes to a test-disease based on information extracted from a different disease. Matching entries were found by querying MEDLINE via PubMed using the expansion of the query with synonyms (such as "cancer" and "tumour") and scanning both the abstract of the reference and the MeSH terms associated with it.
For each test-disease, genes in a 30 Mb region centred on the target gene were scored; on average this region contained about 300 genes. This size was taken because such was the average size of the loci of other diseases mapped to a genomic region but for which no gene had been found yet to be related.
The benchmark results were better than those obtained in our previous analysis [2]. The target gene was identified in 87 of the 100 test diseases (55 previously). The target gene was among the 8 best scoring genes in 47 cases (25 previously), and among the 30 best scoring genes in 62 cases (50 previously) (see details of this and previous benchmark in the Supplement page of the G2D web site). The improved system performance was due in part to the improved quality of the hg17 human genome assembly (build 35, supported by the UCSC Genome Bioinformatics Site; May 2004 freeze) [7]. A performance improvement was likely also obtained from the increased number of sequences in RefSeq, and of their functional annotations with Gene Ontology terms (GO terms; [8]), a system for the description of genetic functionality that is also in continuous expansion and refinement.
Analysis of a monogenic disease: late-onset Alzheimer disease
To illustrate how to use the G2D server for analysis of a monogenic disease, we will be using the recent genetic linkage of a type of late-onset familial Alzheimer disease (LOFAD) to a locus on chromosome 19p13.2 [9] for which no responsible gene has been yet found. Although Alzheimer disease is genetically heterogeneous, there are some common themes among the functions of the proteins related, such as their interaction with the beta-amyloid protein, which can be used to track down a new candidate as we will show.
To search for genes that could be associated with the disease in a chromosomal region two inputs are needed: an OMIM identifier of the disease (or of a similar one), and a chromosomal location. The user defines these two inputs in the "COMBO BOX" (see Figure 2a). Following the example, we type in the "PHENOTYPE BOX" OMIM id window, 104310, which is the OMIM id corresponding to another LOFAD, the AD2 that was linked to the gene APOE4 [10]. In the "LOCATION BOX" we define the chromosomal range to be analyzed as band 19p13 on chromosome 19, intentionally wider than the 19p13.2 locus reported for this disease [9]. To do this, we specify that we are defining the location by band by selecting the option "band(s)", and we type the name of the band, p13 in the first window. Finally we select the chromosome where the band is located, 19.
Figure 2 Example of analysis of a monogenic disease. (a) The data defining the phenotype of the disease (in this case the OMIM identifier of an equivalent disease) and the region where it was mapped are given in the COMBO box. (b) The results window displays the MeSH C terms derived from the links to MEDLINE found in the OMIM entry, and the resulting scores for the GO terms. The green arrows allow traveling the MeSH C/MeSH D/GO network of connections back and forth. (c) Further down in the results window, the list of candidates displays the position of the BLASTx hits [35] in the chromosomal region (dark green bar over the light green bar) and of the hits in the matching protein sequence (dark red bars over the light red bar). Each hit in the genome is linked to the UCSC Genome Browser ("U" link). (d) The UCSC Genome Browser allows examining the genes known or predicted that overlap with the match linking to very useful databases and resources.
The "OUTPUT BOX" allows changing the restrictions on the number of BLAST hits in the region to be displayed. This is done by modifying two parameters: the maximum BLAST E-value threshold in order to report a hit of sequence similarity in the region to a sequence in RefSeq; and the total number of hits to be displayed, considering that those are sorted by decreasing relatedness to the disease (the derived GO score). More restrictive values (a lower BLAST E-value threshold result in a smaller number of candidates) reduce the time needed for the production and download of the results via a web page, although at the risk of missing an interesting candidate. The default values should be appropriate for a first exploratory search. In this example, we keep the default thresholds (E-value = e-10, and 20 candidates displayed).
The result is a search in a region of 19.8 Mb. The only MeSH C term that is extracted from the MEDLINE links from the OMIM entry, is "Alzheimer Disease" (present in four links). The GO term that receives the highest score is "copper homeostasis", then quite close "amyloid protein". Both are pointed through the MeSH D term "Amyloid beta-Protein Precursor" that is strongly associated with the MeSH C term "Alzheimer Disease". This information can be examined by traveling the network of relations between MeSH C, MeSH D, and GO terms, following the links marked as green arrows (Figure 2b).
The list of candidates contains reports for hits in the 19.8 Mb region to protein sequences in RefSeq (with a BLAST E-value of or below 1e-10). Note that multiple hits may be pointing to the same gene in the region, or that a hit may be pointing to multiple genes in the region (if for example there are two similar genes in the area).
The first hit is given by similarity to a sequence from rat, follistatin, identified by the GO term "serine protease inhibitor", which scores 0.0108. The next one is more appealing, given by similarity to the human amyloid beta A4 precursor protein-binding, which is annotated with the better GO-term "amyloid protein" (Figure 2c). The hits from positions 3,705,192 to 3,701,877 in the negative strand of Chromosome 9, match the N-terminal part (positions 420 to the end, 749) of that protein, which contains a PTB domain (phosphotyrosine-binding domain; positions 367–533) and two PDZ domains (positions 577–655 and 669–735), according to the SMART protein domain search online-tool [11]. PDZ domains are implied in polypeptide binding.
The genomic region of each match can be examined via the UCSC Genome Bioinformatics Site following the links marked with a "U" (see Figure 2d). In this case, the UCSC browser allows confirming that the seven matches of candidate 2 actually overlap with the APBA3 gene that encodes a protein with a similar domain distribution in its C-terminal part, which interacts with the Alzheimer's beta-amyloid precursor protein [12]. As such, this is an ideal candidate to search for mutations in patients of Alzheimer disease.
However, in this particular case, the linkage analysis displayed a maximum of intensity ratio between the markers D19S216 (at position 4.9 Mb) and D19S221 (at position 12.5 Mb), therefore 1.2 Mb further from the telomere than the position of the APBA3 gene, and this led to the group researching this disease to disregard this gene that otherwise they found biologically interesting (Gerard D. Schellenberg, personal communication).
Analysis of a complex disease: asthma
G2D can be likewise used for complex diseases linked to multiple loci. Users have to repeat the analysis with the phenotype of the disease for each of the linked regions separately. In order to illustrate the analysis of a complex disease, we have pre-computed the analysis of candidate genes for asthma. The results of this analysis are available at the G2D web site. Here, we describe the steps taken to produce this analysis.
Asthma is a common chronic airway disease caused by multigenic influences, many environmental factors, and a high interaction between various risk factors [13]. Disease loci have been mapped to different chromosomes. Although asthma is the complex disease with the largest number of genome scans (currently more than 15 studies) [14], few genes have been associated with the disease (as far as we know; those are ADAM33, DPP10, PHF11 [15], GPRA [16], and Vitamin D receptor [17]). We used the data stored in the Asthma and Allergy Gene Database [18] as an input to G2D to propose candidates in several genomic regions.
To obtain MeSH C terms related to asthma we queried MEDLINE via PubMed with "asthma [tw] AND 'Case Report' [MH]". A total of 4,317 references was retrieved. The most related GO terms were peptidyl-dipeptidase, interleukin-5 receptor, fluid secretion, peroxidase, and protein kinase inhibitor. We then followed up all ten genome scans fully available in the Asthma and Allergy Gene Database [18]. One study provided genome scans from three populations [19] while all others reported one single genome scan [20-26]. Only one trait per study was selected for this analysis (usually asthma except for two studies where total IgE was taken [20,25]). If available, P-values were taken directly from the data files submitted to us. Otherwise LOD scores were converted into P-values by subtracting 1 minus the probability from a chi-squared distribution with 1 degree of freedom of the individual LOD score that has been multiplied before with the 2 fold logarithm of 10 (P = 1 - Pχ2 (2 log10·LOD, 1)). This is a less conservative approximation than suggested elsewhere [27] but used throughout the database. Finally, 233 unique markers with linkage P-values below 0.05 were then selected from these 10 scans. Microsatellite markers were grouped according to recent Marshfield genetic map data. We then defined linkage regions by having at least four supporting markers in a 20 cM interval. This reduced the linkage dataset to 18 regions that were expanded 6 Mb both to the 3' and 5' direction to allow for some imprecision in marker positioning. G2D was used to prioritize genes according to their relation to asthma in each of the regions. The lists of candidate genes obtained for each of the 18 linkage regions are available at the G2D web site.
As a proof-of-principle, we tested the system with one gene recently linked to asthma: ADAM33 in the cytogenetic band 20p13 [28]. It was discovered in a region so far not linked to asthma by at least two different studies and therefore not in our list of target regions. We applied the system to the band defining the disease via OMIM id 600807 ('Asthma susceptibility') and checking the best 3000 scored sequences in RefSeq. ADAM33 was identified (as the 9th best candidate in band 20p13 after removing redundant hits) because of its sequence similarity to ADAM9 encoding for a protein with a disintegrase and metalloproteinase domain 9.
Discussion
We have implemented in a public web server a method that allows the ranking of genes in a region of the human genome according to their possible relation to a disease. Both the region and the disease can be defined by the user. Since the method is computationally very intensive, mostly due to the amount of genes used for the sequence similarity analysis, we introduced limitations in the maximum size of the genomic region to scan and in the number of candidates to report. Still, the analysis might take a few minutes depending on the load of the server.
We have updated the method and its benchmark with respect to the original version (G2D, [2]) using newer database versions, observing an improvement in performance almost certainly due both to the increased accuracy of the human genome sequence, and to the continuous functional annotation efforts on human sequences and their homologues in other organisms.
In the current version, a test with 100 diseases of known genetic cause indicated that G2D finds the responsible gene in 87 cases out of a pool of 300 genes (on average), the target gene being among the 8 best scoring genes in the 47 of the successful 87 cases.
It must be noted that the identification of candidate genes by G2D relies partly on the sequence similarity comparison of (query) proteins to parts of the genome, and that it is advisable to examine the extent and position of this similarity. For example, the region of similarity could be restricted to a fragment of the query protein not being responsible for the functionality that might be associated to the disease. Moreover, the method could be pointing to a pseudogene. To support the human examination of the results, we indicate the positions of the sequence similarity match both in the query protein and in the genome. We also took advantage of existing information on pseudogene prediction, and we added links to the UCSC genome browser. This allows putting the candidate genes in the context of the latest genetic knowledge, which has been shown to be of great help when identifying genes involved in disease [5].
Although G2D was originally devised for the analysis of single genetic regions, we encourage researchers working on complex diseases to apply the system independently for each of the genomic regions associated with a complex disease (as illustrated in the Results section), provided that the quality of the linkage analysis is good. Actually, it has been shown that when multiple associations of genetic variation to a disease are demonstrated there is a great likelihood that each of them separately constitutes a risk factor contributing to the disease [29]. Accordingly, it is gaining wider acceptance that the classification of a disease as monogenic might be more the result of our lack of knowledge of all the genes involved in that disease than to reality [1,30]. The conceptual separation between monogenic and complex diseases might be illusory.
As far as we know, we have created a unique resource. Other efforts applying data-mining to the study of genes associated with diseases have a different focus. Mainly, none of them uses sequence similarity searches to assess candidate genes so that, in principle, if a disease-related gene lacks functional or protein domain annotation, or if it is not even predicted to be a gene, it will not be detected by such methods.
For example, Freudenberg and Propping [31] also use phenotype information associated with diseases and the GO annotations of genes, but they cluster diseases with similar phenotypes and pool the GO terms of the genes associated. This means that their method cannot distinguish whether a gene will be related to one particular disease but to a pool of diseases. Although the method was tested with a leave-one-out cross-validation in a set of 878 diseases from OMIM with already known associated gene, their results are not as good as ours with 1/3 of cases having the disease-related gene among 160 candidates and the remaining 2/3 being among 1600. However, positional information is not taken into account. The system is not accessible through a web server.
Turner et al. [32] developed POCUS, a method for the prediction of genes related to diseases linked to more than one genomic region. This is exclusively based on the common GO terms and protein domains found among the genes from multiple loci. The performance of the method is shown in a set of 29 diseases and the genes associated with each of them (between three and eleven), by using variable sizes of artificial loci generated around the target genes. Even when using the smallest size (with an average of 20 genes in each loci), the rate of identification of disease genes is of 60 out of 163 disease-genes, with 56 false positives. For the other two larger sizes assayed they find a candidate for only 5 and 4 of the 29 test diseases (using an average of 94, and 187 genes, per loci, respectively) with an increase in the number of false positives. This poor performance in comparison to the methods above is surely due to the fact that they do not use phenotypic information. This, together with the requirement for multiple loci, makes this method complementary to the one above and hard to compare. The method is not available as a web server though some computer programs are given as supplementary material [32].
Finally, we mention the work from van Driel et al. (GeneSeeker, [5]) more focused on the linkage of information related to genes and disease from multiple public databases. GeneSeeker relies on positional information of genetic linkage (to one region), and includes genetic expression information that is extracted for the genes in the region from their entries in sequence databases and MEDLINE references linked therein (but not from ESTs). The method is tested in only ten human malformation syndromes for which the associated gene is known, using an ad-hoc list of organ terms for each one. Obviously, the method would not be able to find a disease-related gene lacking a link indicating expression in an organ. The average of genes in the loci examined was 165. The results vary greatly depending on how the list of terms is applied. In the least restrictive test the gene is found in all the ten cases but among an average of 22 candidate genes. Again, this method is not using any functional information about the genes analyzed or of the disease phenotype, so it is not surprising that its performance is inferior to G2D. Contrary to the methods previously discussed, this method is accessible through a web server.
Conclusion
Irrespective of the variation between these methods and their complementarity, they represent the effort of the scientific community to put together existing resources for the help of the geneticists searching for disease-related genes. It would be interesting to measure how these efforts are accelerating the pace of disease-related gene discovery. In this respect, we note that at the time we applied G2D for the analysis of a number of monogenic diseases in OMIM for the first time (June, 2001), there was a total of 455 monogenic diseases linked to a genomic region without associated gene. In our second analysis of OMIM that is presented in the G2D web server (July, 2005), the amount of such diseases has risen to 552. In the intervening three years, 104 diseases were associated with genes, representing the successful completion of a gene hunt. However, another 201 genetic diseases were newly mapped to the human genome, leading to a net increase of 97 diseases in our list of analyses. Although the human genome sequence is completed and there is a number of resources available to researchers to assist them in completing the last stages of disease-gene hunt, the bottleneck of the process of characterizing genes associated with diseases is still the definition of the disease-related gene at the end of the search and not the genetic linkage of the disease, or the discovery of new diseases with an inheritance pattern.
Further directions for improvement suggested have been to include other types of information about disease related genes [33], and to take into account the functional links existing between genes associated with a complex disease [32].
Finally, we note that another important factor for improvement is the feedback and collaboration with the experimental groups that are benefited by these new data-mining methods. As with other efforts in Bioinformatics, only by close collaboration between computational and experimental groups can we expect a real advance on the methodology. For this, both parts have heavy duties: computational groups must explain these tools cleanly and clearly, making them openly available in a stable and up-to-date fashion, swift to adapt to the suggestions of the users; but we should not forget that experimental groups must try the tools made for them and in doing that give constructive feedback and fair acknowledgment to their authors.
Authors' contributions
CP, PB, and MA conceived G2D and associated benchmarks. CP and MA computed the benchmarks, programmed, and maintain the web server. MW contributed to the asthma study. All authors participated in the preparation of the manuscript. All authors have read and approved the manuscript.
Acknowledgements
We thank the NCBI for the development of a growing number of very useful databases and for providing us with the MEDLINE database for local usage. We thank UCSC for keeping a number of versions of the human genome, for their genome browser, and for their quick and very positive response to our suggestions in a number of occasions. Last but not least, we are grateful to the members of the Biocomputing groups at the European Molecular Biology Laboratory, where large part of this work was developed, for countless inspiring discussions over the years, and especially to Dr. Yuan for initial support to the G2D and other web servers developed by us.
==== Refs
Dean M Approaches to identify genes for complex human diseases: lessons from Mendelian disorders Hum Mutat 2003 22 261 274 12955713 10.1002/humu.10259
Perez-Iratxeta C Bork P Andrade MA Association of genes to genetically inherited diseases using data mining Nat Genet 2002 31 316 319 12006977
Wheeler DL Church DM Edgar R Federhen S Helmberg W Madden TL Pontius JU Schuler GD Schriml LM Sequeira E Suzek TO Tatusova TA Wagner L Database resources of the National Center for Biotechnology Information: update Nucleic Acids Res 2004 32 D35 D40 14681353 10.1093/nar/gkh073
MeSH [http://www.nlm.nih.gov/mesh/]
van Driel MA Cuelenaere K Kemmeren PP Leunissen JA Brunner HG A new web-based data mining tool for the identification of candidate genes for human genetic disorders Eur J Hum Genet 2003 11 57 63 12529706 10.1038/sj.ejhg.5200918
Torrents D Suyama M Zdobnov E Bork P A genome-wide survey of human pseudogenes Genome Res 2003 13 2559 2567 14656963 10.1101/gr.1455503
Kent WJ Sugnet CW Furey TS Roskin KM Pringle TH Zahler AM Haussler D The human genome browser at UCSC Genome Res 2002 12 996 1006 12045153 10.1101/gr.229102. Article published online before print in May 2002
Harris MA Clark J Ireland A Lomax J Ashburner M Foulger R Eilbeck K Lewis S Marshall B Mungall C Richter J Rubin GM Blake JA Bult C Dolan M Drabkin H Eppig JT Hill DP Ni L Ringwald M Balakrishnan R Cherry JM Christie KR Costanzo MC Dwight SS Engel S Fisk DG Hirschman JE Hong EL Nash RS Sethuraman A Theesfeld CL Botstein D Dolinski K Feierbach B Berardini T Mundodi S Rhee SY Apweiler R Barrell D Camon E Dimmer E Lee V Chisholm R Gaudet P Kibbe W Kishore R Schwarz EM Sternberg P Gwinn M Hannick L Wortman J Berriman M Wood V de la Cruz N Tonellato P Jaiswal P Seigfried T White R The Gene Ontology (GO) database and informatics resource Nucleic Acids Res 2004 32 Database issue D258 D261 14681407
Wijsman EM Daw EW Yu CE Payami H Steinbart EJ Nochlin D Conlon EM Bird TD Schellenberg GD Evidence for a novel late-onset Alzheimer disease locus on chromosome 19p13.2 Am J Hum Genet 2004 75 398 409 15248153 10.1086/423393
Corder EH Saunders AM Strittmatter WJ Schmechel DE Gaskell PC Small GW Roses AD Haines JL Pericak-Vance MA Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families Science 1993 261 921 923 8346443
Letunic I Copley RR Schmidt S Ciccarelli FD Doerks T Schultz J Ponting CP Bork P SMART 4.0: towards genomic data integration Nucleic Acids Res 2004 32 D142 D144 14681379 10.1093/nar/gkh088
Tanahashi H Tabira T Genomic organization of the human X11L2 gene (APBA3), a third member of the X11 protein family interacting with Alzheimer's beta-amyloid precursor protein Neuroreport 1999 10 2575 2578 10574372
Tabor HK Risch NJ Myers RM Opinion: Candidate-gene approaches for studying complex genetic traits: practical considerations Nat Rev Genet 2002 3 391 397 11988764 10.1038/nrg796
Altmuller J Palmer LJ Fischer G Scherb H Wjst M Genomewide scans of complex human diseases: true linkage is hard to find Am J Hum Genet 2001 69 936 950 11565063 10.1086/324069
Wills-Karp M Ewart SL Time to draw breath: asthma-susceptibility genes are identified Nat Rev Genet 2004 5 376 387 15143320 10.1038/nrg1326
Laitinen T Polvi A Rydman P Vendelin J Pulkkinen V Salmikangas P Makela S Rehn M Pirskanen A Rautanen A Zucchelli M Gullsten H Leino M Alenius H Petays T Haahtela T Laitinen A Laprise C Hudson TJ Laitinen LA Kere J Characterization of a common susceptibility locus for asthma-related traits Science 2004 304 300 304 15073379 10.1126/science.1090010
Poon A Laprise C Lemire M Montpetit A Sinnett D Schurr E Hudson TJ Association of vitamin D receptor genetic variants with susceptibility to asthma and atopy Am J Respir Crit Care Med 2004 170 967 973 15282199 10.1164/rccm.200403-412OC
Immervoll T Wjst M Current status of the Asthma and Allergy Database Nucleic Acids Res 1999 27 213 214 9847183 10.1093/nar/27.1.213
CSGA A genome-wide search for asthma susceptibility loci in ethnically diverse populations. The Collaborative Study on the Genetics of Asthma (CSGA) Nat Genet 1997 15 389 392 9090385
Daniels SE Bhattacharrya S James A Leaves NI Young A Hill MR Faux JA Ryan GF le Souef PN Lathrop GM Musk AW Cookson WO A genome-wide search for quantitative trait loci underlying asthma Nature 1996 383 247 250 8805698 10.1038/383247a0
Dizier MH Besse-Schmittler C Guilloud-Bataille M Annesi-Maesano I Boussaha M Bousquet J Charpin D Degioanni A Gormand F Grimfeld A Hochez J Hyne G Lockhart A Luillier-Lacombe M Matran R Meunier F Neukirch F Pacheco Y Parent V Paty E Pin I Pison C Scheinmann P Thobie N Vervloet D Kauffmann F Feingold J Lathrop M Demenais F Genome screen for asthma and related phenotypes in the French EGEA study Am J Respir Crit Care Med 2000 162 1812 1818 11069818
Laitinen T Daly MJ Rioux JD Kauppi P Laprise C Petays T Green T Cargill M Haahtela T Lander ES Laitinen LA Hudson TJ Kere J A susceptibility locus for asthma-related traits on chromosome 7 revealed by genome-wide scan in a founder population Nat Genet 2001 28 87 91 11326283 10.1038/88319
Ober C Tsalenko A Parry R Cox NJ A second-generation genomewide screen for asthma-susceptibility alleles in a founder population Am J Hum Genet 2000 67 1154 1162 11022011
Wjst M Fischer G Immervoll T Jung M Saar K Rueschendorf F Reis A Ulbrecht M Gomolka M Weiss EH Jaeger L Nickel R Richter K Kjellman NI Griese M von Berg A Gappa M Riedel F Boehle M van Koningsbruggen S Schoberth P Szczepanski R Dorsch W Silbermann M Wichmann HE A genome-wide search for linkage to asthma. German Asthma Genetics Group Genomics 1999 58 1 8 10333435 10.1006/geno.1999.5806
Xu J Postma DS Howard TD Koppelman GH Zheng SL Stine OC Bleecker ER Meyers DA Major genes regulating total serum immunoglobulin E levels in families with asthma Am J Hum Genet 2000 67 1163 1173 11023809 10.1086/303081
Yokouchi Y Nukaga Y Shibasaki M Noguchi E Kimura K Ito S Nishihara M Yamakawa-Kobayashi K Takeda K Imoto N Ichikawa K Matsui A Hamaguchi H Arinami T Significant evidence for linkage of mite-sensitive childhood asthma to chromosome 5q31-q33 near the interleukin 12 B locus by a genome-wide search in Japanese families Genomics 2000 66 152 160 10860660 10.1006/geno.2000.6201
Nyholt DR All LODs are not created equal Am J Hum Genet 2000 67 282 288 10884360 10.1086/303029
Van Eerdewegh P Little RD Dupuis J Del Mastro RG Falls K Simon J Torrey D Pandit S McKenny J Braunschweiger K Walsh A Liu Z Hayward B Folz C Manning SP Bawa A Saracino L Thackston M Benchekroun Y Capparell N Wang M Adair R Feng Y Dubois J FitzGerald MG Huang H Gibson R Allen KM Pedan A Danzig MR Umland SP Egan RW Cuss FM Rorke S Clough JB Holloway JW Holgate ST Keith TP Association of the ADAM33 gene with asthma and bronchial hyperresponsiveness Nature 2002 418 426 430 12110844 10.1038/nature00878
Lohmueller KE Pearce CL Pike M Lander ES Hirschhorn JN Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease Nat Genet 2003 33 177 182 12524541 10.1038/ng1071
Peltonen L McKusick VA Genomics and medicine. Dissecting human disease in the postgenomic era Science 2001 291 1224 1229 11233446 10.1126/science.291.5507.1224
Freudenberg J Propping P A similarity-based method for genome-wide prediction of disease-relevant human genes Bioinformatics 2002 18 S110 S115 12385992
Turner FS Clutterbuck DR Semple CA POCUS: mining genomic sequence annotation to predict disease genes Genome Biol 2003 4 R75 14611661 10.1186/gb-2003-4-11-r75
Lopez-Bigas N Ouzounis CA Genome-wide identification of genes likely to be involved in human genetic disease Nucleic Acids Res 2004 32 3108 3114 15181176 10.1093/nar/gkh605
Pruitt KD Tatusova T Maglott DR NCBI Reference Sequence project: update and current status Nucleic Acids Res 2003 31 34 37 12519942 10.1093/nar/gkg111
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
|
16115313
|
PMC1208881
|
CC BY
|
2021-01-04 16:30:39
|
no
|
BMC Genet. 2005 Aug 22; 6:45
|
utf-8
|
BMC Genet
| 2,005 |
10.1186/1471-2156-6-45
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-561611783610.1186/1472-6963-5-56Research ArticleAssessing health centre systems for guiding improvement in diabetes care Si Damin [email protected] Ross [email protected] Christine [email protected] Michelle [email protected] Allison [email protected] Gary [email protected] Joan [email protected] Tarun [email protected] Menzies School of Health Research, Charles Darwin University, PO Box 41096, Darwin, NT, Australia2 Northern Territory Department of Health and Community Services, Darwin, NT, Australia3 School for Social and Policy Research, Charles Darwin University, Darwin, NT, Australia2005 24 8 2005 5 56 56 3 5 2005 24 8 2005 Copyright © 2005 Si et al; licensee BioMed Central Ltd.2005Si et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Aboriginal people in Australia experience the highest prevalence of diabetes in the country, an excess of preventable complications and early death. There is increasing evidence demonstrating the importance of healthcare systems for improvement of chronic illness care. The aims of this study were to assess the status of systems for chronic illness care in Aboriginal community health centres, and to explore whether more developed systems were associated with better quality of diabetes care.
Methods
This cross-sectional study was conducted in 12 Aboriginal community health centres in the Northern Territory of Australia. Assessment of Chronic Illness Care scale was adapted to measure system development in health centres, and administered by interview with health centre staff and managers. Based on a random sample of 295 clinical records from attending clients with diagnosed type 2 diabetes, processes of diabetes care were measured by rating of health service delivery against best-practice guidelines. Intermediate outcomes included the control of HbA1c, blood pressure, and total cholesterol.
Results
Health centre systems were in the low to mid-range of development and had distinct areas of strength and weakness. Four of the six system components were independently associated with quality of diabetes care: an increase of 1 unit of score for organisational influence, community linkages, and clinical information systems, respectively, was associated with 4.3%, 3.8%, and 4.5% improvement in adherence to process standards; likewise, organisational influence, delivery system design and clinical information systems were related to control of HbA1c, blood pressure, and total cholesterol.
Conclusion
The state of development of health centre systems is reflected in quality of care outcome measures for patients. The health centre systems assessment tool should be useful in assessing and guiding development of systems for improvement of diabetes care in similar settings in Australia and internationally.
==== Body
Background
Indigenous Australians experience the highest prevalence of diabetes in the country, an excess of preventable complications and early death [1]. Published studies have demonstrated that effective diabetes management in Indigenous primary care can improve the process and outcomes of care [2-4]. Reported approaches to improving quality of care include use of recall and reminder systems, audit and feedback of clinical performance, structured clinical care, and specialist involvement in primary care. This evidence on the effectiveness of approaches to care in the Indigenous Australian population is consistent with the approach reflected in the Chronic Care Model developed in the USA [5,6]. The Chronic Care Model is comprised of six major components that have been shown to be important internationally to chronic illness care: 1) health care organisation, 2) community linkages, 3) self-management support, 4) decision support, 5) delivery system design, and 6) clinical information systems. This model has been extensively implemented in community health centres and hospitals in the USA to assess system support for chronic care and to identify areas for further improvement [7].
In the context of Australia's Northern Territory (NT) health service providers' collaborative efforts to improve prevention and management of chronic disease in primary health care [8], the Audit and Best practice for Chronic Disease (ABCD) project was implemented on the expectation that community health centre staff and managers would benefit from an improved understanding of the status of health centre systems in order to appropriately plan for improvement. A central component of the ABCD project was therefore an assessment of health centre systems as outlined in the Chronic Care Model. This paper reports on the association of the systems assessment with the quality of diabetes care for ABCD participating centres at baseline.
Methods
Study design, setting and selection of participating health centres
This cross-sectional study was conducted in the Top End, an area covering one-third of NT. Of 60 Aboriginal communities with a population of 50 or more that are located in this area, 45 have a health centre within the community [9]. The selection of 12 health centres for participation in the study aimed to reflect the range in size, geographic location and governance arrangements existing in Top End communities (Figure 1). Community populations ranged from 180 to 1500. Five community health centres were managed by the NT Department of Health and Community Services, four by Aboriginal Health Boards, and three by Aboriginal Medical Services.
Figure 1 Distribution of 12 participating communities in the Top End of the Northern Territory, Australia.
Community members who met all of the following criteria were included in the study: 1) a definite diagnosis of type 2 diabetes according to health centre records; 2) identified as Aboriginal; 3) aged 16 years or older; and 4) lived in the community for 6 months or more during the previous 12 months. A random sample of 30 records was drawn from four of the twelve community health centres where more than 30 eligible people were identified. In the other eight centres the records of all eligible people were included.
Measurement and data collection
System mapping
The Assessment of Chronic Illness Care (ACIC, Version 3.5) scale [10], a practical tool based on the Chronic Care Model [11], was used to evaluate the status of health centre systems to support chronic illness care. A minor adaptation was made to facilitate its use in the local setting. The adapted scale consists of 34 items covering the six components of the Chronic Care Model and an additional domain which denotes the level of integration of the six components: health care organisation (3 items), community linkages (4 items), self-management (4 items), clinical decision support (3 items), clinical delivery system (9 items), clinical information system (6 items), and integration (5 items). Compared with the original ACIC scale, this adapted version included three additional items (cultural competence, pathology management, and pharmacy management) in the clinical delivery system domain, to reflect specific features of interest in NT centres.
A group of health centre staff (manager, doctor, nurse, and/or Aboriginal Health Worker when available) was guided by researchers on how to complete the scale. The answer to each item in the adapted scale requires recording of a score in the range of 0–11. The scale includes a set of prompts (see Additional file 1) to increase standardisation and reproducibility in scoring and staff are requested to provide a qualitative justification for their score in relation to these prompts (see example in Additional file 2). The scores are categorised as: 0–2 (limited support); 3–5 (basic support); 6–8 (good support); and 9–11 (fully developed support). The score and justification for each item were obtained by arriving at a consensus among participating staff members. The mean was calculated from individual item scores to create a component score, and the mean of 6 component and integration scores formed the overall system score for the community health centre. The average time to complete the scale was 2 hours.
Quality of diabetes care
Quality of diabetes care was measured in terms of care processes and intermediate outcomes through auditing of medical records. The audit tool lists 28 services which the clinical guidelines in current use across the NT recommend for delivery at regular intervals for all people with diabetes (Table 1) [12]. A service was assessed as delivered if there was a record of delivery within the appropriate period preceding the audit. The overall adherence to delivery of scheduled services for each patient was calculated by dividing the sum of services delivered by 28 (the total number of scheduled services), and expressing this as a percentage.
Table 1 Adherence to delivery of scheduled services for study participants (N = 295)
Process items Scheduled interval (months) % of patients receiving services 95% CI*
Basic measurement
Weight 3 47% 41%–53%
Height Any time 32% 27%–38%
BMI 12 16% 12%–21%
Waist circumference 3 23% 18%–28%
BP 3 63% 57%–69%
Eye check
Visual acuity 12 40% 35%–46%
Cataracts 12 28% 23%–34%
Fundi (dilated pupils) 12 34% 29%–40%
Ophthalmologist review 24 34% 29%–40%
Feet check
Check done 3 20% 16%–25%
Sensation 3 9% 6%–13%
Peripheral pulses 3 8% 5%–12%
Pressure areas 3 7% 5%–11%
Infections 3 8% 6%–12%
Laboratory investigations
BSL (finger prick or venous) 3 61% 55%–67%
HbA1c† 6 41% 35%–47%
Fasting lipids 12 27% 22%–33%
Total cholesterol 12 56% 50%–62%
Urine – Dipstix 3 20% 15%–25%
Creatinine 12 65% 59%–71%
ACR 12 54% 48%–59%
Counselling/advice
Diet 3 15% 11%–19%
Activity 3 13% 9%–17%
Smoking 3 10% 7%–14%
Alcohol 3 9% 6%–13%
Diabetes medications 3 10% 7%–14%
Immunisations
Flu vac. 12 54% 48%–59%
Pneumo vac. 5 yrs 73% 68%–78%
* 95% CIs were calculated adjusting for clustering by health centre.
† During the past 12 month period, 70% (95%CI: 64%–75%) of patients had HbA1c tested.
Intermediate outcomes of diabetes care include three measures: the values of the most recent HbA1c, blood pressure, and total cholesterol within 12 months prior to the audit. Control of these 3 measures is essential in preventing or delaying the onset of macro- and micro-vascular complications [13].
Statistical analysis
Means and proportions were used to summarise normally distributed continuous and binomial data respectively, and 95% confidence intervals were calculated after adjustment for clustering by heath centre. As the ACIC result represented a score ranking from 0 to 11, nonparametric measures were used to describe ACIC results in terms of median, interquartile range, and range.
Multiple linear regression analysis was used to determine the independent association of each health system component with the overall adherence to delivery of scheduled services, with patient level variables (age and sex) treated as covariates.
The associations between system components and intermediate outcomes of diabetes care were assessed using multivariate probit regression [14]. This statistical procedure allows three dependent variables Y1(HbA1c control), Y2 (blood pressure control) and Y3 (total cholesterol control) to be jointly regressed on the same independent variables (6 system components and participants' age and sex) in one model. The dependent variables are dichotomous and defined as follows: Y1 = 1 if HbA1c level < 8.0%, and Y1 = 0 if HbA1c = 8.0% or no HbA1c tested within the past 12 months; Y2 = 1 if blood pressure < 140/90 mmHg, and Y2 = 0 if blood pressure ≥ 140/90 mmHg or no blood pressure checked within the past 12 months; Y3 = 1 if total cholesterol < 5.5 mmol/L, and Y3 = 0 if total cholesterol ≥ 5.5 mmol/L or no total cholesterol tested within the past 12 months. The associations with a single intermediate outcome, and the three as a group, were separately tested. All analyses were performed using Stata version 8.2 [15].
Ethics approval for the study was obtained from the Top End Health Research Ethics Committee, including the Indigenous sub-committee.
Results
The records of 295 people with diabetes (116 males and 179 females) were included in the study. The average age of participants was 49 years (range 16–87). Diabetes duration averaged 6.5 years. Seventy-seven percent (95%CI: 71%–81%) of participants had attended their community health centres within the previous 3 months.
ACIC scores for community health centre systems
The overall ACIC score ranged from 2.6 to 5.3 with a median of 4.3 (Figure 2). The median ACIC scores for system components ranged from 2.5 (component integration) to 5.4 (clinical information system). Community health centres had distinct areas of strengths and weakness as reflected in the qualitative justification in each system component and summarised below.
Figure 2 Assessment of Chronic Illness Care (ACIC) component scores for participating community health centres (N = 12).
Organisational influence
five of twelve health centres included chronic disease care goals in their business plan and had one or more chronic disease coordinators in place. One health centre had an established continuous improvement program. Public health nurse positions, recently developed by the Department of Health and Community Services, were perceived by health service staff as an improvement strategy for chronic illness care. Six health centres reported use of Enhanced Primary Care Medicare claims as incentives for chronic care planning.
Community linkages
All participating health centres reported going out into the community to 'collect' people for specialist visits as a common activity. Some health centres worked together with other organisations and ran community-based programs such as 'school nutrition programs', 'healthy lifestyle promotional days' at the local store, 'healthy kids week', or 'tobacco prevention week'. However, all health service staff reported that acute care demands often prevented the development of community relationships that may have improved chronic care.
Self-management support
There was limited uptake and documentation of self-management activities in health centres, such as goal setting with clients and patient education. One-to-one education was the most commonly mentioned approach for the delivery of patient education. Peer or group education was seldom used.
Clinical decision support
Clinical guidelines (e.g. CARPA standard treatment manual) [12] were universally distributed to centres to facilitate clinical decision-making, and most health centres reported integration of these guidelines into routine care. Involvement of specialists in primary care was mainly through conventional referrals, and visiting specialist services to health centres were perceived generally as not frequent enough to meet needs.
Clinical delivery system
Most health centres suffered from a shortage of staff, and only six communities had a resident doctor. Few centres adopted planned visits for delivering multiple services to clients. All centres reported that systems for collecting and reporting of pathology specimens and for dispensing medication were in place. Most respondents perceived their centres to be delivering culturally safe and appropriate services.
Clinical information systems
Computerised information systems were installed in 11 health centres, with four different clinical information software systems. The remaining centre used a paper-based information system only. Recall systems were operational for eight health centres (6 computerised, 1 paper-based, and 1 with both). Most centres had organised and easily accessible patient records. Most information systems lacked the capacity (or were not used) to supply staff with population-based information on quality of chronic illness care.
Integration of system components
The integration of system components was the least developed area. For example, the existing information system was not integrated with patient self-management in terms of documenting goals and activities; and there was limited sharing of information between out-of-clinic community programs and clinical services.
Quality of diabetes care
Adherence to delivery of scheduled services varied across different categories of services (Table 1). Adherence was relatively higher for immunisation and laboratory investigations, followed by eye checks and basic measurement. Least attention was paid to feet checks and counselling services. The overall adherence to delivery of scheduled services for individuals averaged 31% (range 0–93%).
For intermediate outcomes (Table 2), the most recent value of HbA1c was below 8% for only about a quarter of participants, and most recent value of BP was less than 140/90 for about half of the participants.
Table 2 Intermediate outcomes of diabetes care for study participants (N = 295)
Intermediate outcomes* Mean †(95%CI) Proportion ‡(95% CI)
HbA1c level (%) 9.3 (9.0–9.6)
HbA1c < 8% 26% (21%–31%)
Systolic blood pressure (mmHg) 130 (127–133)
Diastolic blood pressure (mmHg) 79 (78–81)
Blood pressure <140/90 mmHg 54% (48%–59%)
Total cholesterol level (mmol/L) 4.9 (4.7–5.1)
Total cholesterol <5.5 mmol/L 41% (35%–47%)
* The most recent readings in the past 12 months were used.
† Among patients receiving measurements.
‡ Patients not receiving measurements were treated conservatively as having outcomes beyond the cut points.
95% CI was calculated by adjusting for clustering by health centre.
Associations between system ACIC scores and quality of diabetes care
Analysed by the linear regression modelling, each ACIC component was statistically associated with overall adherence to delivery of diabetes services (Table 3). After adjustment for other system components and individual variables, organisational influence, community linkages, and information system were identified as having independent associations with adherence to delivery of diabetes services. For example, an increase of 1 unit in the information system ACIC score was associated with an improvement of 4.5% in overall adherence to delivery of services.
Table 3 Association of health centre system components with overall adherence to delivery of scheduled services§
Variables Unadjusted Coefficients 95% CI Adjusted Coefficients* 95% CI
Individual level variables
Age -0.06 -0.23, 0.11 -0.11 -0.29, 0.07
Sex† 5.10 0.49, 9.70 5.63 -0.07, 11.33
ACIC Components
Organisational influence 6.27 4.10, 8.44 4.30 0.92, 7.69
Community linkages 4.60 2.41, 6.79 3.83 1.89, 5.76
Self-management 2.58 0.25, 4.91 -2.30 -5.39, 0.79
Decision support 2.75 0.68, 4.83 -2.30 -5.09, 0.49
Delivery system 2.30 0.47, 4.14 -0.48 -3.96, 2.99
Information system 3.87 2.30, 5.44 4.52 0.70, 8.34
Integration‡ 2.94 0.76, 5.11
§ Estimated using multiple linear regression models.
* Adjusted for other variables in the table (except variable integration) and for clustering by health centre. The intercept for the linear regression is -7.25.
† Males are referent.
‡ Integration was excluded from adjusted analysis as it was correlated with other ACIC components and caused colinearity in multiple regression models.
Coefficients significant at 0.05 level are shown in bold.
The likelihood of each of HbA1c, blood pressure, and total cholesterol being below the specified cut point rose significantly with an increase in the organisational influence score (Table 4). Higher delivery system design and information system scores were associated with better blood pressure control and total cholesterol control respectively. Organisational influence, delivery system design and information system scores were also significantly associated with higher combined intermediate outcome scores.
Table 4 Association between scores for health centre system components and measures of intermediate outcomes of diabetes care
Variables HbA1c control (Y1) Blood pressure control (Y2) Total Cholesterol control (Y3) Effect of independent variables on joint intermediate outcomes P value ‡
Adjusted Odds Ratios (95%CI) *
Individual level variables
Age 1.02 (0.99,1.03) 0.99 (0.98,1.01) 1.00 (0.99,1.02) 0.234
Sex† 1.24 (0.90,1.71) 1.27 (0.86,1.88) 1.19 (0.86,1.64) 0.068
ACIC Components
Organisational influence 1.47 (1.23,1.76) 1.66 (1.21,2.28) 1.22 (1.03,1.45) 0.000
Community linkages 1.03 (0.91,1.17) 0.97 (0.76,1.22) 1.14 (0.96,1.35) 0.108
Self-management 0.97 (0.83,1.13) 0.85 (0.66,1.11) 0.89 (0.73,1.09) 0.053
Decision support 0.89 (0.78,1.02) 1.10 (0.88,1.38) 0.85 (0.72,1.01) 0.064
Delivery system 1.05 (0.89,1.23) 1.42 (1.06,1.90) 0.88 (0.74,1.06) 0.000
Information system 1.05 (0.89,1.23) 0.75 (0.54,1.03) 1.27 (1.04,1.54) 0.004
Y1, Y2, and Y3 are dichotomous variables (0,1), and having a value of 1 represents HbA1c < 8.0%, blood pressure < 140/90 mmHg, and total cholesterol < 5.5 mmol/L, respectively.
* Adjusted for other variables in the table and for clustering by health centre.
† Males are referent.
‡ Estimated using multivariate probit model. P values below 0.05/8 = 0.0063 (8 because there are 8 independent variables) are declared significant at 5% level.
Odds ratios significant at 0.05 level are shown in bold.
Discussion
This study shows that participating Aboriginal community health centres in Australia had implemented basic systems to support chronic illness care, but there was considerable room for improvement in all system components. Stronger organisational influence and information system components were associated both with better performance in process of diabetes care and in intermediate outcomes. Additionally, community linkages were specifically related to better performance in process of care, and delivery system design was associated with better intermediate outcomes.
When compared with data from the USA [16], health centres had similar ACIC scores for community linkages, decision support, delivery system, and clinical information system, but lower scores for organisational influence and self-management support. In many respects the quality of diabetes care in participating health centres is also comparable with national and international data [17-19] For example, 70% of patients in our study had HbA1c tested in the past year and 26% had values less than 8% – almost identical to experience reported from the USA [19].
To our knowledge, this is the first study to demonstrate quantitative evidence regarding the importance of organisational influence (including goals for chronic care, improvement strategies, and incentives for care) on diabetes care. Wagner and colleagues reported their experience in the chronic condition Breakthrough Series, suggesting the removal of disincentives in practice encourages providers' in delivery of effective chronic illness care [7]. Financial incentives for diabetes care have been introduced to Australian general practice through Enhanced Primary Care (EPC) and Practice Incentive Program (PIP) [20,21]. However, only half of participating centres reported claiming for Medicare rebate using EPC items.
Our study shows better implementation of clinical information systems to be associated with both increased adherence to guideline-recommended processes of diabetes care and improved intermediate outcomes. An ideal information system has three important roles: 1) as a registry for a target population; 2) to provide reminders to primary care teams to comply with guidelines for care; and 3) to provide feedback measures relevant to quality of care [22]. Our data show the third role to be the least-developed area for current information systems in this study setting, and support the appropriateness of external clinical audit as a useful approach to address such system deficiencies [23,24]
The positive relation between delivery system design and intermediate outcomes of diabetes care is consistent with several previous studies [25,26] It is likely that health centres characterised by availability of resident doctors, active specialist outreach, and appropriate client follow-up offer more opportunity for intensive management that might contribute to better diabetes control. Given that many remote community health centres are staffed primarily by nurses and Aboriginal Health Workers (AHWs) and supported by visiting doctors [27], a feasible approach to improve delivery system design is to assign and strengthen nurses' and AHWs' roles in delivering routine care, and to ensure referral to medical practitioners for consultation and medication adjustment where appropriate [28]. Features of delivery system design may also be amenable to improving the relative under-utilisation of primary care services by Aboriginal men – a widely recognised phenomenon that is reflected in the study sample.
The apparent poor integration of system components in participating health centres also needs to be addressed in future system development, as isolated upgrading in one component without integrating with another may lead to an increase in costs but not in effectiveness. For example, computerised information systems can generate "pop-up" reminders for healthcare providers, but poor delivery system design characterised by unclear roles among health staff may result in no appropriate action being taken.
The cross-sectional study design limits the confidence with which the observed associations between health centre systems and processes and outcomes of diabetes care can be defined as causal. However, the findings suggest that the Chronic Care Model and the associated ACIC scale will be valuable in assessing and guiding the development of health centre systems in Aboriginal community settings. More research is needed to formally examine the reproducibility of the methods of assessing systems development, and to define the cause-effect relationship between healthcare systems and quality of diabetes care using longitudinal study designs. Such studies may also help clarify resource and management requirements for sustaining improvements in chronic illness care.
Conclusion
The state of development of health centre systems is reflected in quality of care outcome measures for patients. The health centre systems assessment tool proves to be useful in describing the quality of clinical systems for the prevention and management of diabetes in Australian Aboriginal communities, and providing a guide for development of systems for improving diabetes care.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DS participated in developing the study design, performed the statistical analyses, and drafted the manuscript. RB conceived and designed the study, supervised the data collection and analyses, provided a major role in revising the manuscript. CC participated in the study design and assisted in interpretation of findings. MD and AS developed questionnaires and performed the data collection. GR, JC, and TW participated in the development of study design, editing and revising the manuscript. All authors contributed to, have read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
Assessment of Chronic Illness Care (ACIC scale)
Click here for file
Additional File 2
Example: the score and justification for an item in the ACIC scale, which were made by health centre staff consensus
Click here for file
Acknowledgements
This project would not have been possible without the support of staff and management of the participating community health centres. We also gratefully acknowledge the advice of members of the project reference group and management committee. This project was supported by a grant from the Australian Health Minister's Advisory Council (AHMAC) through the States and Commonwealth Research Issues Forum (SCRIF), Grant number AHMAC PDR 2001/06, and by a grant from the CRC for Aboriginal Health.
==== Refs
National Centre for Monitoring Diabetes Diabetes: Australian Facts 2002 Canberra: AIHW
McDermott R Tulip F Schmidt B Sinha A Sustaining better diabetes care in remote indigenous Australian communities BMJ 2003 327 428 430 12933731 10.1136/bmj.327.7412.428
Simmons D Impact of an integrated approach to diabetes care at the Rumbalara Aboriginal Health Service Intern Med J 2003 33 581 585 14656232 10.1111/j.1445-5994.2003.00491.x
Bailie RS Si D Robinson GW Togni SJ D'Abbs PH A multifaceted health-service intervention in remote Aboriginal communities: 3-year follow-up of the impact on diabetes care Med J Aust 2004 181 195 200 15310253
Wagner EH Chronic disease management: what will it take to improve care for chronic illness? Eff Clin Pract 1998 1 2 4 10345255
Bodenheimer T Interventions to improve chronic illness care: evaluating their effectiveness Dis Manag 2003 6 63 71 14577900 10.1089/109350703321908441
Wagner EH Austin BT Davis C Hindmarsh M Schaefer J Bonomi A Improving chronic illness care: translating evidence into action Health Aff (Millwood) 2001 20 64 78 11816692 10.1377/hlthaff.20.6.64
Weeramanthri T Hendy S Connors C Ashbridge D Rae C Dunn M Fittock M Cleary J O'Donohoe L Morton S Swanson N The Northern Territory preventable chronic disease strategy – promoting an integrated and life course approach to chronic disease in Australia Aust Health Rev 2003 26 31 42 15368818
Hoffmann B Bailie R Health-related housing and infrastructure in the Northern Territory Aboriginal Communities: 2001 data 2003 Darwin: Cooperative Research Centre for Aboriginal and Tropical Health
Improving Chronic Illness Care Assessment of Chronic Illness Care (version 35)
Bonomi AE Wagner EH Glasgow RE VonKorff M Assessment of chronic illness care (ACIC): a practical tool to measure quality improvement Health Serv Res 2002 37 791 820 12132606 10.1111/1475-6773.00049
Central Australian Rural Practitioner Association CARPA standard treatment manual: a clinic manual for primary health care practitioners in remote and rural communities in Central and Northern Australia 2003 4 Alice Springs: Central Australian Rural Practitioner Association
Williams R Herman W Kinmonth AL Wareham NJ The evidence base for diabetes care 2002 Chichester: John Wiley & Sons, Ltd
Cappellari L Jenkins S Multivariate probit regression using simulated maximum likelihood Stata Journal 2003 3 278 294
StataCorp Stata Statistical Software: release 82 2003 College Station, TX: Stata Corporation
Wagner EH Glasgow RE Davis C Bonomi AE Provost L McCulloch D Carver P Sixta C Quality improvement in chronic illness care: a collaborative approach Jt Comm J Qual Improv 2001 27 63 80 11221012
Georgiou A Burns J Wan Q Flack J Penn D Powell DG Analysis of Division-based diabetes register data (2000–2002) 2005 Sydney: National Divisions Diabetes Program, University of New South Wales
Tapp RJ Zimmet PZ Harper CA de Court Balkau B McCarty DJ Taylor HR Welborn TA Shaw JE Diabetes care in an Australian population: frequency of screening examinations for eye and foot complications of diabetes Diabetes Care 2004 27 688 693 14988286
Chin MH Auerbach SB Cook S Harrison JF Koppert J Jin L Thiel F Karrison TG Harrand AG Schaefer CT Takashima HT Egbert N Chiu SC McNabb WL Quality of diabetes care in community health centers Am J Public Health 2000 90 431 434 10705866
Parsons J Diabetes. Can systems change improve outcomes? Aust Fam Physician 2001 30 1116 11838386
Harris M Blakeman T Enhanced Primary Care items. Their use in diabetes management Aust Fam Physician 2001 30 1134 1140 11838391
Bodenheimer T Wagner EH Grumbach K Improving primary care for patients with chronic illness JAMA 2002 288 1775 1779 12365965 10.1001/jama.288.14.1775
Chaves N Weeramanthri T Mak D Bunn L Lines D Morgan S Girrabul J Allen O Diabetes audit can aid practice development in a range of Indigenous health care settings Aust J Rural Health 2001 9 251 253 11736850 10.1046/j.1440-1584.2001.00397.x
Fitzsimons B Wilton L Lamont T McCulloch L Boyce J The Audit Commission review of diabetes services in England and Wales, 1998–2001 Diabet Med 2002 19 73 78 12121342 10.1046/j.1464-5491.19.s4.13.x
Feifer C Ornstein SM Nietert PJ Jenkins RG System supports for chronic illness care and their relationship to clinical outcomes Top Health Inf Manage 2001 22 65 72 11761794
Sperl-Hillen JM Solberg LI Hroscikoski MC Crain AL Engebretson KI O'Connor PJ Do all components of the chronic care model contribute equally to quality improvement? Jt Comm J Qual Saf 2004 30 303 309 15208979
Elliott R Regional statistics, Northern Territory 2004 2004 Canberra: ABS
McDermott RA Schmidt BA Sinha A Mills P Improving diabetes care in the primary healthcare setting: a randomised cluster trial in remote Indigenous communities Med J Aust 2001 174 497 502 11419768
|
16117836
|
PMC1208882
|
CC BY
|
2021-01-04 16:31:51
|
no
|
BMC Health Serv Res. 2005 Aug 24; 5:56
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-56
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-581613140110.1186/1472-6963-5-58Research ArticleThe cost-effectiveness of early noninvasive ventilation for ALS patients Gruis Kirsten L [email protected] Michael E [email protected] Devin L [email protected] Department of Neurology, University of Michigan Health System, Ann Arbor, Michigan, USA2 Department of Health Management and Policy, University of Michigan School of Public Health, Ann Arbor, Michigan, USA2005 30 8 2005 5 58 58 18 4 2005 30 8 2005 Copyright © 2005 Gruis et al; licensee BioMed Central Ltd.2005Gruis et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Optimal timing of noninvasive positive pressure ventilation (NIPPV) initiation in patients with amyotrophic lateral sclerosis (ALS) is unknown, but NIPPV appears to benefit ALS patients who are symptomatic from pulmonary insufficiency. This has prompted research proposals of earlier NIPPV initiation in the ALS disease course in an attempt to further improve ALS patient quality of life and perhaps survival. We therefore used a cost-utility analysis to determine a priori what magnitude of health-related quality of life (HRQL) improvement early NIPPV initiation would need to achieve to be cost-effective in a future clinical trial.
Methods
Using a Markov decision analytic model we calculated the benefit in health-state utility that NIPPV initiated at ALS diagnosis must achieve to be cost-effective. The primary outcome was the percent utility gained through NIPPV in relation to two common willingness-to-pay thresholds: $50,000 and $100,000 per quality-adjusted life year (QALY).
Results
Our results indicate that if NIPPV begun at the time of diagnosis improves ALS patient HRQL as little as 13.5%, it would be a cost-effective treatment. Tolerance of NIPPV (assuming a 20% improvement in HRQL) would only need to exceed 18% in our model for treatment to remain cost-effective using a conservative willingness-to-pay threshold of $50,000 per QALY.
Conclusion
If early use of NIPPV in ALS patients is shown to improve HRQL in future studies, it is likely to be a cost-effective treatment. Clinical trials of NIPPV begun at the time of ALS diagnosis are therefore warranted from a cost-effectiveness standpoint.
==== Body
Background
Respiratory failure is the most common cause of ALS patient death[1]. Prior to respiratory failure, respiratory muscle weakness can be measured by standard pulmonary function tests including forced vital capacity (FVC)[2]. Treatment of ALS patients with noninvasive positive pressure ventilation (NIPPV) when FVC is less than 50% appears to improve ALS patient survival[3,4] and quality of life [5-7]. Improved survival of ALS patients with NIPPV may be explained by a slower rate of pulmonary function decline[4,5].
NIPPV initiated early in the ALS disease course may off-load respiratory muscle work and thereby attenuate the progressive decrease in pulmonary compliance seen in ALS[8]. This treatment may also improve quality of life as ALS patients early in their course may experience non-specific symptoms of fatigue and lethargy, related to subtle respiratory muscle weakness, which goes unrecognized or is attributed to impaired mobility[2]. Whether initiation of NIPPV at diagnosis, when FVC is typically reduced but greater than 50%, slows the rate of pulmonary function decline and improves quality of life and survival remains to be studied in a clinical trial.
Paralleling proposed studies of feasibility and effectiveness of early NIPPV[9] is a need to determine what magnitude of health-related quality of life (HRQL) improvement this proposed treatment needs to achieve to be cost-effective. Quality of life improvement is an essential aspect of ALS treatment as curative treatments are not available[10]. The possible improvement in HRQL from early NIPPV treatment in ALS patients can be analyzed in conjunction with the costs of early NIPPV with a cost-utility analysis by using quality-adjusted life years (QALY) as a measure of effectiveness. Cost-utility analyses traditionally are used to determine which proven therapies are cost-effective by determining a treatment's incremental cost-effectiveness, that is, the cost per QALY gained relative to alternative treatments. In the "traditional" cost-utility analysis, effectiveness has been proven, and an estimate of benefit in health-state utility is already known. As society's willingness-to-pay costs per QALY have been reported,[11] the incremental cost-effectiveness can be compared to this standard to determine whether a newly proposed treatment is cost-effective. We applied this same process to determine a priori, how much benefit early NIPPV treatment in ALS patients would need to provide for this treatment to be cost-effective. We reasoned that should the degree of improvement determined in this analysis seem plausible, future clinical trials testing early NIPPV would be warranted from an economic perspective. If, on the other hand, the analysis showed that an impractical degree of improvement would be necessary for the treatment to be cost-effective, future clinical trials of early NIPPV for ALS would be less worthwhile.
Methods
We calculated the benefit in health-state utility that early NIPPV treatment of ALS patients must achieve to be cost-effective. The primary outcome was the percent utility gained through NIPPV in relation to two common willingness-to-pay thresholds: $50,000 and $100,000 per QALY[11].
Model
A decision tree modeled two alternative strategies: NIPPV starting at the time of diagnosis versus no NIPPV at the time of diagnosis, for a hypothetical cohort of patients with a recent diagnosis of ALS. Eighty percent of ALS patients have some evidence of respiratory muscle weakness at the time of initial diagnosis[12]; while half of patients demonstrate a reduction in FVC to less than 80% (approximately two standard deviations below the normal range) at initial presentation[13]. It was assumed that if early NIPPV is effective in preventing respiratory insufficiency, it should therefore be started at the time of diagnosis.
Patients were allowed to shift through disease states (mild, moderate, severe, terminal, or death) through Markov processes. The probabilities of patients progressing through these disease states over time were obtained from the literature[14]. All patients were modeled to begin in the mild stage, given their recent diagnosis. The Markov models used the average amount of time patients spend in each disease state, the probability of transitioning into a more severe stage of ALS, along with the utility associated with the time spent in each health state, to estimate the clinical and economic disease events over time. Per practice guidelines recommending the initiation of NIPPV based on an FVC < 50%, it was assumed that both groups would be treated with NIPPV when these criteria were met, and thus the analysis modeled only until this point. The time horizon used was 1 year as this is the average time period between diagnosis and meeting the NIPPV treatment criteria[3]. The reference case used a benefit in health-state utility of 20% in the NIPPV group compared with the non-NIPPV group. This is similar to the improvement in patient QOL demonstrated for NIPPV treatment in those who had respiratory muscle weakness, hypoventilation, or sleep-disordered breathing[7]. The improvement in health-state utility associated with NIPPV use was allowed to vary in sensitivity analysis, where one variable is allowed to vary over a plausible range. One-way sensitivity analyses were conducted for each variable across the ranges of values found in Table 1. To account for patients entering the model at varying rates of disease progression, the time horizon was adjusted. The time horizon was varied between 6 months and 2 years, in a one-way sensitivity analysis. As variations in FVC at entry may relate to ALS disease stage at entry, we also conducted a one-way sensitivity analysis on the probability of entering the model in the mild stage, as opposed to the moderate stage. Given that all patients are assumed to have recently been diagnosed with ALS and have an FVC > 50%, it was assumed that no one would enter the model in a severe or terminal state. The decision tree was analyzed by Data 4.0 (TreeAge Inc, Williamstown, MA).
Table 1 Probabilities, utilities, and costs for base case and sensitivity analysis
Variables Reference case Lower value tested Upper value tested
Probabilities:
NIPPV tolerance 0.49[3] 0.40 0.80
Staying mild 0.66[14] 0.60 0.70
Staying moderate 0.77[14] 0.70 0.80
Staying severe 0.76[14] 0.70 0.80
Staying terminal 0.78[14] 0.70 0.80
Utilities:
Mild 0.8[15] 0.7 0.9
Moderate 0.6[15] 0.5 0.7
Severe 0.5[15] 0.4 0.6
Terminal 0.4[15] 0.3 0.5
Costs:
NIPPV and accessories for 1 year $3,132[17] $2,810 $3,306
A trial of NIPPV in those who prove intolerant $467[17] $411 $483
Utilities
Assessing health-state utilities in a patient population allows assignment of a numerical value to patient reports of HRQL or health state "utility" for different stages of disease. Health utilities were determined by assessment of patient's health state at each level of disease by a preference-based method and have been reported previously in control arms of large clinical trials[15]. These measurements were aggregated across individuals to determine utility scores for each health state, ranging from death (0), to perfect health (1). Utilities for each ALS stage, measured by the EuroQol EQ-5D visual analogue scale, were obtained from the literature[15].
Costs
Costs were estimated from the Medicare fee schedule for 2004 (in US dollars) for NIPPV and NIPPV accessories. Medicare reimbursement was selected given the societal perspective[16]. Costs of one month of NIPPV rental and accessory costs were included for those intolerant to NIPPV. No discounting of costs or utilities was needed given the time horizon. Other costs related to ALS patient care were considered equal in both treatment groups given the identical probabilities of transitioning through health states and were therefore not entered into the model.
Results
The average patient receiving NIPPV experienced 0.59 QALYs at a cost of $1,773; a patient not receiving NIPPV experienced 0.54 QALYs at a cost of $0, resulting in an incremental cost-effectiveness ratio of $33,801. Sensitivity analysis performed on the utilities of ALS states demonstrated NIPPV has an incremental cost-effectiveness ratio lower than $50,000 as long as the utility for ALS patients receiving NIPPV is at least 13.5% higher at each stage than those without NIPPV, meaning that early NIPPV is cost-effective as long as the treatment of ALS patients with NIPPV beginning at the time of diagnosis improves HRQL by at least 13.5%. For a willingness-to-pay threshold of $100,000 per QALY, the increase in HRQL with NIPPV would only need to be 6.8% or greater to be cost-effective.
The cost-effectiveness did not exceed the $50,000 willingness-to-pay threshold in any of the cost, transition probability, tolerance, or utility sensitivity analyses, meaning that alterations of each variable across a plausible range (Table 1) did not cause the incremental cost-effectiveness of NIPPV to exceed $50,000 per QALY. Altering the tolerance of NIPPV below 18% would however cause the cost-effectiveness to exceed $50,000 per QALY. No alteration of the probability of entering the model in the mild disease stage caused the incremental cost-effectiveness of NIPPV to exceed the $50,000 per QALY threshold. Shorter time horizons were associated with a lower cost-effectiveness ratio. A time horizon of 6 months was associated with an incremental cost-effectiveness of $76,909, while an 8 month time horizon was associated with incremental cost-effectiveness of $53,001. Time horizons of 10 months or above were associated with an incremental cost-effectiveness less than $50,000.
Discussion
The benefit of early NIPPV use in ALS patients has not yet been studied. However, our cost-effectiveness model suggests that NIPPV begun at the time of diagnosis would be cost-effective if NIPPV were shown to improve HRQL by just 7–14%.
The 7–14% range of HRQL improvement that would be necessary in our model for early NIPPV to be cost-effective may be an overestimate. The $50,000 per QALY threshold for assessing cost-effectiveness is quite conservative. Given more recent estimates of the appropriate cost-effectiveness threshold[11], the improvement that would be necessary for NIPPV to be cost-effective is likely less than 7%. The possibility that NIPPV could slow the transition from less severe to more severe disease states was not taken into account in the model. Should early NIPPV be demonstrated to slow the progression of ALS[9], it would be even more cost-effective than this model suggests.
Tolerability of NIPPV by ALS patients with early disease is unknown. Tolerance of NIPPV (assuming a 20% improvement in HRQL) would only need to exceed 18% in our model for treatment to remain cost-effective using a conservative willingness-to-pay threshold of $50,000 per QALY. This estimate of NIPPV compliance is well lower than that seen in other studies[3]. The base case used a much more conservative estimate of 49% tolerance[3].
The current analysis was limited by the validity of the estimates used in the model. Tolerance of NIPPV administered early in the course of ALS is unknown, but this value was allowed to vary in sensitivity analysis. Utility values were ascertained from estimates in the literature, but previous studies on this topic are limited. The utility values were also allowed to vary in sensitivity analysis. In our model, early NIPPV remained cost-effective in all of the sensitivity analyses, supporting the robustness of the model.
Conclusion
If early use of NIPPV in ALS patients is shown to improve HRQL in future studies, it is likely to be a cost-effective treatment. Further trials of early NIPPV initiation in ALS patients are warranted, and supported from a cost-effectiveness perspective.
Competing interests
The authors declare(s) that they have no competing interests.
Authors' contributions
KLG conceived the study, participated in its design and coordination and drafted the manuscript. MEC participated in the design of the study and helped with the statistical analysis and interpretation. DLB performed the statistical analysis, participated in the design of the study, and helped draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank James W. Albers for his perceptive comments regarding earlier versions of the manuscript.
==== Refs
Kaplan LM Hollander D Respiratory dysfunction in amyotrophic lateral sclerosis Clinics in Chest Medicine 1994 15 675 81 7867282
Perrin C Unterborn JN Ambrosio CD Hill NS Pulmonary complications of chronic neuromuscular diseases and their management Muscle & Nerve 2004 29 5 27 14694494 10.1002/mus.10487
Aboussouan LS Khan SU Banerjee M Arroliga AC Mitsumoto H Objective measures of the efficacy of noninvasive positive-pressure ventilation in amyotrophic lateral sclerosis Muscle & Nerve 2001 24 403 409 11353427 10.1002/1097-4598(200103)24:3<403::AID-MUS1013>3.0.CO;2-3
Kleopa KA Sherman M Neal B Romano GJ Heiman-Patterson T Bipap improves survival and rate of pulmonary function decline in patients with ALS J Neurol Sci 1999 164 82 88 10385053 10.1016/S0022-510X(99)00045-3
Bourke SC Bullock RE Williams TL Shaw PJ Gibson GJ Noninvasive ventilation in ALS: indications and effect on quality of life Neurology 2003 61 171 177 12874394
Jackson CE Rosenfeld J Moore DH Bryan WW Barohn RJ Wrench M A preliminary evaluation of a prospective study of pulmonary function studies and symptoms of hypoventilation in ALS/MND patients J Neurol Sci 2001 191 75 78 11676995 10.1016/S0022-510X(01)00617-7
Lyall RA Donaldson N Fleming T Wood C Newsom-Davis I Polkey MI A prospective study of quality of life in ALS patients treated with noninvasive ventilation Neurology 2001 57 153 156 11445650
Misuri G Lanini B Gigliotti F Iandelli I Pizzi A Bertolini MG Mechanism of CO2 Retention in Patients With Neuromuscular Disease Chest 2000 117 447 453 10669689 10.1378/chest.117.2.447
ALS Association Research Article Accessed 9/15/04
Hardiman O Hickey A O'Donerty LJ Physical decline and quality of life in amyotrophic lateral sclerosis Amyottoph Lateral Scler Other Motor Neuron Disord 2004 5 230 234 15799552
Ubel PA Hirth RA Chernew ME Fendrick AM What is the price of life and why doesn't it increase at the rate of inflation? Arch Intern Med 2003 163 1637 1641 12885677 10.1001/archinte.163.14.1637
Schiffman PL Belsh JM Pulmonary function at diagnosis of amyotrophic lateral sclerosis. Rate of deterioration Chest 1993 103 508 513 8432145
Fallat RJ Jewitt B Bass M Kamm B Norris FH Jr Spirometry in amyotrophic lateral sclerosis Archives of Neurology 1979 36 74 80 420626
Tavakoli M Malek M The cost utility analysis of riluzole for the treatment of amyotrophic lateral sclerosis in the UK J Neurol Sci 2001 191 95 102 11676998 10.1016/S0022-510X(01)00618-9
Kiebert GM Green C Murphy C Mitchell JD O'Brien M Burrell A Patients' health-related quality of life and utilities associated with different stages of amyotrophic lateral sclerosis J Neurol Sci 2001 191 87 93 11676997 10.1016/S0022-510X(01)00616-5
Gold MR Siegel JE Russell LB Weinstein MC Cost-effectiveness in health and medicine 1996 New York: Oxford University Press
Medicare Physician Fee Schedule 2004
|
16131401
|
PMC1208883
|
CC BY
|
2021-01-04 16:31:53
|
no
|
BMC Health Serv Res. 2005 Aug 30; 5:58
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-58
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-591613140510.1186/1472-6963-5-59Research ArticleCaution required when relying on a colleague's advice; a comparison between professional advice and evidence from the literature Schaafsma Frederieke [email protected] Jos [email protected] Carel [email protected] Dijk Frank [email protected] Coronel Institute for Occupational and Environmental Health, Academic Medical Centre, PO Box 22700, 1100 DE Amsterdam, The Netherlands2 Cochrane Collaboration Occupational Health Field, Finnish Institute of Occupational Health, PO Box 93, 70701 Kuopio, Finland2005 31 8 2005 5 59 59 1 2 2005 31 8 2005 Copyright © 2005 Schaafsma et al; licensee BioMed Central Ltd.2005Schaafsma et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Occupational Physicians rely especially on advice from colleagues when answering their information demands. On the other hand, Evidence-based Medicine (EBM) promotes the use of up-to-date research literature instead of experts. To find out if there was a difference between expert-based practice and EBM we compared professional advice on occupational health topics with best evidence from the literature.
Methods
We asked 14 occupational physicians to consult their usual information sources on 12 pre-conceived occupational health problems. The problems were presented in the form of case vignettes which contained sufficient clinical information to be used by the occupational physicians for the consultation of their experts. We had searched the literature for the best available evidence on the 12 problems, which made it possible to answer the clinical questions with a clear yes or no.
Results
The cases could be used by the occupational physicians as arising from their own practice. All together the occupational physicians consulted 75 different experts. Almost half of the consulted experts were near colleagues, 10% were industrial hygienists, 8% medical specialists and the rest had a varied background. Fifty three percent (95% confidence interval 42% to 65%) of all professional advice was not in line with the research literature. In 18 cases (24%) professional advice explicitly referred to up-to-date research literature as their used source. These cases were substantially less incorrect (17%) than advice that had not mentioned the literature as a source (65%) (difference 48%, 95% Confidence Interval from 27% to 69%).
Conclusion
Advice that occupational physicians routinely get in their daily practice differs substantially from best evidence from the literature. Occupational physicians who ask professional advice should always ask about the evidence of this advice.
==== Body
Background
Occupational physicians (OPs) in their daily routine are confronted with a large variety of occupational health problems. From previous research we know that in attending these problems OPs mostly rely on their own experience and on information from consulting an experienced colleague [1]. On the contrary, Evidence-Based Medicine proposes to use evidence from the up-to-date research literature as most reliable source. Reasons for OPs to still prefer working experience- or authority-based are the relatively easy way to obtain and the attributed validity of the information. Evidence-based medicine, although much-supported, is still not a customary way for occupational physicians (OPs) to address problems that arise in their daily work [2]. OPs like other physicians do not quite see its benefits.
Relying on your own or on others' expertise knows some drawbacks. For example, Slawson et al described that the information can be out of date and that there could be the matter of reverse gullibility [3-5]. In this study we want to challenge the belief of OPs that asking for professional advice from a colleague, even if this colleague is considered an expert on the subject, is a good source for information. We will compare professional advice given by experts to answers based on best evidence derived from the literature.
Methods
We asked a convenience sample of fourteen acquainted OPs working scattered over four different regions of the Netherlands to collect data for us. Our main criteria to ask a physician to participate were that he or she had to be professionally sufficiently experienced. Next, we took care that there was variation in location to avoid the situation that the same professional expert would be asked about the same case vignette by different OPs. Even though we tried to vary age, gender and professional experience, the majority was over 40, male and had a long standing professional experience and three OPs had achieved a doctor's degree. (Table 1) All OPs were considered experienced and professionally motivated, and agreed to participate. The OPs were requested to obtain two professional advices on each of three case vignettes which would lead to a maximum of 84 cases. To be able to show that a relevant 15% of the answers would not be in line with the literature with α = 0.05 and β = 80% we would need about 53 cases. A professional advice was defined as an advice from a person who was considered by the OP to be an expert on the subject and who would also be consulted in the normal course of daily routine.
Table 1 Personal characteristics of occupational physicians (N = 14) involved in the study
N (%)
Age (> 40 years) 10 (71)
Gender (male) 12 (86)
Geographical location
North 4 (29)
South 3 (21)
West 4 (29)
East 3 (21)
Certified occupational physician 12 (86)
Professional Experience (> 10 years) 12 (86)
Occupational Health Service
Arbounie 7 (50)
Other 7 (50)
Academic Status (PhD) 3 (21)
Twelve cases were selected on the basis of a clear occupational health problem, resemblance to daily practice for an OP and assumption that there would be sufficient literature (Table 2, See Additional file 1). The cases represent a broad variety of occupational health practice ranging from return to work interventions in workers with musculoskeletal disorders to the causality of stress in case of a myocardial infarction. The case vignettes ended in a clear clinical question that could be answered by a simple yes or no. For example, 'does continuous years of work stress increase the risk of a myocardial infarction?' and 'is it useful to take melatonin to prevent jetlag?'
Table 2 Summary of the case vignettes and correct evidence-based answer
1. For a 36-year old caretaker at a secondary school with a lateral ankle ligament rupture treated with tape for three weeks, is it safe to resume work? Yes
2. Can a rash on the inside of the forearm of a 43-year old production worker be caused by exposure to PVC during the production of bathroom doors? Yes
3. Can continuous years of work stress be the cause of a cardiac infarct in a 54-year-old bank employee with only a slightly raised cholesterol level? Yes
4. For a 38-year old laboratory worker with epicondylitis lateralis, does electro shock wave therapy (ESWT) produce better results in reducing complaints than conventional treatment with physiotherapy and analgesics? No
5. Is a 38-year old sewage worker subject to a higher risk of contracting Hepatitis A as a result of occupational exposure? No
6. For a 48-year old archivist with extrinsic allergic alveolitis, is it useful to investigate the archive more closely for fungal cultures as a possible cause for the lung disease? Yes
7. Is it safe for a 42-year old parking attendant suffering from a whiplash as a result of a car accident to return partially to work after some 10 days? Yes
8. Is Cognitive Behaviour Treatment more effective than other therapies for a 45-year old teacher diagnosed with burnout? Yes
9. Is it effectively useful to take melatonin to prevent jetlag for workers of an ICT firm travelling to Asia? Yes
10. For a 45-year old female teacher diagnosed with mild depression, is St. John's Wort more effective than placebo? Yes
11. Does a return to his physically demanding work after an operation on a lumbal hernia nuclei pulposi in a 45-year-old carpenter, six weeks after the operation, give a higher risk of a recurrence than returning to only light work? No
12. Can a 42-year old male nurse, working on the ambulances safely return to full time work three weeks after his inguinal hernia operation? Yes
The OPs were asked to draw their own conclusion on the case vignettes and to provide the professional advice of all the experts that were consulted. The OP could decide for himself whether or not to rely on the advice received. All cases had to be advised on by the experts with yes or no accompanied by a motivation for the answer. The experts were kept unaware by the consulting OP that the cases presented were fictive.
These professional advices were compared to evidence from the literature in the form of a critically appraised topic (CAT). Critically appraised topics are considered as the best way to retrieve an answer to a question arising from practice from the literature. We followed the guidelines for making critically appraised topics as formulated by Sacket et al.[6] We used Medline, the Cochrane Library and the Dutch clinical guideline database (CBO) to search for relevant evidence to the clinical questions. We used the best available evidence that we could find on a certain topic. In three cases we could use a Cochrane systematic review, in four cases we could use a systematic review and in 5 cases we relied on original studies as the best evidence because no systematic review was available. We felt that for none of the cases the evidence was novel or surprising, but that the available recent literature all pointed in the same direction. All CATs are described in the appendix together with the search strategy and the evidence that was used to answer the clinical question. [See Additional file 1]
A professional advice was considered correct if both the 'yes or no answer' and the motivation were in line with the evidence from the literature as summarised in the CAT. The conclusions of the OPs were assessed only by their 'yes or no answer'.
The first two authors (FS and JV) checked and evaluated both the professional advices and the answers from the OPs separately. We measured the proportion of advices and answers that were not correct.
Results
The occupational physicians consulted 84 different experts of which 75 answered (89% response; 75 out of 84). This resulted in 39 answers to the case vignettes from the 14 participating OPs (93% response; 39 out of 42) on the 12 cases. All cases were perceived as being from daily practice by both the OPs and the consulted experts. Each individual case was advised on at least five times by an expert, except for one case where we had only two advices from experts. Table 3 shows the profession of the consulted experts which are comparable to the type of experts occupational physicians usually consult in daily practice [1]. Most experts were consulted via e-mail (37.3%), by telephone (28.0%) or directly (13.3%). Of the 75 professional advices, 28 (37 %, 95% Confidence Interval from 26% to 48%) were incorrect. If we also took the motivation related to the answers in consideration, 40 answers were incorrect (53%, 95% Confidence Interval from 42% to 65%). Of the 39 conclusions of the OPs, based on the experts' advice 17 (44%, 95% Confidence Interval from 28% to 59%) were incorrect. There was no difference in the rate of incorrect advice per type of profession per consulted expert or per case vignette.
Table 3 Frequency of consulted colleagues
Profession of the consulted colleague Number of consultations N (%)
Occupational Physician 34 (45.3)
Occupational Hygienist 8 (10.7)
Medical Specialist from a local hospital 6 (8.0)
Physiotherapist 6 (8.0)
Professional at a specialized occupational health centre or clinic 6 (8.0)
Psychologist 4 (5.3)
Other 11 (14.7)
Total 75 (100.0)
The motivations of the experts for their advices were based 18 times (24%) on the literature. The rate of incorrect advices by experts was 17% if their advices were explicitly based on the up-to-date research literature versus 65% incorrect if these advices were not based on the literature (difference 48%, 95% Confidence Interval from 27% to 69%).
Discussion
This is a first empirical study about the difference between research literature and the knowledge of professionals within occupational health. The results substantiate the claim by previous authors that physicians should be more aware of the limited value of the information obtained from experts [3]. Less than half of the given professional advice by experts to a practical occupational health problem was in line with evidence from the research literature.
The strength of our study is that we used the information retrieval process such as it occurs in real daily practice of occupational physicians. From our previous study, we know that the information sources that occupational physicians used in this study do conform to the sources they use in general. About half of them ask a colleague, 20% ask other professionals in the occupational health area and another 20% consults medical specialists or other clinical experts. None of the participants in the study commented on the nature of the cases or the questions asked. They were all perceived as relevant and important for clinical occupational health practice. The occupational physicians were situated in different parts of the country and had similar training as occupational physicians in general. There was an overrepresentation of physicians with a doctor's degree in our sample. This might have positively influenced the results in a way that more academic professionals could have been consulted. In turn, we assume this would have resulted in answers more in line with the literature. However, we did not find indications for such a mechanism. The power of the study was sufficient to show at least a 15% deviance from evidence from the literature. Therefore, we feel that there is no reason to believe that the practice of professional advice studied here is different for other OPs or even in other medical disciplines as argued by various authors. [3-5]
Answers to clinical questions arising from practice should not only depend on the available evidence but also on the clinical situation, the patient's preferences and the resources available. The selected case vignettes all required dichotomous answers from the experts and OPs. This obviously distorts to some extent the clinical reality. However, the decision making was rather obvious in all cases with a clear patient preference, and the cases were perceived as being from daily practice even by the experts who were unaware of the fictive nature. As to the resources available, we considered leaving this open for the consulted expert to resemble daily practice most.
The evidence we used to answer to the cases is a selection following the guidance given by the experts.[6] For most cases we found good systematic reviews which can be considered as high quality evidence. However, in some we had to rely on single original studies that were not always evaluation studies. This leaves some room for discussion about the credibility of the evidence. However, none of the results of the studies used as evidence were really novel or surprising but all were in line with general trends in the literature such as the approach to musculoskeletal diseases or advice about return to work. Moreover, the results were not related to the type of case and therefore not to the quality of the evidence provided.
Conclusion
Our findings urge for more and better research into professional knowledge management. For now we conclude that better use of the available research literature is possible and should be stimulated among occupational physicians. If professionals considered an expert on the subject, are asked for advice, occupational physicians should still make sure that the expert also provides the evidence for his advice.
Competing interests
The author(s) declare that they have no competing interests. The views expressed in this article represent those of the authors and are not necessarily the views of the official policy of The Cochrane Collaboration.
Authors' contributions
JV designed the project plan. FS and JV carried out the project. FS wrote the first draft of the manuscript. All authors commented equally on the project plan and all drafts of the manuscript and read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
We give the 12 Critically Appraised Topics that we used as the literature standard to compare the experts' advice with as well as the search strategy and literature references on which the CATs conclusions are based.
Click here for file
==== Refs
Schaafsma FG Hulshof CT van Dijk FJ Verbeek JH Information demands of Occupational Physicians and their attitude towards Evidence-Based Medicine Scand J Work Environ Health 2004 30 327 330 15458017
Verbeek JH van Dijk FJ Malmivaara A Hulshof CT Rasanen K Kankaanpaa EE Mukala K Evidence-based medicine for occupational health Scand J Work Environ Health 2002 28 197 204 12109560
Slawson DC Shaughnessy AF Obtaining useful information from expert-based sources BMJ 1997 314 947 9 9099121
Antman EM Lau J Kupelnick B Mosteller F Chalmers TC A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction JAMA 1992 268 240 8 1535110 10.1001/jama.268.2.240
Riffenburgh RH Reverse Gullibility and Scientific Evidence Arch Otolarygol Head Neck Surg 1996 122 600 601
Sackett DL Straus SE Richardson WS Rosenberg W Haynes RB Evidence-based medicine. How to practice and teach EBM 2000 New York: Churchill Livingston
|
16131405
|
PMC1208884
|
CC BY
|
2021-01-04 16:31:50
|
no
|
BMC Health Serv Res. 2005 Aug 31; 5:59
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-59
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-601613733610.1186/1472-6963-5-60Research ArticleHousehold out-of-pocket medical expenditures and national health insurance in Taiwan: income and regional inequality Chu Tu-Bin [email protected] Tsai-Ching [email protected] Chin-Shyan [email protected] Yi-Wen [email protected] Wen-Ta [email protected] Taipei Municipal Wan Fang Hospital, Taipei, Taiwan2 Department of Public Finance, National Taipei University, Taipei, Taiwan3 Department of Economics, National Taipei University, Taipei, Taiwan4 Division of Health Plicy Research, National Health Research Institues, Taipei, Taiwan2005 2 9 2005 5 60 60 4 1 2005 2 9 2005 Copyright © 2005 Chu et al; licensee BioMed Central Ltd.2005Chu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Unequal geographical distribution of medical care resources and insufficient healthcare coverage have been two long-standing problems with Taiwan's public health system. The implementation of National Health Insurance (NHI) attempted to mitigate the inequality in health care use. This study examines the degree to which Taiwan's National Health Insurance (NHI) has reduced out-of-pocket medical expenditures in households in different regions and varying levels of income.
Methods
Data used in this study were drawn from the 1994 and 1996 Surveys of Family Income and Expenditure. We pooled the data from 1994 and 1996 and included a year dummy variable (NHI), equal to 1 if the household data came from 1996 in order to assess the impact of NHI on household out-of-pocket medical care expenditures shortly after its implementation in 1995.
Results
An individual who was older, female, married, unemployed, better educated, richer, head of a larger family household, or living in the central and eastern areas was more likely to have greater household out-of-pocket medical expenditures. NHI was found to have effectively reduced household out-of-pocket medical expenditures by 23.08%, particularly for more affluent households. With the implementation of NHI, lower and middle income quintiles had smaller decreases in out-of-pocket medical expenditure. NHI was also found to have reduced household out-of-pocket medical expenditures more for households in eastern Taiwan.
Conclusion
Although NHI was established to create free medical care for all, further effort is needed to reduce the medical costs for certain disadvantaged groups, particularly the poor and aborigines, if equality is to be achieved.
==== Body
Background
In 2000, Taiwan was ranked as the second healthiest country in the world, only behind Sweden [1]. While its economic growth, environmental sanitation, and health education have been contributing factors, its implementation of National Health Insurance (NHI) in 1995 is the most important key to its achievements in health. With the comprehensive medical benefit coverage provided to almost all people, NHI has notably increased the availability and accessibility of medical care services and has seen a significant increase in utilization of its medical healthcare services [2-5]. Starting with a national approval rate of 33% in 1995, NHI's approval rating rose to 75.4% in 2000 and to over 80% in 2001 [6].
Although many studies examining whether NHI mitigates the inequality in health care utilization have been conducted and have provided solid evidence that NHI does lessen the socio-demographic and regional disparity, few studies have focused on testing how equitable the financing scheme of Taiwan health care system is. Neither have the changes in relative out-of-pocket expenditures during the pre- and post-NHI period been compared. Since a national insurance program should spread risks over large population groups and reduce medical costs for all patients, particularly the poor, it is important to understand how the health care reform (NHI) reduces and to what extent it reduces out-of-pocket expenditures for medical care. The financial burden from out-of-pocket medical expenditures is regressive financing in developed countries such as USA [7] or in developing countries such as Thailand [8], where financial burden is heavier for the poor. That is, the poor pay a larger share of their income for out-of-pocket medical expenditures than the rich in most countries; only a few, including Colombia and Vietnam, have implemented healthcare reforms capable of remedying regressivity and reducing medical costs for the poor [9,10]. No studies, however, have assessed the impact of Taiwan's NHI on out-of-pocket medical expenditures.
Yeh [11]examined the distribution of out-of-pocket medical expenditures among households under Taiwan's NHI and, like most international studies, found regressivity in out-of-pocket financing scheme of NHI, but that study did not evaluate how or if regressivity especially evolved from implementing NHI. One comparative study of pre- and post-NHI household out-of-pocket medical expenditures [12] reported that NHI reduced the gap between the richer and the poor. That study found household out-of-pocket expenditures of those household with incomes in the highest quintile was 2.64 times that of those in the lowest quintile before NHI was implemented, and that this figure was reduced to 2.30 after NHI was implemented. That study, however, did not delineate the differences in out-of-pocket medical expenditures incurred by different income level of households under NHI. In an effort to resolve these unanswered research questions, we examined the direct and indirect impact of NHI on household out-of-pocket medical expenditures in Taiwan by pooling the 1994 and 1996 cohorts of survey data. Our analysis particularly examines the impact of NHI on reducing financial burdens and regional differences in medical care expenditures.
Public health system in Taiwan
Unequal geographical distribution of medical care resources and insufficient healthcare coverage has been two long-standing problems with Taiwan's public health system.
Regional inequality
Two-thirds of Taiwan's terrain is mountainous, and many of these mountainous areas are isolated and inaccessible. Since most of the high mountains are located in the central eastern regions, the eastern coast area is the least developed and populated area in Taiwan. This area, therefore, is lacking in medical manpower and facilities. Taiwan's northern area, however, is its most developed region. It includes Taiwan's largest city, the capital, Taipei. Most medical care facilities and medical personnel are concentrated in such urban areas. This disparity in accessibility to health care services has raised many concerns and complaints from the public.
In 1985, the Medical Care Network project, established to balance the distribution of medical care resources in various regions and to allow medical manpower and facilities to grow at a reasonable rate, divided Taiwan into 17 medical care regions and aimed to develop medical manpower and facilities. The project had three chronological phases: July 1985 – June 1990, July 1990 – December 1996 and January 1997 – December 2000. The second phase was specifically designed to even out differences in the regional distribution of medical care resources in preparation for the implementation of NHI in 1995.
To reach this objective, the Medical Care Act prohibited establishment or expansion of hospitals in areas with excess medical care resources and encouraged their development in areas with fewer resources. A Medical Care Development Fund was established to provide the private sector with financial incentives (subsidizing the interest on loans) to construct medical care institutions in areas with fewer medical resources [13].
Insufficient health insurance
In the 1980's, around 40% of the population had no health insurance. The uninsured poor faced financial barriers to health care services and their condition of their health was negatively impacted. Industrialization, urbanization and the aging of population, however, increased people's demands for better health and medical care services. To meet that demand and guarantee all people basic health care use, Taiwan began the arduous task of implementing a national health insurance program [6].
Taiwan had three major social insurance programs before NHI was implemented: Labor Insurance, Government Employees' Insurance and Farmers' Insurance. These programs covered employed workers, who accounted for about 58% of the population. The remaining 42%, approximately 9 million children, elderly people, and non-working adults, had no insurance. The availability and accessibility of medical care services for those formerly insured were not high because services were limited to certain contracted medical care institutions. Those formerly insured by Labor Insurance and Farmers' Insurance received medical care services from contracted medical institutions, which accounted for 80% and 47% of hospitals and clinics, respectively[14,15]. Those formerly insured by Government Employee's Insurance had even fewer choices, with only 61% of the hospitals and less than 6% of clinics contracted by to government to provide medical care for them [16].
NHI integrated the former three social insurance programs into one comprehensive program would also include those previously uninsured. All citizens became eligible for NHI, including legally employed foreigners. The enrollment rate during its inception period in 1995 was 92%, and that figure had increased to 96.16% by 2000. Once NHI was implemented, all beneficiaries were also required to receive health care at contracted medical care facilities, but the availability of such facilities increase significantly. By 1998, they constituted about 93.68% of overall medical facilities in Taiwan [6]
With regard to medical benefit coverage, the three insurance schemes in place prior to NHI only covered some a specified types of medical care services. For example, Labor Insurance and Farmers' Insurance covered three items: ambulatory care, inpatient care, and medical care for child delivery. NHI covered inpatient care, ambulatory care, laboratory and X-ray examinations, prescription drugs and certain OTC drugs, dental services, traditional Chinese medicine, day care for mental illness, limited home care and certain preventive services.
Because the previous insurance schemes required no co-payment, moral hazard is widely believed to occur. To prevent these practices, NHI requests patients to make co-payments for outpatient and inpatient care, dental care, emergency care or Chinese medicine services and pharmaceuticals. Co-payment, however, is not required in certain situations. For example, if a beneficiary suffers a major illness or injury and requires long-term, expensive treatment, he/she is exempted from any co-payment obligation under Article 36 of the National Health Insurance Act. Co-payment exemptions have also been established for childbirth and preventive health services, low-income households, veterans and their dependents and people residing in mountainous areas or on offshore islands.
To prevent the public from incurring catastrophic expenses, NHI has established co-payment ceilings. The co-payment ceiling for expenses incurred within 30 days in an acute ward or within 180 days in a chronic ward was calculated as 6% of the average national income, which came to NT$23,000 per admission in 2001. To further reduce financial burden of patients, the cap on cumulative costs for the calendar year was also set and calculated as 10% of the average national income, which came to NT$39,000 in 2001 [6].
Methods
Data
Data used in this study were drawn from the 1994 and 1996 Surveys of Family Income and Expenditure (SFIE) [17,18]. These surveys were conducted by the Directorate-General of the Budget, Accounting and Statistics, Executive Yuan, Republic of China and contained observations for 14,011 and 13,484 households in the pre- and post-NHI period, respectively. The SFIE includes two survey procedures: interviews and account-keeping collection methods. The sampled households, chosen from the population using stratified random sampling, were interviewed once each year to determine their major sources of income and expenditures. Then, to obtain detailed categories of income and expenditure, some of interviewed households were asked to also track daily income and expenditure activity. Because the SFIEs were done by interviews and account-keeping collection methods, they are more accurate than those derived solely from interviews and they have been widely used by domestic and international researchers [19,20].
The data contains detailed information on a series of personal and family characteristics, including age, sex, marital status, highest education level attained, employment status, consumption of health care, individual and household out-of-pocket expenditures, and geographic locations, etc.
Analytical method
The pre-NHI 1994 survey and the post-NHI 1996 survey enabled us to assess the impact of NHI on household out-of-pocket medical care expenditures shortly after its implementation in 1995. We pooled the data from 1994 and 1996 and included a year dummy variable (NHI), equal to 1 if the household data came from 1996. In addition, the estimated model contains two groups of interaction effects, Incomei/NHIi and Regioni/NHIi, to examine whether NHI reduced financial barriers and regional differences in household out-of pocket medical expenditures.
The empirical model is specified below:
Ln(Expi) = β0 + β1Incomei + β2Regioni + β3NHIi + β4Incomei/NHIi + β5Regioni/NHIi + β6Zi + εi
Expi is out-of-pocket expenditures on health care by household i. For linearity in the equation, the dependent variable (Exp) is transferred into logarithm. Incomei is a set of household income dummy variables. Regioni is a set of regional dummy variables. NHIi indicates the national health insurance program. Incomei/NHIi and Regioni/NHIi are two groups of interactions of income and region and NHI. Zi represents a vector of socioeconomic and demographic characteristics of economic household head, which are martial status, sex, employed status, age, education level and family members. The detailed definitions of variables are listed in Table 1.
Table 1 Definitions of Variables
Variable Definition
Dependent Variable
Expenditures Continuous variable: total household annual out-of-pocket medical expenditures
Independent Variable
Demographic variables (Economic Household Head, EHH)
Married Dummy variable = 1 if EHH is currently married, otherwise = 0 including never married, widowed, separated and divorced.
Male Dummy variable = 1 if EHH is male.
Employed Dummy variable = 1 if EHH is employed.
Age
25–44 Dummy variable = 1 if EHH's age in this range.
45–64 Dummy variable = 1 if EHH's age in this range.
65 and above Dummy variable = 1 if EHH's age in this range. (EHH's age under 25 is the default variable.)
Education
Junior high Dummy variable = 1 if EHH finished Junior high school.
Senior (vocational) high Dummy variable = 1 if EHH finished senior (or vocational) high school.
Junior college Dummy variable = 1 if EHH finished Junior college.
College and above Dummy variable = 1 if EHH finished college and above. (Primary school and below is the default variable.)
Family members Dummy variable = 1 if the number of household members ≥ 4.
NHI Dummy variable = 1 if EHH is from the post-national health insurance cohort.
Household income
Lower 20%~2nd quintile Dummy variable = 1 if family income is lower 20%.
Middle 20%~3rd quintile Dummy variable = 1 if family income is middle 20%.
Higher 20%~4th quintile Dummy variable = 1 if family income is higher 20%.
Highest 20%~5th quintile Dummy variable = 1 if family income is highest 20%. (Lowest 20%~1st quintile is the default variable)
Regional variables
Center Dummy variable = 1 if the household is located in: Taichung Hsien, Changhwa Hsien, Nantou Hsien, Yunlin Hsien, Taichung City.
South Dummy variable = 1 if the household is located in: Chiayi Hsien, Tainan Hsien, Kaohsiung Hsien, Pingtung Hsien, Kaohsiung Municipality, Chiayi City, Tainan City.
East Dummy variable = 1 if the household is located in: Taitung Hsien, Hwalien Hsien, Penghu Hsien.
North Dummy variable = 1 if the household is located in: Taipei Hsien, Keelung City, Ilan Hsien, Taoyuan Hsien, Hsinchu Hsien, Miaoli Hsien, and Taipei Municipality. (North is the default variable)
A person in the household who earns the largest personal share of pay in family income is considered to be the economic household head.
Hypotheses
NHI
Prior to NHI, about 40% of the population had no social health insurance. NHI enrolled more than 96%. With the NHI offering generous medical benefit coverage, the medical care for most patients, whose medical expenses had been previous paid out-of-pocket, would be paid by insurance program and their financial burden mitigated. We hypothesized that NHI would reduce out-of-pocket expenditures.
Income
Household incomes were categorized into five categories: the lowest 20%, lower 20%, middle 20%, higher 20%, and the highest 20%. The lowest 20% (first quintile) served as our reference group. It has been reported that the higher the income, the greater the utilization of health care and the higher the household out-of-pocket medical expenditures [21]. In other words, the rich would be generally expected to have more out-of-pocket expenditures than the poor. We hypothesized that the coefficient of four dummy income variables would be significant and positive.
Income/NHI
Since NHI imposed the ceilings on co-payment to avoid catastrophic expenses, income tended to play a relatively smaller role in access to medical care after NHI. In other words, NHI helps reduce out-of-pocket expenditures for medical care, but the highest difference would be observed in the top income quintile. We expect that the negative impact of NHI would dominate the positive impact of income. Four income/NHI interaction variables were expected to have significant and negative coefficients.
Region
As discussed above, uneven distribution of medical care resources in Taiwan has been the target of much criticism. Most resources are concentrated in the North where the population has higher accessibility to health care than residents in non-northern areas (Center, South, East). Out-of-pocket medical expenditures impose a relatively heavier financial burden on residents in non-northern areas than on those in northern areas. The regional factors had four dummy variables, North, Central, South, and East. Therefore, we treated the variable of North as the default variable and hypothesized that the three dummy regional variables would have positive coefficients.
Region/NHI
Because out-of-pocket medical expenditures impose a relatively heavier financial burden on residents in non-northern areas than on northern residents, the Bureau of NHI, has, since its inception in 1995, always endeavored to improve medical care accessibility for residents in remote areas. We expect that NHI mitigates regional inequality in out-of-pocket expenditures and hypothesized that the coefficients of region and NHI interactions would be negative.
Demographic variables
There were six other groups of variables, which were demographic: martial status, sex, employment status, age, education, and family size. The married household heads are likely to result in more household out-of-pocket medical expenditures than the unmarried due to more health care service use either by them or their spouses. Hence, the marital status variable should have a positive coefficient. Females were found to consume somewhat more health care than males do primarily because of childbearing[22]. In addition, since females are more careful about the health condition of the members in their family and possibly more likely to take them for medical care than males, the female household heads would incur more household out-of-pocket medical expenditures than the male. We expect that the sex variable (Male) should have a negative coefficient. Because it would cost the employed work time to seek health care, they should be less likely to seek care and, therefore, have fewer out-of-pocket medical expenditures than the unemployed. The unemployed, due to depression over being jobless, may also seek more health care then the employed. Being employed/Being unemployed would be expected to have a negative coefficient. Because health often deteriorates as a result of aging, we can assume the older an individual, the more health care he or she will seek. We expect that the three dummy age variables would have a positive impact on out-of-pocket medical expenditures. Most often the higher an individual's education, the more socially advantaged he or she will probably be and the more access he or she will have to medical care. In turn, the more medical care one seeks, the more his or her out-of-pocket medical expenditures. We expected four dummy education variables to have a positive impact. Family size is another important factor in the demand for medical care and the amount of more out-of-pocket medical expenditures. An increase in family size should increase the likelihood of medical care use and result in more out-of-pocket medical expenditures. The family size was expected to have a positive coefficient.
Results
After the implementation of NHI, total household out-of-pocket medical expenditures dropped from NT$23,046 in 1994 to NT$17,726 in 1996 (23.08%) (Table 3). This drop suggests that NHI effectively reduced household out-of-pocket medical expenditures. The highly significant and negative coefficient of NHI in the empirical model confirms this observation (Table 2).
Table 2 Multivariate Regression Results
Variables Coefficient Standard Error
Constant 8.6900 *** 0.0604
Demographic variables
Married 0.1511 *** 0.0188
Male -0.1399 *** 0.0194
Employed -0.2753 *** 0.0330
Age
25–44 -0.0284 0.0413
45–64 -0.0158 0.0430
65 and above 0.2983 *** 0.0524
(Under 25 is the default variable)
Education
Junior high 0.0561 ** 0.0184
Senior (vocational) high 0.1269 *** 0.0169
Junior college 0.2568 *** 0.0220
College and above 0.2891 *** 0.0230
(Primary school and below is the default variable)
Family members 0.1668 *** 0.0053
NHI -0.2657 *** 0.0391
Household income
Lower 20%~2nd quintile 0.1638 *** 0.0362
Middle 20%~3rd quintile 0.3405 *** 0.0358
Higher 20%~4th quintile 0.5847 *** 0.0363
Highest 20%~5th quintile 0.8981 *** 0.0371
(Lowest 20%~1 quintile is the default variable)
Regional variables
Center 0.1077 *** 0.0237
South 0.0010 0.0217
East 0.3802 *** 0.0504
(North is the default variable)
Interaction effects
Lower 20% income/NHI 0.0327 0.0443
Middle 20% income/NHI -0.0664 0.0441
Higher 20% income/NHI -0.2465 *** 0.0445
Highest 20% income/NHI -0.4146 *** 0.0447
Center/NHI 0.0785 ** 0.0322
South/NHI 0.1693 *** 0.0286
East/NHI -0.4471 *** 0.0639
Numbers of observation 23925
F 177.81
Prob >F 0.0001
R-squared 0.1621
Adjust R-squared 0.1612
*** Significant at 0.01 level; ** significant at 0.05 level; * significant at 0.1 level.
Table 3 Mean Differences in Out-of-Pocket Expenditures Before and After NHI
Pre-NHI (1994) Post-NHI (1996) Total percentage change
Total expenditures 23046 17726 -23.08%
North 23885 17495 -26.75%
Center 23503 16950 -27.88%
South 20397 18645 -8.59%
East 29606 17389 -41.27%
Family income
Lowest 20%~1st quintile 13551 11390 -15.95%
Lower 20%~2nd quintile 15218 14559 -4.33%
Middle 20%~3rd quintile 18398 16783 -8.78%
Higher 20%~4th quintile 24565 19409 -20.99%
Highest 20%~5th quintile 35738 26227 -26.61%
Unit: NT$, 1994: 1US$ = 26.24NT$; 1996: 1US$ = 27.79NT$
Table 3 shows that average medical out-of-pocket expenditures increased in conjunction with increased income. Pre-NHI expenditures ranged from NT$13,551 to NT$35,738. Post-NHI expenditures ranged from NT$11,390 to NT$26,227. These statistics reveal that households with income in the higher quintiles seemed to have more out-of-pocket medical expenditures both before and after NHI. That is, income should be positively related to out-of-pocket expenditures in Taiwan. The four household income dummy variables were, as expected, found to have significant and positive coefficients. Moreover, the percentage of expenditures is shown to decrease 15.95% for households in the lowest income quintile after NHI in Table 3. This is less than the decreases in the highest 20% (-26.61%) and higher 20% (-20.99%) income quintiles, but greater than in the lower 20% (-4.33%) and middle 20% (-8.78%) income quintile households. In Table 2, Income/NHI interactions were, as expected, found to be significantly related to household out-of-pocket expenditures. All interactions were shown to have negative coefficients except for lower 20% income/NHI. Surprisingly, only two of Income/NHI interactions (higher 20% and highest 20%) reached statistical significance.
The range of out-of-pocket expenditures, all four regions included, was $20,397 – $29,606 pre-NHI and $16,950 – $18,645 post-NHI (Table 3). The eastern area had the highest out-of-pocket expenditures before NHI. After NHI, however, the southern area, previously lowest, had the highest out-of-pocket expenditures. Moreover, household out-of-pocket expenditures in the eastern area decreased by 41.27%, a much greater decrease than was found in the northern (-26.75%) and central (-27.88%) areas, and 5 times greater the -8.59% decrease found for the southern area. As can be seen in the summary of the three regional dummy variables in Table 2, the central and eastern areas were found to have significant and positive coefficients. The coefficient for the southern area was statistically insignificant. The coefficients of three region/NHI interaction effects were all significant. Only one interaction effect (East/NHI) was negative; Center/NHI and South/NHI were surprisingly found to be positive, suggesting that households in the central and southern areas tended to have more out-of-pocket medical expenditures than those in the northern areas, though NHI had been implemented. Moreover, only households in the eastern area tended to have fewer out-of-pocket medical expenditures than those in the North after NHI.
All demographic variables had significant effects except for two age dummy variables, 25 – 44 and 45 – 64 (Table 2). An individual who was older, female, married, unemployed, better educated or the head of a larger family household was likely to have greater household out-of-pocket medical expenditures. Household heads older than 65 tended to have more out-of-pocket medical expenditures than those below that age. There was no significant difference between household heads younger than 25 years and those between 25 and 64 years old.
Discussion
According to the Department of Health, total health expenditures increased steadily over the past decade, from 3.96% of GDP in 1981 to 4.93 of GDP in 1994, but jumped up to over 5.27% in 1995 with the introduction of National Health Insurance. During the same period, the annual growth rate of per capita health expenditures averaged below 10%, but reached 15% in 1994/1995, suggesting that there was a considerable increase in health care utilization and health expenditures with the implementation of NHI. [23]
Although NHI program with its generous medical coverage encourages greater utilization of medical care, it also helps to alleviate financial burden of private sector. The figures showed that private expenditures, which constituted 44.71% of total health expenditures in 1993, dramatically decreased from 43.74% in 1994 to 34.57% in 1995.[23] Our study showed a significant decrement in average household out-of-pocket payments from NT$23,046 in 1994 to NT$17,726 in 1996 (23%), further confirming that NHI significantly reduced household out-of-pocket expenditures.
The finding that higher income households are more likely to have larger out-of-pocket medical expenditures than lower income households is consistent with most international studies as indicated above. However, after the implementation of NHI, lower and middle income quintiles had a relatively small decrease in out-of-pocket medical expenditures, suggesting that NHI led to a significant increase in utilization in the lower income groups. The likely explanation may be that since the coverage rate of social health insurance in 1994 was much lower for the two lower income groups than the highest income group (66.7%, 86.3% vs 94.3%), the effect of greater insurance coverage was matched by increasing utilization. Moreover, NHI is found to reduce out-of-pocket medical expenditures more for households with lowest income than those with lower and middle incomes possibly due to the co-payment exemption provided to the lowest-income households. In general, NHI did reduce the gap in out-of-pocket expenditures among different income groups. Pre-NHI household out-of-pocket expenditures in the highest quintile were approximately 3 times more than in the lowest quintile. After NHI, this multiple was reduced to 2.
The households in the central and eastern areas were found to be likely to have more out-of-pocket medical expenditures than households in the northern area. This difference may be because in 1994 health insurance was less prevalent in these two areas than the northern area (84.7%, 82.8% vs 87.6%). Another likely explanation may be that the some parts of central areas and most of eastern areas are remote areas with more mountainous terrain. Therefore, there are fewer medical resources located in these areas, making medical care less accessible there than in the northern area. This inconvenience may discourage utilization and encourage residents to self-medicate, which may negatively affect their health and possibly result in future catastrophic medical expenditures and more out-of-pocket medical expenditures. No financial difference between southern and northern areas was observed, possibly because a similar proportion of the population was insured (85.6% vs 87.6%) and because had similar levels of economic development.
NHI was not found to effectively reduce regional inequality in medical financial burdens, but it was found to ease the financial burden of eastern residents more than it did for the northern residents, possibly because people residing in mountainous areas or on offshore islands are exempt from co-payment fees. Since the eastern area, including Taitung, Hualian, and Penghu1 Counties, is less developed than other areas in Taiwan, the Bureau of NHI has tried to increase accessibility of medical care there, especially in the aboriginal areas and outlying islands, in many ways. With regard to supply, a number of medical payment incentives have been implemented to encourage healthcare providers to extend medical care to residents of those areas. These incentives include higher insurance reimbursed outpatient physician fees, no limit on the number of insurance-reimbursed visits a patient would make to a physician and increased payment for circuit medical services. With regard to demand, the Bureau of NHI exempts residents in these areas from co-payment requirements and covers the registration fees in aboriginal areas, making the cost of medical care there the lowest in Taiwan. Furthermore, for the people in remote areas, the Bureau of NHI subsidizes transportation costs, including emergency helicopter transportation.
Conclusion
The implementation of NHI is thought to be the most important health care reform in Taiwan. It has not only increased accessibility of health care service, but it has also made impressive strides towards the reduction of out-of-pocket medical expenditures. Such an outstanding achievement in Taiwan's public health has been praised by both domestic residents and international health experts. Nonetheless, major improvements are still required to reduce the financial burden imposed on certain disadvantaged groups, such as the poor and aborigines. The time and money it takes to reach health facilities increases as the distance to these medical institutions increases. Such costs create a greater financial burden on socio-economically disadvantaged groups. In spite of free health services provided by NHI, it is likely that the cost of accessing health care negatively affects their decisions to receive medical assistance. Therefore, outreach services that target groups for whom access to health care remains problematic should be expanded under the auspices of universal health insurance.
Although the data in this study provides information on household consumption of health care, it only focuses on the collection of out-of-pocket health expenditures. This data is limited in that it does not allow us to further investigate any changes in health care use with the implementation of NHI. For example, we were not able to explore the differences in health care use, total health expenditures and estimates of elasticity of utilization among different income quintiles and regions in this paper. Future studies can focus on these issues using insurance claim data collected by the Bureau of NHI.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
TB, TC and CS contributed to conceptualization, analysis, and writing of the article. YW helped to perform the statistical analysis as well as to interpret findings. WT supervised all aspects of policy implications. All of us approved the final version. TC is the guarantor.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was supported by the National Science Council of the Republic of China (NSC-94-2416-H-305-005). The authors gratefully acknowledge editors' and reviewers' constructive comments.
==== Refs
The Economist Intelligence Unit Healthcare Industry: Healthcare International 2000 London, United Kingdom, The Economist Intelligence Unit
Cheng SH Chiang TL The effect of universal health insurance on health care utilization in Taiwan JAMA 1997 278 89 98 9214512 10.1001/jama.278.2.89
Liu TC Chen CS Chen LM The impact of national health insurance on neonatal care and childhood vaccination in Taiwan Health Policy Plan 2002 17 384 392 12424210 10.1093/heapol/17.4.384
Chen CS Liu TC Chen LM National Health Insurance and the antenatal care use: a case in Taiwan Health Policy 2003 64 99 112 12644332 10.1016/S0168-8510(02)00141-0
Chen CS Liu TC The Taiwan National Health Insurance program and full infant immunization coverage Am J Public Health 2005 95 301 311 15671469
Bureau of National Health Insurance National Health Insurance Profile 2001 2002 Department of Health, Taipei, Taiwan, Bureau of National Health Insurance
Rassel E Bernstein J Tang K The impact of health care financing on family budgets Int J Health Serv 1994 24 691 714 7896470
Pannarunothai S Mills A The poor pay more: health-related inequality in Thailand Soc Sci Med 1997 44 1781 1790 9194240 10.1016/S0277-9536(96)00287-0
Castano R Arbelaez J Giedion U Morales L Equitable financing, out-pf-pocket payments and the role of health care reform in Colombia Health Policy Plan 2002 17 5 11 12477736 10.1093/heapol/17.suppl_1.5
Sepehri A Sarma S Simpson W Does non-profit health insurance reduce financial burden? Evidence from the Vietnam living standards survey panel paper presented at the 38th Annual Meetings of the CEA at Ryerson University, Toronto, June 4–6, 2004
Yeh HJ Equity in the out-of-pocket expenditures under NHI Working paper (in Chinese) 1997
Chen CS Liu TC Lin CJ The analysis of family income and out-of-pocket medical expenditure financial burden in Taiwan Taiwan Bank Quarterly Journal 2005 56 200 229 (in Chinese)
Department of Health Taiwan Public Health Reports 2000 Executive Yuan, Taipei, Taiwan, Department of Health
Bureau of Labor Insurance Statistical Data for Taiwan-Fukien Area Labor Insurance 1994 Taipei, Taiwan, Bureau of Labor Insurance
Department of Health Toward The Goal of Health for All – A Health White Paper 1994 Executive Yuan, Taipei, Taiwan, Department of Health
Central Trust of China Statistic Central al Data for Government Employees Insurance 1994 Taipei, Taiwan, Central Trust of China
Directorate-General of Budget Accounting and Statistics 1994 Survey of Family Income and Expenditure in Taiwan Area 1994 Taipei, Taiwan, Directorate-General of Budget, Accounting and Statistics
Directorate-General of Budget Accounting and Statistics 1996 Survey of Family Income and Expenditure in Taiwan Area 1996 Executive Yuan, Taipei, Taiwan, Directorate-General of Budget, Accounting and Statistics
Deaton A Paxson C Intertemporal choice and inequality Journal of Political Economy 1994 102 437 467 10.1086/261941
Liu TC Chen CS An analysis of private health insurance purchasing decisions with national health insurance in Taiwan Soc Sci Med 2002 55 755 774 12190269 10.1016/S0277-9536(01)00201-5
Parker SF Wong R Household income and health care expenditures in Mexico Health Policy 1997 40 237 255 10168755 10.1016/S0168-8510(97)00011-0
Miller L Medical schools put women in curricula Wall St J B1 1994, May 24
Department of Health Statistic of final expenditure for health, 1991–2003 Department of Health Website 2004
|
16137336
|
PMC1208885
|
CC BY
|
2021-01-04 16:31:53
|
no
|
BMC Health Serv Res. 2005 Sep 2; 5:60
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-60
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-631609822710.1186/1471-2334-5-63Research ArticleA nested-PCR with an Internal Amplification Control for the detection and differentiation of Bartonella henselae and B. clarridgeiae: An examination of cats in Trinidad Rampersad Joanne N [email protected] John D [email protected] Michael S [email protected] Raymond [email protected] Shalini [email protected] David R [email protected] Dept. of Life Sciences, The University of the West Indies, St. Augustine, Trinidad, Trinidad and Tobago2 School of Veterinary Medicine, The University of the West Indies, Eric Williams Medical Sciences Complex, Mt. Hope, Trinidad, Trinidad and Tobago2005 12 8 2005 5 63 63 14 4 2005 12 8 2005 Copyright © 2005 Rampersad et al; licensee BioMed Central Ltd.2005Rampersad et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Bartonella species are bacterial blood parasites of animals capable of causing disease in both animals and man. Cat-Scratch Disease (CSD) in humans is caused mainly by Bartonella henselae and is acquired from the cat, which serves as a reservoir for the bacteria. A second species, B. clarridgeiae is also implicated in the disease. Diagnosis of Bartonellosis by culture requires a week or more of incubation on enriched media containing blood, and recovery is often complicated by faster growing contaminating bacteria and fungi. PCR has been explored as an alternative to culture for both the detection and species identification of Bartonella, however sensitivity problems have been reported and false negative reactions due to blood inhibitors have not generally been addressed in test design.
Methods
A novel, nested-PCR was designed for the detection of Bartonella henselae and B. clarridgeiae based on the strategy of targeting species-specific size differences in the 16S-23S rDNA intergenic regions. An Internal Amplification Control was used for detecting PCR inhibition. The nested-PCR was utilized in a study on 103 blood samples from pet and stray cats in Trinidad.
Results
None of the samples were positive by primary PCR, but the Nested-PCR detected Bartonella in 32/103 (31%) cats where 16 were infected with only B. henselae, 13 with only B. clarridgeiae and 3 with both species. Of 22 stray cats housed at an animal shelter, 13 (59%) were positive for either or both species, supporting the reported increased incidence of Bartonella among feral cats.
Conclusion
The usefulness of a single PCR for the detection of Bartonella henselae and B. clarridgeiae in the blood of cats is questionable. A nested-PCR offers increased sensitivity over a primary PCR and should be evaluated with currently used methods for the routine detection and speciation of Bartonella henselae and B. clarridgeiae. In Trinidad, B. henselae and B. clarridgeiae are the predominant species in cats and infection appears highest with stray cats, however B. clarridgeiae may be present at levels similar to that of B. henselae in the pet population.
==== Body
Background
Bartonella are fastidious, gram-negative, bacteria comprised of at least 19 species and 3 subspecies [1] that are obligate parasites of the blood in reservoir animals. Bartonella species are considered emerging zoonotic pathogens [2] and may be involved in a number of disease presentations including angiomatosis [3] and ocular manifestations [4]. Similarly, Bartonella species are being associated with disease in their animal hosts (see reviews [2,5]. The role of cats as a reservoir for human Bartonellosis is well documented however probably incomplete. Studies suggest that other Bartonella species, known to cause disease in humans, are found in the cat (see for example [6] and [7]. Of these, Bartonella henselae and, to a lesser extent, Bartonella clarridgeiae are known to cause Cat-Scratch Disease (CSD) in humans [8]; see also [9] for review of CSD.
As a fastidious organism, Bartonella usually requires over a week of incubation for primary isolation. The slow growth of the organism complicates its isolation since faster growing bacteria and fungi can overrun the plate. Thus various types of tests using Polymerase Chain Reaction (PCR) have been explored as a diagnostic tool for the detection and identification of Bartonella species from blood [10-12]. Previously, Jensen et al., [13] developed a PCR for the detection of Bartonella that targets species-specific size differences in the 16S-23S rDNA intergenic region. However, as a primary PCR, it was questionable if the sensitivity of the test was optimal for the detection of relatively low numbers of bacteria [14] and a control for the detection of false-negative reactions due to inhibition by blood components was not addressed. Herein we describe the development of a nested-PCR for the detection of B. henselae and B. clarridgeiae based on the strategy of species-specific size differences in the 16S-23S rDNA intergenic region that includes an Internal Amplification Control for PCR inhibitors. The test was evaluated on the blood of 103 apparently healthy cats in Trinidad to investigate the presence of these organisms in the local cat population and to verify the test's ability to detect these organisms in the blood of apparently healthy animals.
Methods
Specimen collection
All samples were collected over an 11 month period in 2001. Blood samples were collected in commercial blood collection tubes containing EDTA and transported to the laboratory on ice, where possible the same day, or stored at 4°C until transported. Samples were collected from geographically distinct areas in Trinidad including an animal shelter, private veterinary clinics and the Veterinary Hospital located at the University of the West Indies' School of Veterinary Medicine by both Veterinarians and final year veterinary students.
PCR
DNA was extracted from whole blood according to the method of Boom et al., [15] with modifications as described by Rampersad et al., [16]. One microlitre of blood-extracted DNA template was used in the primary PCR reaction and 1 μl of the primary reaction was used in the nested reaction. In order to minimize contamination, separate rooms were used for preparing the PCR reaction mix, template preparation, gel electrophoresis, and performing nested reactions. Nested reactions were performed in a biosafety hood where 1 μl of primary reaction was delivered to a nested tube with a 1 μl urine loop, the end of the loop was then washed briefly in a 10% bleach solution, rinsed in water then flamed before reusing.
Primary and nested-PCR reactions were optimized by testing 1.5, 2 and 3 mM MgCl2 in the reaction, each at 12 temperatures ranging from 45°C to 65°C using an Eppendorf Gradient Master Cycler. B. henselae and B. clarridgeiae strains used for template were a gift from Dr. Jill Clarridge. Optimized PCR cycle conditions were 94°C -15 s, 48.2°C -30 s and 72°C -30 s for 40 cycles for the primary-PCR and 94°C -15 s, 56°C -30 s and 72°C -30 s for 40 cycles for the nested-PCR. The PCR strategy was an adaptation of a primary PCR described by Jensen et al., [13]. Primary PCR primers were P-bhenfa (5'-TCTTCGTTTCTCTTTCTTCA) and P-benr1 (5'-CAAGCGCGCGCTCTAACC) which gave an approximately 186 bp fragment for B. henselae and a 168 bp fragment for B. clarridgeiae, and nested primers N-bhenf1a (5'-GATGATCCCAAGCCTTCTGGC) and N-bhenr (5'-AACCAACTGAGCTACAAGCC) which gave an approximately 152 bp fragment for B. henselae and a 134 bp fragment for B. clarridgeiae. The Internal Amplification Control (IAC) consisted of a DNA amplicon, randomly amplified from the genome of a chicken with primer IAC (5'-TGTTTGACAGCTTATCAT). All reactions were performed in a 25 μl volume as follows: Primary reaction (0.2 mM each dNTP, 0.5 pmoles/μl each P-bhenfa and P-benr1, 3 mM MgCl2 reaction buffer, 0.4 pmoles/μl primer IAC, 1 μl of IAC template stock (concentration unknown), 0.5 units Taq polymerase (Promega) and Nested (0.2 mM each dNTP, 0.5 pmoles/μl each N-bhenf1a and N-bhenr, 1.5 mM MgCl2 reaction buffer, 0.4 pmoles/μl primer IAC, 0.5 units Taq polymerase).
The B. clarridgeiae-specific PCR was performed using 1 μl of a primary reaction in a PCR reaction with primers bclarF (5'- GCACAAGCCTCTGAGAGGGA) and N-bhenr. Reactions were performed in a 25 μl volume as follows: Primary reaction (0.2 mM each dNTP, 0.5 pmoles/μl each primer, 3 mM MgCl2 reaction buffer, 0.5 units Taq polymerase). PCR cycle conditions were 94°C -15s, 55°C -30 s and 72°C -30 s for 40 cycles.
PCR amplification products were separated on either a 2% or 3% agarose gel, stained with ethidium bromide and visualized under 302 nm ultra-violet light. DNA markers were created by PCR amplification of virulence genes from Escherichia coli generating calculated-sized fragments (bp) of 881, 520, 337, 275 171 and relative sizes confirmed using a commercial DNA size standard (100 bp ladder, Promega).
Culture
One hundred microliters of blood was spread on a Brain Heart Infusion Agar plate (Difco) supplemented with 10% whole sheep blood and incubated for 2–3 weeks at 37°C in 5% CO2. In some cases the blood was frozen at -70°C before being cultured.
DNA sequencing
Amplification products generated in the nested reaction for B. henselae and B. clarridgeiae were electrophoresed through a 1% agarose gel. The agarose containing the amplified DNA was cut from the gel, briefly frozen then thawed and spun at 12,000 rpm through a polypropylene fibre to remove the DNA-containing liquid. The DNA was concentrated by ethanol precipitation and used in the sequencing reaction. An ABI Big Dye sequencing kit was used according to manufacturer's instructions using the nested primers individually in two separate reactions, and the reactions resolved on an ABI 377 DNA sequencing machine.
Results
Of 103 blood samples subjected to PCR, none of the samples were positive for Bartonella in the primary reaction, but 32 were positive in the nested reaction. Inhibition was observed in 3 primary reactions indicated by a relatively light band seen for the IAC. However, one of these reactions was later positive for both B. henselae and B. clarridgeiae with the nested reaction. Two amplification bands of differing mobilities were observed in the nested reaction, both migrating below the 171 bp size marker (Figure 1). The larger fragment co-migrated with an amplification product generated from B. henselae (data not shown) and the smaller band was of a size expected from B. clarridgeiae. When the primary PCR reaction was used as template in a PCR reaction with a primer specific for B. clarridgeiae, only those samples that had the smaller sized fragment were positive. The identity of the respective amplification fragments were further investigated by DNA sequencing wherein the larger fragment was confirmed to be from B. henselae (GenBank: [DQ000494]) and the smaller-sized fragment from B. clarridgeiae (GenBank: [DQ003029]).
Figure 1 PCR amplification products. A 3% agarose gel showing lane 1) DNA size markers (bp) of 881, 520, 337, 275 171 and lanes 2–6) nested-PCR reactions of blood samples. Band a) internal amplification control, band b) B. henselae, and band c) B. clarridgeiae.
The number of Bartonella-negative samples along with those positive for either or both Bartonella species is given for both pet and stray animals in Table 1. Where the age of cats was known, infected cats (not including those from the shelter) ranged from 2–72 months. For stray cats at the animal shelter of unknown age, one half (11) of the cats were classified as adults and the other 11 as kittens, of which 6 of the 13 infected animals were adults and 7 kittens.
Table 1 Bartonella species associated with living environments of cats
Pets Animal shelter Total
B. henselae 7 9 16
B. clarridgeiae 10 3 13
Both species 2 1 3
Bartonella negative 62 9 71
Total 81 22 103
The number of Bartonella-negative samples along with those positive for either or both Bartonella species is given for both pet and stray animals.
Blood cultures were generally contaminated with both spreading bacteria and fungi, to such a degree that it was felt that little meaningful data could be gained through culture, and it was discontinued.
Discussion
As an emerging pathogen known for causing disease in both humans and animals, a simple test for the routine identification of Bartonella species is desirable. Because multiple Bartonella species are associated with disease in both humans and animals, it would be advantageous to have a test capable of both detecting different species and differentiating between them. The strategy first presented by Jensen et al., [13] generally fits these requirements by targeting DNA-length heterogeneity in the 16S rDNA-23S rDNA intergenic region. As such, this test is well suited for the detection of B. henselae and B. clarridgeiae, the causative agents of CSD in humans. However, as a single PCR reaction, this test has questionable sensitivity and has in fact been reported to show less sensitivity than culture for the detection of B. henselae and B. clarridgeiae in blood, [14]; see also [17]. Similarly, the primary reaction used in this study was unable to detect Bartonella DNA in any of the samples although we were able to show the presence of Bartonella DNA in 32 samples with a nested reaction. These results question the usefulness of a single PCR reaction for the detection of Bartonella in blood without further enhancing the overall sensitivity of the test, such as with the addition of a nested-PCR reaction. However, many laboratories are reluctant to use a nested reaction for fear that its high sensitivity will detect contaminating DNAs (mainly amplicons) and deliver a false-positive result. In this work, none of our negative controls were positive. To help control cross contamination, template preparation, assembly of the PCR reaction, nested reactions and gel electrophoresis were performed in different rooms. Especially useful was the use of a UV hood and a 1 μl urine loop instead of barrier tips to transfer an aliquot of the primary reaction to the nested reaction. By immersing the end of the loop, after each transfer, in a 10% bleach solution to chemically degrade any amplicons, and then rinsing with water before flaming, any chance splattering or aerosolization of DNAs from the loop during flaming was reduced.
The test of Jensen et al., [13] also lacked an Internal Amplification Control (IAC) to detect PCR inhibition. In fact, arguments have been made for Internal Amplification Controls to be mandatory when describing diagnostic tests based on PCR [18]. Since blood, which is a known source of PCR inhibitors, is a commonly used sample for Bartonella detection, it is important that PCR-based tests for Bartonella include an IAC. Individuals interested in evaluating this method may write the authors for an aliquot of the IAC template.
Because of the slow growth and long culture time for Bartonella, faster growing contaminating bacteria and fungi can overrun a culture plate and make the detection of Bartonella difficult or even impossible. Early in this study we found that the contamination problem was so severe that culture was discontinued. The unexpectedly high level of contamination may have originated from the blood collection process where blood was collected by both private veterinary clinics and veterinary students as part of their studies, and thus was not strictly controlled. Although perhaps undesirable for the purposes of this study, the uncontrolled nature of its collection perhaps more accurately represented blood as it would arrive to the laboratory for diagnosis, and underscores one advantage of PCR over culture for the detection of Bartonella.
Prevalence studies reported for Bartonella infection in cats are generally characterized by a relatively high prevalence of B. henselae compared to B. clarridgeiae [19-22]. In this study, our results were mostly consistent with other studies where B. henselae was detected in slightly more samples than B. clarridgeiae. However approximately 40% of the infected cats originated from an animal shelter, which were not necessarily representative of the cat population as a whole in Trinidad. For example, discounting the animals from the shelter, 7 (8.6%) animals were positive for B. henselae, 10 (12.3%) for B. clarridgeiae and 2 (2.5%) for both bacteria, suggesting that B. clarridgeiae may be found among pet cats at levels similar to, or even higher than that for B. henselae. As has been reported in other studies [21,22], cases of mixed infections with both species were found. In addition, there was a clear relationship between cats expected to be exposed to fleas (e.g. stray cats at an animal shelter), and the prevalence of infection.
Although this test was designed for the detection of B. henselae and B. clarridgeiae, sequence homology between the primers and other Bartonella species indicates that this test might be useful for the detection of other Bartonella species (Figure 2). Of the 15 species for which sequence data was available, only three species, B. bovis, B. schoenbuchensis and B. birtlesii would not be expected to amplify due to a single base pair mismatch between the template and the terminal 3' position of the primer. Admittedly, our test would benefit from a redesign of the forward primers (Figure 2), so that they would recognize all known Bartonella species, [23]. Jensen et al., [13] had identified sequence homology in the 16S-23S rRNA intergenic region between Bartonella with Brucella species. In this test, sequence homology indicates that the primary PCR should amplify Brucella DNA, however no such amplification should occur in the nested reaction (Figure 2).
Figure 2 Primer and template sequence homology. Bacterial species are followed by their GenBank accession numbers. Primer sequences are given below the primer's name at the top, primary primers are in bold, nested primers are in italics and sequence shared by two primers is underlined. A dot (.) below each nucleotide in the primer represents homology at that position in each species' DNA with the primer and letter signifies the diverging nucleotide found in the bacteria DNA. A dash (-) represents spaces placed in the sequence for alignment purposes. Three asterisks (***) signifies the variable length region of DNA found between the forward and reverse primers and the number given at that site for each species indicates the size of the amplicons generated from the nested reaction.
Conclusion
The usefulness of a single PCR for the detection of B. henselae and B. clarridgeiae in the blood of cats is questionable and enhancement methods, such as a nested-PCR, should be utilized. In Trinidad, B. henselae and B. clarridgeiae are found in cats and infection appears highest among stray cats, however B. clarridgeiae may be present at levels similar to, or higher than, that of B. henselae in pet cats.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JNR and DRA provided project and diagnostic design/development and performed diagnostics, JDW conceived the project and provided project design and management, MSS performed template preparation and sample management, RD and SR collected blood and information on the animals.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Dr. Jill Clarridge for providing us with strains of B. henselae and B. clarridgeiae and her helpful suggestions on the culture of these organisms from blood, and the DNA Unit at Mt. Hope Trinidad for assistance with DNA sequencing. We also acknowledge the financial support provided to DRA by a grant from the University of the West Indies' Research and Publication Committee (Grant# CRP.4Z).
==== Refs
List of Bacterial Names with Standing in Nomenclature – Genus Bartonella
Boulouis H Chang C Henn J Dasten R Chomel B Factors associated with the rapid emergence of zoonotic Bartonella infections Vet Res 2005 36 383 410 15845231 10.1051/vetres:2005009
Santos R Cardoso O Rodrigues P Cardoso J Machado J Afonso A Bacellar F Marston E Proenca R Bacillary angiomatosis by Bartonella quintana in an HIV-infected patient J Am Acad Dermatol 2000 42 299 301 10642694 10.1016/S0190-9622(00)90147-6
Khurana RN Albini T Green RL Rao NA Lim JI Bartonella henselae infection presenting as a unilateral panuveitis simulating Vogt-Koyanagi-Harada syndrome Am J Ophthalmol 2004 138 1063 1065 15629311 10.1016/j.ajo.2004.06.036
Breitschwerdt EB Kordick DL Bartonella infection in animals: Carriership, reservoir potential, pathogenicity, and zoonotic potential for human infection Clin Microbiol Rev 2000 13 428 438 10885985 10.1128/CMR.13.3.428-438.2000
Hjelm E McGill S Blomqvist G Prevalence of antibodies to Bartonella henselae, B. elizabethae and B. quintana in Swedish domestic cats Scand J Infect Dis 2002 34 192 196 12030392 10.1080/00365540110080188
Aboudharam G La VD Davoust B Drancourt M Raoult D Molecular detection of Bartonella spp. in the dental pulp of stray cats buried for a year Microb Pathog 2005 38 47 51 15652295 10.1016/j.micpath.2004.10.004
Kordick DL Hilyard EJ Hadfield TL Wilson KH Steigerwalt AG Brenner DJ Breitschwerdt EB Bartonella clarridgeiae, a newly recognized zoonotic pathogen causing inoculation papules, fever, and lymphadenopathy (Cat-Scratch Disease) J Clin Microbiol 1997 35 1813 1818 9196200
Lamps LW Scott MA Cat-scratch disease historic, clinical, and pathologic perspectives Am J Clin Pathol 2004 121 S71 80 15298152
Del Prete R Fumarola D Ungari S Fumarola L Miragliotta G Polymerase chain reaction detection of Bartonella henselae bacteraemia in an immunocompetent child with cat-scratch disease Eur J Pediatr 2000 159 356 359 10834521 10.1007/s004310051286
Avidor B Graidy M Efrat G Leibowitz C Shapira G Schattner A Zimhony O Giladi M Bartonella koehlerae, a new cat-associated agent of culture-negative human endocarditis J Clin Microbiol 2004 42 3462 3468 15297484 10.1128/JCM.42.8.3462-3468.2004
Ciervo A Ciceroni L Rapid detection and differentiation of Bartonella spp. by a single-run real-time PCR Mol Cell Probes 2004 18 307 312 15294318 10.1016/j.mcp.2004.04.004
Jensen WA Fall MZ Rooney J Kordick DL Breitshwerdt EB Rapid Identification and differentiation of Bartonella species using a single-step PCR assay J Clin Microbiol 2000 38 1717 1722 10790087
Engvall EO Fasth C Brandstrom B Fermer C Blomqvist G Englund L Prevalence of Bartonella henselae in young, healthy cats in Sweden Veterinary Record 2003 152 366 369 12678262
Boom R Sol CJA Salimans MMM Jansen CL Wertheim-van Dillen PME Van der Noordaa J Rapid and simple method for purification of nucleic acids J Clin Microbiol 1990 28 495 503 1691208
Rampersad J Cesar E Campbell MD Samlal M Ammons D A field evaluation of PCR for the routine detection of Babesia equi in horses Vet Parasitol 2003 114 81 87 12781470 10.1016/S0304-4017(03)00129-8
de Souza Zanutto M Mamizuka EM Raiz R Jrde Lima TM Diogo CL Okay TS Hagiwara MK Experimental infection and horizontal transmission of Bartonella henselae in domestic cats Rev Inst Med Trop Sao Paulo 2001 43 257 261 11696847
Hoofar J Cook N Malorny Wagner M De Medici D Abdulmawjood A Fach P Making Internal Amplification Control Mandatory for Diagnostic PCR J Clin Microbiol 2003 41 5835 14662997 10.1128/JCM.41.12.5835.2003
Rolain JM Locatelli C Chabanne L Davoust B Raoult D Prevalence of Bartonella clarridgeiae and Bartonella henselae in domestic cats from France and detection of the organisms in erythrocytes by immunofluorescence Clin Diagn Lab Immunol 2004 11 423 425 15013998 10.1128/CDLI.11.2.423-425.2003
Heller R Artois M Xemar V De Briel D Gehin H Jaulhac B Monteil H Piemont Y Prevalence of Bartonella henselae and Bartonella clarridgeiae in stray cats J Clin Microbiol 1997 35 1327 1331 9163438
Maruyama S Sakai T Morita Y Tanaka S Kabeya H Boonmar S Poapolathep A Chalarmchaikit T Chang CC Kasten RW Chomel BB Katsube Y Prevalence of Bartonella species and 16S rRNA gene types of Bartonella henselae from domestic cats in Thailand Am J Trop Med Hyg 2001 65 783 7 11791975
Chomel BB Carlos ET Kasten RW Yamamoto K Chang CC Carlos RS Abenes MV Pajares CM Bartonella henselae and Bartonella clarridgeiae infection in domestic cats from The Philippines Am J Trop Med Hyg 1999 60 593 7 10348234
Birtles RJ Hazel S Brown K Raoult D Begon M Bennett M Subtyping of uncultured bartonellae using sequence comparison of 16S/23S rRNA intergenic spacer regions amplified directly from infected blood Mol Cell Probes 2000 14 79 87 10799268 10.1006/mcpr.2000.0289
|
16098227
|
PMC1208886
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Aug 12; 5:63
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-63
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-641610916510.1186/1471-2334-5-64Research ArticleInducible nitric oxide synthase (iNOS) expression in monocytes during acute Dengue Fever in patients and during in vitro infection Neves-Souza Patrícia CF [email protected] Elzinandes L [email protected] Sonia MO [email protected] Rogério [email protected] Sonia RNI [email protected] Denise IS [email protected] Rita MR [email protected] Claire F [email protected] Departmento de Virologia, Instituto Oswaldo Cruz Fundação Oswaldo Cruz, Av. Brasil, 4365, CEP 21040-360 Rio de Janeiro, RJ, Brazil2 Departamento de Clínica Médica, Hospital Antonio Pedro, Universidade Federal Fluminense, Niterói, RJ, Brazil3 Instituto de Pesquisas Evandro Chagas, Fundação Oswaldo Cruz, Av. Brasil, 4365, CEP 21040-360 Rio de Janeiro, RJ, Brazil2005 18 8 2005 5 64 64 3 10 2004 18 8 2005 Copyright © 2005 Neves-Souza et al; licensee BioMed Central Ltd.2005Neves-Souza et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
Mononuclear phagocytes are considered to be main targets for Dengue Virus (DENV) replication. These cells are activated after infection, producing proinflammatory mediators, including tumour-necrosis factor-α, which has also been detected in vivo. Nitric oxide (NO), usually produced by activated mononuclear phagocytes, has antimicrobial and antiviral activities.
Methods
The expression of DENV antigens and inducible nitric oxide synthase (iNOS) in human blood isolated monocytes were analysed by flow cytometry using cells either from patients with acute Dengue Fever or after DENV-1 in vitro infection. DENV-1 susceptibility to iNOS inhibition and NO production was investigated using NG-methyl L-Arginine (NGMLA) as an iNOS inhibitor, which was added to DENV-1 infected human monocytes, and sodium nitroprussiate (SNP), a NO donor, added to infected C6/36 mosquito cell clone. Viral antigens after treatments were detected by flow cytometry analysis.
Results
INOS expression in activated monocytes was observed in 10 out of 21 patients with Dengue Fever and was absent in cells from ten healthy individuals. DENV antigens detected in 25 out of 35 patients, were observed early during in vitro infection (3 days), significantly diminished with time, indicating that virus replicated, however monocytes controlled the infection. On the other hand, the iNOS expression was detected at increasing frequency in in vitro infected monocytes from three to six days, exhibiting an inverse relationship to DENV antigen expression. We demonstrated that the detection of the DENV-1 antigen was enhanced during monocyte treatment with NGMLA. In the mosquito cell line C6/36, virus detection was significantly reduced in the presence of SNP, when compared to that of untreated cells.
Conclusion
This study is the first to reveal the activation of DENV infected monocytes based on induction of iNOS both in vivo and in vitro, as well as the susceptibility of DENV-1 to a NO production.
==== Body
Background
Dengue viruses (DENV) have been detected in several lymphoid organs originating from fatal cases of hemorrhagic disease, mainly in cells from the mononuclear phagocyte lineage [1]. Nevertheless, the frequency of disseminated virus detection in autopsies is low and microscopic injury is not sufficient to justify death during the Dengue Hemorrhagic Fever (DHF) or Dengue Shock Syndrome (DSS). It is believed that disease severity is a result of an immunopathological response leading to hemorrhagic and vasodynamic alterations and shock.
Mononuclear phagocytes have been considered as main targets for DENV replication [2,3] and recently the virus was detected in dendritic cells originating from one individual 24 hours post DENV infection [4,5]. Morens et al. [6] have shown strain differences of DENV, originating either from mild or severe cases, with respect to their growth in human monocytes, suggesting that characteristics inherent to virus genome and replication were related with disease severity. Viral RNA was detected by RT-PCR in peripheral blood leukocytes [7] in humans during infection and DENV antigens were present in circulating monocytes [8], Kupffer cells and macrophages from different organs such as liver, spleen and lungs [9].
Pathogenesis may consist of virus penetration into monocytes or dendritic cells, cell activation and synthesis induction of cytokines, arachdonic acids and very likely nitric oxide. These factors would be involved not only in generating dengue disease [10] but also in the elimination of viruses [11].
The in vivo activation of mononuclear phagocytes by interferon (IFN)-γ during DENV infection can be expected, since this cytokine is detected in serum during the acute phase [12,13] and is also produced by peripheral leukocytes from patients with previous dengue records [14]
The in vitro production of IFN-α,β, interleukin-6 and tumour-necrosis factor (TNF)-α by infected fibroblasts and/or macrophages was described [15] as well as the production of platelet activating factors [16]. Recently, viral titres during infection were correlated with the production of prostaglandin E2, which is known to be vasoactive [17].
Considering that dengue is an acute and short-termed disease, and IFN-α,β and TNF-α are molecules that induce antiviral defence mechanisms [11], it is conceivable that DENV replication is greatly inhibited in mononuclear phagocytes, although inflammatory reactions may be activated as well. It is well known that NO is produced by IFN-γ activation of monocytes [18] and has antiviral activities against Herpes Simplex Viruses, Epstein Bar Virus and some Coxsackie and Tick-borne encephalitis viruses [16]. NO levels were found to be increased in patients with dengue [19] or in vitro infected Kupffer cells [20], however the production has not been detected in infected monocytes in culture [21].
The aim of this research is to investigate iNOS monocyte induction during acute DENV infection in patients and after in vitro infection by DENV-1. We demonstrated that monocytes isolated from several patients became activated and express iNOS, which leads to the production of NO by the cell. DENV-1 was susceptible to a NO donor treatment and, in addition, virus was detected at higher rates in infected cells after iNOS inhibition, indicating that NO might play a substantial role in controlling DENV-1 infection of monocytes in culture and in vivo during natural infection.
Methods
Patients and laboratory diagnosis
Blood samples were obtained from DENV infected patients in April 2000 in Foz do Iguaçu, PR, and from February to April 2001 and January to March 2002 in Niterói, RJ, Brazil. Patients were diagnosed based on clinical grounds as Dengue Fever (fever, headache, retro-orbital pain, mialgias, arthralgias, rash and prostration); some had hemorrhagic manifestations, platelet counts under 100,000/mm3, hypotension receiving parenteral hydration and hospitalisation. The diagnosis of DENV infection was confirmed by anti-dengue enzyme-linked immunosorbent assay (ELISA)-IgM, or virus isolation. Informed written consent, approved by Fundação Oswaldo Cruz Ethical Committee under Nr#111/00, was obtained from all dengue patients prior to blood draw.
Cell cultures
Aedes albopictus C6/36 cell clone was grown as monolayers at 34°C on Leibovitz medium (L-15) supplemented with 200 mM glutamine, 1% non-essential amino acids solution, 19% tryptose phosphate broth, 100 U/ml penicillin, 10 μg/ml streptomycin and 5% foetal calf serum (FCS).
Preparation of virus stock and virus titration
DENV serotype 1, strain 16007 was provided by Dr. SB Halstead (Naval Medical Research Center, USA). Virus was titrated by serial dilution cultures in microtiter plates and detected by immunofluorescence as previously described [22]. Virus titre was calculated as 50 percent tissue culture infectious dose or TCID50/ml [23,24]. Inactivated virus was prepared by incubating the inoculum for 24 hours at 37°C and treating it with UV light (60 Hz, with distance of 20 cm) for 1 hour. Virus stock used was at a concentration of 3.18 × 107 TCID50/ml.
Preparation of human peripheral blood mononuclear leukocytes (PBMLs)
Mononuclear leukocytes were obtained from heparinised venous blood originating from either DENV-infected patients or dengue seronegative adult donors. Cells were isolated through density gradient centrifugation (350 g, 20 minutes in Ficoll-Paque Plus Amersham Biosciences Corp, Piscataway, USA). Cells were suspended in RPMI 1640 supplemented with 200 mM glutamine, 100 U/ml penicillin, 10 μg/ml streptomycin and 10% FCS and afterwards incubated at 37°C under humid atmosphere with 5%CO2. Cells isolated from patients with acute dengue were suspended in supplemented RPMI 1640 containing an additional 10% DMSO and 50% FCS and maintained in liquid nitrogen.
Infection of adherent PBMLs and treatment with an iNOS inhibitor
Infections were performed on 24-well plates. Freshly isolated PBML suspended in RPMI 1640 medium and supplemented with 10% FCS were seeded at 2 × 106cells/well. After an 18 hour-incubation, adherent cells were enriched by washing away unattached cells twice. Inoculum was diluted in 1 ml medium containing a multiplicity of infection approximately of 8 TCID50/adherent cell DENV-1. After a 2 hour-incubation for adsorption, 1 ml medium was added to achieve 10% FCS. After an 18 hour-incubation, all culture medium containing virus was removed and cultures were further incubated with fresh medium for up to 6 days. In some experiments the iNOS inhibitor, NG-methyl L-Arginine (NGMLA), was added at a final concentration of 400 μM. Wells were set in triplicates for each different parameter in culture. Cell viability was determined in culture by Trypan blue exclusion during 6 days. Duplicates of cell control, inactivated and infectious DENV were assayed. Viability ratios were calculated by dividing cell counts from each day by counts made just after virus adsorption.
Infection of C6/36 cell line and treatment with a NO donor
Infections were performed on 24-well plates. C6/36 were seeded at 2 × 105 cells/well in medium with 5% FCS and allowed to form a monolayer for 24 h at 34°C. Cell supernatant was removed and adherent cells were washed twice with medium without FCS. Wells were filled with 1 ml fresh medium containing DENV-1 or only medium in a 1:10 dilution from initial inoculum containing a multiplicity of infection of 1.59 TCID50 / cell for DENV-1. After a 90 minute-adsorption, 1.0 ml medium was added to achieve 2% FCS, containing sodium nitroprussiate (SNP), a NO donor, resulting in final concentrations of 10 or 100 μM [25]. Viruses were allowed to grow for two days in incubated cultures. Triplicate wells were set for each culture parameter. Viability of treated cells was confirmed by Trypan blue exclusion or propidium iodine uptake.
Virus labelling of infected C6/36 cells, single and double labelling of infected adherent PBMLs for flow cytometry analysis
Cells were recovered by scratching with plastic microtip using cold medium and were set at 1 × 106/microtube; they were centrifuged (350 g, 5 min) and washed once with 1 ml PBS pH 7.4 with 2% FCS and 0,01% NaN3. Surface labelling was performed with FITC-labelled antibodies to CD14 (DAKO, Denmark, 1:100 dilution) for 45 min directly on adherent viable PBMLs under ice bath. This was done to confirm that ~95% of the monocyte gated cells would be CD14+ on the infection day. Intracellular staining after infection was performed according to previously described [26] with slight modifications. Briefly, it required a fixation with 0.5 ml cold paraformaldehyde at 4% in PBS for 10 min and, after centrifugation, membrane permeabilization was carried out with 1 ml 0.1% saponine in PBS with FCS and NaN3. Monoclonal antibody Dengue Complex-reactive (Chemicon, USA, 1:200 dilution in PBS with saponine, FCS and NaN3) was added to cells for a 60-minute incubation. Cells were washed once with 1 ml PBS with FCS and NaN3 and further incubated with anti-mouse IgG labelled with PE or FITC (DAKO, USA, 1:50 dilution in PBS with saponine, FCS and NaN3) for 30 minutes. After washing, infected adherent PBMLs were further incubated with FITC-labelled antibody to iNOS (B&D Transduction Laboratories, USA, 1:100 dilution) for 45 minutes. Alternatively, for double staining CD14 and DENV, cells were firstly labelled with PE-labelled antibodies to CD14 (DAKO at dilution 1:100) for 30 minutes 4°C, washed, fixed with 0,5 ml cold 2% paraformaldehyde for 10 minutes and, after centrifugation, membrane permeabilization was carried out with 1 ml 0.1% saponine PBS with FCS and NaN3 and further labelled with Alexa 654 – labelled monoclonal antibody to Dengue for 30 minutes. Matching isotype antibodies were used for DENV and iNOS labelings. Finally cells were washed twice, ressuspended in 1% paraformaldehyde and kept at 4°C up to 3 days until acquisition by flow cytometry. Cells were acquired (10,000 events for cell lines and 5,000 for gated monocytes) on a FACS® Calibur flow cytometer (Beckon & Dickinson, USA) and analysed using FlowJo Software (TreeStar Inc., CA, USA). Isotype-matched antibodies were used as a negative control for all stainings.
Statistical analyses
Two-way Student's t test was performed using GraphPad Prism version 4.02 for Windows, GraphPad Software (San Diego, CA, USA, ) in order to determine the significance of differences in percentages of virus labelling found in cells infected under various conditions. Altered parameters were considered significant when P < 0.05.
Labelled cell rates in healthy donors were firstly tested for normality using the Prism program (values passed the normality test of Kolmogorow-Smirnov) and then cell patient frequencies were tested for sample positivity by Student's t-Distribution (tn-1 = 10, α = 0.025 = 2.228 or tn-1 = 9, α = 0.025 = 2.262) calculating a referential limit value for negativity, according to the following formula: Average of values from control samples + [Standard Deviation of values from control samples X t(n-1;α = 0.025)]. Determinations above referential limit values were considered positive.
Results
Characterization of target cells for DENV and iNOS expression in peripheral blood mononuclear leukocytes (PBMLs) originating from patients with Dengue Fever
Previous reports demonstrated human monocytes as targets for DENV in cultures [1,27]. Therefore, monocyte and lymphocyte subsets present in PBMLs originated from patients with acute Dengue Fever were analysed by flow cytometry (FACS). Figure 1 displays the cell size and granularity profiles of PBMLs from a healthy donor (Figure 1A) and a dengue patient (Figure 1C). CD14+ gated cells were selected (Figure 1B) and plotted as logical monocyte gate in the FSCxSSC dot plot (R1 in Figure 1C). Monocyte (R1) and lymphocyte (R2) gates selected were used for further studies. Approximately 95% cells of the R1 are CD14+ cells.
Figure 1 Dot and contour plots showing DENV-Ag and iNOS monocyte and lymphocyte expression in dengue patients. Human PBMLs a from healthy individual (Figures 1A,1D,1G) were used as control. Cells from a 4-day dengue infected patient (Figures 1B,1C,1E,1F,1H,1I) were labelled with anti-CD14-PE and CD14+ gated cells – R1 (Figure 1B) were considered as logical gate or monocyte gate and R2 is as the lymphocyte gate in the FCS vs. SSC dot plot (Figure 1C). R1 and R2 were used for further analysis during this work. Alternatively, human PBMLs were labelled either with an antibody to DENV Ags and anti-mouse IgG-PE (Figures 1D-1F) or with anti-iNOS-FITC (Figures 1G-1I) and analysed by FACS in either the R1, monocyte gate (Figures 1D-1F) or in the R2, lymphocyte gate (Figures 1F-1I). x-axis represent mean of cell population size (FCS) and y-axis represent cell granularity (SSC) or fluorescence intensity.
The monocyte and lymphocyte profile by FACS after DENV and iNOS immune-labelling were studied. Representative dot plots of DENV Antigens (Ag) and iNOS positive detection in monocytes from dengue patients (Figures 1E and 1H, respectively) are compared to monocytes from healthy individuals (Figures 1D and 1G) and lymphocytes from patients (Figures 1F and 1I). Patient monocytes presented DENV-Ag during the first eleven days of disease onset (Figures 1E and 2A) compared to control monocytes (Figures 1D and 2A). Patient cells collected at later stages of disease did not display detectable viral antigen (data not shown). Among the 35 dengue patients 25 expressed DENV-Ag above the referential limit value for negativity in the Student's t-Distribution and no difference was detected between early (1–5 days) and late (6–11 days) infection. Among 22 patients tested for iNOS expression (Figures 1H and 2B), 10 were positive as compared to controls (Figures 1G and 2B). Cells from the lymphocyte gate do express neither DENV-Ag (0.80 ± 0.77%) nor iNOS (3.7 ± 2.57%) (Figures 1F and 1I, respectively). The highest incidence of iNOS activation was from 6 to 10 days of disease (8 out of 20 patients). Among 11 patients with platelet counts<100,000/mm3, six had iNOS+ cells above the referential limit value for negativity. Seven of these patients also had hypotension, were hospitalised and submitted to parenteral rehydration. Among 11 patients with normal platelet levels merely 4 had significantly elevated iNOS+ cell ratios. INOS expression, present in a significant frequency among dengue patients, could not be associated with severity in this sampling.
Figure 2 Frequencies of DENVAg+ and iNOS+ monocytes in dengue patients.
Human PBMLs from dengue individuals were labelled with an antibody to DENV-Ag and an anti-mouse IgG-PE or with an antibody to iNOS FITC-labelled and data obtained by FACS in the monocyte gate. Percentages of DENV-Ag+ infected cells were determined in 11 healthy individuals and 35 with dengue (Figure 2A; 2 patients were measured in duplicate samples at different time points). Percentages of iNOS+ infected cells were determined in 10 healthy individuals and 22 with dengue (Figure 2B; 14 patients were measured in duplicate samples at different time points). Horizontal lines represent percentage averages of labelled cells from the population of individuals in each of the discrete categories.
Characterization of DENV target cells, DENV-Ag detection and iNOS expression in monocytes after in vitro infection
Adherent human PBMLs were incubated for 3–6 days with infectious and inactivated DENV and evaluated for presence of DENV-Ag and iNOS expression. Cell viability was determined daily by Trypan blue exclusion. After 3 days of culture, cells incubated with the inactivated virus remained 42 ± 7% viable and with infectious virus 42 ± 23% viable as compared to the day 0. On day 6 37 ± 2% cells were viable for inactivated and 44 ± 3% for infectious virus, showing no differences between the two virus incubations. Control cell rates of viability were higher, 58 ± 8% at day 3 and 69 ± 19% at day 6. Although infectious viruses may not be crucial in increasing monocyte mortality, viral antigens alter cell viability.
Kinetics of virus detection in mononuclear phagocytes varies according to virus strain and multiplicity of infection (MOI) used (Cologna et al., 2003). At the MOI used here for DENV-1 strain 16007 (8 TCID50/adherent cell) preliminary results displayed low rates of infection on the first day (in average 17 ± %), increasing at day 2 (average 31%). No iNOS labelling was detected on these first days of infection. Therefore, experiments were performed using infections from 3 to 6 days.
Figure 3B shows an image obtained under confocal microscopy in a field with a high frequency of infected adherent mononuclear leukocytes compared to the absence of fluorescent cells in control cells (Figure 3A). DENV-Ag+ cells were detected by flow cytometry among CD14+ cells, although very small rates of CD14- cells were also labelled for DENV-Ag (Figures 4A and 4B are labelling controls for Figure 4C). For further evaluations during in vitro infection, we studied the monocyte population, using the CD14+ logical gate, which presented the same profile as the R1 shown in Figure 1B).
Figure 3 Confocal microscopy of DENV-Ag expression after in vitro infection of monocyte-rich cultures. Adherent human PBMLs were incubated for three days either with cell culture medium (Figure 3A, 189.5 μm/field), or with infectious DENV (Figure 3B, 125 μm/field). Cells were labelled with antibody to DENV-Ag and anti-mouse IgG-FITC.
Figure 4 DENV-Ag and iNOS expression after in vitro Dengue infection of monocytes. Adherent human PBMLs were incubated for three days with infected DENV (Figures 4A, 4B and 4C). Cells were labelled either with isotype matched antibodies (Figures 6A and 6B) or antibodies to CD14-PE (y-axis) and DENV-Complex-Alexa-654 (x-axis) and were analysed by FACS (Figure 4C, ungated). Cells incubated for three to six days with cell culture medium, inactivated DENV or infectious DENV were labelled with antibody to DENV-Ag and anti-mouse IgG-PE and/or anti-iNOS-FITC and analysed by FACS. Isotype matched antibodies were used for both labelings (Figure 4D, 3 days for DENV-Ag and 6 days for iNOS labelling). x-axis represent mean of fluorescence intensity and y-axis represent percentage of maximum cell number counts. Figure 4E shows a kinetic curve comparing DENV-Ag and iNOS expression on cells from the monocyte gate. Each point represent average ± SEM percentages from samples set in triplicates. * P < 0.05 and in Student's T-test showing difference in DENV-Ag and iNOS expression compared to uninfected controls (CT). Data are obtained from one representative experiment out of three performed with different cell donors.
In Figure 4D we observed that infected monocytes displayed specific DENV-Ag expression when compared to monocyte incubation with inactivated virus and an isotype-matched antibody, suggesting virus replication during infection. INOS was specifically expressed in monocytes infected with DENV. Mock-infected monocytes by inactivated DENV did not yield viral antigens during culture. The inactivated virus consistently induced low rates of INOS+ monocytes, however differences were statistically insignificant when compared to controls. The infectious virus in all experiments resulted in significantly higher rates of INOS+ monocytes than in control cultures. Infectious virus, if not crucial, certainly play an important role in inducing iNOS. Control cells displayed only low levels of background labelling for both DENV-Ag and iNOS.
During in vitro infection DENV-Ag were markedly expressed in monocytes and detected on day 3 either by confocal microscopy (Figure 3B) or by flow cytometry (Figure 4). However, from day 4 to 6 Ags detection disappeared rapidly (Figure 4E) with insignificant rates of labelled cells (Student's T-test). INOS is apparently detected in cells already at day 3, but it reaches significantly higher frequencies of positive cells at days 5 and 6 (Figure 4E).
DENV-Ag detection in infected cells after treatment with an iNOS inhibitor – NGMLA or with a NO donor – SNP
Nitric oxide antiviral activity induced by viruses in mononuclear phagocytes may be inhibited by N-monomethyl-L-arginine acetate (NGMLA) [28]. Monocyte-enriched cultures were treated with NGMLA and simultaneously infected with DENV. DENV-Ag+ cells were present in significantly higher frequency among infected cells in treated cultures after 4 days, as compared to infected and untreated cultures (Figure 5A, 5B). Figure 5A presents a flow cytometry histogram with an increased DENV-Ag+ cell population in presence of NGMLA treatment as compared to untreated infected cultures, with significantly higher percentages of labelled cells (Figure 5B). Thus, the inhibition of iNOS facilitates DENV-1 replication, suggesting that NO production might be involved in the control of virus replication.
Figure 5 Effect of an iNOS inhibitor and of a NO donor on DENV-Ag expression after cell infection. Figures 5A and 5B – Effect of iNOS inhibitor – NGMLA on monocyte infection. Adherent PBMLs were incubated with either Mock C6/36 cell supernatant or DENV inoculum for 4 days in presence or absence of 400 μM NGMLA. Figure 4A shows histogram with profile of infected monocytes after NGMLA treatment and labelling controls. Figure 4B shows percentages of DENV-Ag+ in gated monocytes obtained by flow cytometry. Figures 5C and 5D -. Effect of NO donor – SNP, on C6/36 cell infection. C6/36 cells were incubated with DENV for 2 days in absence or presence of 10 or 100 μM SNP. Figure 4C shows a histogram with profiles of infected C6/36 cells obtained by FACS from a culture treated with 10 μM SNP or controls. The histogram gates have been made where ≥99% than cells in the cell control are excluded. Figure 4Dshows percentages of DENV-Ag+ cells calculated from ungated cells after labelling with antibody to DENV-Complex and anti-mouse IgG-PE. In histograms x-axis represent mean of fluorescence intensity and y-axis represent percentage of maximum cell number counts. Cells were labelled with antibody to DENV-Complex followed by anti-mouse IgG-PE. Average ± SEM are represented from samples set in triplicates. * P < 0.05 in Student's T-test showing difference in DENV infection after treatment. Data are obtained from one representative experiment out of two for each cell type using two different monocyte donors.
Dengue virus has limited rates of infection in monocytes and after a few days infection is controlled accompanied by the induction of several factors that may act as inhibitors of viral replication [21,29]. Since C6/36 cells are extremely susceptible to DENV infection and allow intense virus replication, these mosquito cells should be suitable to test the effect of a potential antiviral molecule. Therefore, C6/36 cultures were infected with DENV and simultaneously treated with sodium nitroprussiate (SNP), a known NO donor, which was added at concentrations of 10 or 100 μM. After 2 days of infection and treatment, DENV-Ag+ cells were detected using FACS. An uninfected cell control was used in order to detect unspecific labelling. SNP induced a significant inhibition of DENV-Ag expression in infected C/36 cells in both tested concentrations, when percentages of positive cells were compared to those in untreated infected cells (analysis by Student's T test; Figures 5C and 5D). Therefore, inhibition of DENV-1 replication by NO production is evident during the present experiments.
Discussion
DENV was detected for the first time by flow cytometry in peripheral monocytes of infected patients. Earlier reports demonstrated viral particles in monocytes, lymphocytes or endothelial cells from tissues and blood clot of patients [9,30]. Infectious DENV [3] or PCR detection of nucleic acid [7] in PBMLs were even at higher incidence than in serum. We did not detect viral antigens in lymphocytes despite others having done so [9,30]. This may be due to the viral strain used or alternatively the genetic constitution of the human PBML donors. Monocytes are important targets for DENV either in vitro [31] or in vivo [8]. Supporting our previous report [26], we present here that DENV likely replicates in monocytes since detected antigens increase with time and no antigen is detected when inactivated virus is used. This indicates that viruses probably replicate for a short period and, during cell infection, mechanisms are activated that are able to control further replication.
Interferon (IFN)-mediated antiviral action is still considered to be among the most important intracellular mechanisms against viruses, however recently a mechanism in which viral proteins might be involved in blocking IFN signalling during dengue virus infections has been described [32] supporting the conjecture that other complementary antiviral activities may be acting during dengue infection as well [33].
We have exhibited here that DENV can be detected in peripheral monocytes of patients with acute Dengue Fever, and that these cells are activated as demonstrated by the expression of iNOS studied ex vivo. Equivalent data was obtained with in vitro infected monocytes from healthy individuals, and virus presence can be modulated by inhibiting iNOS or generating NO in cell culture, indicating that this molecule is likely to have a role in controlling DENV-1 genome expression within monocytes.
During acute Dengue Fever, iNOS expression in monocytes differed among patients probably due to individual variations. Inducible NOS is encoded by the NOS2A gene, haplotype 2 is associated with self-limiting hepatitis C infections [34] and NO production was associated with less severe forms of dengue [19].
Viral replication inhibition by NO has been considered with even increasing importance. The ability of NO to inhibit virus replication was first described in 1993 [33,35,36] and thereafter many other studies have been developed. While some viruses were reported to be susceptible to NO action, others were resistant [35]. The mechanisms by which NO induces viral modulation are diverse, such as inactivation of viral cysteine protease by NO-dependent S-nitrosylation [36] or inhibition of Epstein-Barr virus reactivation. The inhibition of virus replication by NO was found to be mediated by IFN-γ [33,37] and dependent upon the induction of the signal transducer and activator of transcription (STAT) STAT1 phosphorylation [38]. Antiviral NO-mediated effect may also be activated by IFN-α, since it has been observed that mononuclear leukocytes activated in vitro by IFN-α or from Hepatitis C virus infected patients undergoing IFN-α therapy produce NO [39]. Some flaviviruses had their NO susceptibility tested and an effective inhibitory mechanism was determined for Japanese encephalitis virus [40], but studies on Tick-borne encephalitis virus (TBE-V) revealed no antiviral effects by NO [35]. The fact that dengue patients may have monocytes activated by circulating IFN-γ [12-14] is consistent with induction of iNOS during acute disease as reported here for both in vivo and in vitro infection.
NO may play a role in severity of viral hemorrhagic fevers(Sanchez et al., 2004) probably when produced in high concentrations, by affecting vascular tone and contributing to virus-induced shock. As previously mentioned, NO was detected in sera from patients with dengue [19] and correlated with the less severe form of the disease. Hence, NO may be exerting both protective and pathological action, depending upon the concentration produced [41]. Since our panel consisted of few patients, a correlation was not possible; iNOS+ cell rates were increased in some patients from both groups presenting mild or severe disease. Espina et al. [21] were not able to detect NO production by monocytes in culture. This apparent contradiction with our data of induced NOS may be explained by the use of a diferent detection technique: Griess reagent detects nitrite in cell culture supernatant which may be below method sensitivity since the density of monocytes in culture is not high. We did not succeed in detecting NO by this method in human monocyte cultures, and we only detected NO in murine macrophages after proper stimulation at high cell culture density [20]. Moreover, detection of NO produced in vitro by infected Kupffer cells [20] was carried out by a fluorimetric assay, much more sensitive than the Griess Reaction.
Only few reports show human mononuclear cells from healthy individuals generating NO production following treatment with cytokines, pathogens, and/or pathogen-derived products [39,42]. Until a few years ago expression of iNOS in humans was questioned [43] or difficult to detect [20,44]. It may be possible that iNOS induction and expression dependent on synergistic signals to macrophage/monocyte [45,46], and cytokines and virus products may be acting as different stimuli.
Replicating dengue viruses may be necessary for optimal iNOS induction, since rates of iNOS+ cells were much higher than when the inactivated virus was used. However, Kupffer cells infected with DENV induced NO and iNOS, in spite of the absence of viral progeny. Non-living products, such as peptides [47], originating from either other viruses [39,48] or parasites [49,50] can stimulate iNOS in macrophages/ monocytes and induce NO production.
Our data demonstrate that, when experimentally infecting their targets (e.g., monocytes), DENV-1 is able to induce a progressive expression of the enzyme iNOS, indicating that NO might be produced during in vitro infection as it is produced in vivo. At 6 days of infection, the iNOS detection is increased as compared to that of 3 days and is coincident with the DENV decrease, favouring the hypothesis that iNOS+ cells were promoting an inhibitory effect on viruses either directly by NO production or by other virus-induced mechanisms during this monocyte activation, such as IFN-α production [39]. Alternatively cells interacting with viral antigens are probably becoming activated and dying either by cell lysis or by virus induced apoptosis [21]. We also observed that surviving cells remained in culture for more than10 days : either viruses are eliminated or cells become refractory by interferon-related mechanisms, in contrast with what is observed, e.g., for mosquito C6/36 cells, which are mostly destroyed by the virus. Here we observed that there is an enhancement of DENV-1 Ags detection, when infected monocytes were cultured with the iNOS inhibitor NGMLA, indicating that DENV-1 is very likely susceptible to NO inhibitory activity. This result is confirmed by the observation that DENV-1 detection is strongly inhibited in C6/36 cultures by the presence of the NO donor, SNP. These findings are similar and supported by observations in other viral models [51].
With regard to DENV replication, further studies are warranted in order to elucidate which pathways are involved in NO-mediated inhibition of DENV-1. Despite C6/36 remain viable by Trypan blue or propide iodine exclusions, it has to be considered that the NO action in virus targeted cells may involve apoptosis. Furthermore, it is not known whether or not other strains and serotypes will activate iNOS and be sensitive to NO action. There may be relevance in studying various strains from different serotypes and evaluate if susceptibility and resistance are strain and/or serotype specific.
Conclusion
The data presented here displays for the first time the presence of iNOS in monocytes from patients with acute dengue fever. Despite the fact that, in excess, NO may provide a deleterious effect on vascular permeability, evidence exhibited here favour a role for NO in the control of in vitro dengue infection within monocytes. This may hold true for natural human infection as well.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PCFNS and ELA participated equally in the study design, in the experimental work and in the preparation of the manuscript.
PCFNS carried out all the in vitro infections and assays and flow cytometry analysis.
ELA carried out all the work with patient blood collecting, storing cells and plasma and performing flow cytometry assays and analysis.
SMOZ established the connection with the Health Units in Niterói, examined the patients and registered the clinical data.
RVS examined the patients and registered the clinical data.
SRNIR helped with patient blood collecting, storing cells and plasma.
DISC helped to perform flow cytometry assays.
RMRN was responsible for the confirmatory diagnosis of dengue in Rio de Janeiro.
CFK oversaw the study and participated in its design and coordination, performed flow cytometry and statistical analysis and prepared the initial and revised drafts of the manuscript and figures.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Oswaldo Cruz Foundation (FIOCRUZ), Brazilian National Council for Research (CNPq) and Rio de Janeiro State Research Support Foundation (FAPERJ) financially supported this work. P.C.F Neves-Souza and D.I.S. Cerqueira were fellows from CNPq and E L Azeredo was a pre-doctoral fellow from FIOCRUZ. We acknowledge the Dr. E Pozzolo and Dr. Gilbert (Secretaria Municipal de Saúde de Foz do Iguaçu) for their support during the fieldwork and confirmatory diagnostics in Paraná, Brazil in 2000. We thank Dr. Marize Miagostovich and Ms. Eliana Saraiva for their assistance during laboratorial diagnosis in our Departament, Dr. Marcelo Pelajo-Machado from Departmento de Patologia, IOC for the confocal microscopy and the technical support of Ms. Farid O F Von Sydow, as graduate student from Instituto de Microbiologia, UFRJ, and Maryrose Lavatori and Mariana de Sousa Lopes and Mr. Mitchell R Lishon for the language revision.
==== Refs
Halstead SB Pathogenesis of dengue: challenges to molecular biology Science 1988 239 476 481 3277268
Monath TP Schlesinger SSMJ Pathobiology of the Flaviviruses he Togaviridae and Flaviviridae 1986 New York , Plenun Press 375 440
Scott RM Nisalak A Cheamudon U Seridhoranakul S Nimmannitya S Isolation of dengue viruses from peripheral blood leukocytes of patients with hemorrhagic fever J Infect Dis 1980 141 1 6 7365271
Tassaneetrithep B Burgess TH Granelli-Piperno A Trumpfheller C Finke J Sun W Eller MA Pattanapanyasat K Sarasombath S Birx DL Steinman RM Schlesinger S Marovich MA DC-SIGN (CD209) Mediates Dengue Virus Infection of Human Dendritic Cells J Exp Med 2003 197 823 829 12682107 10.1084/jem.20021840
Wu SJ Grouard-Vogel G Sun W Mascola JR Brachtel E Putvatana R Louder MK Filgueira L Marovich MA Wong HK Blauvelt A Murphy GS Robb ML Innes BL Birx DL Hayes CG Frankel SS Human skin Langerhans cells are targets of dengue virus infection Nat Med 2000 6 816 820 10888933 10.1038/77553
Morens DM Marchette NJ Chu MC Halstead SB Growth of dengue type 2 virus isolates in human peripheral blood leukocytes correlates with severe and mild dengue disease. Am J Trop Med Hyg 1991 45 644 651 1951875
Wang HL Lin KH Yueh YY Chow L Wu Y Chen HY Sheu MM Chen WJ Efficient diagnosis of dengue infections using patients' peripheral blood leukocytes and serum/plasma Intervirology 2000 43 107 111 10971129 10.1159/000025032
Azeredo EL Reis SRIN Zagne SMO Miagostovich M Nogueira RM Kubelka CF Detection of dengue virus in peripheral mononuclear cells from patients using flow cytometry: Mangaratiba, Rio de Janeiro, Brazil. 2003
Jessie K Fong MY Devi S Lam SK Wong KT Localization of dengue virus in naturally infected human tissues, by immunohistochemistry and in situ hybridization J Infect Dis 2004 189 1411 1418 15073678 10.1086/383043
Rothman AL Ennis FA Immunopathogenesis of Dengue hemorrhagic fever Virology 1999 257 1 6 10208914 10.1006/viro.1999.9656
Guidotti LG Chisari FV Noncytolytic control of viral infections by the innate and adaptive immune response Annu Rev Immunol 2001 19 65 91 11244031 10.1146/annurev.immunol.19.1.65
Azeredo EL Zagne SM Santiago MA Gouvea AS Santana AA Neves-Souza PC Nogueira RM Miagostovich MP Kubelka CF Characterisation of lymphocyte response and cytokine patterns in patients with dengue fever Immunobiology 2001 204 494 507 11776403 10.1078/0171-2985-00058
Green S Vaughn DW Kalayanarooj S Nimmannitya S Suntayakorn S Nisalak A Lew R Innis BL Kurane I Rothman AL Ennis FA Early immune activation in acute dengue illness is related to development of plasma leakage and disease severity J Infect Dis 1999 179 755 762 10068569 10.1086/314680
Kurane I Innis BL Nisalak A Hoke C Nimmannitya S Meager A Ennis FA Human T cell responses to dengue virus antigens. Proliferative responses and interferon gamma production. J Clin Invest 1989 83 506 513 2521489
Kurane I Ennis FA Production of interferon alpha by dengue virus-infected human monocytes J Gen Virol 1988 69 ( Pt 2) 445 449 3339332
Akaike TMH Nitric oxide and virus infection Immunology 2000 101 300 308 11106932 10.1046/j.1365-2567.2000.00142.x
Chen RF Yeh WT Yang MY Yang KD A model of real-time correlation of viral titers with immune reactions in antibody-dependent enhancement of Dengue-2 infections FEMS Immunol Med Microbiol 2001 30 1 7 11172984 10.1016/S0928-8244(00)00225-X
Jiang H Stewart CA Leu RW Tumor-derived factor synergizes with IFN-gamma and LPS, IL-2 or TNF-alpha to promote macrophage synthesis of TNF-alpha and TNF receptors for autocrine induction of nitric oxide synthase and enhanced nitric oxide-mediated tumor cytotoxicity Immunobiology 1995 192 321 342 7544321
Valero N Espina LM Anez G Torres E Mosquera JA Short report: increased level of serum nitric oxide in patients with dengue Am J Trop Med Hyg 2002 66 762 764 12224588
Marianneau P Steffan AM Royer C Drouet MT Jaeck D Kirn A Deubel V Infection of primary cultures of human Kupffer cells by Dengue virus: no viral progeny synthesis, but cytokine production is evident J Virol 1999 73 5201 5206 10233989
Espina LM Valero NJ Hernandez JM Mosquera JA Increased apoptosis and expression of tumor necrosis factor-alpha caused by infection of cultured human monocytes with dengue virus Am J Trop Med Hyg 2003 68 48 53 12556148
Reed LJ Muench H A simple method of estimating fifty per cent endpoints Am J Hyg 1938 27 493 497
Miagostovich MP Nogueira RM Cavalcanti SM Marzochi KB Schatzmayr HG Dengue epidemic in the state of Rio de Janeiro, Brazil: virological and epidemiological aspects Rev Inst Med Trop Sao Paulo 1993 35 149 154 8284599
Schoepp RJ Beaty BJ Titration of dengue viruses by immunofluorescence in microtiter plates. J Clin Microbiol 1984 20 1017 1019 6392318
Akarid K Sinet M Desforges B Gougerot-Pocidalo MA Inhibitory effect of nitric oxide on the replication of a murine retrovirus in vitro and in vivo J Virol 1995 69 7001 7005 7474119
Sydow FF Santiago MA Neves-Souza PC Cerqueira DI Gouvea AS Lavatori MF Bertho AL Kubelka CF Comparison of dengue infection in human mononuclear leukocytes with mosquito C6/36 and mammalian Vero cells using flow cytometry to detect virus antigen. Mem Inst Oswaldo Cruz 2000 95 483 489 10904403 10.1590/S0074-02762000000400007
Cologna R Rico-Hesse R American genotype structures decrease dengue virus output from human monocytes and dendritic cells J Virol 2003 77 3929 3938 12634353 10.1128/JVI.77.7.3929-3938.2003
Benencia F Courreges MC Nitric oxide and macrophage antiviral extrinsic activity Immunology 1999 98 363 370 10583595 10.1046/j.1365-2567.1999.00864.x
Chen YC Wang SY Activation of terminally differentiated human monocytes/macrophages by dengue virus: productive infection, hierarchical production of innate cytokines and chemokines, and the synergistic effect of lipopolysaccharide J Virol 2002 76 9877 9887 12208965 10.1128/JVI.76.19.9877-9887.2002
Lin YW Wang KJ Lei HY Lin YS Yeh TM Liu HS Liu CC Chen SH Virus replication and cytokine production in dengue virus-infected human B lymphocytes J Virol 2002 76 12242 12249 12414963 10.1128/JVI.76.23.12242-12249.2002
Halstead SB Porterfield JS O'Rourke EJ Enhancement of dengue virus infection in monocytes by flavivirus antisera Am J Trop Med Hyg 1980 29 638 642 6157332
Munoz-Jordan JL Sanchez-Burgos GG Laurent-Rolle M Garcia-Sastre A Inhibition of interferon signaling by dengue virus Proc Natl Acad Sci U S A 2003 100 14333 14338 14612562 10.1073/pnas.2335168100
Karupiah G Xie QW Buller RM Nathan C Duarte C MacMicking JD Inhibition of viral replication by interferon-gamma-induced nitric oxide synthase Science 1993 261 1445 1448 7690156
Yee LJ Knapp S Burgner D Hennig BJ Frodsham AJ Wright M Thomas HC Hill AV Thursz MR Inducible nitric oxide synthase gene (NOS2A) haplotypes and the outcome of hepatitis C virus infection Genes Immun 2004 5 183 187 14985675 10.1038/sj.gene.6364054
Kreil TR Eibl MM Nitric oxide and viral infection: NO antiviral activity against a flavivirus in vitro, and evidence for contribution to pathogenesis in experimental infection in vivo Virology 1996 219 304 306 8623546 10.1006/viro.1996.0252
Saura M Zaragoza C McMillan A Quick RA Rohenadl C Lowenstein JM Lowenstein CJ Antiviral mechanims of nitric oxide: inhibition of viral proteinase. Immunity 1999 10 21 28 10023767 10.1016/S1074-7613(00)80003-5
MacLean A Wei XQ Huang FP Al-Alem AH Chan WL Liew FY Mice lacking inducible nitric-oxide synthase are more susceptible to herpes simplex virus infection despite enhanced Th1 responses J Gen Virol 1998 79 825 830 9568978
Chesler DA Dodard C Lee GY Levy DE Reiss CS Interferon-gamma-induced inhibition of neuronal vesicular stomatitis virus infection is STAT1 dependent. J Neurovirol 2004 10 57 63 14982729 10.1080/13550280490261707
Sharara AI Perkins DJ Misukonis MA Chan SU Dominitz JA Weinberg JB Interferon (IFN)-alpha activation of human blood mononuclear cells in vitro and in vivo for nitric oxide synthase (NOS) type 2 mRNA and protein expression: possible relationship of induced NOS2 to the anti-hepatitis C effects of IFN-alpha in vivo J Exp Med 1997 186 1495 1502 9348307 10.1084/jem.186.9.1495
Lin YL Huang YL Ma SH Yeh CT Chiou SY Chen LK Liao CL Inhibition of Japanese encephalitis virus infection by nitric oxide: antiviral effect of nitric oxide on RNA virus replication J Virol 1997 71 5227 5235 9188590
Wolkow PP Involvement and dual effects of nitric oxide in septic shock Inflamm Res 1998 47 152 166 9628258 10.1007/s000110050309
Weinberg JB Misukonis MA Shami PJ Mason SN Sauls DL Dittman WA Wood ER Smith GK McDonald B Bachus KE Human mononuclear phagocyte inducible nitric oxide synthase (iNOS): analysis of iNOS mRNA, iNOS protein, biopterin, and nitric oxide production by blood monocytes and peritoneal macrophages Blood 1995 86 1184 1195 7542498
Denis M Human monocytes/macrophages: NO or no NO? J Leukoc Biol 1994 55 682 684 8182347
Gantt KR Goldman TL McCormick ML Miller MA Jeronimo SM Nascimento ET Britigan BE Wilson ME Oxidative responses of human and murine macrophages during phagocytosis of Leishmania chagasi J Immunol 2001 167 893 901 11441096
Jaramillo M Naccache PH Olivier M Monosodium urate crystals synergize with IFN-gamma to generate macrophage nitric oxide: involvement of extracellular signal-regulated kinase 1/2 and NF-kappa B J Immunol 2004 172 5734 5742 15100320
Plum J Huang C Grabensee B Schror K Meyer-Kirchrath J Prostacyclin enhances the expression of LPS/INF-gamma-induced nitric oxide synthase in human monocytes Nephron 2002 91 391 398 12119468 10.1159/000064278
Perez-Mediavilla LA Lopez-Zabalza MJ Calonge M Montuenga L Lopez-Moratalla N Santiago E Inducible nitric oxide synthase in human lymphomononuclear cells activated by synthetic peptides derived from extracellular matrix proteins FEBS Lett 1995 357 121 124 7528687 10.1016/0014-5793(94)01322-R
Imanishi N Andoh T Sakai S Satoh M Katada Y Ueda K Terasawa K Ochiai H Induction of Inducible Nitric Oxide (NO) Synthase mRNA and NO Production in Macrophages Infected with Influenza A/PR/8 Virus and Stimulated with Its Ether-Split Product Microbiol Immunol 2005 49 41 48 15665452
Camargo MM Andrade AC Almeida IC Travassos LR Gazzinelli RT Glycoconjugates isolated from Trypanosoma cruzi but not from Leishmania species membranes trigger nitric oxide synthesis as well as microbicidal activity in IFN-gamma-primed macrophages J Immunol 1997 159 6131 6139 9550414
Keller CC Kremsner PG Hittner JB Misukonis MA Weinberg JB Perkins DJ Elevated nitric oxide production in children with malarial anemia: hemozoin-induced nitric oxide synthase type 2 transcripts and nitric oxide in blood mononuclear cells Infect Immun 2004 72 4868 4873 15271950 10.1128/IAI.72.8.4868-4873.2004
Rimmelzwaan GF Baars MMJW Lijster P Fouchier RAM Osterhaus ADME Inhibition of influenza virus replication by nitric oxide J Virol 1999 73 8880 8883 10482647
|
16109165
|
PMC1208887
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Aug 18; 5:64
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-64
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-661611783210.1186/1471-2334-5-66Research ArticleDoes age acquired immunity confer selective protection to common serotypes of Campylobacter jejuni? Miller Gordon [email protected] Geoff M [email protected] Thomas MS [email protected] Iain D [email protected] Norval JC [email protected] School of Biological Sciences, University of Aberdeen, Cruickshank Building, Aberdeen, UK2 School of Engineering and Physical Sciences, Department of Engineering, University of Aberdeen, Fraser Nobel Building, Aberdeen, UK3 Microbiology Department, Aberdeen Royal Infirmary, Aberdeen, UK4 Department of Medical Microbiology, University of Aberdeen, Foresterhill, Aberdeen, UK2005 23 8 2005 5 66 66 9 4 2005 23 8 2005 Copyright © 2005 Miller et al; licensee BioMed Central Ltd.2005Miller et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Campylobacter infection is a major cause of bacterial gastrointestinal disease. Exposure to Campylobacter is known to produce an immune response in humans that can prevent future symptomatic infections. Further, studies of the general population have shown that seroprevalence to Campylobacter increases with age.
Methods
A large collection of serotyped Campylobacter isolates, obtained from human clinical faecal samples, were analysed by comparing the ratio of uncommon to common serotypes by different age groups, using χ2 tests.
Results
We have identified that older age groups, as well as having generally lower incidence, are significantly less likely to be infected by the more common serotypes.
Conclusion
These results are indicative of acquired immunity, however, further studies are needed to rule out the confounding effects of the variations in exposure pathways experienced by different age groups.
==== Body
Background
Campylobacter is the greatest cause of human bacterial gastroenteritis in the developed world [1], with more than 50,000 cases reported in the UK alone each year [2]. However, due to under-reporting, it is believed that the true number of cases is around eight times this [2], suggesting humans are infected on average at least once during their lifetime. Symptoms of campylobacteriosis include fever, abdominal cramp, and bloody diarrhoea, which can last for approximately seven days after infection, with one in a 1000 cases leading to the more serious neurological disease, Guillain-Barré syndrome [3].
Approximately 90% of Campylobacter infections are C. jejuni, while the remainder are predominantly C. coli [4]. A variety of typing techniques have been used to subtype these species, including phenotypic Penner heat stable antigen serotyping, genotypic PFGE, AFLP and MLST, which have demonstrated high variability in the genome.
Human seroprevalence studies of C. jejuni specific antigens have shown that IgG, IgM, and IgA antibodies are produced during infection [5,6]. The IgM and IgA antibody levels quickly decrease after infection, however, the IgG antibodies can remain elevated for months or years afterwards [5]. A recent study of seroprevalence in Danish adults looked at IgG antibodies in a sample of 1112 people from Copenhagen [7]. It was found that the percentage of people with C.jejuni-specific IgG antibodies increased with age, from 20.6% in the 15–34 years age group to 32.4% in the 50 – 69 years age group.
Humans are likely to be exposed to Campylobacter a number of times during their lifetime, resulting in immune responses that can last several years [8]. Further, as the types/strains of Campylobacter are diverse, it is probable that people are more likely to develop immune protection to the more common types to which they are exposed. Hence we hypothesise that the general population develops increased immunity to the more common types of Campylobacter as they grow older. Here we test this hypothesis by analysing a large dataset of Penner serotyped cases of human campylobacteriosis, stratified by age, to determine if the prevalence of common serotypes decreases with the age of the population.
Methods
Data were collated from the Grampian NHS trust reporting human infections of C. jejuni in Grampian, Scotland (approximately 2300 cases during 1997 – 1999). These data included each patient's age, gender and date of infection. Isolate serotyping had been performed on these cases, using the Penner method [9]. Cases with no recorded age (105), as well as untyped and C. coli cases (168), were omitted from the data set, leaving 1955 typed cases of C. jejuni.
The C. jejuni cases were separated into 48 Penner serotypes, which were then collected together into recognised groupings, such as HS:1,44, HS:6,7, and HS:4 complex (comprising serotypes HS:4, 13, 16, 43, 50) [10]. The prevalence of each serotype was determined, with serotypes accounting for more than 10% of the cases being defined as common, and those accounting for less than 10% being classed as uncommon. These data were stratified by age (five year age groups), then ratios of uncommon to common types were determined. The incidence (cases per 100,000) of each age group was calculated, using the population data from the 2001 Scottish Census [11]. Finally χ2 tests were used to test the statistical significance of differences between age group ratios.
Results
The overall incidence of Campylobacter infection by five-year age groups (Figure 1A) shows a large number of cases in the 0–4 years age group (241 cases per 100,000). There is a drop off, before a second peak found in the 25–29 years age group (205 cases per 100,000). From 30 years of age onwards there is a gradual reduction in cases down to the minimum of less than 60 cases per 100,000 among the patients 75 years and older.
Figure 1 C. jejuni incidence and ratio of uncommon to common serotypes. Human campylobacteriosis in Grampian Scotland (1997–1999). a) Incidence of reported cases and b) Ratio of uncommon to common serotypes.
The most prevalent Penner serotypes found in these data were HS:4 complex, HS:2, and HS:1,44 (427, 420, and 205 cases respectively) which account for a total of 53.8% of all tested cases (Table 1). These three type groups were classified as common, for comparison against the 40 other uncommon serotypes.
Table 1 C. jejuni Serotypes Number of reported cases of C. jejuni for Grampian, Scotland (1997–1999) by Penner serotypes, including separation into groups less than and greater than 40 years of age.
Penner Serotype Total (%) Age
<40 (%) >40 (%)
4 complexa 427 (21.8) 286 (23.8) 141 (18.7)
2 420 (21.5) 285 (23.8) 135 (17.9)
1,44 205 (10.5) 130 (10.8) 75 (9.9)
11 128 (6.5) 55 (4.6) 73 (9.7)
6,7 113 (5.8) 67 (5.6) 46 (6.1)
8 45 (2.3) 18 (1.5) 27 (3.6)
23 44 (2.3) 25 (2.1) 19 (2.5)
21 27 (1.4) 16 (1.3) 11 (4.5)
5 26 (1.3) 17 (1.4) 9 (1.2)
9 23 (1.2) 14 (1.2) 9 (1.2)
37 23 (1.2) 10 (0.8) 13 (1.7)
Otherb 474 (24.2) 277 (23.1) 197 (26.1)
Total 1955 1200 755
aSerotypes 4, 13, 16, 43, 50
b Types accounting for less than 1% of total.
The ratios of uncommon to common types by five-year age groups (Figure 1B) show that for ages 0 – 39 there is a consistently greater number of infections from the common types (HS:4 complex, HS:2, HS:1,44), with an uncommon/common mean ratio of 0.71 (± 0.10). For patients aged 40 years and older, all age groups but one (45–49 years) have a higher incidence of uncommon types (mean: 1.20 ± 0.18). A χ2 test confirms that the difference is highly statistically significant (p < 0.001).
Discussion
In this study we compiled the Penner serotype data for nearly 2000 cases of human Campylobacter jejuni infections in Grampian, Scotland, determined which were the common types, and identified that these types are significantly more prevalent in younger age groups (less than approximately 40 years). The reduced frequency of common types in older age groups, along with a general reduction in incidence of infection, provides evidence for the hypothesis that people develop immunity to the most common types of C. jejuni during their lifetimes.
From the age-specific incidence data (Figure 1A) it is clear that the highest incidence occurs in the 0–4 years age group, which agrees with previous studies [12]. This could be due to many factors including lack of immunity, higher rates of reporting in the age group, or a higher risk of exposure to a particular infection pathway. For example, in rural regions such as Grampian, Scotland, where farm animals shed high loads into the environment, small children may have a high likelihood of coming into contact with, and ingesting, the bacteria. The reduction of incidence in the 5–9 and again in the 10–14 age groups is similar to previous studies [13], which may be due to improved hygiene reducing the risk of environmental exposure. This trend then reverses with a sharp increase in incidence in the older teens/early twenties. This occurs around the age when people are more likely to be travelling and hence coming into contact with new strains. Case-control studies have found that international and national travel is an important risk factor for infection [14]. In humans >30 years, a gradual reduction in incidence is observed. This, taken together with the results of the Danish study [7] showing that seroprevalence increases with age, and the fact that it has been demonstrated that elevated levels of Campylobacter jejuni immunoglobulin A reduces the risk of Campylobacter diarrhoea in adults traveling to Thailand [15], suggests that immunity may be causing protection from Campylobacter infection, or at least symptomatic infection. However, the lower incidence could also be caused by reduced exposure to the infection pathways more common in younger age groups.
The common serotypes found in these data (Table 1) are consistent with those found in previous studies [16,17], which are also the same types found in poultry and farm animals [17]. Although we do not have the epidemiological data to ascertain whether the human cases can be attributed to animals/poultry it is clear that the majority of cases in all age groups below 40 are from these common serotypes (Figure 1B).
Conclusion
These data presented here demonstrate lower incidence along with the reduced occurrence of infection from the common serotypes with age, supporting the hypothesis of increased immunoprotection in the population with age. However, it is possible that changing exposure to Campyolobacter with age would confound our results, which can only be clarified by performing a comprehensive epidemiological age-stratified exposure assessment.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GM collected the data, performed the analysis, and drafted the manuscript. GMD participated in the design of the study and helped draft the manuscript. TMSR carried out the serotyping and helped collect the data. IDO participated in the design of the study and helped in the interpretation of the data. NJCS conceived of the study, coordinated it, and assisted with the statistical analysis.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Blaser MJ Epidemiologic and clinical features of Campylobacter jejuni infections J Infect Dis 1997 176 S103 S105 9396691
Department for Environment Food and Rural Affairs Zoonoses report: United Kingdom 2002
Pasternack MS Impact and management of Campylobacter in human medicine – US perspective Int J Infect Dis 2002 6 S37 S43 10.1016/S1201-9712(02)90182-7
Gillespie IA O'Brien SJ Frost JA Adak GK Horby P Swan AV A case-case comparison of Campylobacter coli and Campylobacter jejuni infection: a tool for generating hypotheses Emerg Infect Dis 2002 8 937 942 12194770
Strid MA Engberg J Larsen LB Begtrup K Mølbak K Krogfelt KA Antibody responses to Campylobacter infections determined by an enzyme-linked immunosorbent assay: 2-year follow up study of 210 patients Clin Diagn Lab Immunol 2001 8 314 319 11238214 10.1128/CDLI.8.2.314-319.2001
Cawthraw SA Lind l Kaijser B Newell DG Antibodies, direct towards Campylobacter Jejuni antigens, in sera from poultry abattoir workers Clin Exp Immunol 2000 122 55 60 11012618 10.1046/j.1365-2249.2000.01349.x
Linneberg A Andersen LP Madsen F Dirksen A IgG antibodies against micoorganisms and atopic disease in Danish adults: the Copenhagen allergy study J Allergy Clin Immunol 2003 11 847 853 12704368 10.1067/mai.2003.1335
Cawthraw SA Feldman RA Sayers AR Newell DG Long-term antibody responses following human infection with Campylobacter jejuni Clin Exp Immunol 2002 130 101 106 12296859 10.1046/j.1365-2249.2002.01966.x
Penner JL Hennessy JN Passive hemagglutination technique for serotyping Campylobacter fetus subsp. jejuni on the basis of soluble heat-stable antigens J Clin Microbiol 1980 12 732 737 6796598
Owen RJ Slater E Telford D Donovan T M Barnham Subtypes of Campylobacter jejuni from sporadic cases of diarrhoeal disease at different locations in England are highly diverse Eur J Epidemiol 1997 13 837 840 9384275 10.1023/A:1007497005152
General Register Office for Scotland. Census in Scotland 2001
Sibbald CJ Sharp JCM Campylobacter infection in urban and rural populations in Scotland J Hyg (Lond) 1985 95 87 93 3839516
Public Health Laboratory Service Campylobacter sentinel surveillance scheme Commun Dis Rep CDR Wkly 2001 11
The Campylobacter Sentinel Surveillance Scheme Collaborators Foreign and domestic travel and the risk of Campylobacter infection: results from a population based sentinel surveillance scheme J Travel Med 2003 10 136 139 12650661
Walz SE Baqar S Beecham HJ Echeverria P Lebron C McCarthy M Pre-exposure anti- Campylobacter jejuni immunoglobulin A levels associated with reduced risk of Campylobacter diarrhea in adults traveling to Thailand Am J Trop Med Hyg 2001 65 652 656 11716132
Wareing DRA Bolton FJ Fox AJ Wright PA Greenway DLA Phenotypic diversity of Campylobacter isolates from sporadic cases of human enteritis in the UK J Appl Microbiol 2002 92 502 509 11872126 10.1046/j.1365-2672.2002.01552.x
Nielsen EM Engberg J Madsen M Distribution of serotypes of Campylobacter jejuni and C. coli from Danish patients, poultry, cattle and swine FEMS Immunol Med Microbiol 1997 19 47 56 9322068 10.1016/S0928-8244(97)00049-7
|
16117832
|
PMC1208888
|
CC BY
|
2021-01-04 16:28:15
|
no
|
BMC Infect Dis. 2005 Aug 23; 5:66
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-66
|
oa_comm
|
==== Front
BMC Med ImagingBMC Medical Imaging1471-2342BioMed Central London 1471-2342-5-51612238310.1186/1471-2342-5-5Research ArticleNoise correlation in PET, CT, SPECT and PET/CT data evaluated using autocorrelation function: a phantom study on data, reconstructed using FBP and OSEM Razifar Pasha [email protected]öm Mattias [email protected] Harald [email protected]ångström Bengt [email protected] Enn [email protected] Ewert [email protected]öm Mats [email protected] Uppsala University, Centre for Image Analysis, Lägerhyddsv. 3, SE-752 37 Uppsala, Sweden2 Uppsala Imanet AB, Box 967, SE-751 09 Uppsala, Sweden3 Uppsala University Hospital, Department of Hospital Physics, SE-751 85 Uppsala, Sweden4 Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden2005 25 8 2005 5 5 5 26 3 2005 25 8 2005 Copyright © 2005 Razifar et al; licensee BioMed Central Ltd.2005Razifar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Positron Emission Tomography (PET), Computed Tomography (CT), PET/CT and Single Photon Emission Tomography (SPECT) are non-invasive imaging tools used for creating two dimensional (2D) cross section images of three dimensional (3D) objects. PET and SPECT have the potential of providing functional or biochemical information by measuring distribution and kinetics of radiolabelled molecules, whereas CT visualizes X-ray density in tissues in the body. PET/CT provides fused images representing both functional and anatomical information with better precision in localization than PET alone.
Images generated by these types of techniques are generally noisy, thereby impairing the imaging potential and affecting the precision in quantitative values derived from the images. It is crucial to explore and understand the properties of noise in these imaging techniques. Here we used autocorrelation function (ACF) specifically to describe noise correlation and its non-isotropic behaviour in experimentally generated images of PET, CT, PET/CT and SPECT.
Methods
Experiments were performed using phantoms with different shapes. In PET and PET/CT studies, data were acquired in 2D acquisition mode and reconstructed by both analytical filter back projection (FBP) and iterative, ordered subsets expectation maximisation (OSEM) methods. In the PET/CT studies, different magnitudes of X-ray dose in the transmission were employed by using different mA settings for the X-ray tube. In the CT studies, data were acquired using different slice thickness with and without applied dose reduction function and the images were reconstructed by FBP. SPECT studies were performed in 2D, reconstructed using FBP and OSEM, using post 3D filtering. ACF images were generated from the primary images, and profiles across the ACF images were used to describe the noise correlation in different directions. The variance of noise across the images was visualised as images and with profiles across these images.
Results
The most important finding was that the pattern of noise correlation is rotation symmetric or isotropic, independent of object shape in PET and PET/CT images reconstructed using the iterative method. This is, however, not the case in FBP images when the shape of phantom is not circular. Also CT images reconstructed using FBP show the same non-isotropic pattern independent of slice thickness and utilization of care dose function. SPECT images show an isotropic correlation of the noise independent of object shape or applied reconstruction algorithm. Noise in PET/CT images was identical independent of the applied X-ray dose in the transmission part (CT), indicating that the noise from transmission with the applied doses does not propagate into the PET images showing that the noise from the emission part is dominant. The results indicate that in human studies it is possible to utilize a low dose in transmission part while maintaining the noise behaviour and the quality of the images.
Conclusion
The combined effect of noise correlation for asymmetric objects and a varying noise variance across the image field significantly complicates the interpretation of the images when statistical methods are used, such as with statistical estimates of precision in average values, use of statistical parametric mapping methods and principal component analysis. Hence it is recommended that iterative reconstruction methods are used for such applications. However, it is possible to calculate the noise analytically in images reconstructed by FBP, while it is not possible to do the same calculation in images reconstructed by iterative methods. Therefore for performing statistical methods of analysis which depend on knowing the noise, FBP would be preferred.
==== Body
Background
Computed tomography (CT) is a technique based on measurement of X-ray transmission through the object to provide visible thin slices through any section in a human; in other words, it is a technique for creating two dimensional (2D) cross section images of three dimensional (3D) objects [1]. Positron Emission Tomography (PET) and Single Photon Emission Tomography (SPECT) are built on the concept of CT, but use tracing of molecules labelled with positron and gamma ray emitting radionuclides, respectively, to illustrate metabolic and physiological activities in certain organs and tissues. PET/CT combines two state-of-the-art imaging modalities: PET and CT. PET provides high sensitivity functional information of lesions in the body while CT provides detailed information about the location, size, and shape of these lesions, but cannot differentiate pathological lesions from normal structures with the same sensitivity as PET [2-5]. The combined PET/CT scanner has proved to increase the diagnostic value compared to each modality used separately [6].
In PET/CT and SPECT/CT, the CT data are, except for imaging and anatomical delineation, used as transmission data for performing attenuation correction during the reconstruction. PET, SPECT and PET/CT data are usually reconstructed either analytically by filtered back projection (FBP) or iteratively by for instance, ordered subsets expectation maximisation (OSEM), in SPECT followed by 3D filtering. OSEM needs less computation time compared with earlier versions of iterative reconstructions such as maximum likelihood expectation maximisation (ML-EM) [7]. CT data are reconstructed using FBP with corrections for cone beam geometry and filtered using standard filters related to the imaging task.
In PET and PET/CT the convolution kernel applied on projections in an FBP algorithm is a combination of a ramp filter to cope with blurriness of the image after back projecting the projections [8,9] and a low-pass filter e.g. Hanning filter to damp the high frequency behaviour of a ramp filter. FBP is a relatively fast process but it has the drawback that the images are noisier and more sensitive to disturbing factors e.g. patient movements during the scan and in PET studies, between transmission and emission scans, leading sometimes to degradation of image quality.
In SPECT it has been shown that OSEM gives fewer artefacts compared to FBP [10,11]. It has also been shown that changing the detector head orbit from circular to elliptical may improve the isotropy of recovered resolution [12].
One limiting factor with respect to the visualisation of discrete changes, e.g. as induced in pathological lesions is noise in the images. This noise primarily comes from the inherent random variations in the counting of photons and is related to the number of photons detected and used for the generation of images. In CT a high flux of photons is used, giving rise to noise in the images which is about 0.1%, whereas PET and SPECT with fewer photons have noise levels more typically of about 10%.
The main sources of noise in PET images are in decreasing order of magnitude: emission, transmission and blank scans [13]. Detectors, electronics and recorder systems together may add to the noise [14,15]. The choice of reconstruction algorithm and type of convolution kernel used in the reconstruction algorithm significantly affect the magnitude and texture of noise. Other factors which contribute to noise features include mode of attenuation correction and other types of corrections e.g. for variation in detector uniformity, random coincidences, scattered radiation and compensation for radioactive decay as discussed by Alpert et al [16].
Wilson [17] has shown that the magnitude (variance) of noise spreads from regions with higher signal magnitude towards regions with lower signal in FBP reconstruction, but not in iterative reconstruction. Riddell et al [18] has shown that FBP provides a uniform and intensity independent noise distribution over the whole reconstructed image with very low variation within the image even when signal intensity varied significantly from one region to another. In contrast OSEM provides an intensity dependent noise within the image with the noise magnitude being higher in regions with higher intensity compared to the regions with lower intensity. OSEM provides better signal-to-noise ratios (SNR) in regions with high and low intensity compared to FBP yet more dramatic improvement in SNR in regions with low intensity.
Studies have shown that in images reconstructed using OSEM and a low number of iterations, the noise is correlated at shorter distances compared to images reconstructed by FBP. In another study has been shown that images reconstructed using FBP, correlation of each image element influences 1 or 2, pixels of the nearest neighbors in PET [19].
CT has an advantage compared to the two nuclear techniques in its low structural noise in the images. This capacity of low noise and efficient dose utilisation has enabled this technique to visualize low contrast objects [9]. Modern CT equipment has the ability to apply a care dose function, with an automated reduction of the dose for non-circular patient cross-sections based on a reduction of the intensity of X-ray tube current at the angular positions at which the patient diameter is smallest. This procedure is performed online with dose regulation during the scanning with preservation of the image quality and noise magnitude [20-23].
Although different aspects of noise have been covered extensively in the literature, we still feel that one aspect has not been adequately covered: the angular dependence of noise correlation in cases when the investigated object is asymmetrical. With asymmetric objects, the count rates will be different in the different acquisition angles and the relative magnitude of the noise will therefore be different. It is possible that this angular-dependent noise would in the image reconstruction propagate to the images and there generate a noise correlation which is non-isotropic.
In our previous study we explored the pattern of noise correlation in experimentally generated PET images acquired in 2D and 3D mode and reconstructed using FBP and OSEM, with emphasis on the angular dependence of correlation, and evaluated using the autocorrelation function (ACF) [24]. In our present work we extend these observations to other tomographic imaging modalities by applying autocorrelation function on PET, CT, PET/CT and SPECT images reconstructed using FBP and OSEM.
Methods
The PET experiments were performed on an ECAT Exact HR+ PET camera (CTI/Siemens, Knoxville, Tennessee) [25]. This unit contains 32 detector rings separated by removable lead septa, and is capable of performing 2D and 3D data acquisition with an axial field of view (FOV) of 15.5 cm and spatial resolution of 5.1–5.4 mm. The system generates 63, 2D images with a matrix of 128 × 128 pixels. The CT experiments were performed on a Siemens Somatom Sensation 16 CT scanner (Siemens, Erlangen, Germany). This scanner acquires 16 slices/rotation with a rotation time of 0.5 s. Images were reconstructed into 512 × 512 matrices; maximum image FOV is 50 cm. The SPECT measurements were made with Millennium VG dual-headed gamma camera with 5/8 "NaI (TI) detectors and a Hawkeye X-ray tomography for attenuation correction (AC) and anatomical information (General Electric Medical Systems, Haifa, Israel). The image matrix was 128 × 128 pixels.
PET/CT experiments were performed employing Discovery ST (D-ST) (GE medical Systems). The scanner combines a helical multi-slice High Speed Ultra 16 slice, CT scanner and a multi-ring BGO block detector PET tomography, which are arranged in 24 rings, spaced by 3.27 mm covering an axial FOV of 157 mm with different spatial resolution depending on the distance from the centre of FOV. The scanner creates 47 128 × 128 fused images [26].
In the PET study, the radionuclides 11C and 68Ga with 20.3 min and 67.6 min half-lives, respectively, were used. 11C was produced using a Scanditronix MC-17 cyclotron (Scanditronix AB, Uppsala, Sweden). 68Ga was obtained from a 68Ge generator [27] In the SPECT studies the radionuclide 99 mTC with 6.02 h was used. 99 mTc was eluted from a 99Mo/99 mTc generator from Mallinckrodt.
All experiments with each one of the modalities were performed on either a 20-cm-long elliptical torso phantom with 30-cm-diameter long axis and 20-cm short axis or a cylindrical water-filled Nema phantom with 20-cm diameter and 20-cm length [28].
Prior to each PET experiment, a 60-min blank scan was performed with rotating 68Ge/68Ga rod sources without any phantom in the gantry. Then, a 10-min transmission scan was performed where the respective phantom without radioactivity was placed at the centre of the FOV of the scanner. The phantoms were filled with 80 MBq of either 11C or 68Ga and 30-min emission scans in 2D mode were made. Finally, to avoid artefacts in the images caused by movement of the object between transmission and emission scan, a 10-min post-injection ('hot') transmission scan was performed. The SPECT emission acquisitions were made with 60 × 50-s views on phantoms that were filled with 50 MBq of 99 mTc. The used energy window was 140 keV ± 10%. The transmission used for attenuation correction was made with a Hawkeye CT put on half rotation, 140 kVp and 3.0 mA.
In the CT studies the phantoms were filled with water and the tomography was set to create both two- and three-mm-thick slices. Different numbers of slices were obtained in each scan depending on the desired thickness of slices. Each experiment was additionally performed applying the care dose function. In PET/CT, the study was performed on phantoms filled with 60 MBq 68Ga in the Nema and 80 MBq 68Ga in the Torso phantom. The scans were started by transmission (CT) scans using 140 kVp, 10 mA, 30 mA and 100 mA setting of the X-ray tube followed by 12 × 30-min emissions.
The PET images reconstructed using the initial transmission scan for attenuation correction were not used because of slight observed positioning errors when replacing the phantom after filling it with radioactivity. Instead a weighted segmentation technique described elsewhere [29], as included in the ECAT 7.2 software (CTI, Knoxville, Tennesse), was applied to the hot transmission data prior to its use for attenuation correction.
In PET and PET/CT, both FBP and attenuation-weighted OSEM as included in the scanner software were used for reconstructing the images. Different types of filters can be used with each of the reconstruction methods. A Gausssian filter with size of 6 mm (FWHM) was used in this study for reconstruction of images using FBP and OSEM. Also same number of subsets and iterations in OSEM was used for reconstruction of data in PET and PET/CT studies. PET/CT images were reconstructed with three different combinations of CT transmission data (10, 30 and 100 mA) and emission data. All CT images were reconstructed using FBP with corrections for cone beam geometry and filtered using standard abdomen filter, as included in the scanner software. All SPECT images were reconstructed using both FBP and OSEM algorithms, as included in the Entegra workstation, using the default settings for filters and number of iterations in OSEM (Hanning filter with a cut-off of 0.85 and 8 subsets and 4 iterations). Iterative Reconstruction Attenuation Corrected with applied 3D post filtering (IRACF) was used as the nomenclature for iteratively reconstructed SPECT data in this study.
A programme was developed to calculate the ACF of the reconstructed PET images, performed both in frequency and spatial domain. The spatial equation is based on 2D cross-correlation of the matrix:
ai,,j with image resolution (image size) of i × j with itself using the lags k and l
where k and l refer to lags of the function and m and n refer to the number of steps needed for performing the masking and
max(1,1 - k) ≤ i ≤ min(m,m - k)
and
max(1,1 - l) ≤ j ≤ min(n,n - l)
To avoid disturbing effects at the edges of the FOV an arbitrary central slice within each frame was used. Subsequently, an arbitrary chosen matrix with a size of 25 × 25 from the central part of the slice was used as a mask when applying the ACF. After subtraction of the average over this matrix, the ACF image was generated showing the correlation of the noise in the image. The resulting image was then normalized by dividing each pixel value by the maximum pixel value within the generated ACF image. The results from this procedure applied to images from all experiments were studied and compared.
The aim of the ACF application was to describe the noise correlation between the pixels within each image. A specific aim was to analyse the form and the shape of the 2D autocorrelation function in the images in different imaging modalities. The programme results in images, which can be used for the visualization and comparison of images using different techniques obtained with different reconstruction algorithms and for plotting 1D vertical and horizontal profiles through the images.
Another programme was developed to calculate the variance matrix as indicator of the noise magnitude distribution across the image plane. The method compares pixel values through several adjacent slices within a frame and calculates
where n is the total number of slices used, i scans through pixels in-between slices and j defines a position within the two-dimensional image matrices. E.g., in PET, j spans from 1 to 128 × 128. The aim of this application was to study the magnitude of noise in relation to position within the images produced by different modalities. The results are illustrated and studied as 2D variance images and 1D horizontal profiles across the images.
Results
PET studies
In the study on the NEMA phantom (Figure 1), the results indicate an identical and isotropic ACF with a similar pattern of noise texture independent of applied reconstruction methodology and used filter (6 mm Gaussian and 4 mm Hanning produced identical results). In the torso phantom (Figure 2), however, the results of the ACF indicate a non-isotropic correlation of the noise in images reconstructed using FBP, independent of used filter (6 mm Gaussian and 4 mm Hanning). On the other hand, the images reconstructed using OSEM (Figure 3), show an identical and isotropic form with a similar pattern of noise texture independent of used filter.
Figure 1 Results of the cylindrical NEMA phantom study. Acquired image (upper left) reconstructed using FBP with applied 6 mm (FWHM) Gaussian filter, corresponding ACF image (upper right) and vertical (dash point) and horizontal profiles (dash star) through the ACF image.
Figure 2 Results of the elliptical torso phantom study. Image (upper left) reconstructed using FBP with applied 6 mm (FWHM) Gaussian filter, corresponding ACF image (upper right) and vertical (dash point) and horizontal profiles (dash star) through the ACF image.
Figure 3 Results of the elliptical torso phantom study. Image (upper left) reconstructed using OSEM with applied 6 mm (FWHM) Gaussian filter, corresponding ACF image (upper right) and vertical (dash point) and horizontal profiles (dash star) through the ACF image.
CT Studies
In CT images of the circular Nema phantom, the ACF shows the expected isotropic behaviour. When the elliptic torso phantom was scanned, a slightly non-isotropic behaviour was observed as indicated by the ACF (Figures 4, 5). This non-isotropy is only slightly affected by the use of the care dose. Identical results were obtained with 2- and 3-mm slice thickness.
Figure 4 Results of the cylindrical Torso phantom in CT study. FBP image with thickness of 2 mm scanned with constant mA setting (upper left) respective with applied care dose (upper right). ACF images from the normal scan (lower left) respective with applied care dose (lower right).
Figure 5 1D vertical and horizontal profiles through the ACF image based on the normal scan (upper) respective with applied care dose (lower).
SPECT studies
In the studies of noise autocorrelation in SPECT images, the images from the cylindrical phantom showed close to but not entirely isotropic behavior.
Images from the elliptic phantom showed an ACF which was close to isotropic (Figure 6). The same was true when the images were reconstructed with the IRACF iterative reconstruction.
Figure 6 Results of SPECT study on elliptical Torso phantom. Image reconstructed using FBP followed by 3D filtering (upper left) and corresponding ACF (upper right) followed by 1D vertical and horizontal profiles of the ACF (down).
PET/CT studies
The CT images (Figure 7) and the PET images (Figure 8) from the cylindrical phantom using PET/CT showed an expected isotropic behavior, which in the case of PET was independent of reconstruction method. When scanning and reconstructing images from the elliptic phantom; however, both the CT images (Figure 9) and PET images (Figure 10) showed a slightly non-isotropic behavior. When the PET image was reconstructed using OSEM, the noise correlation became more isotropic (Figure 11). There was an identical noise correlation pattern in the PET images reconstructed with different CT doses.
Figure 7 Results of PET/CT study on the cylindrical Nema phantom. The CT image reconstructed using FBP (upper left) and corresponding ACF (upper right) followed by 1D vertical and horizontal profiles of the ACF (down).
Figure 8 Results of PET/CT study on the cylindrical Nema phantom. The PET image reconstructed using FBP with applied 4-mm Hanning filter (upper left) and corresponding ACF (upper right) followed by 1D vertical and horizontal profiles of the ACF (down).
Figure 9 Results of PET/CT study on the elliptical Torso phantom. The CT image reconstructed using FBP (upper left) and corresponding ACF (upper right) followed by 1D vertical and horizontal profiles of the ACF (down).
Figure 10 Results of PET/CT study on elliptical Torso phantom. PET image reconstructed using FBP with applied 4-mm Hanning filter (upper left) and corresponding ACF (upper right) followed by 1D vertical and horizontal profiles of the ACF (down).
Figure 11 Results of PET/CT study on elliptical Torso phantom. PET image reconstructed using OSEM with applied 4-mm Hanning filter (upper left) and corresponding ACF (upper right). 1D vertical and horizontal profiles of the ACF (down).
Variance images
Variance images were generated from PET, SPECT, CT and PET/CT images reconstructed using FBP and OSEM for further illustration of differences in noise behavior. The results obtained from images using the same type of reconstruction method showed similar features, independent of the used imaging modality. In PET and PET/CT reconstructed with FBP the noise variance gradually decreased from inside to outside of the phantom, whereas with OSEM, the noise decreased abruptly at the border of the object (Figures Figure 12, 13).
Figure 12 PET results of PET/CT study on cylindrical NEMA phantom. Variance images reconstructed using FBP (upper left) and OSEM (upper right). 1D horizontal profile through the variance image reconstructed using FBP (lower left) and reconstructed using OSEM (lower right).
Figure 13 Results of PET/CT study on elliptical Torso phantom. PET variance images reconstructed using FBP (upper left) and OSEM (upper right). 1D horizontal profile through the variance image reconstructed using FBP (lower left) and reconstructed using OSEM (lower right).
In the SPECT images, the noise magnitude in the FBP images decreased more abruptly at the border of the object, thereby resembling images reconstructed by IRACF (Figures Figure 14, 15).
Figure 14 Results of the SPECT study on the cylindrical NEMA phantom. Variance image reconstructed using FBP (upper left) and IRACF (upper right). 1D horizontal profile through the variance image reconstructed using FBP (lower left) and reconstructed using IRACF (lower right).
Figure 15 Results of the SPECT study on elliptical Torso phantom. Variance image reconstructed using FBP (upper left) and IRACF (upper right). 1D horizontal profile through the variance image reconstructed using FBP (lower left) and reconstructed using IRACF (lower right).
The variance across the CT image field showed a broad maximum at the centre of the object and gradually decreased towards the periphery. The same features were observed in the cylindrical and the elliptic phantom. The variance at the centre was about three times higher than that at the periphery. Similar behavior was observed independently of slice thickness and application of care dose (Figure 16). Figure 17 visualizes the noise distribution over the image field for the CT in PET/CT. The noise was significantly higher in the centre of the cylindrical phantom, similar to that observed in the previous study with CT. With the cylindrical phantom the variation along the horizontal axis was less than along the vertical axis. Almost identical results were obtained using different magnitude of X-ray dose.
Figure 16 Variance across the CT image of the cylindrical (left) and elliptic (right) phantom. Horizontal profiles over the corresponding variance images are shown in the graphs.
Figure 17 Results of the PET/CT study on cylindrical NEMA phantom. Variance images of CT (transmission scan) of Nema (upper left) and Torso phantom (upper right). Corresponding 1D horizontal profile through the variance images of Nema (lower left) and Torso phantom (lower right).
Sinogram data in PET and SPECT
To understand the noise contribution, profiles were generated across the sinograms from PET studies with the two phantoms. The counts recorded from the circular phantom dipped slightly in the central part (Figure 18). For the elliptic phantom this central dip was more accentuated in the plot over the short axis representation but the profile was rather flat in its central portion for the long axis representation.
Figure 18 Sinograms from PET study of Nema (left) and Torso (right) phantom illustrating detector counts for the different projections (upper images). Profiles through the end and central parts of the sinograms, corresponding to horizontal vs. diagonal projections during acquisition (lower plots).
The corresponding plots for SPECT showed a different behavior, with higher counts centrally and rather similar count values for the two projections in the elliptic phantom (Figure 19).
Figure 19 Projections obtained at two perpendicular angles at the acquisition with SPECT for the Torso phantom. Profiles through the two different projections, corresponding to horizontal vs. diagonal direction in SPECT acquisition (down). The position shift is related to the fact that the phantom was not centred at the rotation axis.
Discussion
The aim of the present study was to explore and compare the properties of noise, notably its correlation, in images from CT, PET, PET/CT and SPECT. One of the main focuses was to illustrate the differences in noise correlation between images reconstructed using different types of reconstruction algorithms such as FBP and OSEM, generated utilizing different imaging modalities. We believed that despite their similarity as tomographic imaging devices, the fact that the acquisition of detector data and modes of reconstruction differ could lead to differences in the expression of noise in the images. CT is based on measurement of transmitted X-ray photons from X-ray tube through the object and to the detectors. PET is based on the simultaneous measurement of two annihilation photons that emerge from the body and hit each of two detectors on opposite sides of the object. SPECT is based on recording single photons that emerge from the body and acquisition of a large number of projections by sequential rotation of the detector system around the body. For these techniques the measured data, typically organised in projections are subjected to different corrections, where especially in PET and SPECT the attenuation correction is dominant. After performing the different corrections the projection data are utilised in a reconstruction algorithm for the generation of images. The reconstruction part can be similar for the three devices with the predominant methods being FBP or iterative reconstruction.
Since noise in the images is a factor which may impair the visualisation of discrete signals and the generation of quantitative values, it is important to understand its features. To illustrate some aspects of noise we developed a programme to generate auto-correlation images which predominantly indicate the noise correlation and the angular dependence of noise correlation. Images depicting the variance across the image planes were created to indicate the spatial dependence of noise. Some of the characters of noise were explained by studying the profiles of counts in different directions in the sinogram domain in PET and SPECT.
Applying the auto-correlation function on the reconstructed PET, PET/CT and SPECT data revealed a clear correlation: the noise was affected by the adjacent 2–3 pixels corresponding to 4.2–6.3 mm in PET, 4.7–7.0 mm in PET/CT and 8.9–13.4 mm in SPECT. This correlation was similar for all devices and all reconstruction algorithms. As expected, an isotropic pattern of correlation of the noise in the images was obtained in the circular Nema phantom. In CT scans from both regular CT and PET/CT, the same behaviour was observed. The radial dependence of noise correlation is highly dependent upon reconstruction algorithm, and ensuing spatial resolution which especially with OSEM in a complicated manner is depending on number of iterations and shape and distribution of the object. Hence in the present work, the width of the autocorrelation function per see can not be generalised, but the concept that the autocorrelation function may have a non-isotropic shape.
Hence we observed that the noise correlation pattern obtained from the torso phantom with elliptical shape was clearly non-isotropic when examined with PET or PET/CT and dissimilar between images using the two reconstruction methods. When studies were performed on this non-circular phantom, ACF images from PET, CT and PET/CT studies reconstructed using FBP showed an asymmetric correlation texture of the noise. This non-isotropic behaviour can be explained in three different ways; the first explanation is that the relative noise in sinogram data in the study on Torso phantom differs in different directions. The sinogram data indicate that the count rates are higher in the direction of the shortest diameter compared to the angle with the longest diameter. This is because along the 30 cm axis the detectors see a 50 % higher amount of radioactivity. However, this radioactivity is in turn subjected to an attenuating path that is 10 cm longer, leading to an overall 43% lower count rate. Since noise in the projection data is Poisson distributed, the noise standard deviation is angular dependent and 25% higher along the long axis of the phantom. The second explanation is related to the attenuation correction whereby the projection data with its noise is multiplied by attenuation correction factors which are different for different projections. The third contribution is related to the way noise is handled in the FBP algorithm when the noise magnitude differs in the different projections. The noise modification and correlation induced by the convolution filter in the reconstruction is the same in all projections. The back projection then distributes noise with different magnitude in different angular directions.
The correlation pattern of the noise in images from PET, CT and PET/CT studies become close to symmetric when the data are reconstructed using OSEM. This behaviour depends on how the iterative reconstruction algorithm handles the noise with an inherent attempt to iterate to similar deviations for each angular projection, which then tends to equalise the noise magnitude for the different projections.
In PET/CT studies the results from fused images were identical independent of the applied dose of X-ray in transmission part (CT). This observation indicates that the noise from the transmission at these doses does not propagate into the fused images because the noise from the emission part is dominant. The result suggests that in human studies it might be possible to use a low dose in the CT transmission part while maintaining the noise behaviour and the quality of the images. This observation supports the outcomes of the study done by Kamel et al [30] evaluating the effect of lowering the CT tube current in human PET/CT studies and suggesting that a lower X-ray dose to the patient could be used without impairment of image quality.
In SPECT studies, the correlation pattern was still close to isotropic in images obtained from an elliptical object and independent of the reconstruction methodology used. This behaviour could be because the number of counts in different directions in the study on either circular or elliptical phantom is almost the same. This similarity is caused by the way radioactivity is sensed, with a detector which records the superficial activity and not that deep inside the object. Since attenuation is a dominant factor with respect to recording of radioactivity, the number of counts will become relatively similar for a uniform elliptic phantom. On the other hand, a focal radioactivity not centred in the phantom will instead give highly variable counts in different detectors. Additionally the detector sensitivity is significantly affected by distance from the detector. This could explain a slightly non-isotropic effect observed in the circular phantom that was not perfectly centred.
ACF images from CT studies reconstructed by FBP on data from the Torso phantom showed a non-isotropic form. The correlation pattern was broader along the horizontal axis compared with the vertical. This non-isotropic behaviour depends on how data is acquired in CT. The X-ray tube rotating around the object emits X-rays, which are detected by detectors that are placed as a block on the opposite side of the tube. The number of photons detected is highly dependent upon the thickness of the object, and the relative noise is hence larger in the horizontal direction of the object. These differences in noise magnitude are, as indicated above, handled by the FBP algorithm such that a non-isotropic noise correlation is given in the images.
The results from studies on CT images with applied care dose function gave slightly different noise correlation compared to those without care dose. Yet the noise correlation was not identical in the vertical and horizontal direction, suggesting that the care dose could not fully equalise the relative noise in different directions.
The variance across the PET images shows a significantly broader distribution with FBP than with OSEM. With OSEM, the variance reduces abruptly at the border of the object. The variance is significantly larger centrally in the object than at the periphery.
Conclusion
The combined effect of noise correlation for asymmetric objects and a varying noise variance across the image field significantly complicates the interpretation of the images when statistical methods are used, such as with statistical estimates of precision in average values, use of statistical parametric mapping methods and principal component analysis. Hence it is recommended that iterative reconstruction methods are used for such applications. However, it is possible to calculate the noise analytically in images reconstructed by FBP, while it is not possible to do the same calculation in images reconstructed by iterative methods. Therefore for performing statistical methods of analysis which depend on knowing the noise, FBP would be preferred.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Authors PR and MB designed the study. They created the method for applying ACF, performed the image and data analysis and drafted the paper.
Author MS helped perform the SPECT studies, acquire and reconstruct the data and write this paper.
Authors HS, EB, EM and BL helped with some practical approaches and the writing of this paper.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to thank the staff at Uppsala Imanet AB, especially Mr. Lars Lindsjö for his assistance with the measurements, and the staff of the chemistry department for the radionuclide production. Our thanks to Uppsala University Hospital, Departments of Hospital Physics and Radiology for their assistance and help. The authors thank Dr. Lennart Norel, Dr. Sergio Estrada and Dr. Jan Axelsson for beneficial scientific discussions.
==== Refs
Swindell W Barrett HH Computerized tomography: taking sectional x-rays Phys Today 1977 30 32 41 Reprinted in Optics today, J. Howard, ed. (American Institute of Physics, New York, 1986)
Townsend DW Beyer T Blodgett TM PET/CT scanners: a hardware approach to image fusion Semin Nucl Med 2003 33 193 204 12931321 10.1053/snuc.2003.127314
Beyer T Townsend DW Brun T Kinahan PE Charron M Roddy R Jerin J Young J Byars L Nutt R A combined PET/CT scanner for clinical oncology J Nucl Med 2000 41 1369 1379 10945530
Beyer T Townsend DW Blodgett TM Dual-modality PET/CT tomography for clinical oncology Q J Nucl Med 2002 46 24 34 12072843
Messa C Bettinardi V Picchio M Pelosi E Landoni C Gianolli L Gilardi MC Fazio F PET/CT in diagnostic oncology Q J Nucl Med Mol Imaging 2004 48 66 75 15243404
Hutton BF Braun M Thurfjell L Lau DY Image registration: an essential tool for nuclear medicine Eur J Nucl Med Mol Imaging 2002 29 559 577 11914898 10.1007/s00259-001-0700-6
Hudson HM Larkin RS Accelerated image reconstruction using Ordered Subsets of Projection data IEEE Transactions on medical imaging 1994 13 601 609 10.1109/42.363108
Brooks RA Chiro GD Principles of computer assisted tomography (CAT) in radiographic and radioisotopic imaging Phys Med Biol 1976 21 689 732 788005 10.1088/0031-9155/21/5/001
Faulkner K Moores BM Analysis of x-ray computed tomography images using the noise power spectrum and autocorrelation function Phys Med Biol 1984 29 1343 1352 6505016 10.1088/0031-9155/29/11/003
Hatton RL Hutton BF Angelides S Choong KK Larcos G Improved tolerance to missing data in myocardial perfusion SPET using OSEM reconstruction Eur J Nucl Med Mol Imaging 2004 31 857 861 14758510 10.1007/s00259-003-1450-4
Leong LK Kruger RL O'Connor MK A comparison of the uniformity requirements for SPECT image reconstruction using FBP and OSEM techniques J Nucl Med Technol 2001 29 79 83 11376099
Pan TS Lou DS Kohli V King MA Influence of OSEM, elliptical orbits and background activity on SPECT 3D resolution recovery Phys Med Biol 1997 42 2517 29 9434304 10.1088/0031-9155/42/12/015
Holm S Toft P Jensen M Estimation of the noise contributions from Blank, Transmission and Emission scans in PET IEEE trans Nucl Sci 1996 43 2285 2291 Part 1 10.1109/23.531893
Metz CE Beck RN Quantitative effects of stationary linear image processing on noise and resolution of structure in radio-nuclide images J Nucl Med 1974 15 164 4813223
Tsui BMW Effects of the Recorder System on Spatial Resolution and Noise in the Nuclear Medicine Dissertation 1977 The University of Chicago, USA
Alpert NM Chesler DA Correia JA Ackerman RH Chang JY Finklestein S Davis SM Brownell GL Taveras JM Estimation of the Local Statistical Noise in Emission Computed Tomography IEEE Trans Med Img, MI-1 1982 2 142 146
Wilson DW Tsui BMW Noise properties of Filtered_Backprojection and ML-EM reconstructed Emission Tomographic Images IEEE trans Nucl Sci 1993 40
Riddell C Carson RE Carrasquillo JA Libutti SK Danforth DN Whatley M Bacharach SL Noise Reduction in Oncology FDG PET Images by Iterative Reconstruction: A Quantitative Assessment J Nucl Med 2001 42 1316 1323 11535719
Bergström M Performance Evaluation and Improvements of Quantitation Accuracy in Transmission and Positron Emission Computer Assisted Tomography PhD thesis 1982 Department of Radiation Physics, University of Stockholm, Sweden
Greess H Wolf H Baum U Lell M Pirkl M Kalender WA Bautz WA Dose reduction in computed tomography by attenuation-based on-line modulation of tube current: evaluation of six anatomical regions Eur Radiol 2000 10 391 394 10663775 10.1007/s003300050062
Gies M Kalender WA Wolf H Suess C Dose reduction in CT by anatomically adapted tube current modulation. I Simulation studies Med Phys 1999 26 2235 2247 10587204 10.1118/1.598779
Greess H Nomayr A Wolf H Baum U Lell M Bowing B Kalender WA Bautz WA Dose reduction in CT examination of children by an attenuation-based on-line modulation of tube current (CARE Dose) Eur Radiol 2002 12 1571 1576 12042970 10.1007/s00330-001-1255-4
Kalender WA Wolf H Suess C Gies M Greess H Bautz WA Dose reduction in CT by on-line tube current control: principles and validation on phantoms and cadavers Eur Radiol 1999 9 323 328 10101657 10.1007/s003300050674
Razifar P Lubberink M Schneider H Langstrom B Bengtsson E Bergstrom M Non-isotropic noise correlation in PET data reconstructed by FBP but not by OSEM demonstrated using auto-correlation function BMC Medical Imaging 2005 5 3 15892891 10.1186/1471-2342-5-3
Brix G Zaers J Adam LE Bellemann ME Ostertag H Trojan H Haberkorn U Doll J Oberorfer F Lorenz WJ Performance evaluation of a whole-body PET scanner using the NEMA protocol J Nucl Med 1997 38 1614 1623 9379202
Bettinardi V Danna M Savi A Lecchi M Castiglioni I Gilardi MC Bammer H Lucignani G Fazio F Performance evaluation of the new whole-body PET/CT scanner: Discovery ST Eur J Nucl Med Mol Imaging 2004 31 867 81 14770270 10.1007/s00259-003-1444-2
Knapp FFJ Brihaye C Callahan AP Wagner HN, Szabo Z, Buchanan JW Generators, Principles of Nuclear Medicine 1995 WB Saunders: Philadelphia 150 165
NEMA NU-2001 Performance standards of positron emission tomographs National Electronics Manufacturers Association
XU M Cutler PD Luk WK Adaptive segmented attenuation correction for whole-body PET imaging IEEE Trans Nucl Sci 1996 43 331 337 10.1109/23.485974
Kamel E Hany TF Burger C Treyer V Lonn AHR Schulthess GKV Buck A CT vs 68Ge attenuation correction in a combined PET/CT system- evaluation of the effect of lowering the CT tube current Eur J Nucl Med Mol Imaging 2002 29 346 350 12002709 10.1007/s00259-001-0698-9
|
16122383
|
PMC1208889
|
CC BY
|
2021-01-04 16:38:04
|
no
|
BMC Med Imaging. 2005 Aug 25; 5:5
|
utf-8
|
BMC Med Imaging
| 2,005 |
10.1186/1471-2342-5-5
|
oa_comm
|
==== Front
BMC Med Inform Decis MakBMC Medical Informatics and Decision Making1472-6947BioMed Central London 1472-6947-5-251608683710.1186/1472-6947-5-25SoftwareThe tissue micro-array data exchange specification: a web based experience browsing imported data Nohle David G [email protected] Barbara A [email protected] Leona W [email protected] The Mid-Region AIDS and Cancer Specimen Resource (ACSR), Department of Pathology, The Ohio State University, Columbus, OH USA2005 8 8 2005 5 25 25 20 4 2005 8 8 2005 Copyright © 2005 Nohle et al; licensee BioMed Central Ltd.2005Nohle et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The AIDS and Cancer Specimen Resource (ACSR) is an HIV/AIDS tissue bank consortium sponsored by the National Cancer Institute (NCI) Division of Cancer Treatment and Diagnosis (DCTD). The ACSR offers to approved researchers HIV infected biologic samples and uninfected control tissues including tissue cores in micro-arrays (TMA) accompanied by de-identified clinical data. Researchers interested in the type and quality of TMA tissue cores and the associated clinical data need an efficient method for viewing available TMA materials. Because each of the tissue samples within a TMA has separate data including a core tissue digital image and clinical data, an organized, standard approach to producing, navigating and publishing such data is necessary.
The Association for Pathology Informatics (API) extensible mark-up language (XML) TMA data exchange specification (TMA DES) proposed in April 2003 provides a common format for TMA data. Exporting TMA data into the proposed format offers an opportunity to implement the API TMA DES. Using our public BrowseTMA tool, we created a web site that organizes and cross references TMA lists, digital "virtual slide" images, TMA DES export data, linked legends and clinical details for researchers.
Microsoft Excel® and Microsoft Word® are used to convert tabular clinical data and produce an XML file in the TMA DES format. The BrowseTMA tool contains Extensible Stylesheet Language Transformation (XSLT) scripts that convert XML data into Hyper-Text Mark-up Language (HTML) web pages with hyperlinks automatically added to allow rapid navigation.
Results
Block lists, virtual slide images, legends, clinical details and exports have been placed on the ACSR web site for 14 blocks with 1623 cores of 2.0, 1.0 and 0.6 mm sizes. Our virtual microscope can be used to view and annotate these TMA images. Researchers can readily navigate from TMA block lists to TMA legends and to clinical details for a selected tissue core.
Exports for 11 blocks with 3812 cores from three other institutions were processed with the BrowseTMA tool. Fifty common data elements (CDE) from the TMA DES were used and 42 more created for site-specific data. Researchers can download TMA clinical data in the TMA DES format.
Conclusion
Virtual TMAs with clinical data can be viewed on the Internet by interested researchers using the BrowseTMA tool. We have organized our approach to producing, sorting, navigating and publishing TMA information to facilitate such review.
We have converted Excel TMA data into TMA DES XML, and imported it and TMA DES XML from another institution into BrowseTMA to produce web pages that allow us to browse through the merged data. We proposed enhancements to the TMA DES as a result of this experience.
We implemented improvements to the API TMA DES as a result of using exported data from several institutions. A document type definition was written for the API TMA DES (that optionally includes proposed enhancements). Independent validators can be used to check exports against the DTD (with or without the proposed enhancements). Linking tissue core images to readily navigable clinical data greatly improves the value of the TMA.
==== Body
Background
AIDS and Cancer Specimen Resource
The National Cancer Institute's (NCI) Division of Cancer Treatment and Diagnosis (DCTD) founded the AIDS Malignancy Bank in 1994 to collect various types of specimens from AIDS patients. It was renamed the AIDS and Cancer Specimen Bank and eventually the AIDS and Cancer Specimen Resource (ACSR) to reflect expansion of its role to include seeking out specimens for researchers when needed. Investigators can search a national on-line database [1] to determine what types of specimens are available.
The ACSR provides Tissue Micro-Array (TMA) tissue sections with de-identified clinical data to approved researchers. Researchers interested in the type and quality of TMA core tissues and the associated clinical data can view digital images of hematoxylin & eosin (H&E) stained TMA tissue sections directly on the web. Delay in determining whether it will be fruitful to initiate a request for TMAs is eliminated and visual evaluation of the quality of available material by the researcher is facilitated while preserving the TMA block from unnecessary sectioning [2-4].
Benefits of using tissue micro-arrays
A tissue micro-array block (TMA) usually is a paraffin recipient block with inserted tiny cylindrical tissue cores taken from donor tissue blocks [5]. A tissue section cut from a TMA block allows large numbers of tissue samples to be tested and stored using less reagent, less effort and less space. Precious tissue is conserved for more research. See the discussion in Berman, et al. [6].
Clinical data entry
De-identified clinical data is assembled and entered so that it will be available along with the TMA recipient (or mother) block, tissue sections and tissue section virtual images. Excel spreadsheets are used to organize the clinical data and TMA tissue core location assignments. Sequential bank identifiers are assigned by the ACSR database software to replace surgical pathology numbers and hospital identifiers in all communications outside the ACSR. De-identified datasets are used for conversion to TMA DES format once the TMA block has been constructed and entered into database.
TMA block and slide production
H&E stained sections of source block tissue on glass or the source embedded tissue paraffin block itself is marked by the pathologist to identify core sites. TMA blocks are then constructed by extracting a tiny (0.6 mm – 2.0 mm diameter) core from all selected source (or donor) tissue blocks and inserting them into a recipient paraffin block (see Figure 1). The most common arrangement is that there is a matrix (rows and columns) of tissue cores in a recipient block and sections are cut from that block.
Figure 1 Producing a TMA block and slide. A slide from a source block is marked to indicate the area of interest. The marked slide is used to locate the place to punch out a core. The core is extracted from the source block and then inserted into a wax TMA block with cores from other source blocks. TMA slides are then cut from the TMA block.
Making digitized cut and stained tissue core images
One glass slide with a cut section from a TMA block is typically prepared with H&E stain. A virtual microscope is used to scan and capture a digital image of the entire matrix of cores [7]. This TMA image (see Figure 2) will be made available via the Internet when potential investigators want to see whether slides from this block would be useful in their research. Sending example slides for this purpose is thus avoided.
Figure 2 Sample image viewed with Virtual Microscope.
When a TMA slide is eventually disbursed, the URL of the TMA web page (which links the above mentioned image) and a one page printout of the image are provided to the investigator. These are useful for locating cores in the matrix.
Data exchange specification
The April 2003 Association for Pathology Informatics (API) TMA data exchange specification (TMA DES) is a proposal for a common XML format for TMA data [8]. A validator (written in Perl) is available with the specification. Exporting TMA data into this format offers an opportunity to apply the specification and validator. Importing and exporting extramural TMA data was used to validate the portability of data using TMA DES.
We used PubMed to access a list of 408 articles related to the TMA DES article [8]. We reviewed the list for implementation and deployment of working examples. We found one other reported implementation [6]. We describe herein our implementation of the TMA DES in a web based experience.
Cancer Biomedical Informatics Grid
The NCI is relying on the Cancer Biomedical Informatics Grid (caBIG) to expedite research leading to new cancer treatment. Tissue banks will share data on a grid as part of the plan [9]. Investigators must be able to merge results from different sources. TMA tissue section slides from a single TMA block are routinely disbursed to multiple research projects. Standard data formats facilitate the merging of study results and comply with the NCI's emphasis on caGRID conformity.
Technology
HTML, XML, XSLT and DTD technologies are those principally used; each is introduced in a section below with description of some aspects important to this work. In addition:
• Microsoft Word mail merge main documents are used to convert Excel tab delimited text exports to XML.
• Windows Script Host (WSH aka batch) files [10] are used to invoke routines that process each TMA block.
• A small amount of JScript [11] is used to extend XSLT scripts and to invoke the XSLT parser [12].
• The XSLT Parser MSXML2 4.0 is used to process XML and XSLT files.
HTML
Most documents on the web are in Hypertext Mark-up Language, HTML. HTML tag delimiters provide instruction for how a document should be displayed [13,14]. For example, the h1 HTML tag delimiter designates a top level heading that a browser will display as such:
<h1>
TMA Block: TA00-050
</h1>
XML
"XML, the Extensible Mark-up Language, is a W3C-endorsed standard for document mark-up. XML is a metamark-up language for text documents. Data is included in XML documents as strings of text. The data is surrounded by text mark-up that describes the data [15]." The tag delimiters that will be used as mark-up are defined in a particular application and comprise a particular language.
The API specification defines an XML mark-up language for the TMA data representation application so that data from various laboratories can be more readily transmitted, shared or merged [8]. The identifier for a tissue micro-array block, in this case TA00-050, is used as the data text string. The API specified tag delimiter for a block identifier is block_identifier [8]. The marked up data is an XML element:
<block_identifier>
TA00-050
</block_identifier>
When displayed in Microsoft Internet Explorer (as shown in our figures), XML will appear in an indented hierarchy with color-coding to separate mark-up from data.
XML data may be well formed or valid. Well-formed data has matching end tags that are properly nested and follows generic XML rules. It may not necessarily be valid, i.e. follow the rules of a particular language. To assure that data is valid, it should be validated more rigorously using a DTD or checked by other software to ensure that the rules for a particular XML language such as TMA DES are followed.
XSLT
XSL Transformation (XSLT) is one part of the extensible stylesheet language (XSL). An XSLT document, called an XSLT stylesheet, uses these rules to convert an XML document into another text document, perhaps an HTML document or another XML document.
This bit of XML defines a block containing a block_identifier:
<block>
<block_identifier>
TA00-050
</block_identifier>
</block>
This XSLT stylesheet template rule converts an API specified XML block_identifier in the above XML to an HTML top-level heading:
<xsl:template match="block">
<h1>
TMA Block: <xsl:value-of select="block_identifier"/>
</h1>
</xsl:template>
Here is the resulting HTML:
<h1>
TMA Block: TA00-050
</h1>
When displayed in a browser the heading will look like this:
TMA Block: TA00-050
"An XSLT processor is a piece of software that reads an XSLT stylesheet, reads an input XML document, and builds an output document by applying the instructions in the stylesheet to the information in the input document. An XSLT processor can be built into a web browser, just as MSXML is in Internet Explorer 6 [15]."
The MSXML processor can be used as a command line processor as well. The processor named MSXML2 version 4.0 is used in BrowseTMA.
DTD
A document type definition (DTD) is used to specify a valid language of XML elements. The DTD can be used with various validators [16,17] to check that XML documents written in this language are valid.
When an external DTD file is referred to and an internal DTD subset is present in an XML file, the two parts collectively are the DTD. The DTD syntax allows definitions to be placed in a file and included into the internal portion as well. The first definition of a given element or entity encountered is used. Internal DTD definitions are encountered before external ones [18].
Implementation
For each TMA block, a digitized image of a stained TMA section is obtained using the virtual microscope [7].
The BrowseTMA tool was constructed and used to create a web site that organizes and cross references TMA lists, virtual slide images, TMA DES export data, linked legends and clinical details. The TMA data is represented in HTML web pages with automatically added hyperlinks to allow rapid navigation. Figure 3 depicts the modules in BrowseTMA.
Figure 3 Implementation. A DTD governs the allowed data elements in each type (TMA block list, TMA LDE, TMA DES export & style) of XML file. The TMA block list uses BrowseTMA to process each listed block file and style. BrowseTMA produces a single HTML TMA block list file, a TMA block HTML file for each specified block and an HTML TMA style file for each style used.
BrowseTMA constructs a single block list web page that has a table listing the TMA blocks (linked to each TMA HTML block file and each image) [see fileList.html in Additional file 1]. Figure 4 illustrates this process.
Figure 4 Producing a block list. The TMA file list is written in a simple XML language. The language is defined by tmaList.dtd, a document type definition that allows 13 common data elements (ex. tma-file-name, style-id). Referencing this DTD causes validation of the entries indicating which TMA DES export files and blocks to use with which styles. For each TMA file in the TMA file list, BrowseTMA processes the specified TMA blocks and displays a row of information about each. In processing a TMA block, BrowseTMA uses that block's data from the XML export file to display the export file name, block identifier, purpose and description and the TMA file list entry to display the style file name and links to the style and possibly to an image.
BrowseTMA constructs a web page for each TMA block that contains a details table and a legend table [see TA00-050.html in Additional file 1]. Figures 5 and 6 illustrate these processes. The details table has core specific demographic data copied from the database. The data in each cell is linked to that tissue core's place in the legend table. The legend table, a "roadmap" of the slide, has a link in each cell to the corresponding core's row in the details table.
Figure 5 Producing a legend. The style file specifies the layout of the array (or matrix) and the legend cell contents and colors using a simple XML language defined in the addStyles conditional section of tmades.dtd. These 26 definitions (ex. core-location-id, row-label, cell-color) allow specification of the TMA layout. They are supplemented by common and locally defined data elements that are nested under the core data element (ex. core_TISSUE, core_histo-repository_donor-block) which may be used to determine the legend cell color and contents and the details columns. Referencing this DTD causes validation of the entries that specify labelling, row and column grouping of cell identifiers, legend cell contents and colors. Cell identifiers (ex. A1, R1C1, or 1.1) are specified in rows with row and column labels interspersed inside the matrix element. The legend-cell-content element contains a list of clinical data items that should be placed in each cell in the HTML legend table. The legend-cell-colors element contains a clinical data item and a list of cell-color elements. The data from the XML export file corresponding to the specified core data item (core_TISSUE is used here) is compared against the substring and exact-match entries inside the cell-color elements and the corresponding color is used for the legend cell in the HTML legend table when a match is found.
Figure 6 Producing the clinical details table. The style file specifies an ordered list of columns and their headings using a simple XML language defined in the addStyles conditional section of tmades.dtd, definitions that allow any data element nested inside the core CDE in the XML export file (ex. core_histo-repository) including locally defined data elements to be used. Activating addStyles causes validation of the entries that specify each details table column. Entries nested inside one or more data elements inside core are allowed. Each core data entry in the XML export file provides one row of cells in the resulting HTML details table.
The various formats involved are discussed below. Descriptions of the activities involved follow.
DTD for TMA DES language
A DTD governs the allowed common data elements (CDE) in TMA DES export XML files. This DTD is implemented as an external file that is referenced by the TMA DES export XML files and supplemented by internal DTD extensions [18].
CDEs are defined in the external DTD file written at OSU [see tmades.dtd in Additional file 1]. Locally defined data elements (LDE) are in a supplementary DTD file [see myLDEs.dtd in Additional file 1]; this file gets included into the internal DTD section in the TMA DES XML file.
TMA styles language
Auxiliary XML files are used to specify the TMA styles including fields and color-coding for the legend cells and columns for the details table. Data elements are defined and placed in a conditional section in the external DTD file to govern the style language; this file is referenced as the external DTD in the TMA style XML files. A definition is there activating the style CDEs.
A TMA style file is made for each set of TMA blocks that used a similar pattern. TMA5by7Style is an example [see TMA5by7Style.xml in Additional file 1]. The BrowseTMA Style XSLT stylesheet [see BrowseTMAStyle.xsl in Additional file 1] produces an HTML TMA style web page file for each style [see TMA5by7Style.html in Additional file 1].
TMA DES XML data
Data for a variety of TMA blocks from several institutions is used. OSU produced the first exports in the TMA DES format [2]; the Cooperative Prostate Cancer Tissue Resource (CPCTR) has since produced exports in XML [6,19]. Data from Brigham and Women's Hospital (BWH) and the Tissue Array Research Program (TARP) [20] has been converted to XML at OSU. The BrowseTMA Legend Details XSLT stylesheet [see BrowseTMALegendDetails.xsl in Additional file 1] produces a TMA block HTML file for each block.
To produce the TMA DES file for each block, three sets of information are created for each block:
• A text file is entered containing description, tissue, # patients, #cores, Core Size (mm), Autopsy/Surgical/Mixed and ID columns and a row for each block. A block list Word main merge file is merged with the text file to produce the block portion of the TMA DES data for each block.
• The table containing the clinical data in that block's Excel file is exported from the Excel program to a tab delimited text file. It is used for two operations:
A hash Word main merge file is merged with the text file to produce the block_array-hash element contents for that block.
A core Word main merge file is merged with the text file to produce the core elements for that block.
These three sets of information are merged by hand for each block.
TMA block list language
A DTD governs the allowed data elements in a TMA block list XML file [see tmaList.dtd in Additional file 1]. A block list file is used in two ways. The BrowseTMA List XSLT stylesheet [see BrowseTMAList.xsl in Additional file 1] uses a TMA block list XML file (as well as export and style files) to produce single HTML TMA block list file [see fileList.html in Additional file 1]. The BrowseTMA Make WSH batch file [see BrowseTMAMakeBatch.bat in Additional file 1] invokes the BrowseTMA Batch XSLT stylesheet [see BrowseTMABatch.xsl in Additional file 1] which uses a TMA block list XML file [see fileList.xml in Additional file 1] to produce a single batch file [see BrowseTMA.bat in Additional file 1].
Producing TMA exports
Mail merge was used to produce the bulk of an export from an Excel file [2]. Berman, et al. have discussed how to produce an export from an Excel table using a Perl script and hand editing [8].
Validating TMA exports with a DTD
We use both cited validation sites to validate a sample TMA block, TA00-050 [see TA00-050.xml in Additional file 1]. No errors are present; warnings may indicate that certain entities are redefined. This is expected as placeholders are defined in the common (external) DTD file and the placeholders are overridden by the local (internal) portion of the DTD.
A combination of XSLT and JScript is used to make a reorder program, BrowseTMAReorder, which accepts a TMA DES XML file and creates a version with certain data elements (mainly the identifiers) first within their parent data element. If no identifier is present one is created using automatic numbering. This allows validation with an improved version of the DTD [18].
TMA styles
There are a variety of ways to arrange tissue cores within a TMA. The number of rows and columns and the size of the cores, space between cores and between groupings of cores vary. Likewise, there are various ways to designate cells, rows and columns. It is common for a TMA producer to assemble multiple TMA blocks that follow the same pattern.
The way that the matrix of cores is laid out is referred to herein as a TMA style. A simple XML language is used to specify the block layout and the data content and appearance of the legend and details HTML tables. Once specified, a TMA style can be reused for similar TMA blocks. A style file may contain multiple TMA styles.
The TMA style uses a simple XML language specified in the style extensions in tmades.dtd which supplement 33 of the CDEs (those nested under core) from the TMA DES and 31 LDEs from the local DTD with 26 style data elements. Activating these definitions (in an internal DTD subset by redefining the entity addStyles as INCLUDE) enables validation of the entries that specify the style (layout), legend (labelling, spacing, rows and implied columns of core location, cell contents and colors) and details (clinical data column order and labels) tables.
Specifying matrix layout and legend
A legend is a map of the location of cores in a TMA block, slide or image. In our HTML legend tables, each cell that represents a core in the legend has a hyperlink to that core's row in the HTML details table.
In the matrix element in a TMA style, core location (and matrix cell) identifiers (ex. A1, R1C1, or 1.1) are specified in rows with row and column labels interspersed.
In a TMA style, a few clinical data core attributes are specified so as to appear in HTML legend cells. The legend-cell-content element contains a list of clinical data items that should be placed in each cell in the HTML legend table. Any data elements nested under the core element, whether defined locally or in the TMA DES, are allowed.
In the legend-cell-color element in a TMA style, a single clinical data core attribute may be identified so as to govern the background color of HTML legend cells. Any data element nested under the core element, whether defined locally or in the TMA DES is allowed. Multiple cell-color entries may follow. Inside each cell-color entry, a color (ex. Blue or #8080FF) is followed by either a substring or exact-match element containing a string. The data from the XML export file corresponding to the specified core data item is compared against the substring and exact-match string entries inside the cell-color elements and when the first match is found, the corresponding color is used as the background for the legend cell in the HTML legend table.
Specifying the details table contents
Each row in the details table has clinical data associated with a core. Each cell in the row has a hyperlink to the legend cell for this core.
The style file specifies an ordered list of columns (and their headings). Each column list entry could be any data element used inside the core CDE in the XML export file (ex. core_histo-repository) including any of the locally defined LDEs. Referencing this DTD causes validation of the entries that specify which clinical data item to include in each HTML details table column. Entries nested inside one or more CDEs inside core are allowed. Each core entry in the XML export file provides one row of cells in the resulting HTML details table.
Specifying the block list
A TMA publisher should determine which TMA XML export files are to be used and which blocks and styles are to be used together. This information (block and style names and filenames, etc.) should be manually typed in a text file named fileList.xml [see Additional file 1] in the file list format specified by tmaList.dtd [see Additional file 1].
Producing an HTML block list table
For each TMA file in the TMA file list, BrowseTMA processes the specified TMA blocks (or all when none specified) and displays a row of information about each indicated TMA block. Refer to Figure 4. In processing each TMA block, BrowseTMA uses that block's data from the XML export file to display the export file name, block identifier, purpose and description.
Producing an HTML legend table
The style file specifies the layout of the array (or matrix) and the legend cell contents and colors [see tmades.dtd in Additional file 1]. Activating the style definitions in the DTD allows validation of the entries that specify labelling, row and column grouping of cell identifiers, legend cell contents and colors. For each TMA file, BrowseTMA processes the specified TMA blocks (or all when none specified) and creates an HTML file containing a legend table (followed by a clinical details table as described below) for each indicated TMA block. Refer to Figure 5. In processing each TMA block, BrowseTMA uses that block's data from the XML export file to display the export file name, block identifier, purpose and description as the header and specified attributes in each cell in the legend table.
Producing an HTML clinical details table
The style file specifies the details table columns. Activating the style definitions in the DTD allows validation of the entries that specify the core attributes, headings and order. Refer to Figure 6. BrowseTMA processes the specified TMA block and displays a row of information about each core in the block. In processing each core, BrowseTMA uses that block's data from the XML export file to display the fields specified for each details column.
Results and discussion
Block lists, virtual slide images, legends, clinical details and exports are on the ACSR web site for 14 blocks with 1623 cores of 2.0, 1.0 and 0.6 mm sizes. Exports for 11 blocks with 3812 cores from three other institutions are processed with the BrowseTMA tool. Researchers can readily navigate from TMA block lists to TMA legends and to clinical details for a selected tissue core. Our virtual microscope can be used to view and annotate linked TMA images. Researchers can download TMA clinical data in the TMA DES format.
Fifty common data elements (CDE) from the 80 defined by the API TMA DES are used. 42 local data elements (LDE) are defined to store data that cannot be mapped to the API TMA DES definitions. This information can help indicate whether additional CDEs should be added to the specification. Table 1 shows the LDEs with indication of which institutions they were used for. For example, as race and tissue elements are needed for data from three institutions, adding CDEs for these might be appropriate. Detailed comments about converting TMA data to the TMA DES format are provided [see Additional file 2].
Table 1 LDE with use by institution
ParentCDE LDE OSU CPCTR BWH TARP
block block_NUMBER-OF-PATIENTS x
block_AUTOPSY-SURGICAL-MIXED x
core core_HIV-STATUS x
core_EXCEL-LOCATION x x
core_AGE x x
Year_of_Birth x
Year_of_Diagnosis x
Year_of_Prostatectomy x
core_GENDER x x
core_RACE x x
Race x
core_TISSUE x x x
core_LN-CONT-NORM-ETC x
core_COMMENTS x x
core_COORDINATES x
IMS_Case_Identifier x
Location_Code x
Is_Residual_Carcinoma_Present x
Most_Prominent_Histologic_Type x
Gleason_Primary_Grade x
Gleason_Secondary_Grade x
Gleason_Sum_Score x
Number_of_Nodes_Examined x
Number_of_Nodes_Positive x
Distant_Mets__1_at_Time_of_Diagn x
pT_Stage x
pN_Stage x
pM_Stage x
Vital_Status x
Year_of_PSA_Recurrence x
PSA_Recurrence_Status x
Recurrence_Free_Year x
core_OTHER-INFO-NOTES x
core_COUNT x
core_BLOCK-SIZE x
core_STATUS x
core_PROTOCOL x
core_FIXATION x
core_ORGAN x
core_IHC x
core_ORGANISM x
core_GRADE x
The data from the CPCTR is modified to make it compatible with the BrowseTMA capabilities. Identifier LDEs were changed to CDEs to avoid additional programming that is necessary when required information (such as a block, core or slide identifier) is stored in locally defined data elements minimizing the benefit of using a common format. Detailed comments about modifying CPCTR data are provided [see Additional file 2].
Six recommendations are prompted by difficulties experienced exporting, importing and using data in the TMA DES format. Several relate to confusion about 'sections'. The data structure section of the specification [8] describes dividing any TMA DES XML file into 4 sections. This is misleading because when there are multiple blocks, the core CDEs are not all together like the chapters of a book. The structure of such XML files should be understood to be a hierarchy or tree. It would be clearer to say that data elements are grouped into 4 categories: header, block, slide and core. A data element of the respective type is the root of a sub-tree that contains data elements about that respective header, block, slide or core. The idea of 'sections' breaks down because each core and slide element must be inside some block element (i.e. within that block section). Detailed recommendations are provided [see Additional file 2].
Conclusion
An organized approach to producing, sorting, navigating and publishing TMA information using the BrowseTMA tool has been put in place. Virtual TMA sections with clinical data from the ACSR can be evaluated on the Internet by interested researchers. Linking images to readily navigable clinical data facilitates reasearcher evaluation of the TMA.
A DTD has been written and applied to the API TMA DES specification and is in routine use. Using a DTD (optionally reflecting our proposed enhancements) can provide stronger validation. The BrowseTMA tool can be used to merge XML exports from various sources and to construct an integrated website. The BrowseTMA tool and DTD are available publicly to assist TMA DES users in reviewing and validating the XML exports they have constructed.
Next steps
We are considering enhancing BrowseTMA to:
• use more slide and block information from exports as defaults to define the block list including multiple slides per block (displayed in the block list), slide image locations, multiple blocks per file,
• allow selection of block list fields as in details,
• navigate to the specific core image (not just the image for the whole block),
• handle namespaces,
• provide details table sorting options,
• improve element attribute handling,
• allow result column addition and editing,
• define and publish allowed data element values and
• produce printer friendly versions of all pages (smaller font, minimum width, no links, legend separated from details).
We plan to rework the style elements to fit inside the histo/tma hierarchy.
We hope to interact with other interested parties about exporting and merging TMA data and improvements to the TMA DES.
Availability and requirements
Example TMA HTML and XML data files and all source code for the BrowseTMA tool and related software are available at the public ACSR Mid-Region web site Personal computers running Microsoft Windows 98, 2000, NT and XP have been used with Internet Explorer 6.0, Word 2002, and FrontPage 2002 in this work.
List of abbreviations used
ACSR – AIDS and Cancer Specimen Resource
AIDS – acquired immunodeficiency syndrome
AMB – AIDS Malignancy Bank
API – Association for Pathology Informatics
caBIG – cancer Biomedical Informatics Grid
CDE – common data elements
DCTD – Division of Cancer Treatment and Diagnosis
DES – data exchange specification
DTD – document type definition
H&E – hematoxylin & eosin
HIV – human immunodeficiency virus
HTML – hypertext markup language
IRB – institutional review board
LDE – locally defined data elements
NCI – National Cancer Institute
OSU – The Ohio State University
TMA – tissue micro-array
TMA DES – tissue micro-array data exchange specification
URI – universal resource identifier
W3C – World Wide Web Consortium
XML – extensible mark-up language
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DGN developed the Microsoft Mail Merge scripts, Excel formulas, API DES DTD, JScript programs, WSH batch scripts and BrowseTMA XSLT scripts and wrote the first draft of the manuscript. BAH supervised the construction of the TMA blocks and slides, provided clinical data and scanned slides with the virtual microscope. LWA was responsible for the TMA design including case selection, block selection, core focus selection and core quality assurance. All authors reviewed and commented on successive drafts of the manuscript and versions of the software and have approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional file 1
This .zip file contains the following 16 files which may be extracted using standard unzip software: Sample TMA DES export: TA00-050.xml This TMA block contains 35 cores in a 5 by 7 array. This XML file can be viewed with Internet Explorer or another browser or editor. Sample TMA file list: fileList.xml Entries for all but one TMA block are commented out to reduce the number of files needed to try the software while illustrating more substantive use. This XML file can be viewed with any text editor or browser. Sample TMA style file: TMA5by7Style.xml The sample TMA block uses the TMA style specified in this TMA style file. TMA5by7Style specifies a 5 row by 7 column array. In legend tables based on this style: • Column labels are repeated at top and bottom. • The label end of the block/slide is at left. • Row labels are repeated at left and right. • The bank ID and organ are shown in each legend cell, which is color-coded based on diagnosis. In details tables based on this style, common and locally defined data elements containing column labels are ordered so that top-to-bottom data elements can be map to left-to-right output table columns. This XML file can be viewed with Internet Explorer or another browser or editor. TMA DES DTD: tmades.dtd This DTD defines the CDEs (tags) and allowed structure of API TMA DES exports. Also present are elements to define the allowed structure of TMA styles in TMA style files. It can be viewed with any text editor. TMA file list DTD: tmaList.dtd This DTD defines the CDEs (tags) and allowed structure of a TMA file list. It can be viewed with any text editor. TMA local data element file DTD: myLDEs.dtd This file defines the LDEs (tags) and their allowed structure. It can be viewed with any text editor. BrowseTMA XSLT script: BrowseTMABatch.xsl This XSLT file produces a single batch file, BrowseTMA.bat, which contains commands to run the BrowseTMAList.xsl XSLT script once and the BrowseTMAStyle.xsl and BrowseTMALegend.xsl XSLT scripts for each TMA block specified in fileList.xml. BrowseTMA WSH batch file: BrowseTMA.bat This WSH batch file runs the BrowseTMAList.xsl XSLT script once and the BrowseTMAStyle.xsl and BrowseTMALegend.xsl XSLT scripts for each TMA block specified in fileList.xml. Standard output is redirected to the appropriate HTML file. BrowseTMA XSLT script: BrowseTMAStyle.xsl This XSLT file produces a single TMA style HTML file hyperlinked to the TMA block list and from the blocks that use this style. BrowseTMA XSLT script: BrowseTMAList.xsl This XSLT file produces a single TMA file list HTML file hyperlinked to each TMA block and style HTML file referenced in it. BrowseTMA XSLT script: BrowseTMALegendDetails.xsl This XSLT file produces a single TMA block HTML file containing a legend and details table hyperlinked to each other. Sample TMA legend and details: TA00-050.html This displays the legend and details tables for the sample TMA block represented in the sample TMA DES block export in TA00-050.xml. This HTML file can be viewed with Internet Explorer or other browser. Sample TMA file table: fileList.html This displays the files, blocks and styles given in the sample TMA file list, fileList.xml. This HTML file can be viewed with Internet Explorer or other browser. Sample TMA style table: TMA5by7Style.html This displays the TMA5by7Style style specified in TMA5by7Style.html. This HTML file can be viewed with Internet Explorer or other browser. XSLT test JScript: xsltTest.js This JScript program runs the Microsoft XSLT Parser, MSXML2 4.0, on a passed XML file and XSLT file. BrowseTMA Make WSH batch file: BrowseTMAMakeBatch.bat This WSH batch file runs the BrowseTMABatch.xsl XSLT script once and outputs the BrowseTMA.bat file.
Click here for file
Additional file 2
This HTML document contains detailed comments and recommendations regarding use of the TMA DES. It can be viewed with Internet Explorer or other browser.
Click here for file
Acknowledgements
This work was supported by a grant from the National Cancer Institute for support of the AIDS and Cancer Specimen Resource: U01 CA66531. The authors are funding recipients of this grant. Dr. Jules Berman, of the National Cancer Institute, is acknowledged for providing XML data from the CPCTR in advance of publication. Jeffrey Tang, of the Brigham and Women's Hospital, is acknowledged for providing Excel data for a Harvard TMA block. Dr. Stephen Hewitt, of the National Cancer Institute, is acknowledged for answering questions about the TARP data.
==== Refs
AIDS and Cancer Specimen Resource web site
Nohle DG Hackman BA Ayers LW Web-based virtual tissue micro-array slides with clinical data for researchers in HTML, Excel® and API standard XML data exchange specification formats produced with Microsoft Office ®applications [abstract] Proceedings of the Advancing Pathology Informatics Imaging and the Internet 8th annual conference: 8–10 October 2003; Pittsburgh, Pennsylvania
Nohle DG Hackman BA Ayers LW Tissue micro-array data TMA standard and public tool for linked legend and clinical details at [abstract] Proceedings of the 8th International Conference on Malignancies in AIDS and Other Immunodeficiencies (ICMAOI): Basic, Epidemiologic and Clinical Research: 29–30 April 2004 98
Nohle DG Hackman BA Ayers LW A document type definition for the API standard tissue micro-array XML data exchange specification [abstract] Proceedings of the Advancing Practice, Instruction and Innovation through Informatics 9th annual conference: 6–8 October 2004; Pittsburgh, Pennsylvania
Kononen J Bubendorf L Kallioniemi A Barlund M Schraml P Leighton S Torhorst J Mihatsch MJ Sauter G Kallioniemi OP Tissue microarrays for high-throughput molecular profiling of tumor specimens Nat Med 1998 4 844 847 9662379 10.1038/nm0798-844
Berman JJ Datta M Kajdacsy-Balla A Melamed J Orenstein J Dobbin K Patel A Dhir R Becich MJ The tissue microarray data exchange specification: implementation by the Cooperative Prostate Cancer Tissue Resource BMC Bioinformatics 5 19 2004 Feb 27 15040818 10.1186/1471-2105-5-19
Romer DJ Yearsly KH Ayers LW Using a modified standard microscope to generate virtual slides Anat Rec 2003 272B 91 7 10.1002/ar.b.10017
Berman JJ Edgerton ME Friedman BA The tissue microarray data exchange specification: A community-based, open source tool for sharing tissue microarray data BMC Med Inform Decis Mak 3 5 2003 May 23 12769826 10.1186/1472-6947-3-5
National Cancer Institute's cancer Biomedical Informatics Grid (caBIG) web site
Microsoft Windows 2000 Scripting Guide web site
Microsoft Corporation Microsoft JScript Tutorial 2004 Microsoft Developers Network Library
Microsoft Corporation Initiate XSLT in a Script 2004 Microsoft Developers Network Library
Arpajian S Mullen R How To Use HTML 32 1996 Macmillan Computer Publishing USA
Niederst J HTML Pocket Reference 2002 2 O'Reilly & Associates, Inc
Harold ER Means WS XML In A Nutshell 2002 2 O'Reilly & Associates, Inc
Brown University Scholarly Technology Group's XML Validation Form
Richard Tobin's XML well-formedness checker and validator
Nohle DG Ayers LW The tissue microarray data exchange specification: A document type definition to validate and enhance XML data BMC Med Inform Decis Mak 2005 5 12 15871741 10.1186/1472-6947-5-12
Cooperative Prostate Cancer Tissue Resource (CPCTR) web site
Tissue Array Research Program (TARP) web site
Bray T Paoli J Sperberg-McQueen CM Maler E Yergeau F Extensible Markup Language (XML) 1.0 (Third Edition) W3C Recommendation 04 February 2004 W3C Recommendation 2004 February 4
Mangano S XSLT Cookbook 2003 O'Reilly & Associates, Inc
Ray ET McIntosh J Perl & XML 2002 O'Reilly & Associates, Inc
Flanagan D JavaScript Pocket Reference 2003 2 O'Reilly & Associates, Inc
|
16086837
|
PMC1208890
|
CC BY
|
2021-01-04 23:52:11
|
no
|
BMC Med Inform Decis Mak. 2005 Aug 8; 5:25
|
utf-8
|
BMC Med Inform Decis Mak
| 2,005 |
10.1186/1472-6947-5-25
|
oa_comm
|
==== Front
BMC Med Inform Decis MakBMC Medical Informatics and Decision Making1472-6947BioMed Central London 1472-6947-5-281610917710.1186/1472-6947-5-28Research ArticleHandheld computers and the 21st century surgical team: a pilot study Aziz Omer [email protected] Sukhmeet S [email protected] Gopalakrishnan [email protected] Paraskevas [email protected] Aziz [email protected] Ara [email protected] Department of Surgical Oncology & Technology, Imperial College London, UK2 Division of Primary Care & Population Health Sciences, Imperial College London, UK3 Division of Community Health Sciences: GP Section, University of Edinburgh, UK2005 18 8 2005 5 28 28 15 12 2004 18 8 2005 Copyright © 2005 Aziz et al; licensee BioMed Central Ltd.2005Aziz et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The commercial development and expansion of mobile phone networks has led to the creation of devices combining mobile phones and personal digital assistants, which could prove invaluable in a clinical setting. This pilot study aimed to look at how one such device compared with the current pager system in facilitating inter-professional communication in a hospital clinical team.
Methods
The study looked at a heterogeneous team of doctors (n = 9) working in a busy surgical setting at St. Mary's Hospital in London and compared the use of a personal digital assistant with mobile phone and web-browsing facilities to the existing pager system. The primary feature of this device being compared to the conventional pager was its use as a mobile phone, but other features evaluated included the ability to access the internet, and reference data on the device. A crossover study was carried out for 6 weeks in 2004, with the team having access to the personal digital assistant every alternate week. The primary outcome measure for assessing efficiency of communication was the length of time it took for clinicians to respond to a call. We also sought to assess the ease of adoption of new technology by evaluating the perceptions of the team (n = 9) to personal digital assistants, by administering a questionnaire.
Results
Doctors equipped with a personal digital assistant rather than a pager, responded more quickly to a call and had a lower of failure to respond rate (RR: 0.44; 95%CI 0.20–0.93). Clinicians also found this technology easy to adopt as seen by a significant reduction in perceptions of nervousness to the technology over the six-week study period (mean (SD) week 1: 4.10 (SD 1.69) vs. mean (SD) week 6: 2.20 (1.99); p = 0.04).
Conclusion
The results of this pilot study show the possible effects of replacing the current hospital pager with a newer, more technologically advanced device, and suggest that a combined personal digital assistant and mobile phone device may improve communication between doctors. In the light of these encouraging preliminary findings, we propose a large-scale clinical trial of the use of these devices in facilitating inter-professional communication in a hospital setting.
==== Body
Background
Worldwide, there has been increasing interest in the use of wireless handheld technologies such as the personal digital assistant (PDA) in hospitals, with recent reports suggesting these devices may soon earn their position on a physician's desk next to the stethoscope [1]. A key reason for this interest is the possibility that PDAs can help facilitate safer decision making and improved efficiency of healthcare delivery [2].
The commercial development and expansion of mobile phone networks has led to the creation of many combined mobile phone and PDA devices, which could prove invaluable in a clinical setting. This is because these devices have the advantage of providing mobile information access, making it possible to retrieve clinically important information at any time of day and in any location. Information resources that doctors may benefit from include easy access to hospital addresses, protocols, evidence-based guidelines, textbooks, electronic patient records and drug formularies, to mention but a few [3,4].
Since being first introduced in 1948, the United Kingdom National Health Service (NHS) faces increasing costs as it attempts to provide free healthcare to all citizens of the United Kingdom, based on need rather than the ability to pay. Investment in information technology (IT) and communication infrastructure is an important part of healthcare expenditure as hospitals in the NHS aim to provide efficient and standardised healthcare delivery to a large patient population [5]. Nevertheless, a recent report by a parliamentary advice committee recommended that the NHS must overcome its preference for short term savings and develop strategies to stop the current underuse of new medical technologies. Currently, the UK only spends 0.36% of its gross domestic product (GDP) on medical technology, unlike Europe which spends 0.55% [6]. Hospital communication in the UK currently consists of pagers and landline telephones, and although they have had a ban on mobile phone use on their premises since the early 1990s (prompted by a warning issued by the UK Medical Devices Agency), many doctors ignore this, using their personal mobile phones for convenience at work.
With a greater appreciation that mobile phone and other wireless technologies are on the whole safe to use in the proximity of medical equipment [7], it is conceivable that devices combining the PDA and mobile phone could soon become the standard means for communication between hospital practitioners. The controlled use of mobile phone systems for hospital-specific communication by medical staff even in sensitive areas has already been deployed in a number of U.S. hospitals [8]. That said there is very limited and empirical evidence that investment in such technologies will actually translate into any clinically meaningful outcome. This pilot study aims to test the hypothesis that a PDA with a built-in mobile telephone is more efficient in facilitating communication between healthcare providers than a hospital pager device.
Methods
The study group consisted of a heterogeneous team of doctors (n = 9) working in a busy surgical setting at the Academic Surgical Unit at St. Mary's Hospital (London). This unit was selected because it was a general surgical team with all members having a clinical commitment. There were varied levels of enthusiasm about the PDA, with some clinicians more critical and others more enthusiastic. All members of the team had a similar basic knowledge of computers and mobile telephones.
All members of the team were given a Palm Tungsten W™ PDA with mobile phone and web-browsing facilities, connected to a Vodafone UK network. In is important to note that these palm devices are no longer on the market and have been replaced by Treo™ smartphones. The Palm Tungsten W™ measures 12.1 × 7.79 × 1.65 cm, and weighs 184 grams, and has a 320 × 320 16-bit colour TFT display screen. The biggest connectivity feature of the Tungsten W™ (provided via a 1-cm long rounded Antenna) is the Class 10 GSM/GPRS radio [9].
This GSM/GPRS enabled PDA was connected to a Vodafone network. General Packet Radio Service (GPRS) keeps you permanently connected to the Internet but only charges you when your phone is sending or receiving data, and works at similar speeds to a dial-up modem on your home PC (100 kbps – 125 kbps) [10]. Global system for mobile communications (GSM) is an open, non-proprietary system that is constantly evolving. Voice is digitally encoded via a unique encoder, which emulates the characteristics of human speech. This method of transmission permits a very efficient data rate/information content ratio [11].
The Tungsten W™ is intended primarily as a data-centric device using the GPRS network. Any application that requires an Internet connection will automatically cause the device to establish a GPRS connection, which takes anywhere from 20–40 seconds depending on the signal strength, and the connection automatically shuts itself off after a period of inactivity. The device runs on a 33 MHz Dragonball VZ processor and includes 16 MB of RAM, of which 15 is available to the user. The manufacturer's specifications state that the lithium ion battery provides up to 10 hours of talk time [9].
All PDAs were also equipped with Dr. Companion® software provided by MedHand©, containing electronic versions of commonly used UK medical reference textbooks such as the British National Formulary and the Oxford Handbook of Clinical Medicine. Other reference software available on this card included a drug interactions compendium, and anatomy atlas, the International Classification of Disease-10 guidelines, and medical calculators. Some users also chose to make use of the address book and diary functions of the PDA. However, detailed use of the various functions of the PDA was not monitored.
The crossover study was carried out for 6 weeks (17th May to 25th June 2004), with the team having access to the PDA every alternate week. During weeks 1, 3 and 5, study participants adopted the conventional pager system for communication, and used PDAs during weeks 2, 4 and 6. The pagers used by the hospital were alpha-numeric with no internet text paging, and no two-way communication ability. The main feature of this device being compared to the conventional pager was its use as a mobile phone, but other features such as access to the internet and reference data were also considered and evaluated.
The primary outcome measure of interest for assessing efficiency of communication was the length of time it took for clinicians to respond to a call. To test this, one investigator, on a randomly selected day of the week, called all members of the team (n = 8 i.e. excluding the consultant) and measured the time taken for doctors to respond to the call. If after 5 minutes, there was no reply, it was noted that the doctor had 'failed to respond'. We used this measurement of 'response time' to measure efficiency of communication with the respective devices. The mobile phones were called directly and the paging was done via the standard paging system of using the hospital extension. In order to minimise the risk of bias we ensured that doctors could not differentiate a test call from regular communication traffic by making calls from different hospital extensions on each occasion. For the purpose of computing mean response time for each period of the study, the failure to respond was given a value of 301 seconds (5minutes + 1 second). Undoubtedly, not all members of the team would be in the same place; some were in the operating theatre, some on the wards and some in clinic. Whether this affected the response times is beyond the scope of this pilot-study. The aim was however to determine the response times at a randomly allocated point in time as in a real-life situation. We determined the rate ratios of those failing to respond for each pair of adjacent Pager/PDA periods and pooled the results using a fixed effects model to produce an overall rate ratio for failure to respond. This method was chosen so as to take into account the clustering of responses within individuals and the small sample size.
Furthermore, this study sought to assess the ease of adoption of new technology by evaluating the perceptions of the team including the consultant (n = 9), to PDAs by administering a questionnaire at the start of PDA use (week 2) and end of the study (week 6). The questions were adapted from the following validated rating scales: Computer Anxiety Rating Scale (CARS-C); the Computer Thoughts Survey Scale (CTS-C); and the General Attitude Towards Computer Scale (GATC-C) [12]. During the final week of the study, we also administered a questionnaire to the entire team which asked the user to evaluate the usefulness of the PDA and Dr. Companion® software on a 5-point Likert scale (1 = not at all useful; 5 = extremely useful).
Results
The response times to paging were measured on the eight participating junior doctors, although the number of participants was on occasions reduced because of night-duty and absence (Table 1). The point estimates of the response times were always smaller in the PDA weeks than in the pager weeks, but the wide confidence intervals made all but one comparison non-significant (significant difference in mean response times between pager week 5 and PDA week 4). However, when the rates of not responding to the bleep between adjacent pager/PDA periods were compared, in every comparison the periods of PDA use showed lower rates than those during the periods of pager use (overall RR: 0.44; 95%CI 0.20–0.93). We did not take into account the location of the respondents who may have been on the ward round or in the operating theatre, as we do not believe that this would alter the overall result.
Perceptions of nervousness associated with PDA use dropped significantly (p = 0.04) during the study, suggesting positive uptake of new technology by the team. Negative perceptions and disagreement (as measured by the validated questionnaires) regarding PDAs also decreased, albeit non- significantly [12]. Initially not all members of the team were keen to use the PDA-mobile phone device as they would have preferred using the traditional paging system owing to the reluctance to learn how to operate a new device. Dr. Companion® software was rated as being moderately useful (mean = 3.77, SD 0.97) by the end of the study, with most doctors reporting using of the software between 5 to 10 times a day. Detailed usage of the various functions of the PDA and the software were not assessed, except that the favourite reference texts for most users were the British National Formulary, the Oxford Handbook of Clinical Medicine and the Evidence-based medicine (EBM) guidelines.
In addition, 7/9 (78%) of the staff preferred electronic-based reference material compared to paper-based material. 6/9 (67%) members of the team found the PDA to be user-friendly and easy to learn (67% and 56% respectively). 4/9 (44%) team members thought that it decreased their work load while 7/9 (78%) agreed that it enhanced the efficacy of communication between each other. However, 5/9 (56%) doctors did not think that the PDA gave them more time for other hospital tasks.
Finally, 7/9 (78%) agreed with the statement that having a PDA meant that they could deliver faster, more efficient patient care. Issues about battery life were raised by some members. Some participants charged their PDAs on a daily basis, others did not. Comments on battery life and poor signal strength in certain areas of the hospital such as the lifts were not explored in any further detail.
Table 1 Mean response time for each week of the study
Period Device Number Response time (seconds)
Total† Not responding Mean 95%CI
1 Pager 8 4 182 70 to 293
2 PDA 7 2 97 0* to 266
3 Pager 6 2 133 0* to 277
4 PDA 7 0 18 4 to 33
5 Pager 6 3 178 34 to 322
6 PDA 7 2 92 0* to 224
† Varies from week to week due to absence or night duty
* Adjusted to avoid implausible values
Discussion
The use of combined PDA-mobile phone devices in our pilot study suggests that this technology might reduce the time doctors take to respond to a call. This is not surprising as the PDA with a mobile phone is a bi-directional device and enables faster communication between the caller and the doctor. Furthermore, it highlights an important potential limitation of pagers. Doctors often find themselves on busy hospital wards where phones are often otherwise engaged, or lifts, corridors, and just outside the hospital, where the absence of a landline makes the pager an ineffective communication device. Although the study lasted only six weeks, there is no evidence to suggest that these devices will not be as effective in the longer term. In addition, it is important to note that the members of the study group were not all 'PDA enthusiasts', but a rather diverse group consisting of pro-PDA users as well as sceptics even though it is often felt that resident physicians or surgeons are more likely to use newer technology and are more receptive to change. The group on the whole was largely heterogeneous in this respect. Only 2/9 (22%) participants in the study had prior experience with a PDA (1 using a Palm device and 1 using a Pocket PC device).
The assessment of doctors' perceptions regarding introduction of PDAs during the six-week study demonstrated an increasing level of confidence with the devices. This is an important characteristic that any device proposed to replace the hospital pager must have. Widespread mobile phone and PDA use in the commercial setting have meant that the majority of doctors working in the UK already own such technology and are therefore likely to be comfortable with using the devices at work.
The ability to store electronic versions of commonly used reference material such as the British National Formulary and the Oxford Handbook of Clinical Medicine on a single chip is a potentially valuable attribute of the PDA. It means that clinicians can now carry on their person a quantity of reference material that would previously have been inconceivable in hard-copy format. This is, we believe, particularly important for junior doctors and during night-time periods when there are fewer people around to ask for advice. With regards to standard of patient care, it is conceivable that these reference materials may aid correct diagnosis and management, particularly with regards to correct drug dosage.
A major limitation of most research involving mobile phone devices in healthcare is that this has been carried out under laboratory conditions, as opposed to a field test such as in this study [13]. Research on the use of PDAs however, has been in the form of field tests, and supports the finding of this study. An example of this is a recent trial in the United Kingdom, where authors looked at the use of PDAs to access reference information as well as local and national guidelines by members of an on call service for health protection, with results suggesting that this was a fast, reliable and easily maintained source of information that could be used by other groups of practitioners [14].
Conclusion
PDAs have been increasingly used in clinical medicine for the delivery of information at the point of care, the collection of patient information, updating of clinical records, electronic prescribing and medical education. This pilot study integrated a combined PDA and mobile phone device into the daily schedule of a team of general surgeons, showing its usefulness primarily as a communication tool. Although we were unable measure a direct patient benefit from the use of these devices, the majority of doctors in the trial felt that having a PDA with a mobile phone as an in-built feature did help them to deliver better health care, and found this technology easy to adopt.
In light of these promising initial findings, we now propose a large-scale clinical trial of the use of PDAs with built-in mobile telephones in the hospital setting. This may be a first step in developing the evidence base for a new hospital communication system that may eventually replace the quaint, but hopelessly outdated hospital pager. Other wireless technologies such as blue-tooth and wireless local area networks must also be considered in any communication system proposed for use in hospitals.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Omer Aziz concieved and coordinated the study. Sukhmeet S. Panesar was responsible for data acquisition, statistical analysis, and write-up. Gopalakrishnan Netuveli and Aziz Sheikh were responsible for study design and statistical analysis. Paraskevas Paraskeva and Ara Darzi were responsible for study implementation, result interpretation, and oversaw the trial.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank Henrik Andersson (MedHand International), Martin Day (Palm Europe), David Headley (Vodafone UK), and Urooj Shaikh (Workstream) for their contribution of hardware, software, mobile phone line connection, and expertise without which this trial would not be possible.
==== Refs
Gillingham W Holt A Gillies J Handheld computers in health care: what software programs are available? N Z Med J 2002 115 U180 12386659
McAlearney AS Schweikhart SB Medow MA Doctors' experience with handheld computers in clinical practice BMJ 2004 328 1162 15142920 10.1136/bmj.328.7449.1162
Ammenwerth E Buchauer A Bludau B Haux R Mobile information and communication tools in the hospital Int J Med Inform 2000 57 21 40 10708253 10.1016/S1386-5056(99)00056-8
Cimino JJ Bakken S Personal digital educators N Engl J Med 2005 352 860 2 2005 Mar 3 15745975 10.1056/NEJMp048149
Humber M National programme for information technology BMJ 2004 328 1145 6 15142891 10.1136/bmj.328.7449.1145
Mayor S NHS is failing to reap benefit of new technologies, MPs say BMJ 2005 330 861 16April 15831853 10.1136/bmj.330.7496.861
Aziz O Sheikh A Paraskeva P Darzi A Use of mobile phones in hospital: time to lift the ban? Lancet 2003 361 788 12620772 10.1016/S0140-6736(03)12652-9
Morrissey JJ Mobile phones in the hospital: improved mobile communication and mitigation of EMI concerns can lead to an overall benefit to healthcare Health Phys 2004 87 82 8 15194927 10.1097/00004032-200407000-00011
Garfield L Palm Tungsten W 26th March 2003. Last accessed on 15th April 2005
Vodafone website. GPRS Last accessed on 15th April 2005
GSM world – frequently asked questions Last accessed on 15th April 2005
Korukonda AR Finn A An investigation of framing and scaling as confounding variables in information outcomes: the case of technophobia Information Sciences 2003 15 79 88 10.1016/S0020-0255(03)00153-1
Ammenwerth E de Keizer N An inventory of evaluation studies of information technology in health care: Trends in evaluation research 1982 – 2002 Accepted for Methods of Information in Medicine 2004 Last accessed on 15th April 2005
Abubakar I Williams CJ McEvoy M Development and evaluation of a hand held computer based on-call pack for health protection out of hours duty: A pilot study BMC Public Health 2005 5 35 15823207 10.1186/1471-2458-5-35
|
16109177
|
PMC1208891
|
CC BY
|
2021-01-04 23:52:11
|
no
|
BMC Med Inform Decis Mak. 2005 Aug 18; 5:28
|
utf-8
|
BMC Med Inform Decis Mak
| 2,005 |
10.1186/1472-6947-5-28
|
oa_comm
|
==== Front
BMC Med Inform Decis MakBMC Medical Informatics and Decision Making1472-6947BioMed Central London 1472-6947-5-291611150310.1186/1472-6947-5-29Research ArticleExploring cancer register data to find risk factors for recurrence of breast cancer – application of Canonical Correlation Analysis Razavi Amir R [email protected] Hans [email protected]ål Olle [email protected] Marie [email protected] Sten [email protected]Åhlfeldt Hans [email protected] Nosrat [email protected] South-East Swedish Breast Cancer Study Group [email protected] Department of Biomedical Engineering, Division of Medical Informatics, Linköping University, Sweden2 Department of Biomedicine and Surgery, Division of Oncology, Linköping University, Sweden3 Department of Surgery, County Hospital, Kalmar, Sweden4 Department of Pathology, County Hospital, Kalmar, Sweden5 Oncology Centre, University Hospital, Linköping University, Sweden2005 22 8 2005 5 29 29 1 3 2005 22 8 2005 Copyright © 2005 Razavi et al; licensee BioMed Central Ltd.2005Razavi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A common approach in exploring register data is to find relationships between outcomes and predictors by using multiple regression analysis (MRA). If there is more than one outcome variable, the analysis must then be repeated, and the results combined in some arbitrary fashion. In contrast, Canonical Correlation Analysis (CCA) has the ability to analyze multiple outcomes at the same time.
One essential outcome after breast cancer treatment is recurrence of the disease. It is important to understand the relationship between different predictors and recurrence, including the time interval until recurrence. This study describes the application of CCA to find important predictors for two different outcomes for breast cancer patients, loco-regional recurrence and occurrence of distant metastasis and to decrease the number of variables in the sets of predictors and outcomes without decreasing the predictive strength of the model.
Methods
Data for 637 malignant breast cancer patients admitted in the south-east region of Sweden were analyzed. By using CCA and looking at the structure coefficients (loadings), relationships between tumor specifications and the two outcomes during different time intervals were analyzed and a correlation model was built.
Results
The analysis successfully detected known predictors for breast cancer recurrence during the first two years and distant metastasis 2–4 years after diagnosis. Nottingham Histologic Grading (NHG) was the most important predictor, while age of the patient at the time of diagnosis was not an important predictor.
Conclusion
In cancer registers with high dimensionality, CCA can be used for identifying the importance of risk factors for breast cancer recurrence. This technique can result in a model ready for further processing by data mining methods through reducing the number of variables to important ones.
==== Body
Background
Breast cancer is the most common type of cancer diagnosed in women in Western countries. Sweden has had a high incidence of breast cancer for several decades, although mortality rates have been lower than in most other Western countries [1].
Breast cancer prognosis is influenced by many factors such as morphological and pathological tumor specifications and biological tumor markers. Studying these predictors and finding those of most importance can give clinicians better insight regarding the prognosis.
As a rule, data on cancer patients have been collected in all regions of Sweden since 1960. The data have been used for different purposes including epidemiological studies, monitoring and evaluating medical interventions, and finding risk factors for specific types of cancer [2-4]. A common approach in using data in registers for finding relationships between outcomes and predictors is to use multiple regression analysis (MRA). If the aim of the study is to determine the degree of importance of each predictor with more than one outcome variable, then the analysis has to be repeated, and the results must be combined in some arbitrary fashion.
In MRA, identification of important predictors is usually done by looking at the regression weights associated with each predictor. If the variables in the analysis are correlated among themselves (multicollinearity), it is then difficult to interpret the importance of the individual variables [5]. However, a seldom-used alternative approach is to compute loadings (structure coefficients) and use them as indicators of important predictors.
Hotelling (1936) developed Canonical Correlation Analysis (CCA) as a method for evaluating linear correlation between sets of variables [6]. The method allows investigation of the relationship between two sets of variables that can identify important variables in a set of multiple predictors and a set of multiple outcomes. Loadings and weights can be calculated by CCA with a module that is included in commercial statistical packages.
In the analysis of cancer recurrence, it is necessary to consider time as a fundamental factor, since different parameters are related to recurrence after a certain period of time. Some methods like Cox regression analysis are specially developed for handling events that occur during different times and when some cases are censored [7] (outcome not known when the study period ends).
In this study, CCA was used for identifying the importance of risk factors for breast cancer recurrence within specified time intervals. CCA was applied to data from breast cancer patients in the south-east region of Sweden.
Methods
In this study, data from 637 female patients, mean age 59.5 years, were analyzed. The same patients as in the study by Sundquist et al. was used [8]. The aim of that study was to assess the applicability of histopathological grading as a prognostic index applied to a defined breast cancer population. Only patients without any sign of distant metastasis at the time of surgery were included in the analysis. Tumors with an invasive component of 2 millimetres or less in diameter were excluded from analysis because the small size did not permit proper grading in accordance with the protocol.
To obtain a more comprehensive dataset for this study, patient data were retrieved from three different registers, i.e. the regional breast cancer-, tumor markers- and cause of death registers.
Predictors and outcomes
In order to answer important questions such as: "Which variables might be important predictors for recurrence of breast cancer? Is the time interval after diagnosis important? Is there a way to determine the importance of each predictor when there is more than one type of recurrence?" two sets of predictors and outcomes (see Table 1) were selected by consulting and collaborating with oncologists and studying the literature in the domain.
Table 1 List of variables in both sets
Predictor Set Outcome Set ‡
Age DM, first two years
Tumor location DM, 2–4 years
Side DM, more than 4 years
Tumor size * LRR, first two years
LN involvement * LRR, 2–4 years
LN involvement † LRR, more than 4 years
Periglandular growth *
NHG
Multiple tumors *
Estrogen receptor
Progesterone receptor
S-phase fraction
DNA index
DNA ploidy
Abbreviations: LN: lymph node, NHG: Nottingham Histologic Grade, DM: Distant Metastasis, LRR: Loco-regional Recurrence
* from pathology report, † N0: Not palpable LN metastasis, ‡ all periods are time after diagnosis.
Age of the patient and variables regarding tumor specifications based on pathology reports, physical examination and tumor markers were selected as predictors. Two variables in the outcome set, distant metastasis and loco-regional recurrence were observed at different time intervals after diagnosis.
Data preprocessing
After retrieving information in different registers about selected predictors and outcomes (Table 1) for 637 patients, the raw data were transformed and converted as illustrated in Table 2.
Table 2 Transformation rules and the study population characteristics
Variable Categories Coded as n
Age >50 years 0 177
≤ 50 years 1 459
Tumor location (not) Superior medial (0)1 144
(not) Inferior medial (0)1 70
(not) Superior lateral (0)1 368
(not) Inferior lateral (0)1 112
(not) Nipple area (0)1 58
Side Left 0 315
Right 1 322
Tumor Size ≤ 20 mm 0 233
>20 mm 1 404
LN involvement No LN involvement 0 373
Positive LN involvement 1 260
LN involvement (N0) No palpable LN 0 100
Palpable and/or fixed LNs 1 533
Periglandular growth Absence of growth 0 515
Presence of growth 1 122
Nottingham Histologic Grade I 1 145
II 2 228
III 3 264
Multiple tumors Absence of multiple tumors 0 502
Presence of multiple tumors 1 134
Estrogen receptor ≥ 0.3 fmol/mg 0 181
<0.3 fmol/mg 1 456
Progesterone receptor ≥ 0.3 fmol/mg 0 232
<0.3 fmol/mg 1 405
S-phase fraction <10% 0 439
≥ 10% 1 198
DNA index (DI) 0.9 ≤ DI and DI < 1.3 0 345
0.9 > DI or DI ≥ 1.3 1 292
DNA ploidy DNA diploidy or tetraploidy 0 368
DNA aneuploid 1 269
For some variables such as LN involvement, periglandular growth and multiple tumors, dichotomization was done based on their presence or absence in the patients. Other variables such as tumor location, side, Nottingham Histologic Grade and DNA ploidy were already categorical. The remaining variables, i.e. age, tumor size, estrogen and progesterone receptors, S-phase fraction and DNA index, were transformed from continuous to dichotomous variables (Table 2).
Missing values were substituted using the Expectation maximization (EM) algorithm [9]. This algorithm is a parameter estimation method, which falls within the general framework of maximum likelihood estimation and is an iterative optimization algorithm.
Canonical Correlation Analysis
Because the outcome set consists of several variables, CCA, which is a technique for analyzing the relationship between two sets of variables, was performed.
The fundamental principle behind CCA is the creation of a number of canonical solutions [5], each consisting of a linear combination of one set of variables, which has the form:
Ui = a1(predictor1) + a2(predictor2) + ... + am(predictorm)
and a linear combination of the other set of variables, which has the form:
Vi = b1(outcome1) + b2(outcome2) + ... + bn(outcomen)
The goal is to determine the coefficients (a's and b's) that maximize the correlation between canonical variates Ui and Vi. The number of solutions is equal to the number of variables in the smaller set. The first canonical correlation is the highest possible correlation between any linear combination of the variables in the predictor set and any linear combination of the variables in the outcome set.
A way of interpreting the canonical solutions is to look at the correlations between the canonical variates and the variables in each set. These correlations are called structure coefficients or loadings. The logic here is that variables that are highly correlated with a canonical variate have more in common with it and they should be considered more important when deriving a meaningful interpretation of the related canonical variate. This way of interpreting canonical variates is identical to the interpretation of factors in factor analysis [10]. The criterion for choosing the important variables in each canonical variate is the structure coefficients (loadings). As a rule of thumb for meaningful loadings, an absolute value equal to or greater than 0.3 is often used [11,12].
Significance of the canonical correlations was tested with randomization tests, and robustness of the estimates of the loadings was tested with bootstrapping [13].
SPSS version 11 [14] was used for data transformation and replacing missing values. For running CCA, the CANCORR macro, a part of the Advanced Statistics module of SPSS, was used. Tests of significance for canonical correlations and bootstrapping were done using MATLAB Ver 6.5 [15].
Results
Table 2 shows the study population characteristics. The relationships between predictors and outcomes (Table 1) were analyzed by CCA, which generated six solutions, equal to the number of outcome variables.
For the first solution, the canonical correlation coefficient (rc) was equal to 0.547 with the p value ≤ .001.
Table 3 gives the individual Structure Coefficients (loadings) between the tumor specifications and their canonical variate (U1) and between the recurrences of breast cancer and their canonical variate (V1) for the first solution. Important variables (absolute value of loading ≥ 0.3) are shown in bold type in the table. Other variables with lower loadings are not considered important for the interpretation.
Table 3 Canonical Structure Matrix for Predictor and outcome Variates
Predictor Set U1 Outcome Set‡ V1
Age .223 DM, first two years .837
Tumor location DM, 2–4 years .332
Superior medial .138 DM, more than 4 years .193
Inferior medial .159 LRR, first two years .486
Superior lateral -.056 LRR, 2–4 years -.030
Inferior lateral .155 LRR, more than 4 years -.013
Nipple area .160
Side -.017
Tumor size * .432
LN involvement * .567
LN involvement (N0) † .580
NHG .697
Perigland growth * .566
Multiple tumors * .110
Estrogen receptor .370
Progesterone receptor .365
S-phase fraction .629
DNA index .325
DNA ploidy .342
Abbreviations: LN: lymph node, NHG: Nottingham Histologic Grade, DM: Distant Metastasis, LRR: Loco-regional Recurrence
* from pathology report, † N0: Not palpable LN metastasis, ‡ all periods are time after diagnosis.
If the signs in the sets are the same, then if one increases the other also increases, and vice versa.
When displaying loadings, signs help to identify the character of the relationship between variables in the predictor and outcome sets. If both have the same sign then they change in the same direction; if one increases then the other will also increase, and vice versa.
The first solution is illustrated in Figure 1. The variables in both sets are arranged by the absolute values of the loadings, which show their importance within each canonical variate. By considering loadings and signs in the first solution, patients with higher NHG, higher S-phase fraction, presence of lymph node involvement (based on pathology reports and physical examination), presence of periglandular growth, larger tumor size, negative estrogen and progesterone receptor status, DNA aneuploidy and abnormal DNA index are associated with an increased risk for distant metastasis (DM) and loco-regional recurrence (LRR) during the first two years after diagnosis of breast cancer, and DM 2–4 years after diagnosis. In the predictor set, age, existence of multiple malignant tumors, and side and tumor location did not get meaningful loadings so they are assumed to be unimportant as predictors for recurrence of the disease.
Figure 1 The first canonical correlation solution. Variables are sorted by the absolute value of their loadings. Abbreviations: LN: lymph node, DM: Distant Metastasis, LRR: Loco-regional Recurrence. * N0: Not palpable LN metastasis, † from pathology report, ‡ all periods are time after diagnosis. If the signs in the sets are the same, then if one increases the other also increases, and vice versa.
Discussion
The use of CCA loadings facilitates the detection of important predictors, particularly when there are many variables in the dataset and there are high correlations among those variables. The ability to find important predictors when we are not restricted to using just one outcome variable means that we have a more general tool for analyzing the data. CCA can be used for the purpose of dimension reduction prior to a data mining step for knowledge discovery in databases. Using exploratory multivariate statistics such as CCA, the effective number of variables can be reduced while preserving the information content.
Methodological consideration
In the present study we show that CCA provides overall associations between tumor specifications and breast cancer outcomes derived from the datasets in the registers.
If the interpretation is based on the level of loadings, high correlations between variables do not disturb the interpretation. This is in contrast to analyzing variables based only on the significance of the weights. Cooley and Lohnes suggest examining loadings as a better criterion for finding important predictors, especially when the goal is to determine which variables relate most strongly to the linear composite that best predicts the outcomes [16].
Since MRA is in fact a special case of CCA when the outcome set consists of just one variable, both loadings and weights can also be calculated for MRA, but in standard statistics software only the weights are calculated [17].
The number of cases studied is important in CCA. If there are too few cases, the results will not be reliable. Barcikowski and Stevens, in a Monte Carlo study on the stability of the coefficients and the correlations in canonical correlation analysis, found that a ratio of about 20:1 between the number of records and the number of variables is sufficient for accurate interpretation [18,19]. In our study, the ratio was about 25:1, which means that we had a sufficient number of cases.
CCA is not commonly used in medicine. Its limited use may be due to a lack of familiarity with the method and complexity in the calculations, but CCA is now included in statistics software packages such as SPSS and SAS.
The significance of the analysis and the robustness of the results were successfully tested with randomization tests and bootstrapping.
The results support the correctness and validity of the registers that were used, since many of the important variables were confirmed based on the knowledge and experience of oncology specialists.
Medical interpretation
In the first solution (Figure 1), Nottingham Histologic Grading (NHG) got the highest loading. This grading technique involves semiquantitative evaluation of three morphological features and a numerical scoring system. This score is also a well-known and important predictor for the prognosis of breast cancer [20].
The S-phase fraction is a measure of the percentage of cells in cancer cells that are in the phase of the cell cycle during which DNA is synthesized. Other studies have shown that higher fractions are generally associated with poorer overall survival [21]. In our study, S-phase fraction got second place among predictors.
Examining lymph node involvement is essential in assessing the probability of breast cancer recurrence. In this study, variables showing nodal involvement are important and got third and fourth place among predictors. The overall survival of patients has been shown to decrease as nodal involvement increases [22].
Periglandular growth of the malignant tumor [23], size of the tumor [8], receptors for estrogen and progesterone [24], DNA ploidy and DNA index [25,26] were also found to be important predictors in the present study.
Some studies indicate that age is an important factor [27] and the younger the age of the patient, the poorer the prognosis for the disease. However, in this study, age, used either as a continuous variable or as a categorical one, did not get any meaningful loading.
There were other variables such as side and location of the tumor that based on their loadings were not considered important predictors (Table 3).
Looking at the outcome set for the first solution, we see that the important predictors are related to the occurrence of DM during the first four years and LRR during the first two years after diagnosis.
In this study, we have presented the CCA method for exploring registered data using a proposed model for predictors and multiple outcome variables. CCA can analyse different models including combination of predictors or further predictors such as genetic risk factors, node ratio or family history of breast cancer. Flexibility in creating different models can also cover several outcomes in different time intervals. However, construction of a model is an important step because predictors and outcomes should be meaningful medically or epidemiologically as far as clinical decision making is concerned.
Conclusion
CCA is suggested as an appropriate method when there are many variables in the input set and more than one variable in the output set. Applying CCA to the available dataset and reducing the number of variables to the important ones can promote further analysis in data mining methods. This can be assumed as the dimension reduction step in the whole process of knowledge discovery in the databases. By reducing the effort involved in manual interventions in analyzing data, CCA can also be helpful in real-time analytical processes.
We analyzed the relationship between tumor specifications and outcomes for breast cancer during different time intervals. The results of the main analysis successfully detected well known predictors for breast cancer recurrence in the input set. Nottingham Histologic Grade (NHG) was the most important prognostic variable in breast cancer patients. The next most important factors were S-phase fraction and nodal status.
List of abbreviations
CCA: Canonical Correlation Analysis
MRA: Multiple Regression Analysis
NHG: Nottingham Histologic Grading
LN: Lymph Node
DM: Distant Metastasis
LRR: Loco-regional Recurrence
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
This research is done as a part of ARR's PhD education under the supervision of NS with the HG, HÅ, OS as co-supervisors. Pathologic part of the study was done by MS and ST. The South-east Swedish breast cancer study group approves the project.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was performed in the framework of Semantic Mining, a Network of Excellence funded by EC FP6. It was also supported by grant No. F2003-513 from FORSS, the Health Research Council in the South-East of Sweden.
==== Refs
Tejler G Norberg B Dufmats M Nordenskjold B Survival after treatment for breast cancer in a geographically defined population The British journal of surgery 2004 91 1307 1312 15376206 10.1002/bjs.4697
Berg P Wennberg AM Tuominen R Sander B Rozell BL Platz A Hansson J Germline CDKN2A mutations are rare in child and adolescent cutaneous melanoma Melanoma Research 2004 14 251 255 15305154 10.1097/01.cmr.0000131014.79262.bf
Bull A Mountney L Sanderson H Use of a cancer register for monitoring and evaluating a breast cancer screening programme Community Medicine 1989 11 220 224 2605888
Feltelius N Ekbom A Blomqvist P Cancer incidence among patients with ankylosing spondylitis in Sweden 1965-95: a population based cohort study Annals of the rheumatic diseases 2003 62 1185 1188 14644856 10.1136/ard.2002.004721
Vogel RL Ackermann RJ Is primary care physician supply correlated with health outcomes? International journal of health services 1998 28 183 196 9493759
Levine MS Canonical Analysis and Factor Comparison Quantitative Applications in the Social Sciences series 1977 6 Sage Publications Inc
Cox DR Regression models and life tables (with discussion) Journal of the Royal Statistical Society: Series B 1972 34 187–220
Sundquist M Thorstenson S Brudin L Nordenskjold B Applying the Nottingham Prognostic Index to a Swedish breast cancer population. South East Swedish Breast Cancer Study Group Breast cancer research and treatment 1999 53 1 8 10206067 10.1023/A:1006052115874
McLachlan GJ Krishnan T The EM algorithm and extensions 1997 John Wiley & Sons
McKeon JJ Canonical analysis: Some relations between canonical correlation, factor analysis, discriminant function analysis, and scaling theory Psychometric Monographs 1965 13 Chicago , Univ. of Chicago Press
Lambert ZV Durand RM Some precautions in using canonical analysis Journal of Marketing Research 1975 7 468 475
Thompson B Canonical correlation analysis: Uses and interpretation Quantitative Applications in the Social Sciences 1984 47 Thousand Oaks, CA , Sage Publications
Ridderstolpe L Canonical correlation analysis of risk factors and clinical outcomes in cardiac surgery Dept of Biomedical Engineering 2003 Linköping University, Sweden
SPSS Inc. SPSS for Windows 2001 11.0.1 Chicago , SPSS Inc.
MATLAB 6.5 The MathWorks Inc.
Cooley WW Lohnes PR Multivariate data analysis 1971 New York , Wiley
Dunlap WP Landis RS Interpretations of multiple regression borrowed from factor analysis and canonical correlation Journal of General Psychology 1998 125 397 407
Barcikowski R Stevens JP A Monte Carlo study of the stability of canonical correlations, canonical weights, and canonical variate-variable correlations Multivariate behavioral research 1975 10 353 364
Pugh RC Hu Y Use and interpretation of canonical correlation analyses in Journal of Educational Research articles: 1978-1989 Journal of Educational Research 1991 84 147 152
Elston CW Ellis IO Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up Histopathology 1991 19 403 410 1757079
Wenger CR Clark GM S-phase fraction and breast cancer--a decade of experience Breast Cancer Res Treat 1998 51 255 265 10068083 10.1023/A:1006188512927
Sundquist M Thorstenson S Brudin L Wingren S Nordenskjold B Incidence and prognosis in early onset breast cancer Breast 2002 11 30 35 14965642 10.1054/brst.2001.0358
Adami HO Graffman S Johansson H Rimsten A Survival and recurrences five years after selective treatment for breast carcinoma British Journal of Cancer 1978 38 624 630 728351
Ciocca DR Elledge R Molecular markers for predicting response to tamoxifen in breast cancer patients Endocrine 2000 13 1 10 11051041 10.1385/ENDO:13:1:1
Rzymowska J Skierski J Kurylcio L Dyrda Z DNA index as prognostic factor in breast cancer Neoplasma 1995 42 239 242 8552202
Wenger CR Beardslee S Owens MA Pounds G Oldaker T Vendely P Pandian MR Harrington D Clark GM McGuire WL DNA ploidy, S-phase, and steroid receptors in more than 127,000 breast cancer patients Breast Cancer Res Treat 1993 28 9 20 8123871 10.1007/BF00666351
Lyman GH Lyman S Balducci L Kuderer N Reintgen D Cox C Baekey P Greenberg H Horton J Age and the Risk of Breast Cancer Recurrence Cancer Control 1996 3 421 427 10764500
|
16111503
|
PMC1208892
|
CC BY
|
2021-01-04 23:52:11
|
no
|
BMC Med Inform Decis Mak. 2005 Aug 22; 5:29
|
utf-8
|
BMC Med Inform Decis Mak
| 2,005 |
10.1186/1472-6947-5-29
|
oa_comm
|
==== Front
BMC Med Inform Decis MakBMC Medical Informatics and Decision Making1472-6947BioMed Central London 1472-6947-5-301613524410.1186/1472-6947-5-30Research ArticleAutomation of a problem list using natural language processing Meystre Stephane [email protected] Peter J [email protected] Department of Medical Informatics, University of Utah School of Medicine, Salt Lake City, Utah, USA2005 31 8 2005 5 30 30 18 3 2005 31 8 2005 Copyright © 2005 Meystre and Haug; licensee BioMed Central Ltd.2005Meystre and Haug; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The medical problem list is an important part of the electronic medical record in development in our institution. To serve the functions it is designed for, the problem list has to be as accurate and timely as possible. However, the current problem list is usually incomplete and inaccurate, and is often totally unused. To alleviate this issue, we are building an environment where the problem list can be easily and effectively maintained.
Methods
For this project, 80 medical problems were selected for their frequency of use in our future clinical field of evaluation (cardiovascular). We have developed an Automated Problem List system composed of two main components: a background and a foreground application. The background application uses Natural Language Processing (NLP) to harvest potential problem list entries from the list of 80 targeted problems detected in the multiple free-text electronic documents available in our electronic medical record. These proposed medical problems drive the foreground application designed for management of the problem list. Within this application, the extracted problems are proposed to the physicians for addition to the official problem list.
Results
The set of 80 targeted medical problems selected for this project covered about 5% of all possible diagnoses coded in ICD-9-CM in our study population (cardiovascular adult inpatients), but about 64% of all instances of these coded diagnoses. The system contains algorithms to detect first document sections, then sentences within these sections, and finally potential problems within the sentences. The initial evaluation of the section and sentence detection algorithms demonstrated a sensitivity and positive predictive value of 100% when detecting sections, and a sensitivity of 89% and a positive predictive value of 94% when detecting sentences.
Conclusion
The global aim of our project is to automate the process of creating and maintaining a problem list for hospitalized patients and thereby help to guarantee the timeliness, accuracy and completeness of this information.
==== Body
Background
The problem list is an important piece of the medical record as well as a central component of the problem-oriented electronic medical record in development in our institution (Intermountain Health Care, Utah). To serve the functions it is designed for, the problem list has to be as accurate and timely as possible. In most of our inpatient settings, we are converting to an electronic problem list from a paper-based tool. However, the current problem list, hand written on a form in the paper chart, is usually incomplete and inaccurate, and is often totally unused. While we hope that the advantages of a universally available problem list will change this behavior, as an addition incentive, we are building an environment where the problem list is easily and effectively maintained.
The medical problem list
More than three decades ago, Larry Weed proposed the Problem-Oriented Medical Record as a remedy for the complexity of the medical knowledge and clinical data, and for weaknesses in the documentation of medical care [1,2]. He noted the lack of consistent structure and content in the progress notes that make up a large part of the medical record. He proposed a standard approach emphasizing a list of patient medical problems that is scrupulously maintained by those caring for the patient. This problem list serves the dual purpose of providing a brief, formal summary of the patient's illnesses and of acting as a tool for organizing the routine documentation of the physician's decision-making process and the plan for and results of care. The problem-oriented, Computer-based Patient Record (CPR) and the problem list have seen renewed interest as an organizational tool in the recent years [3-10], but most of today's patient records retain a time-oriented structure. The Institute of Medicine report on the CPR [11] recommends that it contain a problem list that specifies the patient's medical problems and the status of each. It mentions advantages to this approach: the problem list can be the central place for clinicians to obtain a concise view of all patients' medical problems; this list facilitates associating clinical information in the record to a specific medical problem; and the problem list can encourage an orderly process of medical problem solving and clinical judgment. The problem list in a problem-oriented patient record also provides a context in which continuity of care is supported, preventing both redundant and peripheral actions [3].
Problem list entries coding
To enable most of these potential advantages, problem list entries in the electronic medical record should be coded, which means that medical problems entered must have a corresponding code in a controlled vocabulary. Medical vocabularies used in problem lists are numerous, ranging from ICD-9-CM [12], to SNOMED [13], the Unified Medical Language System (UMLS®) [14-16], and to locally developed vocabularies [17]. SNOMED-CT has been shown to allow 98.5% coverage of problem list terms [13]. Coding of medical problems may be achieved by manually assigning a code when the problem is entered, or by using NLP techniques to map free-text problem list entries with an appropriate code, as described below. The former method is usually eased by the use of pick lists or search engines [18], both features available in our institution's application for managing the problem list. While tools for structured problem entry provide a simple way to assure the availability of coded problems, using NLP techniques to extract coded data from free text allows the use of natural language as the input medium. Natural language remains the most user-friendly and expressive way of recording information, and the application of NLP can still provide the advantages of coded data.
Medical text mapping to standard vocabularies
Many authors have reported on methods to automatically map clinical text concepts to a standardized vocabulary such as the Unified Medical Language System (UMLS®) [19-26]. An example is MetaMap [22,27], an application developed by the US National Library of Medicine (NLM) as an element of the SPECIALIST™ system [28,29]. MetaMap is used to index text or to map concepts in the analyzed text with UMLS concepts. It returns ranked medical concepts, but no negation detection is performed. Five steps are needed, beginning with noun phrase identification using the SPECIALIST minimal commitment parser [30], followed by variant generation, candidate phrase retrieval, and computation of a score for each candidate by comparing it with the input phrase. The process ends with the mapping and the ranking of candidates using the computed score. MetaMap has been shown to identify most concepts present in MEDLINE titles [31]. It has been used for Information Retrieval [32-34], for Information Extraction in biomedical text [31,35], and to extract different types of information like anatomical concepts [36] or molecular binding concepts [37]. It has also been used with electronic messages submitted by patients to automatically provide relevant health information to the patients [24]. Finally, the system was able to extract the most critical findings in 91% of the documents in a previous study done on pathology reports [38].
Another approach to mapping concepts in text to UMLS concepts is found in an application called IndexFinder. IndexFinder functions by generating all valid UMLS concepts by permuting the set of words in the text, and selecting relevant concepts using syntactic and semantic filtering [26].
Negation detection algorithms
The techniques described above can find medical problems, but they are not designed to detect negation. MetaMap, in particular, does not do negation detection. In the medical domain finding negatives is essential due to the fact that findings and diseases are often described as absent.
Several independent negation detection algorithms have been developed. These include NegEx, a computationally simple algorithm using regular expressions [39], and the more complex general-purpose Negfinder [40]. These algorithms have been evaluated and have shown good results. NegEx has shown a sensitivity of 94.5% and a specificity of 77.8% [39], and Negfinder demonstrated a sensitivity of 95.3% and a specificity of 97.7% [40]. After its evaluation, NegEx was updated to a version 2, available on the main author's website [41].
Natural language processing in medicine
The patient record contains a considerable amount of information in addition to problem list entries. However, much of the recorded clinical information is unstructured text, also called free-text. These free-text documents represent patient history and reports of therapeutic interventions or clinical progress and make up a substantial part of the medical record. The increasing use of encoded data in decision support and the requirement for standard medical data sets creates a need for coded information instead. As a possible answer to this issue, NLP can convert narrative text into coded data, and therefore extend the use of the CPR [42].
Several groups have evaluated techniques for automatically encoding textual documents from the medical record. The Linguistic String Project [43,44] has developed a series of tools for analyzing medical text. X-ray reports appear to be an especially fertile ground for NLP. Two groups have developed systems whose focus is the radiologist's report of the chest x-ray. Zingmond [45] has applied a semantic encoding tool to these reports to recognize abnormalities that should receive follow-up, and Friedman has studied techniques for encoding interpretations found in these reports [46-50]. In addition, Friedman and her colleagues have studied NLP in mammography reports [51], neuroradiology reports [19], discharge summaries [52], and pathology reports [53]. Good performance was demonstrated. The application developed, called MedLEE (Medical Language Extraction and Encoding system) [54], has also been used to help develop and maintain vocabularies [55], and recently adapted to extract UMLS concepts from medical text documents, achieving 83% recall and 89% precision [56]. MedLEE was also the first biomedical NLP system that was applied to data in an institution different than the one where it was developed. This resulted in a small drop in performance, but after making some adjustments, it performed as well as in the original institution [49].
The understanding of natural language has been a topic of interest for our Medical Informatics group at the LDS Hospital and the University of Utah (Salt Lake City) for a number of years. SPRUS (Special Purpose Radiology Understanding System) [57,58] was the first NLP application developed at the Medical Informatics group at the University of Utah, and was only semantically driven. Later came SymText (Symbolic Text processor) [59-61], with syntactic and probabilistic semantic analysis. The latest version is called MPLUS [62], provides interleaved syntactic analysis based on a context-free grammar with a bottom-up chart parser, interleaved with object-oriented semantic analysis.
SymText and its successor, MPLUS, make extensive use of semantic networks for semantic analysis. These networks are implemented as Bayesian networks (also called belief networks) [63], trained to infer probabilistic relationships between extracted terms and their meaning. This approach has the advantages of being tolerant to noisy data, and of allowing training to refine performances.
A key realm for testing various approaches has been in the Radiology Department. Here we have focused on reports for chest radiographs [58-60] with a further focus on the data in these reports that supports the diagnosis of pneumonia [64,65]. Admit diagnoses [66] and reports describing the results of pulmonary perfusion scans have also been used to test these NLP approaches [67]. For the project described in this paper, MPLUS has been adapted and trained to extract medical problems from electronic free-text documents [68].
Medical document models
Research into modeling medical documents has been a focus in the development of standards for the electronic medical record. A central example is the first ANSI-approved healthcare standard: the HL7 Clinical Document Architecture (CDA) [69]. It uses XML (eXtensible Markup Language) [70] to facilitate the exchange of documents between users. XML is a data modeling toolkit, a configurable vehicle for any kind of information, and an evolving open standard consistent with a variety of Internet applications. It can store and organize most kinds of data, offers many ways to check the quality of documents, and is easy to read and parse by humans and programs alike.
Several examples of specific medical documents exist. In one case, a successful prototype of a CDA-based, structured discharge summary system was implemented for use in the clinical and community environments of a family practice [71]. In Germany discharge and referral information has been exchanged between Hospital Information Systems and Physician Office Systems in the SCIPHOX project [72]. And in other medical settings this formalism has been used to format discharge letters [73,74].
Regular expressions
In this project, we have also made heavy use of tools to process regular expressions (sometimes abbreviated as RE, regexp or regex). These are templates that describe a whole set of strings, according to certain syntactic rules. Regular expressions are used by many text editors and utilities to search bodies of text for certain patterns and, for example, replace the found strings with a certain other string. Regular expressions trace back to the work of Stephen Kleene, a mathematician, who described these models using his mathematical notation called "regular sets" [75]. His work was used to develop text-manipulation tools on the Unix platform (including ed [76] and grep) as well as in many other scripting languages, editors, programming environments, and specialized tools such as: expr, awk, Emacs, vim, lex, Perl, Java™ and Microsoft® Word. Regular expressions are well described in many publications [77].
Methods
Targeted medical problems
The first step in the process of developing a NLP application to support a problem list was to identify the medical problems for which the system would be responsible. Since the focus of this project has been in the cardiovascular domain, we developed the problem list application with an emphasis on the patient population seen there. Problems appropriately entered into the problem list include both simple observations, representing basic abnormalities noted during a patient's workup (ex: dyspnea, hyperkalemia), and interpretive statements that may represent either syndromic descriptions (ex: Respiratory Failure) or the etiologies for the simple abnormalities that may appear there (ex: Myocardial Infarction as an etiology for Chest Pain). Simple medical problems, whose etiology is not yet understood, are included as they are recognized. Etiologic problems are entered as the physician becomes convinced that they represent a reasonable interpretation of the patient's condition.
For our prototype, we focused on a limited set of 80 medical problems. These problems are listed in Table 1. They were selected for their frequency of use in our institution and in the field of evaluation of our system (cardiovascular). Two sources of information were used to create this list: the log of all coded concepts stored in the central clinical database, called Clinical Data Repository (CDR) in our institution, and a list of the top 25 diagnoses for cardiovascular patients at the LDS Hospital (Salt Lake City, Utah). The first author extracted all concepts coded as medical problems from the log, and ordered them by number of uses during the year 2002. The most common general diagnoses (37 of them) and cardiovascular diagnoses (34 of them) were selected, and 9 of the top 25 cardiovascular diagnoses at the LDS Hospital were added to this list. The resulting list was finally submitted to two expert physicians (a board-certified Internal Medicine specialist and a board-certified Critical Care specialist with more than 20 years of experience) for validation. The second expert is the head of one of the departments where our system will be evaluated at the LDS Hospital.
Table 1 List of the selected medical problems. Origin of the problems: 37 from most frequent diagnoses at IHC°, 9 from the field of evaluation (LDS Hospital 7th floor)*, and 34 from other IHC diagnoses (cardiovascular).
Anemia Hyperlipidemia°
Angina° Hypertension°
Anxiety° Hypotension
Aortic stenosis° Hypothyroidism°
Aortic valve insufficiency Hypovolemia*
Arrhythmia° Infectious Endocarditis°
Arthralgias° Ischemic Heart Disease*
Arthritis° Left bundle branch block
Asthma° Left ventricular hypertrophy
Atrial fibrillation° Melena
Atrial Septal Defects Mitral stenosis°
Back pain° Mitral valve insufficiency
Cancer° Myocardial Infarction°
Cardiac arrest Obesity°
Cardiogenic Shock° Pain°
Cardiomyopathy Paroxysmal supraventric. tachycard.
Cerebrovascular accident° Peptic Ulcer*
Coma Pericardial tamponade
Congenital heart disease Pericarditis*
Constipation° Peripheral vascular disease°
Coronary artery disease° Pneumonia*
Deep vein thrombosis Pneumothorax*
Depression° Pulmonary edema
Diabetes mellitus° Pulmonary embolus*
Diplopia° Pulmonary hypertension
Dysphagia° Renal insufficiency°
Dyspnea Restless legs
Emphysema° Rheumatic heart disease
Epistaxis Right bundle branch block
Fatigue° Septicemia
Gastroesophageal reflux° Sinusitis°
Gout° Streptococcal sore throat
Headache° Syncope*
Heart block Tobacco Use Disorder
Heart failure* Tricuspid valve insufficiency
Heart Murmur Urinary tract infection°
Hematemesis Varicose veins
Hematuria° Venous insufficiency
Hemoptysis Ventricular ectopic beats
Hypercholesteremia° Wheeze
To estimate the proportion of patient's coded diagnoses covered by our set of targeted problems, we retrieved ICD-9-CM codes assigned for administrative purposes to all cardiovascular adult inpatients during 2003 in our institution (LDS Hospital, Salt Lake City, Utah), and compared it with our set of problems. ICD-9-CM codes were selected because they were the only coded medical problems available for cardiovascular inpatients at the LDS Hospital at this time.
Automated Problem List system
The Automated Problem List (APL) system was designed to extract potential medical problems from free-text medical documents, and uses NLP to achieve this task. It is constructed of two main components: a background application and the problem list management application.
The background application does all text processing and analysis and stores extracted medical problems in the Clinical Data Repository. These medical problems can then be accessed by the problem list management application integrated into our Clinical Information System. Clinicians use this application to access and confirm the proposals of the APL system. The problem list management application allows the viewing, editing and creation of medical problems, and gives access to Internet-based medical knowledge sources allowing rapid review of medical facts for each encoded problem. This latter feature is called "infobutton" [78]. The medical problems extracted by our background application are listed in this application, and link back to a customized view of the source document(s). This display of the document(s) that each problem was extracted from helps users of the problem list determine if a medical problem should be part of the "official" problem list.
Information model for medical documents and problems
The information model used by our system was created to ease the exchange of data between the background application and the problem list management application. It models medical documents and medical problems.
Medical documents are represented using HL7's Clinical Document Architecture (CDA) standard [69], with detected medical problems coded as Observations. CDA XML Schemata are therefore used for validation (Figure 1). The XML format allows us to link the extracted medical problem back to its source sentence(s): We use it to display the document(s) a medical problem was extracted from, with the sentence(s) containing mentions of the problem highlighted for easier reading. Medical problems are represented using an information model currently associated with the Clinical Information System in use at our institution: the Clinical Event model [79], instead of the custom model implemented in XML and described in a previous publication [68]. The Clinical Event model is implemented in ASN.1 [80].
Figure 1 Information model diagram. Analysis of free-text documents results in the creation of a CDA version of the analyzed document and of an ASN.1 Problem record for each medical problem detected.
Background application
The application for processing clinical documents runs in the background and follows the sequence of steps depicted in Figure 2 and explained below.
Figure 2 Automated Problem List system diagram. The two main components of the Automated Problem List system (the background application and the problem list management application) are displayed with the elements of our clinical information system they interact with. The ASN.1 data model used is called MultiMedia, with a GenericBigXMLTextObs data type. It allows storage of XML files as simple text but recognized as XML.
In our organization, all data stored in the central database (CDR) pass through an event monitor called the "data driver". This tool recognizes medical events that may require additional processing and allows applications to subscribe to these events and to receive notification of their arrival and storage in the database. A subscription was created to route notification of free-text clinical documents to our system. The notification includes related information such as the patient identifier. We use the patient identifier to determine whether the patient is in the targeted group of cardiovascular patients. If this is the case, the corresponding document is retrieved from the CDR. The document processing phase then begins, starting with section and sentence detection.
Section detection
Sections are main paragraphs of the document, sharing a common class of information, like the "History of present illness", the "Physical examination", the "Family history", and many others. To develop this processing, we analyzed 200 documents randomly selected from a set of 5271 documents from cardiovascular adult inpatients at the LDS Hospital in 2002. Section titles were recognized using specific regular expressions. For example, in some "Progress Notes", section titles start with an uppercase letter after a carriage return and a space, and are followed by two colons. In some "Surgery Reports", all section titles are in uppercase letters, following a carriage return, and ending with a colon. The resulting section detection algorithm was implemented using regular expressions, and first used on this set of 5271 medical text documents to extract all possible section titles. 539 different titles were recognized and manually mapped to a list of 20 generic section titles. This latter step was required because we wished to use a version of the NLP module designed to integrate the title of a section as context in recognizing medical problems.
Sentence detection
After a section title is detected, the text following it until the next section title is extracted as the section's textual content. This text is then split into sentences using a sentence detection algorithm also implemented using regular expressions. The same 200 documents mentioned above and generic sentence boundary information described in another publication [81] were used to define the required regular expressions.
Before applying the sentence detection algorithm, some data cleaning was needed: white spaces (space, tabs, carriage return, etc.) before the first letter or number character and after the last one were removed. The regular expressions were then used to split the section text into sentences. Text was split at periods, exclamation and interrogation marks, and at white spaces, when those were preceded and followed by some specified characters, as in the following example:
(?<=[0-9a-z\.][0-9a-z%"])\.(?=[ \n\r]+[A-Z0-9])
This regular expression splits text at periods preceded by one number (0–9), lowercase letter (a-z), or period (.), this character being followed by one number, lowercase letter, percent sign (%), or quote ("). The period must be followed by one to many spaces or carriage returns, the latter being followed by one uppercase letter (A-Z) or number. For all twelve regular expressions used, see [Additional file 1].
A small pilot evaluation of our section and sentence detection algorithms was executed to determine their effectiveness in the clinical text documents planned for our subsequent studies. For this small evaluation, 20 documents from the test set described above were randomly selected, and sections and sentences were determined by the first author to build the reference standard. Each document was analyzed three times, in different random order, and at an interval of at least 3 days. Sections and sentences determined by the majority of the three reviews were finally used as reference standard. The section and sentence detection algorithms were then applied to these documents and results compared with the reference standard.
Natural language processing module
After the algorithms described above split the document's text into sections and sentences, each sentence is passed to the NLP module with context information including the document type and the section descriptor. The NLP module then analyzes each sentence and extracts potential medical problems, inferring the state of the problem using this contextual information. For example, a problem found in the Family History section will be stated as absent in the patient, if not found in another section of the document. At the sentence level, priority is given to the state present if the same problem is found more than once in the sentence. Thus, if the same problem is found once absent and once present in the same sentence, it is categorized as present at this sentence level. The priority of the state present is justified by examples like "the patient is known for angina, but hasn't suffered from anginal pain for more than two years". The first mention of the problem is stated as present, and the second as absent, and it will finally be considered present, since mention of this problem in the problem list with an inactive status is desirable.
At the document level, priority is given to the state absent. This means that if the same problem is found present in one sentence and absent in another, it will finally be categorized as absent in the document, unless the absent problem reference was found in the Family History section. Medical problems are often mentioned more than once in a document, and their states usually match, but examples of multiple mentions of the same problem with different states are common in documents like Discharge Summaries. A problem could be mentioned in the patient medical history, and then negated at the end of the document, when explaining that the problem was successfully treated during the hospitalization and is now absent. In this case, no mention of the problem in the problem list is desirable (considering that this type of document is analyzed by our system after discharge of the patient).
The NLP module was developed in two different versions. The first version was based on the NLP application developed locally and called SymText. This application is designed to do syntactic and semantic analysis, the latter using Bayesian Networks. It is trainable for different contexts and the semantic part was adapted to accommodate the clinical documents and medical concepts necessary to identify medical problems (e.g. history and physical, surgery report, consultation note, etc...). Training for this tool applies principally to the semantic model. Its semantics are represented as collections of Bayesian Networks representing the relationship between the words and phrases in a sentence and the concepts that they represent. Once the structure of the network is defined, the relationships among the semantic elements are captured as tables of probabilities. The Bayesian Network used by our application was made of 11 nodes and is displayed in Figure 3, with nodes corresponding to the word(s) or phrase(s) in the text at the bottom, and related concepts at the top.
Figure 3 Medical problems Bayesian Network. Bayesian Network with example values for each node when analyzing the sentence "The patient presents with shortness of breath" in the "History of Present Illness" section of a "Consultation Note". Note the application of within-document context represented by the Document Type and Document Section nodes.
To capture these relationships, training cases are needed. We created those training cases using a semi-automated technique depicted in Figure 4. We started by applying the section and sentence detection algorithms on 3000 free-text documents and 91483 sentences were detected. Those documents were randomly selected from the set of 5271 test documents described above. Regular expressions and a list of phrases representing possible ways of describing each of the 80 targeted problems were then used to select sentences with mentions of the targeted problems. Lexical variants of phrases were partially taken into account by using regular expressions. The phrase table was built using the UMLS Metathesaurus MRCONSO table and a manually built table linking the 80 targeted problem concepts with all related subconcepts (e.g. Right Bundle Branch Block was linked to Incomplete Right Bundle Branch Block, Complete Right Bundle Branch Block, and Other or unspecified Right Bundle Branch Block). All phrases corresponding to the 80 concepts and their subconcepts were first retrieved from the MRCONSO table (6928 phrases). We then did some filtering, removing all phrases containing brackets, angled brackets, commas, forward slashes, squared brackets, dashes, and the words NOS or unspecified (e.g. "Anemia unspecified", "Anemia <1>", "Anemia (disorder)", "Anemia, aplastic", etc.). After removal of duplicates, the final table contained 4570 phrases.
Figure 4 Training cases creation process. Sentences were first selected from the set of sentences resulting from the section and sentence detection of free-text documents. Regular expressions and a list of phrases representing possible ways of describing each of the 80 targeted problems were used for this task. The resulting pre-training cases were then augmented by a human reviewer adding the state and state phrase in each sentence. The resulting file contained 4436 training cases.
Once this algorithm had selected target sentences, each sentence and its corresponding medical problem, along with the document type, the section title, and the sentence text were then proposed to a human reviewer (the first author). He proceeded to extract the word or phrase expressing the state of the problem, and added the state of the problem (present or absent) to the case. A medical problem was considered present if mentioned in the text as probable or certain in the present or the past (e.g. "the patient has asthma"; "past history positive for asthma"; "pulmonary edema is probable"), and considered absent if negated in the text or not mentioned at all (e.g. "this test excluded diabetes..."; "he denies any asthma"). The resulting file contained 4436 training cases. Training cases were prepared in a tab-delimited format as required by for training by Netica [82], the Bayesian Networks processing application used by SymText. A fine-tuning phase followed, using the NLP module to analyze sentences and eventually add training cases to improve the application's accuracy.
After the training, the tables in the NLP application's Bayesian Networks contained a statistical representation of the relationship between the words and phrases in the sentences and the medical problems that we had identified for extraction.
The second version of the NLP module was based on MetaMap Transfer (MMTx) [83], the Java version of MetaMap, and on the negation detection algorithm called NegEx [39]. This version worked in two steps. The first step used MMTx with the default full Metathesaurus data set to detect medical problems. The second step then used NegEx to infer the state of the concept detected, determining whether the concept was absent (negated) or present.
CDA document creation
In addition to detecting problems referenced in medical documents, the system formats the documents according to the CDA standard. After section and sentence detection and NLP analysis, this XML-based CDA version of the source document is created, with each medical problem embedded as an encoded Observation, and with markup of the source sentences to allow the creation of the customized view of the document when a problem list user views the source documents associated with a problem. An example of a CDA document in both its XML form and its customized and rendered HTML (Hypertext Markup Language) form is displayed in Figures 5 and 6.
Figure 5 Example CDA document. XML Clinical Document Architecture version of the analyzed document.
Figure 6 Example rendered HTML version of the document. Example of the customized HTML version of the document, as seen if linked from the problem headache
Document and problems storage
To avoid displaying repeated instances of a previously recognized problem, a patient's problems that are already stored in the CDR are analyzed. If the same problem has already been previously stored, and if this problem is not of "Family History" type, then the medical problem found by our system will not be stored in the CDR. In all other cases, the problem will be stored with a reference to the CDA version of the source documents. This link is required to allow recovery of this source document for display to users in the problem list management application. Finally, the medical problems recognized by the system as present are stored in the CDR, along with the resulting CDA document. These are given the status of "proposed" to distinguish them from problems stored manually in the CDR.
Problem list management application
The problem list management application is the users' interface to the Automated Problem List system. It is the window through which end users interact with the medical problems proposed by the Automated Problem List system described above. For this project, an earlier problem list management application was rewritten to take advantage of the proposed problems. This new application for managing the problem list includes additional functionality focused on the proposed medical problems. It is designed to prompt the user to consider adding these extracted problems into the problem list.
Our clinical information system's user interface, called the Clinical Desktop, offers secured access to clinical data through specialized modules like the "Patient search", "Labs", "Medications", or "Problems" module.
The "Problems" module is our problem list management application. Features already present allow viewing, modifying, and adding medical problems in the problem list. An "infobutton" also allows access to medical knowledge relevant to the problem listed [78]. Filters control the display of medical problems based on their status (active, inactive, resolved, error, or proposed) and other personal preferences.
Functions added to take advantage of the proposed problems include the capability to list these problems with a new "proposed" status, and the provision of a link back to the source document to allow viewing of the document from which the problem was extracted. Human intervention is required to officially add a medical problem to the problem list, but for problems mentioned in the clinical documentation, addition is guided and simplified by this extended interface. To accept a problem, the user simply changes the status of the problem from "proposed" to "active" or "inactive", and to reject it, changes the status to "error". A "source button" has been added to each proposed problem listed, giving access to a viewer pop-up window displaying the source document with the source sentences of the medical problem highlighted in red (Figure 7). The user can locate the sentences and read them in a few seconds to determine whether the automatic system correctly proposed a problem for addition. To this end, the CDA version of the source document is retrieved from the CDR and transformed to a customized HTML version of the document. After preprocessing, an XSLT (XML Stylesheet Transformation) [84] stylesheet can be used to generate the HTML version of the document.
Figure 7 Screenshot of the problem list management application. Problem list management application with the viewer window showing the source document of the problem headache with the source sentence highlighted in red.
Preliminary processing of the XML document is required due to some missing features in XSLT, the most striking being an inability to change variables' values after their declaration and instantiation. This latter feature is needed to link the encoded Observation elements with the corresponding content elements in CDA documents. This preliminary phase therefore uses some more advanced XML manipulation features provided by a library for working with XML and XSLT on the Java™ platform. It searches the CDA document for Observation elements with codes corresponding to the medical problem's code. If Observation element(s) are found, corresponding content elements in the text element of the same section are searched and their tag name changed from content to bold when found, as described in Figure 8. The next step is the XSLT transformation, using the stylesheet included as [Additional file 2]. The whole retrieval and transformation process is fast, taking less than a second to retrieve, transform, and display the source document.
Figure 8 Preliminary XML manipulation example. In this extract of a CDA document, the code of the Observation element (dyspnea problem) and the reference identifiers are in bold characters. Reference identifiers link Observation elements (i.e. coded problems) to content elements (i.e. sentence(s) they were extracted from).
Results
Targeted medical problem set coverage
To estimate the proportion of coded diagnoses covered by our set of 80 targeted problems, all unique ICD-9-CM codes assigned to cardiovascular adult inpatients during 2003 in our institution (LDS Hospital, Salt Lake City, Utah) were retrieved. A total of 1531 different codes had been used. Our set of 80 targeted problems therefore covered only 5.2% of all possible codes. In the data set we found a total of 24160 ICD-9-CM code instances, and 15449 of them corresponded to one of our 80 targeted problems. Our set of problems therefore covered 63.9% of all code instances.
Section and sentence detection
A small pilot evaluation of the section and sentence detection algorithms showed good performance. The section detection algorithm reached a sensitivity of 1.00 and a positive predictive value of 1.00. It detected all 146 sections present in the 20 randomly selected documents from our dataset, and detected only those 146 sections. With the sentence detection algorithm, a sensitivity of 0.889 (95% confidence interval 0.78-0.998) and a positive predictive value of 0.946 (0.907-0.985) were measured. 687 out of 731 sentences present in the test set of 20 documents were detected. Of those 687 sentences, 662 were correctly detected (i.e. true positives), and 25 were false positives.
Discussion
The development and functions of the Automated Problem List system have been reported in this paper, along with results of two preliminary evaluations: The coverage of the set of 80 targeted problems was only about 5% of all possible diagnoses coded in ICD-9-CM, but about 64% of all instances of these coded diagnoses. The section and sentence detection algorithms performed well, with a sensitivity and positive predictive value of 100% when detecting sections, and a sensitivity of 89% and a positive predictive value of 94% when detecting sentences.
Some potential issues concerning this system have to be discussed. A key issue is the level of use of the problem list. Our goal is a system that eases the effort of maintaining an accurate, up-to-date problem list. However, our system can only be beneficial if users feel a commitment to maintain a problem list.
In the environment represented by the electronic medical record, there are much improved incentives to do so. Electronic implementation helps to mitigate key reasons for a lack of proper problem list maintenance. With the list available in one place, the EMR, rather than independently in each of the sites where the patient receives care, there is a significant incentive to both maintain and consult this document.
Another issue could be insufficient accuracy in the NLP portion of our system, with deficient sensitivity, positive predictive value, or speed. A sufficient sensitivity is required to significantly improve the quality of the problem list. We are seeking a sensitivity higher than 80%. A sufficient positive predictive value is also desirable to avoid overloading the collection of proposed problems with false positives. Those incorrectly proposed medical problems could make the use of the APL slower by requiring the user to reject an excessive number of incorrectly proposed problems. We are seeking a positive predictive value higher than 60%.
A last issue, which might effect our evaluation of this system, would be speed with which proposed medical problems are returned from the target documents. Our focus here is on the speed of the NLP module. This is clearly the slowest part of our system, requiring heavy computing power to allow deep text analysis. The evaluation of the NLP module will allow us to select the version to use (i.e. based on SymText or on MMTx/NegEx) for subsequent use and evaluation in a clinical setting. We will continue to study the issues of speed, along with scalability and accuracy as we evolve the system to manage problems beyond the cardiovascular domain.
In the future, the NLP module will be evaluated first, and, if satisfactory, the system described above will be evaluated in a clinical setting to determine its effectiveness in guaranteeing the accuracy, completeness, and timeliness of the medical problem list. We expect an increased proportion of correct problems, a reduced proportion of incorrect problems, and a reduced time between problem identification and addition to the problem list.
Coverage of instances of problem mentions in the free-text medical documents by the set of UMLS-derived phrases described in the Methods section could affect the accuracy of the NLP module, but this will not be evaluated in the future. We will focus on the bottom line question: performance of the NLP module when detecting medical problems.
The proposed Automated Problem List will be beneficial for many reasons: A better problem list should improve patient outcomes and reduce costs by reducing omissions and delays, improving the organization of care, and reducing adverse events. It will enhance decision-support for applications that require knowledge of patient medical problems to function optimally. A timely and accurate problem list should improve patient safety, an important and timely issue that has received substantial attention since the 1999 Institute of Medicine report [85].
Conclusion
We have developed an Automated Problem List management system using NLP to extract potential medical problems from free-text documents in a patient's EMR. This system's goal is to improve the problem list's quality by increasing its completeness, accuracy and timeliness. By encouraging the use of a problem list of good quality, this system could potentially improve patient outcomes and security, improve the organization of care, reduce costs, and reduce adverse events.
The medical problem list figures prominently in our plans for computerized physician order entry and medical documentation in the new Clinical Information System (HELP2) currently under development at IHC. A well-maintained list will significantly enhance HELP2's applications. We believe that this clinical tool, which has been taught in medical schools and used sporadically in medical practice for decades, will become a key component for managing clinical information in systems that are developed to provide a longitudinal view of the health record.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SM has conceived the Automated Problem List, developed it with the help of some acknowledged programmers, and implemented it. SM also has drafted this manuscript. PJH has proposed the general design and aim of the project, has guided its development and implementation, and has critically revised this manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
Sentence detection algorithm regular expressions Text file listing all regular expressions used for sentence detection.
Click here for file
Additional File 2
XSLT stylesheet Stylesheet used to transform the CDA version of the medical problem source document into its customized HTML version for viewing in the problem list management application.
Click here for file
Acknowledgements
This work is supported by a Deseret Foundation Grant (#444).
We would like to thank Lee Christensen for his work on MPLUS and SymText, and Min Bowman for her help with the modified Problems module. We would also like to thank Greg Gurr for his advices and his help. Finally, Scott Narus and Stan Huff also gave us helpful advice and guidance for which we are grateful.
==== Refs
Weed LL Medical records that guide and teach N Engl J Med 1968 278 593 600 5637758
Weed LL Medical records that guide and teach N Engl J Med 1968 278 652 657 concl 5637250
Bayegan E Tu S The helpful patient record system: problem oriented and knowledge based Proc AMIA Symp 2002 36 40 12463782
Campbell JR Payne TH A comparison of four schemes for codification of problem lists Proc Annu Symp Comput Appl Med Care 1994 201 205 7949920
Campbell JR Strategies for problem list implementation in a complex clinical enterprise Proc AMIA Symp 1998 285 289 9929227
Donaldson MS Povar GJ Improving the master problem list: a case study in changing clinician behavior QRB Qual Rev Bull 1985 11 327 333 3935996
Elkin PL Mohr DN Tuttle MS Cole WG Atkin GE Keck K Fisk TB Kaihoi BH Lee KE Higgins MC Suermondt HJ Olson N Claus PL Carpenter PC Chute CG Standardized problem list generation, utilizing the Mayo canonical vocabulary embedded within the Unified Medical Language System Proc AMIA Annu Fall Symp 1997 500 504 9357676
Goldberg H Goldsmith D Law V Keck K Tuttle M Safran C An evaluation of UMLS as a controlled terminology for the Problem List Toolkit Medinfo 1998 9 609 612 10384527
Hales JW Schoeffler KM Kessler DP Extracting medical knowledge for a coded problem list vocabulary from the UMLS Knowledge Sources Proc AMIA Symp 1998 275 279 9929225
Starmer J Miller R Brown S Development of a Structured Problem List Management System at Vanderbilt Proc AMIA Annu Fall Symp 1998 1083
Institute of Medicine (U.S.) Dick RS, Steen EB, Detmer DE Committee on Improving the Patient Record The computer-based patient record: an essential technology for health care 1997 Rev Washington, D.C.: National Academy Press
Scherpbier HJ Abrams RS Roth DH Hail JJ A simple approach to physician entry of patient problem list Proc Annu Symp Comput Appl Med Care 1994 206 210 7949921
Wasserman H Wang J An Applied Evaluation of SNOMED CT as a Clinical Vocabulary for the Computerized Diagnosis and Problem List Proc AMIA Symp 2003 699 703 14728263
Unified Medical Language System (UMLS)
Payne T Martin DR How useful is the UMLS metathesaurus in developing a controlled vocabulary for an automated problem list? Proc Annu Symp Comput Appl Med Care 1993 705 709 8130567
Goldberg H Hsu C Law V Safran C Validation of clinical problems using a UMLS-based semantic parser Proc AMIA Symp 1998 805 809 9929330
Zelingher J Rind DM Caraballo E Tuttle M Olson N Safran C Categorization of free-text problem lists: an effective method of capturing clinical data Proc Annu Symp Comput Appl Med Care 1995 416 420 8563314
Wang SJ Bates DW Chueh HC Karson AS Maviglia SM Greim JA Frost JP Kuperman GJ Automated coded ambulatory problem lists: evaluation of a vocabulary and a data entry tool Int J Med Inf 2003 72 17 28 10.1016/j.ijmedinf.2003.08.002
Elkins JS Friedman C Boden-Albala B Sacco RL Hripcsak G Coding neuroradiology reports for the Northern Manhattan Stroke Study: a comparison of natural language processing and manual review Comput Biomed Res 2000 33 1 10 10772780 10.1006/cbmr.1999.1535
Tuttle MS Olson NE Keck KD Cole WG Erlbaum MS Sherertz DD Chute CG Elkin PL Atkin GE Kaihoi BH Safran C Rind D Law V Metaphrase: an aid to the clinical conceptualization and formalization of patient problems in healthcare enterprises Methods Inf Med 1998 37 373 383 9865035
Cooper GF Miller R An experiment comparing lexical and statistical method for extracting MeSH terms from clinical free text J Am Med Inform Assoc 1998 5 62 75 9452986
Aronson AR Effective mapping of biomedical text to the UMLS Metathesaurus: the MetaMap program Proc AMIA Symp 2001 17 21 11825149
Nadkerni P Chen R Brandt C UMLS concept indexing for production databases: a feasibility study J Am Med Inform Assoc 2001 8 80 91 11141514
Brennan PF Aronson AR Towards linking patients and clinical information: detecting UMLS concepts in e-mail J Biomed Inform 2003 36 334 341 14643729 10.1016/j.jbi.2003.09.017
Huang Y Lowe H Hersh W A pilot study of contextual UMLS indexing to improve the precision of concept-based representation in XML-structured clinical radiology reports J Am Med Inform Assoc 2003 10 580 587 12925544 10.1197/jamia.M1369
Zou Q Chu WW Morioka C Leazer GH Kangarloo H IndexFinder: A Method of Extracting Key Concepts from Clinical Texts for Indexing Proc AMIA Symp 2003 763 767 14728276
Aronson AR Bodenreider O Chang HF Humphrey SM Mork JG Nelson SJ Rindflesch TC Wilbur WJ The NLM Indexing Initiative Proc AMIA Symp 2000 17 21 11079836
McCray AT Sponsler JL Brylawski B Browne AC The role of lexical knowledge in biomedical text understanding SCAMC 87 1987 IEEE 103 107
McCray AT Extending a natural language parser with UMLS knowledge Proc Annu Symp Comput Appl Med Care 1991 194 198 1807586
Rindflesch TC Tanabe L Weinstein JN Hunter L EDGAR: extraction of drugs, genes, and relations from the biomedical literature Pac Symp Biocomput 2000 517 528 10902199
Pratt W Yetisgen-Yildiz M A Study of Biomedical Concept Identification: MetaMap vs. People Proc AMIA Symp 2003 529 533 14728229
Aronson AR Query expansion using the UMLS Metathesaurus Proc AMIA Symp 1997 485 489
Wright LW Hierarchical Concept Indexing of Full-Text Documents in the Unified Medical Language System Information Sources Map Journal of the American Society for Information Science 1998 50 514 523 10.1002/(SICI)1097-4571(1999)50:6<514::AID-ASI6>3.0.CO;2-Q
Pratt W Wassermann H QueryCat: Automatic categorization of MEDLINE Queries Proc AMIA Symp 2000 Los Angeles 655 659 11079965
Weeber M Klein H Aronson AR Mork JG de Jong-van den Berg LT Vos R Text-based discovery in biomedicine: the architecture of the DAD-system Proc AMIA Symp 2000 903 907 11080015
Sneiderman CA Rindflesch TC Bean CA Identification of anatomical terminology in medical text Proc AMIA Symp 1998 428 432 9929255
Rindflesch TC Hunter L Aronson AR Mining molecular binding terminology from biomedical text Proc AMIA Symp 1999 127 131 10566334
Shadow G McDonald C Extracting structured information from free text pathology reports Proc AMIA Symp 2003 584 588 14728240
Chapman WW Bridewell W Hanbury P Cooper GF Buchanan BG A simple algorithm for identifying negated findings and diseases in discharge summaries J Biomed Inform 2001 34 301 310 12123149 10.1006/jbin.2001.1029
Mutalik PG Deshpande A Nadkarni PM Use of general-purpose negation detection to augment concept indexing of medical documents: a quantitative study using the UMLS J Am Med Inform Assoc 2001 8 598 609 11687566
NegEx 2
Spyns P Natural language processing in medicine: an overview Methods Inf Med 1996 35 285 301 9019092
Chi E Lyman M Sager N Friedman C IEEE Database of computer-structured narrative: methods of computing complex relations SCAMC 85 1985 221 226
Sager N Friedman C Chi E The analysis and processing of clinical narrative Medinfo 86 1986 Amsterdam (Holland) 1101 1105
Zingmond D Lenert LA Monitoring free-text data using medical language processing Comput Biomed Res 1993 26 467 481 8243070 10.1006/cbmr.1993.1033
Friedman C Alderson PO Austin JH Cimino JJ Johnson SB A general natural-language text processor for clinical radiology J Am Med Inform Assoc 1994 1 161 174 7719797
Friedman C Hripcsak G Shagina L Liu H Representing information in patient reports using natural language processing and the extensible markup language J Am Med Inform Assoc 1999 6 76 87 9925230
Hripcsak G Friedman C Alderson PO DuMouchel W Johnson SB Clayton PD Unlocking clinical data from narrative reports: a study of natural language processing Ann Intern Med 1995 122 681 688 7702231
Hripcsak G Kuperman GJ Friedman C Extracting findings from narrative reports: software transferability and sources of physician disagreement Methods Inf Med 1998 37 1 7 9550840
Knirsch CA Jain NL Pablos-Mendez A Friedman C Hripcsak G Respiratory isolation of tuberculosis patients using clinical guidelines and an automated clinical decision support system Infect Control Hosp Epidemiol 1998 19 94 100 9510106
Jain NL Friedman C Identification of findings suspicious for breast cancer based on natural language processing of mammogram reports Proc AMIA Annu Fall Symp 1997 829 833 9357741
Friedman C Knirsch C Shagina L Hripcsak G Automating a severity score guideline for community-acquired pneumonia employing medical language processing of discharge summaries Proc AMIA Symp 1999 256 260 10566360
Xu H Friedman C Facilitating Research in Pathology using Natural Language Processing Proc AMIA Symp 2003 1057 14728560
Friedman C Johnson SB Forman B Starren J Architectural requirements for a multipurpose natural language processor in the clinical environment Proc Annu Symp Comput Appl Med Care 1995 347 351 8563299
Friedman C Liu H Shagina L A vocabulary development and visualization tool based on natural language processing and the mining of textual patient reports J Biomed Inform 2003 36 189 201 14615228 10.1016/j.jbi.2003.08.005
Friedman C Shagina L Lussier Y Hripcsak G Automated Encoding of Clinical Documents Based on Natural Language Processing J Am Med Inform Assoc 2004
Ranum DL Knowledge-based understanding of radiology texts Comput Methods Programs Biomed 1989 30 209 215 2684493 10.1016/0169-2607(89)90073-4
Haug PJ Ranum DL Frederick PR Computerized extraction of coded findings from free-text radiologic reports. Work in progress Radiology 1990 174 543 548 2404321
Haug P Koehler S Lau LM Wang P Rocha R Huff S A natural language understanding system combining syntactic and semantic techniques Proc Annu Symp Comput Appl Med Care 1994 247 251 7949928
Haug PJ Koehler S Lau LM Wang P Rocha R Huff SM Experience with a mixed semantic/syntactic parser Proc Annu Symp Comput Appl Med Care 1995 284 288 8563286
Koehler SB SymText : a natural language understanding system for encoding free text medical data 1998
Christensen L Haug P Fiszman M MPLUS: a probabilistic medical language understanding system Proceedings of the Workshop on Natural Language Processing in the Biomedical Domain 2002 29 36
Nivre J Bubenko JjW, Benkt On Statistical Methods in Natural Language Processing Promote IT Second Conference for the Promotion of Research in IT at New Universities and University Colleges in Sweden 2002 Skövde (Sweden): University of Skövde; 684 694
Fiszman M Chapman WW Aronsky D Evans RS Haug PJ Automatic detection of acute bacterial pneumonia from chest X-ray reports J Am Med Inform Assoc 2000 7 593 604 11062233
Fiszman M Haug PJ Using medical language processing to support real-time evaluation of pneumonia guidelines Proc AMIA Symp 2000 235 239 11079880
Haug PJ Christensen L Gundersen M Clemons B Koehler S Bauer K A natural language parsing system for encoding admitting diagnoses Proc AMIA Annu Fall Symp 1997 814 818 9357738
Fiszman M Haug PJ Frederick PR Automatic extraction of PIOPED interpretations from ventilation/perfusion lung scan reports Proc AMIA Symp 1998 860 864 9929341
Meystre S Haug PJ Medical problem and document model for natural language understanding Proc AMIA Symp 2003 455 459 14728214
Dolin RH Alschuler L Beebe C Biron PV Boyer SL Essin D Kimber E Lincoln T Mattison JE The HL7 Clinical Document Architecture J Am Med Inform Assoc 2001 8 552 569 11687563
Extensible Markup Language (XML) 1.0 (Second Edition)
Paterson G Shepherd M Wang X Watters C Zitner D Using the XML-based Clinical Document Architecture for Exchange of Structured Discharge Summaries Proceedings 35 th Hawaii Int Conf on System Sciences 2002
Heitmann K Schweiger R Dudeck J Discharge and referral data exchange using global standards – the SCIPHOX project in Germany Int J Med Inf 2003 70 195 203 10.1016/S1386-5056(03)00036-4
Muller ML Butta R Prokosch HU Electronic discharge letters using the Clinical Document Architecture (CDA) Stud Health Technol Inform 2003 95 824 828 14664090
Bludau H Wolff A Hochlehnert AJ Presenting XML-based medical discharge letters according to CDA Methods Inf Med 2003 42 552 556 14654890
Kleene SC Shannon C, McCarthy J Representation of events in nerve nets and finite automata Automata Studies 1956 Princeton, NJ: Princeton University Press 3 41
Thompson K Regular expression search algorithm Communications of the ACM 1968 11 419 422 10.1145/363347.363387
Friedl JEF Mastering regular expressions 1997 Cambridge: O'Reilly
Reichert JC Glasgow M Narus SP Clayton PD Using LOINC to link an EMR to the pertinent paragraph in a structured reference knowledge base Proc AMIA Symp 2002 652 656 12463904
Huff SM Rocha RA Bray BE Warner HR Haug PJ An event model of medical information representation J Am Med Inform Assoc 1995 2 116 134 7743315
International Organization for Standardization International Standard ISO/IEC 8824: specification of Abstract Syntax Notation One (ASN1) 1990 Second Geneva, Switzerland: International Organization for Standardization
Manning CD Schutze H Foundations of Statistical Natural Language Processing 2003 6 Cambridge, Massachusetts, London, England: MIT Press
Netica™ Application
MetaMap Transfer (MMTx)
XSL Transformations (XSLT) Version 1.0
Institute of Medicine Committee on Quality of Health Care in America Kohn LT Corrigan JM Donaldson MS To Err is Human: Building A Safer Health System 1999 National Academy Press, Washington, DC
|
16135244
|
PMC1208893
|
CC BY
|
2021-01-04 23:52:11
|
no
|
BMC Med Inform Decis Mak. 2005 Aug 31; 5:30
|
utf-8
|
BMC Med Inform Decis Mak
| 2,005 |
10.1186/1472-6947-5-30
|
oa_comm
|
==== Front
BMC Med Res MethodolBMC Medical Research Methodology1471-2288BioMed Central London 1471-2288-5-271611532310.1186/1471-2288-5-27Research ArticleEvaluation of the Total Design Method in a survey of Japanese dentists Nakai Yukie [email protected] Peter [email protected] Toshiko [email protected] Chikako [email protected] Tsutomu [email protected] Department of Behavioral Pediatric Dentistry, Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama, Japan2 Department of Dental Public Heath Sciences, University of Washington, Seattle, USA2005 23 8 2005 5 27 27 28 5 2005 23 8 2005 Copyright © 2005 Nakai et al; licensee BioMed Central Ltd.2005Nakai et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
This study assessed the application of the Total Design Method (TDM) in a mail survey of Japanese dentists. The TDM was chosen because survey response rates in Japan are unacceptably low and the TDM had previously been used in a general population survey.
Methods
Four hundred and seventy eight dentist members of the Okayama Medical and Dental Practitioner's Association were surveyed. The nine-page, 27-item questionnaire covered dentist job satisfaction, physical practice, and dentist and patient characteristics. Respondents to the first mailing or the one-week follow-up postcard were defined as early responders; others who responded were late responders. Responder bias was assessed by examining age, gender and training.
Results
The overall response rate was 46.7% (223/478). The response rates by follow-up mailing were, 18% after the first mailing, 35.4% after the follow-up postcard, 42.3% after the second mailing, and 46.7% after the third mailing. Respondents did not differ from non-respondents in age or gender, nor were there differences between early and late responders.
Conclusion
The application of TDM in this survey of Japanese dentists produced lower rates of response than expected from previous Japanese and US studies.
==== Body
Background
Mail survey questionnaires of dentists as well as the general public have been used widely in the U.S. and response rates are generally high. In contrast the use of mail surveys in Japan has been less successful. Japanese textbooks on social science research techniques report return rates of no more than 20–40% [1-3]. A mail survey conducted by one of the local Japanese dental associations had a response rate of 10% (unpublished data). Mail surveys reported in the Japanese medical literature had response rates ranging from 49 to 90% [4-8]. Research subjects in the various studies were the physicians and residents working at two private University hospitals (Response rate 49.1%) [4], the institutions belonging to an oncology group (Response rate 90.2%) [5], the council members of the Japanese society of child neurology (Response rate 72.8%) [6], ophthalmologists in hospitals and clinics (Response rate 73%) [7], and psychologists (Response rate not given) [8]. However, the publications lack methodological detail. Only two of the five, for example, provide the source of the mailing lists. In two of the surveys, questionnaires were sent to a representative at each hospital or institution rather than to individuals directly [4,5]. None of the five papers indicated whether the studies were sponsored by a professional association, or university or other group. One of five publications indicated that an advance letter was sent before the questionnaire [7]. Only one paper specified whether participants were told how the data would be used [7]. None of the papers explained whether an incentive was included in the mailing of the questionnaire. Other details generally missing were the length of the questionnaire (missing in 2/5) [4,6,8], telephone contacts for more information (missing in all 5) or assurance of confidentiality (missing in all 5).
The Total Design Method (TDM), which was developed by Dillman and includes personalization of the cover letter and repeated follow-ups, was designed to achieve high response rates and minimize the potential influence of systemic nonresponse bias [9]. The response rate generally is lower in surveys of the general public and higher in surveys of professionals although this varies by group and subject. Locker and colleagues reported a 71.6% response rate when an oral health questionnaire using the TDM was used to survey the general population from voters' lists [10]. Fiset and colleagues mailed questionnaires concerning dental malpractice claims to dentists using the TDM, and reported a 69.6% response rate [11].
In the only application of the TDM in Japan to date, Jussaume and colleagues reported a 55.6% response rate for a survey of the general population on the subject of 748 when those surveyed were selected from telephone listings [12].
No work has been done on adapting the TDM to Japanese dental populations. The aim of this study was to assess the application of the TDM in a mail survey of Japanese dentists.
Methods
Subjects
The questionnaire was mailed to all 482 dentist members on Okayama Medical and Dental Practitioner's Association list. Out of 482 questionnaires sent out, four dentists were excluded because they had closed their office due to sickness or had shared replying survey with a spouse dentist. The final survey population was 478 dentists. Potential subjects were informed in the cover letter that participation in the study was voluntary and that individual responses would be confidential.
Questionnaire development
A nine-page, 27- item questionnaire was designed in English using questions derived from earlier surveys. It covered four categories: 1) dentist job satisfaction, 2) physical practice, 3) dentist and 4) patient characteristics. Instrumentation was translated from English to Japanese by a native speaker, and then back-translated by another native speaker to ensure comparability to the original English form (see Figure 1). The questionnaire booklet was organized so that easier and less personal questions were asked initially and more difficult or personal questions were asked at the end of the questionnaire. The questionnaire was pretested among the alumni practicing out of Okayama prefecture before use. The questionnaire was formatted into a 182 × 257 mm booklet style to make it appear easier and less time-consuming to complete.
Figure 1 The first page of the questionnaire and its English translation.
Procedures
The Okayama Medical and Dental Practitioner's Association agreed to participate and endorse the study.
The procedures followed were generally those recommended by Jussaume and Yamada [12] who had previously adapted the TDM to Japan. In designing the letters, a strong emphasis was placed on three essential features of the TDM. First, respondents were told how their names were selected, that their responses would represent those of many other Japanese dentists, and that their participation was invaluable. Second, the confidentiality of the survey was emphasized and participants were promised that their names would never be placed on the questionnaire. Finally, as an incentive for participation, a decision was made with the Association that respondents would be offered a report of the results of this study. No personal incentive was included in the survey because Japanese culture values service to the group rather than the individual [12].
Approximately one week before the first mailing of the questionnaire, an advance letter including the Association endorsement was sent to all the dentists introducing the researchers and explaining the importance of the study (see Figure 2). The letters were not personalized and not individually signed. The letter noted that the participant would receive the questionnaire in a couple of days. The envelopes were personally addressed and stamped. In the first questionnaire mailing, the participants received a letter again explaining the importance of the study and assuring confidentiality (see Figure 3), the questionnaire booklet, and a stamped self addressed return envelope. Identification number markers were used on questionnaires so that respondents could be checked off the mailing list. A follow-up postcard (see Figure 4), encouraging participation, was sent about one week later. Three weeks after the first mailing, a replacement questionnaire, a stamped return envelope and a cover letter (see Figure 5) were sent to any dentists who had not responded. Dentists who did not respond within six weeks after the original mailing received a cover letter (see Figure 6), a second replacement questionnaire, and a stamped self addressed return envelope.
Figure 2 Advance letter.
Figure 3 Initial letter sent in the first mailing.
Figure 4 Follow-up card.
Figure 5 Second letter sent in the second mailing.
Figure 6 Third letter sent in the third mailing.
Japanese standard number 3 size (235 × 120 mm) envelopes of an light yellow green color were used. Addresses were written on envelopes from left to right in the manner of most Japanese business correspondence.
Data handling and analysis
The data from questionnaires received within four months of the first mailing were entered into a database in Excel 2000 (Microsoft), and were checked for accuracy. Data management and analyses were conducted using SPSS version 11.5.
A two-pronged strategy was used to assess bias. First, age and gender of respondents and non-respondents, provided by the association list, were compared. Second, we compared study variables for early and late respondents. Respondents to the first mailing or the one-week follow-up postcard were defined as early responders; others who responded were late responders. Study variables included age, gender, years in practice, practice satisfaction, practice status, practice location, patient number seen per day, having postgraduate training, total hours of continuing dental education taken for the past 12 months, employment status (owner vs. non-owner), number of practice locations, yearly gross income before any expenses or taxes. T-tests, Fisher's exact test and Chi-square analyses were used to compare differences between the groups to assess respondent representativeness.
Results
Response rate
The overall response rate was 46.7% (223/478). The cumulative response rates by each follow-up mailing are shown in Table 1. Ten dentists declined to participate. The primary reason given for refusal was that the dentist was not comfortable in answering personal questions.
Table 1 Cumulative response rate to Japanese dental questionnaire using the TDM
1st Mailing Follow-up Card 2nd Mailing 3rd Mailing
Response rate 18.0 % 21.2% 10.7% 7.6%
(86/478) (83/392) (33/309) (21/276)
Cumulative 35.4% 42.3% 46.7%
response rate (169/478) (202/478) (223/478)
Respondent representativeness
Respondents did not differ from non-respondents for the gender and age, nor were there differences between early and late responders for any of the 12 of variables that were compared except that the late responders have taken less hours of continuing dental education during the past 12 months (17.5 vs. 29.2 hours; t = 1.95, p = 0.05) (Table 2). There was a trend for late responders to be more likely to have received postgraduate training (27.3% vs. 16.3%; Fisher's exact test, p = 0.08)
Table 2 Responses on study variable for early/late responders
Variables N Early responders Late Responders p
Mean age (SD) 223 45.6 (9.8) 45.6 (10.9) NS
% female 223 9.5 7.4 NS
Mean months in practice (SD) 220 163 (117) 172 (129) NS
Practice satisfaction (% dissatisfied) 215 24.7 22.6 NS
Busyness (%not busy enough) 220 28.7 30.2 NS
Practice location (% patients from rural areas) 212 28.7 29.2 NS
Patient visits/day (mean, SD) 212 34.5 (25.6) 32.2 (26.2) NS
Postgraduate training (%no) 215 25.5 15.1 0.08
Total hrs CDE/12 mos (Mean, SD) 197 29.2 (58.7) 17.5 (24.9) 0.05
Employment status (% non owner) 219 15.7 13.2 NS
>1 practice location 214 4.9 7.8 NS
Annual gross income before expense/taxes (% less than ¥30,000,000 194 23.0 34.8 NS
Year of graduation (mean, SD) 212 1981 (10.7) 1981 (11.7) NS
Discussion
The TDM, as generally adapted by Jussaume and Yamada [12], was used in a survey of Japanese dentists. Previously Jussaume and Yamada obtained nearly identical response rates when they surveyed the general public in Japan (55.6%) and the U.S. (57.5%) using this method. The application of the TDM in this survey of dental practice produced a lower response rate (46.6%) than expected but with little response bias. The results of a low response rate (43%) without non-response bias was previously reported in the US dentist population [13] although other studies using this method have produced higher response rates. Dentists can be considered to have sufficiently similar education, income, and interest to be considered a homogeneous group. If there is little difference between the respondents and non-respondents, a smaller percentage of return might be acceptable. Parashos and colleagues, who reported a dentist survey in Australia and New Zealand also using the TDM, found significant differences between early and late respondents in responses to a specific survey question of topical interest despite the absence of differences in the demographic data [14]. They emphasize the importance of using methods to achieve a high response rate to overcome such bias.
The lower than anticipated response rate of this study may have resulted from our failure to follow all aspects of the TDM fully. One of the differences found in procedures between Jussaume's [12] and this study was that neither Inkan (personal seal) nor signature was used to the letter in this study. Jussaume said that it could convey to the respondent the importance which researchers placed on the project. The other difference was that the letter was not written in longhand in this study. Japanese respondents are hypothesized to react more positively to a survey seeing the effort taken to write out their names in longhand. Japanese dentists may also be more reluctant to answer the questions that they feel too personal. Ten dentists had refused to participate in this study due to such reasons. One of the authors asked the primary reasons why they didn't participate to another five dentists. Two of these five dentists said the questions were too personal and that there were too many questions to answer. One dentist said that he didn't want his practice to be compared with others. Two said that topic was not interesting enough to make them want to participate.
The response rate was, however, much greater than that of another unpublished survey of Japanese dentists that achieved a 10 % response rate. The follow-up contact and repeated mailing to non-respondents increased our sample size by more than a quarter (28.7%). Before using the postcard reminder, the response rate was one-third of the final rate, suggesting that follow-up contact is critical to bolstering mail survey response rates. This is consistent with research indicating that follow-up contact has the most positive effect on return rates [11,15-17].
Our results are encouraging, and demonstrate the feasibility of using TDM to study a population of Japanese dentists.
Conclusion
The application of TDM in this survey of Japanese dentists produced lower rates of response than expected from previous Japanese and US studies with little response bias.
List of abbreviations
TDM Total Design Method
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
YN participated in the design of the study, made the instruments (translated English version to Japanese), negotiated with Okayama Medical and Dental Practitioner's Association to be given endorsement, collected the data, performed the statistical analyses, and drafted the manuscript. PM participated in the design of the study and in writing the manuscript. TY participated in the design of the study and developed the English version of the instruments for this paper. CI collected the data and performed the statistical analyses. TS participated in the design of the study. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to thank all participating dentists who responded our mail survey and Okayama Medical and Dental Practitioner's Association who agreed to participate in our mail survey. The study was supported in part by funds from Okayama University.
==== Refs
Hara J Umino M Shakai Chosa Enshu 1984 Tokyo: Tokyo Daigaku Shuppan Kai in Japanese
Igaki S Shakai Chosa Nyumon 1968 Kyoto: Mineruba Shobo in Japanese
Kobayashi S Shakai Chosa Ron 1981 Tokyo: Bunshindo in Japanese
Ota H Tanimoto S Takayanagi K Kimura T Oida T A preliminary study on the knowledge and attitudes of physicians at two university hospitals towards the medical insurance system of Japan Tohoku J Exp Med 2000 190 143 155 10770622 10.1620/tjem.190.143
Konno R Sato S Yajima A A questionnaire survey on current surgical procedures for endometrial cancer in Japan Tohoku J Exp Med 2000 190 193 203 10778803 10.1620/tjem.190.193
Sakakihara Y Ethical attitudes of Japanese physicians regarding life-sustaining treatment for children with severe neurological disabilities Brain Dev 2000 22 113 117 10722963 10.1016/S0387-7604(99)00127-8
Funatsu H Hori S Present status of ophthalmological care for diabetic patients in Japan Jpn J Ophthalmol 2000 44 75 81 10698029 10.1016/S0021-5155(99)00144-6
Oshima A Higuchi T Fujiwara Y Iida M Iwanami A Kanba S Motohashi N Uchitomi Y Yamada K Yamawaki S Questionnaire survey on the prescribing practice of Japanese psychiatrists for mood disorders Psychiatry Clin Neurosci 1999 53 S67 72 10560902
Dillman DA Mail and telephone surveys: The Total Design Method 1978 New York: John Wiley & Son
Locker D Grushka M Response trends and nonresponse bias in a mail survey of oral and facial pain J Public Health Dent 1988 48 20 25 3422313
Fiset L Milgrom P Tarnal J Dentists' response to financial incentives in a mail survey of malpractice liability experience J Public Health Dent 1994 54 68 72 8046692
Jussaume RA Yamada Y A comparison of the viability of mail surveys in Japan and the United States Public Opin Q 1990 54 219 228 10.1086/269199
Hovland EJ Romberg E Moreland EF Nonresponse bias to mail survey questionnaires within in a professional population J Dent Educ 1980 44 270 274 6928881
Parashos P Morgan MV Messer HH Response rate and nonresponse bias in a questionnaire survey of dentists Community Dent Oral Epidemiol 2005 33 9 16 15642042 10.1111/j.1600-0528.2004.00181.x
Fox RJ Crask MR Kim J Mail survey response rate: A meta-analysis of selected techniques for inducing response Public Opin Q 1988 52 467 491 10.1086/269125
Weathers PL Furlong MJ Solorzano D Mail survey research in counseling psychology: Current practices and suggested guidelines J Couns Psychol 1993 40 238 244 10.1037//0022-0167.40.2.238
Gore-Felton C Koopman C Bridges E Thoresen C Spiegel D An example of maximizing survey return rates. Methodological issues for health professionals Eval Health Prof 2002 25 152 168 12026750 10.1177/01678702025002002
|
16115323
|
PMC1208894
|
CC BY
|
2021-01-04 16:32:52
|
no
|
BMC Med Res Methodol. 2005 Aug 23; 5:27
|
utf-8
|
BMC Med Res Methodol
| 2,005 |
10.1186/1471-2288-5-27
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-411603364610.1186/1471-2474-6-41Study ProtocolTesting the effectiveness of an innovative information package on practitioner reported behaviour and beliefs: The UK Chiropractors, Osteopaths and Musculoskeletal Physiotherapists Low back pain ManagemENT (COMPLeMENT) trial [ISRCTN77245761] Evans David W [email protected] Nadine E [email protected] Martin [email protected] Steven [email protected] Alan C [email protected] Tamar [email protected] School of Health and Rehabilitation, Keele University, Staffordshire, UK2 Primary Care Sciences Research Centre, Keele University, Staffordshire, UK3 Centre for General Practice and Primary Care, Barts and The London, London, UK4 Research Centre, The British School of Osteopathy, London, UK5 Institute for Musculoskeletal Research and Clinical Implementation, Bournemouth, UK6 Department of Psychology, Royal Holloway, University of London, London, UK2005 20 7 2005 6 41 41 16 5 2005 20 7 2005 Copyright © 2005 Evans et al; licensee BioMed Central Ltd.2005Evans et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Low back pain (LBP) is a common and costly problem. Initiatives designed to assist practitioner and patient decisions about appropriate healthcare for LBP include printed evidence-based clinical guidelines. The three professional groups of chiropractic, osteopathy and musculoskeletal physiotherapy in the UK share common ground with their approaches to managing LBP and are amongst those targeted by LBP guidelines. Even so, many seem unaware that such guidelines exist. Furthermore, the behaviour of at least some of these practitioners differs from that recommended in these guidelines.
Few randomised controlled trials evaluating printed information as an intervention to change practitioner behaviour have utilised a no-intervention control. All these trials have used a cluster design and most have methodological flaws. None specifically focus upon practitioner behaviour towards LBP patients. Studies that have investigated other strategies to change practitioner behaviour with LBP patients have produced conflicting results. Although numerous LBP guidelines have been developed worldwide, there is a paucity of data on whether their dissemination actually changes practitioner behaviour. Primarily because of its low unit cost, sending printed information to large numbers of practitioners is an attractive dissemination and implementation strategy. The effect size of such a strategy, at an individual practitioner level, is likely to be small. However, if large numbers of practitioners are targeted, this strategy might achieve meaningful changes at a population level.
Methods
The primary aim of this prospective, pragmatic randomised controlled trial is to test the short-term effectiveness (six-months following intervention) of a directly-posted information package on the reported clinical behaviour (primary outcome), attitudes and beliefs of UK chiropractors, osteopaths and musculoskeletal physiotherapists. We sought to randomly allocate a combined sample of 1,800 consenting practitioners to receive either the information package (intervention arm) or no information above that gained during normal practice (control arm). We collected questionnaire data at baseline and six-months post-intervention. The analysis of the primary outcome will assess between-arm differences of proportions of responses to questions on recommendations about activity, work and bed-rest, that fall within categories previously defined by an expert consensus exercise as either 'guideline-consistent' and 'guideline-inconsistent'.
==== Body
Background
Low back pain (LBP) is a common and costly problem [1,2]. As a result, the volume of research produced worldwide on the subject has been increasing steadily for several decades. This has allowed for the formulation of evidence-linked recommendations based on a process of evidence synthesis, expert consensus and summarising evidence in systematically developed statements [3]. Healthcare for acute LBP based on these recommendations can be expected to be safe, effective and more cost-effective than current usual care. Over 70% of patients managed in this way are likely to become pain-free, with a recurrence rate of less than 25% over the course of twelve months [4,5].
With the aim of assisting practitioner and patient decisions about appropriate healthcare for specific clinical circumstances [6], several initiatives have delivered recommendations about LBP to various stakeholder groups, including healthcare personnel, patients and the general public [7-15]. Amongst such initiatives, printed evidence-based clinical guidelines have been developed for primary care and occupational settings, in several countries [16-19].
In the UK, the National Health Service (NHS) Executive commissioned the Effective Clinical Practice Programme of the Royal College of General Practitioners (RCGP) to develop evidence-based clinical guidelines for the management of acute LBP in primary care. These were developed and printed for the use of first contact primary care practitioners, including the professional groups of chiropractic, osteopathy and musculoskeletal physiotherapy in the UK. They were first published in 1996 [20], then revised and updated in 1999 [21]. Since then, the Faculty of Occupational Medicine of the Royal College of Physicians commissioned the development of equivalent multi-disciplinary guidelines for the occupational setting [22,23].
Chiropractors, osteopaths and musculoskeletal physiotherapists in the UK share some common ground with their approaches to managing LBP [24]. As a result, these professions have been considered to be a collective stakeholder group [1,20,21]. Together, they are involved in the management of approximately one-fifth of all cases of LBP in the UK [25,26], making them an important provider of care. Indeed, Maniadakis and Gray [2] estimated that in 1998, the annual combined direct healthcare costs for LBP delivered by these professions was £493 million (US$820 million, based on 1998 exchange rates).
The majority of these practitioners work in private practice, rather than within the NHS [15]. As a result, they are excluded from both the constraints [27-32] and potential support [8,33] offered by working within large organisations of healthcare delivery. Unlike general practitioners [34], few UK guidelines are aimed at chiropractors, osteopaths and musculoskeletal physiotherapists. With LBP accounting for more than half of their workload [15,35-40], recommendations for optimal LBP management should be relevant to these practitioners. Whilst anecdote and some empirical data suggest that many of these practitioners are naïve to the existence of LBP guidelines, they do express a desire for a clear rationale in their LBP clinical decision-making [41]. Several studies show that the clinical behaviour of members of these professional groups diverges from guideline recommendations [38-40,42] and that this may relate to the beliefs of these practitioners [43-49]. It is therefore timely to explore ways to inform and optimise aspects of their clinical practice.
Evidence for interventions to change practitioner behaviour
Healthcare research findings are usually published in peer-reviewed scientific journals. However, this neither ensures that these findings reach practitioners (dissemination) nor that they are used in routine clinical practice (implementation) [50-54]. Dissemination involves raising awareness of research messages whereas implementation involves getting the findings of research adopted into practice [55].
Interventions that aim to change the behaviour of healthcare practitioners essentially fall into two groups: those that use broadly educational approaches and those that use financial or organisational prompts [56]. There is a diverse array of educational interventions [57,58]. 'Active' educational interventions, with which practitioners are directly engaged, are usually regarded as more effective at changing behaviour than 'passive' interventions, which act as vehicles for disseminating information such as behaviour recommendations, and are commonly in a written format [59-62]. Multifaceted interventions are widely thought to be more effective than single interventions, even though empirical evidence does not fully support this opinion [63].
The most recent and thorough systematic review of the evidence for printed information as an intervention to change practitioner behaviour [61] included several randomised controlled trials (RCTs) with a no-intervention control [64-70]. These studies reported mixed results. Notably, all were of a cluster design, with the unit of randomisation for intervention allocation being groups of practitioners, such as regions, practices or clinical teams. Cluster designs have potentially limited external validity since the results may not be applicable to other locations and settings. Most of these RCTs suffered from methodological flaws, with their study designs and analyses failing to allow for effects of clustering [71,72]. Similar methodological problems were identified in a further cluster RCT [73], published after the search period of the review. Of all these RCTs, only one [67] included practitioner behaviour towards LBP patients, however, the intervention was poorly described. In addition, none of these RCTs focused on members of the chiropractic, osteopathy and musculoskeletal physiotherapy professional groups in the UK.
Given the methodological problems associated with cluster RCTs [74-76], it is surprising that they dominate the scientific literature on changing practitioner behaviour with printed guidelines. Indeed, one large RCT that recently attempted to assess guideline implementation using a cluster design had to abandon this comparison because of methodological problems [77]. Furthermore, the practical ease of posting written educational material to individual practitioners, in contrast to more 'active' interventions, should lend itself to a non-cluster RCT. The effectiveness of disseminating printed information on practitioner behaviour is therefore yet to be tested for independent effects on individual practitioners in a RCT. In addition, the impact of printed guidelines on the management of LBP has not been adequately assessed.
To date, studies investigating strategies to change practitioner behaviour towards LBP patients have produced conflicting results. Consistent with studies from other healthcare topics, a cluster RCT by Bekkering et al [78] found a statistically significant positive change in some of the targeted behaviours of Dutch physiotherapists using a multifaceted educational intervention, compared to passive dissemination of printed guidelines alone. The multifaceted intervention consisted of mostly active components, utilising education, discussion, role-playing, feedback and reminders. In another cluster RCT, the same research group produced small changes in some of the targeted behaviours of Dutch GPs [80]. In this study, the multifaceted intervention consisted of active and passive components, including the Dutch LBP guideline for GPs, the LBP guideline for occupational physicians, a 2-hour educational and clinical practice workshop, 2 scientific articles on LBP management, a tool for patient education and a tool for reaching agreement on LBP care with physical, exercise, and manual therapists. In a similar manner, yet another cluster RCT [81] found a small but statistically significant positive behaviour change in groups of US primary care physicians, following an active educational intervention based on an education/audit/feedback model with local peer opinion leaders. Conversely, a cluster RCT in a sample of UK general practitioners showed that an active, tailored, multifaceted educational intervention, including the identification and targeting of specific barriers to behaviour change via outreach visits, failed to demonstrate a statistically significant change in prescribing or referral behaviour [82]. The variability of these results may be partly explained by contextual factors, such as the specific behaviour(s) targeted, the professional group of the practitioners, and the healthcare system and practice setting in which the practitioners work.
Given the expense of active interventions, the comparative low unit cost of sending printed information to large numbers of practitioners makes it an attractive dissemination and implementation strategy. It may be seen as more efficient to adopt a less effective but less costly implementation strategy [61,63,83]. Whilst the effect size of passively disseminated printed information is likely to be small at an individual practitioner level, the potential of the strategy relies on the capacity for small behaviour changes in large numbers of practitioners to have an overall substantive impact upon a healthcare system [84]. Even quite small changes at an individual practitioner level could translate, at a population level, into a worthwhile change in the management of large numbers of patients. This notion may have particular merit for an economically costly healthcare issue such as LBP [2]. Nevertheless, the evident international popularity of printed guidelines for the management of LBP [16-19] ought to be matched by rigorous evaluation through systematic research [52].
We have designed a prospective, pragmatic RCT for a large number of practitioners working predominantly in primary care settings. The primary aim of the study is to test the short-term effectiveness (six-months following intervention) of a directly-posted information package on the reported behaviour and beliefs of a combined sample of chiropractors, osteopaths and musculoskeletal physiotherapists in the UK. We are not aware of any previous implementation research that has utilised a randomly allocated no-intervention control, in which the individual practitioner was the unit of randomisation. Additionally, to our knowledge no research has investigated these issues across these three professional groups.
Our secondary aims are:
i) to describe patterns of reported behaviour, attitudes and beliefs, within and between professional groups at baseline,
ii) to identify baseline factors that are associated with a change in reported behaviour of these practitioners (such as demographic data, attitudes and beliefs of practitioners),
iii) to make recommendations for future research on implementation strategies in these professional groups.
Methods
Trial design
This prospective, pragmatic, two-arm RCT was conducted with practitioners from the three groups of chiropractic, osteopathy and musculoskeletal physiotherapy in the UK. The unit of randomisation and analysis throughout the study was the individual practitioner. This allowed for a nationally representative, non-clustered study population. Participants from each professional group were randomly allocated to receive either a directly-posted information package (intervention arm) or no information over and above that gained during normal practice (control arm). Ethical approval was obtained from the London Multi-centre Research Ethics Committee. We collected data using self-completed questionnaires at baseline and six-months post-intervention. Questionnaire content, format and response rates were tested in a pilot survey (n = 150).
Study population
Participants were members of the professional groups of chiropractic, osteopathy and musculoskeletal physiotherapy, working in the UK and predominantly in a primary care setting. A substantial majority of chiropractors and osteopaths work exclusively in the private sector. Musculoskeletal physiotherapists work in both private and NHS sectors. Based on the pilot study response rates for each of the three professional groups, roughly proportional random samples were drawn from the respective professional register of each professional group (Table 1). Each selected practitioner was then sent an invitation to take part in the study (Figure 1). The study began in November 2003. Invitation letters, study information sheets and baseline questionnaires were sent together, with one reminder including further copies of all documents for non-responders. Only completed baseline questionnaires received prior to our predetermined baseline cut-off date (1st March 2004) were included. Following this, the intervention package was posted. Follow-up questionnaires were posted six-months following intervention. Non-responders to the follow-up questionnaire were also sent a reminder. The eligibility (and exclusion) criteria are in Table 1.
Table 1 Sample eligibility and recruitment
Professional group Eligibility criteria Total population meeting criteria Available number in population following initial exclusions* Number randomly selected (% of total population)
Chiropractors Registered with the General Chiropractic Council and the British Chiropractic Association 1071 (100%) 611 611 (57.0%)
Osteopaths Registered with the General Osteopathic Council 2718 (100%) 1868 1368 (50.3%)
Physiotherapists Registered with the Chartered Society of Physiotherapy and self-selected as 'musculoskeletal' by speciality 3586** (100%) 3230 1625 (45.3%)
Total = 7375 (100%) 5709 3604 (48.9%)
* Practitioners with addresses outside of the UK and in Scotland were excluded prior to posting invitations, and only one practitioner from each practice was eligible, being randomly selected from each address to avoid contamination (chiropractors and osteopaths only).
** At the time of sampling, there were approximately 30,000 physiotherapists in the UK, of which only 3587 were listed by the Chartered Society of Physiotherapy as 'musculoskeletal' by speciality.
Figure 1 Trial design
Chiropractors
Following the Chiropractors Act (1994), the legal definition of a chiropractor in the UK is a registrant with the General Chiropractic Council (GCC). McTimoney chiropractors are a minority group within the profession, comprising approximately one third of these registrants. McTimoney chiropractors are known to practice in substantially different ways compared to other chiropractors [85]. For this reason, they have previously been excluded from another multidisciplinary study [24,86]. Therefore, in an attempt to increase the homogeneity of the chiropractors in this study, only British Chiropractic Association (BCA) members, who are not McTimoney trained, were included (i.e. non-BCA members were excluded). With relevant permissions, the GCC register was cross referenced with the BCA register to provide the chiropractic sampling frame in this study. This information was less than twelve months old at the time of recruitment (November 2003).
Osteopaths
Following the Osteopaths Act (1993), the legal definition of an osteopath in the UK is a registrant with the General Osteopathic Council (GOsC). Whilst there is known to be considerable variation in practice within the profession [42], there are no established minority groups comparable to McTimoney chiropractors that needed excluding. Therefore, with permission, the GOsC register was used as the osteopathy sampling frame in this study. This information was less than twelve months old at the time of recruitment.
Musculoskeletal physiotherapists
Almost all of the approximately 35,000 physiotherapists in the UK are registered with the Chartered Society of Physiotherapy (CSP). Within the register of the CSP at the time of sampling, there were self-selected clinical speciality groups, one of which was 'musculoskeletal'. In an attempt to increase the likelihood that contacted physiotherapists were directly involved in managing LBP patients, this subgroup of the CSP register was used, with permission, as the physiotherapy sampling frame in this study. This information was the best available at the time of recruitment, although it was approximately four years old.
Exclusion criteria
Non-UK practitioners were excluded from the study by removing them from the professional register database sampling frames before the study sample was randomly selected. Likewise, all practitioners living or working in Scotland were excluded due to the likely confounding effects of the simultaneous, ongoing national multifactorial educational campaign, Working Backs Scotland [15,87].
Where possible, one practitioner was randomly selected from each practice prior to randomly drawing the study sample. Practice addresses were only available from the GCC and GOsC registers. The CSP register provided mostly home addresses and thus did not allow for the exclusion of practitioners working from the same practice. Inspection of chiropractic and osteopathy registers also indicated that many practitioners worked at more than one practice. Therefore, to reduce possible contamination, we excluded any practitioner (regardless of professional group) if the preferred contact address given on the baseline questionnaire was identical to that of a practitioner already recruited to the study.
Further exclusions were made from information given on returned baseline questionnaires. Practitioners who had retired, who were non-practicing, or who reported that they were not currently involved directly in the management of LBP patients were excluded (Figure 1).
Participant recruitment
We drew a random sample from the respective professional registers of each professional group. Random selection from the electronic register databases was performed by allocating a randomly generated number in a computer spreadsheet (MS Excel 2002) to each practitioner in their respective register and then sorting the list by this number. The sample was drawn from the descending list as per the order of the sorted random numbers (Figure 1). The size of each sample drawn was approximately proportional to the size of the group in the UK at the time of recruitment (Table 1).
Each practitioner selected was sent, via 'recorded delivery' (signature required) post, a letter of invitation to take part in the study, a brief study information sheet based on the guidance from the Multi-Centre Research Ethics Committee, a pre-paid reply envelope and the baseline questionnaire. A removable consent form was incorporated in the baseline questionnaire, which also gave participants opportunity to provide a preferred contact address. Only those respondents providing written consent to participate were included in the study.
We calculated response and recruitment rates from an 'adjusted sample' that excluded undelivered questionnaires and those returned to the research team (e.g. 'addressee gone away') plus those respondents who indicated that they were retired, no longer in clinical practice, or whose workload did not include LBP patients. As expected, the older physiotherapy register database provided more exclusions in this way than the other register databases. Practitioners not responding to the first posting of these documents were sent a second and final copy of each after eight weeks, via normal post.
'Non-responders' were those where no response was received to either posting of the baseline questionnaire. In total, 20% of the non-responders from each professional group were followed up with a shortened version of the questionnaire, in order to collect demographic information and to explore potential selection bias.
Consent and participant anonymity
Each practitioner selected from their professional register was allocated a unique study number prior to the posting of the baseline questionnaires. This number appeared on both consent and data collection sections of the baseline questionnaire, and allowed for the efficient tracking of both responders and non-responders throughout the study.
On receipt of completed baseline questionnaires, one member of the research team removed the consent section from the accompanying data collection section and stored them separately. Thus, the unique study number became the only source of identification on each questionnaire. This ensured that participant confidentially was maintained, and ensured that the study team were blind to intervention allocation. Only one member of the research team, not involved in data entry or intervention allocation, had access to a password-protected database on a secure computer containing the personal identifying information associated with each unique study number.
Randomisation and intervention allocation
The baseline questionnaire was completed prior to subsequent intervention allocation (Figure 1). Practitioners that responded to the baseline questionnaire and consented to take part further in the study became the study 'participants'. A member of the research team, based in another institution and who was not involved in the data collection, produced a blind, random allocation of each participant to either control or intervention arms, using only a list of participant's anonymous unique study numbers in StatsDirect software package. The intervention packages were posted with a brief covering letter to those allocated to the intervention arm in March 2004.
Interventions
Information package
The intervention consisted of an innovative information package, posted directly to participants. The package clearly details the evidence-based management of acute back pain, based on recommendations from the UK evidence-based and evidence-linked multi-disciplinary primary care [21] and occupational [22] LBP guidelines. We originally considered producing our own information package. However, when designing the study we identified an existing, high quality, package that met our needs. This package was originally developed as part of an ongoing national, multi-intervention educational campaign in Scotland led by the Health Education Board of Scotland and the Health and Safety Executive, entitled 'Working Backs Scotland' (WBS) [15]. We used exactly the same package as the WBS package with an additional covering letter, which explained the origins of the package. We did not have sufficient resources to produce a new version of the pack to the same quality, in order to remove its Scottish origins. Since the issues in the management of LBP are likely to be similar in all parts of the UK, this should not affect the appropriateness of the package for our current purpose.
The novel approach taken by the WBS initiative was to produce a comprehensive and cohesive information package, suitable for multiple stakeholders (summarised in Table 2), as part of a national initiative similar to a high-profile and largely successful Australian study [9,10,88].
Table 2 Components of the 'Working Backs Scotland' information package
Targeted stakeholder Component of package
Patients The Back Book (1st edition)
Health & Safety Executive: Back in Work booklet
Contextualised A4 sheet with recommendations
Poster with recommendations for patients
All providers Introduction letter from Working Backs Scotland
Therapy Providers Contextualised A4 sheet with recommendations
Royal College of General Practitioners: A4 guideline summary pamphlet
Yellow Flags guideline summary
General Practitioners Contextualised A4 sheet with recommendations
Royal College of General Practitioners: A4 guideline summary pamphlet
Yellow Flags guideline summary
Pharmacists Contextualised A4 sheet with recommendations
Employers Contextualised A4 sheet with recommendations
Health & Safety Executive: Back in Work booklet
In order to maximise the likely success of the package, the contents were deliberately designed to target previously identified contextual 'barriers' to behaviour change. One suggested strategy for increasing the uptake of guideline recommendations has been local adaptation, in order to provide a sense of relevance and local 'ownership' of the recommendations [54,89,90]. However, there is some evidence that local adaptation of national guidelines has no effect on practitioner behaviour [91], suggesting that relevance and ownership should be provided in some other way. One alternative for providing these has been the production of unidisciplinary guidelines. An example of this has been the recent formation of physiotherapy guidelines in the Netherlands [13], an equivalent of which is also currently under development in the UK [92]. Unidisciplinary guidelines have the further advantage of being able to provide best practice recommendations on technical areas of clinical practice that have little relevance outside a particular healthcare professional group.
In the WBS package, an attempt has been made to simultaneously provide relevance for all stakeholder groups by supplying each with an individual, tailored A4 recommendation sheet, every sheet being consistent with the others [87]. Each A4 sheet includes only those recommendations most likely to be relevant to that particular group. This allows every stakeholder to see the consistent messages provided by the recommendations, with the aim of reinforcing these messages. Importantly, the specific behaviours targeted for all stakeholder groups are kept to a minimum [84] and focus predominantly upon LBP patients staying active, avoiding bed-rest, and staying at or returning to work. Intended specifically for the healthcare practitioners is a copy of the brief summary A4 pamphlet version of the RCGP guidelines [21], and an A4 summary sheet explaining the concept of psychosocial 'Yellow Flags' [93]. The package also includes a copy of the patient-orientated advice booklet, the first edition of 'The Back Book' [94], which has been formally tested by Burton et al [95], and a leaflet designed for employers and workers, 'Back in Work' [96], the messages of which are consistent with the guidelines.
Much like the principal recommendations of the original sources on which the WBS package was based [21-23], the wording of the recommendations on each contextualised A4 sheet are clearly written in the form of behavioural instructions; an important attribute in any intervention designed to change practitioner behaviour [97,98]. Credibility of the package is reinforced by several declarations that the contents of the package are based on the best international research evidence [97]. Furthermore, the commissioning body was expected to be well respected by all stakeholders (Waddell G, personal communication). The issue of perceived ownership of the material [99] was addressed through considerable effort in enlisting and displaying the support of a wide range of local and national professional groups and societies (22 in total). These groups, including bodies representing the chiropractic, osteopathy and physiotherapy professions, endorsed both the initiative and the information package.
Control arm
Participants in the control arm were sent the standard brief description of the study with the baseline questionnaire, but were sent no further information prior to the six-months post-intervention follow-up questionnaire. Therefore, they received nothing other than they would during usual clinical practice. Of course no control could, nor should, have been exerted over other sources of information available to either arm of the trial. Being a pragmatic RCT with a large sample size, any other influences on clinical practice are likely to have been equal for each arm.
Questionnaire development
The choice of outcome measures was guided by the findings from previous qualitative studies [41,44]. Permission to use previously published material was gained from the original authors of each outcome measure. The contents of the baseline and sixmonths post-intervention follow-up questionnaires are summarised in Table 3. The two most important outcomes measured in the study were (1) reported clinical behaviour (primary outcome), and (2) practitioners' attitudes and beliefs. Each outcome measure was slightly modified from the original version for the purposes of this study. A description and justification for each modification is given below.
Table 3 Information collected in the self-completed questionnaires
Information Measure Original authors Items Baseline 6 months
Demographic details Age Professional group Years qualified NHS/Private Back pain workload Personal history of back pain - - Yes No
Reported behaviour relating to a patient vignette with non-specific acute low back pain and no 'red flags'. Questions relating to a low back pain patient vignette (modified from original studies) Modified from:
Bombardier et al 1995 [105]
Rainville et al 2000 [109]
Buchbinder et al 2001 [10]
Bishop & Foster 2005 [49]
3 Yes Yes
Practitioner beliefs concerning the way that low back pain is likely to affect function Modified Health Care Providers Pain and Impairment
Relationship scale (HC-PAIRS) Modified from:
Rainville et al 1995 [110]
Houben et al 2004 [46] 13 Yes Yes
Practitioner self- confidence in managing patients with low back pain Practitioner self-confidence scale Modified from:
Bush et al 1993 [117]
Smucker et al 1998 [118] 4 Yes Yes
Practitioner attitudes towards the use of research and evidence in practice Practitioners connections with research
Authority of information sources to influence practice Modified from:
Connolly et al 2001 [119] 12 Yes Yes
Outcome measure 1: Reported behaviour
The primary outcome measure was a measure of reported clinical practice behaviour, captured using questions related to a single vignette of a patient with non-specific acute LBP and no 'red flags' for possible serious spinal pathology (Figure 2).
Figure 2 Patient vignette with non-specific acute low back pain and no 'red flags'.
Vignettes are hypothetical clinical presentations of patients intended to elicit from practitioners an underlying practice pattern or 'medical signature' [100]. Research has been conducted on the discordance between what physicians report they would do in response to a patient vignette and what they actually do in practice. Some investigators have raised concern to the value and validity of vignettes as an outcome measure in assessing practitioner behaviour [100-103]. Specifically, critics have argued that practitioner responses to vignette are likely to report what they perceive should be done rather than what they would actually do. However, a large, prospective study [104] has shown that vignettes are not only reliable, stable and valid measures of practitioner behaviour, but that they are more accurate than patient case notes (chart abstraction), which are often used as a measure of practitioner behaviour. Thus, despite understandable concerns of social desirability or acquiescence bias, changing responses towards vignettes may be considered a single, valid point along a causal pathway towards actual behaviour change.
We chose to use a vignette in the present study as we wanted to assess the practitioners' decisions about appropriate healthcare for a specific clinical scenario, which is precisely what clinical guidelines are intended to facilitate [6]. The vignette used was originally developed by Bombardier et al [105], and has since been used in several studies [10,106-108]. Some details of the vignette were anglicised and clarified prior to use, and further details regarding occupational duties and domestic responsibilities were added following comments received during the pilot study.
In light of key recommendations consistent across guidelines from many countries [16-19], and in a manner similar to the work of Rainville et al [109] with US physicians and Bishop and Foster [49] with UK physiotherapists, the primary questions relating to the vignette were designed to explore practitioner recommendations for patient behaviour in relation to (i) Work, (ii) Activity, and (iii) Bed-rest. Other secondary questions were added to explore investigations (including x-rays) and initial treatment interventions (e.g. advice, manual therapy, electrotherapy and exercise). The format chosen to capture responses to each of these primary questions was a 5-point Likert-type scale (Figure 3), based on questions originally used by Rainville et al [109].
Figure 3 Questions and corresponding response options relating to a low back pain patient vignette The points of dichotomisation for each scale, based on expert consensus, are shown (with arrows). For each scale, the response option(s) to the right of each arrow were interpreted as consistent with guideline recommendations. The numbers above and between response options represent the frequency of expert opinion for the point of dichotomy relating to that scale. Lenient interpretations (L) are placed on the top row above response options, and strict interpretations (S) on the bottom row. The five-point scales are considered ordinal. However, two items of the scale related to Work will be combined (highlighted in bold italics) to form a four-point scale in this case.
In order to inform the analysis of the results, and as the responses to these questions were being interpreted in light of guideline recommendations, an expert consensus exercise was performed. A group of 12 experts with extensive prior experience in LBP research were contacted independently and provided with a copy of the vignette and questions. Feedback from the experts confirmed that the five-point scales could be considered ordinal. However, it was recommended that two items of the scale related to Work should be combined (highlighted in bold italics in Figure 3) to form a four-point scale in this case.
There are two research questions relating to the primary outcome of reported behaviour. Each of these requires a different analysis to test for a statistically significant change between the two arms of the trial at the six-month post-intervention time-point. The two research questions are:
1. For each scale, can the intervention change the response proportions of 'guideline-consistent' and 'guideline-inconsistent' reported behaviour?
2. For each scale, can the intervention change the response proportions of reported behaviour across the entire range of responses?
The first of these questions represents the primary outcome measure for this study. It relates to changes in the proportions of responses either side of a predetermined cut-off point for each reported behaviour research question. Therefore, each of the original five-point scales will be dichotomised to create a new variable with categories defined as 'guideline-consistent' and 'guideline-inconsistent'. The location of the cut-off points for each scale is based on the expert consensus exercise. All experts were asked to give, where possible their 'lenient' and 'strict' interpretation of internationally agreed recommendations from primary care and occupational evidence-based guidelines [16-19] for each scale. These two interpretations will allow for a sensitivity analysis.
The wording and points of dichotomisation for each scale question are shown (with arrows) in Figure 3. For each scale, the response option(s) to the right of the arrow were interpreted as recommendations consistent with those of the guidelines. The positions of the red arrows indicate the location of cut-off points for the most frequent 'strict' expert interpretation of the guideline recommendations. The blue arrows indicate the location of cut-off points for the most frequent 'lenient' expert interpretation.
Outcome measure 2: Practitioners' beliefs
Practitioners' beliefs regarding the relationship between back pain and physical function were measured using a modified version of the 13-item Health Care Providers Pain and Impairment Relationship Scale (HC-PAIRS) (Table 4), which was originally developed and validated by Rainville et al [110], and further studied by Houben et al [46]. The, responses for each item were collected with a 7-point Likert scale anchored from 'Completely disagree' (1) to 'Completely agree' (7). The scores are summed to form a total HC-PAIRS score, giving a theoretical range of possible scores from 13 to 91. Items 1, 6 and 12 were positively worded and were therefore reverse-scored for scoring and subsequent analysis. Factor analysis has shown the HC-PAIRS to measure a single factor of beliefs concerning the way in which LBP affects physical function [46]. Higher scores on the HC-PAIRS indicate a stronger belief that pain should affect daily function.
Table 4 Modified items of the HC-PAIRS. Modified components are highlighted in bold.
Item No. Item
1 Low back pain patients can still be expected to fulfil work and family responsibilities despite pain
2 An increase in pain is an indicator that a low back pain patient should stop what they are doing until the pain decreases
3 Low back pain patients cannot go about normal life activities when they are in pain
4 If their pain would go away, low back pain patients would be every bit as active as they used to be
5 Low back pain patients should have the same benefits as the handicapped* because of their painful problem
6 Low back pain patients owe it to themselves and those around them to perform their usual activities even when their pain is bad
7 Most people expect too much of low back pain patients, given their pain
8 Low back pain patients have to be careful not to do anything that might make their pain worse
9 As long as they are in pain, low back pain patients will never be able to live as well as they did before
10 Low back pain patients have to accept that they are disabled persons, due to their pain
11 There is no way that low back pain patients can return to do the things that they used to unless they first find a cure for their pain
12 Even though their pain is always there, low back pain patients often don't notice it at all when they are keeping themselves busy
13 All of low back pain patients' problems would be solved if their pain would go away
* The following caveat to item 5 was added as a footer in the modified version: We are aware that the term 'handicapped' is not liked by many people who have completed this questionnaire in the past but the research team are unfortunately unable to change this previously validated questionnaire as we may affect the results if we did so.
The original version of the HC-PAIRS was intended to assess health care providers' attitudes and beliefs about the degree to which chronic LBP justifies impairments and disability. It was published in 1995, and was itself adapted from a questionnaire developed for chronic pain patients, the Pain and Impairment Relationship Scale (PAIRS) that was first published in 1988 [111].
Modifications were made to the original HC-PAIRS in the present study for several reasons. Firstly, the evidence-based primary care [21] and occupational [22] guidelines, upon which the intervention package was based, were specifically developed for the management of low back pain rather than 'back pain' per se. Furthermore, the primary care guideline in particular is specifically based on the evidence for acute LBP and is clearly labelled as such. Moreover, the term 'chronic back pain patients' that Rainville et al [110] used to modify the PAIRS scale was based on the premise that the attitudes and beliefs measured would be valid only in chronic pain patients. Thus, the assumptions behind the design of these scales were representative of the knowledge at the time of publication. Certainly, one major development in LBP research since the late 1980's and early 1990's is that psychosocial factors have been shown to be involved much earlier in the course of LBP than was originally thought. Therefore, even though the PAIRS has not been formally tested in any other population than chronic pain patients [111-114], it is reasonable to argue that the practitioner beliefs measured by the modified version of the HC-PAIRS will be appropriate in the context of acute LBP patients.
Sample size calculation
The primary outcome measure for this trial was the proportion of 'guideline-consistent' and 'guideline-inconsistent' reported behaviour, relating to the patient vignette for the key variables of work, activity, and bed-rest (Figure 3). This analysis is also in need of more statistical power than the alternative analysis using change in responses across the whole scale. It is therefore the appropriate criterion by which to estimate the target sample size for the trial. In light of previous studies on practitioner behaviour change [59,73], we defined a change of 10% in reported behaviour between the proportions of practitioners from baseline to the six-months post-intervention follow-up as a clinically meaningful change. Thisdifference is the same as the median effect size reported in a review of guideline implementation studies reported in a subsequent review [61,63].
There are no existing data on 'guideline consistent' responses on which we could base our sample size calculation. To show a 10% difference in the proportion of these practitioners who, at the six-month post-intervention follow up, would give 'guideline consistent' responses to this questionnaire, the greatest statistical power is needed when proportions in each group are 45% and 55%. To show this difference at the 5% significance level with 90% power, valid data from at least 543 participants in each arm of the trial are needed for analysis (1086 total). Previous studies using a similar questionnaire format to measure outcomes have shown a response rate of between 31% and 65% [10,41,73,109,115]. Based on these figures, and the overall response rate of the pilot study, a recruitment rate of approximately 50% was anticipated at baseline. Allowing for 40% loss to follow-up, 900 practitioners had to be recruited into each arm of the trial at baseline (1800 total). Further allowing for the estimated recruitment rate of 50% at baseline, approximately 3600 practitioners needed to be invited to participate at the start of the study. A total of 3604 were actually invited. Such over-sampling strategies are typical of studies on healthcare practitioners based on survey methodology [e.g. [41,115]].
Data handling and analyses
The anonymised questionnaire data have been transferred into electronic format using a pre-coded datafile via Statistical Package for Social Scientists (SPSS) Data Entry (version 3.0) software, which is designed to reduce systematic errors during data entry. Data were independently entered by an experienced data entry administrator, who was blind to the participant identities. For purposes of reducing error, 20% of all data entered in the SPSS datafile was checked by one of the research team, blind to participant intervention allocation at the time.
As this is a pragmatic trial, the primary analysis will be performed on an intention to treat basis, as recommended by Bland [71]. It will include all participants randomised, even if they did not subsequently receive or read the intervention package when allocated. A proposed secondary analysis will focus on participants who appear to have read the information package, as identified by means of responses to a question on the follow-up questionnaire acting as a proxy. Data will be analysed using SPSS (version 11.5). All significance tests will be two-sided and statistical significance will be set at the 5% level.
The reported behaviour outcomes require a different analysis for each of the two research questions. The dichotomous variable, with categories defined as either 'guideline-consistent' or 'guidelineinconsistent', will be analysed using a binomial regression model (primary analysis). This will allow for between-arm differences of the proportions of responses that fall within each category at follow-up to be tested. The points of dichotomy will be based on the 'lenient interpretation' for each variable, as per the expert consensus exercise (Figure 3). The 'strict interpretation' for each variable will be used to provide sensitivity analyses. The original ordinal data will be analysed using an ordinal regression model, in which between-arm differences of proportions across all follow-up reported behaviour data may be tested. Baseline reported behaviour responses will be included in each model as a covariate. The secondary clinical and demographic data collected at baseline will be inspected and will also be included as covariates in the analyses, depending on the proportions of variance shared between these variables.
Conclusion
The COMPLeMENT trial is a prospective, pragmatic RCT of printed, evidence-based educational material that incorporates a no-intervention control. We have presented the rationale, design, and strategy for implementation of the trial, which incorporates a large number of practitioners working predominantly in a primary care setting. The primary objective of this study is to test the short-term effectiveness (six-months following intervention) of a directly-posted information package, that contains contextualised evidence-based recommendations for the management of acute back pain, on the reported behaviour and beliefs of a combined sample of chiropractors, osteopaths and musculoskeletal physiotherapists in the UK. The secondary objectives are to describe patterns of reported behaviour, attitudes and beliefs, within and between professional groups at baseline; to identify baseline factors that are associated with a change in reported behaviour of these practitioners; and to make recommendations for future research on implementation strategies in these professional groups. The results of this trial will be presented as soon as they are available.
Abbreviations
BCA = British Chiropractic Association
CSP = Chartered Society of Physiotherapy
HC-PAIRS = Health Care Providers Pain and Impairment Relationship Scale
GCC = General Chiropractic Council
GOsC = General Osteopathic Council
LBP = Low back pain
NHS = National Health Service
RCT = Randomised controlled trial
RCGP = Royal College of General Practitioners
SPSS = Statistical Package for Social Scientists
WBS = Working Backs Scotland
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DE conceived, operationalised and co-ordinated the trial and drafted the manuscript. All authors participated in the design of the trial and have read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study is supported financially by grants from the School of Health and Rehabilitation, Keele University, the Osteopathic Educational Foundation, the Anglo-European College of Chiropractic, the General Osteopathic Council, and the British Chiropractic Association.
The authors would like to thank Health Scotland for permission to use the Working Backs Scotland package. In particular, thanks are due to Professor Gordon Waddell and Miriam O'Connor for facilitating this. Thanks are also due to the General Chiropractic Council, the General Osteopathic Council and the Chartered Society of Physiotherapy for providing electronic copies of their register databases, and to the experts and many practitioners who contributed to the study. We would also like to acknowledge the hard work of the staff at the Birmingham Back Pain Clinic, and the staff in the School of Health and Rehabilitation and the Primary Care Research Sciences Centre at Keele University.
==== Refs
Clinical Standards Advisory Group (CSAG) Clinical Standards Advisory Group Report on Back Pain 1994 London: HMSO
Maniadakis N Gray A The economic burden of back pain in the UK Pain 2000 84 95 103 10601677 10.1016/S0304-3959(99)00187-6
Breen AC The implementation of consensus guidelines 2003
McGuirk B King W Govind J Lowry J Bogduk N Safety, efficacy, and cost effectiveness of evidence-based guidelines for the management of acute low back pain in primary care Spine 2001 26 2615 22 11725244 10.1097/00007632-200112010-00017
Rossignol M Abenhaim L Séguin P Neveu A Collet J Ducruet T Shapiro S Coordination of primary health care for back pain: A randomized controlled trial Spine 2000 25 251 8 10685491 10.1097/00007632-200001150-00018
Field MJ Lohr KN Eds Clinical Practice Guidelines: Directions for a New Program 1990 Institute of Medicine, Washington, DC: National Academy Press
Cherkin D Deyo RA Berg AO Bergman JJ Lishner DM Evaluation of a physician education intervention to improve primary care for low-back pain. I. Impact on physicians Spine 1991 16 1168 72 1836676
Deyo RA Schall M Berwick DM Nolan T Carver P Continuous quality improvement for patients with back pain J Gen Intern Med 2000 15 647 55 11029679 10.1046/j.1525-1497.2000.90717.x
Buchbinder R Jolley D Wyatt M Population based intervention to change back pain beliefs and disability: three part evaluation BMJ 2001 322 1516 20 11420272 10.1136/bmj.322.7301.1516
Buchbinder R Jolley D Wyatt M 2001 Volvo award winner in clinical studies: effects of a media campaign on back pain beliefs and its potential influence on management of low back pain in general practice Spine 2001 26 2535 42 11725233 10.1097/00007632-200112010-00005
Underwood M O'Meara S Harvey E The UK BEAM Trial Team The acceptability to primary care staff of a multidisciplinary training package on acute back pain guidelines Fam Pract 2002 19 511 5 12356704 10.1093/fampra/19.5.511
Bekkering GE Engers AJ Wensing M Hendriks HJ van Tulder MW Oostendorp RA Bouter LM Development of an implementation strategy for physiotherapy guidelines on low back pain Aust J Physiother 2003 49 208 14 12952520
Bekkering GE Hendriks HJM Koes BW Oostendorp RAB Ostelo RWJG Thomassen JMC van Tulder MW Clinical guidelines on physiotherapy for low back pain Physiotherapy 2003 89 82 96
Breen A Carr E Mann E Crossen-White H Acute back pain management in primary care: a qualitative pilot study of the feasibility of a nurse-led service in general practice J Nurs Manag 2004 12 201 9 15089958
Waddell G The Back Pain Revolution 2004 2 Edinburgh: Churchill Livingstone
Burton AK Waddell G Clinical guidelines in the management of low back pain Baillieres Clin Rheumatol 1998 12 17 35 9668955
Koes BW van Tulder MW Ostelo R Burton AK Waddell G Clinical guidelines for the management of low back pain in primary care: an international comparison Spine 2001 26 2504 13 11707719 10.1097/00007632-200111150-00022
Staal JB Hlobil H van Tulder MW Waddell G Burton AK Koes BW van Mechelen W Occupational health guidelines for the management of low back pain: an international comparison Occup Environ Med 2003 60 618 26 12937181 10.1136/oem.60.9.618
van Tulder MW Tuut M Pennick V Bombardier C Assendelft WJ Quality of primary care guidelines for acute low back pain Spine 2004 29 E357 62 15534397 10.1097/01.brs.0000137056.64166.51
Waddell G McIntosh A Hutchinson A Feder G Lewis M Clinical Guidelines for the Management of Acute Low Back Pain 1996 London: Royal College of General Practitioners
Waddell G McIntosh A Hutchinson A Feder G Lewis M Clinical Guidelines for the Management of Acute Low Back Pain 1999 London: Royal College of General Practitioners
Carter JT Birrell LN Eds Occupational health guidelines for the management of low back pain at work – principal recommendations 2000 London: Faculty of Occupational Medicine of the Royal College of Physicians
Waddell G Burton AK Occupational health guidelines for the management of low back pain at work – evidence review 2000 London: Faculty of Occupational Medicine of the Royal College of Physicians
Harvey E Burton AK Klaber-Moffett J Breen AC Spinal manipulation for low-back pain: a treatment package agreed by the UK chiropractic, osteopathy and physiotherapy professional associations Man Ther 2003 8 46 51 12635637 10.1054/math.2002.0472
Government Statistical Service The prevalence of back pain in Great Britain in 1998 Bulletin based on the Office for National Statistics (ONS) Omnibus Survey module conducted in March, April and June 1998 2000 London: ONS
Mason V The prevalence of low back pain in Great Britain A report on Office of Census & Population Surveys (OPCS) omnibus survey data produced on behalf of the Department of Health 1994 London: OPCS
Funk SG Champagne MT Wiese RA Tornquist EM BARRIERS: the barriers to research utilization scale Appl Nurs Res 1991 4 39 45 1741634
Funk SG Champagne MT Wiese RA Tornquist EM Barriers to using research findings in practice: the clinician's perspective Appl Nurs Res 1991 4 90 5 1741642
Barta KM Information-seeking, research utilization, and barriers to research utilization of pediatric nurse educators J Prof Nurs 1995 11 49 57 7844284 10.1016/S8755-7223(95)80073-5
Dunn V Crichton N Roe B Seers K Williams K Using research for practice: a UK experience of the BARRIERS Scale J Adv Nurs 1997 26 1203 10 9429972 10.1046/j.1365-2648.1997.00462.x
Metcalfe C Lewin R Wisher S Perry S Bannigan K Klaber-Moffett J Barriers to implementing the evidence base in four NHS therapies Physiotherapy 2001 87 433 41
Schers H Wensing M Huijsmans Z van Tulder M Grol R Implementation barriers for general practice guidelines on low back pain a qualitative study Spine 2001 26 E348 53 11474367 10.1097/00007632-200108010-00013
Gabbay J le May A Evidence based guidelines or collectively constructed "mindlines?" Ethnographic study of knowledge management in primary care BMJ 2004 329 1013 8 15514347 10.1136/bmj.329.7473.1013
Hibble A Kanka D Pencheon D Pooles F Guidelines in general practice: the new Tower of Babel? BMJ 1998 317 862 3 9748185
Breen AC Chiropractors and the treatment of back pain Rheumatol Rehabil 1977 16 46 53 139670
Burton AK Back pain in osteopathic practice Rheumatol Rehabil 1981 20 239 46 6458082
Pringle M Tyreman S Study of 500 patients attending an osteopathic practice Br J Gen Pract 1993 43 15 18 8457357
Pedersen P A survey of chiropractic practice in Europe Euro J Chiro 1994 42 1 28
Foster NE Thompson KA Baxter GD Allen JM Management of nonspecific low back pain by physiotherapists in Britain and Ireland. A descriptive questionnaire of current clinical practice Spine 1999 24 1332 42 10404576 10.1097/00007632-199907010-00011
Gracey JH McDonough SM Baxter GD Physiotherapy management of low back pain: a survey of current practice in Northern Ireland Spine 2002 27 406 11 11840108 10.1097/00007632-200202150-00017
Foster NE Doughty G Using clinical guidelines and research evidence in the management of back pain patients: a qualitative study of musculoskeletal physiotherapists in Britain Presented at the International Forum V for Primary Care Research on Low Back Pain, Montréal, Canada 2002
Rebain R Baxter GD McDonough S The passive straight leg raising test in the diagnosis and treatment of lumbar disc herniation: a survey of United kingdom osteopathic opinion and clinical practice Spine 2003 28 1717 24 12897499 10.1097/00007632-200308010-00017
Linton SJ Vlaeyen J Ostelo R The back pain beliefs of health care providers: are we fear-avoidant? J Occup Rehabil 2002 12 223 32 12389475 10.1023/A:1020218422974
Evans DW Foster NE Vogel S Breen AC Implementing evidence-based practice in the UK physical therapy professions: do they want it and do they feel they need it? Presented at the International Forum VI for Primary Care Research on Low Back Pain, Linkoping, Sweden 2003
Daykin AR Richardson B Physiotherapists' pain beliefs and their influence on the management of patients with chronic low back pain Spine 2004 29 783 95 15087802 10.1097/01.BRS.0000115135.19082.97
Houben RMA Vlaeyen JWS Peters PMJC Ostelo RWJG Wolters PMJC Stomp-van den Berg SGM Health care providers' attitudes and beliefs towards common low back pain: Factor structure and psychometric properties of the HC-PAIRS Clin J Pain 2004 20 37 44 14668655
Houben RM Gijsen A Peterson J de Jong PJ Vlaeyen JW Do health care providers' attitudes towards back pain predict their treatment recommendations? Differential predictive validity of implicit and explicit attitude measures Pain 2005 114 491 8 15777874 10.1016/j.pain.2005.01.017
Houben RM Ostelo RW Vlaeyen JW Wolters PM Peters M Stomp-van den Berg SG Health care providers' orientations towards common low back pain predict perceived harmfulness of physical activities and recommendations regarding return to normal activity Eur J Pain 2005 9 173 83 15737810 10.1016/j.ejpain.2004.05.002
Bishop A Foster N Do physical therapists in the United Kingdom recognize psychosocial factors in patients with acute low back pain? Spine 2005 30 1316 1322 15928559
Grol R Thomas S Roberts R Development and implementation of guidelines for family practice: lessons from The Netherlands J Fam Pract 1995 40 435 9 7730766
Grol R Zwaard A Mokkink H Dalhuijsen J Casparie A Dissemination of guidelines: which sources do physicians use in order to be informed? Int J Qual Health Care 1998 10 135 40 9690886 10.1093/intqhc/10.2.135
Grol R Grimshaw J Evidence-based implementation of evidence-based medicine Jt Comm J Qual Improv 1999 25 503 13 10522231
Grol R Jones R Twenty years of implementation research Fam Pract 2000 17 S32 5 10735266 10.1093/fampra/17.suppl_1.S32
van Tulder MW Croft PR van Splunteren P Miedema HS Underwood MR Hendriks HJM Wyatt ME Borkan JM Disseminating and implementing the results of back pain research in primary care Spine 2002 27 E121 7 11880848 10.1097/00007632-200203010-00018
NHS Centre for Reviews and Dissemination Getting evidence into practice Effective Health Care 1999 5 1 16
Freemantle N Implementation strategies Fam Pract 2000 17 S7 10 10735261 10.1093/fampra/17.suppl_1.S7
Davis DA Thomson MA Oxman AD Haynes RB Changing physician performance. A systematic review of the effect of continuing medical education strategies J A M A 1995 274 700 5 7650822 10.1001/jama.274.9.700
Thorsen T Mäkelä M Eds Changing Professional Practice: Theory and Practice of Clinical Guidelines Implementation 1999 Danish Institute for Health Services Research and Development
Freemantle N Harvey EL Wolf F Grimshaw JM Grilli R Bero LA Printed educational materials: effects on professional practice and health care outcomes Cochrane Database Syst Rev 2000 2 CD000172 10796502
Grimshaw JM Shirran L Thomas R Mowatt G Fraser C Bero L Grilli R Harvey E Oxman A O'Brien MA Changing provider behavior: an overview of systematic reviews of interventions Med Care 2001 39 II2 45 11583120
Grimshaw JM Thomas RE MacLennan G Fraser C Ramsay CR Vale L Whitty P Eccles MP Matowe L Shirran L Wensing M Dijkstra R Donaldson C Effectiveness and efficiency of guideline dissemination and implementation strategies Health Technol Assess 2004 8 1 72 14960256
Oxman AD Thomson MA Davis DA Haynes RB No magic bullets: a systematic review of 102 trials of interventions to improve professional practice CMAJ 1995 153 1423 31 7585368
Grimshaw J Eccles M Tetroe J Implementing clinical guidelines: current evidence and future implications J Contin Educ Health Prof 2004 24 S31 7 15712775 10.1002/chp.1340240506
Cohen SJ Weinberger M Hui SL Tierney WM McDonald CJ The impact of reading on physicians' nonadherence to recommended standards of medical care Soc Sci Med 1985 21 909 14 4071124 10.1016/0277-9536(85)90147-9
Emslie C Grimshaw J Templeton A Do clinical guidelines improve general practice management and referral of infertile couples? B M J 1993 306 1728 31
Bearcroft PW Small JH Flower CD Chest radiography guidelines for general practitioners: a practical approach Clin Radiol 1994 49 56 8 8299334
Grimshaw J Evaluation of four quality assurance initiatives to improve out-patient referrals from general practice to hospital Dissertation University of Aberdeen 1998
Mesters I Meertens R Kok G Parcel GS Effectiveness of a multidisciplinary education protocol in children with asthma (0–4 years) in primary health care J Asthma 1994 31 347 59 7928930
Oakeshott P Kerry SM Williams JE Randomised controlled trial of the effect of the Royal College of Radiologists' guidelines on general practitioners' referral for radiographic examination Br J Gen Pract 1994 44 197 200 8204331
Rabin DL Boekeloo BO Marx ES Bowman MA Russell NK Willis AG Improving office-based physician's prevention practices for sexually transmitted diseases Ann Intern Med 1994 121 513 19 8067649
Bland JM Sample size in guidelines trials Fam Pract 2000 17 S17 20 10735263 10.1093/fampra/17.suppl_1.S17
Wood J Freemantle N Choosing an appropriate unit of analysis in trials of interventions that attempt to influence practice J Health Serv Res Policy 1999 4 44 8 10345566
O'Brien K Wright J Conboy F Bagley L Lewis D Read M Thompson R Bogues W Lentin S Parr G Aron B The effect of orthodontic referral guidelines: a randomised controlled trial Br Dent J 2000 188 392 97 10816930 10.1038/sj.bdj.4800492a
Bland JM Cluster randomised trials in the medical literature: two bibliometric surveys BMC Med Res Methodol 2004 4 21 15310402 10.1186/1471-2288-4-21
Eldridge SM Ashby D Feder GS Rudnicka AR Ukoumunne OC Lessons for cluster randomized trials in the twenty-first century: a systematic review of trials in primary care Clinical Trials 2004 1 80 90 16281464 10.1191/1740774504cn006rr
Hahn S Puffer S Torgerson DJ Watson J Methodological bias in cluster randomised trials BMC Med Res Methodol 2005 5 10 15743523 10.1186/1471-2288-5-10
Farrin A Russel I Torgerson D Underwood M Differential recruitment in a cluster randomised trial in primary care – the experience of the UK Back pain, Exercise, Active management and Manipulation (UK BEAM) feasibility study Clinical Trials 2005 2 119 124 16279133
Bekkering GE van Tulder MW Hendriks EJM Koopmanschap MA Knol DL Bouter LM Oostendorp RAB Implementation of clinical guidelines on physical therapy for patients with low back pain: randomized trial comparing patient outcomes after a standard and active implementation strategy Phys Ther 2005 85 544 55 15921475
Bekkering GE Hendriks HJ van Tulder MW Knol DL Hoeijenbos M Oostendorp RA Bouter LM Effect on the process of care of an active strategy to implement clinical guidelines on physiotherapy for low back pain: a cluster randomised controlled trial Qual Saf Health Care 2005 14 107 12 15805455 10.1136/qshc.2003.009357
Engers AJ Wensing M van Tulder MW Timmermans A Oostendorp RA Koes BW Grol R Implementation of the Dutch low back pain guideline for general practitioners: a cluster randomized controlled trial Spine 2005 30 559 600 15770171 10.1097/01.brs.0000155406.79479.3a
Schectman JM Schroth WS Verme D Voss JD Randomized controlled trial of education and feedback for implementation of guidelines for acute low back pain J Gen Intern Med 2003 18 773 80 14521638 10.1046/j.1525-1497.2003.10205.x
Dey P Simpson CW Collins SI Hodgson G Dowrick CF Simison AJ Rose MJ Implementation of RCGP guidelines for acute low back pain: a cluster randomised controlled trial Br J Gen Pract 2004 54 33 7 14965404
Sculpher M Evaluating the cost-effectiveness of interventions designed to increase the utilization of evidence-based guidelines Fam Pract 2000 17 S26 31 10735265 10.1093/fampra/17.suppl_1.S26
Suarez-Almazor ME Russell AS The art versus the science of medicine. Are clinical practice guidelines the answer? Ann Rheum Dis 1998 57 67 9 9613333
Harding S McTimoney Chiropractic: the first 25 years 1997 Abingdon, Oxford: McTimoney Chiropractic Association
UK BEAM Trial Team UK Back pain Exercise And Manipulation (UK BEAM) trial – national randomised trial of physical treatments for back pain in primary care: objectives, design and interventions BMC Health Serv Res 2003 3 16 12892566 10.1186/1472-6963-3-16
Working Backs Scotland Online
Buchbinder R Jolley D Population based intervention to change back pain beliefs: three year follow up population survey B M J 2004 328 321
Feder G Grange W Hamlyn P Hull S Hussaini A Kidd B Kidner S Montgomery J Clinical guidelines for the management of acute low back pain in East London primary care 1997 London : Joint Academic Department of General Practice and Primary Care, St Bartholomew's and the Royal London School of Medicine and Dentistry
Stevenson K Hay EM A survey of GP's and physiotherapists (PT) views on a physiotherapy led back pain service Presented at the Society for Back Pain Research, Bristol, UK 2001
Silagy CA Weller DP Lapsley H Middleton P Shelby-James T Fazekas B The effectiveness of local adaptation of nationally produced clinical practice guidelines Fam Pract 2002 19 223 30 11978710 10.1093/fampra/19.3.223
Deane K Development of guidelines on lower back pain Presented at the Chartered Society of Physiotherapists Annual Congress, Birmingham, UK 2004
Kendall N Linton SJ Main CJ Guide to Assessing Psychological Yellow Flags in Acute Low Back Pain: Risk factors for long-term disability and work loss 1997 Wellington, New Zealand: Accident Rehabilitation & Compensation Insurance Corporation of New Zealand and the National Health Committee
Roland M Burton K Waddell G Klaber-Moffett J Main C The Back Book 1996 1 London: The Stationery Office
Burton AK Waddell G Tillotson KM Summerton N Information and advice to patients with back pain can have a positive effect. A randomized controlled trial of a novel educational booklet in primary care Spine 1999 24 2484 91 10626311 10.1097/00007632-199912010-00010
Health and Safety Executive Back in work: managing back pain in the workplace A leaflet for employers and workers in small businesses 2000 London: Health and Safety Executive
Grol R Dalhuijsen J Thomas S Veld C Rutten G Mokkink H Attributes of clinical guidelines that influence use of guidelines in general practice: observational study B M J 1998 317 858 61
Michie S Johnston M Changing clinical behaviour by making guidelines specific B M J 2004 328 343 5
Hayward RS Guyatt GH Moore KA McKibbon KA Carter AO Canadian physicians' attitudes about and preferences regarding clinical practice guidelines Can Med Assoc J 1997 156 1715 23 9220923
Wennberg DE Dickens JD JrBiener L Fowler FJ JrSoule DN Keller RB Do physicians do what they say? The inclination to test and its association with coronary angiography rates J Gen Intern Med 1997 12 172 6 9100142 10.1046/j.1525-1497.1997.012003172.x
Hartley RM Charlton JR Jarman B Harris CM Case history questionnaires in the study of doctors' use of resources: Are they measuring what we want? Med Care 1985 23 1163 8 3903368
Jones TV Gerrity MS Earp J Written case simulations: Do they predict physicians' behavior? J Clin Epidemiol 1990 43 805 15 2200852 10.1016/0895-4356(90)90241-G
Morrell DC Roland MO Analysis of referral behaviour: responses to simulated case histories may not reflect real clinical behaviour Br J Gen Pract 1990 40 182 5 2114132
Peabody JW Luck J Glassman P Dresselhaus TR Lee M Comparison of vignettes, standardized patients, and chart abstraction. A prospective study of 3 methods for measuring quality J Am Med Assoc 1997 283 1715 22 10.1001/jama.283.13.1715
Bombardier C Jansz G Maetzel A Primary care physicians' knowledge, confidence, and attitude in the management of acute low back pain (ALBP) Arthritis Rheum 1995 38 S385
Ammendolia C Bombardier C Hogg-Johnson S Glazier R Views on radiography use for patients with acute low back pain in an Ontario community J Manip Physiol Ther 2002 25 11 20
Li LC Bombardier C Physical therapy management of low back pain: an exploratory survey of therapist approaches Phys Ther 1999 81 1018 28 11276184
Maetzel A Johnson SH Woodbury M Bombardier C Use of grade membership analysis to profile the practice styles of individual physicians in the management of acute low back pain J Clin Epidemiol 2000 53 195 205 10729692 10.1016/S0895-4356(99)00120-1
Rainville J Carlson N Polatin P Gatchel RJ Indahl A Exploration of physicians' recommendations for activities in chronic low back pain Spine 2000 25 2210 20 10973405 10.1097/00007632-200009010-00012
Rainville J Bagnall D Phalen L Health care providers' attitudes and beliefs about functional impairments and chronic back pain Clin J Pain 1995 11 287 95 8788576
Riley JF Ahern DK Follick MJ Chronic back pain and functional impairment: assessing beliefs about their relationship Arch Phys Med Rehab 1988 69 579 82
Barrios FX Riley JF Cognitive factors and disability in chronic pain: is there more than just depression? Presented at the 21st Annual Meeting for the Advancement of Behavioural Therapy, Boston 1987
Put CL Witkower A Pain and impairment beliefs in patients treated in an interdisciplinary inpatient pain program Presented at the American Pain Society 10th Annual Meeting, New Orleans, LA 1991
Slater MA Hall HF Atkinson JH Garfin SR Pain and impairment beliefs in chronic low back pain; validation of the Pain and Impairment Relationship Scale (PAIRS) Pain 1991 44 51 6 1828111 10.1016/0304-3959(91)90146-O
Russell ML Verhoef MJ Injeyan HS McMorland DG Response rates for surveys of chiropractors J Manip Physiol Ther 2004 27 43 8 10.1016/j.jmpt.2003.11.005
Konstantinou K Foster N Rushton A Baxter D The use and reported effects of mobilization with movement techniques in low back pain management; a cross-sectional descriptive survey of physiotherapists in Britain Man Ther 2002 7 206 14 12419655 10.1054/math.2002.0469
Bush T Cherkin D Barlow W The impact of physician attitudes on patient satisfaction with care for low back pain Arch Fam Med 1993 2 301 5 8252151 10.1001/archfami.2.3.301
Smucker DR Konrad TR Curtis P Carey TS Practitioner self-confidence and patient outcomes in acute low back pain Arch Fam Med 1998 7 223 8 9596455 10.1001/archfami.7.3.223
Connolly BH Lupinnaci NS Bush AJ Changes in attitudes and perceptions about research in physical therapy among professional physical therapist students and new graduates Phys Ther 2001 81 1127 34 11319938
|
16033646
|
PMC1208895
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Jul 20; 6:41
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-41
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-431608079410.1186/1471-2474-6-43Research ArticleOsteopathic manipulative treatment for low back pain: a systematic review and meta-analysis of randomized controlled trials Licciardone John C [email protected] Angela K [email protected] Linda N [email protected] Osteopathic Research Center, University of North Texas Health Science Center, Fort Worth, TX 76107, USA2 Department of Family Medicine, University of North Texas Health Science Center, Fort Worth, TX 76107, USA3 Gibson D. Lewis Health Science Library, University of North Texas Health Science Center, Fort Worth, TX 76107, USA2005 4 8 2005 6 43 43 8 11 2004 4 8 2005 Copyright © 2005 Licciardone et al; licensee BioMed Central Ltd.2005Licciardone et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Osteopathic manipulative treatment (OMT) is a distinctive modality commonly used by osteopathic physicians to complement their conventional treatment of musculoskeletal disorders. Previous reviews and meta-analyses of spinal manipulation for low back pain have not specifically addressed OMT and generally have focused on spinal manipulation as an alternative to conventional treatment. The purpose of this study was to assess the efficacy of OMT as a complementary treatment for low back pain.
Methods
Computerized bibliographic searches of MEDLINE, EMBASE, MANTIS, OSTMED, and the Cochrane Central Register of Controlled Trials were supplemented with additional database and manual searches of the literature.
Six trials, involving eight OMT vs control treatment comparisons, were included because they were randomized controlled trials of OMT that involved blinded assessment of low back pain in ambulatory settings. Data on trial methodology, OMT and control treatments, and low back pain outcomes were abstracted by two independent reviewers. Effect sizes were computed using Cohen's d statistic and meta-analysis results were weighted by the inverse variance of individual comparisons. In addition to the overall meta-analysis, stratified meta-analyses were performed according to control treatment, country where the trial was conducted, and duration of follow-up. Sensitivity analyses were performed for both the overall and stratified meta-analyses.
Results
Overall, OMT significantly reduced low back pain (effect size, -0.30; 95% confidence interval, -0.47 – -0.13; P = .001). Stratified analyses demonstrated significant pain reductions in trials of OMT vs active treatment or placebo control and OMT vs no treatment control. There were significant pain reductions with OMT regardless of whether trials were performed in the United Kingdom or the United States. Significant pain reductions were also observed during short-, intermediate-, and long-term follow-up.
Conclusion
OMT significantly reduces low back pain. The level of pain reduction is greater than expected from placebo effects alone and persists for at least three months. Additional research is warranted to elucidate mechanistically how OMT exerts its effects, to determine if OMT benefits are long lasting, and to assess the cost-effectiveness of OMT as a complementary treatment for low back pain.
==== Body
Background
Historically, low back pain has been the most common reason for visits to osteopathic physicians [1]. More recent data from the Osteopathic Survey of Health Care in America has confirmed that a majority of patients visiting osteopathic physicians continue to seek treatment for musculoskeletal conditions [2,3]. A distinctive element of low back care provided by osteopathic physicians is osteopathic manipulative treatment (OMT). A comprehensive evaluation of spinal manipulation for low back pain undertaken by the Agency for Health Care Policy and Research in the United States concluded that spinal manipulation can be helpful for patients with acute low back problems without radiculopathy when used within the first month of symptoms [4]. Nevertheless, because most studies of spinal manipulation involve chiropractic or physical therapy [5], it is unclear if such studies adequately reflect the efficacy of OMT for low back pain.
Although the professional associations that represent osteopaths, chiropractors, and physiotherapists in the United Kingdom developed a spinal manipulation package consisting of three common manual elements for use in the UK Back pain Exercise and Manipulation (UK BEAM) trial [6], there are no between-profession comparisons of clinical outcomes [7,8]. It is well known that OMT comprises a diversity of techniques [9] that are not adequately represented by the UK BEAM trial package. Professional differences in spinal manipulation are more pronounced in research studies, where chiropractors have focused almost exclusively on high-velocity-low-amplitude techniques [10]. For example, a major trial of chiropractic manipulation as adjunctive treatment for childhood asthma used a high-velocity-low-amplitude thrust as the active treatment [11]. The simulated treatment provided in the sham manipulation arm of this chiropractic trial, which ostensibly was thought to have no therapeutic effect, had a marked similarity to OMT [10,12]. Further, because differences in professional background and training lend themselves to diverse manipulation approaches, clinicians have been warned about generalizing the findings of systematic reviews to practice [13].
In addition to professional differences in the manual techniques themselves, osteopathic physicians in the United States, unlike allopathic physicians, chiropractors, or physical therapists, can treat low back pain simultaneously using both conventional primary care approaches and complementary spinal manipulation. This represents a unique philosophical approach in the treatment of low back pain. Consequently, there is a need for empirical data that specifically address the efficacy of OMT for such conditions as low back pain [14]. The present study was undertaken to address this need by conducting a systematic review of the literature on OMT and performing a meta-analysis of all randomized controlled trials for low back pain performed in ambulatory settings.
Methods
Search
A search of the English language literature was performed through August 2003 to identify reports of randomized controlled trials of OMT. We searched MEDLINE, OLDMEDLINE, EMBASE, MANTIS, OSTMED, Alt Health Watch, SciSearch, ClinicalTrials.gov, CRISP, and the Cochrane Central Register of Controlled Trials. A detailed description of the search strategy is provided in the Appendix [see Additional file 1]. Additionally, reports were sought from relevant reviews or meta-analyses of spinal manipulation [9,15-32] and manual searches of reference citations in the reviewed literature sources.
Selection
The search bibliographies and relevant reports were reviewed by the authors to identify randomized controlled trials involving OMT in human subjects. To assess the efficacy of OMT in primary care, eligibility was limited to randomized controlled trials of OMT performed by osteopaths, osteopathic physicians, or osteopathic trainees that included blinded assessment of low back pain in ambulatory settings. Trials that involved manipulation under anesthesia, industrial settings, or hospitalized patients were not included. Because there is potential confusion regarding the type of manipulation performed in some trials [33], the reported methods in each trial were carefully reviewed to assess eligibility for the meta-analysis. Overall, seven studies known or purported to involve OMT for low back pain were reviewed and excluded for not meeting all eligibility criteria [34-40]. A subsequent source [41] indicated that an osteopathic manipulation technique was used in the Irvine study [42]. Although several of the six included OMT trials were identified in multiple bibliographic databases, five [42-46] were indexed in MEDLINE. The Cleary [47] trial was identified exclusively through the Cochrane Central Register of Controlled Trials.
Data extraction
Each eligible trial was independently reviewed by two of us to abstract data on methodological characteristics, OMT and control treatments, and low back pain outcomes. Conflicting data were resolved by consensus.
Trial characteristics
As shown in Table 1, the six OMT trials were conducted between 1973 and 2001 in the United Kingdom or the United States [42-47]. Two of the six trials each included two control treatments [43,46], thus providing eight OMT vs control treatment comparisons. The trials generally were comparable in their methodology, with the possible exception of the Cleary [47] trial. Twenty contrasts were reported in the six trials (a contrast refers to a within-trial comparison between OMT and a control treatment with respect to a low back pain outcome at a given point in time). Following randomization, nine contrasts were reported within one month (short-term outcomes), another seven contrasts were reported within three months (intermediate-term outcomes), and the remaining four contrasts were reported within 12 months (long-term outcomes).
Table 1 Summary of trials.
Hoehler 1981 [42] Gibson 1985 [43] Cleary 1994 [47]
Years conducted 1973–1979 ... 1991–1992
Country United States United Kingdom United Kingdom
Setting University clinic Hospital outpatient clinic Ambulatory clinic
No. of subjects randomized 95 109 30*
Comparison OMT vs soft tissue massage and sham manipulation OMT vs short-wave diathermy
OMT vs detuned short-wave diathermy OMT vs sham manipulation
Subject
characteristics
Age, y
Mean ± SD OMT, 30.1 ± 8.4
Controls, 32.1 ± 9.8 OMT, 34 ± 14
Short-wave diarthermy controls, 35 ± 16
Detuned short-wave diathermy controls, 40 ± 16 Overall age range, 50–60
Sex
% male OMT, 59
Controls, 59 OMT, 49
Detuned short-wave diathermy controls, 68
Short-wave diarthermy controls, 53 OMT, 0
Controls, 0
Type of low back pain Referred patients with acute or chronic low back pain Referred patients with low back pain of greater than 2 months' and less than 12 months' duration Recruited subjects with chronic low back pain in conjunction with menopausal symptoms
OMT protocol
Technique High-velocity, low-amplitude thrust only Variety of techniques Low-force techniques
No. of treatments
Mean ± SD OMT, 4.8 ± 2.7
Controls, 3.9 ± 2.5 4, per protocol 10, per protocol
Outcomes assessment Blinded Blinded Assessment independent of treatment, blinding not specified
No. of pain contrasts 3 6 (3 for each of the two OMT vs control treatment comparisons) 1
Type of pain outcome Dichotomous pain outcomes Dichotomous pain outcomes Dichotomous pain outcome
Timing of pain contrasts
Short-term First treatment and mean, 20–30 days following randomization 2 and 4 weeks ...
Intermediate-term Mean, 41–51 days following randomization ... ...
Long-term ... 12 weeks 15 weeks
Andersson 1999 [44] Burton 2000 [45] Licciardone 2003 [46]
Years conducted 1992–1994 ... 2000–2001
Country United States United Kingdom United States
Setting Health maintenance organization Hospital orthopedic department University clinic
No. of subjects randomized 178 40 91
Comparison Usual care and OMT vs usual care only OMT vs chemonucleolysis Usual care and OMT vs usual care and sham manipulation
Usual care and OMT vs usual care only
Subject characteristics
Age, y
Mean ± SD OMT, 28.5 ± 10.6
Controls, 37.0 ± 11.0 Overall, 41.9 ± 10.6 Usual care and OMT, 49 ± 12
Usual care and sham manipulation controls, 52 ± 12
Usual care only controls, 49 ± 12
Sex
% male OMT, 41
Controls, 44 Overall, 48 Usual care and OMT, 31
Usual care and sham manipulation controls, 43
Usual care only controls, 35
Type of low back pain Patients with low back pain of 3 or more weeks' and less than 6 months' duration Recruited patients with low back pain and sciatica; mean duration, 30 and 32 weeks in OMT and chemonucleolysis groups, respectively Recruited subjects with low back pain of at least 3 months' duration
OMT protocol
Technique Variety of techniques, individualized to patient Variety of techniques, individualized to patient Variety of techniques, individualized to subject
No. of treatments
Mean ± SD 8, per protocol Mean for OMT, 11; range 6–18 7, per protocol
Outcomes assessment Blinded Blinded Blinded
No. of pain contrasts 1 3 6 (3 for each of the two OMT vs control treatment comparisons)
Type of pain outcome Pain scale Pain scales Pain scales
Timing of pain contrasts
Short-term ... 2 weeks 1 month
Intermediate-term 12 weeks 6 weeks 3 months
Long-term ... 12 months 6 months
OMT denotes osteopathic manipulative treatment.
*A total of 30 subjects with menopausal symptoms were randomized; however, only 12 subjects had low back pain.
The methodological quality of four of the OMT trials [42-45] was independently confirmed in a recent systematic review that included a best evidence synthesis incorporating eight explicit quality criteria, including similarity of baseline characteristics of subjects or reporting of adjusted outcomes; concealment of treatment allocation; blinding of subjects; blinding of providers or other control for attention bias; blinded or unbiased outcomes assessment; subject dropouts reported and accounted for in the analysis; missing data reported and accounted for in the analysis; and intention-to-treat analysis or absence of differential co-interventions between groups in studies with full compliance [13]. The Cleary [47] trial was not eligible for this review because it did not include a sufficiently large number of subjects. Although the Licciardone [46] trial was not eligible for the review because it was published after the closing date of December 2002, it has been characterized as an innovative and important trial with many rigorous design features [48], and more recently has been identified as an evidence-based supplement relative to the previous review from the Cochrane Library [49].
Quantitative data synthesis
We used the effect size, computed as Cohen's d statistic, to report all trial results [50]. A negative effect size represented a greater decrease in pain among OMT subjects relative to control treatment subjects. Dichotomous pain measures were transformed to effect sizes by first computing the relevant P-value and then determining the effect size and 95% confidence interval (CI) that would obtain under the assumption of a two-tailed t-test for measuring the standardized mean difference between OMT and control treatments in the relevant number of subjects [50]. The meta-analysis results were weighted by the inverse variance for each OMT vs control treatment comparison. The Q statistic was used to test the homogeneity of trials included in each analysis [51].
The overall meta-analysis included the eight OMT vs control treatment comparisons. Four of the six trials, involving six of the eight OMT vs control treatment comparisons, each reported three contrasts [42,43,45,46] (Table 1). The median contrast, as identified by the intermediate effect size among the three reported pain outcomes for a given OMT vs control treatment comparison, was used to represent the pain outcome for each of these six comparisons. These median contrasts were then combined with the lone contrasts reported in each of the two remaining OMT vs control treatment comparisons [44,47]. Based on the similarity among trials (Table 1), a fixed-effects model initially was used to perform meta-analysis and the results were then compared with those of a random-effects model.
A series of sensitivity analyses were then performed. First, to address the possibility of bias by using the median contrasts method, analyses were repeated using the best-case and worst-case scenarios for the six relevant OMT vs control treatment comparisons [42,43,45,46]. Second, to address the possibility of bias by including comparisons involving the same OMT group vs two different control treatment groups in two trials [43,46], analyses were repeated using only one OMT vs control treatment comparison for each of these trials. Each of the four possible combinations of contrasts was analyzed. Third, the analysis was repeated after excluding the Cleary [47] trial. Finally, an analysis was performed using all 20 low back pain contrasts. Similar analyses were performed after stratifying trials according to control treatment, country where the trial was performed, and duration of follow-up.
As summarized in Table 2, there were 43 analyses performed, including the overall meta-analysis, seven stratified meta-analyses, and 35 sensitivity analyses. Meta-analysis was performed only when there were at least three contrasts available for data synthesis. Database management and analyses were performed using the Comprehensive Meta-Analysis software package (Version 1.0.23, Biostat, Inc, Englewood, NJ 07631, USA).
Table 2 Summary of analyses.*
Meta-Analyses Sensitivity Analyses
Overall Median Contrasts Best-case and worst-case scenarios
4 possible combinations of contrasts including one control treatment per trial
Cleary [47] trial excluded
All 20 contrasts
Stratified Median Contrasts
A. Control Treatment
1. Active treatment or placebo control Best-case and worst-case scenarios
2 possible combinations of contrasts including one control treatment per trial
Cleary [47] trial excluded
All 16 contrasts
2. No treatment
B. Country Where Trial was Performed
3. United Kingdom Best-case and worst-case scenarios
2 possible combinations of contrasts including one control treatment per trial
Cleary [47] trial excluded
All 10 contrasts
4. United States Best-case and worst-case scenarios
2 possible combinations of contrasts including one control treatment per trial
All 10 contrasts
C. Duration of Follow-Up
5. Short-term Best-case and worst-case scenarios
All 9 contrasts
6. Intermediate-term 4 possible combinations of contrasts including one control treatment per trial
7. Long-term 2 possible combinations of contrasts including one control treatment per trial
Cleary [47] trial excluded
*There were insufficient contrasts to perform sensitivity analyses for the no treatment stratified analysis. For the short-term stratified analysis, the median contrast was defined to be that corresponding to the eighth combination when effect sizes for the 16 possible contrast combinations were rank ordered from least to greatest. For the intermediate-and long-term stratified analyses, the median contrasts defaulted to the all-contrasts analyses because there were no repeated measures within these time intervals in any trial. All possible contrast combinations were included in the sensitivity analyses for intermediate-and long-term follow-up because of the limited numbers of combinations for these analyses.
Results
Overall analyses
The search for reports is summarized in Figure 1. A total of 525 subjects with low back pain were randomized in the eligible trials. The overall results are presented in Figure 2. There was a highly significant reduction in pain associated with OMT (effect size, -0.30; 95% CI, -0.47 – -0.13; P = .001). The Q statistic was non-significant, thus supporting the assumption of homogeneity among trials. The primary sensitivity analyses are presented in Table 3. Using a random-effects model, the results were virtually identical to those observed with a fixed-effects model. There were 729 (36 × 12) possible combinations of contrasts for analysis based on three contrasts for each of six OMT vs control treatment comparisons [42,43,45,46] and one contrast for each of the two remaining OMT vs control treatment comparisons [44,47]. The efficacy of OMT for low back pain was supported in both the best-case (effect size, -0.37; 95% CI, -0.55 – -0.20; P < .001) and worst-case (effect size, -0.18; 95% CI, -0.35 – 0.00; P = .046) scenarios. Similarly, when each trial was limited to one OMT vs control treatment comparison, OMT was found to be efficacious in each of the four analyses. OMT also demonstrated significantly greater low back pain reduction than control treatment in analyses with the Cleary [47] trial excluded and with all 20 contrasts included.
Figure 1 Flowchart of trials.
Figure 2 Effect size for low back pain. CI denotes confidence interval; OMT, osteopathic manipulative treatment. Overall effect size, -0.30; 95% CI, -0.47 – -0.13; P = .001.
Table 3 Overall results.
No. of Subjects
Model No. of
Contrasts OMT Control Effect
Size 95% CI P
Median contrasts
Fixed-effects model* 8 318 231 -0.30 -0.47 – -0.13 .001
Random-effects model 8 318 231 -0.31 -0.49 – -0.13 .001
Best-case scenario 8 293 220 -0.37 -0.55 – -0.20 <.001
Worst-case scenario 8 298 221 -0.18 -0.35 – 0.00 .046
Median contrasts, one OMT vs control treatment comparison per trial
Gibson [43] active treatment control and Licciardone [46] placebo control 6 237 181 -0.30 -0.49 – -0.10 .003
Gibson [43] active treatment control and Licciardone [46] no treatment control 6 247 179 -0.39 -0.59 – -0.20 <.001
Gibson [43] placebo control and Licciardone [46] placebo control 6 238 187 -0.26 -0.45 – -0.06 .01
Gibson [43] placebo control and Licciardone [46] no treatment control 6 248 185 -0.35 -0.54 – -0.15 <.001
Median contrasts, Cleary [47] trial excluded 7 310 227 -0.29 -0.47 – -0.12 .001
All contrasts 20 727 520 -0.29 -0.40 – -0.17 <.001
CI denotes confidence interval; OMT, osteopathic manipulative treatment.
*Test for homogeneity, P = .37.
Stratified analyses
The results of stratified meta-analyses are presented in Table 4. There was a significant reduction in low back pain associated with OMT in trials vs active treatment or placebo control (effect size, -0.26; 95% CI, -0.48 – -0.05; P = .02), independent of fixed-effects or random-effects model specification. There were 243 (35 × 11) possible contrast combinations based on three contrasts for each of five OMT vs control treatment comparisons [42,43,45,46] and one contrast for another remaining OMT vs control treatment comparison [47]. Both the best-case and worst-case scenarios demonstrated a greater reduction in pain with OMT than active treatment or placebo control, although the worst-case results did not achieve statistical significance. OMT was found to significantly reduce pain in the remaining analyses that limited OMT vs active treatment or placebo control comparisons to one per trial, excluded the Cleary [47] trial, and included all 16 contrasts. The OMT vs no treatment control comparisons were observed in trials in which all subjects received usual low back care in addition to their allocated treatment (ie, OMT and usual care vs only usual care) [44,47]. For these trials, the all-contrasts model (ie, the only model with sufficient contrasts for data synthesis) demonstrated a highly significant reduction in pain with OMT.
Table 4 Stratified results.
No. of Subjects
Model No. of
Contrasts OMT Control Effect
Size 95% CI P
OMT vs. Active Treatment or Placebo Control
Median contrasts
Fixed-effects model* 6 193 142 -0.26 -0.48 – -0.05 .02
Random-effects model 6 193 142 -0.26 -0.48 – -0.05 .02
Best-case scenario 6 174 132 -0.34 -0.57 – -0.11 .004
Worst-case scenario 6 183 134 -0.07 -0.29 – 0.16 .54
Median contrasts, one OMT vs control treatment comparison per trial
Gibson [43] active treatment 5 154 109 -0.33 -0.58 – -0.08 .01
Gibson [43] placebo control 5 155 115 -0.26 -0.51 – -0.02 .03
Median contrasts, Cleary [47] trial excluded 5 185 138 -0.24 -0.47 – -0.02 .03
All contrasts 16 534 400 -0.21 -0.34 – -0.08 .002
OMT vs. No Treatment Control
All contrasts 4 193 120 -0.53 -0.76 – -0.30 <.001
Trials Performed in the United Kingdom
Median contrasts
Fixed-effects model* 4 105 84 -0.29 -0.58 – 0.00 .050
Random-effects model 4 105 84 -0.30 -0.63 – 0.02 .06
Best-case scenario 4 105 88 -0.36 -0.64 – -0.07 .01
Worst-case scenario 4 100 83 -0.11 -0.40 – 0.19 .48
Median contrasts, one OMT vs control treatment comparison per trial
Gibson [43] active treatment 3 66 51 -0.46 -0.83 – -0.09 .02
Gibson [43] placebo control 3 67 57 -0.30 -0.66 – 0.05 .10
Median contrasts, Cleary [47] trial excluded 3 97 80 -0.26 -0.56 – 0.04 .09
All contrasts 10 294 247 -0.23 -0.40 – -0.06 .01
Trials Performed in the United States
Median contrasts
Fixed-effects model* 4 213 147 -0.31 -0.52 – -0.10 .004
Random-effects model 4 213 147 -0.32 -0.57 – -0.06 .01
Best-case scenario 4 188 132 -0.38 -0.61 – -0.16 .001
Worst-case scenario 4 198 138 -0.22 -0.44 – 0.00 .050
Median contrasts, one OMT vs control treatment comparison per trial
Licciardone [46] placebo control 3 171 130 -0.24 -0.47 – -0.01 .04
Licciardone [46] no treatment control 3 181 128 -0.36 -0.59 – -0.14 .002
All contrasts 10 433 273 -0.33 -0.48 – -0.18 <.001
Short-Term Follow-Up
Median contrasts
Fixed-effects model* 5 181 130 -0.28 -0.51 – -0.06 .01
Random-effects model 5 181 130 -0.31 -0.61 – -0.01 .046
Best-case scenario 5 196 142 -0.41 -0.62 – -0.19 <.001
Worst-case scenario 5 181 136 -0.10 -0.32 – 0.12 .38
All contrasts 9 357 258 -0.23 -0.39 – -0.07 .01
Intermediate-Term Follow-Up
Median (all) contrasts
Fixed-effects model* 7 283 209 -0.33 -0.51 – -0.15 <.001
Random-effects model 7 283 209 -0.36 -0.63 – -0.10 .01
Median contrasts, one OMT vs control treatment comparison per trial
Gibson [43] active treatment and Licciardone [46] placebo control 5 209 161 -0.31 -0.52 – -0.10 .004
Gibson [43] active treatment and Licciardone [46] no treatment control 5 209 158 -0.45 -0.65 – -0.24 <.001
Gibson [43] placebo control and Licciardone [46] placebo control 5 209 166 -0.25 -0.46 – -0.05 .02
Gibson [43] placebo control and Licciardone [46] no treatment control 5 209 163 -0.39 -0.59 – -0.18 <.001
Long-Term Follow-Up
Median (all) contrasts
Fixed-effects model* 4 87 53 -0.40 -0.74 – -0.05 .03
Random-effects model 4 87 53 -0.41 -0.82 – -0.01 .046
Median contrasts, one OMT vs control treatment comparison per trial
Licciardone [46] placebo control 3 55 38 -0.23 -0.65 – 0.19 .28
Licciardone [46] no treatment control 3 55 34 -0.64 -1.08 – -0.20 .01
Median contrasts, Cleary [47] trial excluded 3 79 49 -0.36 -0.72 – 0.01 .054
CI denotes confidence interval; OMT, osteopathic manipulative treatment.
*Tests of homogeneity, P = .45 and P = .06 for active treatment or placebo control, and no treatment control groups, respectively; P = .32 and P = .26 for trials in the United Kingdom and the United States, respectively; and P = .14, P = .06, and P = .28 for short-term, intermediate-term, and long-term follow-up, respectively.
Trials in both the United Kingdom (effect size, -0.29; 95% CI, -0.58 – 0.00; P = .050) and the United States (effect size, -0.31; 95% CI, -0.52 – -0.10; P = .004) demonstrated significant reductions in low back pain associated with OMT. In the sensitivity analyses, effect sizes were generally of comparable magnitude in both countries, although results in American trials consistently achieved statistical significance as a consequence of the larger sample sizes in these trials (Table 4).
There were significant reductions in low back pain associated with OMT during the short-term (effect size, -0.28; 95% CI, -0.51 – -0.06; P = .01), intermediate-term (effect size, -0.33; 95% CI, -0.51 – -0.15; P < .001), and long-term (effect size, -0.40; 95% CI, -0.74 – -0.05; P = .03) follow-up periods. Sensitivity analyses for temporal outcomes demonstrated that intermediate-term results consistently achieved statistical significance, generally because of the greater number of subjects in these analyses (Table 4). The results of the meta-analyses and sensitivity analyses are further summarized in Figure 3.
Figure 3 Summary of meta-analysis results. A denotes all-contrasts model; B, best-case scenario model; C, Cleary [47] trial excluded model; M, median contrasts model; NT, no treatment control; OMT, osteopathic manipulative treatment; W, worst-case scenario model. 1, 2, 3, and 4 indicate alternative models restricted to one OMT vs control treatment comparison per trial. A diamond indicates the inclusion of the relevant contrast or observation of the stated result. Sensitivity analyses are shaded in gray. Results are presented for each of the 43 analyses, including the overall meta-analysis, seven stratified meta-analyses, and 35 sensitivity analyses.
Discussion
Efficacy of osteopathic manipulative treatment
The overall results clearly demonstrate a statistically significant reduction in low back pain with OMT (Figure 2). Further, the meta-analysis results are quite robust as indicated by the comprehensive sensitivity analyses (Figure 3). Stratified meta-analyses to control for moderator variables demonstrated that OMT significantly reduced low back pain vs active treatment or placebo control and vs no treatment control. If it is assumed, as shown in a review [52], that the effect size is -0.27 for placebo control vs no treatment in trials involving continuous measures for pain, then the results of our study are highly congruent (ie, effect size for OMT vs no treatment [-0.53] = effect size for OMT vs active treatment or placebo control [-0.26] + effect size for placebo control vs no treatment [-0.27]).
It has been suggested that the therapeutic benefits of spinal manipulation are largely due to placebo effects [53]. A preponderance of results from our sensitivity analyses supports the efficacy of OMT vs active treatment or placebo control and therefore indicates that low back pain reduction with OMT is attributable to the manipulation techniques, not merely placebo effects. Also, as indicated above, OMT vs no treatment control demonstrated pain reductions twice as great as previously observed in clinical trials of placebo vs no treatment control [52]. Thus, OMT may eliminate or reduce the need for drugs that can have serious adverse effects [44].
Because osteopathic physicians provide OMT to complement conventional treatment for low back pain, they tend to avoid substantial additional costs that would otherwise be incurred by referring patients to chiropractors or other practitioners [54]. With respect to back pain, osteopathic physicians make fewer referrals to other physicians and admit a lower percentage of patients to hospitals than allopathic physicians [1], while also treating back pain episodes with substantially fewer visits than chiropractors [55]. Although osteopathic family physicians are less likely to order radiographs or prescribe nonsteroidal anti-inflammatory drugs, aspirin, muscle relaxants, sedatives, and narcotic analgesics for low back pain than their allopathic counterparts, osteopathic physicians have a substantially higher proportion of patients returning for follow-up back care than allopathic physicians [56]. In the United Kingdom, where general practitioners may refer patients with spinal pain to osteopaths for manipulation, it has been shown that OMT improved physical and psychological outcomes at little extra cost [57].
In our study, the effect sizes for OMT in the United Kingdom, where osteopaths are not licensed physicians, were generally comparable to those in the United States, where OMT is provided by licensed physicians. This consistency suggests that the results truly reflect the effects of OMT itself, and not other elements of low back care. It is not surprising that osteopaths in the United Kingdom achieved pain reduction with OMT similar to that of their physician counterparts in the United States. The training of osteopaths in the United Kingdom is highly focused on OMT, whereas osteopathic physicians undertake a medical curriculum that necessarily relegates OMT to one of many therapeutic approaches, albeit a fundamental one for osteopathic practitioners. Regardless of the career training path of the provider, it appears that OMT achieves clinically important reductions in low back pain.
Potential limitations
There are several potential limitations of this study that should be addressed. First, as with any meta-analysis, the individual trials varied somewhat with respect to methodology, including trial setting, subject characteristics, OMT and control treatment interventions, and pain measures (Table 1). Such heterogeneity has been commonly observed in previous meta-analyses of spinal manipulation, including a recent meta-analysis performed in collaboration with the Cochrane Back Review Group [31]. The latter study addressed potential heterogeneity by presenting stratified results according to chronicity of low back pain, type of control group, and duration of follow-up. This approach is analogous to the methods used in our study. Further, it should be noted that the assumption of homogeneity among trials was not rejected statistically in any of our eight overall or stratified median contrasts meta-analyses.
Second, because five trials each included repeated pain measures and two trials each included two control treatments, there was no unique set of independent outcomes for meta-analysis. Such interdependencies were noted to be a problem in an early meta-analysis of spinal manipulation [15]. We used the median contrasts method to address this problem because the median outcome represents an observed outcome that is easy to compute and is less vulnerable to extreme observations than other measures of central tendency. Further, sensitivity analysis was used to assess the range of possible combinations of outcomes. Thus, for the overall meta-analysis, there were 729 potential contrast combinations. Of these, both the best-case and worst-case scenarios demonstrated statistically significant results favoring OMT, thereby providing unequivocal evidence for the efficacy of OMT. Robust findings were also observed for trials performed in the United States and for intermediate-term outcomes.
Third, because there were a relatively small number of eligible trials, there were not sufficient contrasts for certain analyses and some results were imprecise. The latter phenomenon likely obviated the statistical significance of some results. Nevertheless, it is important to note that the direction of results favored OMT in each of the 43 meta-analyses and sensitivity analyses presented herein (Figure 3).
Fourth, there exists the possibility that the results of unpublished trials of OMT for low back pain may have altered significantly the conclusions of this study. To address this issue, we performed file drawer analysis by computing the fail-safe N [58]. This represents the number of unpublished trials of OMT for low back pain that would have met our inclusion criteria, and that also would have demonstrated an effect size averaging ≥ -0.10, which is assumed to reflect clinically insignificant levels of pain reduction. A total of 16 unpublished trials (assuming one control group per trial) with, in the aggregate, clinically insignificant pain reduction outcomes would have been needed to obviate the significance of our results. Only recently has government funding for research in the area of complementary and alternative medicine become more widely available, in response to the public's interest in such treatments. Historically, it is highly unlikely that 16 trials of OMT for low back pain would have been sponsored, conducted, and subsequently not published.
Finally, this study focused only on the efficacy of OMT with respect to pain outcomes. Generic health status, back-specific function, work disability, and back-specific patient satisfaction are other recommended outcome domains [59] that were not assessed because the included OMT trials did not consistently report these data.
Conclusion
The present study indicates that OMT is a distinctive modality that significantly reduces low back pain. The level of pain reduction is greater than expected from placebo effects alone and persists for at least three months. Additional research is warranted to elucidate mechanistically how OMT exerts its effects, to determine if OMT benefits are long lasting, and to assess the cost-effectiveness of OMT as a complementary treatment for low back pain.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JCL, AKB, and LNK conceived and designed the study. LNK performed the literature searches. JCL and AKB extracted the data. JCL performed the statistical analyses. JCL, AKB, and LNK interpreted the data and drafted the manuscript. JCL will act as guarantor for the paper. The guarantor accepts full responsibility for the conduct of the study, had access to the data, and controlled the decision to publish. All authors approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
this file provides the timetable, databases, and search terms used to identify relevant studies for the meta-analysis.
Click here for file
Acknowledgements
This research was supported in part by a grant (No. D56HP00170) from the Health Resources and Services Administration, United States Department of Health and Human Services. The funding organization had no role in the design, conduct, and reporting of this study.
==== Refs
Cypress BK Characteristics of physician visits for back symptoms: a national perspective Am J Public Health 1983 73 389 395 6219588
Licciardone JC Herron KM Characteristics, satisfaction, and perceptions of patients receiving ambulatory healthcare from osteopathic physicians: a comparative national survey J Am Osteopath Assoc 2001 101 374 385 11476027
Licciardone JC Awareness and use of osteopathic physicians in the United States: results of the Second Osteopathic Survey of Health Care in America (OSTEOSURV-II) J Am Osteopath Assoc 2003 103 281 289 12834101
Agency for Health Care Policy and Research [AHCPR] Acute low back problems in adults Clinical Practice Guideline 14 1994 Rockville, MD, US Department of Health and Human Services
Kuchera ML DiGiovanna EL Greenspan PE Ward RC Efficacy and complications Foundations for Osteopathic Medicine 2003 2 Philadelphia, PA: Lippincott Williams & Wilkins 1143 1152
Harvey E Burton AK Moffett JK Breen A Spinal manipulation for low-back pain: a treatment package agreed by the UK chiropractic, osteopathy and physiotherapy professional associations Man Ther 2003 8 46 51 12635637 10.1054/math.2002.0472
UK BEAM Trial Team United Kingdom back pain exercise and manipulation (UK BEAM) randomised trial: effectiveness of physical treatments for back pain in primary care BMJ 2004 329 1377 doi:10.1136/bmj.38282.669225.AE. 15556955 10.1136/bmj.38282.669225.AE
UK BEAM Trial Team United Kingdom back pain exercise and manipulation (UK BEAM) randomised trial: cost effectiveness of physical treatments for back pain in primary care BMJ 2004 329 1381 doi:10.1136/bmj.38282.607859.AE. 15556954 10.1136/bmj.38282.607859.AE
Lesho EP An overview of osteopathic medicine Arch Fam Med 1999 8 477 484 10575385 10.1001/archfami.8.6.477
Mein EA Greenman PE McMillin DL Richards DG Nelson CD Manual medicine diversity: research pitfalls and the emerging medical paradigm J Am Osteopath Assoc 2001 101 441 444 11526876
Balon J Aker PD Crowther ER Danielson C Cox PG O'Shaughnessy D Walker C Goldsmith CH Duku E Sears MR A comparison of active and simulated chiropractic manipulation as adjunctive treatment for childhood asthma N Engl J Med 1998 339 1013 20 9761802 10.1056/NEJM199810083391501
Nelson CD McMillin DL Richards DG Mein EA Redwood D Manual healing diversity and other challenges to chiropractic integration J Manipulative Physiol Ther 2000 23 202 207 10771505 10.1016/S0161-4754(00)90250-1
Bronfort G Haas M Evans RL Bouter LM Efficacy of spinal manipulation and mobilization for low back pain and neck pain: a systematic review and best evidence synthesis Spine J 2004 4 335 356 15125860 10.1016/j.spinee.2003.06.002
Howell JD The paradox of osteopathy N Engl J Med 1999 341 1465 1468 10547412 10.1056/NEJM199911043411910
Ottenbacher K DiFabio RP Efficacy of spinal manipulation/mobilization therapy: a meta-analysis Spine 1985 10 833 837 2935951
Koes BW Assendelft WJJ van der Heijden GJMG Bouter LM Knipschild PG Spinal manipulation and mobilisation for back and neck pain: a blinded review BMJ 1991 303 1298 1303 1836153
Abenhaim L Bergeron AM Twenty years of randomized clinical trials of manipulative therapy for back pain: a review Clin Invest Med 1992 15 527 535 1286535
Anderson R Meeker WC Wirick BE Mootz RD Kirk DH Adams A A meta-analysis of clinical trials of spinal manipulation J Manipulative Physiol Ther 1992 15 181 194 1533416
DiFabio RP Efficacy of manual therapy Phys Ther 1992 72 853 864 1454861
Shekelle PG Adams AH Chassin MR Hurwitz EL Brook RH Spinal manipulation for low-back pain Ann Intern Med 1992 117 590 598 1388006
Scheer SJ Radack KL O'Brien DR Randomized controlled trials in industrial low back pain relating to return to work. Part 1. Acute interventions Arch Phys Med Rehabil 1995 76 966 973 7487440 10.1016/S0003-9993(95)80076-X
Koes BW Assendelft WJJ van der Heijden GJMG Bouter LM Spinal manipulation for low back pain: an updated systematic review of randomized clinical trials Spine 1996 21 2860 2873 9112710 10.1097/00007632-199612150-00013
Scheer SJ Radack KL O'Brien DR Randomized controlled trials in industrial low back pain relating to return to work. Part 2. Discogenic low back pain Arch Phys Med Rehabil 1996 77 1189 1197 8931535 10.1016/S0003-9993(96)90147-1
Scheer SJ Watanabe TK Radack KL Randomized controlled trials in industrial low back pain. Part 3. Subacute/chronic pain interventions Arch Phys Med Rehabil 1997 78 414 423 9111463 10.1016/S0003-9993(97)90235-5
van der Weide WE Verbeek JHAM van Tulder MW Vocational outcome of intervention for low-back pain Scand J Work Environ Health 1997 23 165 178 9243726
van Tulder MW Koes BW Bouter LM Conservative treatment of acute and chronic nonspecific low back pain: a systematic review of randomized controlled trials of the most common interventions Spine 1997 22 2128 2156 9322325 10.1097/00007632-199709150-00012
Bronfort G Spinal manipulation: current state of research and its indications Neurol Clin 1999 17 91 111 9855673 10.1016/S0733-8619(05)70116-X
Ernst E Harkness E Spinal manipulation: a systematic review of sham-controlled, double-blind, randomized clinical trials J Pain Symptom Manage 2001 22 879 889 11576805 10.1016/S0885-3924(01)00337-2
Ferreira ML Ferreira PH Latimer J Herbert R Maher CG Does spinal manipulative therapy help people with chronic low back pain? Aust J Physiother 2002 48 277 284 12443522
Pengel HM Maher CG Refshauge KM Systematic review of conservative interventions for subacute low back pain Clin Rehabil 2002 16 811 820 12501942 10.1191/0269215502cr562oa
Assendelft WJJ Morton SC Yu EI Suttorp MJ Shekelle PG Spinal manipulative therapy for low back pain: a meta-analysis of effectiveness relative to other therapies Ann Intern Med 2003 138 871 881 12779297
Cherkin DC Sherman KJ Deyo RA Shekelle PG A review of the evidence for the effectiveness, safety, and cost of acupuncture, massage therapy, and spinal manipulation for back pain Ann Intern Med 2003 138 898 906 12779300
Haldeman S Hooper PD Phillips RB Scaringe JG Traina AD Frymoyer JW Spinal manipulative therapy The Adult Spine: Principles and Practice 1997 2 Philadelphia, PA: Lippincott-Raven Publishers 1837 1861
Siehl D Olson DR Ross HE Rockwood EE Manipulation of the lumbar spine with the patient under general anesthesia: evaluation by electromyography and clinical-neurologic examination of its use for lumbar nerve root compression syndrome J Am Osteopath Assoc 1971 70 433 440 5203536
Doran DML Newell DJ Manipulation in treatment of low back pain: a multicentre study Br Med J 1975 2 161 164 123815
Hadler NM Curtis P Gillings DB Stinnett S A benefit of spinal manipulation as adjunctive therapy for acute low-back pain: a stratified controlled trial Spine 1987 12 703 706
Ellestad SM Nagle RV Boesler DR Kilmore MA Electromyographic and skin resistance responses to osteopathic manipulative treatment for low-back pain J Am Osteopath Assoc 1988 88 991 997 2975645
MacDonald RS Bell CMJ An open controlled assessment of osteopathic manipulation in nonspecific low-back pain Spine 1990 15 364 370 2141951
Boesler D Warner M Alpers A Finnerty EP Kilmore MA Efficacy of high-velocity low-amplitude manipulative technique in subjects with low-back pain during menstrual cramping J Am Osteopath Assoc 1993 93 203 214 8432669
Hoffman KS Hoffman LL Effects of adding sacral base leveling to osteopathic manipulative treatment of back pain: a pilot study J Am Osteopath Assoc 1994 94 217 226 8200825
Patterson MM Ward RC Foundations for osteopathic medical research Foundations for Osteopathic Medicine 2003 2 Philadelphia, PA: Lippincott Williams & Wilkins 1167 1187
Hoehler FK Tobis JS Buerger AA Spinal manipulation for low back pain JAMA 1981 245 1835 1838 6453240 10.1001/jama.245.18.1835
Gibson T Grahame R Harkness J Woo P Blagrave P Hills R Controlled comparison of short-wave diathermy treatment with osteopathic treatment in non-specific low back pain Lancet 1985 1 1258 1261 2860453 10.1016/S0140-6736(85)92323-2
Andersson GBJ Lucente T Davis AM Kappler RE Lipton JA Leurgans S A comparison of osteopathic spinal manipulation with standard care for patients with low back pain N Engl J Med 1999 341 1426 1431 10547405 10.1056/NEJM199911043411903
Burton AK Tillotson KM Cleary J Single-blind randomised controlled trial of chemonucleolysis and manipulation in the treatment of symptomatic lumbar disc herniation Eur Spine J 2000 9 202 207 10905437 10.1007/s005869900113
Licciardone JC Stoll ST Fulda KG Russo DP Siu J Winn W Swift J Osteopathic manipulative treatment for chronic low back pain: a randomized controlled trial Spine 2003 28 1355 1362 12838090 10.1097/00007632-200307010-00002
Cleary C Fox JP Menopausal symptoms: an osteopathic investigation Complement Ther Med 1994 2 181 186 10.1016/0965-2299(94)90017-5
Ernst E Another negative trial of manipulative treatment for back pain Focus Altern Complement Ther 2004 9 43 44
Margo K Cochrane for clinicians: putting evidence into practice. Spinal manipulative therapy for low back pain Am Fam Physician 2005 71 464 465 15712619
Rosenthal R Cooper H, Hedges LV Parametric measures of effect size The Handbook of Research Synthesis 1994 New York, NY: Russell Sage Foundation 231 244
Hedges LV Olkin I Statistical Methods for Meta-Analysis 1985 Boston, MA: Academic Press
Hróbjartsson A Gøtzsche PC Is the placebo powerless? An analysis of clinical trials comparing placebo with no treatment N Engl J Med 2001 344 1594 1602 11372012 10.1056/NEJM200105243442106
Ernst E Does spinal manipulation have specific treatment effects? Fam Pract 2000 17 554 556 11120730 10.1093/fampra/17.6.554
Reilly BM Hart A Evans AT Part II. Evidence-based medicine: a passing fancy or the future of primary care? Dis Mon 1998 44 370 399 9735941 10.1016/S0011-5029(98)90006-2
Shekelle PG Markovich M Louie R Factors associated with choosing a chiropractor for episodes of back pain care Med Care 1995 33 842 850 7637405
Hart LG Deyo RA Cherkin DC Physician office visits for low back pain: frequency, clinical evaluation, and treatment patterns from a U.S. national survey Spine 1995 20 11 19 7709270
Williams NH Wilkinson C Russell I Edwards RT Hibbs R Linck P Muntz R Randomized osteopathic manipulation study (ROMANS): pragmatic trial for spinal pain in primary care Fam Pract 2003 20 662 669 14701889 10.1093/fampra/cmg607
Wolf FM Lewis-Beck MS (Series) Meta-analysis: quantitative methods for research synthesis Quantitative Applications in the Social Sciences 1986 Newbury Park, CA: Sage Publications, no. 07–059
Bombardier C Outcome assessments in the evaluation of treatment of spinal disorders: summary and general recommendations Spine 2000 25 3100 3103 11124724 10.1097/00007632-200012150-00003
|
16080794
|
PMC1208896
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Aug 4; 6:43
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-43
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-451611783010.1186/1471-2474-6-45Research ArticleKnee disorders in primary care: design and patient selection of the HONEUR knee cohort Heintjes Edith M [email protected] Marjolein Y [email protected] Bart W [email protected] Sita M [email protected] Department of General Practice, Erasmus University Medical Centre, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands2005 23 8 2005 6 45 45 22 4 2005 23 8 2005 Copyright © 2005 Heintjes et al; licensee BioMed Central Ltd.2005Heintjes et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Knee complaints are a frequent reason for consultation in general practice. These patients constitute a specific population compared to secondary care patients. However, information to base treatment decisions on is generally derived from specialistic settings. Our cohort study is aimed at collecting knowledge about prognosis and prognostic factors of knee complaints presented in a primary care setting. This paper describes the methods used for data collection, and discusses potential selectiveness of patient recruitment.
Methods
This is a descriptive prospective cohort study with one-year follow-up. 40 Dutch GPs recruited consecutive patients with incident knee complaints aged 12 years and above from October 2001 to October 2003. Patients were assessed with questionnaires and standardised physical examinations. Additional measurements of subgroups included MRI for recent knee traumas and device assessed function measurements for non-traumatic patients.
After the inclusion period we retrospectively searched the computerized medical files of participating GPs to obtain a sample to determine possible selective recruitment. We assessed differences in proportions of gender, traumatic onset of injury and age groups between participants and non-participants using Odds Ratios (OR) and 95% confidence intervals.
Results
We recruited 1068 patients. In a sample of 310 patients visiting the GP, we detected some selective recruitment, indicating an underrepresentation of patients aged 12 to 35 years (OR 1.70; 1.15–2.77), especially among men (OR 2.16; 1.12–4.18). The underrepresentation of patients with traumatic onset of injury was not statistically significant.
Conclusion
This cohort is unique in its size, setting, and its range of both age and type of knee complaints. We believe the detected selective recruitment is unlikely to introduce significant bias, as the cohort will be divided into subgroups according to age group or traumatic onset of injury for future analyses. However, the underrepresentation of men in the age group of 12 to 35 years of age warrants caution. Based on the available data, we believe our cohort is an acceptable representation of patients with new knee complaints consulting the GP, and we expect no problems with extrapolation of the results to the general Dutch population.
==== Body
Background
Knee complaints rank among the most frequent reasons for consulting primary care physicians. A nationwide study into the incidence and prevalence of diseases and complaints in Dutch General Practices revealed that the incidence of unspecified knee complaints in General Practice is 13.7 per 1000 patients per year, ranking 16th in the list of most frequent reasons for visiting the General Practitioner (GP). Specified knee complaints (knee distortion, acute injury to meniscus or ligaments, chronic internal traumatic knee injuries, knee osteoarthritis, and Osgood Schlatter) account for an incidence 11.3 per 1000 on top of that [1].
Nonetheless, clinical research in this area is usually carried out in hospital settings and only covers serious or persistent injuries, usually meeting stringent inclusion criteria. The applicability of results from this research to patients presenting knee complaints in general practice is therefore limited. Open population studies [2,3] offer a broader view of knee complaints, but often target specific age groups and also include patients that do not seek medical care for their complaints. To our knowledge, publication of studies dealing with patients with knee disorders in general practice is limited to cross-sectional registration studies that report incidence and prevalence of diagnostic codes and their corresponding referral rates to physical therapy or specialist care [1]. This type of study is not informative with respect to disease burden, the (natural) course of complaints, treatments strategies or even diagnosis, because the diagnostic codes are often non-specific. As a result, our understanding of knee complaints in primary care is far from complete. But knowledge about the determinants of the clinical course is essential for making management decisions and to inform patients about their prognosis. Furthermore, decisions about management and referral of knee complaints in primary care are to a large extent based on test results from physical examination. Physical signs and symptoms may also play an important part in predicting the course of knee complaints. Nevertheless, the value of physical examination in general practice has never been evaluated.
To fill in the gaps in the information available to GPs, we performed a prospective, observational cohort study including the whole range of incident knee complaints presented to the GP, by adolescents as well as adults. The primary objectives of our cohort study are as follows:
1. What type of knee complaints are presented to the GP, and what is their severity and impact on daily activities?
2. What is the one-year prognosis of knee complaints presented to the GP?
3. What are the factors predicting prognosis?
4. How are knee complaints managed by GPs?
The wide range of knee complaints included in our cohort study enables us to focus on specific aspects for specific subgroups and on the validity of measurement tools in a primary care setting. Therefore our secondary objectives for specific subgroups are as follows:
1. What is the predictive value of physical examination and history taking for detecting lesions that can be seen with Magnetic Resonance Imaging (MRI) in patients with acute traumatic knee injuries in General Practice?
2. What is the additive predictive value of MRI over physical examination and history taking for the prognosis of knee complaints in patients with acute traumatic knee injuries in General Practice?
3. What is the validity and responsiveness of disease specific questionnaire assessed disability measurements compared to device assessed disability measurements?
In this paper we will outline the composition of our cohort and define its subgroups. The objectives of our cohort demand that we give an accurate account of the population of patients that visit the GP with knee complaints. As we depended on active cooperation from the GPs for recruitment of patients, we need to ascertain that our cohort represents this population. Therefore objectives for the present paper are twofold:
1. To describe the methods used for data collection
2. To determine whether the recruitment procedures resulted in a patient selection that accurately represents the patients visiting the GP.
Methods
Design
This is a prospective, observational cohort study, with a follow-up period of one year. Data were collected using questionnaires and physical examinations. The researchers did not interfere with usual care with respect to advice, diagnostics or treatment. The study was approved by the ethics committee of the Erasmus Medical Centre Rotterdam.
Inclusion criteria
Patients aged 12 years or above, consulting their GP for a new episode of knee complaints, were invited to participate in the study. New complaints were defined as complaints that were presented to the GP for the first time. Recurrent complaints for which the GP was not consulted within the last 3 months were also considered new complaints. Knee complaints that required urgent medical attention, such as fractures or infections were excluded. Patients with malignancies, neurological disorders or systemic musculoskeletal diseases that affect the outcome measures used in this study (i.e. Parkinson's disease, Rheumatoid Arthritis, Amyotrophic Lateral Sclerosis, etc.), as well as patients that were incapable of understanding the ramifications of participation, were excluded from participation.
Recruitment
40 GPs from 5 municipalities in the southwest region of the Netherlands, connected to the Erasmus Medical Centre GP Research Network HONEUR and representing a total patient population of around 84.000 patients, participated in this prospective cohort study. We started recruitment in October 2001 in 1 municipality and a new municipality was added approximately every 3 months. All GPs recruited up to October 2003.
Patients were alerted to the existence of the study through posters in the waiting room. Participation of patients was voluntary and did not affect the care given to the patient. Patients received no compensation for participation. During consultation, the GP briefly informed the patient of the existence of the study and handed over written information and a baseline questionnaire. Interested patients forwarded their contact details to the researchers. The researchers contacted the patients to further inform patients of the study and to make an appointment for signing informed consent and performing a comprehensive standardized physical examination of both knees. Informed consent forms for minors (aged 12 through 17) were co-signed by a parent or guardian.
Participating GPs agreed to note the following items in their computerised medical files: relevant anamnestic findings, treatment details, a preliminary prognosis, and a diagnostic code from the International Classification of Primary Care (ICPC) [4], chosen from a list provided by the researchers.
Physical examination
Two physiotherapists employed as research assistants (DC and EB) developed the standardized protocol for physical examination under supervision of two physiotherapists with over ten years of experience in both physiotherapy and research (SMAB and HW). Standardisation of the examinations among research assistants was accomplished by a series of training sessions before starting the inclusion of patients. These training sessions were repeated regularly over the course of the inclusion period. In total five physiotherapists (DC, EB, CV, AV and RvB) with clinical experience varying from one to 14 years performed the physical examinations of the patients.
The physical examination was planned as close to the date of consultation of the GP as possible. Irrespective of the type of symptoms presented, a standard range of tests was performed on both knees. The physical examination covered inspection of postural aspects, signs of inflammation, tests of swelling, locating tender areas, patellofemoral joint compression, crepitus, knee extensor and flexor strength, joint laxity, range of motion and meniscus tests (see table 1).
Table 1 Item list for physical examination
inspection [18, 19] palpation [18, 19] specific diagnostic tests
coloration temperature sustained flexion test [20]
valgisation/varisation swelling: balottable patella sign patellar grinding test [19]
overextension/limited extension swelling: fluid shift/fluctuation sign patellar axial pressure test [21]
tibial tuber swelling pain tibial tuber patellar apprehension test [21]
atrophy quadriceps pain joint line Steinmann II test [22]
flexion contracture hip pain patellar edges McMurray test [23]
internal/external rotation femur pain patellar ligament Apley's grind/traction tests [24]
internal/external rotation tibia pain collateral lateral/medial ligaments valgus/varus test [25]
foot pronation pain insertion pes anserinus anterior drawer test [25]
leg length difference pain insertion iliotibial band Lachman test [26]
swelling fossa poplitea/Baker's cyst pivot shift test [27]
function assessment hypertrophy synovial plica posterior drawer test [25]
flexion/extension active/passive [18, 19] bursa prepatellaris pain/swelling tibial posterior sag [28]
resisted flexion/extension [18, 19] bursa infrapatellaris pain/swelling
Discussion about diagnosis and/or appropriate management between patient and physiotherapist was discouraged, to avoid influencing the management initiated by the GP. The physical examination was repeated after one year, to enable comparison of perceived recovery with changes in test results.
Self-report questionnaires
Baseline questionnaires were filled in by the patients before the baseline visit, and checked for completeness by the physiotherapist during the baseline visit after physical examination. Any uncertainties on behalf of the patient were discussed at that point and any necessary corrections made accordingly. The three monthly follow-up questionnaires were mailed to the participants, and returned by mail, except for the last questionnaire, which coincided with the follow-up physical examination. The questionnaires included possible prognostic factors as well as outcome measures. Details of questionnaire items are listed in table 2. For possible prognostic factors we enquired after socio-economic status, comorbidity, history of knee complaints, characteristics of the knee complaints, daily activities and coping behaviour. To determine whether the complaints were recurrent, we asked patients if they had experienced similar knee complaints in the past, with complaints disappearing at least several weeks before returning again now. We also asked if they had consulted the GP for that previous episode. Occupations were accredited with a level of knee loading ranging from 1 (e.g. office jobs) to 3 (e.g. construction workers and mail men) and sports activities with a level from 1 (strolling and swimming) to 5 (contact sports). Physical activities from level 2 upward are considered substantial knee loading sports activities. Contact sports and sports involving rapid changes of direction are considered heavy loading activities (levels 4 and 5).
Table 2 Questionnaire items
item evaluation at (months) validation/reliability
demographics
age 0 -
gender 0 -
composition of the household 0 -
type of medical insurance 0 -
education level 0 -
comorbidity 0 -
knee complaints
history, duration, recurrence, consultation previous episode, perceived cause of knee complaint, mechanism of traumatic injury 0 -
pain 11 point numeric rating scale 0, 3, 6, 9, 12 Likert scales have compared favourably to visual analogue scales for children and adults [5-7]
Lysholm knee scale [8] 0, 3, 6, 9, 12 developed for ligament ruptures, sensitive and reliable for meniscus tears, (patellar) chondral disorders [11, 12]
Knee Society Score [29] 0, 12 Intra/interobserver reliabililty poor [30]
- function score (patient questionaire) 0, 12 - function score moderate agreement
- knee score (observer, physical exam) 0, 12 - knee score poor agreement
WOMAC osteoarthritis index [9, 10] 0, 3, 6, 9, 12 validated and reliable for osteoarthritis [10]
pain and difficulty with cycling, running, jumping, squatting, kneeling 0, 3, 6, 9, 12
knee loading
daily activities: employment, volunteer jobs, household chores, study: 0, 3, 6, 9, 12 -
physical exercise/sports participation Frequency, intensity, duration, association with knee complaints 0, 12 -
impact of knee complaint
hindrance during daily activities sick leave from daily activities 0, 3, 6, 9, 12
health related quality of life
SF-36 [31-33] 0, 3, 6, 9, 12 sensitive to change in common orthopaedic diagnoses [14], invalid for adolescents [15]
COOP/WONCA charts [34, 35] 0, 3, 6, 9, 12 valid for adults [34]
treatment
advise given by the GP 0 -
knee medication, dose, frequency, duration, form of administration 0, 3, 6, 9, 12 -
medication for comorbidity 0, 3, 6, 9, 12 -
visits to health care professionals follow-up -
operations follow-up -
Coping
Tampa Kinesiofobia Scale, (TKS) [36] catastrophizing 0
Medical advice and interventions by the GP were recorded at baseline. During follow-up patients also recorded visits to other medical professionals with a short description of interventions.
Outcome measures
Patients filled in their experienced recovery after one year on a 7 point Likert scale, ranging from 'fully recovered' to 'worse than ever'. Pain intensity was determined using a numeric rating scale (NRS) ranging from 0 (no pain) to 10 (unbearable pain). Numeric rating scales have compared favourably to visual analogue scales for children [5,6] as well as adults [7], though the number of points on these scales differed. Function assessments on disability level were determined using the Lysholm knee scoring scale (0–100) [8] and the WOMAC Hip and Knee Osteoarthritis Index (0–100) [9,10]. The Lysholm knee scoring scale was developed for ligament injuries, but was validated for use in various other knee disorders as well [11,12]. The questions from the WOMAC Hip and Knee Osteoarthritis Index were adapted to specifically address only the knee complaints.
The SF-36 was chosen for the assessment of health related quality of life because of its responsiveness [13] and its sensitivity to change in common orthopaedic diagnoses [14], though it has been shown to be invalid in adolescents [15]. We therefore also included the COOP/WONCA charts, which have not been validated in adolescents, but can be easily interpreted through illustrations. From the age of 18, patients filled in both SF-36 and COOP/WONCA charts, younger patients only filled in the COOP/WONCA charts.
Definition of subgroups
As different pathologies are expected to show different prognoses, we defined three subgroups. Patients with non-traumatic knee complaints are divided into a group aged 12 to 35 (I) and a group aged 36 years and over (II), because around 35 years of age the predominance of specific diagnoses shifts from patellofemoral pain syndrome [16] to osteoarthritis [1]. The group of patients with traumatic knee complaints (III) includes all patients whose knee complaints were caused by a sudden impact or wrong movement within one year before consulting the GP. All other patients were considered to have non-traumatic complaints, based on the assumption that the immediate effects of traumatic injuries will have worn off after one year.
Additional MRI
Patients aged between 18 and 65 years of age with an onset of trauma up to 5 weeks before consulting the GP were invited to participate in an additional MRI study. Participants were informed that patient and GP would not be informed of the presence or absence of detected lesions to prevent influencing the treatment strategy employed by the GP. Exceptions to this rule were lesions where urgent intervention was deemed necessary. For this additional study patients signed an additional informed consent form. MRI was performed between 3 to 6 weeks after the initial trauma, to allow swelling to subside while still observing the relatively acute stages of trauma. Following MRI a trained physiotherapist repeated the standardised physical examination. The assessors performing MRI and physical examination were blind to each other's results. The patients themselves recorded pain intensity and Lysholm score. After one year MRI and physical examination were repeated. If participants consulted medical specialists at a later date, the specialists were able to request MRI reports to prevent unnecessary duplication of diagnostic procedures.
Device assessed disability
Adult patients with non-traumatic knee complaints living in three municipalities close to the research facility were invited to participate in the additional device assessed disability measurement using the Dynaport knee scoring system [17]. This system registers accelerations of torso, hip, upper and lower legs during simulations of daily activities like walking stairs, sitting down, or walking with grocery bags or loaded trolleys. Measurements were repeated after 6 months.
Assessment of selective recruitment
To check whether the cohort adequately represents the patients that consulted the GP with a new episode of knee complaints during the inclusion period, the GPs computerized patient records were searched retrospectively for all occurrences of the relevant ICPC codes after the inclusion period had ended. As data collection from medical files is very labour-intensive, the search in each general practice was limited to a randomly assigned 4-month period within the total recruitment period. Within each municipality we made sure the 4-month periods covered all seasons. From all identified patients ICPC-code, diagnosis, age, sex and possible reasons for exclusion were registered anonymously on structured forms. ICPC-code and textual notes were both taken into account to determine whether onset of symptoms was considered traumatic by the GP. From the collected data we determined whether patients were eligible for inclusion in the cohort study. Eligible patients were then dichotomized according to age (12 to 35 years or above), gender, and traumatic onset of knee complaints. Participation rates within these dichotomized subgroups were compared using Odds Ratios (OR) and 95% confidence intervals, indicating their relative chances for inclusion into the study. Within each dichotomised subgroup we again compared the participation rates for the two other patient characteristics.
As the randomly chosen sample period was in itself a potential cause for bias of our analysis, we also compared the proportions of age, gender, and traumatic onset of knee complaints between participants from the sample periods and participants from the rest of the inclusion period. Furthermore, we compared proportions of ICPC codes in participants and non-participants from the sample periods, and in the participants of the sample periods and participants of the rest of the inclusion period, using Chi-square tests. For the statistical analyses we used SPSS for Windows, release 11.0.1.
Study sample
General cohort
Of the 1261 patients that forwarded their contact details to the researchers, 1068 (85%) signed informed consent. Reasons stated by contacted patients for non-participation are listed in table 3. The majority stated lack of time for participation (37%) or lack of personal gain (24%). The category miscellaneous included family circumstances, other health problems, language problems, and several patients judged themselves too old for participation. Ten patients were excluded: five because they were under 12 years of age, three because their complaints were not new, one because of rheumatoid arthritis and one patient was hospitalized with a bacterial infection of the knee.
Table 3 Reasons for non-participation of patients that forwarded their contact details to the researchers
N
Lack of time/could get no time off from work/could not make an appointment for examination 72
No personal gain/too much bother 47
No longer any complaints at time of contact and no longer interested 19
Could not be contacted 15
Miscellaneous 30
Non-compliance with inclusion criteria 10
The flow diagram (fig. 1) shows the distribution of participants over different subgroups of the cohort and the additional measurements. 51% of the participants were assigned to the subgroup of non-traumatic knee complaints in patients aged over 35 years of age. 18% were assigned to the group of non-traumatic knee complaints in the age of 12 to 35 years. 31% were assigned to the traumatic knee injury subgroup. Figure 2 shows the age distribution of the entire cohort, identifying subgroups and additional measurements. The percentage of female participants in each subgroup was 50%, 47% and 44% respectively.
Figure 1 Patient recruitment and subgroup assignment.
Figure 2 Age distributions of subgroups. Proportion of traumatic injuries and additional measurements per age category.
Of 1031 of the 1068 participants both questionnaire and physical examination results are available; 27 (2.5%) underwent a physical examination, but did not return their baseline questionnaire. Ten patients (0.9%) underwent no physical examination due to external circumstances like holidays and intervening commitments, but did return their baseline questionnaire. Data from the computerized medical files of 13 patients were not available.
Additional assessments
Since starting inclusion for the MRI study in April 2002 there were 184 eligible patients, of which 134 patients participated. Reasons for non-participation were (in order of their contributions) unwillingness or inability to find time for these extra measurements, distance to the research facility, and the fact that detected lesions would only be reported to the patient and their GP if urgent intervention was deemed necessary.
Since starting inclusion for the knee function assessments study in August 2002 there were 330 eligible patients, of which 87 patients participated. Reasons for non-participation were unwillingness or inability to attend the extra visits required for these measurements.
Patient selection
The search in the computerized patient records for occurrences of defined ICPC codes during the 4-month sample periods identified 310 eligible patients. 153 (49%) of those forwarded their contact details to the researchers, and 130 (42%) were included in the study and signed informed consent. The actual number of patients from which we received contact details during those same sample periods was 176, of which 150 patients were included in the cohort study (15% declined). When we looked up the medical files of the 150 participants we found that 20 of them lacked ICPC codes, explaining the 130 participants that were identified during the search. Likewise, a lack of ICPC coding in the medical records explains the discrepancy between the 176 contacted patients and the 153 that were identified in the search. Over the entire inclusion period the medical files of 15% of all participants lacked ICPC-codes.
Comparing the 130 participants and 180 non-participants identified through ICPC codes in the sample periods, we find significant selection with respect to age groups (table 4): we recruited relatively more patients over 35 years of age (OR 1.70; 1.15–2.77). This selective recruitment was more pronounced in the male population (OR 2.16; 1.12–4.18), than the female population (OR 1.22; 0.58–2.55). Overall, participation rates of women were not significantly higher than that of men (OR 1.13; 0.72–1.78). Participation rates of traumatic patients were lower than those of non-traumatic patients, though not significantly (OR 0.60; 0.26–1.43). Figures 3 and 4 show graphical representations of the proportions of included patients for each age group, subdivided for gender and traumatic onset of complaints.
Table 4 Patient characteristics of participants and non-participants
all participants in cohort participants in sample non-participants in sample comparison# of participation rates
N n
(%) N n
(%) N n
(%) OR
(95% CI)
gender (nwomen) 1045 494
(47%) 130 64
(49%) 180 83
(46%) 1.20
(0.78–1.85)
age (n>35 years) 1045 741
(71%) 130 94
(72%) 180 109
(61%) 1.79
(1.12–2.86)*
in men 551 380
(69%) 66 46
(69%) 97 50
(52%) 2.08
(1.11–3.93)*
in women 494 361
(73%) 64 48
(78%) 83 59
(71%) 1.41
(0.69–2.90)
in traumatic 197 134
(68%) 16 11
(73%) 34 17
(50%) 2.60
(0.80–8.98)
in non-traumatic 848 607
(72%) 114 83
(73%) 146 92
(63%) 1.63
(0.98–2.72)
trauma$ (npositive) 1045 197
(19%) 130 16
(12%) 180 34
(19%) 0.59
(0.32–1.09)
in men 551 109
(20%) 66 9
(14%) 97 20
(21%) 0.60
(0.26–1.38)
in women 494 88
(18%) 64 7
(11%) 83 14
(17%) 0.58
(0.23–1.47)
ICPC codes
L15: unspecified 519
(51%) 69
(53%) 82
(46%)
L78: acute distortion 107
(10%) 7
(5%) 20
(11%)
L90: osteoarthritis 77
(7%) 9
(7%) 28
(16%)
L94.2: Osgood-Schlatter 10
(1%) 1
(1%) 0 0%
L96: acute meniscus/ligament ruptures 87
(8%) 12
(9%) 14
(8%)
L97: chronic internal trauma 245
(23%) 32
(25%) 36
(20%)
$ patient described onset of complaints in questionnaire as immediate, due to impact or twisting, maximally 1 year before consultation
# comparing participation rates of age groups and traumatic injuries in sample
* p-value < 0.05
N total number of patients
n number of patients in a subset
Figure 3 Inclusion rate of eligible patients per age group and gender.
Figure 4 Inclusion rate of eligible patients per age group and traumatic onset.
When comparing participants from the sample periods with participants from the entire inclusion period, we found equal proportions of gender and age groups (see table 4). However, the proportions of traumatic injuries differed significantly: 12% of the patients in the sample periods were labelled 'traumatic injury' against 19% in the rest of the inclusion period (OR 0.59; 0.35 – 0.98).
We compared ICPC codes of participants and non-participants from the sample periods with a Chi-square test, pooling the codes L15 and L94.2 to prevent empty cells. We found a significant difference between the groups (Chi-statistic 11.2, p = 0.025). The differences are caused mainly by codes L78 and L96 for acute traumatic injuries and code L90 for osteoarthritis of the knee, all of which are less frequent in the participants. Comparison of ICPC codes of participants from the sample period with those of the rest of the inclusion period using the Chi-square test reveals no significant difference (Chi-statistic 5.6, p = 0.234).
Discussion
We succeeded in starting a unique cohort study of patients with incident knee complaints in general practice. From October 2001 to October 2003 we included 1068 patients. Apart from its size, this cohort is unique in the range of knee complaints we studied: we included all ages from adolescents to the elderly, and we included both traumatic and non-traumatic complaints. Furthermore, this is the first cohort to include a standardised physical examination as well as questionnaires in patients who seek medical care for their knee complaints in general practice. We therefore think our cohort has a high potential for giving insight into the natural course of a range of knee complaints, and will give valuable information to base future effectiveness studies in primary care on. But in order to extrapolate the results of future publications ensuing from this cohort to clinical practice, we need to determine whether selective recruitment could induce bias.
Selective recruitment
Patients below the age of 36 years were significantly less inclined to participate in our cohort study, and this trend was even stronger in the male population. Other comparisons did not produce statistically significant differences. However, the sample size may have been too small to prove that patients with traumatic injuries were underrepresented, again to a greater extent in the younger age group. Comparison of ICPC codes of the non-participants with those of the participants from the sample periods using a Chi-square test reveals a significant difference with respect to types of knee complaints. The difference is mainly caused by lower frequencies of the codes for the acute traumatic injuries L78 and L96, but lower frequencies of osteoarthritis of the knee (L90) also contribute.
The lack of ICPC codes in 15% of the participants indicates that our method for determining patient selection depends on the coding behaviour of the GPs. So is our sample a good representation of the situation during the entire inclusion period? We cannot identify the non-participants without ICPC codes to verify that, so we compared the proportions of gender, age groups and traumatic injuries of cohort and sample (table 4). We found a significantly smaller proportion of participants with traumatic injuries in the sample (12%) than in the cohort (19%) (OR 0.59; 0.35 – 0.98). As we made sure that the 4-month sample periods were distributed over all seasons in each municipality before randomly assigning them to the resident practices, we have ruled out seasonal fluctuations as a possible cause. But the working definition of 'traumatic injury' might explain something. In the medical records traumatic injuries can be recognised either by their ICPC code, or by the textual notes made by the GP. Some GPs tend to choose non-specific codes (L15) for any knee complaint, in which case recognition of traumatic injuries depends on the amount of detail in the textual notes. However, for further analyses in our cohort we use the patients perceived cause of the knee complaint together with the duration of the complaint to determine whether the complaint was of recent traumatic onset. With this definition we no longer detected any differences (29% in sample versus 31% in cohort). This indicates that the seemingly low participation rates of traumatic patients may have been an artefact caused by variations in the amount of detail in the medical files, rather than reflecting a non-representative sample. Furthermore, comparison of ICPC codes from the sample period with those of the rest of the inclusion period using the Chi-square test revealed no significant difference. One limitation remains: we have no insight into the possible differences in severity of knee complaints of participants and non-participants.
Comparing our results with those reported for the nationwide registration study [1], we found similar distributions of ICPC codes, suggesting that our population does not substantially deviate from patients with knee complaints in other Dutch general practices.
Conclusion
Based on these results, we expect that the effects of selective recruitment will not cause significant bias, as future analyses will be performed separately for subgroups of patients, and adjustments will be made for gender and other possible risk factors and confounders.
We are confident that the present cohort study will provide new insights into the prognosis and management of knee complaints in primary care, and that the results can be extrapolated to all Dutch general practices.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
EMH coordinated the cohort study, data collection and data management, performed the statistical analyses and drafted the manuscript. SMAB conceived of the study, participated in its design and coordination, and helped to draft the manuscript. MYB recruited participating GPs. SMAB, MYB and BWK have helped in interpreting the data and critically revised the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank the following parties for facilitating this research:
• All participating GPs from Brielle, Capelle aan den IJssel, Etten-Leur, Krimpen aan den IJssel and Rotterdam for recruiting patients for our research
• Physical therapists Dominique Crema (DC), Evelien Beckers (EB), Corinne Vrijland (CV), Ankie Verstijnen (AV) and Renée van Batenburg (RvB) for performing the physical examination of the patients and Dynaport measurements and Harry Wagemakers (HW) for supervising their training
• Medical student Meelan Bul for collecting data from GP medical files
• All personnel from the Radiology Department of Medical Centre Rotterdam Zuid who facilitated the MRI measurements
• Erasmus MC for internal funding of our research: "Revolving Fund" for the study of non-traumatic knee complaints in adolescents and younger adults and device assessed disability and "VAZ doelmatigheid" for the MRI study
• The health insurance companies Trias, Zilveren Kruis Achmea and OZ zorgverzekeringen for their financial support of the GPs participating in our research network HONEUR.
==== Refs
van der Linden MW Westert GP de Bakker DH Schellevis FG Tweede Nationale Studie naar ziekten en verrichtingen in de huisartspraktijk: klachten en aandoeningen in de bevolking en in de huisartspraktijk. [Second National Study into diseases and actions in general practice: complaints and disorders in the population and in general practice.] 2004 Utrecht, Bilthoven, NIVEL, Rijksinstituut voor Volksgezondheid en Milieu
Jinks C Jordan K Ong BN Croft P A brief screening tool for knee pain in primary care (KNEST). 2. Results from a survey in the general population aged 50 and over Rheumatology (Oxford) 2004 43 55 61 12923283 10.1093/rheumatology/keg438
Peat G Thomas E Handy J Wood L Dziedzic K Myers H Wilkie R Duncan R Hay E Hill J Croft P The Knee Clinical Assessment Study - CAS(K). A prospective study of knee pain and knee osteoarthritis in the general population BMC Musculoskelet Disord 2004 5 4 15028109 10.1186/1471-2474-5-4
Hofmans-Okkes IM Lamberts H The International Classification of Primary Care (ICPC): new applications in research and computer-based patient records in family practice Fam Pract 1996 13 294 302 8671139
Shields BJ Cohen DM Harbeck-Weber C Powers JD Smith GA Pediatric pain measurement using a visual analogue scale: a comparison of two teaching methods Clin Pediatr (Phila) 2003 42 227 234 12739921
van Laerhoven H van der Zaag-Loonen HJ Derkx BH A comparison of Likert scale and visual analogue scales as response options in children's questionnaires Acta Paediatr 2004 93 830 835 15244235 10.1080/08035250410026572
Herr KA Spratt K Mobily PR Richardson G Pain intensity assessment in older adults: use of experimental pain to compare psychometric properties and usability of selected pain scales with younger adults Clin J Pain 2004 20 207 219 15218405
Tegner Y Lysholm J Rating systems in the evaluation of knee ligament injuries Clin Orthop 1985 43 49 4028566
Roorda LD Jones CA Waltz M Lankhorst GJ Bouter LM van der Eijken JW Willems WJ Heyligers IC Voaklander DC Kelly KD Suarez-Almazor ME Satisfactory cross cultural equivalence of the Dutch WOMAC in patients with hip osteoarthritis waiting for arthroplasty Ann Rheum Dis 2004 63 36 42 14672889 10.1136/ard.2002.001784
Bellamy N Buchanan WW Goldsmith CH Campbell J Stitt LW Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee J Rheumatol 1988 15 1833 1840 3068365
Bengtsson J Mollborg J Werner S A study for testing the sensitivity and reliability of the Lysholm knee scoring scale Knee Surg Sports Traumatol Arthrosc 1996 4 27 31 8819060 10.1007/BF01565994
Kocher MS Steadman JR Briggs KK Sterett WI Hawkins RJ Reliability, validity, and responsiveness of the Lysholm knee scale for various chondral disorders of the knee J Bone Joint Surg Am 2004 86-A 1139 1145 15173285
Angst F Aeschlimann A Steiner W Stucki G Responsiveness of the WOMAC osteoarthritis index as compared with the SF-36 in patients with osteoarthritis of the legs undergoing a comprehensive rehabilitation intervention Ann Rheum Dis 2001 60 834 840 11502609
di Fabio RP Boissonnault W Physical therapy and health-related outcomes for patients with common orthopaedic diagnoses J Orthop Sports Phys Ther 1998 27 219 230 9513868
Vitale MG Levy DE Johnson MG Gelijns AC Moskowitz AJ Roye BP Verdisco L Roye DPJ Assessment of quality of life in adolescent patients with orthopaedic problems: are adult measures appropriate? J Pediatr Orthop 2001 21 622 628 11521031 10.1097/00004694-200109000-00014
Thomee R Augustsson J Karlsson J Patellofemoral pain syndrome: a review of current issues Sports Med 1999 28 245 262 10565551
van den Dikkenberg N Meijer OG van der Slikke RM van Lummel RC van Dieen JH Pijls B Benink RJ Wuisman PI Measuring functional abilities of patients with knee problems: rationale and construction of the DynaPort knee test Knee Surg Sports Traumatol Arthrosc 2002 10 204 212 12211185
Reider B The Orthopaedic Physical Examination 1999 WB Saunders 202 245
Hoppenfield S R. H Physical examination of the spine and extremities 1976 East Norwalk Conn, Appleton-Century-Crofts 171 196
Arnoldi CC Patellar pain Acta Orthop Scand Suppl 1991 244 1 29 1882690
Hughston JC Subluxation of the patella J Bone Joint Surg Am 1968 50 1003 1026 5676827
Bhagia SM Weinik M Xing Y Gupta K Meniscal injury
Stratford PW Binkley J A review of the McMurray test: definition, interpretation, and clinical usefulness J Orthop Sports Phys Ther 1995 22 116 120 8535469
Fowler PJ Lubliner JA The predictive value of five clinical signs in the evaluation of meniscal pathology Arthroscopy 1989 5 184 186 2775390
Daniel DM Stone ML Feagin JA Diagnosis of knee ligament injury: Tests and measurements of joint laxity The Crucial Ligaments 1988 1st New York, Churchill Livingstone 287 300
Torg JS Conrad W Kalen V Clinical diagnosis of anterior cruciate ligament instability in the athlete Am J Sports Med 1976 4 84 93 961972
Galway HR MacIntosh DL The lateral pivot shift: a symptom and sign of anterior cruciate ligament insufficiency Clin Orthop 1980 45 50 7371314
Smith BW Green GA Acute knee injuries: Part I. History and physical examination Am Fam Physician 1995 51 615 621 7863957
Insall JN Dorr LD Scott RD Scott WN Rationale of the Knee Society clinical rating system Clin Orthop 1989 13 14 2805470
Liow RY Walker K Wajid MA Bedi G Lennox CM The reliability of the American Knee Society Score Acta Orthop Scand 2000 71 603 608 11145388 10.1080/000164700317362244
Ware JEJ Sherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Ware JE Gandek BL Keller SD Group IQOLAP Spilker B Evaluating instruments used crossnationally: Methods from the IQOLA Project. Quality of life and pharmacoeconomics in clinical trials 1996 2nd Philadelphia, Lippincott-Raven Publishers 681
Aaronson NK Muller M Cohen PD Essink-Bot ML Fekkes M Sanderman R Sprangers MA te Velde A Verrips E Translation, validation, and norming of the Dutch language version of the SF-36 Health Survey in community and chronic disease populations J Clin Epidemiol 1998 51 1055 1068 9817123 10.1016/S0895-4356(98)00097-3
van Weel C Functional status in primary care: COOP/WONCA charts Disabil Rehabil 1993 15 96 101 8513163
Nelson E Wasson J Kirk J Keller A Clark D Dietrich A Stewart A Zubkoff M Assessment of function in routine clinical practice: description of the COOP Chart method and preliminary findings J Chronic Dis 1987 40 Suppl 1 55S 69S 3597698
Swinkels-Meewisse EJ Swinkels RA Verbeek AL Vlaeyen JW Oostendorp RA Psychometric properties of the Tampa Scale for kinesiophobia and the fear-avoidance beliefs questionnaire in acute low back pain Man Ther 2003 8 29 36 12586559 10.1054/math.2002.0484
|
16117830
|
PMC1208897
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Aug 23; 6:45
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-45
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-471614304610.1186/1471-2474-6-47Research ArticleEvaluation of easily measured risk factors in the prediction of osteoporotic fractures Bensen Robert [email protected] Jonathan D [email protected] Alexandra [email protected] George [email protected] Wojciech P [email protected] Rolf J [email protected] Timothy M [email protected] Robert G [email protected] Jacques P [email protected] David A [email protected] Annie [email protected] Mark [email protected] Charlie H [email protected] W [email protected] Medical Science, McMaster University, Hamilton, Ontario, Canada2 Medicine, McMaster University, Hamilton, Ontario, Canada3 Medicine, University of Saskatchewan, Saskatoon, Saskatchewan, Canada4 Medicine, University of Toronto, Toronto, Ontario, Canada5 Medicine, Laval University, Ste-Foy, Quebec, Canada6 Medicine, University of Calgary, Calgary, Alberta, Canada2005 5 9 2005 6 47 47 24 12 2004 5 9 2005 Copyright © 2005 Bensen et al; licensee BioMed Central Ltd.2005Bensen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Fracture represents the single most important clinical event in patients with osteoporosis, yet remains under-predicted. As few premonitory symptoms for fracture exist, it is of critical importance that physicians effectively and efficiently identify individuals at increased fracture risk.
Methods
Of 3426 postmenopausal women in CANDOO, 40, 158, 99, and 64 women developed a new hip, vertebral, wrist or rib fracture, respectively. Seven easily measured risk factors predictive of fracture in research trials were examined in clinical practice including: age (<65, 65–69, 70–74, 75–79, 80+ years), rising from a chair with arms (yes, no), weight (< 57, ≥ 57kg), maternal history of hip facture (yes, no), prior fracture after age 50 (yes, no), hip T-score (>-1, -1 to >-2.5, ≤-2.5), and current smoking status (yes, no). Multivariable logistic regression analysis was conducted.
Results
The inability to rise from a chair without the use of arms (3.58; 95% CI: 1.17, 10.93) was the most significant risk factor for new hip fracture. Notable risk factors for predicting new vertebral fractures were: low body weight (1.57; 95% CI: 1.04, 2.37), current smoking (1.95; 95% CI: 1.20, 3.18) and age between 75–79 years (1.96; 95% CI: 1.10, 3.51). New wrist fractures were significantly identified by low body weight (1.71, 95% CI: 1.01, 2.90) and prior fracture after 50 years (1.96; 95% CI: 1.19, 3.22). Predictors of new rib fractures include a maternal history of a hip facture (2.89; 95% CI: 1.04, 8.08) and a prior fracture after 50 years (2.16; 95% CI: 1.20, 3.87).
Conclusion
This study has shown that there exists a variety of predictors of future fracture, besides BMD, that can be easily assessed by a physician. The significance of each variable depends on the site of incident fracture. Of greatest interest is that an inability to rise from a chair is perhaps the most readily identifiable significant risk factor for hip fracture and can be easily incorporated into routine clinical practice.
==== Body
Background
Osteoporosis affects one in four women and one in eight men in Canada [1]. Its consequences are devastating both to the individual and society in terms of suffering, disability, and increased health care expenses. It is expected that over the next 40 years the number of hip fractures alone will increase exponentially, leading to a problem of epidemic proportions [2]. Fragility fractures are the most important and disabling consequence of osteoporosis and result in a loss of functional ability and increased morbidity and mortality. Estimated conservatively, a 50 year old woman has a 40 percent lifetime risk of a hip, vertebral or wrist fracture [3]. In 1993, the Canadian health care system spent more than $1.3 billion on osteoporosis-related fractures [4]. Recent research has shown that the mean one-year cost of a hip fracture, including direct and indirect costs, is (CAN) $26 527 [5].
Predicting incident fractures is critical today, but in light of the aging demographics, will have even greater importance in the future. Early recognition and treatment of osteoporotic patients is crucial to the prevention of these fractures. A number of risk factors have come to define those at increased risk for osteoporotic fracture and much work has been done in an attempt to delineate the critical variables [6-11]. In order for risk assessment to be effective and efficient, it must be practical and have high predictive value for the identification of fractures. Black et al., with use of data from the Study of Osteoporotic Fractures (SOF), developed the Fracture Index [12]. This assessment tool for predicting fracture risk in osteoporotic post-menopausal women was shown to be predictive of incident osteoporotic fractures (hip, vertebral, wrist and rib) in post-menopausal women and was validated using the EPIDOS study [12]. The Fracture Index assessment tool considers seven variables: age, fracture after age 50, history of maternal hip fracture after age 50, weight less than 125 lb (57 kg), smoking status, use of arms to stand up from a chair and bone mineral density (BMD) T-score [12]. These variables were chosen not only based on their predictive value but also because of their ease of assessment in a clinical setting. As a result, we sought to examine the usefulness of these risk factors in a clinical specialist setting.
The purpose of this study was to examine whether the risk factors outlined in the Fracture Index could be used to predict new incident osteoporotic fractures of the hip, vertebra, wrist, and rib in postmenopausal women who are registered in The Canadian Database for Osteoporosis and Osteopenia (CANDOO).
Methods
Study design
An analysis using prospective collected patient information from the CANDOO database was conducted. The CANDOO registry is a multi-site (Calgary, Winnipeg, Saskatoon, Toronto, Hamilton, Montreal, Quebec City) database consisting of more than 10,000 men and women who have been referred to specialists for osteoporosis [13]. The CANDOO database is designed to gather osteoporosis-related clinical information in a prospective manner displaying a record at each clinical visit. Patient information is entered and recorded into a central database [13]. It includes information on patient demographics; vertebral, rib, wrist and hip fracture history; gynecological history; use of osteoporosis related drugs; drug side effects; use of corticosteroids and other medications; dietary calcium intake; smoking habits; physical activity; fall history; prior medical history; family history including fractures; a self administered osteoporosis health related quality of life instrument; basic laboratory results and bone density measurements [13]. Patients were followed yearly with detailed assessments at clinical centres. Each yearly visit involved the collection of information similar to that collected at baseline.
All centres with a baseline CANDOO assessment (visit 1) between 1990 and 1999 with follow-up data available were selected. Only postmenopausal women ≥ 55 years of age with follow-up data available entered in CANDOO were considered for inclusion. Individuals with no follow-up assessments or males were excluded. Any individuals experiencing multiple fractures were also excluded in order to group subjects into hip, vertebral, wrist or rib fracture groups respectively. This exclusion criteria limited the number of applicable subjects from CANDOO and simultaneously implicated only females in the analysis. Fractures in CANDOO were determined based on either self-reports or X-ray confirmation.
For the current study we examined the seven easily measured clinical risk factors for fracture outlined in the Fracture Index. Each risk factor was expressed as an independent variable. The major dichotomous risk factors abstracted were: vertebral, rib, wrist or hip fractures after age 50 (yes/no); maternal family history of fracture (yes/no); weight < 57 kg (≤ 57 kg, > 57 kg); smoking status (yes/no); use of arms to stand from seated position (yes/no). Non-dichotomous risk factors abstracted included: age (<65, 65–69, 70–74, 75–79, 80–84, 85+ years); bone mineral density total hip T-score (> -1, -1 to -2.5, ≤ -2.5) (measurements were made using dual energy X-ray absorptiometry using Lunar or Hologic densitometers) [13]. Adjustments for osteoporotic drug treatment and inclusion of the osteoporosis quality of life questionnaire (OQLQ) as a variable were considered.
Statistical analysis
Using the seven independent variables we conducted multivariate logistic regression analyses for each fracture type to identify which of the seven risk factors were related to fracture in patients from the CANDOO database.
While longitudinal databases are a powerful research tool, a common weakness is that of missing or incomplete data. To minimize the bias associated with missing data, multiple imputation was utilized to replace missing data prior to the analyses [14,15]. In this case, 10 complete data sets were generated. The multiple imputed datasets were then analyzed using standard procedures for complete data and the results were pooled. Odds ratios and the associated 95% confidence intervals were calculated.
Results
Baseline characteristics
A total of 3426 patients were evaluated. Of those patients examined, 99 developed a wrist fracture, 64 a rib fracture, 158 a vertebral fracture, 40 a new hip fracture and 3065 experienced no fracture. Individuals in CANDOO who developed multiple fractures, 53 in total, were excluded from the analyses. The mean age and weight of the patients was 68 years and 64 kg, respectively. Mean total hip T-score was -1.9.
Baseline characteristics for the risk factors examined are displayed in Table 1. More than one third (36.8 %) of the women included were <65 years of age. While many subjects did not have a BMD assessment, 54.8 % of those that did had a T-score between -1 and -2.5. Only 5.6 % of postmenopausal women in CANDOO reported having a maternal history of hip fracture while 41.2 % had a prior fracture after the age of 50 years and 10.5 % were current smokers. Of the entire group, approximately half were only able to rise from a chair with the use of arms (as compared with no arms) and 30.2 % of patients weighed less the 57 kg.
Table 1 Baseline characteristics of risk factors for patient cohort.
RISK FACTOR n (% of cohort)
Age < 64.5 years 1259 (36.8)
Age 64.5–69.5 740 (21.6)
Age 69.5–74.5 751 (21.9)
Age 74.5–79.5 426 (12.4)
Age 79.5–84.5 186 (5.4)
Age ≥ 84.5 64 (1.9)
Total Hip T-score, T > -1 89 (19.9)
Total Hip T-score, -2.5<T<-1 245 (54.8)
Total Hip T-score, T ≤ -2.5 113 (25.3)
Maternal history of hip fracture 192 (5.6)
Prior fracture >50 years 1413 (41.2)
Current smokers 361 (10.5)
Rise from a chair with arms 1242 (49.2)
Weight < 57 kg 867 (30.2)
Hip fractures
The results of multivariable logistic regression analyses for each of the seven risk factors examined to predict a new fracture event in the hip are presented in Table 2. Prior fracture after the age of 50 years, those currently smoking, those between 75–79 years of age and those with a T-score of < -2.5 indicated a trend towards higher risk of developing a new hip fracture. The one factor that was statistically significant for predicting subsequent hip fractures were those individuals only able to rise from a chair with the aid of their arms. These individuals had an approximate odds ratio of 3.6. The 95% confidence interval of each risk factor predicting higher risk of developing a new fracture at the hip site is shown in Figure 1. All the factors had fairly wide confidence intervals, in part due to the overall low incidence of hip fractures. Maternal history of hip fracture was excluded from this model as the hip fracture group contained no data on maternal history of hip fracture.
Table 2 Hip Fracture Results Odds ratios and 95% confidence intervals point estimates for each risk factor in the development of new hip fracture
RISK FACTOR Odds Ratio 95% Confidence Interval
Weight (< 57 Kg) *Reference Level: ≥ 57 Kg 1.70 0.719; 4.037
Use of Arms to Rise *Reference Level: rise without arms 3.58 1.173; 10.927
Smoking Status *Reference Level: no smoking 1.48 0.485; 4.536
Age (65 to 69 y) *Reference Level: < 65 1.53 0.505; 4.667
Age (70–74 y) *Reference Level: < 65 0.98 0.280; 3.421
Age (75–79 y) *Reference Level: < 65 1.62 0.451; 5.840
Age (80–84 y) *Reference Level: < 65 1.52 0.296; 7.780
Age (85+y) *Reference Level: < 65 4.36 0.823; 23.078
Previous Fracture After 50 y
*Reference Level: no fracture after 50 1.08 0.477; 2.463
Osteoporotic (<-2.5 SD)**
*Reference Level: normal BMD 1.69 0.233; 12.253
Figure 1 Odds ratios and confidence intervals for each risk factor that predicts higher risk of hip fracture.
Vertebral fractures
The results of multivariate logistic regression analyses for each of the seven risk factors thought to predict a new vertebral fracture within the CANDOO database are depicted in Figure 2 and Table 3. Of all the risk factors, three were found to be significant predictors of new spinal fractures. Current smokers and those between the ages of 75–79 years were approximately two times more likely to develop a new vertebral fracture. Individuals weighing less than 57 kg were about 1.6 times more likely to develop a new fracture at this site. The other four risk factors did not reach statistical significance in terms of their predictability of a new fracture at vertebral sites in CANDOO.
Figure 2 Odds ratios and confidence intervals for each risk factor that predicts higher risk of vertebral fracture.
Table 3 Vertebral Fracture Results Odds ratios and 95% confidence intervals point estimates for each risk factor in the development of a new vertebral fracture
RISK FACTOR Odds Ratio 95% Confidence Interval
Weight (< 57 Kg) *Reference Level: ≥ 57 Kg 1.57 1.035; 2.373
Use of Arms to Rise *Reference Level: rise without arms 1.72 0.981; 3.023
Smoking Status *Reference Level: no smoking 1.95 1.199; 3.184
Age (65 to 69 y) *Reference Level: < 65 1.36 0.805; 2.309
Age (70–74 y) *Reference Level: < 65 1.28 0.753; 2.174
Age (75–79 y) *Reference Level: < 65 1.96 1.096; 3.508
Age (80–84 y) *Reference Level: < 65 0.93 0.368; 2.334
Age (85+y) *Reference Level: < 65 1.32 0.375; 4.631
Previous Fracture After 50 y
*Reference Level: no fracture after 50 1.37 0.931; 2.012
Maternal History of Fracture
*Reference Level: no family history 0.85 0.296; 2.412
Osteoporotic (<-2.5 SD)**
*Reference Level: normal BMD 1.85 0.448; 7.676
Wrist fractures
Table 4 shows odds ratios and 95% confidence intervals for each risk factor for fractures at the wrist. Each of the seven risk factors' independent ability to predict a new wrist fracture is illustrated in Figure 3. Prior fracture after the age of 50 and a body weight less than 57 kg were deemed significant risk factors for a new incident wrist fracture. Individuals with a prior fracture after the age of 50 years and those weighing less than 57 kg were about two and 1.7 times more likely to develop a new wrist fracture, respectively. The other five risk factors were not statistically significant.
Figure 3 Odds ratios and confidence intervals for each risk factor that predicts higher risk of wrist fracture.
Table 4 Wrist Fracture Results Odds ratios and 95% confidence intervals point estimates for each risk factor in the development of a new fracture at the wrist
RISK FACTOR Odds Ratio 95% Confidence Interval
Weight (< 57 Kg) *Reference Level: ≥ 57 Kg 1.71 1.007; 2.897
Use of Arms to Rise *Reference Level: rise without arms 1.53 0.790; 2.981
Smoking Status *Reference Level: no smoking 0.62 0.259; 1.466
Age (65 to 69 y) *Reference Level: < 65 0.76 0.387; 1.485
Age (70–74 y) *Reference Level: < 65 1.00 0.539; 1.843
Age (75–79 y) *Reference Level: < 65 0.57 0.238; 1.387
Age (80–84 y) *Reference Level: < 65 0.46 0.134; 1.603
Age (85+y) *Reference Level: < 65 0.83 0.183; 3.776
Previous Fracture After 50 y
*Reference Level: no fracture after 50 1.96 1.191; 3.223
Maternal History of Fracture
*Reference Level: no family history 1.25 0.430; 3.627
Osteoporotic (<-2.5 SD)**
*Reference Level: normal BMD 2.26 0.476; 10.682
Rib fractures
Individuals with a maternal history of a hip fracture and a prior fracture after the age of 50 years were approximately three and two times more likely, respectively, to develop a new rib fracture during the course of the study. Those with any of the other five risk factors were not at significantly increased risk of sustaining a new fracture at the rib. These data are displayed in Table 5 and Figure 4.
Table 5 Rib Fracture Results Odds ratios and 95% confidence intervals point estimates for each risk factor in the development of a new fracture at the rib
RISK FACTOR Odds Ratio 95% Confidence Interval
Weight (< 57 Kg) *Reference Level: ≥ 57 Kg 0.47 0.222; 1.011
Use of Arms to Rise *Reference Level: rise without arms 1.93 0.868; 4.272
Smoking Status *Reference Level: no smoking 1.94 0.903; 4.179
Age (65 to 69 y) *Reference Level: < 65 1.42 0.631; 3.210
Age (70–74 y) *Reference Level: < 65 1.58 0.709; 3.516
Age (75–79 y) *Reference Level: < 65 2.26 0.928; 5.486
Age (80–84 y) *Reference Level: < 65 2.69 0.950; 7.625
Age (85+y) *Reference Level: < 65 1.04 0.126; 8.478
Previous Fracture After 50 y
*Reference Level: no fracture after 50 2.16 1.201; 3.874
Maternal History of Fracture
*Reference Level: no family history 2.89 1.035; 8.081
Osteoporotic (<-2.5 SD)**
*Reference Level: normal BMD 3.89 0.367; 41.263
Figure 4 Odds ratios and confidence intervals for each risk factor that predicts higher risk rib fracture.
Discussion
Even with the advent of the recent publication of the 2002 Osteoporosis Society of Canada's Clinical Practice Guidelines, there remains no universal way of identifying those with osteoporosis and those at increased risk of fracture. The clinical goal must be to identify the substantial percentage of the population at high risk for fracture while simultaneously limiting any unnecessary testing to the increasingly burdened health care system [16,17]. Prevention of an incident fracture and the cascade of subsequent fractures is the ultimate objective.
We found a variety of clinical risk factors in conjunction with BMD to be extremely helpful in distinguishing those at high risk of fracture in CANDOO.
The association of increasing age with declining bone density has long been recognized as being the predominant risk factor for fracture. Age was shown to be an independent predictor of hip fracture in the EPIDOS study [6]. For ages 45 through 85, the ten year probability of a fracture in the forearm, humerus, spine or hip increases five times in men and eight times in women [18]. It has been estimated by Kanis et al. that risk of fracture in the forearm alone can increase eight times between the ages of 45 and 85 [18]. The results from the CANDOO database showed that although age, especially between 75 to 79 years, was associated with heightened risk of fracture at all sites, it only reached significance for predicting future fractures at vertebral sites. These results indicate that it is possible that age alone may not be definitive in predicting overall fracture risk.
In a clinical setting, BMD remains the gold-standard in assessing fracture risk so long as it is considered within the context of age. Cummings et al. showed that for each standard deviation decline in femoral neck BMD is associated with 2.6 times the risk of hip fracture in postmenopausal women aged 65 years or more [19]. Although BMD can identify people who are at increased risk of experiencing a fracture, it cannot, with any certainty, identify those individuals who will necessarily sustain a future fracture. Moreover, BMD testing is both inconvenient and expensive. Many under-serviced areas do not have the technology available for assessing BMD and even in those areas where machines are available, such procedures may be difficult to access due to mobility issues for osteoporotic patients and elderly subjects in long-term care facilities. Results from the CANDOO database did not indicate the overwhelming importance of BMD values in predicting those at increased fracture risk at any of the four sites considered. While a clinically relevant association was seen between BMD and fracture, its association was not statistically significant. This could at least in part arise from the small number of eligible subjects in CANDOO with BMD assessments.
Previous fracture history is well known to be predictive of future fracture risk [10,20-23]. The number of prior fractures at the site of incident fracture (i.e. hip, spine) combined with age has been shown to increase fracture risk 1.5 to 9.5 times [10,20,21,23-27]. Prior fracture at vertebral sites increases the risk of future fracture by as much as four times [28,29]. A previous history of fracture represented an important factor in evaluating the risk of future non-vertebral (wrist and rib) fractures in the CANDOO patients. At hip related sites within the CANDOO population, previous fracture history was not significant in predicting future hip fracture risk. This may be explained by the low incidence of hip fractures and the fact that many vertebral fractures remain undiagnosed. In fact, it has been estimated that less than one third of all vertebral fractures are clinically diagnosed [30].
Much evidence has indicated that those with a maternal history of fracture, especially at the hip, are at increased risk of future fracture. Moreover, those whose maternal history involves a grandmother, carry an even greater risk of hip fracture [11]. Within the CANDOO patients, the predictive ability of this variable appeared to be less important. At all sites, with the exception of the rib, maternal family history of fracture showed no significant association with increased fracture risk. This could possibly be explained by poor recollection or documentation of maternal history within the baseline assessment.
Low body weight was shown to be a significant predictor of future fracture at both vertebral and wrist sites. It played a significant role in assessing fracture risk at these sites, but appeared to be less important in determining risk of future fracture in the hip or rib.
The direct effects of smoking resulting in declining BMD and increased fracture risk have been identified by a variety of different studies [31-34]. Individuals who reported smoking in CANDOO were only at significantly increased risk of fracture at vertebral sites. These results indicate that at least in CANDOO, smoking as a risk factor is limited at other sites.
Many risk factors incorporated into the Fracture Index and other publications focus on biological, historical or BMD-affecting variables in risk factor analysis. Although the importance of these particular risk factors is unquestionable, a considerable number of osteoporotic fractures result from falls. Easily assessed neuromuscular measures of fall-related hip fracture have been examined in a few studies [10,35,36]. With the use of data from EPIDOS, Dargent-Molina et al. found that four predictors of fall-related fracture (slow gait speed, difficulty in tandem walking, reduced visual acuity, small calf circumference) were significantly associated with increased risk of future hip fracture [35]. A similar study confirmed that a simple and efficient measure of gait speed had the same discriminant value for fracture prediction as femoral BMD at all cutoff values [36]. It was the prospective study by Cummings et al. [10] that indicated the importance of examining one's ability to rise from a chair without the use of arms in risk assessment. They determined that this factor alone was the most significant at predicting hip fracture and that the addition of other neuromuscular assessment tests added little to the prediction of subsequent hip fracture [10].
The use of arms to rise from a seated position was found to be the single most important predictor of increased fracture risk at the hip in CANDOO patients. These results indicate that both research and clinical settings should place a greater emphasis than is currently standard procedure on this variable or others that examine neuromuscular ability. Such an easily assessed risk factor would make for an efficient and effective mechanism to gauge risk across the entire population, especially in those for which BMD testing is not feasible.
Many of these risk factors have been successfully incorporated into clinical risk assessment tools, while others such as the use of arms to rise from a chair tend to be ignored. A variety of current risk assessment mechanisms are used routinely in clinical practice, including the Osteoporosis Risk Assessment Instrument (ORAI), Simple Calculated Osteoporosis Risk Estimation (SCORE) and the module Physicians' Information and Education Resource (PIER) [37-39].
The ORAI is a clinical assessment tool that was designed to identify women over the age of 45 who are at increased risk for osteoporosis who should undergo BMD testing. This simple tool uses three items (age, weight and estrogen use) to gauge risk and has been validated to identify over 90% of women at increased risk of osteoporotic fracture while ensuring that less than 50 % of those with normal BMD are selected [37]. Although this procedure has proven to be useful in clinically identifying patients at risk, it only allows for examination of females and relies completely on historical risk factors.
A similar mechanism for identifying individuals at increased risk in clinical practice was set forth in SCORE. This tool relies on six risk factors rather than three in order to assess risk of osteoporotic fracture. These six variables include: race, age, rheumatoid arthritis, history of non-traumatic fracture after age 45, weight and estrogen use [37]. Despite its incorporation of more risk factors, SCORE has been shown to have similar sensitivity to ORAI but greater selection of individuals with normal BMD [40].
The effectiveness of determining those individuals at increased risk of future fracture not only revolves around public awareness of risk factors, but relies heavily on the ability of physicians (especially in primary care settings) to effectively and efficiently identify at-risk individuals. Module PIER [39] is a web and computer based resource designed to guide physicians through the diagnosis, treatment and management of a plethora of diseases including osteoporosis. Their recommendations include the use of the Fracture Index for risk assessment and analyses of other variables including use of corticosteroid therapy for more than 3 months, impaired vision, low calcium intake, low physical activity, dementia, alcohol consumption of greater than 2 drinks per day and estrogen deficiency before 45 years of age. Subsequently, any post-menopausal woman over 65 years of age or those under 65 who have at least one of these risk factors is recommended to have her BMD assessed. The inclusion of risk factors to assess frailty and mobility in PIER, such as the use of arms to rise from a chair and impaired vision, takes into consideration the importance of preventing falls and future fracture rather than solely basing risk assessment on historical risk factors.
Our study incorporated a large sample of post-menopausal women from the CANDOO database. The subject data from this database were homogeneous as they stemmed from a group of patients who were assessed in the tertiary care setting and represented a "real world" group [13]. Moreover, the multivariable analyses involved detailed and controlled delineation of potential confounding variables.
Although the study design attempted to minimize limiting factors, it should be recognized that much of the information in the CANDOO database relied on factual recall from patients, some of which was reviewed at considerable length after the experience had occurred. In addition, although careful delineation of confounding variables was carried out, it remains uniquely possible that other potential confounders played a role. Assessment of fractures for the CANDOO database were either confirmed or self-reported and as a result sub-clinical fractures in patients could have been missed. Also, the lack of data on male patients meant only females were assessed and, as such, results should not necessarily be extrapolated to the male population.
Further studies will need to be undertaken in attempt to identify a more definitive and indicative group of risk factors that can be used across genders and populations to assess future fracture risk and the subsequent development of osteoporosis.
Conclusion
Despite extensive research and new treatment strategies, osteoporosis remains one of the "sleeping giants of health care". Much of the population remains undiagnosed and unaware of the importance of early recognition and preventative treatment. As the "boomer bulge" continues to progress through middle age and into senior brackets, the effects of osteoporosis on both the individual and health care system will be enormous. This study has shown that there exists a variety of predictors of future fracture, besides BMD, that can be easily assessed by a physician. The predictive significance of each of these risk factors has been shown to be dependent on the site of incident fracture. The assessment of unconventional risk factors such as examining one's ability to rise from a chair without the use of arms to gauge proprioception, strength and coordination are simple, convenient and valuable in terms of predicting fracture risk. These risk assessment factors can be easily incorporated into routine clinical practice, especially where BMD testing is unfeasible. A better understanding of which factors lead to an increase in the incidence of fractures is a crucial step in evaluating patients at risk and designing therapeutic strategies.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RB- conceived design, participated in analysis and coordination and drafted manuscript
JDA- conceived design, participated in CANDOO database, participated in analysis and manuscript
AP- conceived design, participated in CANDOO database, coordinated analysis and participated in manuscript
GI- performed statistical analysis and participated in study design
WO- participated in CANDOO, study design, and analysis
RS- participated in CANDOO, study design, and analysis
TM- participated in CANDOO, study design, and analysis
RJ- participated in CANDOO, study design, and analysis
JB- participated in CANDOO, study design, and analysis
DH- participated in CANDOO, study design, and analysis
AP- participated in CANDOO, study design, and analysis
MP- participated in CANDOO, study design, and analysis
CG- participated in statistical analysis and study design
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
McMaster University is the institution to which this work should be attributed.
==== Refs
Brown JP Josse RG Scientific Advisory Council of the Osteoporosis Society of Canada 2002 clinical practice guidelines for the diagnosis and management of osteoporosis in Canada CMA 2003 1 34 14728730
Matthis C Weber U O'Neill TW Raspe H Health impact associated with vertebral deformities: results from the European Vertebral Osteoporosis Study (EVOS) Osteoporos Int 1998 8 364 372 10024907 10.1007/s001980050076
Melton LJIII Chrischilles EA Cooper C Lane AW Riggs BL Perspective. How many women have osteoporosis? J Bone Miner Res 1992 7 1005 1010 1414493
Goeree ROB Pettitt DB Cuddy L Ferraz M Adachi J An assessment of the burden of illness due to osteoporosis in Canada J Soc Obstet Gynaecol Can 1996 18 15 24
Wiktorowicz ME Goeree R Papaioannou A Adachi JD Papadimitropoulos E Economic implications of hip fracture: health service use, institutional care and cost in Canada Osteoporos Int 2001 12 271 278 11420776 10.1007/s001980170116
Schott AM Cormier C Hans D Favier F Hausherr E Dargent-Molina P Delmas PD Ribot C Sebert JL Breart G Meunier PJ How hip and whole-body bone mineral density predict hip fracture in elderly women: The EPIDOS prospective study Osteoporosis International 1998 8 247 254 9797909 10.1007/s001980050061
Marshall D Johnell O Wedel H Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures British Medical Journal 1996 312 1254 1259 8634613
Genant HK Li J Wu CY Shepherd JA Vertebral fractures in osteoporosis - A new method for clinical assessment Journal of Clinical Densitometry 2000 3 281 290 11090235 10.1385/JCD:3:3:281
Nelson HD Helfand M Woolf SH Allan JD Screening for postmenopausal osteoporosis: A review of the evidence for the US Preventive Services Task Force Annals of Internal Medicine 2002 137 529 541 12230356
Cummings SR Nevitt MC Browner WS Risk factors for hip fracture in white women The New England Journal of Medicine 1995 332 767 773 7862179 10.1056/NEJM199503233321202
Torgerson DJ Campbell MK Thomas RE Reid DM Prediction of perimenopausal fractures by bone mineral density and other risk factors J Bone Mine Res 1996 11 293 297
D.M B M S L P P DM R L M.S H O J An Assessment Tool for Predicting Fracture Risk in Postmenopausal Women Osteoporosis International 2001 12 519 528 11527048 10.1007/s001980170072
Olszynski WP Ioannidis G Sebaldt RJ Hanley DA Petrie A Brown JP Josse RG Murray TM Goldsmith CH Stephenson GF Papaioannou A Adachi JD The association between iliocostal distance and the number of vertebral and non-vertebral fractures in women and men registered in the Canada Database of Osteoporosis and Osteopenia (CANDOO) BMC Musculoskelet Disord 2002 3 22 12361480
Rubin D Inference and missing data Biometrika 1976 63 581 590
Schafer JL Analysis of Incomplete Multivariate Data 1997 London, UK, Chapman & Hall 2 6
McClung MR Therapy for fracture prevention JAMA 1999 282 687 689 10517723 10.1001/jama.282.7.687
Kanis JA Delmas P Burckhardt P Cooper C Torgerson D Guidelines for diagnosis and management of osteoporosis. The European Foundation for Osteoporosis and Bone Disease Osteoporos Int 1997 7 390 406 9373575 10.1007/BF01623782
Kanis JA O J Oden A Dawson A C DL Jonsson B Ten Year Probabilities of Osteoporotic Fractures According to BMD and Diagnostic Thresholds Osteoporosis International 2001 12 989 995 11846333 10.1007/s001980170006
Cummings SR Black DM Nevitt MC Browner W Cauley J Ensrud K Genant HK Palermo L Scott J Vogt TM Bone density at various sites for prediction of hip fractures. The Study of Osteoporotic Fractures Research Group Lancet 1993 341 72 75 8093403 10.1016/0140-6736(93)92555-8
Wasnich RD Davis JW Ross PD Spine fracture risk is predicted by non-spine fractures Osteoporos Int 1994 4 1 5 8148565 10.1007/BF01623426
Davis JW Grove JS Wasnich RD Ross PD Spatial relationships between prevalent and incident spine fractures Bone 1999 24 261 264 10071920 10.1016/S8756-3282(98)00176-8
A.A I W C C C J.D F Prevalent Vertebral Deformity Predicts Incident Hip though not distal Forearm Fracture: Results from the European Prospective Osteoporsis Study Osteoporosis International 2001 12 85 90 11303719 10.1007/s001980170138
Klotzbuecher CM Ross PD Landsman PB Abbott TAIII Berger M Patients with prior fractures have an increased risk of future fractures: a summary of the literature and statistical synthesis J Bone Miner Res 2000 15 721 739 10780864
Ross PD Davis JW Epstein RS R.D W Pre-existing fractures and bone mass predict vertebral fracture incidence in women Annals of Internal Medicine 1991 114 919 923 2024857
Tromp AM Smit JH Deeg DJ Bouter LM Lips P Predictors for falls and fractures in the Longitudinal Aging Study Amsterdam J Bone Miner Res 1998 13 1932 1939 9844112
Black DM Palermo L Nevitt MC Genant HK Christensen L Cummings SR Defining incident vertebral deformity: a prospective comparison of several approaches. The Study of Osteoporotic Fractures Research Group J Bone Miner Res 1999 14 90 101 9893070
Fox KM Cummings SR E W K S Femoral Neck and Intertrochanteric Fractures Have Different Risk Factors: A Prospective Study Osteoporosis International 2000 11 1018 1023 11256892 10.1007/s001980070022
Ettinger B Black DM Mitlak BH Knickerbocker RK Nickelsen T Genant HK Christiansen C Delmas PD Zanchetta JR Stakkestad J Gluer CC Krueger K Cohen FJ Eckert S Ensrud KE Avioli LV Lips P Cummings SR Reduction of vertebral fracture risk in postmenopausal women with osteoporosis treated with raloxifene: results from a 3-year randomized clinical trial. Multiple Outcomes of Raloxifene Evaluation (MORE) Investigators JAMA 1999 282 637 645 10517716 10.1001/jama.282.7.637
Black DM Arden NK Palermo L Pearson J Cummings SR Prevalent vertebral deformities predict hip fractures and new vertebral deformities but not wrist fractures. Study of Osteoporotic Fractures Research Group J Bone Miner Res 1999 14 821 828 10320531
Cooper C Atkinson EJ O'Fallon WM Melton LJIII Incidence of clinically diagnosed vertebral fractures: a population-based study in Rochester, Minnesota, 1985-1989 J Bone Miner Res 1992 7 221 227 1570766
Honkanen R Tuppurainen M Kroger H Alhava E Saarikoski S Relationships between risk factors and fractures differ by type of fracture: a population-based study of 12,192 perimenopausal women Osteoporos Int 1998 8 25 31 9692074 10.1007/s001980050044
Siris ES Miller PD Barrett-Connor E Faulkner KG Wehren LE Abbott TA Berger ML Santora AC Sherwood LM Identification and fracture outcomes of undiagnosed low bone mineral density in postmenopausal women: results from the National Osteoporosis Risk Assessment JAMA 2001 286 2815 2822 11735756 10.1001/jama.286.22.2815
Law MR Hackshaw AK A meta-analysis of cigarette smoking, bone mineral density and risk of hip fracture: recognition of a major effect BMJ 1997 315 841 846 9353503
Law MR Cheng R Hackshaw AK Allaway S Hale AK Cigarette smoking, sex hormones and bone density in women Eur J Epidemiol 1997 13 553 558 9258568 10.1023/A:1007389712487
Dargent-Molina P Favier F Grandjean H Baudoin C Schott AM Hausherr E Meunier PJ Breart G Fall-related factors and risk of hip fracture: the EPIDOS prospective study Lancet 1996 348 145 149 8684153 10.1016/S0140-6736(96)01440-7
Dargent-Molina P Schott AM Hans D Favier F Grandjean H Baudoin C Meunier PJ Breart G Separate and combined value of bone mass and gait speed measurements in screening for hip fracture risk: results from the EPIDOS study. Epidemiologie de l'Osteoporose Osteoporos Int 1999 9 188 192 10367048 10.1007/s001980050134
Cadarette SM Jaglal SB Kreiger N McIsaac WJ Darlington GA Tu JV Development and validation of the Osteoporosis Risk Assessment Instrument to facilitate selection of women for bone densitometry CMAJ 2000 162 1289 1294 10813010
Cadarette SM Jaglal SB Murray TM McIsaac WJ Joseph L Brown JP Evaluation of decision rules for referring women for bone densitometry by dual-energy X-ray absorptiometry Journal of the American Medical Association 2001 286 57 63 11434827 10.1001/jama.286.1.57
Papaioannou A Ferko N Adachi JD Watts N Physicians' Information and Education Resource (PIER) 2004
Cadarette SM Jaglal SB Murray TM Validation of the simple calculated osteoporosis risk estimation (SCORE) for patient selection for bone densitometry Osteoporos Int 1999 10 85 90 10501785 10.1007/s001980050199
|
16143046
|
PMC1208898
|
CC BY
|
2021-01-04 16:32:04
|
no
|
BMC Musculoskelet Disord. 2005 Sep 5; 6:47
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-47
|
oa_comm
|
==== Front
BMC NeurolBMC Neurology1471-2377BioMed Central London 1471-2377-5-151612021010.1186/1471-2377-5-15Research ArticleSelf-reported parkinsonian symptoms in the EPIC-Norfolk cohort Ishihara Lianna S [email protected] Kay-Tee [email protected] Robert [email protected] Sheila [email protected] Ailsa [email protected] Nicholas [email protected] Carol [email protected] Department of Public Health and Primary Care, University of Cambridge, Forvie Site, Robinson Way, Cambridge CB2 2SR, UK2 Clinical Gerontology Unit, Addenbrookes Hospital, University of Cambridge, Cambridge CB2 2QQ, UK3 MRC Dunn Nutrition Unit, Wellcome Trust/MRC Building, Hills Road, Cambridge CB2 2XY, UK2005 24 8 2005 5 15 15 25 2 2005 24 8 2005 Copyright © 2005 Ishihara et al; licensee BioMed Central Ltd.2005Ishihara et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Parkinsonian symptoms have been associated with increased morbidity and mortality. Several studies have reported on the prevalence of signs and symptoms. Symptoms questionnaires can identify potential PD cases for further neurological examination to save resources. They can also provide information about how much of the population reports specific signs and symptoms. The objective of the study was to determine the self-reported prevalence of parkinsonian symptoms from a questionnaire, and to examine their association with age and self-reported Parkinson's disease in a large cohort.
Methods
A cross-sectional study was conducted within a sub-cohort of the EPIC-Norfolk (European Prospective Investigation of Cancer) cohort study.
Results
The prevalence of six self-reported parkinsonian symptoms are reported for 11539 individuals who answered all symptoms questions (62% of sub-cohort): rest tremor (4%), difficulty starting to walk (4%), difficulty getting out of a chair (6%), slower walking (34%), smaller handwriting (micrographia- 9%), and less acute sense of smell (olfactory dysfunction- 9%). The presence of individual symptoms increased with age except for difficulty getting out of a chair.
Conclusion
The results support previous findings that the presence of self-reported parkinsonian symptoms is strongly associated with age and self-reported PD diagnosis. The data also provide information regarding the prevalence of symptoms in a large, younger population of adults than previously reported in the literature.
==== Body
Background
Parkinsonian symptoms have been associated with increased morbidity and mortality in Alzheimer's disease and in individuals without dementia [1,2]. In one prospective cohort study, progression in gait disturbance, measured with the UPDRS, was associated with mortality [3]. Individuals with Parkinson's Disease (PD) by definition will have a greater prevalence of symptoms and signs than those without PD, and the age-dependence of parkinsonism predicts that older age groups will have higher proportions of reported signs and symptoms. Several studies have reported the presence of parkinsonian signs and symptoms in elderly individuals without a diagnosis of PD [1,4-11].
The diagnosis of PD is based on clinical signs because there is currently no conclusive diagnostic test available [12]. Clinical characteristics manifest only when the substantia nigral cell loss has reached a threshold of 60 to 80%, and most parkinsonian symptoms and signs are not specific to PD [12-14].
There are four cardinal motor signs of Parkinsonism: resting tremor at 4–6 Hertz, bradykinesia, rigidity, and impaired postural and righting reflexes. Other PD symptoms include: shuffling gait, cogwheel rigidity, unilateral onset, persistent asymmetry in severity, micrographia, masked facies, and retropulsion [15]. Reduced olfactory function has recently been suggested as a feature indicating pre-clinical PD, although it is not specific to PD [16-19].
The purpose of this study was to determine the prevalence of self-reported parkinsonian symptoms within a sub-cohort of EPIC-Norfolk [20]. We report the prevalence of six symptoms by age and PD status.
Methods
The design and methods of the EPIC-Norfolk cohort study have been described previously [20]. We provide a brief summary of the details regarding the sub-cohort used for the current analyses. The cohort includes men and women who were aged between 45 and 74 years at the time they entered the study. Subjects were recruited through selected General Practitioner (GP) practices in Norfolk, United Kingdom. Invitations were sent to 77 630 individuals and the response rate was 39% (N = 30 447). The main purpose of cohort recruitment was to conduct a prospective study of healthy participants willing to be followed up over time, so the aim was not to recruit a representative population sample. Nevertheless, the EPIC population is similar in age, sex and race to the national sample in the Health Survey for England though with a slightly lower proportion of current smokers.
The population used for studying Parkinsonian symptoms in the EPIC cohort includes only those that completed a second health questionnaire eighteen months after baseline (N = 18 465), when there were specific questions about parkinsonian symptoms. Baseline characteristics are reported for the total cohort, individuals who did not respond to the Parkinson's questionnaire, and those in the current sub-cohort (Table 1).
Table 1 Comparison of baseline characteristics between the total EPIC-Norfolk cohort and those completing the 18-month follow-up health questionnaire (current sub-cohort).
Baseline Characteristics 2nd health questionnaire (Sub-cohort) Non-respondents to the 2nd health questionnaire Total Cohort
Total Number 18465 11980 30445
Age (Years), Mean ± SD 59.0 ± 9.2 60.0 ± 9.7 59.4 ± 9.4
Male (%) 8029 (43.5) 5672 (47.3) 13701 (45.0)
Female (%) 10436 (56.5) 6308 (52.7) 16744 (55.0)
Smoking Status N (%)
Current 1873 (10.2) 1858 (15.7) 3731 (12.4)
Former 7548 (41.2) 5158 (43.6) 12706 (42.1)
Never 8897 (48.6) 4824 (40.7) 13721 (45.5)
Education Level
Below O level 7067 (38.3) 5585 (46.8) 12652 (41.6)
O level 2441 (13.2) 1339 (11.2) 3780 (12.4)
A level 6468 (35.0) 3806 (31.9) 10274 (33.8)
Degree 2489 (13.5) 1216 (10.2) 3705 (12.2)
An algorithm was created to relate the questionnaire to specific parkinsonian symptoms (Table 2).
Table 2 Summary of questions related to parkinsonian symptoms from the follow-up questionnaire
Parkinsonian symptom Present if answered yes to these questions
Resting Tremor Tremor or shakiness in hands? When relaxed?
Difficulty walking Difficulty in starting to walk? Not due to arthritis?
Difficulty getting out of a chair Difficulty getting out of a chair? Not due to arthritis?
Walking slower Has walking become slower?
Micrographia Has handwriting changed? Smaller?
Olfactory dysfunction Has sense of smell changed? Less acute?
PD case ascertainment was based on self-report at baseline or 18-month follow-up questionnaire, ongoing reports of ICD-9 diagnosis of PD from computerised hospital discharge records, and ongoing searches of death certificates reports for mention of PD. Diagnostic criteria were not standardised and relied on individual clinician opinions.
Data analyses
Analyses were conducted using Intercooled STATA 7.0. The symptom prevalence is reported for the subset of individuals who answered all symptom questions (62%), and for the total sub-cohort with missing responses as a separate response category. Chi-square tests for trend were used to examine the differences in reported parkinsonian symptom prevalence across 10-year age groups.
Results
Of 30 447 study participants, the sub-cohort used for the present analyses included 18 465 subjects that were followed up at least through the health questionnaire eighteen months after baseline (61%). There were 8029 men (43%), mean age 59.6 ± 9.1 years and 10 436 women, mean age 58.5 ± 9.2 years. A comparison of the sub-cohort with the non-respondents indicates that the non-respondents were older, less educated, and had a higher proportion of males and smokers (Table 1). There were 39 (0.9%) men and 30 (0.3%) women with prevalent Parkinson's disease in the sub-cohort identified through medical records, death certificates and self-report. Hospital records found only one additional baseline case, and death certificates revealed two previously unreported cases.
Only 11 539 (62%) of the participants answered all questions regarding parkinsonian symptoms on the follow-up questionnaire, including 53 (30 men, 23 women) PD cases. Symptom prevalence is reported excluding individuals with incomplete responses. The prevalences are also reported for the total sub-cohort for individuals answering yes or no, or missing a response. An exploration of the pattern of missing data revealed that most individuals either answered all, none, all except one, or only one of the questions (Table 3).
Table 3 Patterns of missing responses according to the order of questions on the questionnaire (N = 18 465)
Order of 6 symptom questions N (% of total)
Answered all 11539 (62.5)
Answered none 1196 (6.5)
Answered first question only 2160 (11.7)
Missing one question only 2219 (12.0)
Miscellaneous missing 1351 (7.3)
The parkinsonian symptoms were each reported by less than 10% of individuals except for slower walking, which was present in 34% of the individuals who answered all symptoms questions (Table 4). When missing answers were included as a response category, the prevalence of slower walking went down to 28%. The prevalence of slower walking which ranged from 11% to 71% in those with complete responses (8% to 56% including missing answers) from youngest to oldest age groups.
Table 4 The % (n) of subjects reporting symptoms and diagnosed Parkinson's disease, by 10-year age group and for PD cases
Age Group, Years (N)
Excluding individuals with any missing symptom answers
Symptom 40–49 (818) 50–59 (4008) 60–69 (3849) 70–79 (2803) 80+ (61) All Ages (11 539) P- value for age trend* PD Cases (53)
Resting Tremor 3.9 (32) 3.6 (144) 4.3 (166) 5.4 (150) 6.6 (4) 4.3 (496) 0.0008 49.1 (26)
Difficulty Walking 4.7 (38) 5.0 (200) 3.8 (147) 3.9 (110) 6.6 (4) 4.3 (499) 0.0435 35.9 (19)
Difficulty Chair 6.6 (54) 6.4 (255) 5.7 (220) 7.2 (203) 8.2 (5) 6.4 (737) 0.3255 39.6 (21)
Walking Slower 11.3 (92) 19.6 (787) 35.9 (1383) 57.3 (1607) 70.5 (43) 33.9 (3912) <0.0001 84.9 (45)
Micrographia 5.1 (42) 5.5 (219) 8.7 (336) 16.2 (454) 27.9 (17) 9.3 (1068) <0.0001 73.6 (39)
Smell less acute 7.7 (63) 7.9 (315) 9.3 (357) 11.4 (319) 13.1 (8) 9.2 (1062) <0.0001 34.0 (18)
Any of symptoms 25.7 (210) 32.6 (1306) 46.1 (1775) 65.1 (1826) 77.1 (47) 44.8 (5164) <0.0001 96.2 (51)
Self-reported PD 0.12 (1) 0.35 (14) 0.36 (14) 0.82 (23) 1.64 (1) 0.46 (53) 0.0012
Including Missing answers as a separate category
Symptom 40–49 (1342) 50–59 (6341) 60–69 (6063) 70–79 (4608) 80+ (111) All Ages (18 465) P- value for age trend* PD Cases (69)
Resting Tremor
Yes 3.5 (47) 3.2 (203) 3.7 (224) 4.5 (206) 6.3 (7) 3.7 (687) 0.0026 42.0 (29)
No 84.8 (1138) 87.1 (5526) 88.8 (5385) 87.9 (4052) 83.8 (93) 87.7 (16194) 50.7 (35)
Missing 11.7 (157) 9.7 (612) 7.5 (454) 7.6 (350) 9.9 (11) 8.6 (1584) 7.3 (5)
Difficulty Walking
Yes 3.4 (46) 4.7 (295) 3.4 (208) 3.4 (157) 6.3 (7) 3.9 (713) 0.0042 33.3 (23)
No 68.6 (921) 71.5 (4537) 76.4 (4630) 75.2 (3464) 72.1 (80) 73.8 (13632) 59.4 (41)
Missing 28.0 (375) 23.8 (1509) 20.2 (1225) 21.4 (987) 21.6 (24) 22.3 (4120) 7.3 (5)
Difficulty Chair
Yes 5.3 (71) 5.6 (354) 5.2 (312) 6.1 (279) 6.3 (7) 5.5 (1023) 0.6787 37.7 (26)
No 66.5 (892) 69.5 (4405) 72.9 (4421) 71.2 (3281) 68.5 (76) 70.8 (13075) 58.0 (40)
Missing 28.2 (379) 24.9 (1582) 21.9 (1330) 22.7 (1048) 25.2 (28) 23.7 (4367) 4.3 (3)
Walking Slower
Yes 8.4 (113) 16.1 (1024) 29.5 (1787) 47.0 (2165) 55.9 (62) 27.9 (5151) <0.0001 82.6 (57)
No 61.0 (818) 56.7 (3594) 47.2 (2862) 30.7 (1417) 18.0 (20) 47.2 (8711) 14.5 (10)
Missing 30.6 (411) 27.2 (1723) 23.3 (1414) 22.3 (1026) 26.1 (29) 24.9 (4603) 2.9 (2)
Micrographia
Yes 4.6 (62) 4.6 (294) 7.4 (446) 13.1 (604) 21.6 (24) 7.7 (1430) <0.0001 71.0 (49)
No 65.2 (875) 69.5 (4406) 70.6 (4283) 65.4 (3012) 54.1 (60) 68.5 (12636) 24.6 (17)
Missing 30.2 (405) 25.9 (1641) 22.0 (1334) 21.5 (992) 24.3 (27) 23.8 (4399) 4.4 (3)
Smell less acute
Yes 5.4 (73) 6.1 (388) 7.4 (450) 8.8 (407) 10.8 (12) 7.2 (1330) <0.0001 30.4 (21)
No 61.5 (825) 66.2 (4199) 68.7 (4164) 68.9 (3175) 65. 8 (73) 67.4 (12436) 56.5 (39)
Missing 33.1 (444) 27.7 (1754) 23.9 (1449) 22.3 (1026) 23.4 (26) 25.4 (4699) 13.1 (9)
Any of symptoms
Yes 15.7 (210) 20.6 (1306) 29.3 (1775) 39.6 (1826) 42.3 (47) 28.0 (5164) <0.0001 73.9 (51)
No 45.3 (608) 42.6 (2702) 34.2 (2074) 21.2 (977) 12.6 (14) 34.5 (6375) 2.9 (2)
Missing 39.0 (524) 36.8 (2333) 36.5 (2214) 39.2 (1805) 45. 1 (50) 37.5 (6926) 23.2 (16)
Self-reported PD 0.07 (1) 0.25 (16) 0.31 (19) 0.69 (32) 0.90 (1) 0.37 (69) <0.0001
*Calculated using chi-square for trend for each individual symptom across the age groups.
The proportion of subjects reporting each symptom significantly increased with age, with the exception of difficulty getting out of a chair. Difficulty walking was relatively constant across the age groups, until the age of 80 when there was a sharp increase. The age trends were similar for males and females, with the exception of difficulty getting out of a chair, which was age-dependent for males but less so for females (Results not shown). The percentage of PD cases reporting each parkinsonian symptom ranged from 34% to 85% (30% to 83% including missing responses), and as expected all symptoms were reported in proportionally more PD than non-PD subjects (Table 4).
The proportion of individuals self-reporting PD increased with age. The prevalence of PD in the individuals answering all symptoms questions was 0.46%, and 0.37% in the total sub-cohort (Table 4). Anti-parkinsonian medication use was recorded according to the British National Formulary. Nineteen identified PD cases were using at least one of selegiline, pergolide, and levodopa. The use of bromocriptine and procyclidine, which are not specific treatments for PD, was reported by nine individuals without self-reported PD.
The sensitivity and specificity for each symptom question as well as cumulative numbers of symptoms questions are reported in Table 5. The question with the highest sensitivity (85%) asked about walking slower, but the specificity was very low (66%). If a positive response to any of the six symptoms questions was defined as screening positive, then the sensitivity was 96% and specificity was 56%, but if slower walking was excluded then sensitivity was 91% and specificity 76%.
Table 5 The sensitivity and specificity of the self-reported questionnaire for PD identified by self-report and routine records (N = 11 539)
TP FN Sensitivity TN FP Specificity
Rest Tremor 26 27 49.1 11016 470 95.9
Difficulty Walking 19 34 35.9 11006 480 95.8
Difficulty Chair 21 32 39.6 10770 716 93.8
Walking Slower 45 8 84.9 7619 3867 66.3
Micrographia 39 14 73.6 10457 1029 91.0
Olfactory dysfunction 18 35 34.0 10442 1044 90.9
Any of 6 symptoms 51 2 96.0 6373 5113 55.5
Any of symptoms except Walking slower 48 5 90.6 8684 2802 75.6
≥2 symptoms 46 7 86.8 9705 1781 84.5
≥3 symptoms 38 15 71.7 10930 556 95.2
≥4 symptoms 22 31 41.5 11363 123 98.9
≥5 symptoms 11 42 20.8 11455 31 99.7
TP = True Positive; FN = False Negative; TN = True Negative; FP = False Positive
The prevalence of parkinsonian symptoms in the total sub-cohort was similar for the age groups < 65 and 65+ years, with the exception of micrographia and walking slower, which were reported twice as often by the older age group (Table 6).
Table 6 Prevalence (%) of parkinsonian symptoms in elderly populations, with and without parkinsonism, from screening questionnaire studies and current* study.
Total Populations
Current Mutch [4] Pramstaller [5] Rocca [6] Teresi [7]
Number of Subjects 6771 4768 87 36 16 164
Age (Years) <65 >65 71–76 40+ 62–96 65+
Symptom
Arms or legs ever shake? 5.0 35.0 6.2
Resting Tremor 3.8 5.0 9.9
Shuffle feet 3.0 8.3 31.2 10.1
Feet suddenly freeze in doorways 2.8 6.2
Poor balance 10.0 27.8 37.5
Difficulty Walking 4.7 3.8
Walking Slower 22.1 50.7 3.8#
Trouble rising from chair 6.4 6.4 13.0 13.9 37.5 21.4
Micrographia 6.0 13.9 14.0 5.6 12.5 13.2
Olfactory dysfunction 7.7 11.3
Patients with Parkinsonism
Current Duarte [28] Meneghini [31] Mutch [4] Pramstaller [5] Rocca [6]
Number of Subjects 53 36 21 35 40 37
Age (Years) 46–81 30–89 40+ 57–89 40+ 58–94
Symptom
Arms or legs ever shake? 89.1 76.2 77.0 65.0 64.9
Resting Tremor 49.1
Shuffle feet 72.9 51.0 82.5 67.6
Feet suddenly freeze in doorways 51.3 52.5 56.8
Poor balance 83.7 69.0 75.0 81.1
Difficulty Walking 35.9
Walking Slower 84.9
Trouble rising from chair 39.6 59.0 66.0 67.5 75.7
Micrographia 73.6 64.8 74.0 90.0 73.0
Olfactory dysfunction 34.0
*Current study including only the individuals who answered all symptoms questions (N = 11 539)
# Interviewer observation
Discussion
The purpose of this paper was not to validate a possible screening questionnaire, but to report the responses of the EPIC cohort to questions about the prevalence of various parkinsonian symptoms, and examine how these relate to age and self-reported PD. Similar questionnaires have been validated and various cut-off points used to identify individuals for further examination in a two-step screening process [5,6,21-32]. However, most of these did not report the responses for each question, but instead reported the sensitivity and specificity for a specific chosen cut-off point. Sensitivity was often quite high, however the positive predictive value for Parkinson's disease after screening was often low. Sensitivity and specificity are reported for the questionnaire, with self-reported PD as the disease outcome. Neurological examinations were not conducted to validate diagnoses or to identify de novo cases.
Study population
The current study population was a sub-cohort that completed the first follow-up questionnaire of the EPIC-Norfolk study. The population was restricted by age at baseline (45–74 years) and was ethnically homogeneous. The consequence of the homogeneous ethnicity and limited study area is that the results of this study may only be generalizable to Caucasians of similar socio-economic status and environmental conditions. Ethnic and geographical differences must be considered as possible explanations for differences in prevalence between studies, once methodological differences are accounted for. We have combined men and women for the analyses because there were few PD cases, and age was considered a more pertinent variable.
Selection bias was introduced at the original recruitment of participants for EPIC and in the exclusion of participants who had not answered the follow-up health questionnaire. Individuals who were ill at baseline were excluded from the study, and those who became ill after the baseline exam would be more likely to have dropped out of the study by the questionnaire at eighteen months. The prevalence estimates from this study are expected to be underestimates of the true measures in the Norfolk population because the sub-cohort is healthier than the general population and the original study population. However, the expected number of 72 cases, as calculated by indirect standardization using a prevalence study carried out in general practices in London, is similar to the 69 prevalent cases identified in the EPIC sub-cohort [33].
A further attrition of participants occurred because 38% of the sub-cohort did not respond to one or more of the symptom questions. The number of missing responses was associated with age, education, and smoking. All analyses of symptom prevalence were conducted with the subset of 62% of individuals who answered all questions, and repeated assuming that missing answers indicated the absence of the symptom.
Subjects missing the date of the second health examination were excluded to avoid an inconsistent denominator for analyses. Among the excluded participants were less than 3% of the sub-cohort, but this included 4 PD cases.
Methods
The questions regarding the presence of parkinsonian symptoms were subject to participant interpretation and relied on complete answers in the follow-up health questionnaire. The patterns of missing data according to the order of the questions in the questionnaire showed that most individuals answered all, none, all except one, or only one of the questions. It is difficult to hypothesize why individuals did not answer certain questions. Those answering none of the questions or only the first symptoms question could be due to the format of the questionnaire. The first part of the question asks about the diagnosis of PD, followed by the symptoms questions. It might improve the response if the symptoms questions were asked first, so those without PD diagnoses would not assume that the following questions did not apply to them.
It was not possible to conduct neurological examinations for all cohort participants, as Parkinson's disease is a secondary outcome for the EPIC-Norfolk study. However, a review reports that self-rated health status has been shown in many studies to be an accurate predictor of mortality, even better than medical records [34]. Self-rated health and self-reported parkinsonian symptoms are indeed two different measures, however the review emphasizes the value of self-report, even with its limitations. A cross-sectional study reported correlation between mild parkinsonian symptoms, measured with a shortened version of the UPDRS, and both self-reported measured of function and performance-based test scores [35]. Some literature comparing self-report with neurological examination, although not specifically for PD, offers further support [36].
The PD ascertainment methods were incomplete and relied heavily on the accuracy of self-report. PD self-report can be assumed to be relatively accurate because it is a serious condition without a negative social stigma [37-40]. However, all ascertainment methods relied on a previous diagnosis of PD by medical practitioners, when it is known that misdiagnosis in the community setting can be greater than 25%, and underdiagnosis more than 20% [33,41]. Another possible cause for underreporting was that the baseline questionnaire did not specifically ask about past diagnosis of PD, so the participant had to write the disease in a part of the form to list 'other serious illnesses.' Underascertainment and underdiagnosis would dilute the observed difference in reported symptoms between PD cases and controls because there would be unidentified cases in the control group. However, the control group is relatively large so the effect would be minimal.
Sensitivity and specificity were calculated for each question and combinations of questions, however the outcome measure was self-reported PD. In the absence of a true gold standard pathological marker for PD, a thorough neurological examination by a movement disorders specialist would be preferred. The measures in this study show how well the self-reported symptom questions relate to self-reported PD. De novo PD cases would not be included in the cases, therefore the true sensitivity and specificity of the questionnaire are unknown. The values do give a good indication of which questions might not be useful for distinguishing cases from controls, such as slower walking.
Comparisons and conclusions
Although this study was not an evaluation of a screening instrument, the studies reporting the responses to similar questionnaires were all either for screening validation or prevalence surveys [4-7,28,31]. Positive responses are compared between the individuals in the sub-cohort who answered all symptoms questions, for above and below 65 years of age, and the unaffected controls from other studies. The symptom responses are also compared between PD cases (Table 6). The proportion of individuals with complete responses reporting individual parkinsonian symptoms in the sub-cohort was approximately 4–10% for resting tremor, difficulty walking, difficulty getting out of a chair, micrographia, and olfactory dysfunction. In addition, five studies conducted neurological examinations to determine the presence of neurological signs in elderly populations [1,8-11]. The neurological signs tremor, gait disturbance, and shuffling feet correspond with some of the previous questionnaire symptoms, but only tremor was included in the current study questionnaire.
If resting tremor and shaking of arms or legs are considered to be the same symptom, most of the questionnaire studies report prevalence less than 10% except for the 35% reported by a relatively young control group in Pramstaller et al [5]. There is a large range across the neurological examination from 1% to 30%, which is not explained by age. A number of differences could explain the variation including the differences in study populations and assessment methods. Resting tremor was reported by 42% of the cases, which is low compared to the 75–90% reported in several other studies [42-45]. Some of the studies asked about tremor or shaking without specifying that it was occurring at rest. If the tremor question alone is used then 43 (81.1%) of PD cases and 888 (8%) of individuals with complete symptoms responses answered positively.
Walking slower was the most commonly reported symptom in cases and in all EPIC subjects who responded to all questions. The only other study to report on this symptom used interviewer observation rather than a questionnaire for assessment, therefore the much lower prevalence in that study is not truly comparable.
Trouble rising from a chair was reported less frequently by EPIC participants with complete responses compared with other questionnaires. The same was true for PD cases. The presence of micrographia within EPIC corresponded with other questionnaires for the two age groups, below and above 65 years, and for PD patients. Difficulty walking was not specifically asked on comparison questionnaires, although other symptoms related to walking are reported (Table 6). Olfactory dysfunction was only reported by EPIC.
The lower prevalence of parkinsonian symptoms in the EPIC sub-cohort compared to the previous studies could be explained by selection bias for a healthy cohort, although the observed prevalence of PD from self-report was close to the expected prevalence of PD [33]. The lower prevalence of symptoms in EPIC PD cases compared to cases from other studies could be due to less severe or more recently diagnosed disease in the EPIC participants, although dates of diagnosis are not available at this time.
The age division of EPIC participants showed that only walking slower and micrographic were much more frequent in individuals above 65 years. The other symptoms were similar in both groups, which is unexpected, however the questions may not be specific to the symptom of interest since the question is dependent on individual interpretation.
Despite the limitations of the study, there were important findings concerning the reported prevalence of parkinsonian symptoms in the sub-cohort and stratified by self-reported PD status. The results are not unexpected but it is useful to support evidence that individual and cumulative self-reported parkinsonian symptoms are strongly associated with self-reported PD diagnosis and increasing age. The data also provide information regarding the prevalence of symptoms in a younger population of adults than previously reported in the literature.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LI participated in the design of the study, carried out statistical analyses, and drafted the manuscript. KK, RL, SB, AW, and ND conceived of the study, and participated in its design and coordination. CB participated in the design of the study, data analysis methods, and critical revision of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
EPIC-Norfolk is supported by programme grants from the Medical Research Council UK and Cancer Research UK and with additional support from the European Union, Stroke Association, British Heart Foundation, Department of Health, Food Standards Agency and the Wellcome Trust. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication. LI was partially financially supported by the Cambridge Overseas Trust and Christ's College, Cambridge. The authors are grateful to the EPIC-Norfolk group, Sara Godward, Tom Foltynie, Cynthia Ishihara, Adrian Mander, Brian Tom, and Anna Marnik.
==== Refs
Bennett DA Beckett LA Murray AM Shannon KM Goetz CG Pilgrim DM Evans DA Prevalence of parkinsonian signs and associated mortality in a community population of older people N Engl J Med 1996 334 71 76 8531961 10.1056/NEJM199601113340202
Mitchell SL Rockwood K The association between parkinsonism, Alzheimer's disease, and mortality: a comprehensive approach J Am Geriatr Soc 2000 48 422 425 10798470
Wilson RS Schneider JA Beckett LA Evans DA Bennett DA Progression of gait disorder and rigidity and risk of death in older persons Neurology 2002 58 1815 1819 12084882
Mutch WJ Smith WC Scott RF A screening and alerting questionnaire for parkinsonism Neuroepidemiology 1991 10 150 156 1922649
Pramstaller PP Falk M Schoenhuber R Poewe W Validation of a mail questionnaire for parkinsonism in two languages (German and Italian) J Neurol 1999 246 79 86 10195401 10.1007/s004150050312
Rocca WA Maraganore DM McDonnell SK Schaid DJ Validation of a telephone questionnaire for Parkinson's disease J Clin Epidemiol 1998 51 517 523 9636001 10.1016/S0895-4356(98)00017-1
Teresi JA Albert SM Holmes D Mayeux R Use of latent class analyses for the estimation of prevalence of cognitive impairment, and signs of stroke and Parkinson's disease among African-American elderly of central Harlem: results of the Harlem Aging Project Neuroepidemiology 1999 18 309 321 10545783 10.1159/000026226
Mitchell SL Rockwood K Defining parkinsonism in the Canadian Study of Health and Aging Int Psychogeriatr 2001 13 Supp 1 107 113 11892957 10.1017/S1041610202008049
Waite LM Broe GA Creasey H Grayson D Edelbrock D O'Toole B Neurological signs, aging, and the neurodegenerative syndromes Arch Neurol 1996 53 498 502 8660150
Prettyman R Extrapyramidal signs in cognitively intact elderly people Age Ageing 1998 27 557 560 12675093
Kumamoto T Sannomiya K Ueyama H Aoki K Nakashima T Nakamura R Tsuda T Neurological abnormalities in cognitively impaired but not demented elderly Acta Neurol Scand 2000 102 292 298 11083505 10.1034/j.1600-0404.2000.102005292.x
Tanner CM Ben Shlomo Y Epidemiology of Parkinson's disease Adv Neurol 1999 80 153 159 10410715
Lang AE Lozano AM Parkinson's disease. First of two parts N Engl J Med 1998 339 1044 1053 9761807 10.1056/NEJM199810083391506
Y BS Martyn C and Hughes R Parkinson's disease The Epidemiology of Neurological Disorders 1998 London, BMJ Books 1 33
Siderowf A Parkinson's disease: clinical features, epidemiology and genetics Neurol Clin 2001 19 565 78, vi 11532644 10.1016/S0733-8619(05)70035-9
Wszolek ZK Markopoulou K Olfactory dysfunction in Parkinson's disease Clin Neurosci 1998 5 94 101 10785834 10.1016/S0967-5868(98)90213-2
Doty RL Bromley SM Stern MB Olfactory testing as an aid in the diagnosis of Parkinson's disease: development of optimal discrimination criteria Neurodegeneration 1995 4 93 97 7600189 10.1006/neur.1995.0011
Mesholam RI Moberg PJ Mahr RN Doty RL Olfaction in neurodegenerative disease: a meta-analysis of olfactory functioning in Alzheimer's and Parkinson's diseases Arch Neurol 1998 55 84 90 9443714 10.1001/archneur.55.1.84
Stern MB Doty RL Dotti M Corcoran P Crawford D McKeown DA Adler C Gollomp S Hurtig H Olfactory function in Parkinson's disease subtypes Neurology 1994 44 266 268 8309571
Day N Oakes S Luben R Khaw KT Bingham S Welch A Wareham N EPIC-Norfolk: study design and characteristics of the cohort. European Prospective Investigation of Cancer Br J Cancer 1999 80 95 103 10466767
Benito-Leon J Bermejo-Pareja F Morales-Gonzalez JM Porta-Etessam J Trincado R Vega S Louis ED Incidence of Parkinson disease and parkinsonism in three elderly populations of central Spain Neurology 2004 62 734 741 15007123
de Rijk MC Breteler MM Graveland GA Ott A Grobbee DE van der Meche FG Hofman A Prevalence of Parkinson's disease in the elderly: the Rotterdam Study Neurology 1995 45 2143 2146 8848182
Meara RJ Bisarya S Hobson JP Screening in primary health care for undiagnosed tremor in an elderly population in Wales J Epidemiol Community Health 1997 51 574 575 9425469
Bergareche A De La PE Lopez M Sarasqueta C de Arce A Poza JJ Marti-Masso JF Prevalence of Parkinson's disease and other types of Parkinsonism. A door-to-door survey in Bidasoa, Spain J Neurol 2004 251 340 345 15015016 10.1007/s00415-004-0333-3
Hoglinger GU Rissling I Metz A Ries V Heinermann A Prinz H Spieker S Deuschl G Baum E Oertel WH Enhancing recognition of early Parkinsonism in the community Mov Disord 2004 19 505 512 15133813 10.1002/mds.20033
Sevillano MD de Pedro-Cuesta J Duarte J Claveria LE Field validation of a method for population screening of parkinsonism Mov Disord 2002 17 258 264 11921110 10.1002/mds.10088
Tison F Dartigues JF Dubes L Zuber M Alperovitch A Henry P Prevalence of Parkinson's disease in the elderly: a population study in Gironde, France Acta Neurol Scand 1994 90 111 115 7801736
Duarte J Claveria LE de Pedro-Cuesta J Sempere AP Coria F Calne DB Screening Parkinson's disease: a validated questionnaire of high specificity and sensitivity Mov Disord 1995 10 643 649 8552118 10.1002/mds.870100518
Chan DK Dunne M Wong A Hu E Hung WT Beran RG Pilot study of prevalence of Parkinson's disease in Australia Neuroepidemiology 2001 20 112 117 11359078 10.1159/000054769
Chan DK Cordato D Karr M Ong B Lei H Liu J Hung WT Prevalence of Parkinson's disease in Sydney Acta Neurol Scand 2005 111 7 11 15595932 10.1111/j.1600-0404.2004.00348.x
Meneghini F Rocca WA Anderson DW Grigoletto F Morgante L Reggio A Savettieri G di Perri R Validating screening instruments for neuroepidemiologic surveys: Experience in Sicily Journal of Clinical Epidemiology 1992 45 319 331 1314889 10.1016/0895-4356(92)90033-J
Kis B Schrag A Ben-Shlomo Y Klein C Gasperi A Spoegler F Schoenhuber R Pramstaller PP Novel three-stage ascertainment method: prevalence of PD and parkinsonism in South Tyrol, Italy Neurology 2002 58 1820 1825 12084883
Schrag A Ben Shlomo Y Quinn NP Cross sectional prevalence survey of idiopathic Parkinson's disease and Parkinsonism in London BMJ 2000 321 21 22 10875828 10.1136/bmj.321.7252.21
Idler EL Benyamini Y Self-rated health and mortality: a review of twenty-seven community studies J Health Soc Behav 1997 38 21 37 9097506
Louis ED Tang MX Schupf N Mayeux R Functional Correlates and Prevalence of Mild Parkinsonian Signs in a Community Population of Older People Arch Neurol 2005 62 297 302 15710859 10.1001/archneur.62.2.297
Hoogervorst ELJ van Winsen LML Eikelenboom MJ Kalkers NF Uitdehaag BMJ Polman CH Comparisons of patient self-report, neurologic examination, and functional impairment in MS Neurology 2001 56 934 937 11294932
Fowles JB Fowler EJ Craft C Validation of claims diagnoses and self-reported conditions compared with medical records for selected chronic diseases J Ambul Care Manage 1998 21 24 34 10181337
Sherbourne CD Meredith LS Quality of self-report data: a comparison of older and younger chronically ill patients J Gerontol 1992 47 S204 S211 1624716
Harlow SD Linet MS Agreement between questionnaire data and medical records. The evidence for accuracy of recall Am J Epidemiol 1989 129 233 248 2643301
Simpson CF Boyd CM Carlson MC Griswold ME Guralnik JM Fried LP Agreement between self-report of disease diagnoses and medical record validation in disabled older women: factors that modify agreement J Am Geriatr Soc 2004 52 123 127 14687326 10.1111/j.1532-5415.2004.52021.x
Meara J Bhowmick BK Hobson P Accuracy of diagnosis in patients with presumed Parkinson's disease Age Ageing 1999 28 99 102 10350403 10.1093/ageing/28.2.99
Chalmanov VN Epidemiological studies of parkinsonism in Sofia Neuroepidemiology 1986 5 171 177 3785531
Li SC Schoenberg BS Wang CC Cheng XM Rui DY Bolis CL Schoenberg DG A prevalence survey of Parkinson's disease and other movement disorders in the People's Republic of China Arch Neurol 1985 42 655 657 4015461
Morgante L Rocca WA Di Rosa AE De Domenico P Grigoletto F Meneghini F Reggio A Savettieri G Castiglione MG Patti F . Prevalence of Parkinson's disease and other types of parkinsonism: a door-to-door survey in three Sicilian municipalities. The Sicilian Neuro-Epidemiologic Study (SNES) Group Neurology 1992 42 1901 1907 1407570
Melcon MO Anderson DW Vergara RH Rocca WA Prevalence of Parkinson's disease in Junin, Buenos Aires Province, Argentina Mov Disord 1997 12 197 205 9087978 10.1002/mds.870120210
|
16120210
|
PMC1208899
|
CC BY
|
2021-01-04 16:28:53
|
no
|
BMC Neurol. 2005 Aug 24; 5:15
|
utf-8
|
BMC Neurol
| 2,005 |
10.1186/1471-2377-5-15
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-511612022610.1186/1471-2202-6-51Research ArticleHemispheric biases and the control of visuospatial attention: an ERP study Spencer Kevin M [email protected] Marie T [email protected] Department of Psychiatry, Harvard Medical School/VA Boston Healthcare System, Brockton, Massachusetts USA2 Department of Psychology, University of Colorado at Boulder, Boulder, Colorado USA2005 24 8 2005 6 51 51 30 3 2005 24 8 2005 Copyright © 2005 Spencer and Banich; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
We examined whether individual differences in hemispheric utilization can interact with the intrinsic attentional biases of the cerebral hemispheres. Evidence suggests that the hemispheres have competing biases to direct attention contralaterally, with the left hemisphere (LH) having a stronger bias than the right hemisphere. There is also evidence that individuals have characteristic biases to utilize one hemisphere more than the other for processing information, which can induce a bias to direct attention to contralateral space. We predicted that LH-biased individuals would display a strong rightward attentional bias, which would create difficulty in selectively attending to target stimuli in the left visual field (LVF) as compared to right in the performance of a bilateral flanker task.
Results
Consistent with our hypothesis, flanker interference effects were found on the N2c event-related brain potential and error rate for LH-biased individuals in the Attend-LVF condition. The error rate effect was correlated with the degree of hemispheric utilization bias for the LH-Bias group.
Conclusion
We conclude that hemispheric utilization bias can enhance a hemisphere's contralateral attentional bias, at least for individuals with a LH utilization bias. Hemispheric utilization bias may play an important and largely unrecognized role in visuospatial attention.
==== Body
Background
In this study we investigated the hypothesis that the control of spatial attention is influenced by hemispheric utilization bias, a characteristic bias of individuals to utilize one hemisphere more than the other for information processing [1]. Spatial selective attention is commonly thought to be mediated by a network in which attentional control systems based in prefrontal and posterior parietal areas modulate processing in perceptual areas [2,3]. Each of the cerebral hemispheres contains a set of mechanisms for attentional control and perceptual representation [4,5], which may be capable of operating independently in the intact brain [6]. These hemispheric attention systems appear to have mutually inhibitory biases that compete to direct attention to contralateral space, with the left hemisphere (LH) having a stronger and/or more focused intrinsic bias than the right hemisphere (RH) [7].
Evidence for interhemispheric attentional competition comes from studies showing asymmetrical gradients in spatial attention in healthy individuals [8,9], split-brain patients [10-12], and neglect patients [13,14]. Also in neglect patients, disruption of the contralesional hemisphere by a subsequent stroke [15] or transcranial magnetic stimulation (TMS) [16] can cause recovery from neglect symptoms, consistent with the lesioned hemisphere being released from inhibition by the contralesional hemisphere. Furthermore, in healthy individuals, reduction of the excitability of the parietal cortex in one hemisphere by TMS can lead to an increase in subjects' ability to attend to ipsilateral stimuli [17]. Together, these lines of evidence point to the focus of spatial attention being influenced by mutually competitive hemispheric attention systems.
Hemispheric utilization bias appears to represent another, largely unexplored type of interhemispheric competition. Individuals appear to have a characteristic and consistent bias to utilize one hemisphere more than the other for processing information. A reliable finding from many studies addressing the issue of between-subject variance on laterality tasks has been that the major source of this variability is not random measurement error nor variability in hemispheric specialization. Rather, a significant source of variation comes from characteristic biases that individuals possess for consistently utilizing one hemisphere more than the other for processing information [1,18-22].
For example, Levine and colleagues [21] tested individuals in three perceptual recognition tasks with bilateral stimulus presentations using words, faces, and pictures of chairs. As a group, the participants demonstrated a right visual field (RVF)/LH advantage on the word task, a left visual field (LVF)/RH advantage on the face task, and no visual field advantage on the chair task, as was expected. Yet individuals differed with respect to the size and direction of these advantages. Levine et al. found that individuals' asymmetry scores on the tasks were correlated, such that participants with a RVF/LH advantage on the chair task had a larger-than-average RVF/LH advantage on the word task, and a smaller-than-average LVF/RH advantage on the face task. Hence, across all tasks these individuals were biased to utilize their LH more than average across all tasks. The reverse pattern was found for participants with a LVF/RH advantage on the chair task. The common factor affecting a given individual's performance on the three tasks is what we call here hemispheric utilization bias (referred to as "characteristic perceptual asymmetry" by some researchers). When the inter-subject variability due to hemispheric utilization bias is taken into account, the pattern of hemispheric specialization across a population of individuals is consistent with that predicted by studies of neurological patients [23]. Thus, hemispheric utilization bias is stable within an individual but variable across individuals.
Levy and colleagues [1] proposed that hemispheric utilization bias induces a characteristic bias to direct attention to the side of space contralateral to the more activated hemisphere, which leads to enhanced processing of information in the contralateral field regardless of the nature of the stimulus. Thus, it seems plausible that individuals' hemispheric utilization biases could interact with the intrinsic attentional biases of the hemispheres.
In the present study we investigated this hypothesis, predicting that the intrinsic attentional bias of a hemisphere to the contralateral side of space would be enhanced when an individual is also biased to utilize that same hemisphere. Competition between the hemispheres in attentional control was assessed using a bilateral version of the Eriksen flankers task [24], in which participants responded to target stimuli in the attended hemifield while ignoring a simultaneous flanking distractor in the opposite hemifield (Figure 1). Flanker items could be Compatible (same response as target) or Incompatible (opposite response). The attended hemifield was varied across blocks.
We expected that the effect of the flanker would be larger when presented to the hemisphere for which an individual has a utilization bias, compared to when the it was presented to the opposite hemisphere. Furthermore, this effect would be modulated by the greater strength of the LH's intrinsic bias compared to the RH. In particular, we predicted that LH-biased individuals would show the largest flanker effects when they attended to target items in the left visual field (LVF) and tried to ignore flankers in the right visual field (RVF), for which they would have a strong attentional bias. Hemispheric utilization bias was measured using the chimeric faces test (CFT; [25]) (see Methods).
Studies of neglect and extinction patients have found evidence that stimuli presented to each hemisphere are processed to a post-identification level without reaching awareness, suggesting that interhemispheric attentional competition occurs at a late stage of processing [26-29]. Therefore, we focused the present investigation on the N2c component of the ERP, rather than on early sensory components such as the P1 and N1 that reflect the spatial distribution of early selective attention processes [30]. The N2c is elicited by stimuli that are associated with response conflict, such as a "no-go" stimulus in a "go/no-go" task [31], or by stimulus displays containing response-incompatible distractors, as in flanker [32,33] or Stroop tasks [34]. N2c amplitude is related to the degree of response conflict on correct-response trials [32], even when no overt signs of response conflict are present, suggesting that this component may reflect both the detection and suppression of incorrect response information [35]. The N2c is maximal at midline fronto-central sites with a peak around 200–300 ms post-stimulus in tasks with simple stimuli. Source localization studies suggest that the N2c may be generated in the anterior cingulate cortex, similar to the error related negativity [33,34].
Results
The behavioral data are presented in Fig. 2, and the ERP data in Figs. 3 and 4. As is typical in an oddball paradigm, rare stimuli compared to frequents were associated with higher error rates, slower reaction times (RTs), and elicited larger N2 and P300 components (see Fig. 3), so these effects will not be discussed further. We focused our analyses on the responses elicited by Compatible and Incompatible Rares, which are matched for stimulus probability.
As shown in Fig. 2, Incompatible Rares elicited more errors than Compatible Rares (Flanker main effect: F[1,22] = 15.8, p < 0.001). This difference was larger in the Attend-LVF condition than in Attend-RVF (F[1,22] = 12.7, p < 0.01), and the Attention × Flanker interaction was statistically significant for the LH-Bias group but not the RH-Bias group (LH-Bias: F[1,11] = 19.8, p < 0.001; RH-Bias: F[1,11] = 0.861, n.s.; Group × Attention × Flanker: F[1,22] = 4.57, p < 0.05). Planned comparisons confirmed these effects. Thus, LH-biased individuals were more likely to make an error on rare trials when a response-incompatible flanker was presented to their biased hemisphere. In the RT data, the overall Attention × Flanker interaction approached significance (F[1,22] = 3.99, p = 0.058), but the Group × Attention × Flanker interaction was not significant (F[1,22] = 0.177, n.s.).
As predicted, Incompatible Rare stimuli elicited a larger N2c component (245–325 ms) than Compatible Rares only in the Attend-LVF condition for the LH-Bias group (Group × Attention × Flanker: F[1,22] = 8.86, p < 0.01; LH-Bias, Attention × Flanker: F[1,11] = 13.7, p < 0.01; RH-Bias, Attention × Flanker: F[1,11] = 0.695, n.s.). Direct comparison of the responses for the LH-Bias group confirmed that Incompatible Rares elicited larger N2s than Compatible Rares in Attend-LVF (F[1,11] = 9.22, p < 0.05), but there was no difference in Attend-RVF (F[1,11] = 2.79, n.s.). For the RH-Bias group there was no interaction between attention and flanker compatibility (Attention × Flanker: F[1,11] = 0.695, n.s.; Group × Attention × Flanker: F[1,22] = 8.86, p < 0.01).
Inspection of the waveforms in Fig. 4 suggests that the P2 component (160–240 ms), which peaks before the N2c, shows a complementary pattern of differences between Compatible and Incompatible Rares across the two attention conditions: the P2 is increased rather than decreased in amplitude for Compatible Rares compared to Incompatible Rares. For the P2, the Group × Attention × Flanker interaction approached significance (F[1,22] = 3.75, p = 0.066), and the Attention × Flanker interaction was significant for the LH-Bias group (F[1,11] = 10.52, p < 0.01) but not the RH-Bias group (F[1,11] = 0.304, n.s.). For the LH-Bias group, there was a trend for P2 amplitude to be larger for Compatible relative to Incompatible Rares in Attend-LVF (F[1,11] = 3.50, p = 0.088), and P2s elicited by Incompatible Rares were significantly larger than Compatible Rares in Attend-RVF (F[1,11] = 5.25, p < 0.05). The marginal P2 effect in the Attend-LVF condition may reflect overlap with the N2c, especially given the similar topography of the effects (maximal at Fz), but the P2 effect in Attend-RVF is harder to interpret. The P2 is seldom reported to be modulated by spatial attention, and to our knowledge, has not been reported to be modulated by response compatibility. At this point, we can only speculate that the Attend-RVF P2 effect could be due to a registration of the incompatible flanker at a perceptual, rather than response-related level.
Since we found relationships between hemispheric utilization bias and some of the dependent variables, we investigated whether individual differences in these measures might be related. Non-parametric Spearman's correlation coefficients (2-tailed) were computed between CFT score, the Attend-LVF error rate effect (Incompatible Rare minus Compatible Rare), and the Attend-LVF N2c effect (Incompatible Rare minus Compatible Rare) separately for each group. For the LH-Bias group, CFT score and the error rate effect were significantly correlated (ρ = 0.591, p < 0.05), indicating that the greater the degree of LH bias, the more the individuals were influenced by conflicting items presented to their LH. For the RH-Bias group, a negative correlation between CFT score and error rate approached significance (ρ = -0.515, p = 0.087). No other correlations with CFT score were found.
Discussion
When attending to the LVF, the LH-Bias group made more errors when the flanker presented to their biased hemisphere conflicted with the target item in the opposite visual field. The size of this effect was correlated with the degree of LH utilization bias as measured by the CFT. Furthermore, on correct trials, Incompatible Rare arrays elicited larger N2c components than Compatible Rare arrays, indicating that the conflict between the flanker and target reached a late, response-related stage of processing. Taken together, these data suggest that when LH-biased individuals focus attention in the LVF, they still attend at some level to items in the task-irrelevant RVF because of their hemispheric utilization bias.
Flanker compatibility had a much stronger effect on error rate than RT for the Rare arrays. The most likely explanation for this discrepancy is that the slowing of RT due to the low probability of the Rares overrode the effects of flanker compatibility. In unpublished data, we have obtained flanker effects on RT using the same displays but with equal stimulus probabilities [36]. Another possibility is that the manner in which individuals allocated attention between the visual fields varied across trials. On the majority of Incompatible Rare trials, the potency of the RVF flankers influenced processing (manifested by the N2c) but ultimately not overt response production (no effect on RT). However, on a sizable proportion of Incompatible Rare trials, the influence of the conflicting RVF flankers did reach the level of response production, leading to errors.
These data demonstrate that interhemispheric competition for the control of spatial attention can indeed be influenced by individual differences in hemispheric utilization. We predicted that when individuals have a LH utilization bias, it would augment the LH's already strong intrinsic attentional bias to the contralateral visual field. This enhancement of the LH's attentional bias would allow task-irrelevant information presented in the RVF to interfere with task-relevant LVF items. For the RH-Bias group, however, no effects of interhemispheric attentional competition were detected for the corresponding set of conditions (incompatible LVF flankers in the Attend-RVF condition). This outcome may be due to the weaker intrinsic attentional bias of the RH to the contralateral side of space. Further research is necessary to examine whether hemispheric utilization bias can influence attentional control by the RH.
The N2c data are consistent with the existence of a late locus of interhemispheric attentional competition, as demonstrated by behavioral studies [26-29]. The N2c results could also be interpreted in terms of the LH attempting to control response output, rather than as a failure to control spatial attention. Such a proposal would be consistent with the evidence from split-brain patients for an interhemispheric "bottleneck" in response control [37,38].
An important question to be examined in future studies is whether hemispheric utilization bias influences the early stages of spatial selective attention that are manifested by the P1 and posterior N1 components [30]. The present task was not well-suited to examine these early attention effects due to the bilateral stimulus displays, relatively small number of stimuli, and long stimulus onset asynchrony (SOA). The P1 and N1 attention effects are typically obtained in tasks involving unilateral displays (but see [39]) and hundreds of trials with rapid stimulus presentation (~300 ms SOA). SOA may be a particularly important factor, as we have found in unpublished data that as SOA decreases, flanker compatibility effects also decrease. Nevertheless, the examination of early spatial attention effects will be necessary to fully determine the mechanisms by which hemispheric attentional biases operate.
Conclusion
This study provides evidence that hemispheric utilization bias influences visuospatial attention at a late, response-related stage of processing, by enhancing a hemisphere's intrinsic contralateral attentional bias. These data are consistent with the hypothesis of Kinsbourne [7] that the cerebral hemispheres have mutually inhibitory and asymmetric biases to direct attention to contralateral space. The results of this study suggest that hemispheric utilization bias may play an important and largely unrecognized role in visuospatial attention.
Methods
Chimeric faces test (CFT)
On each trial of the CFT [25], participants are shown two chimeric faces, each composed of a smiling half-face and a neutral half-face, which are mirror images of one another (Fig. 5). Participants are given several seconds to scan the faces in free vision, and their task is to decide which of the two faces appears happier than the other. The version of the CFT used here was administered in booklet form, with responses marked on a response sheet. The CFT score is computed according to the formula (R - L)/(R + L), where R and L are the total number of times an individual chooses a face with a right or left smile. Hence, the CFT score indicates the degree and direction of an individual's perceptual asymmetry. It ranges from -1.0 (complete leftward bias) to +1.0 (complete rightward bias).
Across individuals there is a left hemispatial bias on the CFT, in agreement with a RH advantage for processing faces and facial expressions in particular [40]. It is the degree and direction in which an individual differs from this mean that reflects hemispheric utilization bias. Some of the features of this test that make it a useful measure of hemispheric utilization bias are its high test-retest and split-half reliability [25] and its sensitivity to attentional biases, such as those seen in neglect [41].
Participants
Participants were recruited from the university community and received monetary reimbursement. All gave their informed consent to participate in the study, as per the guidelines of the Institutional Review Board of the University of Illinois at Urbana-Champaign and in accordance with the Helsinki Declaration. Participants were right-handed as assessed by the Edinburgh inventory [42] and had normal or corrected-to-normal vision.
Twenty-seven individuals participated in the study. The data from 3 were unusable due to data acquisition problems, resulting in a final sample of 24 (13 females, 11 males; age range 18–32 years). They were divided into two groups (N = 12 each) based on their performance on the CFT. The LH-Bias group comprised all individuals with a score indicating more LH involvement than the mean (-0.377, an estimate of the population mean based on a sample of 1000 individuals; Heller, unpublished data, 1996), and the RH-Bias group comprised all individuals with a score indicating more RH involvement than the mean. The overall mean CFT score was -0.202, which did not significantly differ from the population estimate (t[23] = 1.56, p = 0.132). The LH-Bias group consisted of 7 females and 5 males, with a mean CFT score of +0.262. The RH-Bias group consisted of 6 females and 6 males, with a mean CFT score of -0.667. By subtracting the population mean CFT score from each individual's score, we found that the absolute degree of hemispheric utilization bias was larger for the LH-Bias group than the RH-Bias group (t[22] = 3.054, p < 0.01).
Stimuli and task design
Stimulus arrays (Fig. 1) consisted of two items placed to the left and right of the center of the display on the horizontal meridian. Stimulus items were line drawings of circles and squares presented in white on a black background. All stimulus elements were 2° wide, with their inner edges 4° from the center of the display. A fixation cross was continuously present at the center of the display. The stimulus arrays were presented for 150 ms in white on a black background, with a 2000 ms SOA.
After giving informed consent, participants filled out medical history and handedness questionnaires and took the CFT. Following electrode application, they were seated in a darkened sound-attenuation booth and rested their head on a chinrest in front of the computer monitor on which the stimuli were presented.
Participants performed a bilateral version of the Eriksen flanker task [24] in an "oddball" format. They responded to target stimuli in one hemifield while ignoring a simultaneous flanking distractor in the opposite hemifield (Fig. 1). The flanker could be Compatible (same stimulus as the target) or Incompatible (the other potential target item). To increase the potential for response conflict, the stimulus arrays were presented with different probabilities. Arrays with circle targets were rare (20%) and arrays with square targets were frequent (80%). Thus, stimulus arrays were Compatible Rare (circle for target and flanker), Incompatible Rare (circle target, square flanker), Compatible Frequent (square for target and flanker), and Incompatible Frequent (square target, circle flanker). The different stimulus probabilities ensured that participants would develop a bias to make a "square" response, so that response conflict would be maximal on trials in which the target item was a circle and the flanker was a square (Incompatible Rares).
The flanker task was performed while attending to the LVF and RVF (Attend-LVF and Attend-RVF conditions). The task was to press one button for circles and another for squares in the attended visual field using separate fingers on the same hand, responding as quickly and accurately as possible. Each attention condition was performed twice, once with each hand. Practice trials were given prior to each attention condition to familiarize the participants with task requirements.
Participants were instructed to maintain central fixation, which was monitored by the horizontal electro-oculogram (EOG) (see below). They were also instructed to make verbal responses (saying the word "check") to infrequent fixation check stimuli. On fixation check trials (5% of the total trials), an "X" (1° wide) was presented at the center of the display, overlapping the fixation cross. The fixation check stimuli were presented simultaneously with Compatible Frequent arrays. No participant missed more than 1 or 2 fixation check stimuli.
The order of the attention conditions and response hands was counterbalanced across participants. In each combination of attention condition and response hand, participants received 24 trials (10%) each of Compatible Rare, Incompatible Rare, and Incompatible Frequent arrays; 156 (65%) Compatible Frequent arrays; and 12 (5%) Fixation Check arrays.
Data acquisition and processing
Electroencephalographic (EEG) and EOG signals (0.01–100 Hz passband, 60 Hz notch filter) were digitized at 200 Hz. The EEG was recorded from 12 scalp sites (F3/Fz/F4, C3/Cz/C4, P3/Pz/P4, O1/Oz/O2) and the right mastoid. Scalp electrode sites were initially referenced to the left mastoid and re-referenced off-line to averaged mastoids. The vertical EOG was recorded from bipolar-referenced electrodes above and below the left eye. The horizontal EOG was recorded from bipolar electrodes at the outer canthi of the eyes.
Trials were excluded from further analysis for the following reasons: 1) the horizontal EOG exceeded 1° of visual angle (calculated from calibration data obtained for each participant); 2) amplifier blocking occurred; and 3) no response was made by 1750 ms. After horizontal eye-movement rejection was performed (see results in Fig. 6), vertical EOG artifacts were removed from the EEG by an eye-movement correction method [43]. Averaged ERPs for each combination of attention condition and stimulus array were derived from correct response trials, collapsed over response hand, and baseline-corrected with a 100-ms prestimulus baseline.
Data analysis
Behavioral performance measures were error rate and median RT. Errors were defined as incorrect target classifications (i.e., circle as square). ERP average amplitudes were measured at the electrode sites and latency windows where the components were maximal, based on the grand averages. ANOVAs consisted of the design: hemispheric utilization bias Group (LH-Bias/RH-Bias) × Attention condition (Attend-LVF/Attend-RVF) × Flanker (Compatible/Incompatible Rare) × Electrode site (for ERP amplitudes, Fz/Cz). ANOVAs were performed with the Greenhouse-Geisser correction reflected in the reported p values.
Authors' contributions
KMS and MTB designed the study. KMS acquired and analyzed the data, and drafted the manuscript. MTB also helped to draft the manuscript.
Acknowledgements
This study was conducted at the University of Illinois at Urbana-Champaign, and was supported by predoctoral Individual National Research Service Award 11516 from NIMH to KMS. Preliminary reports of this study were presented at the 1998 Meeting of the Society for Psychophysiological Research, Denver, USA, and at the 2000 Meeting of the Cognitive Neuroscience Society, San Francisco, USA. The authors gratefully acknowledge the support of Michael G. H. Coles, and the advice of Emanuel Donchin, Arthur Kramer, and Joseph Malpeli.
Figures and Tables
Figure 1 Stimulus arrays. The stimulus arrays in each attention condition are shown, along with the percentage of trials on which they were presented.
Figure 2 Behavioral performance data. Error rate and median RT data for each hemispheric utilization bias group. Bars indicate standard error. Comp = Compatible, Inc = Incompatible.
Figure 3 Grand average ERPs (whole epoch). ERPs in each condition at electrodes Fz, Cz, and Pz are shown for the LH-Bias and RH-Bias groups. The N2c and P300 components are indicated.
Figure 4 Grand average ERPs (early portion of epoch). ERPs in the early portion of the epoch at Fz and Cz are shown to highlight the N2c and P2 effects. In the bottom panel, only the Rare stimulus ERPs are shown for the LH-Bias Group.
Figure 5 Chimeric Faces Test (CFT). A sample page from the CFT (see Methods), which was used to measure hemispheric utilization bias.
Figure 6 Horizontal EOG data. Grand average horizontal EOG data in each condition for the hemispheric utilization bias groups are shown. Note that stimulus offset was at 150 ms. In the recording procedures used, 1° of visual angle was equal to approximately +/- 20 μV EOG amplitude.
==== Refs
Levy J Heller W Banich MT Burton LA Are variations among right-handed individuals caused by characteristic arousal differences between hemispheres? J Exp Psychol: Hum Percept Perform 1983 9 329 359 6223975 10.1037//0096-1523.9.3.329
Desimone R Duncan J Neural mechanisms of selective visual attention Annu Rev Neurosci 1995 18 193 222 7605061 10.1146/annurev.ne.18.030195.001205
Posner MI Petersen SE The attention system of the human brain Annu Rev Neurosci 1990 13 25 42 2183676 10.1146/annurev.ne.13.030190.000325
Arguin M Lassonde M Quattrini A Del Pesce M Foschi N Papo I Divided visuo-spatial attention systems with total and anterior callosotomy Neuropsychologia 2000 38 283 291 10678694 10.1016/S0028-3932(99)00077-9
Luck SJ Hillyard SA Mangun GR Gazzaniga MS Independent attentional scanning in the separated hemispheres of split-brain patients J Cog Neurosci 1994 6 84 91
Corballis PM Gratton G Independent control of processing strategies for different conditions in the visual field Biol Psychol 2003 64 191 209 14602362 10.1016/S0301-0511(03)00109-1
Kinsbourne M Robertson IH, Marshall JC Orientational bias model of unilateral neglect: Evidence from attentional gradients within hemispace Unilateral Neglect: Clinical and Experimental Studies 1993 Hove, UK, Erlbaum
Jewell G McCourt ME Pseudoneglect: A review and meta-analysis of performance factors in line bisection tasks Neuropsychologia 2000 38 93 110 10617294 10.1016/S0028-3932(99)00045-7
Reuter-Lorenz PA Kinsbourne M Moscovitch M Hemispheric control of spatial attention Brain Cogn 1990 12 240 266 2340154 10.1016/0278-2626(90)90018-J
Berlucchi G Aglioti S Tassinari G Rightward attentional bias and left hemisphere dominance in a cue-target light detection task in a callosotomy patient Neuropsychologia 1997 35 941 952 9226656 10.1016/S0028-3932(97)00022-5
Corballis MC Line bisection in a man with complete forebrain commissurotomy Neuropsychol 1995 9 147 156 10.1037//0894-4105.9.2.147
Proverbio AM Zani A Gazzaniga MS Mangun GR ERP and RT signs of a rightward bias for spatial orienting in a split-brain patient NeuroReport 1994 5 2457 2461 7696579
Behrmann M Watt S Black SE Barton JJS Impaired visual search in patients with unilateral neglect: an oculographic analysis Neuropsychologia 1997 35 1445 1458 9352522 10.1016/S0028-3932(97)00058-4
Smania N Martini MC Gambina G Tomelleri G Palamara A Natale E Marzi CA The spatial distribution of visual attention in hemineglect and extinction patients Brain 1998 121 1759 1770 9762963 10.1093/brain/121.9.1759
Vuilleumier P Hester D Assal G Regli F Unilateral spatial neglect recovery after sequential strokes Neurology 1996 46 184 189 8559371
Oliveri M Rossini PM Traversa R Cicinelli P Filippi MM Pasqualetti P Tomaiuolo F Caltagirone C Left frontal transcranial magnetic stimulation reduces contralesional extinction in patients with unilateral right brain damage Brain 1999 122 1731 1739 10468512 10.1093/brain/122.9.1731
Hilgetag CC Théoret H Pascual-Leone A Enhanced visual spatial attention ipsilateral to rTMS-induced 'virtual lesions' of human parietal cortex Nat Neurosci 2001 4 953 957 11528429 10.1038/nn0901-953
Kim H Levine SC Sources of between-subjects variability in perceptual asymmetries: a meta-analytic review Neuropsychologia 1991 29 877 888 1834960 10.1016/0028-3932(91)90053-B
Kim H Levine SC Variations in characteristic perceptual asymmetry: modality specific and modality general components Brain Cogn 1992 19 21 47 1605949 10.1016/0278-2626(92)90036-L
Kim H Levine SC Kertesz S Are variations among subjects in lateral asymmetry real individual differences or random error in measurement? Putting variability in its place Brain Cogn 1990 14 220 242 2285515 10.1016/0278-2626(90)90031-I
Levine SC Banich MT Kim H Ottoson O Variations in arousal asymmetry: implications for face processing Duality and Unity of the Brain 1987 New York, Plenum Press
Levine SC Banich MT Koch-Weser M Variations in patterns of lateral asymmetry among dextrals Brain Cogn 1984 3 317 334 6536331 10.1016/0278-2626(84)90024-1
Kim H Levine SC Inferring patterns of hemispheric specialization for individual subjects from laterality data: A two-task criterion Neuropsychologia 1991 3 93 105 10.1016/0028-3932(91)90095-P
Eriksen BA Eriksen CW Effects of noise letters on the identification of a target letter in a nonsearch task Percept Psychophys 1974 16 143 149
Levy J Heller W Banich MT Burton LA Asymmetry of perception in free viewing of chimeric faces Brain Cogn 1983 2 404 419 6546034 10.1016/0278-2626(83)90021-0
Berti A Allport A Driver J Dienes Z Oxbury J Oxbury S Levels of processing for visual stimuli in an "extinguished" field Neuropsychologia 1992 30 403 415 1620321 10.1016/0028-3932(92)90088-4
Fuentes L Humphreys GW On the processing of "extinguished" stimuli in unilateral visual neglect: An approach using negative priming Cog Neuropsychol 1996 13 111 136 10.1080/026432996382088
McGlinchey-Berroth R Milberg WP Verfaellie M Grande L D'Esposito M Alexander M Semantic processing and orthographic specificity in hemispatial neglect J Cog Neurosci 1996 8 291 304
Volpe BT Ledoux JE Gazzaniga MS Information processing of visual stimuli in an "extinguished" visual field Nature 1979 282 722 724 514351 10.1038/282722a0
Luck SJ Woodman GF Vogel EK Event-related potential studies of attention Trends Cog Sci 2000 4 432 440 10.1016/S1364-6613(00)01545-X
Pfefferbaum A Ford JM Weller BJ Kopell BS ERPs to response production and inhibition Electroencephal clin Neurophysiol 1985 71 55 63 10.1016/0168-5597(88)90019-6
Kopp B Rist F Mattler U N200 in the flanker task as a neurobehavioral tool for investigating executive control Psychophysiol 1996 33 282 294
Van Veen V Carter CS The timing of action-monitoring processes in the anterior cingulate cortex J Cog Neurosci 2002 14 593 602 10.1162/08989290260045837
West R Neural correlates of cognitive control and conflict detection in the Stroop and digit-location tasks Neuropsychologia 2003 41 1122 1135 12667546 10.1016/S0028-3932(02)00297-X
Bartholow BD Pearson MA Dickter CL Sher KJ Fabiani M Gratton G Strategic control and medial frontal negativity: beyond errors and response conflict Psychophysiol 2005 42 33 42 10.1111/j.1469-8986.2005.00258.x
Spencer KM Banich MT Hemispheric biases in spatial attention Cognitive Neurosci Soc Abstracts 2000 S96
Pashler H O'Brien S Dual-task interference and the cerebral hemispheres J Exp Psychol: Hum Percept Perform 1993 19 315 330 8473842 10.1037//0096-1523.19.2.443
Reuter-Lorenz PA Nozawa G Gazzaniga MS Hughes HC Fate of neglected targets: a chronometric analysis of redundant target effects in the bisected brain J Exp Psychol: Hum Percept Perform 1995 21 211 230 7714469 10.1037//0096-1523.21.2.211
Heinze HJ Luck SJ Mangun GR Hillyard SA Visual event-related potentials index focused attention within bilateral stimulus arrays. I. Evidence for early selection Electroenceph clin Neurophysiol 1990 75 511 527 1693896 10.1016/0013-4694(90)90138-A
Heller W Levy J Perception and expression of emotion in right-handers and left-handers Neuropsychologia 1981 19 263 272 7254505 10.1016/0028-3932(81)90110-X
Mattingley JB Bradshaw JL Phillips JG Bradshaw JA Reversed perceptual asymmetry for faces in left unilateral neglect Brain Cogn 1993 23 145 165 8292323 10.1006/brcg.1993.1052
Oldfield RC The assessment and analysis of handedness: The Edinburgh inventory Neuropsychologia 1971 9 97 113 5146491 10.1016/0028-3932(71)90067-4
Gratton G Coles MGH Donchin E A new method for off-line removal of ocular artifact Electroencephal clin Neurophysiol 1983 75 468 484 10.1016/0013-4694(83)90135-9
|
16120226
|
PMC1208900
|
CC BY
|
2021-01-04 16:03:47
|
no
|
BMC Neurosci. 2005 Aug 24; 6:51
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-51
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-551612488110.1186/1471-2202-6-55Research ArticleEpidermal growth factor (EGF) withdrawal masks gene expression differences in the study of pituitary adenylate cyclase-activating polypeptide (PACAP) activation of primary neural stem cell proliferation Sievertzon Maria [email protected] Valtteri [email protected] Alex [email protected]én Jonas [email protected] Joakim [email protected] Royal Institute of Technology, AlbaNova University Center, KTH Genome Center, Dept. of Biotechnology, S-106 91 Stockholm, Sweden2 NeuroNova AB, S-114 33 Stockholm, Sweden3 Department of Cell and Molecular Biology, Medical Nobel Institute, Karolinska Institute, S-171 77 Stockholm, Sweden2005 28 8 2005 6 55 55 17 3 2005 28 8 2005 Copyright © 2005 Sievertzon et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The recently discovered adult neural stem cells, which maintain continuous generation of new neuronal and glial cells throughout adulthood, are a promising and expandable source of cells for use in cell replacement therapies within the central nervous system. These cells could either be induced to proliferate and differentiate endogenously, or expanded and differentiated in culture before being transplanted into the damaged site of the brain. In order to achieve these goals effective strategies to isolate, expand and differentiate neural stem cells into the desired specific phenotypes must be developed. However, little is known as yet about the factors and mechanisms influencing these processes. It has recently been reported that pituitary adenylate cyclase-activating polypeptide (PACAP) promotes neural stem cell proliferation both in vivo and in vitro.
Results
We used cDNA microarrays with the aim of analysing the transcriptional changes underlying PACAP induced proliferation of neural stem cells. The primary neural stem/progenitor cells used were neurospheres, generated from the lateral ventricle wall of the adult mouse brain. The results were compared to both differentiation and proliferation controls, which revealed an unexpected and significant differential expression relating to withdrawal of epidermal growth factor (EGF) from the neurosphere growth medium. The effect of EGF removal was so pronounced that it masked the changes in gene expression patterns produced by the addition of PACAP.
Conclusion
Experimental models aiming at transcriptional analysis of induced proliferation in primary neural stem cells need to take into consideration the significant effect on transcription caused by removal of EGF. Alternatively, EGF-free culture conditions need to be developed.
==== Body
Background
During the last decade it has become evident that neurogenesis occurs in certain restricted regions of the adult mammalian brain, particularly in the dentate gyrus of the hippocampus and the subventricular zone (SVZ) of the lateral ventricle [1-3]. The neurogenic cells in these areas have been isolated and can be propagated in vitro where they form clonal aggregates of cells denoted neurospheres [4,5]. However, there seems to be some differences between the populations of neurospheres obtained from the different regions. SVZ neurospheres possess the two cardinal properties of stem cells, i.e. they are self-renewable and multipotent [5], whereas hippocampal neurospheres appear to be more precursor cell-like, in that they are more limited in their self-renewal and individual spheres exclusively give rise to neurons or glia [6].
The rather recently discovered neural stem cell (NSC) population, sustaining continuous neurogenesis in the adult brain, has become a promising source of cells for cell-replacement therapies for various neurological diseases [7-9]. Cell replacement with fetal mesencephalic or striatal tissue has previously been shown to lead to functional improvement in patients with Parkinson's disease and Huntington's disease [10-12]. However, the use of fetal cells is hampered by a number of hurdles, in addition to ethical concerns. Fetal tissue is only available in limited quantities and the fetal cells are mostly postmitotic and cannot be expanded, nor stored for long periods of time. Furthermore these cell populations are heterogeneous, and their purity and viability cannot be reliably controlled, perhaps explaining the variation in functional outcome between different transplantation studies. In contrast, stem cells represent a source of cells that is more readily obtainable, expandable and could potentially be maintained as a more homogeneous, pure cell population. By supplying the appropriate factors endogenous stem/progenitor cells could be recruited to generate new neural or glial cells of specific phenotypes (e.g. dopaminergic neurons in Parkinson's disease or oligodendrocytes in multiple sclerosis). Alternatively the NSCs could be expanded and properly manipulated in vitro, before transplanting them into the appropriate area of the brain. Both strategies have been evaluated with limited but promising success [13-16]. However, more knowledge on the control of NSC proliferation and differentiation into specific phenotypes, both in vivo and in vitro, is needed before any clinical trails can be made.
Some factors stimulating neurogenesis are already known (for reviews see [7,8,17]). These include epidermal growth factor (Egf), basic fibroblast growth factor (Fgf2), brain-derived neurotrophic factor (Bdnf) and transforming growth factor-α (Tgfa). Recently there was a report that neurogenesis can also be stimulated through a G-protein coupled receptor [18]. The pleiotropic neuropeptide pituitary adenylate cyclase-activating polypeptide (PACAP) promotes NSC proliferation both in vivo and in vitro, via the PACAP receptor 1 (PAC1). PACAP belongs to the vasoactive intestinal peptide (VIP)/secretin/glucagon family of peptides and exerts a number of biological effects in addition to neurogenesis (for reviews see [19,20]).
We have recently performed a gene expression analysis of primary neural stem cells (neurospheres), demonstrating the need to consider biological fluctuations of these cells when performing comparative transcriptome analysis [21]. Here we continue the work to investigate the transcriptional changes underlying the proliferative effect of PACAP on SVZ neurospheres by using microarrays, including controls for differentiation and proliferation.
Results
To investigate the molecular changes underlying the proliferative effects of PACAP we performed a gene expression analysis using a 16 k cDNA microarray. The experimental design depicted in Figure 1 was utilised. NSCs were isolated from a pool of lateral ventricular wall tissue from 15 mice and grown in culture as neurospheres, after which they were dissociated and cultured to form secondary neurospheres. Secondary neurospheres were dissociated and split into eight parallel cultures. The cells were used to study the culture variance and three different treatments, each replicated in two individual cultures (indicated by letters a and b in Figure 1.) We used amplification and culture replicates to control for technical and biological variation respectively, and controls for differentiation and proliferation activation to verify our results.
In all comparisons undifferentiated neurospheres (NS), maintained in culture medium supplemented with epidermal growth factor (EGF), was used as a reference control sample. In the first of three treatment regimes neurospheres were induced to proliferate in response to PACAP by replacing the EGF supplemented culture medium with medium supplemented with PACAP. In the proliferation control, the cells were induced to proliferate in response to a transmembrane receptor agonist (TMR agonist) by replacing EGF with the agonist in the neurosphere culture medium. In the third treatment, the differentiation control, neurospheres were induced to differentiate into neurons, astrocytes and oligodendrocytes by replacing the EGF supplemented culture medium with medium supplemented with fetal calf serum, and plating the cells onto poly-D-lysine plates.
mRNA was isolated from all cell cultures and used in a series of cDNA microarray hybridisations. To avoid extensive passaging of the neurospheres, a limited amount of cells were obtained, and the generated RNA was not sufficient for labelling and subsequent microarray analysis. The RNA was therefore amplified using a previously described protocol [22,23], that has recently also been evaluated for neurosphere analysis [21]. The principle relies on incorporating a biotin moiety into the cDNA during the first-strand cDNA synthesis, by using a biotinylated oligo(dT) primer. The population of cDNAs is fragmented and the biotinylated 3'ends captured onto a streptavidin-coated solid support. The isolated cDNA tags are released from the support, amplified by PCR and labelled for subsequent microarray hybridisation.
Amplified and labelled cDNA from each treated sample was hybridised against the neurosphere control sample (NS). In order to measure the technical variation self-to-self hybridisations were made with material from NS sample a. In addition, to measure the variation between two identical cultures, the two NS replicates (a and b) were hybridised against each other. For each comparison two replicate and two dye-swap hybridisations were performed.
We have previously shown that neurosphere culture passaging or prolonged culturing per se, is sufficient to induce differential expression and that this should be taken into account in the design of the experiment [21]. To address these issues, i.e. to get a variance measure from as many slides as possible, the results from all hybridisations were divided into three groups prior to data analysis, as indicated in Figure 1. Group 1 contains eight hybridisations comparing the technical amplification replicates (NS a-NS a) and the biological culturing replicates (NS a-NS b). Groups 2 and 3 consist of sixteen hybridisations each and include the NS a-NS b hybridisations as well as one replicate of each treatment comparison (NS vs. PACAP treated, NS vs. differentiation control and NS vs. proliferation control). This scheme allows for estimation of technical and biological noise (Group 1). Also, by using the NS a and NS b samples as reference samples in each group, the "contrasts" (i.e. the calculated differences between two treatments) can be calculated and compared between the groups (data not shown). The data in all three groups was filtered (for details see Methods) and print-tip lowess normalised using identical criteria. For each group individual gene-wise variances were calculated, and taken into account in the identification of differentially expressed genes using the empirical Bayes moderated t-test [24-26]. For each comparison the log-odds ratio (B-value) was used to rank the genes in order of evidence for differential expression. Higher B-value indicates higher probability of differential expression.
In order to investigate the magnitude of differential expression in each comparison the M-value (log2(sample X intensity/sample Y intensity)) for each gene was compared to the corresponding B-value (Figure 2). Genes with a high M-value usually receive a high B-value, which gives the plots the characteristic volcano shape. Figure 2 shows that the number of differentially expressed (DE) genes and the magnitude of differential expression is much lower for the technical and biological replicates than for the treated samples. This indicates that the RNA amplification and the biological fluctuations during culturing do not contribute substantially to the observed differential expression for PACAP and control treatments. Noteworthy, the distributions of B- and M-values are very similar for all treated samples, implicating that the magnitude of gene expression changes are similar for all treatments. The correlation between the replicated comparisons in Groups 2 and 3 was investigated by visualising the M-values from a comparison in Group 2 against the corresponding M-values in Group 3 (Figure 3). This shows that there is a high correlation between the replicates, with correlation coefficients ranging from 0.85 to 0.88. For the A-values (intensity values) the correlations are even higher ranging from 0.996 to 0.997 (data not shown). The M-value correlations between the contrasts were analysed in a similar fashion, but yielded much lower correlation coefficients (0.10 for PACAP treatment vs. proliferation control, 0.33 for differentiation control vs. proliferation control and 0.56 for differentiation control vs. PACAP treatment, data not shown). These low correlations are a consequence of the small differences between the contrasts as shown in the M-value distributions for the different comparisons (figure 4)
To further explore the overlap between DE genes in the replicated treatments Venn diagrams shown in Figure 5 were created. The comparisons include genes with a Holm adjusted p-value < 0.0001 and an M-value > +/- 0.6 (corresponding to a fold change > 1.5). Also included in the comparison are 29 DE genes (corresponding to 40 redundant probes on array) identified in the NS a-NS b comparison that could be considered as technical noise. Figure 5 shows that the overlap between the biological replicates is high. For all three treatments the majority of the genes (60–70%) are differentially expressed in both replicates; 814 genes (1109 probes) for the NS vs. PACAP treatment, 741 genes (986 probes) for the NS vs. proliferation control and 604 genes (797 probes) for the NS vs. differentiation control. Also, a large proportion of the non-overlapping genes showed a similar M-value in the two groups, indicating that the true overlap is even higher. The genes shared by both replicates for a certain treatment were further compared between the different treatments to visualise the effects of the different treatments on gene expression levels. Surprisingly, the great majority of the genes (435, 579 probes) fall within the overlap of all treatments, further suggesting that the different stimuli results in very similar effects on gene expression level.
A likely explanation is that the removal of growth factor (EGF) from the neurosphere culture medium, coinciding with the treatment initiation, masks the effect on gene expression changes caused by the different stimuli. To further investigate whether the remaining unique DE genes were true differences or related to the EGF effect we performed additional analysis of the 213 (265 probes), 135 (168 probes) and 79 (92 probes), non-overlapping genes. A comparison was made by taking the lists of the unique genes for one treatment and visualising their corresponding M-values in the other treatments (Figure 6). The results depicted in Figure 6 demonstrate that the majority of genes are clustered around the threshold values of the criteria for differential expression, either with an M-value just above or below 0.6 or with a p-value greater than 0.0001. Thus, the majority of unique (non-overlapping) genes are borderline cases, nearly included in the category of overlapping genes. Genes that would have been truly unique to the treatment in question would have had M-values in both replicates that were centred round zero. No such genes can be found when comparing the results from NS vs. PACAP treatment and NS vs. proliferation control. A few genes can be found when comparing NS vs. differentiation control to either one of the other two treatments, indicating that the serum treatment gives a somewhat more different gene expression profile, as expected. To facilitate further analysis of the results, annotated gene lists corresponding to the genes that were considered differentially expressed only in the NS vs. Diff Cont comparison, or shared between NS vs. PACAP and NS vs. Prol Cont comparisons are provided [see additional file 1 and additional file 2, respectively]. The complete results for all comparisons are available in ArrayExpress using experiment accession number E-MEXP-322.
These findings indicate that the differentially expressed genes in the different treatments are due to the withdrawal of EGF rather than to the treatment itself. A list of the 435 genes identified as the EGF treatment/withdrawal genes (see Figure 5) is provided [see additional file 3]. A short version of the list, with the top 40 genes, is shown in Table 1. The genes were further grouped and ranked according to their Gene Ontology annotation. We focused on the 'Biological Processes' branch of the GO theme structure and analysed the functional categories represented in the data. In total, 241 genes received a functional annotation. The results of the analysis, which was carried out at the detail level 3 (intermediate level that gives a general overview of the data), are provided [see additional file 4]. The themes with the highest representation were 'cell growth and/or maintenance' (118 genes), 'nucleobase, nucleoside, nucleotide and nucleic acid metabolism' (65), 'protein metabolism' (46), 'signal transduction' (29), 'catabolism' (20) and 'organogenesis' (20). In general, the list of expressed functional categories is enriched for various metabolism-related themes, but further down also contains themes such as 'cell adhesion', 'cell death', 'response to external stimulus' and 'cell-cell signalling' Next we analysed the overrepresentation of functional categories by using the genes represented on the array as background for the significance calculations. The top GO terms of the biological process class overrepresented in the data are shown in Table 2. A large proportion of these GO themes are related to the cell cycle and/or DNA replication, as expected for EGF-related effects. Many of the transcripts found and classified within "mitotic cell cycle" and "DNA replication and chromosome cycle" are down-regulated in the treated samples lacking EGF. In contrast many genes within "neurogenesis", "organogenesis" and "development" are up-regulated in these samples, probably reflecting the removal of inhibitory regulation on differentiation as EGF is withdrawn from the culture medium.
Discussion
We have recently performed a pilot study investigating both the amplification technology and various neurosphere isolation and culture conditions [21]. In that study we could show a low technical variability in the microarray analysis, large transcriptional differences between passages of neurospheres and a smaller differences between parallel cultures. The variability is addressed in this study by increasing the number of replicates, using parallel cultures to improve the statistical power in identification of differentially expressed genes. The observed differential gene expression (Figure 2) is low for the culture and technical replicates, confirming previous findings, and has a minor impact on the results obtained in the three treatment comparisons.
The primary goal in this study was to identify the genes activated by PACAP treatment. PACAP acts as a neurotransmitter or neuromodulator in the brain, and regulates the secretion of certain neurohormones. PACAP also acts as a neurotrophic factor during brain development and as a neuroprotective agent in the adult brain and acts via three G-protein coupled receptors, PAC1 and the vasoactive intestinal peptide receptors VIP1 and VIP2. In the recent study [18] it was shown that PAC1 is expressed in the neurogenic regions of the adult mouse brain and in neurospheres generated from the lateral ventricular wall in the adult mouse. Importantly it was also shown that PACAP stimulates proliferation of NSC in the brain and of cultured neurospheres, and a number of new proliferation targets were sought to be discovered by the array analysis of neurospheres. The previous study [18] also demonstrated a synergistic proliferative effect of the combined stimulation of EGF and PACAP. It has previously been described that PAC1 signalling shows cross-talk to the signalling of receptor tyrosine kinase cascades, such as that elicited by nerve growth factor (NGF), suggesting that PACAP may act in concert with other molecules to maximize or regulate their effect [27]. In order to investigate the PACAP effect without EGF interference, we chose to culture the neurospheres for three days in the absence of EGF, but presence of PACAP.
However, the obtained results could not demonstrate a PACAP-specific transcriptional signature. Instead, by comparing the differentially expressed genes from PACAP treatment with control experiments representing proliferation or differentiation activation we could conclude that the removal of EGF was the main contributor to the observed differential expression. We demonstrate this by the large overlap in the Venn diagram of DE genes and we could also conclude that the group of "unique" (non-overlapping) genes were not truly unique by demonstrating that these would have been shared among the different treatments if slightly less stringent criteria had been used in the definition of differential expression. So why does EGF removal have such a strong effect in comparison to the addition of other factors, such as PACAP? EGF is an extremely powerful mitogen for adult mouse neural stem cells, exhibiting a four-fold greater proliferative potency than PACAP at 1 nM and 100 nM, respectively [18]. Such a profound mitogenic effect as that elicited by EGF is likely to require large alterations in levels of gene transcription and consequently the removal of this agent necessitates large reversals. PACAP, being a substantially less powerful proliferative agent, may well trigger less dramatic transcriptional changes, be it on genes in common (such as through the EGF pathway system) or not to those utilized by EGF to promote mitogenesis. In essence, it appears that transcriptional fluctuations induced by the presence and absence of EGF dwarf that elicited by PACAP, and indeed the TMR agonist and even differentiating factors.
Consequently, the shared DE genes reflect the removal of the mitogenic signal of EGF. Gene annotation analysis also supports the conclusions by demonstrating a down-regulation of proliferation signals and up-regulation of differentiation signals in the treated neurosphere cultures. Some of the genes observed in the differentiation control are however unique, which probably reflects a broader activation spectrum induced by fetal calf serum and solid support plating, as compared to EGF withdrawal. Previous studies of neurospheres using microarrays have also used culture conditions with withdrawal of growth factors to study differentiation [28-30]. These showed decreased expression of cell cycle related genes as well as increased expression of differentiation markers as a response to the respective treatments.
Conclusion
In conclusion, we report that using the current study design, EGF removal in neurosphere culture medium creates too large transcriptional changes to be able to identify other parallel transcriptional events relating to proliferation by PACAP. This requires future development of modified culture conditions and assay designs to enable monitoring and proof of these transcriptional alterations.
Methods
Adult mouse neural stem cell culture
Adult mouse neural stem cell cultures were initiated originating from tissue isolated from 15 mice using identical dissection, dissociation and culture protocols. Briefly, the lateral wall of the lateral ventricle of 6 week-old mice was enzymatically dissociated in 0.8 mg/ml hyaluronidase and 0.5 mg/ml trypsin in Dulbecco's modified Eagle medium (DMEM) containing 4.5 mg/ml glucose and 80 U/ml DNase at 37°C for 20 min. The cells were gently triturated and mixed with three volumes of neurosphere medium (DMEM/F12, B27 supplement, 12.5 mM HEPES pH7.4) containing 20 ng/ml EGF, 100 U/ml penicillin and 100 μg/ml streptomycin. After passing through a 70-μm strainer, the cells were pelleted at 160 × g for 5 min. The supernatant was subsequently removed and the cells resuspended in neurosphere medium supplemented as above, plated in uncoated culture dishes and incubated at 37°C. Neurospheres were ready to be split 7–8 days after plating.
To split neurosphere cultures, neurospheres were collected by centrifugation at 160 × g for 5 min. The neurospheres were resuspended in 0.5 ml Trypsin/EDTA in HBSS (1×), incubated at 37°C for 2 min and triturated gently to aid dissociation. Following a further three-min incubation at 37°C and trituration, 3 volumes of ice-cold neurosphere medium containing EGF were added. The cells were pelleted at 220 × g for 4 min, resuspended in fresh neurosphere medium supplemented with 3 nM EGF.
Dissociated cells were plated and grown in neurosphere medium supplemented with EGF for a further 3–4 days by which time secondary neurospheres had developed. Secondary neurospheres were dissociated and divided into a 4 duplicated fractions. The culture control, proliferation control and PACAP-treated cultures were grown in neurosphere medium supplemented with 3 nM EGF, 1 nM transmembrane receptor agonist and 100 nM PACAP, respectively, for 3 days, and harvested for mRNA isolation. The differentiation control was replated in neurosphere medium supplemented with 1% fetal calf serum (FCS) onto poly-D-lysine plates to which the cells adhered. After incubating, overnight FCS concentration was reduced to 0.5%, and the cells cultured a further 2 days before centrifugation and subsequent mRNA isolation. All experiments were approved by the Karolinska Institute Ethical Committee.
cDNA synthesis
Messenger RNA was isolated using Dynabeads® mRNA DIRECT™ Kit from Dynal (Dynal A.S., Norway). First and RNaseH-dependent second-strand cDNA synthesis (SuperScript Choice System for cDNA Synthesis) was performed according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA) using 45 pmol biotinylated NotI-oligo(dT) primer (5'-biotin-GAGGTGCCAACCGCGGCCGC (T)15-3'). The cDNA was phenol-chloroform extracted and ethanol precipitated and the pellet was dissolved in 40 μl of 1× TE (10 mM Tris-HCl, 1 mM EDTA). Excess NotI-oligo(dT) primer was removed by Chromaspinn TE-100 column (Clontech, CA, USA) purification.
Amplification of 3'-end signature tags
The cDNA was fragmented and amplified according to a protocol previously described [22,23]. Shortly, fragmentation of the cDNA was performed in 40 μl 1× TE using an inverted sonication probe, using 16 × 10 s pulses at 90% effect (Sonifier® B-12, Branson Sonic Power Company, CT, USA). Biotinylated 3'-end signature tags from the fragmented cDNA population were isolated onto 20 μl of paramagnetic streptavidin-coated beads (10 mg/ml) (Dynal A.S.) in 40 μl sample plus 40 μl Binding/Washing buffer (2 M NaCl, 0.1% Tween-20 in 1× TE, pH 7.7) at 37°C for one hour with rotation. The immobilised signature tags were end repaired using 1.5 U T4 DNA polymerase (New England BioLabs, MA, USA) in a 30-μl reaction volume at 12°C for 20 minutes according to the supplier's recommendations. Blunt-end adapters (Sima18: 5'-GGATCCGCGGTG-3'; Sima19: 5'-TCTCCAGCCTCTCACCGCGGATCC-3') were pre-annealed and ligated onto the immobilised repaired 3'-end signature tags using a solution comprising 1.1 nmol adapter, ligase buffer (66 mM Tris-HCl, pH 7.6, 5 mM MgCl2, 5 mM DTT, 50 μg/ml BSA), 0.2 mM ATP, 1200 U T4 DNA ligase (New England BioLabs) in a final volume of 60 μl. Ligation was performed overnight at room temperature with constant rotation to keep beads in suspension. The signature tags were released from the magnetic beads by restriction with NotI (New England BioLabs) for 2 hours in a volume of 60 μl while keeping the beads in suspension. Five microlitres of the eluate containing the 3'-end signature tags was used as template in a subsequent PCR. The PCR was performed in 100 μl containing 200 μM of each dNTP, 0.75 μM Sima19, 0.75 μM NotI-oligo(dT) primer, 65 mM Tris-HCl, pH 8.8, 4 mM MgCl2, 16 mM (NH4)2SO4, 0.5 μM BSA and 3 U AmpliTaq DNA polymerase (Perkin Elmer, Boston, USA). Cycling was performed according to the following procedure, initial incubation at 72°C for 3 min, followed by addition of Taq DNA polymerase and subsequent cycling: 72°C for 20 min, 95°C for 1 min, 45°C for 5 min, 72°C for 15 min, followed by four cycles (95°C for 1 min, 50°C for 1 min, 72°C for 15 min), and 13 cycles (as previously optimised) (95°C for 1 min, 50°C for 1 min, 72°C for 2 min).
Target labelling and microarray hybridisation
The 3'-end signature tags were purified using QIAquick® PCR purification kit (Qiagen, Germany). Direct labelling was performed using Cy3-dCTP or Cy5-dCTP (Perkin Elmer, Boston, USA) in a linear, asymmetric PCR. The reaction was performed in a 50-μl labelling mix containing 100–200 ng purified 3'-end signature tags, 80 μM dATP, dGTP and dTTP, 20 μM dCTP, 5 μM Sima 19 primer, 2 mM MgCl2, 1 × PCR Buffer II (Applied Biosystems, Ca, USA), 3 U AmpliTaq Gold® (Applied Biosystems) and 60 pM Cy3-dCTP or Cy5-dCTP. The labelling mix was cycled as follows: 95°C for 12 min, then 20 cycles (95°C for 30 s, 50°C for 30 s, 72°C for 10 min). Excess primer and nucleotides were removed using QIAquick® PCR purification kit. The eluted labelling products were speed vacuumed until dry, then dissolved in 55 μl hybridisation buffer (24% formamide, 5× SSC and 0.1% SDS). Cy-3 and Cy-5 labellings were blended and mixed with 25 μg mouse Cot-1 DNA (Invitrogen) and 50 μg polyA DNA (Operon Biotechnologies GmbH, Germany). The microarray contained 14121 probes printed in duplicate and originating mainly from a lateral ventricle wall (Unigene Library ID 16789), a neurosphere (Lib. 16808) and a hematopoietic cell-line cDNA (Lib. 16809) library. Details regarding the array manufacturing are available through ArrayExpress (accession number A-MEXP-175). Briefly, probes were generated through PCR amplification and subsequent Multiscreen-384 filter plates (Millipore) purification of the library clones. Purified products in 50% DMSO were printed onto ultraGAPS slides (Corning Inc) using the QArray arrayer (Genetix) and covalently attached to the slide surface using 250 mJ UV-light (Stratalinker). The arrays were first prehybridised for 30 min in a 42°C prehybridisation solution (1% BSA, 5× SSC, 0.1% SDS), then washed in water and isopropanol and dried through centrifugation. The sample was denatured in 95°C for 3 min, applied to the array and incubated in a hybridisation chamber at 42°C for 18 hours. After hybridisation the arrays were washed in three successive wash buffers with increasing stringency: (1) 1× SSC and 0.2% SDS, 42°C, (2) 0.1× SSC and 0.2% SDS, room temperature, (3) 0.1× SSC, room temperature. All wash steps were made on a shaking table for 4 min. After the last step the arrays were immediately centrifuged in a slide centrifuge and kept in the dark until scanned with the DNA Microarray scanner G2565BA (Agilent Technologies, CA, USA).
Image and data analysis
All image and data analysis was conducted in GenePix Pro 5.1 (Axon Instruments Inc, CA, USA) and the R environment for statistical computing and programming [31] using packages aroma [32], Bioconductor [33], limma [25] and kth [34]. The analysis was conducted according to the following workflow. (1) Spot identity and foreground/background intensities were extracted from the tiff files using the irregular feature-finding algorithm implemented in GenePix Pro 5.1. (2) GenePix results files were imported into R and the median of the foreground signal was used as expression measurements without subtracting the local background signal. (3) Flagged features (either automaticly by GenePix or manually by the user), too small or large features (<51 and >250 μm, respectively), saturated features (both channels > 65100 intensity units) or features for which the signal-to-noise ratio (as defined by GenePix 5.1) was < 3, were removed from further analysis. (4) Remaining data was normalised separately for each block on the slide using the robust intensity-dependent print-tip lowess normalisation approach [35]. (5) Differentially expressed genes were identified using an empirical Bayes moderated t-test [24-26] and ranked in order of evidence for differential expression. The p-values associated with the t-test were adjusted for multiple testing by using the Holm's approach [36]. Genes with an adjusted p-value < 0.0001 were considered differentially expressed. Genes with very small fold-changes (< 1.5 fold) were excluded by using an additional M-value cut-off (> +/- 0.6). ESTs on the array were mapped against build 144 of Mus musculus Unigene. Over-representation analysis of functional cateogories or themes was carried out using the EASE software [37]. Themes with an EASE-score < 0.05 were considered overrepresented. The complete set of raw data and transformed data is available through ArrayExpress (experiment accession number E-MEXP-322).
List of abbreviations
NSC; neural stem cell
SVZ; subventricular zone
DE; differentially expressed
NS; neurosphere
DC; differentiated cells
PACAP; pituitary adenylate cyclase-activating polypeptide
EGF; epidermal growth factor
Authors' contributions
MS participated in the design of the study, drafted the manuscript, coordinated and carried out microarray experiments as well as performed data processing and data analysis. VW coordinated and carried out the manufacturing of the microarrays, performed data analysis and statistical analysis as well as assisted with the manuscript. AM participated in the design of the study, cultured cells and assisted with the manuscript. JF participated in its design and coordination and assisted with the manuscript. JL conceived of the study, participated in its design and coordination of the study and helped to draft the manuscript, principle investigator. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Differentially expressed genes in the NS vs. Diff Cont comparison only. Differentially expressed genes (with an M-value > | 0.6 | and a p-value < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing) in the NS vs. Differentiation Control only, corresponding to the 79 genes in Fig 5. To facilitate comparisons the M- and Holm adjusted p-values for all treatment replicates are provided. Genes that have positive M-values are up-regulated in the treated (EGF withdrawn) samples, compared to the undifferentiated neurospheres (NS). Genes that have negative M-values are up-regulated in the NS sample.
Click here for file
Additional File 2
Differentially expressed genes in the NS vs. Proliferation samples only. Differentially expressed genes (with an M-value > | 0.6 | and a p-value < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing) in the overlap between the NS vs. PACAP and NS vs. Prol Cont comparisons, corresponding to the 151 genes in Fig 5. To facilitate comparisons the M- and Holm adjusted p-values for all treatment replicates are provided. Genes that have positive M-values are up-regulated in the treated (EGF withdrawn) samples, compared to the undifferentiated neurospheres (NS). Genes that have negative M-values are up-regulated in the NS sample.
Click here for file
Additional File 3
Differentially expressed genes in all treatments. Differentially expressed genes in the overlap between all treatments (NS vs. PACAP, NS vs. Prol Cont and NS vs. Diff Cont). Genes that have positive M-values are up-regulated in the treated (EGF withdrawn) samples, compared to the undifferentiated neurospheres (NS). Genes that have negative M-values are up-regulated in the NS sample. Provided are the M-, B- and p-values for all treatment replicates, as well as the average M- and B-values and their standard deviations. All genes have an M-value > | 0.6 | and a p-value < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing, in all treatments.
Click here for file
Additional File 4
GO theme representation among the EGF treatment/withdrawal genes. Representation of gene ontology themes among the genes that were differentially expressed in all treatments. Analysis was restricted to themes of category biological process. Genes with M-value > | 0.6 | and p < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing, are included.
Click here for file
Acknowledgements
We thank Anna Westring, Lilian Wikström, Peter Nilsson and Cecilia Williams for valuable assistance and comments. This work was supported by grants from the Knut and Alice Wallenberg Foundation, the Wallenberg Consortium North, the Swedish Cancer Foundation and the Swedish Scientific Research Council.
Figures and Tables
Figure 1 Neurosphere culturing and experimental design. Neural stem/progenitor cells were isolated from the lateral ventricle wall region of brains from a pool of mice and grown as neurospheres. RNA was isolated from different treatments as indicated and used for subsequent microarray hybridisations. a and b indicate biological replicates. Each arrow represents the number of hybridisations, arrowhead represents labelling with Cy5 and arrow tail represents labelling with Cy3. Groups 1, 2 and 3 indicate hybridisations grouped in the data analysis, to optimise the variance estimates for each gene. NS = neurosphere control, PACAP = pituitary adenylate cyclase-activating polypeptide treated samples, Prol Cont = proliferation control (transmembrane receptor agonist treated samples), Diff Cont = differentiation control (fetal calf serum treated, and solid support plated samples).
Figure 2 B-value distributions for each comparison. The x-axis shows the M-value (log2(Cy5/Cy3)) for each gene and the y-axis the corresponding B-value (calculated by an empirical Bayes moderated t-test).The B-value scores the genes according to their probability of differential expression. Higher B-value means higher probability of differential expression. Dotted lines are drawn at M-values 0.6 and -0.6, i.e. at a 1.5-fold difference in signal intensity between the compared samples, and at B = 9.3, corresponding to a Holm adjusted p-value of 0.0001. These values correspond to the thresholds set for differential expression in this study.
Figure 3 Graphs displaying the correlation between replicated samples. The x-axis and y-axis display M-values (log2(Cy5/Cy3)) for replicated samples. The values of the Pearson correlation coefficient (r) and the coefficient of determination (R2) are also included.
Figure 4 Box plots displaying the M-value (log2(Cy5/Cy3)) distribution for each comparison.
Figure 5 Correlation between biological replicates and between treatments. The number of differentially expressed genes in each comparison are presented and compared. Genes with an M-value > | 0.6 | and a p-value < 0.0001, calculated by empirical Bayes moderated t-test and a Holm's adjustment for multiple testing are included. Figures without parentheses show the number of probes included. Figures within parentheses show the number of corresponding genes. NS = neurosphere control, PACAP = pituitary adenylate cyclase-activating polypeptide treated samples, Prol Cont = proliferation control (transmembrane receptor agonist treated samples), Diff Cont = differentiation control (fetal calf serum treated, and solid support plated samples).
Figure 6 M-value analysis of non-overlaping genes from the treatment comparisons. M-values (log2(Cy5/Cy3)) for genes identified as DE in either only neurosphere vs. PACAP treatment (A), neurosphere vs. proliferation control (B) or neurosphere vs. differentation control (C) are shown using data from the other two treatment comparisons. Colouring indicates if a gene reaches the statistical significance required for differential expression (black, both replicates have p-values < 0.0001; red, p-value in replicate a > 0.0001: green, p-value in replicate b > 0.0001). Dotted lines depict the cut-off values for scoring differential expression (M-value > | 0.6 |).
Table 1 Top differentially expressed genes in the overlap between all treatments.
Unigene ID Gene ID GeneName Avg M Stdev M Avg B Stdev B
Mm.305535 69719 Carbamoyl-phosphate synthetase 2 4.53 0.26 53.33 4.34
Mm.1239 14580 Glial fibrillary acidic protein 4.47 0.39 69.36 3.97
Mm.252063 17196 Myelin basic protein 3.84 0.57 47.80 7.00
Mm.200608 12759 Clusterin 3.72 0.54 60.76 7.92
Mm.240850 76960 Breast carcinoma amplified sequence 1 3.72 0.17 46.89 1.30
Mm.291129 70877 RIKEN cDNA 4921517J23 gene 3.68 0.44 49.64 6.06
Mm.1268 18823 Proteolipid protein (myelin) 1 3.64 0.49 51.72 4.02
Mm.305152 11816 Apolipoprotein E 3.62 0.46 45.79 11.39
Mm.334994 53867 Procollagen. type V. alpha 3 3.61 0.45 49.05 8.48
Mm.272443 64383 Sirtuin 2 3.33 0.34 72.98 3.98
Mm.7420 22153 Tubulin. beta 4 3.13 0.38 53.65 4.65
Mm.39053 71770 Adaptor-related protein complex 2. beta 1 subunit 2.92 0.61 32.53 10.40
Mm.228362 NA Transcribed locus 2.69 0.15 39.82 3.13
Mm.358573 11858 Ras homolog gene family. member N 2.68 0.39 52.12 7.88
Mm.276739 20665 SRY-box containing gene 10 2.65 0.34 33.25 1.34
Mm.241355 17136 Myelin-associated glycoprotein 2.57 0.59 32.28 6.80
Mm.278672 14810 Glutamate receptor. ionotropic. NMDA1 (zeta 1) 2.56 0.52 35.15 7.11
Mm.260601 242521 Kelch-like 9 (Drosophila) 2.54 0.41 44.19 9.26
Mm.192991 17748 Metallothionein 1 2.47 0.54 21.61 5.08
Mm.37199 14862 Glutathione S-transferase. mu 1 2.44 0.16 38.61 2.15
Mm.41580 69908 RAB3B. member RAS oncogene family -3.55 0.31 47.83 4.75
Mm.290563 12615 Centromere autoantigen A -3.14 0.16 49.02 3.58
NA NA GenBank ID: CX204578 -3.09 0.34 34.48 2.55
NA NA GenBank ID: CX207566 -2.99 0.27 51.46 6.04
Mm.273049 12443 Cyclin D1 -2.89 0.33 44.78 3.46
Mm.289747 107995 Cell division cycle 20 homolog (S. cerevisiae) -2.68 0.15 55.41 5.67
Mm.285723 14793 Cell division cycle associated 3 -2.65 0.08 42.97 0.95
Mm.24337 52033 PDZ binding kinase -2.58 0.12 40.31 2.70
Mm.29254 16009 Insulin-like growth factor binding protein 3 -2.49 0.29 41.37 3.97
Mm.233799 16010 Insulin-like growth factor binding protein 4 -2.46 0.31 23.86 5.98
Mm.16525 18817 Polo-like kinase 1 (Drosophila) -2.43 0.12 26.12 1.92
Mm.28038 52276 Cell division cycle associated 8 -2.39 0.12 30.79 5.44
Mm.4857 12323 Calcium/calmodulin-dependent protein kinase II. beta -2.38 0.49 42.14 5.06
Mm.335368 NA CDNA. clone:Y2G0133B03:ENSEMBL -2.35 0.25 50.92 4.35
NA NA GenBank ID: CX206925 -2.32 0.13 28.39 1.66
Mm.271711 21346 Transgelin 2 -2.25 0.12 38.33 1.70
Mm.4237 21973 Topoisomerase (DNA) II alpha -2.14 0.12 46.22 2.01
Mm.1408 11535 Adrenomedullin -2.07 0.17 26.68 5.70
Mm.37802 69544 WD repeat domain 5B -2.03 0.10 59.04 2.30
Mm.35389 13063 Cytochrome c. somatic -2.02 0.12 51.69 2.15
The top of the table shows genes that are up-regulated in the treated (EGF withdrawn) samples, compared to the undifferentiated neurospheres (NS) (positive M-values). The lower half of the table shows genes that are up-regulated in the NS sample (negative M-values). Average M- and B-values for all comparisons (NS vs. PACAP, NS vs. Prol Cont and NS vs. Diff Cont) are shown, together with their standard deviations. All genes have a p-value < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing.
Table 2 Gene ontology analysis.
Biological process enriched in neurosphere (NS) control sample No of DE genes in all treatments No of genes on array Fisher Exact
mitotic cell cycle 35 136 3.840E-22
DNA replication and chromosome cycle 24 72 3.320E-18
cell cycle 43 307 2.360E-16
cell proliferation 46 362 7.620E-16
DNA replication 20 61 3.770E-15
S phase of mitotic cell cycle 20 62 5.400E-15
DNA-dependent DNA replication 11 24 1.160E-10
DNA metabolism 29 218 1.830E-10
DNA replication initiation 7 8 4.430E-10
mitosis 12 48 5.240E-08
M phase of mitotic cell cycle 12 48 5.240E-08
cell growth and/or maintenance 79 1411 2.740E-07
nuclear division 12 58 4.950E-07
M phase 12 61 8.830E-07
cellular physiological process 82 1564 2.770E-06
cytokinesis 11 63 8.760E-06
nucleosome assembly 6 21 5.920E-05
regulation of cell cycle 15 138 7.790E-05
physiological process 135 3454 6.310E-04
DNA packaging 9 80 1.750E-03
Biological process enriched in treated (PACAP. Prol Cont and Diff Cont) samples No of DE genes in all treatments No of genes on array Fisher Exact
microtubule-based process 7 64 1.340E-03
development 22 456 3.400E-03
neurogenesis 8 99 4.300E-03
organogenesis 14 243 4.390E-03
morphogenesis 15 282 6.730E-03
cytoskeleton organization and biogenesis 9 134 8.430E-03
organelle organization and biogenesis 10 175 1.650E-02
cellular process 65 2144 6.560E-02
cytoplasm organization and biogenesis 10 222 6.720E-02
lipid metabolism 8 173 8.570E-02
membrane lipid metabolism 3 42 9.840E-02
cell adhesion 7 162 1.360E-01
cell communication 27 835 1.440E-01
cell organization and biogenesis 12 331 1.620E-01
regulation of biological process 5 110 1.650E-01
ion transport 7 188 2.280E-01
cellular physiological process 45 1564 2.640E-01
cation transport 5 135 2.850E-01
cell death 5 144 3.330E-01
cell surface receptor linked signal transduction 9 286 3.440E-01
The most highly overrepresented gene ontology themes among the genes that were differentially expressed in all treatments. Analysis was restricted to themes of category biological process. Genes with M-value > | 0.6 | and p < 0.0001, calculated by empirical Bayes moderated t-test and a Holm adjustment for multiple testing, are included. Jacknife Fisher's exact probability test was used to identify over-represented themes in the data. Corresponding p-values are listed.
==== Refs
Seki T Arai Y Highly polysialylated neural cell adhesion molecule (NCAM-H) is expressed by newly generated granule cells in the dentate gyrus of the adult rat J Neurosci 1993 13 2351 2358 7684771
Corotto FS Henegar JA Maruniak JA Neurogenesis persists in the subependymal layer of the adult mouse brain Neurosci Lett 1993 149 111 114 8474679 10.1016/0304-3940(93)90748-A
Luskin MB Restricted proliferation and migration of postnatally generated neurons derived from the forebrain subventricular zone Neuron 1993 11 173 189 8338665 10.1016/0896-6273(93)90281-U
Gritti A Parati EA Cova L Frolichsthal P Galli R Wanke E Faravelli L Morassutti DJ Roisen F Nickel DD Vescovi AL Multipotential stem cells from the adult mouse brain proliferate and self-renew in response to basic fibroblast growth factor J Neurosci 1996 16 1091 1100 8558238
Reynolds BA Weiss S Generation of neurons and astrocytes from isolated cells of the adult mammalian central nervous system Science 1992 255 1707 1710 1553558
Seaberg RM van der Kooy D Adult rodent neurogenic regions: the ventricular subependyma contains neural stem cells, but the dentate gyrus contains restricted progenitors J Neurosci 2002 22 1784 1793 11880507
Lindvall O Kokaia Z Martinez-Serrano A Stem cell therapy for human neurodegenerative disorders-how to make it work Nat Med 2004 10 Suppl S42 50 15272269 10.1038/nm1064
Rossi F Cattaneo E Opinion: neural stem cell therapy for neurological diseases: dreams and reality Nat Rev Neurosci 2002 3 401 409 11988779 10.1038/nrn809
Horner PJ Gage FH Regenerating the damaged central nervous system Nature 2000 407 963 970 11069169 10.1038/35039559
Bachoud-Levi AC Remy P Nguyen JP Brugieres P Lefaucheur JP Bourdet C Baudic S Gaura V Maison P Haddad B Boisse MF Grandmougin T Jeny R Bartolomeo P Dalla Barba G Degos JD Lisovoski F Ergis AM Pailhous E Cesaro P Hantraye P Peschanski M Motor and cognitive improvements in patients with Huntington's disease after neural transplantation Lancet 2000 356 1975 1979 11130527 10.1016/S0140-6736(00)03310-9
Kordower JH Freeman TB Snow BJ Vingerhoets FJ Mufson EJ Sanberg PR Hauser RA Smith DA Nauert GM Perl DP Olanow CW Neuropathological evidence of graft survival and striatal reinnervation after the transplantation of fetal mesencephalic tissue in a patient with Parkinson's disease N Engl J Med 1995 332 1118 1124 7700284 10.1056/NEJM199504273321702
Bjorklund A Lindvall O Cell replacement therapies for central nervous system disorders Nat Neurosci 2000 3 537 544 10816308 10.1038/75705
Nakatomi H Kuriu T Okabe S Yamamoto S Hatano O Kawahara N Tamura A Kirino T Nakafuku M Regeneration of hippocampal pyramidal neurons after ischemic brain injury by recruitment of endogenous neural progenitors Cell 2002 110 429 441 12202033 10.1016/S0092-8674(02)00862-0
Sun Y Jin K Xie L Childs J Mao XO Logvinova A Greenberg DA VEGF-induced neuroprotection, neurogenesis, and angiogenesis after focal cerebral ischemia J Clin Invest 2003 111 1843 1851 12813020 10.1172/JCI200317977
Studer L Tabar V McKay RD Transplantation of expanded mesencephalic precursors leads to recovery in parkinsonian rats Nat Neurosci 1998 1 290 295 10195162 10.1038/2774
Svendsen CN Caldwell MA Shen J ter Borg MG Rosser AE Tyers P Karmiol S Dunnett SB Long-term survival of human central nervous system progenitor cells transplanted into a rat model of Parkinson's disease Exp Neurol 1997 148 135 146 9398456 10.1006/exnr.1997.6634
Picard-Riera N Nait-Oumesmar B Baron-Van Evercooren A Endogenous adult neural stem cells: limits and potential to repair the injured central nervous system J Neurosci Res 2004 76 223 231 15048920 10.1002/jnr.20040
Mercer A Ronnholm H Holmberg J Lundh H Heidrich J Zachrisson O Ossoinak A Frisen J Patrone C PACAP promotes neural stem cell proliferation in adult mouse brain J Neurosci Res 2004 76 205 215 15048918 10.1002/jnr.20038
Arimura A Perspectives on pituitary adenylate cyclase activating polypeptide (PACAP) in the neuroendocrine, endocrine, and nervous systems Jpn J Physiol 1998 48 301 331 9852340 10.2170/jjphysiol.48.301
Vaudry D Gonzalez BJ Basille M Yon L Fournier A Vaudry H Pituitary adenylate cyclase-activating polypeptide and its receptors: from structure to functions Pharmacol Rev 2000 52 269 324 10835102
Sievertzon M Wirta V Mercer A Meletis K Erlandsson R Wikstrom L Frisen J Lundeberg J Transcriptome analysis in primary neural stem cells using a tag cDNA amplification method BMC Neurosci 2005 6 28 15833137 10.1186/1471-2202-6-28
Hertzberg M Sievertzon M Aspeborg H Nilsson P Sandberg G Lundeberg J cDNA microarray analysis of small plant tissue samples using a cDNA tag target amplification protocol Plant J 2001 25 585 591 11309148 10.1046/j.1365-313x.2001.00972.x
Sievertzon M Agaton C Nilsson P Lundeberg J Amplification of mRNA populations by a cDNA tag strategy Biotechniques 2004 36 253 259 14989090
Lönnstedt I Speed T Replicated microarray data Statistica Sinica 2002 12 31 46
Smyth G Linear models and empirical Bayes methods for assessing differential expression in microarray experiments Statistical Applications in Genetics and Molecular Biology 2004 3 Article 3
Smyth GK Michaud J Scott H The use of within-array duplicate spots for assessing differential expression in microarray experiments Submitted 2003
Vaudry D Stork PJ Lazarovici P Eiden LE Signaling pathways for PC12 cell differentiation: making the right connections Science 2002 296 1648 1649 12040181 10.1126/science.1071552
Geschwind DH Ou J Easterday MC Dougherty JD Jackson RL Chen Z Antoine H Terskikh A Weissman IL Nelson SF Kornblum HI A genetic analysis of neural progenitor differentiation Neuron 2001 29 325 339 11239426 10.1016/S0896-6273(01)00209-4
Karsten SL Kudo LC Jackson R Sabatti C Kornblum HI Geschwind DH Global analysis of gene expression in neural progenitors reveals specific cell-cycle, signaling, and metabolic networks Dev Biol 2003 261 165 182 12941627 10.1016/S0012-1606(03)00274-4
Gurok U Steinhoff C Lipkowitz B Ropers HH Scharff C Nuber UA Gene expression changes in the course of neural progenitor cell differentiation J Neurosci 2004 24 5982 6002 15229246 10.1523/JNEUROSCI.0809-04.2004
R Development Core Team R: A language and environment for statistical computing
Bengtsson H aroma - An R Object-oriented Microarray Analysis environment
Gentleman R Carey V Bates D Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Hornik K Hothorn T Huber W Iacus S Irizarry R Leisch F Li C Maechler M Rossini A Sawitzki G Smith C Smyth G Tierney L Yang J Zhang J Bioconductor: open software development for computational biology and bioinformatics Genome Biology 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Wirta V Lindberg J Gry Björklund M The kth-package for microarray data analysis 2004
Yang YH Dudoit S Luu P Lin DM Peng V Ngai J Speed TP Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation Nucleic Acids Res 2002 30 e15 11842121 10.1093/nar/30.4.e15
Holm S A simple sequentially rejective multiple test procedure Scandinavian Journal of Statistics 1979 6 65 70
Hosack DA Dennis GJ Sherman BT Lane HC Lempicki RA Identifying biological themes within lists of genes with EASE Genome Biol 2003 4 R70 14519205 10.1186/gb-2003-4-10-r70
|
16124881
|
PMC1208901
|
CC BY
|
2021-01-04 16:03:48
|
no
|
BMC Neurosci. 2005 Aug 28; 6:55
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-55
|
oa_comm
|
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-181609295410.1186/1471-2415-5-18Technical AdvanceThe macular mapping test: a reliability study Bartlett Hannah [email protected] Leon N [email protected] Frank [email protected] Ophthalmic and Physiological Optics Research Group, Aston University, Birmingham, UK2005 10 8 2005 5 18 18 18 3 2005 10 8 2005 Copyright © 2005 Bartlett et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Age-related macular degeneration (ARMD) is the leading cause of visual disability in people over 60 years of age in the developed world. The success of treatment deteriorates with increased latency of diagnosis. The purpose of this study was to determine the reliability of the macular mapping test (MMT), and to investigate its potential as a screening tool.
Methods
The study population comprised of 31 healthy eyes of 31 participants. To assess reliability, four macular mapping test (MMT) measurements were taken in two sessions separated by one hour by two practitioners, with reversal of order in the second session. MMT readings were also taken from 17 age-related maculopathy (ARM), and 12 AMD affected eyes.
Results
For the normal cohort, average MMT scores ranged from 85.5 to 100.0 MMT points. Scores ranged from 79.0 to 99.0 for the ARM group and from 9.0 to 92.0 for the AMD group. MMT scores were reliable to within ± 7.0 points. The difference between AMD affected eyes and controls (z = 3.761, p = < 0.001) was significant. The difference between ARM affected eyes and controls was not significant (z = -0.216, p = 0.829).
Conclusion
The reliability data shows that a change of 14 points or more is required to indicate a clinically significant change. This value is required for use of the MMT as an outcome measure in clinical trials. Although there was no difference between MMT scores from ARM affected eyes and controls, the MMT has the advantage over the Amsler grid in that it uses a letter target, has a peripheral fixation aid, and it provides a numerical score. This score could be beneficial in office and home monitoring of AMD progression, as well as an outcome measure in clinical research.
==== Body
Background
Age-related macular degeneration (AMD) is the advanced from of age-related macular disease (ARMD) and is the leading cause of visual disability in patients over the age of 60 years in the developed World [1]. In the United States, the demographic right shift is expected to cause an increase in the number of people affected by this condition from 1.75 million to almost 3 million by 2020 [2].
Exudative AMD and/or geographic atrophy (GA) can result in the severe visual loss, and their prevalence in the US population over 40 years of age has been estimated at 1.47% [95% confidence interval (CI), 1.38% – 1.55%] [2]. The likelihood of visual deterioration in those with exudative AMD may be reduced with laser photocoagulation and photodynamic therapy [3-6], but the success rate deteriorates with increasing latency of diagnosis as lesions extend towards the foveal avascular zone. Consequently, early diagnosis is crucial, and at risk patients are often advised to use an Amsler grid at home for detection of scotomas and metamorphopsia [7-9].
The Amsler grid has been used in clinical practice since the 1940s [10] for detecting and monitoring retinal conditions such as age-related macular disease [11-13]. It has been shown, however, that a report of distortion of the Amsler grid can come from perceived lines filling-in across scotomas, or equally from non-scotomatous retinal impairments; the clinician has no way of knowing which. Seventy-seven percent of standard and 87 % of threshold scotomas that are 6° or less in diameter are not detected by Amsler grid testing [10]. The poor performance of the Amsler grid may be related to the inability to maintain fixation while testing the peripheral visual field [14], crowding effects caused by peripheral presentation of multiple lines [15], inability to assess factors such as quality of examination performance, and low compliance to perform the Amsler grid at home [14]. Despite the shortcomings of this test, the Amsler has been used as a measure of visual change in a randomised controlled trial (RCT) investigating the effect of nutritional supplementation in atrophic AMD [16].
The MMT was designed primarily for quick assessment of residual vision in patients with maculopathies. However, the peripheral fixation aid, along with the use of letter targets may make it a useful tool for monitoring progression of macular disease. The objective of this study was to assess reliability (repeatability and reproducibility), and to assess the effect of ARMD on MMT scores.
The results of this study will allow the use of the MMT as an outcome measure in clinical research, and may also be of interest to those who evaluate scotomas for the purpose of eccentric viewing training.
Methods
Design and setting
This was a reliability study designed to investigate the reproducibility and repeatability of the MMT. Data was also collected from ARMD affected eyes to assess the impact of the disease on MMT score. The study was carried out in the Neurosciences Research Institute at Aston University, Birmingham, UK, in a clinical research setting.
Subjects
Thirty-one normally sighted participants were recruited from staff, students, and patients of the Division of Optometry, Aston University, Birmingham, UK. This research adhered to the tenets of the Declaration of Helsinki. All participants gave informed consent to take part in the study, which was approved by the Institutional Human Ethics Committee. These participants varied in age from 23 to 77 years (mean ± SD, 50.3 ± 19.0 years). Visual acuity (VA) ranged from -0.20 to 0.0 (0.00 ± 0.07 logMAR).
Exclusion criteria were: best-corrected logMAR VA of worse than 0.1 logMAR (VA was measured under standard testing conditions using a logMAR chart, retro illuminated to a luminance of 130 cdm-2 [17] and each letter seen was scored as 0.02 log units, with guessing encouraged); retinal disease detected using a direct ophthalmoscope; glaucoma; lenticular opacities greater than grade 1 on the LOCS I grading scale [18]; prescribed medication associated with changes in retinal function.
The ARMD affected eyes were classified according to the International Classification and Grading System for Age-related Maculopathy (ARM) and Age-related Macular Degeneration (AMD) [19]. ARM refers to large soft drusen and pigmentary abnormalities of the retinal pigment epithelium (RPE) and the retina, whereas AMD refers to later stages of the disease such as GA, choroidal neovascularization, pigment epithelium detachment, and fibrous scaring of the macula [19]. Digital fundus photographs were taken using the Topcon non-mydriatic TRC-NW5S retinal camera (Topcon House, Bone Lane. Kennet Side, Newbury, Berkshire RG14 2PX, UK). Eyes were not dilated and one investigator classified all photographs.
The MMT was carried out on 17 ARM affected eyes of 17 participants aged from 55 to 82 (69.4 ± 7.7 years) and 12 AMD affected eyes of 12 participants aged from 65 to 78 (71.8 ± 4.3 years). VA ranged from -0.08 to 0.2 (0.04 ± 0.09 logMAR) for the ARM group, and from 0.2 to 0.76 (0.50 ± 0.21) for the AMD group. The eyes included did not have lenticular opacities greater than grade 1 on the LOCS I grading scale [18], and were not affected by any other ocular condition. The MMT scores from these groups were compared with those from 16 normal participants aged > 50 years, aged from 51 to 77 (67.1 ± 8.9 years).
Materials
The MMT is a software program used in conjunction with a desktop or laptop computer, designed specifically to map visual defects due to macular disease. The test screen displays a characteristic background pattern throughout the test, which resembled a 'wagon wheel'. Eight spokes point inwards, but do not reach the centre of the circular display area. This pattern provides the patient with sufficient peripheral landmarks to indicate the location of the centre of the circular display area [20], even if the centre is not directly visible to the patient (fig. 1).
When loaded for the first time, the program prompts the user to calibrate the system. It is necessary to measure the diameter of the wagon wheel, to the check the horizontal/vertical proportion of the monitor, and to set the monitor to maximum contrast and brightness. The diameter of the wagon wheel on the screen allows the system to suggest a correct viewing distance, such that the wagon wheel has the correct angular extent of 18° diameter. The viewing distance for our system was 76.2 cm.
The targets comprise of the Sloan letters [21] with a change in the width-to-height ratio from 5 × 5 to 4 × 5 for better legibility. Serifs are also added to the D to reduce the probability of confusion with O. Letter sizes vary according to eccentricity, starting at 5 pixels high, and increments of 5 pixels to 45 each. Thirty-two standard locations ensure that the letters do not 'collide' with the spokes of the wagon wheel. There are an additional four central locations.
Procedure
All participants were seated comfortably, with their head supported using a chin rest, and the eye not being tested occluded. In order to ensure control of direction of gaze, each participant was asked, 'Do you have a sense of where the centre of the wagon wheel is?'. All participants responded positively, and were then instructed to direct their eye to the centre and to keep it there as still as possible. Participants with AMD were also advised that the centre of the wagon wheel may seem to disappear. As the background illuminance was low (6 lux), and our participants did not have lens opacities greater than grade one of the LOCS I scale, we were not concerned about glare, and therefore selected the black letters on a white background presentation mode. The following explanation of the test was given to every participant; 'Black letters will appear one at a time anywhere on the white wagon wheel. It is important that you do not move your eye to try and look at them. When you see a letter, say the name of it out loud. If you are not sure what the letter is, have a guess'.
The same testing room was used for each test with background lighting switched off. When both eyes matched the inclusion criteria, the right eye was tested, and when only one eye was suitable for inclusion, this eye was tested. The prescription providing best-corrected visual acuity at the test distance was placed into a trial frame, with the eye not being tested occluded. Full aperture trial lenses were used.
When each participant was ready the examiner pressed the mouse button and a letter appeared in an unpredictable location in the visual field, remaining visible for 234 msec. The participant told the examiner which letter they perceived and the examiner entered this response into the computer for scoring. This prompts display of the next letter, and so on. There are three categories of response; 1) target not detected, 2) target detected, but not recognized (including incorrect responses), 3) target correctly recognized. The final score is calculated using the following system; target not detected = 0, target detected but not recognized = 1, target detected = 2. The score is displayed at the end of the test and the participant is able to view a 'map' of the areas seen and unseen.
Data were collected by two optometrists, HB and LD, during two sessions separated by one hour from the same subjects. A trial run was completed for each subject by HB in the first session and LD in the second session to encourage test familiarity and reduce learning effects. In session 1 the first experimental test was carried out by HB, and the second by LD, and in session two this order was reversed The study was designed to assess reproducibility (HB1-LD1, LD2-HB2), and repeatability (HB1-HB2, LD1-LD2),
Data analysis employed the independent-samples t-test for comparing ages and the chi-squared test to compare proportion of males by group. The MMT scores are non-continuous data, and so we used nonparametric statistical tests in our analysis. The Mann-Whitney U test was used to compare MMT scores between groups. When assessing the effect of ARMD on MMT score, power analysis shows that the group sizes were sufficient to have an 80% chance of detecting a difference in means of 5 MMT points at the 5% level of significance using the independent samples t-test.
Results
Reliability
Average MMT scores ranged from 85.5 to 100.0 MMT points. Reproducibility was determined by comparing HB1-LD1 and LD2-HB2, and repeatability was determined by comparing HB1-HB2 and LD1-LD2. The differences between data sets for two of the comparisons (HB1-HB2 and HB1-LD1) were normally distributed and these were used for analysis. See figures 2 to 5 for a graphical representation of test-retest data of the HB1-HB2, HB1-LD1, LD1-LD2, and LD2-HB2 comparisons respectively.
Accurate analysis of test-retest data can be achieved using the coefficient of repeatability [22,23]. This gives the 95% confidence limits for the amount of difference between two sets of results. It is calculated as 1.96 multiplied by the standard deviation of the mean differences between the two sets of data. The coefficient of repeatability is ± 6.70 for the HB1-HB2 comparison, ± 7.24 for the HB1-LD1 comparison, ± 7.4 for the LD1-LD2 comparison, and ± 4.3 for the LD2-HB2 comparison.
A comparison between HB1 and LD2 has also been made. The coefficient of repeatability is ± 4.9 and the data is shown in figure 6.
Effect of ARM
Of the ARM participants, 12 had soft distinct drusen, three had soft distinct drusen with hyperpigmentation of the RPE, and two had soft indistinct drusen. The ARM group and the subset of nomals were matched for age (t = -0.783; p = 0.439) and gender [χ2 (1) = 0.501, p = 0.579]. Scores ranged from 85.5 to 100.0 for the subset of normals and from 79.0 to 99.0 for the ARM group. The mean MMT score was 94.5 points (94.6 ± 4.8) for the over 50 normal group and 95.0 points (94.9 ± 4.7) for the ARM group, and there was no significant difference between the groups (z = -0.216, p = 0.829). Power analysis shows that the group sizes were sufficient to have an 80% chance of detecting a difference in means of 5 MMT points at the 5% level of significance using the unpaired t-test.
Effect of AMD
All AMD participants had GA, characterised by any sharply delineated roughly round or oval area of hypopigmentation or depigmentation, or apparent absence of the RPE, in which choroidal vessels were more visible than in surrounding areas. The size of the lesion was at least 1/8 disc diameters in size. The AMD group and the subset of normals were matched for age (t = -1.613, p = 0.119) and gender [χ2 (1) = 0.619, p = 0.821). Scores ranged from 9.0 to 92.0 for the AMD group. The mean MMT score was 62.0 points (61.9 ± 23.8) for the AMD group. The difference in scores between the over 50 and the AMD group is statistically significant (z = 3.761, p < 0.001).
Figure 7 shows the effect of age on MMT score for the normal, ARM and AMD groups.
Discussion
The primary objective of this study was to determine the reliability of the MMT. The developers of the MMT assessed its reliability by performing the test twice on 20 eyes. The conditions were made harder between the first and second run by decreasing the size of the letters by 5 pixels, and 92 % of all tested locations in the participants behaved within expectation. It was concluded that the test procedure reflects the functional topography with reasonable accuracy and reliability [24]. Our reliability data has been used to determine the decrease in MMT score that is needed to indicate a clinical change between tests. The MMT scores are repeatable to within ± 7.0 points, indicating that the MMT score has to change by more than 14 points for the change to be clinically significant. This value could be useful if MMT score was being used as an outcome measure for monitoring progression of ARMD, or for monitoring the effect of an intervention, e.g. a nutritional supplement.
A limitation of this study design is that the order of the assessor in each of the two sessions was not randomly assigned. This would have permitted assessment of the variance components, assessor, order, and session. However, a comparison has been made between HB1 and LD2, and the coefficient of repeatability was smaller than for other comparisons. This, therefore, does not affect the reliability conclusions draw.
Comparisons between ARM-affected eyes and age- and gender-matched controls yielded no significant difference, although this part of the study was powered to detect a difference between groups. There was significant reduction in MMT score between AMD-affected eyes and age- and gender-matched controls. Our results show that the MMT is not suitable for screening of ARM.
Conclusion
Although the reliability data indicates variability, the MMT has the advantage over, for example, the Amsler grid that it uses a letter target, has a peripheral fixation aid, and it provides a numerical score. The score could be beneficial in clinic and home monitoring of AMD progression, as well as provide an outcome measure in clinical research.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HB participated in the design of the study, the acquisition, analysis, and interpretation of the data, and drafting of the final manuscript. LD participated in the design of the study, aquisitiion of data, and the final drafting of the manuscript. FE participated in the design of the study and the final drafting of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
Hannah Bartlett is funded by the College of Optometrists, UK. This work was presented as a paper at the American Academy of Optometry meeting, Tampa, 2004.
Figures and Tables
Figure 1 Subject view of the MMT results. Black squares indicate letters that were not seen, half-shaded squares indicate those that were seen but not correctly identified, and white squares indicate letters that were correctly identified.
Figure 2 Difference in MMT score between HB1 and HB2 compared with the mean (n = 31 eyes). The mean bias is represented by the solid line and the 95 % confidence limits are represented by the dotted lines.
Figure 3 Difference in MMT score between HB1 and LD1 compared with the mean (n = 31 eyes). The mean bias is represented by the solid line and the 95 % confidence limits are represented by the dotted lines.
Figure 4 Difference in Macular Mapping Test (MMT) value between LD1 and LD2, compared with the mean (N = 30 eyes). The mean bias is represented by the solid line, and the 95% confidence limits are known by the dashed lines.
Figure 5 Difference in Macular Mapping Test (MMT) value between LD2 and HB2, compared with the mean (N = 30 eyes). The mean bias is represented by the solid line, and the 95% confidence limits are known by the dashed lines.
Figure 6 Difference in Macular Mapping Test (MMT) value between HB1 and LD2, compared with the mean (N = 30 eyes). The mean bias is represented by the solid line, and the 95% confidence limits are known by the dashed lines.
Figure 7 First MMT scores taken from normal (over 50 years), ARM and AMD participants plotted against age.
==== Refs
Klein R Klein BE Jensen SC Meuer SM The five-year incidence and progression of age-related maculopathy: the Beaver Dam Eye Study Ophthalmology 1997 104 7 21 9022098
The Eye Diseases Prevalence Research Group Prevalence of Age-Related Macular Degeneration in the United States Archives of Ophthalmology 2004 122 564 572 15078675 10.1001/archopht.122.4.564
American Academy of Ophthalmology Photodynamic therapy with verteporfin for age-related macular degeneration Ophthalmology 2000 107 2314 2317 11186878 10.1016/S0161-6420(00)00562-5
Bressler NM Photodynamic therapy of subfoveal choroidal neovascularization in age-related macular degeneration with verteporfin: two-year results of 2 randomized clinical trials: TAP Report 2 Archives of Ophthalmolgy 2001 119 198 207
Macular Photocoagulation Study Group Laser Photocoagulation of subfoveal neovascular lesions of age-related macular degeneration: updated findings from two clinical trials Archives of Ophthalmolgy 1993 111 1200 1209
Macular Photocoagulation Study Group Laser photocoagulation of subfoveal neovascular lesions in age-related macular degeneration: results of a randomized clinical trial Archives of Ophthalmolgy 1991 109 1220 1231
Macular Photocoagulation Study Group Early detection of extrafoveal neovascular membranes by daily field evaluation Ophthalmology 1985 92 603 609 2409502
Yannuzzi LA A modified Amsler grid Ophthalmology 1982 89 157 159 6175935
Singerman LJ Important points in management of patients with choroidal neovascularization Ophthalmology 1985 92 610 614 2409503
Schuchard RA Validity and interpretation of Amsler Grid reports Archives of Ophthalmolgy 1993 111 776 780
Boldrey EE Foveal ablation for subfoveal choroidal neovascularization Ophthalmology 1989 96 1430 1435 2476702
Walsh AW Magargal LE Wright F Donoso LA The early natural history of subfoveal neovascular membranes in eyes with age-related macular degeneration Annals of Ophthalmology 1989 21 348 350 2817663
Swann PG Lovie-Kitchin JE Age-related maculopathy, II: the nature of the central visual field loss Ophthalmic Physiol Opt 1991 11 59 70 2034457 10.1016/0275-5408(91)90012-8
Loewenstein A Malach R Goldstein M Leibovitch I Barak A Baruch E Alster Y Rafaeli O Avni I Yassur Y Replacing the Amsler Grid. A New Method for Monitoring Patients with Age-related Macular Degeneration Ophthalmology 2003 110 966 970 12750099 10.1016/S0161-6420(03)00074-5
Parkes L Lund J Angelucci A Compulsary averaging of crowded orientation signals in human vision Nature Neuroscience 2001 4 739 744 11426231 10.1038/89532
Richer S Stiles W Statkute L Pulido J Frankowski J Rudy D Pei K Tsipursky M Nyland J Double-masked, placebo-controlled, randomized trial of lutein and antioxidant supplementation in the intervention of atrophic age-related macular degeneration: the Veterans LAST study (Lutein Antioxidant Supplementation Trial) Optometry 2004 75 216 230 15117055
Bailey F Lovie-Kitchin JE New design principles for visual acuity letter charts American Journal of Optometry and Physiological Optics 1976 53 745 753
Chylack LT Leske MC Sperduto R Khu P McCarthy D Lens Opacities Classification-System Arch Ophthalmol 1988 106 330 334 3345149
Bird AEC Bressler NM Bressler SB Chisholm IH Coscas G Davis MD Dejong P Klaver CCW Klein BEK Klein R Mitchell P Sarks JP Sarks SH Sourbane G Taylor HR Vingerling JR An International Classification and Grading System for Age- Related Maculopathy and Age-Related Macular Degeneration Survey of Ophthalmology 1995 39 367 374 7604360
Steinbach M Pursuing the perceptual rather than the retinal stimulus Vision Research 1976 16 1371 1007015 10.1016/0042-6989(76)90154-1
Sloan LL New test charst for the measurement of visual acuity at far and near distances American Journal of Ophthalmology 1959 48 807 813 13831682
Bland JM Altman DG Statistical methods for assessing agreement between two methods of clinical measurement Lancet 1986 1 307 310 2868172
Elliott DB Sheridan M The use of accurate visual acuity measurements in clinical anti-cataract formulation trials Ophthalmic Physiol Opt 1988 8 397 401 3253632
Mackeben M Colenbrander A Mapping the Topography of Residual Vision after Macular Vision Loss IOS Press 1994 59 67
|
16092954
|
PMC1208902
|
CC BY
|
2021-01-04 16:03:49
|
no
|
BMC Ophthalmol. 2005 Aug 10; 5:18
|
utf-8
|
BMC Ophthalmol
| 2,005 |
10.1186/1471-2415-5-18
|
oa_comm
|
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-201610518210.1186/1471-2415-5-20Case ReportBilateral neuro-retinitis following chick embryo cell anti-rabies vaccination – a case report Saxena Rohit [email protected] Harinder Singh [email protected] Harminder Kumar [email protected] Vimla [email protected] Dr. R. P. Centre for Ophthalmic Sciences, All India Institute of Medical Sciences, New Delhi 110029, India2005 17 8 2005 5 20 20 18 5 2005 17 8 2005 Copyright © 2005 Saxena et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The Optic nerve is rarely involved after sheep brain anti-rabies vaccination in the form of retrobulbar neuritis or papillitis. Bilateral neuroretinitis after chick embryo cell antirabies vaccination has not been reported.
Case presentation
We report the case of a 56 year old male who developed bilateral neuro-retinitis following three injections of antirabies vaccine prepared from the chick embryo.
Conclusion
The chick embryo cell antirabies vaccine can cause bilateral neuroretinits which has not been reported previously.
==== Body
Background
The Optic nerve is rarely involved after sheep brain anti-rabies vaccination. Neurological complications are usually seen with sheep brain vaccines but can be rarely seen after chick embryo cell vaccines [1]. The main cause in such cases is presumed to be the antigenic cerebral tissue used in the preparation of sheep brain vaccine [2-5]. We report the case of a 56 year old male who developed bilateral neuro-retinitis following three injections of antirabies vaccine prepared from the chick embryo. Retrobular neuritis and papillitis following sheep brain antirabies vaccine have been reported [6-11]. The present report describes a case with bilateral neuroretinitis after chick embryo antirabies vaccine, which to best of our knowledge has not been reported earlier.
Case presentation
A 56 year old male presented with complaint of acute onset painless diminution of vision in the right eye of 3 days duration followed by similar complaint in left eye after 1 day which deteriorated over the next 2 days. This was associated with headache and pain on ocular movements. There was a history of being bitten by a stray dog 8 days before the visual symptoms, and the patient had received three injections of chick embryo cell anti-rabies vaccine (Rabipur, Hoeshst Marion Roussel) on day 0, 3 and 7 after the dog bite. He developed these symptoms after the third injection. There was no history of any other ocular or systemic problems.
The general physical examination was within normal limits. The best-corrected visual acuity was 6/60 in the right and 6/24 in the left eye. IOP was 16 mm of Hg by applanation tonometry. Anterior segment examination using slitlamp was normal. Fundoscopy using + 90 D revealed hyperemic optic disc with blurred margins and disc edema. Vessels were tortuous and macular star was present in the nasal part of macula of both the eyes (Fig 1). Pupils were sluggishly reacting in both the eyes. Goldmann perimetry revealed centrocaecal scotoma in both eyes which was more dense in the right eye as compared to the left.
The routine blood investigations were all within normal limits. The aerobic and aerobic cultures were sterile. The VDRL test and indirect haemagglutination test for syphillis and toxoplasmosis were normal. Serology for bartonella was negative. Mantoux test showed a reading of 8 mm. and the chest X-ray was normal. Magnetic resonance imaging (MRI) brain and optic nerves revealed slight signal enhancement of both the optic nerves with slight increase in optic nerve thickness on axial T2-weighted magnetic resonance imaging. No other significant abnormality was seen in MRI of brain.
A diagnosis of bilateral neuroretinitis was made. As the dog which had bitten the patient was kept under observation and was alive after one week, further injections of antirabies vaccine were stopped. The patient was given three doses of pulse dexamethasone 100 mg in 150 ml of 5% dextrose. The visual acuity improved to 6/6 in both the eyes over the next two weeks. The disc oedema start resolving within one week but macular exudates took 4 months to resolve.
Discussion
Neuroretinitis is an acute swelling of the optic disc associated with hard exudates or edema around the macula. They are frequently associated with infectious diseases but may also occur without any apparent cause (Leber's Idiopathic Stellate neuroretinitis). Infectious causes reported in literature include syphilis, toxoplasmosis, toxocariosis, Lyme's disease and Cat scratch disease. Antecedent viral infection especially herpes simplex virus (HSV), Hepatitis B and mumps have also been reported. It is a self limiting disease with a tendency to recover spontaneously. No treatment is also an option as the condition recovers spontaneously but treatment may speed the recovery and shorten the course of the disease. The cases with underlying bacterial etiology require the specific antibiotic therapy [12,13].
Neurologic complications after sheep brain anti rabies vaccination, although rare can be very serious. These include encephalitis, myelitis, encephalomyelitis and paralysis of various cranial nerves[2]. Optic nerve involvement is not a frequently reported complication and is usually a part of diffuse encephalomyelitis. Antigenecity of the sheep brain is generally held responsible for the neuroparalytic complications [1,2]. The onset of symptoms may vary from 1 to 3 weeks to as much as 100 days after vaccination. Immune complex mediated vascular injury jeopardizing the blood-brain barrier has been proposed as a possible mechanism, which leads to lymphocyte mediated demyelinaton and inflammation [4]. In all previously reported cases of optic neuritis following antirabies vaccination, sheep brain vaccine had been used. However our present case had received chick embryo vaccine. Chakravarty has reported a single case of neurologic illness simultating Guillain Barre syndrome after post-exposure prophylaxis with purified chick embryo cell anti-rabies vaccine [1]. Apart from this, no published reports are available regarding the neurological complication following purified chick embryo cell vaccination (although manufacturer's information on their websites does mention the risk of neurological complications with some chick embryo vaccines). This case was also unusual in the sense that patient has neuroretinits as compared to previously reported cases which have retrobulbar neuritis or papillitis [6-11]. Although good recovery is reported spontaneously, pulse steroids speed up the recovery. Our patient had good visual recovery after three doses of pulse steroids. As brain tissue and perhaps chick embryo tissue (as in our case) is thought to be the cause of the neuroparalytic complications, it is suggested that recombinant vaccines may offer a solution to this problem.
Conclusion
The chick embryo antirabies vaccine can cause bilateral neuroretinits which has not been reported previously.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RS : Helped in making the clinical diagnosis and management of the case.
HSS : Maintained the follow-up record, and prepared the manuscript.
HKR : Performed the literature search and helped in documentation of case
VM : Helped in making the clinical diagnosis and management of the case and supervised the case report.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
None
Figures and Tables
Figure 1 Fundus photograph of right and left eye depicting bilateral neuroretinitis.
==== Refs
Chakravarty A Neurologic illness following post-exposure prophylaxis with purifiled chick embryo cell antirabies vaccine J Assoc Physicians India 2001 49 927 8 11837768
Raghvan Veera N Rabies and its prevention Vepery, Madras, India 1955 The Diocesan Press 17
Brain WR Diseases of the Nervous System 1962 6 New York: Oxford University Press
Stevenson VL Acheson JF Ball J Plant GT Optic neuritis following measles/rubella vaccination in two 13-year-old children Br J Ophthalmol 1996 80 1110 1111 9059281
Fenichel GM Neurological complication after immunization Ann Neurol 1982 12 119 28 6751212 10.1002/ana.410120202
Cormack HS Anderson LAP Bilateral papillitis following antirabies innoculation recovery Br J Ophthalmol 1934 18 167 8
Consul BN Purohit GK Chabra HN Antirabies vaccine optic neuritis Indian Med Sci 1968 22 630 632
Srisupan V Konyama K Bilateral retrobulbar optic neuritis following antirabies vaccination Sriraj Hosp Gaz 1970 4 403 8
Chayakul V Ishikawa S Chotibut S Konyama K Convergence insufficiency and optic neuritis due to antirabies innoculation – a case study Jap J ophthalmol 1975 19 307 14
Gupta V Bandyopadhyay S Bapuraj JR Gupta A Bilateral optic neuritis complicating rabies vaccination Retina 2004 24 179 181 15076969 10.1097/00006982-200402000-00033
Dadeya s Guliani BP Gupta VS Mallick KPS Jain DC Retrobulbar neuritis following rabies vaccination Tropical doctor 2004 34 174 175 15267054
Neuroretinitis Miller NR, Newman J Walsh & Hoyt's clinical Neuro – ophthalmology 1998 I 5 williams & Wilkins Baltimore Mariland, USA 634 647
Neuroretinitis Liu GT, Volpe NJ, Galette SL Neuro – Ophthalmology Diagnosis and management 2001 1 WB Saunders company, Philadelphia, pennsylvania 140 141
|
16105182
|
PMC1208903
|
CC BY
|
2021-01-04 16:03:49
|
no
|
BMC Ophthalmol. 2005 Aug 17; 5:20
|
utf-8
|
BMC Ophthalmol
| 2,005 |
10.1186/1471-2415-5-20
|
oa_comm
|
==== Front
BMC OphthalmolBMC Ophthalmology1471-2415BioMed Central London 1471-2415-5-211610722410.1186/1471-2415-5-21Technical AdvanceClosed chamber globe stabilization and needle capsulorhexis using irrigation hand piece of bimanual irrigation and aspiration system Sethi Harinder S [email protected] Tanuj [email protected] Harminder K [email protected] Prabhpreet [email protected] Rajendra Prasad Centre for Ophthalmic Sciences, All India Institute of Medical Sciences, New Delhi 110029, India2 Lala Ram Saroop Institute, Mehrauli, New Delhi, India2005 18 8 2005 5 21 21 1 6 2005 18 8 2005 Copyright © 2005 Sethi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The prerequisites for a good capsulorhexis include a deep, well maintained anterior chamber, globe stabilization and globe manipulation. This helps to achieve a capsulorhexis of optimal size, shape and obtain the best possible position for a red glow under retroillumination. We report the use of irrigation handpiece of bimanual irrigation aspiration system to stabilize the globe, maintain a deep anterior chamber and manipulate the globe to a position of optimal red reflex during needle capsulorhexis in phacoemulsification.
Methods
Two side ports are made with 20 G MVR 'V' lance knife (Alcon, USA). The irrigation handpiece with irrigation on is introduced into the anterior chamber through one side port and the 26-G cystitome (made from 26-G needle) is introduced through the other. The capsolurhexis is completed with the needle.
Results
Needle capsulorhexis with this technique was used in 30 cases of uncomplicated immature senile cataracts. 10 cases were done under peribulbar anaesthesia and 20 under topical anaesthesia. A complete capsulorhexis was achieved in all cases.
Conclusion
The irrigating handpiece maintains deep anterior chamber, stabilizes the globe, facilitates pupillary dilatation, and helps in maintaining the eye in the position with optimal red reflex during needle capsulorhexis. This technique is a safe and effective way to perform needle capsulorhexis.
==== Body
Background
The anterior capsulorhexis has got several intra and post operative advantages over can opener or endocapsular capsulotomies and has become the standard capsulotomy technique for phacoemulsification [1-3]. Anterior capsulorhexis can be performed using 26 G bent needle cystitome or Utratas forceps [1,2]. The needle capsulorhexis can be performed through side port incision using a viscoelastic device or an anterior chamber maintainer [1,2,4,5]. During the performance of capsulorhexis, the globe can be stabilized either using a second instrument such as a Sinskey hook, or by holding limbal conjunctiva with a Lim's forceps. Sinskey hook introduced through a separate side port incision can lead to egress of viscoelastic from the eye and hence risk of radial extension of capsular flap. Holding conjunctiva with Lim's forceps can be traumatic or undesirable under topical anaesthesia.
The prerequisites for a good capsulorhexis include a deep well maintained anterior chamber, globe stabilization and globe manipulation to achieve best position for a red glow under retroillumination. All these can be achieved by the technique described by us, which is a modification of anterior capsulorhexis under anterior chamber maintainer described previously by Blumenthal [4].
Methods
Two side ports are made with 20 G V lance knife (Alcon, Fort Worth Texas, USA) at 10 'O' clock and 2 'O' clock. A bent cystitome is made from 26-G needle. The irrigation hand piece of Bimanual irrigation aspiration system (Appasamy Associates, Chennai, India) is introduced into the anterior chamber through 2 'O' Clock side port with irrigation on. The irrigation hand piece is attached to balanced salt solution with bottle height of 90 – 95 cm above eye level. The 26-G cystitome is introduced through 10 'O' clock side port. The relaxing incision is made in the central area of anterior capsule, and a flap is created. This flap is flipped and reflected upon the underlying capsule so that epithelial side now faces the cornea. The reflected flap is then engaged near the tearing edge and rotated in a circular manner. When almost entire circumference of the central capsular opening is achieved, the flap is pulled centripetally inwards. This joins the capsular edges from outside in to complete the capsulorhexis (Figure 1). During this period, the irrigation hand piece held in the left hand maintains deep anterior chamber and stabilizes the globe. It also helps to keep globe in the best position for a good red glow.
Results
We have used this technique in 30 cases of uncomplicated immature senile cataracts. The cases with intumuscent cataracts, white cataracts, corneal opacities or other ocular pathologies were not included.10 cases were done under peribulbar anaesthesia and 20 under topical anaesthesia. A complete capsulorhexis was achieved in all cases.
Discussion
The anterior capsulorhexis is one of the prerequisites for successful and uncomplicated phacoemulsification [1]. The needle capsulorhexis can be performed under an anterior chamber maintainer, which requires an extra side port incision inferiorly, whereas in our technique the side port made for the second instrument is used for introducing the irrigation handpiece. In addition one needs to stabilize the globe using a Lim's forceps or Sinskey hook introduced through side port. The Sinskey hook introduced through side port can lead to egress of viscoelastic or fluid, which can lead to shallowing of the anterior chamber. Holding conjunctiva with lims's forceps can lead to hemorrhage, tear and increased patient discomfort. This is undesirable, especially under topical anaesthesia. These problems can be overcome by using the above described technique.
We prefer to use irrigation hand piece to achieve well maintained deep anterior chamber and relaxed zonules throughout the capsulorhexis. In addition, the irrigation handpiece snuggly fits into the side port incision minimizing the fluid leaks. This also helps in moving the globe to an optimal position for the best red glow under retroillumination. The slight manipulation of the globe in a direction opposite to the needle movement facilitates the rotation of the anterior capsular flap with needle. Our technique is especially useful for the phacoemulsificaton under topical anaesthesia, where holding the conjunctiva for globe rotation is undesirable and can be painful for the patient. This can also cause conjunctival tear and subconjunctival hemorrhage. Our technique has a special role in the developing countries where reducing the amount of viscoelastic used can decrease the overall cost of surgery.
With continuous irrigation, the capsular flap floats freely in the anterior chamber. This improves visualization and maneuverability. This increased mobility of the anterior capsular flap may be troublesome for the beginners. It can be minimized by keeping the direction of flow of fluid away from the site of flap rotation. The capsulorhexis under the irrigation fluid demands a tight incision and undue posterior pressure on these incisions is undesirable as this may lead to sudden egress of fluid and the flap may extend radially outwards [2].
The described technique by us is simple and needs reapplication of already described techniques, hence can be easily mastered by all surgeons.
Conclusion
The irrigating handpiece maintains deep anterior chamber, stabilizes the globe, facilitates pupillary dilatation, helps in maintaining the eye in the position with optimal red reflex during needle capsulorhexis. This technique is a safe and effective way to perform needle capsulorhexis.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HSS : Conceived the idea of above technique, conducted, coordinated the study and performed the surgeries.
TD : Performed the surgeries and helped in documentation of cases
HKR : Performed the surgeries and helped in documentation of cases
PS : Compiled the data and helped in the preparation of the manuscript
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
None.
Figures and Tables
Figure 1 Closed chamber globe stabilization with irrigation hand piece and needle capsulorhexis with 26 G cystitome.
==== Refs
Zamini M Buratto L Savin G Buratto L, Werner L, Zanini M Apple D Capsulorhexis 2003 2 Phacoemulsification : Principles and Techniques, Slack Incorporated, Thorafore NJ, USA 83 92
Neuhan TF Steinert RF, Fine H, Gimbel HV, Koch HD, Lindstrom RL, Neuhan TF, Osher RH Capsulorhexis 2004 2 Cataract Surgery, Elsevier Science, USA 137 46
Gimbel HV Neuhan T Development, advantages and methods of the continous circular capsulorhexis J Cataract Refract Surg 1990 16 31 37 2299571
Blumenthal M Arsia E Schochot Y Lens anatomical principles and their technical implications in cataract surgery. Part I: The lens capsule J Cataract Refract Surg 1991 17 205 10 2040978
Polat A needle capsulorhexis: The importance of using an anterior chamber maintainer J Cataract Refract Surg 2003 29 1248 9 12900223 10.1016/S0886-3350(03)00475-9
|
16107224
|
PMC1208904
|
CC BY
|
2021-01-04 16:03:49
|
no
|
BMC Ophthalmol. 2005 Aug 18; 5:21
|
utf-8
|
BMC Ophthalmol
| 2,005 |
10.1186/1471-2415-5-21
|
oa_comm
|
==== Front
BMC Palliat CareBMC Palliative Care1472-684XBioMed Central London 1472-684X-4-51610916310.1186/1472-684X-4-5Research ArticleAcupuncture for dyspnea in advanced cancer: a randomized, placebo-controlled pilot trial [ISRCTN89462491] Vickers Andrew J [email protected] Marc B [email protected] Gary E [email protected] Barrie R [email protected] Integrative Medicine Service, Memorial Sloan-Kettering Cancer Center, New York, USA2 Biostatistics Service, Memorial Sloan-Kettering Cancer Center, New York, USA3 Pulmonary Service, Memorial Sloan-Kettering Cancer Center, New York, USA2005 18 8 2005 4 5 5 11 4 2005 18 8 2005 Copyright © 2005 Vickers et al; licensee BioMed Central Ltd.2005Vickers et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Dyspnea, or shortness of breath, is a common symptom in patients with advanced cancer. Pharmacologic management is of proven benefit, but it does not help all patients. Preliminary data suggest that acupuncture can relieve dyspnea in a variety of populations, including cancer patients. We conducted a pilot study (ISRCTN89462491) preparatory to a fully powered randomized, placebo-controlled trial to determine whether acupuncture reduces dyspnea in patients with lung or breast cancer.
Methods
The study sample was comprised of forty-seven patients with lung or breast cancer presenting with dyspnea. Patients receiving symptomatic treatments were not excluded as long as no changes in management were planned during the trial. Patients were randomized to receive a single session of true or placebo acupuncture in addition to their existing dyspnea treatments. Semi-permanent acupuncture "studs" were then inserted: patients applied pressure to these studs twice a day to provide ongoing stimulation to acupuncture points. The subjective sensation of dyspnea was assessed with a 0 – 10 numerical rating scale immediately before and after acupuncture treatment and daily for a week thereafter.
Results
All but two of 47 randomized patients provided follow-up data. Dyspnea scores were slightly higher for patients receiving true versus placebo acupuncture, for both the period immediately following acupuncture treatment and for the daily one week follow-up (differences between means of 0.34, 95% C.I. -0.33, 1.02 and 0.56, 95% C.I. -0.39, 1.51). The 95% confidence interval excludes the prespecified minimum clinically significant difference of a 20% greater improvement in dyspnea for patients receiving acupuncture.
Conclusion
The acupuncture technique used in this trial is unlikely to have effects on dyspnea importantly larger than placebo for patients with advanced cancer.
==== Body
Background
Dyspnea, or shortness of breath, is defined as the subjective experience of difficulty breathing. It is a common symptom in cancer patients, particularly those with advanced cancer. For example, in a general survey of 100 outpatients and 140 inpatients at a Veterans' Affairs Medical Center, at least some dyspnea was reported by 50% of patients [1]. Studies of patients with advanced disease find that 50 –70% experience significant dyspnea[2,3]. Although dyspnea is most commonly associated with lung cancer[3], high rates also occur in patients with breast cancer with pulmonary metastases[4].
Dyspnea in cancer patients has numerous causes. It may result from complications of the cancer itself, such as pulmonary embolism, pleural effusion, anemia, and endobronchial obstruction or from conditions associated with a risk factor such as chronic obstructive pulmonary disease (COPD) in smokers. It may also result from disorders not directly attributable to the lungs, such as congestive heart failure or anemia. Treatment of these patients is typically guided by the identification of a specific underlying abnormality. Pleural effusions, for instance, are treated by thoracentesis, COPD is treated by bronchodilators and/or steroids, and anemia is treated by transfusion.
Unfortunately, a specific lung or cardiac pathology is not identifiable in approximately one-quarter of patients[2]. Moreover, dyspnea may result from causes in which definitive treatment has been unsuccessful or not practical. A number of interventions have been attempted for such patients, who often have advanced cancer within the lung parenchyma or endobronchial disease refractory to external beam radiation. Several randomized trials have shown that both opioids and supplemental oxygen can alleviate the subjective sensation of dyspnea among patients with advanced cancer.
For example, in a randomized, double-blind, crossover trial, dyspnea sensation fell by 25 mm on a 100 mm visual analog scale after morphine, with little change on placebo (p < 0.01 for difference between groups)[5]. Using a similar randomized, crossover design, Bruera et al. reported highly significant differences between periods when patients received oxygen as compared to periods when unsupplemented air was administered [6]. However, such measures are not effective in all patients, and the problem of dyspnea in the cancer patient is often frustratingly difficult to treat. A multinational study of end-of life-care found that dyspnea required sedation in 25 to 50% of patients during the last week of life[7]. It is for these difficult to treat cancer patients that additional therapeutic options are needed.
A number of randomized controlled trials have examined acupuncture for shortness of breath in patients without cancer. For example, a randomized, double-blind, placebo-controlled trial of acupuncture for methacholine-induced asthma reported significant differences in favor of acupuncture[8]. A similar study involved a standardized running test to induce bronchorestriction in asthmatics. Exercise-induced reductions in lung function were significantly lower in real compared to placebo acupuncture[9]. A trial in COPD reported significant differences between groups for subjective breathlessness. Thirty-one patients were randomized to self-administer finger pressure ("acupressure") at true or sham acupuncture points on a double-blind, crossover basis. Acupressure led to approximately a one-third reduction of scores on a VAS of dyspnea compared to about a 20% improvement in placebo controls[10].
Only one trial has been reported in cancer. Although uncontrolled, the results are provocative. Thirty cancer patients in palliative care received a single session of acupuncture. The mean visual analog score for breathlessness before acupuncture treatment was 42; this fell to 24 immediately following 10 minutes of needle insertion, an improvement that was maintained at six hour follow-up. Symptom scores returned to baseline 24 hours later[11].
Given these data, we believed it would be worthwhile to investigate whether cancer-related dyspnea can be relieved by acupuncture. Acupuncture is a complementary therapy, and therefore was administered as a supplement to use of pharmacologics or oxygen for dyspnea. Placebo control was employed because our endpoint was subjective and because acupuncture had not previously been found superior to placebo for cancer-related dyspnea. Our overall objective was to determine whether true acupuncture was more effective than placebo for alleviating dyspnea in patients with advanced lung or breast cancer. Here we report the results of a pilot study that aimed to refine the methodology for an anticipated definitive trial.
Methods
Recruitment for the trial took place between July 2001 and March 2004 among patients under treatment for advanced lung or breast cancer at Memorial Sloan-Kettering Cancer Center (MSKCC). Patients age 18 or above with subjective complaint of shortness of breath and scoring grade two or higher on the American Thoracic Society Breathlessness Scale were eligible. Patients must have pursued a trial of steroid medication for dyspnea, if indicated, for at least 48 hours. Patients were excluded if any of the following applied: shortness of breath predated cancer diagnosis (e.g. asthma); recent onset of symptoms (< 7 days); anemia (defined as hemoglobin < 8 gm/dl); recent acupuncture; contraindications to acupuncture such as heart valve dysfunction or pancytopenia; planned initiation or change in oncologic therapy or symptomatic management of breathlessness (continuation of any existing management was allowed), or likelihood of patient death during the course of anticipated participation in the trial. Patients were also excluded if the primary cause of dyspnea was thought to be congestive heart failure, sarcoid disease, hypersensitivity pneumonitis, cryptogenic organizing pneumonitis, pneumothorax, chest wall deformity, obesity, neuromuscular disorders, pulmonary vascular disease, hepatomegaly or phrenic nerve paralysis syndrome. If the primary cause of dyspnea was ascites, effusion, pneumonia, large airway obstruction, superior vena cava syndrome or pulmonary embolism, patients were eligible only if they remained short of breath despite conventional therapy administered by their primary oncologist, or if they had refused such therapy.
Eligible consenting patients were randomized by telephone using the MSKCC clinical research database. Randomization used randomly permuted blocks with the cancer diagnosis (lung/breast) and breathless at rest (yes/no) as strata. The use of independent telephone registration and randomization ensured concealment of treatment allocation. Patients, researchers and others involved in patient care were blind to study group; only the acupuncturists and a researcher not associated with the study were aware of which patients received true and which placebo treatment.
Treatment consisted of two phases, acupuncture and acupressure. Patients received either true acupuncture followed by true acupressure, or placebo acupuncture followed by placebo acupressure. The first "acupuncture phase" consisted of a single treatment in which real or placebo needles were applied for 15 minutes at true or sham points, respectively. In the true acupuncture group, the needles used were stainless steel AsiaMed No.16 30 mm × 0.30 mm inserted to the traditional Chinese depth of 0.5 cm – 1.5 cm. Auricular points were needled with stainless steel Seirin D type No. 1 15 mm × 0.16 mm. As is common in traditional acupuncture, practitioners attempted to elicit de qi to help determine exact point location, but there was no manipulation of needles after placement. Placebo needles consist of a blunted needle that moves up inside its handle instead of into the skin. This technique has previously been demonstrated to be indistinguishable from true acupuncture[12]. Patients were told that in placebo acupuncture "the needles are placed so that they do not stimulate the correct acupuncture points. However, they do not look or feel any different from real needles." Immediately after insertion of the needles, patients were asked to assess the credibility of their treatment using a previously published scale[13,14].
The point prescription is described in Table 1. True points were chosen on the basis of the prior case series[11] and points traditionally used for breathlessness[15]; sham points were chosen in body areas away from true acupuncture points. The point prescription was modified slightly during the trial (details available on request), primarily to improve patient comfort. Patient outcome was not affected by the prescription used (see Results). Immediately after needle insertion, acupuncturists completed an audit sheet verifying the acupuncture points used. These records were routinely reviewed, and acupuncturists were found to have followed the acupuncture point prescription.
Table 1 Location of acupuncture points.
Point Laterality Anatomical location
Ren6 Unilateral On the midline of the lower abdomen, 1.5 cun* inferior to the umbilicus and 3.5 cun* superior to the pubic symphysis
LU1 Bilateral On the lateral aspect of the chest, in the first intercostal space, 6 cun* lateral to the midline,
LU7 Bilateral On the radial aspect of the forearm, in the cleft between the tendons of brachioradialis and abductor pollicis longus
ST36 Bilateral Below the knee, 3 cun* inferior to Dubi ST35, one finger breadth lateral to the anterior crest of the tibia.
KI6 Bilateral 1 cun* below the prominence of the medial malleolus, in the groove formed by two ligamentous bundles.
Auricular lung point Bilateral Medial and distal aspect of the tragus
Auricular kidney point Bilateral Medial and proximal aspect of the cymba conchae
Sternal points OR Ren17 Unilateral Two needles inserted in the top two inches of the sternum and inserted down to the periosteum. Weak or cachectic patients were treated at Ren17, which is found on the anterior midline, at the level with the fourth intercostal space, midway between the nipples
Sham sternal Unilateral Two points anywhere in the upper two inches of the sternum: placebo needle only
Sham1 Bilateral Posterior wrist, between radius & ulna, 1 cun* distal to radial & ulnar heads
Sham2 Bilateral Anterior arm, 3 cun* proximal and 3 cun* medial to the antecubital crease
Sham3 Bilateral Anterior arm, center of biceps brachii (midway between shoulder joint and antecubital crease)
Ear Sham Bilateral Adjacent to the "finger point", just below the helix of the auricle, at the apex
* a "cun" is a practical measurement used by acupuncturists equivalent to the greatest width of a patient's thumb at the distal phalanx
One hour after removal of needles, patients started the second "acupressure phase." Stainless steel AcuMedic acupressure studs, sometimes described as "press," "semi-permanent" or "intradermal" acupuncture needles, were used. These consist of a 2 mm × 0.28 mm acupuncture needle attached to a metal ring embedded in surgical tape. When the surgical tape is pressed on the skin, the needle pierces the skin but its diameter is so fine that this sensation is neither painful nor even immediately obvious. Acupressure studs have been used in a number of studies with cancer patients[16,17]. Placebo studs have no needle: they were specially designed for research in end-of-life populations by the author of the previous research on acupuncture for cancer related breathlessness[11].
To ensure retention of studs and guard against infection, the studs were covered with Tegaderm following application. Patients in the true acupuncture group were treated with true studs at ST36; Sternal points or Ren17; Auricular lung point and Auricular kidney point (see Table 1). Placebo patients received placebo studs at sham1, sham2 and ear sham. We thought it possible that patients might remove the studs before the end of the trial and that studs with a visible needle might be more credible than those without. Therefore, we also applied a placebo stud at sham1 to the acupuncture group and a true stud at sham3 to the placebo group. We told patients that some studs had small needles, others did not and that the choice depended on where they were placed on the body. Patients were also told that those in the placebo group received treatment at points not thought to help breathlessness.
Following application of the studs, patients were instructed to apply pressure to the study by making small circular movements with the fingers of the opposite hand, 2 – 3 cycles per second for 1 – 2 minutes per point. As is typical for self-administered acupressure, patients were encouraged to apply acupressure this way on waking, in the early afternoon and during any exacerbation of symptoms. Initial instruction was provided verbally, at which time patients were asked to confirm their understanding by demonstrating the procedure. Patients also were given easy-to-read written materials describing the acupressure procedure.
Study acupuncturists are certified by the National Certification Commission for Acupuncture and Oriental Medicine (NCCAOM) and are licensed to practice acupuncture in New York State. They have used acupuncture in clinical practice for 3 – 25 years. All were employed at MSKCC during the study and had considerable experience in treating cancer patients.
Outcome was measured in two ways. Every 15 minutes for 75 minutes immediately before acupuncture treatment and one hour immediately after, patients rated their current level of breathlessness on a 0 to 10 rating scale. They then completed a dairy daily for seven days, recording their average level of breathlessness through the day with the same 0 to 10 scale, and recorded compliance with acupressure.
The protocol was modified after 16 patients were accrued, as we found that many in-patients had been excluded because changes in management were likely in the subsequent seven days. We therefore modified the protocol so that inpatients did not take part in the acupressure phase. Moreover, many outpatients who complained of breathlessness in everyday life did not record significant breathlessness during the waiting period immediately before acupuncture. Accordingly, we amended the protocol so that outpatients completed the daily breathlessness diary for a week at baseline. Only patients scoring a mean of two or more were eligible. Out-patients who reported no breathlessness at rest were not asked to report on symptoms in the period immediately before and after acupuncture.
Preliminary power calculations on the basis of published data[10,11] suggested that a sample of 120–150 patients would be required to provide power to detect a clinically significant difference between groups, defined as a 20% lower follow-up dyspnea score in the acupuncture group compared to placebo. For this pilot we sought a sample of 40–50 patients. We felt that a sample of this size would give us sufficient methodologic experience to conduct an adequate fully powered study. Comparisons between groups were by analysis of covariance (ANCOVA) with baseline score and randomization strata as covariates. Baseline score for the acupressure phase was the one week baseline diary, if available, otherwise the immediate pre-treatment breathlessness score was used. Patients were analyzed in their randomized groups regardless of treatment received. Prespecified sensitivity analyses were to examine the effects of credibility on outcome and to assess outcome for the acupuncture phase only in patients with pre-treatment dyspnea scores greater than two. Statistical analysis was conducted using Stata 8 (College Station, Texas). No interim analyses were planned and the data were not analyzed before study closure.
The study was approved by the institutional review board at MSKCC in accordance with an assurance filed with and approved by the Department of Health and Human Services. Written informed consent was obtained from each participant.
Results
Flow of participants through the trial is shown in figure 1. Raw data are provided in Additional file 1. Table 2 shows data on study completers. Groups are balanced for baseline characteristics such as age, sex and diagnosis, and for postrandomization characteristics such as credibility and compliance. The similarity among credibility scores suggests that blinding was maintained, a conclusion also supported by analysis of research assistant notes: of the 21 patients who commented on allocation during post-study debriefing, 12 claimed to be unaware of allocation and only four of the remaining nine made a correct guess as to treatment received. Use of steroids, slightly more common in the placebo group, was not a strong predictor of outcome and there was no interaction between steroid use and group allocation. No patient changed use or dose of diuretics, opiates, bronchodilators or steroids during study participation.
Figure 1 Flow of participants through the trial.
Table 2 Baseline characteristics of patients providing post randomization data. Values are number (percentage) or mean (standard deviation)
Acupuncture n = 25 Placebo n = 20
Female 15 (60%) 13 (65%)
Age 63.0 (12.8) 67.0 (11.4)
Diagnosis
breast 5 (20%) 4 (20%)
lung 20 (80%) 16 (80%)
Breathless at rest 9 (36%) 6 (30%)
Credibility score* 2.24 (0.46) 2.10 (0.59)
Steroid use:**
None 15 (60%) 9 (47%)
<40 mg prednisone equivalents 6 (24%) 8 (42%)
40+ mg prednisone equivalents 4 (16%) 2 (11%)
Diuretics** 3 (12%) 0 (0%)
Opiates** 5 (20%) 3 (16%)
Bronchodilators** 12 (48%) 7 (37%)
Fully compliant with acupressure+ 8 (80%)° 6 (75%)°°
* two missing observations in acupuncture group; three missing observations in placebo group
** one missing observation in placebo group
+ n = 10 acupuncture, n = 8 placebo
° one patient completely non-compliant, one patient 85% compliant
°° one patient 33% compliant, one patient 85% compliant
Breathlessness scores are reported in Table 3. We give results separately for dyspnea measured immediately after acupuncture, and for dyspnea assessed by daily diary. Some patients reporting dyspnea scores immediately after acupuncture treatment had negligible pre-treatment dyspnea. Hence we also provide a sub-group analysis of the immediate post-treatment scores, excluding patients scoring a mean of 2 or below. For all analyses, patients receiving true acupuncture reported slightly higher scores at follow-up, approximately equivalent to 10% of baseline. Patients in both groups improved, but no important differences between groups emerged. The upper bound of the 95% confidence interval for the difference between means excludes our prespecified minimum clinically significant difference of a 20% greater improvement in dyspnea for patients receiving acupuncture. Credibility did not predict breathlessness scores, either in the immediate post-treatment period (p = 0.4) or during the one week of diary recording (p = 0.9). Moreover, including credibility in the model for treatment effect did not importantly modify estimates.
Table 3 Breathlessness scores. Baseline scores only given for patients providing relevant follow-up data. Values are mean (standard deviation)
Endpoint Group Baseline Follow-up p value for within group change Difference between means+ 95% C.I. p value for between group comparison
Immediately post-treatment Acupuncture n = 19: 4.09 (2.32) n = 19: 3.36 (2.21) 0.003 0.34 -0.33, 1.02 0.3
Placebo n = 14: 3.41 (2.79) n = 14: 2.42 (2.64) 0.003
Immediately post-treatment (pre-treatment score >2) Acupuncture n = 15: 4.87 (1.92) n = 15: 3.99 (2.03) 0.003 0.45 -0.55, 1.46 0.4
Placebo n = 8: 5.28 (2.18) n = 8: 3.92 (2.50) 0.01
Mean of seven day breathlessness diary Acupuncture n = 10*: 6.58 (1.71) n = 16*: 5.07 (2.12) 0.4 0.56** -0.39, 1.51 0.2
Placebo n = 7*: 5.99 (1.71) n = 14*: 3.77 (2.39) 0.07
+ Positive indicates worse score in acupuncture group.
* Some patients providing post-treatment diaries gave baseline data only for the immediately pre-treatment period
**n = 13 for placebo group: one patient had no baseline data.
We conducted exploratory analyses to determine whether acupuncture prescription affected outcome. Outcome did not differ by point prescription in either the subgroup analysis (Table 4) or by including prescription and prescription by group interaction in the ANCOVA model.
Table 4 Effect of acupuncture point prescription on outcome.
Endpoint Point prescription n Difference between groups+ 95% C.I.
Immediately post-treatment Initial 15 -0.23 -1.23, 0.77
Modified 18 0.87 -0.19, 1.94
Daily Breathlessness diary Initial 12 0.63 -0.54, 1.79
Modified 17 0.43 -0.92, 1.77
+ Positive indicates worse score in acupuncture group
No adverse events were causally related to acupuncture treatment.
Discussion
Although intended as a methodologic pilot, our results are unexpectedly precise enough to warrant clinical recommendations. The confidence interval for the difference between groups is sufficiently narrow for us to conclude that, even in the best case scenario, the acupuncture technique tested is not importantly superior to placebo for dyspnea in cancer patients. Based on our clinical judgment of how much improvement we should expect to make worthwhile the time and trouble associated with acupuncture, we pre-specified a 20% difference between groups as clinically meaningful. It is against this criterion that we draw our conclusion of "no effect". Yet we can exclude even smaller levels of benefit, such as 10%. We find it unlikely that any patient or clinician would deem acupuncture to be worthwhile for a benefit of one third of a point on a 0 – 10 scale.
There are several possible limitations of our study. First, protocol changes concerning intervention, eligibility and evaluation, although common for a methodologic pilot, are not ideal when drawing conclusions about the effectiveness of a treatment. Nonetheless, we feel that these protocol changes did not have an important influence on our results. Changes in the methods of evaluation were designed to reduce drop-out and improve the measurement of baseline symptoms. These changes should increase the precision of results. Amendments to eligibility requirements were made to ensure that all patients had non-zero symptom scores at baseline and thus ensure that the trial included only those for whom benefit was measurable. Modification of the intervention itself is potentially more problematic. Most of the acupuncture point prescription changes were made to increase patient comfort. For example, we initially used the PC6 point on the wrist, but patients complained that studs inserted there caused radiating pain, apparently due to the proximity of PC6 to the median nerve. If our initial, but not modified, prescription were of value, it would be clinically irrelevant as an effective but intolerable treatment has little clinical role. Moreover, we saw no evidence that outcome was affected by the particular prescription used.
Patients in this study received only a single acupuncture treatment. This does not reflect traditional acupuncture practice, and it is possible that repeated treatment might have a cumulative effect. Our study design was based on prior research that suggested immediate effects on dyspnea following a single session of acupuncture[8,11] Delayed or cumulative effects are of questionable relevance for symptomatic management in patients with late stage cancer. Moreover, daily acupressure treatment had no apparent effect.
It is possible that more encouraging findings may have emerged had we used a different acupuncture technique, such as modifying the point prescriptions for each patient. Our study population reflected a typical spectrum of cases requiring symptomatic management of dyspnea, and was thus rather heterogeneous. Therefore, the possibility that acupuncture might be effective for dyspnea in a sub-group of patients with a particular pathology cannot be ruled out. There is a dearth of systematically collected data to inform decisions about study design with respect to either the most effective acupuncture technique or the most responsive patient group.
Our study provides some evidence of a placebo effect. Symptom scores fell by approximately 20% immediately after acupuncture in both groups (p = 0.003), a result unlikely to be explained by regression to the mean or other effects. It is plausible that such a result would have been considered "positive" in a single-arm study, underscoring the importance of randomized trials for evaluating complementary therapies used to treat cancer-related symptoms.
Accrual to the study was relatively slow, with only approximately 15 patients entering per year. This is due in part to the fact that MSKCC does not provide hospice services, which limits the number of end of life patients, when dyspnea is most prevalent. Accrual was also limited by patient refusal, as nearly half of patients approached declined participation. Patients with advanced cancer admitted to MSKCC generally have severe medical problems. Accordingly, they often feel overwhelmed, and a clinical trial, regardless of its perceived benefits and risks, is perceived as an additional burden. Patients commonly noted in response to the invitation to participate that "I just can't face dealing with this right now".
Conclusion
We conclude that the acupuncture technique used in this trial is unlikely to have effects importantly larger than placebo for dyspnea in patients with advanced cancer.
Abbreviations
ANCOVA: analysis of covariance
COPD: chronic obstructive pulmonary disease
VAS: visual analog scale
MSKCC: Memorial Sloan-Kettering Cancer Center
Competing interests
The author(s) declare that they have no competing interests
Authors' contributions
AJV, BRC and MBF designed the study. AJV analyzed data and wrote the manuscript. GED gave input to the analysis and manuscript with respect to the interpretation of the findings.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional File 1
This is an Excel file with raw data from the study. The first worksheet contains the data; the second, a list of variable descriptors.
Click here for file
Acknowledgements
Bertha Fearon and Tina Chuck were the Research Study Assistants.
Acupuncture was provided by Simon Yeung, Lily Zhang, Jin Cheng Han and Mitch Chalek. The study was funded by the National Center for Complementary and Alternative Medicine at the National Institutes of Health (AT 01030)
==== Refs
Chang VT Hwang SS Feuerman M Kasimis BS Symptom and quality of life survey of medical oncology patients at a veterans affairs medical center: a role for symptom assessment Cancer 2000 88 1175 1183 10699909 10.1002/(SICI)1097-0142(20000301)88:5<1175::AID-CNCR30>3.0.CO;2-N
Reuben DB Mor V Dyspnea in terminally ill cancer patients Chest 1986 89 234 236 3943383
Donnelly S Walsh D The symptoms of advanced cancer Semin Oncol 1995 22 67 72 7537907
Kreisman H Wolkove N Finkelstein HS Cohen C Margolese R Frank H Breast cancer and thoracic metastases: review of 119 patients Thorax 1983 38 175 179 6857580
Mazzocato C Buclin T Rapin CH The effects of morphine on dyspnea and ventilatory function in elderly patients with advanced cancer: a randomized double-blind controlled trial Ann Oncol 1999 10 1511 1514 10643545 10.1023/A:1008337624200
Bruera E de Stoutz N Velasco-Leiva A Schoeller T Hanson J Effects of oxygen on dyspnoea in hypoxaemic terminal-cancer patients Lancet 1993 342 13 14 8100289 10.1016/0140-6736(93)91880-U
Fainsinger RL Waller A Bercovici M Bengtson K Landman W Hosking M Nunez-Olarte JM deMoissac D A multicentre international study of sedation for uncontrolled symptoms in terminally ill patients Palliat Med 2000 14 257 265 10974977 10.1191/026921600666097479
Tashkin DP Bresler DE Kroening RJ Kerschner H Katz RL Coulson A Comparison of real and simulated acupuncture and isoproterenol in methacholine-induced asthma Annals of Allergy 1977 39 379 387 339785
Fung KP Chow OK So SY Attenuation of exercise-induced asthma by acupuncture Lancet 1986 2 1419 1422 2878275
Maa SH Gauthier D Turner M Acupressure as an adjunct to a pulmonary rehabilitation program J Cardiopulm Rehabil 1997 17 268 276 9271771 10.1097/00008483-199707000-00008
Filshie J Penn K Ashley S Davis CL Acupuncture for the relief of cancer-related breathlessness Palliative Medicine 1996 10 145 150 8800822
Streitberger K Kleinhenz J Introducing a placebo needle into acupuncture research Lancet 1998 352 364 365 9717924 10.1016/S0140-6736(97)10471-8
Vincent CA A controlled trial of the treatment of migraine by acupuncture Clinical Journal of Pain 1989 5 305 312 2520420
Vickers AJ Placebo controls in randomized trials of acupuncture Eval Health Prof 2002 25 421 435 12449085 10.1177/0163278702238055
Deadman P Al Khafaj M A Manual of Acupuncture East Sussex, England: Journal of Chinese Medicine Publications (monograph) 1998
Dillon M Lucas C Auricular stud acupuncture in palliative care patients Palliat Med 1999 13 253 254 10474714 10.1191/026921699667999073
Towlerton G Filshie J O'Brien M Duncan A Acupuncture in the control of vasomotor symptoms caused by tamoxifen Palliat Med 1999 13 445 10659122
|
16109163
|
PMC1208905
|
CC BY
|
2021-01-04 16:30:52
|
no
|
BMC Palliat Care. 2005 Aug 18; 4:5
|
utf-8
|
BMC Palliat Care
| 2,005 |
10.1186/1472-684X-4-5
|
oa_comm
|
==== Front
BMC PediatrBMC Pediatrics1471-2431BioMed Central London 1471-2431-5-331613140210.1186/1471-2431-5-33Research ArticleDalhousie dyspnea scales: construct and content validity of pictorial scales for measuring dyspnea McGrath Patrick J [email protected] Paul T [email protected] Anita M [email protected] Chloe P [email protected] Departments of Psychology, Pediatrics, and Psychiatry, Dalhousie University, NS, B3H 4J1, Canada2 IWK Health Centre, Halifax, NS, B3K 6R8, Canada3 Department of Pediatrics Dalhousie University and IWK Health Centre, Halifax, NS, B3K 6R8, Canada4 School of Health and Human Performance and School of Occupational Therapy, Dalhousie University, 1459 Oxford Street, Halifax, NS, B3H 4R2, Canada2005 30 8 2005 5 33 33 4 1 2005 30 8 2005 Copyright © 2005 McGrath et al; licensee BioMed Central Ltd.2005McGrath et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Because there are no child-friendly, validated, self-report measures of dyspnea or breathlessness, we developed, and provided initial validation, of three, 7-item, pictorial scales depicting three sub-constructs of dyspnea: throat closing, chest tightness, and effort.
Methods
We developed the three scales (Throat closing, Chest tightness, and Effort) using focus groups with 25 children. Subsequently, seventy-nine children (29 children with asthma, 30 children with cystic fibrosis. and 20 children who were healthy) aged 6 to 18 years rated each picture in each series, using a 0–10 scale. In addition, each child placed each picture in each series on a 100-cm long Visual Analogue Scale, with the anchors "not at all" and "a lot".
Results
Children aged eight years or older rated the scales in the correct order 75% to 98% correctly, but children less than 8 years of age performed unreliably. The mean distance between each consecutive item in each pictorial scale was equal.
Conclusion
Preliminary results revealed that children aged 8 to 18 years understood and used these three scales measuring throat closing, chest tightness, and effort appropriately. The scales appear to accurately measure the construct of breathlessness, at least at an interval level. Additional research applying these scales to clinical situations is warranted.
==== Body
Background
Dyspnea or breathlessness is a subjective phenomenon that can be perceived, regardless of the presence or absence of disease [1]. The measurement of the severity of dyspnea is challenging. The most commonly employed measure is the Borg scale and modifications thereof [2-4] even though it was initially designed to measure the effects of perceived exertion rather than dyspnea. The Borg scale has proven to be remarkably useful clinically as it correlates well with various physiologic parameters. The Borg scale uses simple, descriptive, adjectives such as slight, moderate, and severe in an open-ended scale that is, however, usually presented with numbers from 6–20 or 0–10. Studies in adults required a level of comprehension and considerable briefing of subjects, rendering these scales difficult to apply in children [5]. Throughout this paper, the word "children" is used to refer to both children and adolescents.
The other commonly used scale to rate degree of breathlessness is a Visual Analogue Scale (VAS) which usually consists of a 100-mm line with two anchors such as "not at all breathless" at one end and "maximally breathless" at the other [2,6]. Qualitative components of the sensory aspect, such as perceived "tightness" or "heaviness", which could be measured have not been seen as particularly useful for measuring severity of breathlessness, but may be related to the nature of the underlying disease [7,8]. Virtually no systematic use of the measurement of dyspnea severity has been made in children [9]. On the other hand, scales for measurement of magnitude of effort required to perform a task (such as exercise) have been described [10,11]. These measurements are particularly challenging in children because of the need for understanding and distinguishing the separate concepts of breathlessness and exertion, and differentiating the respective sensation from the affective response to it [12].
We sought to devise a pictorial scale that would encompass the full range of the perception of breathlessness by children. The aim was to develop a category scale or scales with at least ordinal properties. By intentionally avoiding an open magnitude scale like the Borg scale, it should be possible to use the scale to compare sensation within (e.g. before versus after intervention) and between individuals. Once the scales were devised, psychometric validation was undertaken to determine initial properties of the scales.
Methods
The IWK Health Center Research Ethics Board approved this research, and all participants and/or their parents signed an informed consent or assent form.
Content validity of the pictorial scales was achieved by using three focus group sessions for children to tell us how they perceived dyspnea or breathlessness. One was attended by six children with asthma, aged 8–16 years; another by five children with cystic fibrosis, aged 12–19 years; and a third consisted of 14 healthy children with no history of chronic cardiopulmonary disease, aged 8–16 years. Altogether there were 13 boys and 12 girls. Children with asthma and cystic fibrosis were targeted as they represent the commonest chronic respiratory diseases in childhood. Each session was attended by a pediatric respirologist, an occupational therapist with experience in having children use drawings to express their feelings, and a graphic artist. The children and adolescents first were asked to recall instances when they experienced breathlessness, and what images were conjured up in their minds as a result. A roundtable discussion ensued, and the children and adolescents were then asked to draw one or more picture as best they could of what they thought breathlessness would look like, The children and adolescents also described different sensations of breathlessness or dyspnea.
In consultation with the investigators, the illustrator drew pictures based on the drawings and descriptions of the children and adolescents who had been in the focus groups. The final sets of pictures were determined by consensus among the investigators. Our selection of three scales was determined by the fact that some children drew and described throat constriction and some described and drew chest tightness. Many children also commented or drew pictures of effort, 'tired' or exhaustion. In each series, we coached the illustrator to draw pictures that we thought represented breathlessness of equal intervals from no breathlessness to maximal breathlessness.
Initial psychometric validation and determination of scale properties was then carried out with 20 healthy controls (mean age 8.7 ± 2.4 years, 10 female), 29 children with asthma (mean age 10.4 ± 3.3 years, 18 female), and 30 children with cystic fibrosis (mean age 11.6 ± 3.6 years, 16 female). Figure 1 shows the age breakdown of the participants.
Figure 1 Age breakdown of the participants.
The back of each card was numbered such that the picture depicting no breathlessness was given a "1" and the picture depicting maximal breathlessness was given a "7".
Each child or adolescent completed three tasks on each set of pictures. The order of the sets was randomized. First of all, a set containing seven pictures was shuffled and the pictures randomly placed in a row in front of the child. Beginning with the left-most picture, the child was asked to rate each of the seven pictures one after another on a zero to ten scale, where zero meant not at all breathless and ten meant a whole lot of breathlessness. These ratings were done to determine whether the child was able to judge the magnitude of breathlessness represented by each picture independent of the pictures being in order. Secondly, the child was asked to lay the pictures out from left to right, in the order he or she felt represented the least to the most breathlessness. Finally, participants were asked to place each of the seven pictures on a one-metre long frame, and were asked to place the cards in relations to the severity of breathlessness using the verbal descriptors "not at all" and "a whole lot" as anchors at either end. A 100 cm ruler with 1 cm gradations was attached to the side of the frame facing the research assistant, hidden from the child. The placement recorded by the research assistant was the measurement on the ruler that lined up with a small mark on the centre of the back of the card. This allowed the research assistant to identify where the card had been placed on the VAS. In this manner, we were able to ascertain the ratings of each picture, ordinal properties of each scale (whether or not the cards were placed in rank order on the VAS) and interval properties (distance between cards).
Multivariate, between-subjects, analysis of variance tests were used to test for the main effects and interactions between age, gender, and diagnosis, on VAS placements for each pictorial set. Repeated measures analysis of variance was used to evaluate the distances along the 100 cm VAS line between each picture within a given set of seven.
Results
Development of the scales
Following the focus groups, all drawings were reviewed by the investigators and common themes were chosen for inclusion in the scale, based on the responses by children in each group. The common themes chosen were throat closing, chest tightness, and effort because these were the sub-constructs drawn and described in the focus groups.
Although negative affect was mentioned by the children, we decided not to include an affective scale as it was mentioned only in conjunction with physical sensations represented by the three constructs: throat constriction, chest tightness and exhaustion.
The illustrator, with assistance from the investigators, integrated the drawings and verbal description of the children to develop illustrations that reflected what we had been told by the children and adolescents. For example, the ropes used in the figures were depicted as rough and heavy to convey the sensation which children described as sharp, burning, or needle-like and that several children and adolescents had portrayed using a rope or cord. We decided to use seven pictures of increasing severity in each series, as a compromise between providing sufficient resolution of degrees of breathlessness without overwhelming a viewer with too many choices. Once the drawings were completed, each scale consisted of seven individual pictures on laminated 12 × 15 cm cards. The final scales are shown in Figure 2.
Figure 2 Dalhousie Dyspnea Scales.
Evaluation of the scales
Multivariate, between-subjects, analysis of variance was used to test for the main effects and interactions between age, gender, and diagnosis, on VAS placements for each pictorial set. Repeated measures analysis of variance was used to test for significance the distances along the 100 cm VAS line between each picture within a given set of seven.
Each of the three dyspnea scales (Throat closing, Chest tightness, and Effort) was initially rated on a 0 (none at all) to 10 (a whole lot) scale by each child. Figure 3 shows the mean ratings of breathlessness for each picture on the three scales, with the anchors 0 representing "no breathlessness at all", and 10 representing "the worst breathlessness imaginable". There was a consistent increase in ratings from the first to the seventh pictures.
Figure 3 Mean Rating (SE) of breathlessness for each picture in each pictorial set.
Inspection of a scatter plot of the data revealed that there was a natural break, with 6 and 7 year old children rating in a less consistent manner than older children. We divided the sample into two age groups to better examine the performance of younger participants. Data from children aged six to seven years were grouped (n = 22, mean age 6.6 ± 5 years) and compared with the remaining participants aged 8–18 years (n = 56, mean age 12.0 ± 2.9 years). Table 1 shows the percentage of pictures correctly ranked by both age groups within each scale. The number of perfect rankings (i.e. all seven pictures within a set correctly ranked) was also investigated for each age group (Table 2). Since the ranking by 6–7 year old participants did not support the requirement that the scales have ordinal properties when employed in this age group, the data from 6 and 7 year olds were excluded from subsequent analysis.
Table 1 Percentage of children who correctly ranked individual pictures within each pictorial set. n = 22, 6–7 year olds, n = 56, 8–18 year olds.
Picture no. Effort Throat narrowing Chest tightness
Age 6–7 Age 8–18 Age 6–7 Age 8–18 Age 6–7 Age 8–18
1 54.5 85.7 87.0 98.2 73.9 94.6
2 31.8 85.7 73.9 98.2 47.8 94.6
3 40.9 85.7 56.5 96.4 56.5 96.4
4 40.9 76.8 60.9 92.9 47.8 96.4
5 45.5 80.4 52.2 83.9 52.2 96.4
6 50.0 94.6 56.5 75.0 47.8 96.4
7 59.1 92.9 65.2 87.5 69.6 98.2
Table 2 Percentage of children who perfectly ranked individual pictures within each pictorial set. n = 22, 6–7 year olds, n = 56, 8–18 year olds.
Age group Effort Throat narrowing Chest tightness
6–7 22 39 43
8–18 66 73 91
The remaining group of 56 participants comprised of 14 healthy controls (mean age 9.8 ± 2.1 years), 18 children with asthma (mean age 12.7 ± 2.4 years), and 24 children with cystic fibrosis (mean age 12.8 ± 2.9 years). Children with cystic fibrosis had mean (SD) FEV1 of 75 (22)% predicted. Our children with cystic fibrosis had the full spectrum of mild to severe of breathing difficulty. Children with asthma had mean (SD) FEV1 of 103% predicted, though we do not know how low their FEV1 dropped during exercise induced bronchoconstriction, or during flare-ups. Overall, they had mild to moderate disease.
Figure 4 shows a gradual increase in the mean VAS placements when children were asked to place the pictures on a one-metre board. The mean distances between successively ranked pictures in each pictorial set were 12.5 ± 0.4 cm, 12.5 ± 1.2 cm, and 12.2 ± 0.5 cm respectively for the Chest tightness, Effort, and Throat closing pictorial scales. These distances between consecutive pictures were not significantly different.
Figure 4 Mean (SE) VAS placements for each picture in each pictorial set
Multivariate, between-subjects analysis of variance showed a main effect of underlying pulmonary disease on the sixth (p = .015) and seventh (p = .013) pictures in the Chest Tightness pictorial set. Children with asthma placed the sixth picture at 75.6 cm and the seventh picture at 89.7 cm along the VAS line. Children with cystic fibrosis and healthy children placed the sixth picture at 69.6 cm, 74.7 cm, and the seventh at 82.7 cm, 87.7 cm, respectively. No significant main effects or interactions were observed for the VAS placements of the Throat closing or Effort sets of pictures.
Discussion
All three pictorial scales were understood by children over 8 years of age and were found to have at least interval properties and be valid psychophysical measures of self report of dyspnea in children with cystic fibrosis, asthma and in healthy children older than 8 years. Only some children less than 8 years of age used the pictures in a consistent way.
Our focus groups were conducted with children 8–19 years and the validation sample included children 6–18 years. We designed the scales for use of 8–19 year olds and then challenged the scales with younger children to see if the scales were robust in this age group. We believe that the constructs are difficult for children below 8 years to use. However, a useful scale might be developed for the younger age group.
The pictures in both the Chest tightness and Throat narrowing scales were rated and ranked more correctly than the items in the Effort pictorial scale. This may have been the result having more experience with, throat constriction or chest tightness when they experience dyspnea. In other words, the tightness images may have evoked better recall of the perceived intensity of sensation that the children experience during episodes of breathlessness. The sense of breathing effort, while related to dyspnea, can occur independently of difficulty breathing, and may require more cognitive complexity to understand. In this study, the Effort pictorial scale is not as robust as the other two scales and future research will determine if it needs to be revised.
The reasons why children with asthma placed the sixth and seventh pictures of the chest tightness scale differently than the normal children and the children with cystic fibrosis is not clear. We don't know if it is a reflection of true differences, or a quirk of the way this sample used the scale.
These scales were developed with Caucasian children in our centre in Canada. The results may not apply to other cultures that may conceptualize breathlessness in different ways. Moreover, children whose breathlessness is from causes other than cystic fibrosis and asthma such as children with vocal cord dysfunction or other upper airway problems may react differently. Similarly, we did not test children who suffer breathlessness because of lack of conditioning. Future studies should include these populations.
The Dalhousie Dyspnea Scales describe the sensation of breathing effort, as well as breathing difficulty. On the other hand, they do not measure the affective response [13]. This was purposefully avoided by showing only the lungs and throat, and not showing facial features on the Effort scale.
We did not control for multiple comparisons in our analyses. This was because, for the most part, the research questions were conceptually independent. Moreover our sample size was modest and a more conservative approach would reduce the power of the statistics used.
It will take further research to determine the best way to use the scales. For example, at this point, we do not know if combining the scales or if letting children use whichever scale is most appropriate to them yields the most meaningful data and the best scaling properties.
Our study did not correlate the self-report with physical measurements. These data speak only to the psychometrics of the self report of dyspnea. Future studies should examine how these scales relate to physical measures.
Conclusion
The Dalhousie Dyspnea Scales were developed to ensure content validity and have been shown to have at least interval scale characteristics to measure breathlessness by children more than eight years old. Children readily use the scales with minimal instruction. Use of the scales in clinical studies and determination of the relationship between subjective breathlessness and objective physical stimuli is warranted.
Abbreviations
Throughout this paper the term "children" is used to refer to children and adolescents up to 18 years of age.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The study was conceived and designed by PJM & PTP with assistance from all remaining authors. The study was conducted under the supervision of PJM, PTP and conducted by CPB, PTP and AMU. Statistical analyses were conducted by PTP and CPB, with assistance from PJM. Interpretation of results were conducted by PJM, PTP, AMU. The manuscript was prepared by PJM and PTP and edited by AMU with review and assistance from all remaining authors. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The research was supported by a seed grant from Merck Frosst Canada Co., and by an operating grant from the Foundation for Clinical Research of the Lung Association of Nova Scotia. Dr. McGrath was supported by a Distinguished Scientist Award from the Canadian Institutes of Health Research and a Canada Research Chair. Elizabeth Owen was the illustrator.
==== Refs
Manning HL Schwartzstein RM Pathophysiology of dyspnea NEJM 1995 333 1547 1553 7477171 10.1056/NEJM199512073332307
Adams L Chronos N Lane R Guz A The measurement of breathlessness induced in normal subjects: individual differences Clin Sci 1986 70 131 140 3956105
Borg GAV Psychophysical basis of perceived exertion Med Sci Sport and Exerc 1982 14 377 381
Wilson RC Jones PW A comparison of the visual analogue scale and modified Borg scale for the measurement of dyspnea during exercise Clin Sci 1989 76 277 282 2924519
Lamb KL Eston RG Effort perception in children Sports Med 1997 23 139 148 9108634
Harty HR Heywood P Adams L Comparison between continuous and discrete measurements of breathlessness during exercise in normal subjects using a visual analogue scale Clin Sci 1993 85 229 236 8403792
Elliot MW Adams L Cockcroft A Macrae KD Murphy K Guz A The language of breathlessness. Use of verbal descriptors by patients with cardiopulmonary disease Am Rev Respir Dis 1991 144 826 832 1928956
Mahler DA Harver A Lentine T Scott JA Beck K Schwartzstein RM Descripitors of breathlessness in cardiopulmonary diseases Am J Respir Crit Care Med 1996 154 1357 1363 8912748
Carrieri-Kohlman V Kieckhefer G Janson-Bjerklie S Souza J The sensation of pulmonary dyspnea in school-age children Nurs Res 1991 40 81 85 2003078
Robertson RJ Goss FL Boer NF Peoples JA Foreman AJ Dabayebeh IM Millich NB Balasekaran G Riechman SE Gallagher JD Thompkins T Children's OMNI Scale of perceived exertion: mixed gender and race validation Med Sci Sports Exerc 2000 32 452 458 10694131 10.1097/00005768-200002000-00029
Williams JG Eston R Furlong B CERT: A perceived exertion scale for young children Percept Motor Skills 1994 79 1451 1458 7870530
Wilson RC Jones PW Differentiation between the intensity of breathlessness and the distress it evokes in normal subjects during exercise Clin Sci 1991 80 65 70 1846793
Carrieri-Kohlman V Gormley JM Douglas MK Paul SM Stulbarg MS Differentiation between dyspnea and its affective components West J Nurs Res 1996 18 626 642 9000871
|
16131402
|
PMC1208906
|
CC BY
|
2021-01-04 16:31:06
|
no
|
BMC Pediatr. 2005 Aug 30; 5:33
|
utf-8
|
BMC Pediatr
| 2,005 |
10.1186/1471-2431-5-33
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-851610915710.1186/1471-2458-5-85Research ArticleHealth disparities and advertising content of women's magazines: a cross-sectional study Duerksen Susan C [email protected] Amy [email protected] Laura [email protected] Annie [email protected] Janina [email protected] Xavier [email protected] Reynaldo [email protected] Maria [email protected] Brenda [email protected] Georgia Robins [email protected] Rebecca and John Moores UCSD Cancer Center, University of California-San Diego School of Medicine, 9500 Gilman Drive, La Jolla, CA 92093-0658 USA2 Graduate School of Public Health, San Diego State University, San Diego, California, USA2005 18 8 2005 5 85 85 11 10 2004 18 8 2005 Copyright © 2005 Duerksen et al; licensee BioMed Central Ltd.2005Duerksen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Disparities in health status among ethnic groups favor the Caucasian population in the United States on almost all major indicators. Disparities in exposure to health-related mass media messages may be among the environmental factors contributing to the racial and ethnic imbalance in health outcomes. This study evaluated whether variations exist in health-related advertisements and health promotion cues among lay magazines catering to Hispanic, African American and Caucasian women.
Methods
Relative and absolute assessments of all health-related advertising in 12 women's magazines over a three-month period were compared. The four highest circulating, general interest magazines oriented to Black women and to Hispanic women were compared to the four highest-circulating magazines aimed at a mainstream, predominantly White readership. Data were collected and analyzed in 2002 and 2003.
Results
Compared to readers of mainstream magazines, readers of African American and Hispanic magazines were exposed to proportionally fewer health-promoting advertisements and more health-diminishing advertisements. Photographs of African American role models were more often used to advertise products with negative health impact than positive health impact, while the reverse was true of Caucasian role models in the mainstream magazines.
Conclusion
To the extent that individual levels of health education and awareness can be influenced by advertising, variations in the quantity and content of health-related information among magazines read by different ethnic groups may contribute to racial disparities in health behaviors and health status.
==== Body
Background
Disparities in health status among racial and ethnic groups favor the Caucasian population in the United States on almost all major indicators [1-4]. African Americans have the highest annual death rates of any major racial or ethnic group from lung cancer, breast cancer, heart disease and stroke, as well as the highest incidence rates of AIDS and hypertension [5-9]. Hispanics have among the highest rates of hypertension, diabetes and AIDS, along with low rates of access to health care, diagnostic screenings and early prenatal care [1,8,10,11]. Both Blacks and Hispanics are more likely than non-Hispanic Whites to be obese and to report low levels of leisure time physical activity [5,8]. The observed disparities are only partially explained by differences in socioeconomic status [3,7,12].
Women, in particular, are generally less healthy and less likely to use health services if they belong to a minority group [13-16]. Black females born in the United States have a life expectancy 5.2 years shorter than that of their White counterparts [15]. African American women have elevated mortality rates from breast cancer, cardiovascular disease, and AIDS [5,7,16,17]. Both Black and Hispanic women have high rates of diabetes, diabetic complications and mortality, compared to Whites, and Mexican American women are particularly likely to have undiagnosed diabetes [10,13]. Relative to non-Hispanic Whites, Hispanic women also have high rates of cervical cancer, AIDS, teen pregnancy, and obesity, and are less likely to access basic health services [12,14,18,19].
The racial and ethnic imbalance in disease burden has been linked to a variety of factors such as income, education, health care access, cultural barriers, genetics, co-morbidities and environmental factors [15,20,21]. Among the many environmental factors may be a disparity in access to health-related information and ethnically aligned role models in the mass media [22]. Within the framework of the Health Belief Model and other cognitive theories of health behavior, knowledge is the essential underpinning of individual perceptions of health risks and the benefits of preventive action [23]. Repeated health-related messages in the mass media can contribute to individuals' knowledge and serve as cues, prompting individuals to take protective action [24-26]. Differential levels of awareness and knowledge about risks, symptoms and available treatments have been linked to ethnic and racial disparities in health status [13,27-32]. Manton et al. [33] noted in a report on racial health disparities that information shapes attitudes and "black attitudes toward disease, treatment, screening, and outcome may significantly affect their help-seeking behavior...." For instance, a disparity in access to basic information about breast cancer and mammography and in cues to screening action may contribute to the delayed detection of breast cancer in Black women compared to White women. That lag in detection is believed to be one reason Black women have a higher breast cancer death rate despite a lower incidence rate [16].
Social marketing theory suggests that the delivery of messages encouraging healthy behavior or other social change requires the same multi-pronged marketing methods employed in product sales [34-36]. While information alone is rarely sufficient to change behavior, it is a crucial element [24-28,30-37]. Beyond informing individuals, the mass media can play an important role in setting the public agenda, stimulating public attention to issues, and offering repeated cues to action [38,39]. Innovation diffusion theory, also known as technology transfer, holds that the mass media can be pivotal in influencing beliefs and attitudes that lead to behavior changes [37,40].
Marketing messages often are targeted to specific population segments, presumably for the purpose of increasing the company's market share among that segment of the population. Garfield reported that advertising for alcoholic beverages focused on magazines with high adolescent readership [41]. Smith and Malone explored industry's decision to focus on marketing to the gay and lesbian community as a strategy for bolstering market share [42]. Balbach et al. showed the importance of strategically placed advertising designed to secure an increased share of the African American market when they examined the marketing and advertising strategies associated with the Uptown brand of the R.J. Reynolds Tobacco Company [43].
While women access health information from a variety of sources, multiple surveys have shown lay magazines to be one key source of health information for women [44-46]. A study by Sadler et. al found that more African American women listed mass media as a key source of information on breast cancer and diabetes than listed physicians, and print media were listed three times as often as electronic media [47]. In a national Gallup poll in September 2002, 51 percent of Americans – and 56 percent of those aged 50 and above – said they got a great deal or moderate amount of health and medical information from magazines [48].
Although many Black and Hispanic women read mainstream White-oriented magazines [49], it is likely that readers give more serious consideration to, and are more influenced by, health-related content in magazines that present them with culturally aligned role models and information of personal relevance [50-52]. The few studies that have examined the health-related content of magazines by the magazines' ethnic orientations have found some significant differences in the amount of health information directed at Black female and White female audiences. Gerlach et al. found that coverage of colon cancer was greatly underrepresented in both White and Black women's magazines, compared to the risk the disease poses to women, but African American magazines contained many fewer articles on colon cancer than did White-oriented magazines [53]. Omonuwa reported more than four times as many advertisements for prescription and nonprescription drugs in White women's magazines as in those directed at Black women, without adjusting for the number of pages in the magazines [54]. In a six-year study of nutritional advertising in three magazine, Pratt and Pratt found what they called a "deluge" of alcohol advertisements in two major magazines for Black women: 62% of the food and beverage ads in Ebony and 47% in Essence were for alcohol [22]. By contrast, the White-oriented Ladies' Home Journal had only 2% alcoholic beverage ads.
This study reports on the development and testing of a methodology designed to facilitate a more in-depth qualitative and quantitative content analysis of the health information and cues provided in lay magazines [55]. For this first assessment, the researchers focused on an analysis of the advertising content of women's magazines.
Methods
The authors tested the hypothetical statement: Magazines oriented primarily toward Caucasian women contain significantly more health-promoting and less health-diminishing advertising content than those targeting African American and Hispanic women.
To test this hypothesis for health-related advertisements, assessments were made of advertising content in the four highest circulating, general interest magazines aimed primarily at Black women and the four comparable magazines aimed at Hispanic women, compared to advertising content in the four highest-circulating magazines aimed at a predominantly White readership. General interest magazines were defined as those covering a wide range of topics of interest to women; magazines more narrowly focused on fashion, beauty and relationships, such as Cosmopolitan, were excluded. Circulation and orientation were assessed using the SRDS Consumer Magazine Advertising Source. O, The Oprah Magazine, although launched by an African American celebrity, was not included among the African American magazines because the majority of the readership was White and a content review of role models pictured in the magazine revealed a mainstream rather than an African American orientation. Nor was the magazine's circulation high enough to be included among the mainstream magazines analyzed. Table 1 displays the characteristics of the magazines for which comparable information was available. While health disparities affecting the Asian-Pacific Islander community are of concern, comparable women's journals for that population were not available to be included in this analysis. This is presumed to be at least partly because the Pan Asian community is composed of many smaller ethnic groups, each with its own unique language, culture, and print media.
Table 1 Magazine readership demographics
Magazines Paid Circulationa Issues/Yeara Aver. Pages/Issue % Femalec % Whiteb % Blackb Median Agec Median Household Incomec
Mainstream
Family Circle 4,712,548 15 173 90 89 9 48.2 $49,221
Good Housekeeping 4,527,447 12 201 87 87 11 47.3 $52,609
Ladies' Home Journal 4,100,675 12 159 94 90 8 48.0 $52, 634
Woman's Day 4,257,742 17 155 94 87 12 47.7 $50,356
African American
Ebony 1,785,139 12 182 61 7 93 39.4 $38,678
Essence 1,053,484 12 171 75 6 92 35.8 $40,698
Heart and Soul 308,899 10 104 85 NA NA 37.0 $47,082
Upscale 224,649 9 124 70 NA NA 31.0 $72,500
Hispanic American
Latina (English) 230,503 12 115 94 NA NA 31.0 $45,807
Latina Style (English) 152,561 5 62 NA NA NA 30.0 $73,000
Cristina (Spanish) 102,194 12 103 NA NA NA 18–49 37% over $35,000
Vanidades (Spanish) 90,598 26 132 NA NA NA 25–54 54% over $35,000
a Sources: SRDS Consumer Magazine Advertising Source, May 2002 and Audit Bureau of Circulations, December 2001
b Source: Mediamark Fall 2000. White and Black categories include Hispanic.
c Source: magazine publishers
The sample included all issues of the 12 selected magazines published in June, July, and August 2002. During this time frame, holiday specific advertising would have been focused on American holidays (e.g. the Fourth of July) rather than culturally specific holidays like Kwanzaa or Cinco de Mayo. Publication frequency varied from bi-monthly to bi-weekly, resulting in a total of 12 issues of mainstream magazines, 10 African American, and 14 Hispanic American during the three months. These differences in publication frequency, as well as differences in the magazines' page sizes and total number of pages, were accounted for by assessing each of the magazines' total square inches of printed pages during the three-month assessment period. This allowed the investigators to make a proportional assessment, as well as a "dose intensity" assessment of the potentially healthy and potentially unhealthy advertising impact on each of the three ethnic groups. Comparisons were then made between the absolute and relative amounts of health-related ads in the mainstream magazines and in those aimed at African Americans and Hispanics.
Data were collected and analysis completed in 2002 and 2003. Abstracting guides were developed to code the size and page numbers of all advertisements and to code the advertising content into 48 categories. (Classified ads were not individually analyzed because of their small size and hence limited impact.) Of the 48 content categories, 35 were considered health-related because the advertised products had the potential to affect health. The remaining 13 were broad categories such as entertainment, retail, beauty supplies, and financial services. The 35 health-related categories included:
• 16 advertising categories related to health promoting medical treatments or products judged to have a potentially positive impact on health (i.e., drugs, vitamins, exercise equipment, sunscreen, clinical trials participation, and prevention of smoking or substance abuse). For this analysis, prescription drug ads were considered positive because they provided general health information or at least awareness about related health conditions, not because of the drugs themselves.
• 6 advertising categories promoting products with potentially negative health consequences (i.e., cigarettes, alcohol, and medical treatments with no apparent value).
• 13 food and non-alcoholic beverage advertising categories, including:
5 with potentially positive or healthy impact;
4 with potentially negative or unhealthy impact;
4 with potentially mixed health impact.
Food and beverage ads were designated as having potentially healthy or unhealthy impact based on subjective evaluations of the usual nutritive value, fat content and energy density of the primary product advertised. Therefore, all advertisements for vegetables, fruit, 100% fruit juice, milk, soymilk, yogurt, water, tea, plain coffee, plain nuts, veggieburgers, baked beans, plain bread, plain rice and non-chocolate granola bars were coded as healthy. All foods and vegetables advertised as weight-loss, low-fat or reduced-calorie, including diet sodas, also were coded as healthy. Foods and beverages coded as unhealthy included: candy, ice cream, gelatin desserts, other desserts, chips and other high-fat/low-nutrient snacks (e.g. cheese puffs), granola bars containing chocolate, fast food, non-diet sodas, juice drinks with added sugar and coffee drinks with added sugar and creamer. To avoid subjective decisions in coding, all fast food restaurant advertisements were considered unhealthy, even if relatively healthy menu items were highlighted. The advertisements were considered promotions of fast food chains, where most food and drink products sold are high-fat or high-sugar.
Some types of foods were coded as having mixed health impact because they had both healthy and unhealthy aspects (e.g., foods with high energy density but also substantial nutrient value). These included red meat/processed meat products such as hot dogs, cheeses, all cereals except plain rolled oats, all condiments not advertised as low-fat or low-calorie, macaroni and cheese mixes, and all refrigerated or frozen packaged meals that were not advertised as weight-loss or low-fat products. All condiments and cereals were assigned the neutral category of "mixed" to avoid the need for subjective distinctions by the reviewers and to simplify the process. The majority of advertisements in those categories were for relatively high-fat or high-sugar items.
Coding was also developed to quantify the photographic racial and ethnic role models presented in advertisements as White, Black, Hispanic, and Other. The standard used for counting people in the photos was to include everyone whose face was at least half visible but not photos taken from the back or of body parts other than faces. Photos of all sizes were counted equally, but the number of advertising pages with any faces also was reported to partially control for the differential impact of very differently sized faces. In the very rare cases in which there was doubt about whether a face was White or Black, the photo was coded as "other" and was excluded from analysis. In the magazines reviewed, there were almost no faces coded as "other," which included Pan Asian. Since distinguishing Hispanic faces with even a modest degree of confidence proved difficult, even for the four Hispanic research team members, the attempt to count Hispanic faces was abandoned. (Coincidentally, one of the magazines studied contained an article about the wide racial variation among Hispanics: "Will the real Latina please stand up?" Latina, July 2002, page 80).
Two reviewers independently analyzed each of the 36 magazines under study, with the pairs selected randomly from among eight researchers. The two magazines published in Spanish were reviewed by varying pairs of the five researchers literate in Spanish.
To confirm the consistent application of the content analysis guide, pilot testing and honing of the guide was done to establish a baseline inter-rater reliability rate of 95% (kappa = 0.92) and maintenance of this rate was reconfirmed throughout the data collection process. A third reviewer rechecked the coding of both primary reviewers.
When the first two reviewers disagreed and the third was unable to reach a decision, the coding was discussed at a meeting of all reviewers and senior authors, and a consensus decision was reached in all cases. In the absence of recommended guidelines for health-related advertising content, the advertising content and quantity in the mainstream magazines were used as the points of reference to which the African American and Hispanic magazines were compared.
Results
Table 2 displays the cumulative advertising space for each of the three clusters of ethnic magazines for the three-month period under investigation. After taking the number of pages, page sizes, and publication frequency into account, the proportion of magazine space devoted to all advertising was similar for the African American and mainstream magazines (approximately 50%), but lower for the Hispanic magazines (30%). Approximately half of all advertisements in mainstream magazines were health-related, more than double the proportion in the Black and Hispanic magazines.
Table 2 Space devoted to health-related advertising
Square Inches Health-Related Ads
Magazine Type Total Sq. Inches over 3 monthsa Advertising as % of Magazine Space Health-Related Ads as % of All Ad Space % Positive % Negative % Mixed
Mainstream 171,671 49.9 51.9 65.0 17.1 18.0
African American 130,666 47.3 22.4 58.0 28.7 13.3
Hispanic American 135,456 29.5 25.9 48.9 43.7 7.5
Number of Ads All Health-Related Ads
Magazine Type Number of Pages over 3 months Number of Adsb Health-Related Ads as % (N) of all Ads % Positive % Negative % Mixed
Mainstream 2,066 943 50.0 (470) 61.5 17.4 21.1
African American 1,518 684 20.5 (140) 52.9 30.0 17.1
Hispanic American 1,570 466 23.6 (110) 52.7 39.1 8.2
a Magazine pages average 84 square inches
b Some ads are more than one page
In terms of the proportion of both square inches of advertising space and the actual number of advertisements, the African American and Hispanic magazines contained fewer advertisements with potentially healthy impact and more with potentially unhealthy impact than the mainstream magazines (Table 2). Table 3 lists the types of advertisements with positive, negative, and mixed impact, clustered by the magazines' ethnic orientation.
Table 3 Number of health-related ads by topic
Mainstream N African American N Hispanic American N
Health Related Ads (N=) 470 140 110
Positive Health Related Ads
Prescription Drugs 65 20 1
Over the Counter Drugs, Vitamins, and Suppliesa 121 25 22
Medical Treatments and Procedures 2 4 2
Smoking and Drug Abuse Prevention 18 4 6
Sun Protection 5 1 0
Exercise Equipment and Facilities 1 0 1
Healthy Foods and Beverages 47 14 12
Weight Loss and Low Calorie Foods and Beverages 20 3 3
Health-Related Organizations and Charities 10 3 11
Clinical Trials Participation Promotion 0 0 0
Subtotal Positive Ads 289 74 58
Negative Health Related Ads
Cigarettes 15 11 12
Alcohol 0 12 3
Unhealthy Foods and Beverages 64 15 25
Fast Food Restaurants 3 4 1
Pseudo-Treatmentsb 0 0 2
Subtotal Negative Ads 82 42 43
Health Related Ads of Mixed Impact
Meats 14 4 5
Condiments 53 7 2
Cereals 11 2 0
Miscellaneous and Packaged Meals 21 11 2
Subtotal Mixed Ads 99 24 9
a Includes dietary supplements, vision and hearing products, vitamins, medical supplies, and drugs purchased without a prescription
b Ads for questionable "blood treatments"
The mainstream magazines contained 28% of all cigarette and alcohol ads published in the 12 magazines during the three-month study period, while they had almost two-thirds of all preventive advertisements related to smoking cessation and alcohol and drug abuse prevention (Table 3). Similar to the earlier findings by Pratt and Pratt [22], there were no alcohol advertisements in the White-oriented magazines, 12 in the African American magazines and 3 in the Hispanic magazines. Advertisements for over-the-counter drugs, vitamins, and medical supplies were placed in mainstream magazines most frequently; 72% of 168 such ads were in the mainstream magazines, 15% in African American magazines and 13% in Hispanic magazines. Except for alcohol and fast food, mainstream magazines had a higher percentage of all categories of food and beverage advertisements than either of the other two ethnic groups of magazines. In all three types of magazines, 28–30% of the food and beverage ads were for healthy and low-calorie products, but overtly unhealthy food and drink ads (other than alcohol) comprised 52% of food and beverage advertisements in Hispanic magazines, compared to 32% in Black magazines and 29% in mainstream magazines.
Of the 86 prescription drug ads in the 12 magazines during the three-month study period, 76% were placed in the four mainstream magazines. Only one prescription drug ad appeared in any of the Hispanic magazines during the three months studied. Prescription drugs and cigarettes were the products with the most multi-page ads. Table 4 displays the health conditions for which prescription drugs were advertised, illustrating a disparity in the placement of prescription drug advertising. Although hypertension disproportionately affects African Americans, ads for hypertension drugs were confined to mainstream magazines, as were most of the allergy, diabetes, arthritis and cholesterol drug ads. However, ads for birth control, HIV medications and Viagra appeared only in the African American magazines. Advertisements to promote clinical trials participation were completely absent from all magazines.
Table 4 Prescription drug ads by category
Drug Category Mainstream African American Hispanic American
Allergy 13 2 1
Diabetes 11 6
Heartburn/reflux 7
Arthritis 6
Chemo-related Anemia 5
Cholesterol 4 1
Hypertension 4
Acne 3
Hormone Replacement Therapy 3
Nail Fungus 3
Yeast Infection 3 4
Osteoporosis 2
Alzheimer's 1
Birth Control 3
HIV 3
Erectile Dysfunction (Viagra) 1
TOTAL 65 20 1
Table 5 displays the analysis of the identifiable White and Black faces. Those advertisements selling products with negative health impact in the African American magazines frequently (55%) featured faces of African Americans. In contrast, comparable health-diminishing ads in the mainstream magazines rarely (6%) included any people at all, although 58% of the health-promoting ads in mainstream magazines featured White faces.
Table 5 Racial/ethnic representation on health-related ad pages (by face count in photographs)
Mainstream African American
Positive Health-Related Ads Number of ad pages = 238* Number of ad pages = 73*
#Ad Pages with White Faces 138 (58%) 8
#Ad Pages with Black Faces 28 30 (41%)
Total # of White Faces 232 15
Total # of Black Faces 43 47
Negative Health-Related Ads Number of ad pages = 83* Number of ad pages = 42*
#Ad Pages with White Faces 5 (6%) 4
#Ad Pages with Black Faces 1 23 (55%)
Total # of White Faces 8 9
Total # of Black Faces 3 47
* The total number of ad pages includes pages with no faces or indistinguishable faces
Analysis of the specific page locations of different types of ads within the magazines yielded cells that were too small to draw conclusions with only the three-month period of evaluation used for this study. Further evaluation of this outcome measure is warranted in a larger study.
Discussion
While many factors that contribute to health disparities are extremely difficult to change, lifestyle choices can be influenced by selective advertising, marketing, and public relations. In addition to selling health-related products, advertisements can contribute to public awareness and access to information about health conditions. This can contribute to readers' perceptions of personal risk status, knowledge of how to lower risks and ability to weigh the pros and cons of selecting healthy alternatives. These are all key elements of the Health Belief Model and other theoretical frameworks. Indeed, that informative potential is one of the primary rationales for allowing direct-to-consumer advertising of prescription drugs [56].
To the extent that individual levels of health education and awareness can be influenced by advertising, variations in the quantity, quality, and content of health-related information among magazines read by different ethnic groups may contribute to disparities in health behaviors and health status. This study compared the dose intensity of health-promoting and health-diminishing magazine advertisements in African American, Hispanic and mainstream magazines. The shibboleth of marketing is that multiple messages must be received to achieve top-of-mind awareness regarding a specific message. From this perspective, the actual number of messages delivered to each cultural group is as important as the proportion of messages compared to other types of messages. In addition to the "dose intensity" of messages, the Health Belief Model suggests that the narrowness of the time frame in which social marketing messages are delivered and the alignment of the message with the intended recipient's culture should influence the intensity of the message.
In comparison to mainstream magazines, magazines specifically created for African American and Hispanic American women were smaller, less frequently published (with the exception of Vanidades), and had proportionally more potentially harmful health-related advertising and less potentially health-promoting advertising. None of the magazines in the study had any advertisements recruiting women for clinical trials or supporting the idea of participation in health-related research, indicating a missed opportunity to address a critical issue in women's health, and especially minority women's health. The limited pharmaceutical advertising in minority women's magazines further reduced those readers' exposure to information related to the benefits of clinical trials.
The finding that Black faces are more frequently used to advertise products with negative health impact than are White faces deserves further research. In the African American magazines, a higher proportion of negative health-related ads than positive ads contained Black faces. In the mainstream magazines, however, White faces were almost completely absent from negative health-related ads and appeared frequently in positive health-related ads. The possible reasons for this advertising discrepancy are beyond the scope of this paper, but the ramifications of the disproportionate exposure to negative role models are potentially significant.
Generalizations from this study must be drawn with caution. While this pilot study evaluated the advertising content of lay magazines with national readership, it included only a three-month period of publication. Some of the magazines studied did not precisely fit the inclusion criterion of "general women's magazines." Ebony is not officially a women's magazine, but is "a Black-oriented, general, picture magazine dealing primarily with contemporary topics." It was included because the magazine regularly contains articles on fashion, beauty, recipes, relationships, tips for mothers and other features typical of women's magazines, and 61% of its readers are women [49,57].
In addition, since journals focused on Pan Asian women were not included in this analysis, generalizations cannot be extended to this community. Inclusion of Pan-Asian journals might have helped distinguish between socio-economic factors and ethnicity since Pan Asian families, on average, have a higher household income than any other minority group [58]. However, data on median household income of magazine readers in this study (Table 1) do not suggest widely disparate socio-economic status between the mainstream, African American, and Hispanic American readers.
Another limitation of this study is that age disparities existed among readers of the highest-circulation women's magazines in the three racial/ethnic categories. The median age of readers of the mainstream magazines was about 10 years higher than that of the African American magazines, as reported by the publishers. Median ages were not available for two of the Hispanic magazines, but the age ranges provided indicate that the median was probably similar or slightly lower than that of the African American readers. This is a potential confounder that warrants further investigation.
One of the goals of this study was to test and refine our data collection instruments and coding guidelines. Categorizing food advertisements as having positive or negative health impact was problematic because the health value of many foods depends on the context and manner in which they are eaten. While this is especially true for the food and beverage items classified as mixed, it also applies to all other foods. Milk, for instance, was classified as a healthy beverage although drinking large quantities of whole milk can contribute to high cholesterol and obesity. A more precise system might evaluate the health impact of each ad, so that, for example, a cereal advertisement focusing on a recipe for making a high-fat dessert with the cereal would be classified as negative.
Conclusion
These findings suggest that health information disparities in the advertising content of lay magazines could contribute to health disparities among women of differing ethnic communities. Since women often make health-related and purchasing decisions that affect their family members, disparities in women's health knowledge, attitudes, and behaviors can have implications for entire families.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SD and AM were involved with study design, data collection, data analysis, and manuscript preparation. LT, AP, JL, XA, RV, and MV participated in data collection and manuscript preparation. BK oversaw the integrity of the database and assisted with mathematical calculations within the database. GS conceived of the study, secured the research funding, participated in study design, coordination, and manuscript preparation.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This research was prepared with the support of NCI Grant #R25 CA65745, CURE Grant #3 P30 CA 23100 22S2, National Institutes of Health, National Center on Minority Health and Health Disparities EXPORT Grant #P60MD00220, and the National Cancer Institute Minority Institution/Cancer Center Partnership Program Grants #U56 CA92079 and #U56 CA92081. Its content is the sole responsibility of the authors and does not necessarily represent the official beliefs of the funding agencies.
This research received the John and Lola Ross Award in the Science and Culture of Medicine, UCSD School of Medicine.
==== Refs
Plepys C Klein R Health status indicators: differentials by race and Hispanic origin Healthy People 2000 Stat Notes 1995 1 8
Keppel KG Pearcy JN Wagener DK Trends in racial and ethnic-specific rates for the health status indicators: United States, 1990-98 Healthy People 2000 Stat Notes 2002 1 16 11808619
Williams DR Race, socioeconomic status, and health. The added effects of racism and discrimination Ann N Y Acad Sci 1999 896 173 188 10681897
Hogue CJR Hargraves MA Collins KS eds. Minority Health in America: Findings and Policy Implications from the Commonwealth Fund Minority Health Survey 2000 New York, NY, Commonwealth Fund
Sowers JR Ferdinand KC Bakris GL Douglas JG Hypertension-related disease in African Americans. Factors underlying disparities in illness and its outcome Postgrad Med 2002 112 24 6, 29-30, 33-4 passim 12405099
Feldman RH Fulwood R The three leading causes of death in African Americans: barriers to reducing excess disparity and to improving health behaviors J Health Care Poor Underserved 1999 10 45 71 9989006
Bradley CJ Given CW Roberts C Disparities in cancer diagnosis and survival Cancer 2001 91 178 188 11148575 10.1002/1097-0142(20010101)91:1<178::AID-CNCR23>3.0.CO;2-S
Bolen JC Rhodes L Powell-Griner EE Bland SD Holtzman D State-specific prevalence of selected health behaviors, by race and ethnicity--Behavioral Risk Factor Surveillance System, 1997 MMWR CDC Surveill Summ 2000 49 1 60
Cancer Facts and Figures 2001 2001 Atlanta, American Cancer Society, Inc.
Black SA Diabetes, diversity, and disparity: What do we do with the evidence? Am J Public Health 2002 92 543 548 11919048
Stewart SH Silverstein MD Racial and ethnic disparity in blood pressure and cholesterol measurement J Gen Intern Med 2002 17 405 411 12133153 10.1046/j.1525-1497.2002.10524.x
Alegria M Canino G Rios R Vera M Calderon J Rusch D Ortega AN Inequalities in use of specialty mental health services among Latinos, African Americans, and non-Latino whites Psychiatr Serv 2002 53 1547 1555 12461214 10.1176/appi.ps.53.12.1547
Graham-Garcia J Raines TL Andrews JO Mensah GA Race, ethnicity, and geography: disparities in heart disease in women of color J Transcult Nurs 2001 12 56 67 11988986
Glanz K Croyle RT Chollette VY Pinn VW Cancer-related health disparities in women Am J Public Health 2003 93 292 298 12554589
Williams DR Racial/ethnic variations in women's health: the social embeddedness of health Am J Public Health 2002 92 588 597 11919058
Satcher D American women and health disparities J Am Med Womens Assoc 2001 56 131 2, 160 11759778
CDC Trends in ischemic heart disease death rates for blacks and whites--United States, 1981-1995 MMWR Morb Mortal Wkly Rep 1998 47 945 949 9832470
Fiscella K Franks P Doescher MP Saver BG Disparities in health care by race, ethnicity, and language among the insured: findings from a national sample Med Care 2002 40 52 59 11748426 10.1097/00005650-200201000-00007
Amaro H de la Torre A Public health needs and scientific opportunities in research on Latinas Am J Public Health 2002 92 525 529 11919044
Ashton CM Haidet P Paterniti DA Collins TC Gordon HS O'Malley K Petersen LA Sharf BF Suarez-Almazor ME Wray NP Street RLJ Racial and ethnic disparities in the use of health services: bias, preferences, or poor communication? J Gen Intern Med 2003 18 146 152 12542590 10.1046/j.1525-1497.2003.20532.x
Lillie-Blanton M Parsons PE Gayle H Dievler A Racial differences in health: not just black and white, but shades of gray Annu Rev Public Health 1996 17 411 448 8724234
Pratt CA Pratt CB Nutrition Advertisements in Consumer Magazines: Health Implications for African Americans Journal of Black Studies 1996 26 504 523
Becker M The Health Belief Model and Personal Health Behavior 1974 Thorofare, NJ, Slack
Meischke H Kuniyuki A Yasui Y Bowen DJ Andersen R Urban N Information women receive about heart attacks and how it affects their knowledge, beliefs, and intentions to act in a cardiac emergency Health Care Women Int 2002 23 149 162 11868962 10.1080/073993302753429022
Prentice-Dunn S Rogers RW Protection Motivation Theory and preventive health: beyond the Health Belief Model Health Education Research 1986 1 153 161
Ryan M Dunwoody S Tankard J Risk information for public consumption: print media coverage of two risky situations Health Educ Q 1991 18 375 390 1917512
Peruga A Rivo M Racial differences in AIDS knowledge among adults AIDS Educ Prev 1992 4 52 60 1543644
Peragallo NP Fox PG Alba ML Breast care among Latino immigrant women in the U.S Health Care Women Int 1998 19 165 172 9526336 10.1080/073993398246494
Peragallo NP Alba ML Tow B Cervical cancer screening practices among Latino women in Chicago Public Health Nurs 1997 14 251 255 9270290
Scroggins TG Bateman M Allen S Rittenberg CN Cancer incidence and mortality among African Americans J La State Med Soc 1994 146 147 151 8006497
Baumann LC Chang MW Hoebeke R Clinical outcomes for low-income adults with hypertension and diabetes Nurs Res 2002 51 191 198 12063418 10.1097/00006199-200205000-00008
Ayala C Greenlund KJ Croft JB Keenan NL Donehoo RS Giles WH Kittner SJ Marks JS Racial/ethnic disparities in mortality by stroke subtype in the United States, 1995-1998 Am J Epidemiol 2001 154 1057 1063 11724723 10.1093/aje/154.11.1057
Manton KG Patrick CH Johnson KW Health differentials between Blacks and Whites: Recent trends in mortality and morbidity Milbank Quarterly 1987 65 129 199 3327005
Winett LB Wallack L Advancing public health goals through the mass media J Health Commun 1996 1 173 196 10947359 10.1080/108107396128130
Siegel M Donor L Marketing public health: Strategies to promote social change 1998 Gaithersburg, Maryland, Aspen Publishers, Inc. 530
Kotler P Zaltman G Social marketing: an approach to planned social change J Mark 1971 35 3 12 12276120
Brawley EA Human services and the media : developing partnerships for change Applied social problems and intervention strategies v 2 1995 Australia, Harwood Academic Publishers viii, 379
Ross-Degnan D Changing behavior to maintain a healthy home Pediatr Infect Dis J 2000 19 S117 9 11052401 10.1097/00006454-200002000-00007
Wallack L Two approaches to health promotion in the mass media World Health Forum 1990 11 143 54; discussion 155-64 2271092
Rogers EM Diffusion of innovations 1995 4th New York, Free Press xvii, 519
Garfield CF Chung PJ Rathouz PJ Alcohol advertising in magazines and adolescent readership Jama 2003 289 2424 2429 12746369 10.1001/jama.289.18.2424
Smith EA Malone RE The outing of Philip Morris: advertising tobacco to gay men Am J Public Health 2003 93 988 993 12773366
Balbach ED Gasior RJ Barbeau EM R.J. Reynolds' targeting of African Americans: 1988-2000 Am J Public Health 2003 93 822 827 12721151
Price JH Casler SM African-American mothers' perceptions of cholesterol and its effects on their children J Natl Med Assoc 1996 88 145 150 8839029
O'Malley AS Kerner JF Johnson L Are we getting the message out to all? Health information sources and ethnicity Am J Prev Med 1999 17 198 202 10987635 10.1016/S0749-3797(99)00067-7
Kahn C How the media influences women's perceptions of health care Mark Health Serv 2001 21 12 17 11291506
Sadler GR Escobar RP Ko CM White M Lee S Neal T Gilpin EA African-American women's perceptions of their most serious health problems J Natl Med Assoc 2005 97 31 40 15719869
Newport F Poll analysis: Americans get plenty of health news on TV, but tend not to trust it
Mediamark Magazine Total Audiences -- Fall 2000 2000 New York, Mediamark Research Inc.
Bandura A Social Foundations of Thought and Action 1986 Englewood Cliffs, N.J., Prentice Hall
Bandura A The self-system in reciprocal determinism American Psychologist 1978 37 121 147
Festinger L A theory of social comparison processes. Human Relations 1954 7 117 140 10.1177/001872675400700202
Gerlach KK Marino C Weed DL Hoffman-Goetz L Lack of colon cancer coverage in seven women's magazines Women Health 1997 26 57 68 9472955
Omonuwa SC Health disparity in black women: lack of pharmaceutical advertising in black vs. white-oriented magazines J Natl Med Assoc 2001 93 263 266 11491276
Berger AA Media Research Techniques 1998 , Sage Publications Inc. 184 9577214
Wofford JL Pinson JA Folmar SJ Moran WP Health-related messages in consumer magazine advertising J Gen Intern Med 1995 10 488 490 8523150
Ebony Audience Profile: Adults 18+ 2001 Los Angeles, Johnson Publishing Co. Inc.
Current Population Survey, 2001, 2002, and 2003 Annual Social and Economic Supplements: Income 2002
|
16109157
|
PMC1208907
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Aug 18; 5:85
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-85
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-861610915910.1186/1471-2458-5-86Research ArticleFalls in young, middle-aged and older community dwelling adults: perceived cause, environmental factors and injury Talbot Laura A [email protected] Robin J [email protected] Erica K [email protected] E Jeffery [email protected] Graduate School of Nursing, Uniformed Services University of the Health Sciences, 1335 East West Highway, Silver Spring, Maryland 20910, USA2 Holy Cross Hospital, 1500 Forrest Glen Road, Silver Spring, Maryland 20910, USA3 The Henry M. Jackson Foundation, 1401 Rockville Pike, Suite 600, Rockville, Maryland 20852, USA4 Intramural Research Program, National Institute on Aging, National Institutes of Health, Harbor Hospital, 3001 South Hanover Street, Baltimore, Maryland 21225, USA2005 18 8 2005 5 86 86 17 1 2005 18 8 2005 Copyright © 2005 Talbot et al; licensee BioMed Central Ltd.2005Talbot et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Falls in older people have been characterized extensively in the literature, however little has been reported regarding falls in middle-aged and younger adults. The objective of this paper is to describe the perceived cause, environmental influences and resultant injuries of falls in 1497 young (20–45 years), middle-aged (46–65 years) and older (> 65 years) men and women from the Baltimore Longitudinal Study on Aging.
Methods
A descriptive study where participants completed a fall history questionnaire describing the circumstances surrounding falls in the previous two years.
Results
The reporting of falls increased with age from 18% in young, to 21% in middle-aged and 35% in older adults, with higher rates in women than men. Ambulation was cited as the cause of the fall most frequently in all gender and age groups. Our population reported a higher percentage of injuries (70.5%) than previous studies. The young group reported injuries most frequently to wrist/hand, knees and ankles; the middle-aged to their knees and the older group to their head and knees. Women reported a higher percentage of injuries in all age groups.
Conclusion
This is the first study to compare falls in young, middle and older aged men and women. Significant differences were found between the three age groups with respect to number of falls, activities engaged in prior to falling, perceived causes of the fall and where they fell.
==== Body
Background
Falls are a health problem in need of preventive intervention and research [1] with prior studies focused on older adults [2]. In older people, falls are a risk factor for disability and frailty, and exacerbate the disablement process [3,4]. Numerous studies have reported on the circumstances surrounding falls in older adults including descriptive studies detailing falls causation, related injuries, and risk factor determination [5,6]. This information on falls has also led to the development and testing of various preventive interventions for older adults including preventive screening for those at risk, strength and balance training, home hazard assessments with modifications, and Tai Chi [5]. In contrast to studies in older people, little attention is given to falls in young and middle-aged adults where falls represent a risk for injury with related expenses and potential interference with work and family [7,8]. Publications are lacking that describe the incidence, prevalence and circumstances surrounding falls in these age groups. This descriptive study examines and compares falls in young, middle and older aged men and women. The focus is on the faller's activity, environment, and perception as to what caused the fall.
Methods
Sample
Participants included 1497 volunteers from the Baltimore Longitudinal Study of Aging (BLSA), a prospective study of the National Institute on Aging, initiated in 1958 as an open panel cohort of participants recruited primarily from the Baltimore-Washington, DC area. Participants return for evaluation at one to five year intervals for 1–3 days of physiological and psychological tests at the Gerontology Research Center (GRC). BLSA participants are community residing, predominantly upper middle class and health conscious. A falls questionnaire was completed by 757 men and 740 women aged 20 to 92 years from 1996 to 2001 with 96.5% scoring < 7 on the Blessed Information-Memory-Concentration Test [9] and none having a diagnosis of dementia at that visit. The young age group (20–45 years) included 292 participants, the middle aged group (46–65 years) had 616 participants and the older aged group (> 65 years) included 589 participants.
Descriptive variables
Physical activity was based on a self-report of 97 activities performed over the previous 2 years and converted into metabolic equivalents of resting oxygen consumption (MET-minutes) per 24 hours (see Talbot et. al. [10] for details). Body mass index (BMI) was calculated from weight in kilograms divided by height in meters squared. The Center for Epidemiologic Study Depression (CES-D) Scale, a validated self-report instrument, was used to measure depressive symptoms [11,12].
Instruments
The History of Falls questionnaire is a 17-item survey developed in 1996 by Dr. Ann Myers based on an extensive literature review, a panel of experts, and her research on falls [13]. The questionnaire queried the circumstances surrounding a fall, specifically, activities prior to falling, perceived cause, environmental factors, and a description of injuries. The participant could check multiple responses for contributing environmental hazards and their perceived cause for falling. A specific definition of a fall was not used in this questionnaire.
For these analyses, participants were classified as fallers if they had at least one fall in the past two years. Only one fall per faller (the most serious fall) was used in the analyses. If injured in the fall, participants were then prompted with a list of possible injuries.
Activities prior to falling were categorized according to type of activity: (1) Ambulation – included walking, turning and standing; (2) Transferring – included getting in/out of bed, chair, wheelchair or car, getting on/off toilet, climbing stool, chair, or ladder, and falling off stool or chair; (3) Running; (4) Sports – included exercising, dancing, cycling; (5) Stairs/curb; and (6) Other – such as reaching overhead and bending down.
Perceived causes of falling were grouped into: 1) accident/environment related – such as slipped, missed seat, was bumped/pushed, slid off surface, bumped into object, or furniture/equipment broke; 2) collapse episode – included passing out or legs gave way; 3) dizziness/vertigo/weakness – included dizziness, light-headed, weakness or paralysis; 4) balance/gait impairment – included tripped/stumbled, quick movement, or lost balance; and 5) other – which contained causes not listed on the questionnaire and circumstances where the faller was uncertain of the cause.
Environmental factors were grouped into: 1) wet surface/slippery footwear; 2) uneven surface/steps – includes steps of unequal height, and indistinguishable surface colors; 3) objects on surface/rug – includes tripped/slid on rug; 4) external forces – furniture/stair/step/railing broke and something moved or bumped the participant; 5) icy surface; and 6) other – which included eyeglasses, darkness/dimness or environment, glare, "other", or "uncertain". Participants were given three choices as to where the fall occurred: 1) inside the home, 2) inside a building but not at home, or 3) outside.
Injuries sustained from a fall were grouped into four treatment categories: fractures, treated injury, untreated injury, and no injury. For these analyses, the most serious injury reported was included. Treated injury was any injury that caused the participant to see a physician, and/or go to the emergency room or hospital. Fractures, though treated, were categorized separately from treated injury. An untreated injury was any injury caused by a fall where the participant did not see a physician, or go to the emergency room or hospital. The no injury category included participants who reported falling but did not report injuring themselves as a result of the fall. This categorization of injury has been used previously [14] and represents the severity of the injury as initially perceived by the participant.
Procedure
The History of Falls questionnaire was mailed to the participants and completed prior to visiting the GRC. A nurse practitioner or physician's assistant reviewed the questionnaire with the participant for accuracy. The study was approved by the Institutional Review Board at Johns Hopkins Bayview Medical Center, and all participants gave signed, informed consent.
Data analysis
Descriptive statistics of mean, standard deviation, percentages and frequencies were calculated for all variables. The Pearson chi-square test was used to examine associations between gender and categorical variables related to falling. Analysis of variance was used to determine activity differences between fallers for the three age groups. The proportion of injuries sustained from falling was grouped according to the severity of the injury. The significance was set at the 0.05 level. Statistical analyses were performed using SPSS (version 10.1).
Results
Participant characteristics are shown in Table 1. Participants are generally well educated, Caucasian, married, overweight (BMI > 25), screened negative for depression and report being in good to excellent health. Young fallers were more active than older fallers > 65 years (F(2,288) = 3.16, p > 0.04).
Table 1 Sample characteristics of the population
Total Population
(N = 1497) Young (20–45 years)
(N = 292) Middle (46–65 years)
(N = 616) Old (> 65 years) (N = 589)
Fallers
(N = 54) Non-Fallers
(N = 238) Fallers
(N = 131) Non-Fallers
(N = 485) Fallers
(N = 205) Non-Fallers
(N = 384)
Age (years) 59.5 ± 16.5 34.0 ± 6.9 35.6 ± 6.8 55.9 ± 5.5 55.1 ± 5.6 78.2 ± 7.1 75.2 ± 6.7
Gender
(% male) 50.6% (757) 46.2% (23) 47.9% (114) 31.3% (41) 44.4% (220) 51.7% (106) 65.9% (253)
Race
(% Caucasian) 72.1% (962) 65.2% (30) 56.2% (109) 69.2% (83) 66.2% (286) 88.4% (167) 81.3% (287)
Marital Status (% subj.)
Single
(Single, Divorced) 23.5% (299) 37.5% (15) 31.5% (58) 27.7% (31) 30.7% (126) 14.4% (27) 12.3% (42)
Married 66% (842) 62.5% (25) 62.5% (115) 67.0% (75) 67.2% (276) 62.0% (116) 68.9% (235)
Widowed 10.5% (134) NONE 6.0% (11) 5.4% (6) 2.2% (9) 23.5% (44) 18.8% (64)
Annual Income
(% > $65,000) 57.3% (718) 57.5% (23) 54.9% (100) 61.8% (68) 66.3% (268) 44.2% (80) 53.4% (176)
College Graduates (%) 75.9% (933) 84.2% (32) 80.1% (141) 76.2% (80) 75.3% (298) 71.0% (130) 76.1% (252)
Body Mass Index
(kg*m-2)
Male 27.6 ± 3.8 (739) 27.9 ± 4.1 (21) 27.5 ± 3.7 (113) 28.0 ± 4.5 (41) 27.8 ± 3.6 (219) 27.1 ± 3.6 (100) 27.7 ± 4.1 (245)
Female 26.3 ± 4.9 (685) 24.2 ± 3.6 (30) 24.9 ± 4.9 (120) 27.6 ± 5.2 (82) 27.0 ± 5.4 (246) 25.7 ± 3.9 (92) 26.0 ± 4.6 (115)
CES-D Depression Score1
(% > 15) 9.5% (136) 18.5% (10) 12.2% (28) 3.3% (4) 7.9% (37) 11.6% (21) 9.7% (36)
Self-Reported Health
(% Excellent or Good) 93.7% (1152) 100.0% (41) 96.7% (175) 90.8% (99) 98.5% (391) 84.0% (152) 91.9% (294)
Activity Level
(MET-minutes/24 hours) 2228 ± 660 (1092) 2403 ± 528 (36) 2339 ± 521 (172) 2218 ± 492 (95) 2163 ± 487 (356) 2160 ± 546 (160) 2262 ± 977 (273)
Values are mean ± SD except where indicated
1CES-D = Center for Epidemiologic Study Depression Scale; Score > 15 indicates significant depressive symptoms
Twenty-six percent of the participants fell in the previous two years (390 fallers of 1497 participants). One or more falls occurred in 18.5% of young adults (54 fallers of 292 young participants), 21% of middle-aged adults (131 fallers of 616 middle-aged participants) and 35% of older adults (205 fallers of 589 older participants). Significant differences were found between women and men (χ2 = 10.27; p < 0.001) and between age groups ((χ2 = 39.41; p < 0.001). Older men and women were more likely to report a fall in the previous two years than middle-aged and young men and women.
Table 2 details the responses to the questionnaire in relation to their most serious fall. The activity most frequently cited as causing the fall was ambulation in all age-groups and both genders. Younger participants tended to fall more with running, older participants while walking. The young participants fell more outdoors, while the percentage of falls indoors increased from the middle to the older age group. Participants' perceived the cause of the fall to be balance/gait impairment and accidents/environment related. From the middle to old age group, the reporting of accidents/environment related falls declined while balance/gait impairment increased. Figure 1 illustrates the severity of the most serious injury suffered (by treatment groups) based on age group and gender.
Table 2 Factors preceding the most serious fall
Total Cohort Men Women
Young
(N = 54) Middle
(N = 131) Old
(N = 205) Young
(N = 23) Middle
(N = 41) Old
(N = 106) Young
(N = 31) Middle
(N = 90) Old
(N = 99)
ACTIVITIES PERFORMING PRIOR TO THE FALL*
Ambulation 31.5(17) 45(58) 56.5(113) 13(3) 42.5(17) 58.3(60) 45.2(14) 46.1(41) 54.6(53)
Transferring 9.3(5) 10.9(14) 9(18) 8.7(2) 17.5(7) 8.7(9) 9.7(3) 7.9(7) 9.3(9)
Running 20.4(11) 6.2(8) 5.5(11) 21.7(5) 5(2) 4.9(5) 19.4(6) 6.7(6) 6.2(6)
Sports/exercise/dancing/ bicycling 22.2(12) 13.2(17) 6.5(13) 39.1(9) 10(4) 10.7(11) 9.7(3) 14.6(13) 2.1(2)
Stairs/curb 7.4(4) 11.6(15) 9(18) 13(3) 12.5(5) 7.8(8) 3.2(1) 11.2(10) 10.3(10)
Other1 9.3(5) 13.2(17) 13.5(27) 4.3(1) 12.5(5) 9.7(10) 12.9(4) 13.5(12) 17.5(17)
PERCEIVED CAUSE OF THE FALL **
Accident/Environment 37(20) 30.2(38) 15.8(32) 39.1(9) 32.5(13) 16.2(17) 35.5(11) 29.1(25) 15.5(15)
Collapse Episode 5.6(3) 6.3(8) 3.5(7) 13(3) 10(4) 4.8(5) 4.7(4) 2.1(2)
Dizziness/Vertigo/Weakness 5.6(3) 1.6(2) 2.5(5) 1(1) 9.7(3) 2.3(2) 4.1(4)
Balance/Gait Impairment 38.9(21) 49.2(62) 61.9(125) 39.1(9) 47.5(19) 62.9(66) 38.7(12) 50(43) 60.8(59)
Other/Uncertain2 13(7) 12.7(16) 16.3(33) 8.7(2) 10(4) 15.2(16) 16.1(5) 14(12) 17.5(17)
ENVIRONMENTAL FACTORS LEADING TO THE FALL***
Wet surface/slippery footwear 9.3(4) 17(16) 9.5(14) 11.1(2) 13.8(4) 7.6(6) 8(2) 18.5(12) 11.6(8)
Uneven surface/steps 20.9(9) 25.5(24) 27(40) 16.7(3) 24.1(7) 32.9(26) 24(6) 26.2(17) 20.3(14)
Objects on surface/rug 9.3(4) 7.4(7) 16.2(24) 5.6(1) 6.9(2) 15.2(12) 12(3) 7.7(5) 17.4(12)
External forces 7(3) 4.3(4) 2(3) 5.6(1) 6.9(2) 2.5(2) 8(2) 3.1(2) 1.4(1)
Icy surface 20.9(9) 12.8(12) 14.2(21) 22.2(4) 24.1(7) 13.9(11) 20(5) 7.7(5) 14.5(10)
Other3 32.6(14) 33(31) 31.1(46) 38.9(7) 24.1(7) 27.8(22) 28(7) 36.9(24) 34.8(24)
Values are percentages (frequency)
*Activities Performing Prior to The Fall (Total Cohort: χ2 = 31.01; p < 0.001; Men:χ2 = 31.51; p < 0.001; Women: χ2 = 17.44; p > 0.05).
**Perceived Cause (Total Cohort:χ2 = 29.03; p < .001; Men: χ2 = 15.61; p < 0.05; Women: χ2 = 15.54; p < 0.05).
***Environmental Factors (Total Cohort:χ2 = 11.52; p > 0.05; Men: χ2 = 8.04; p > 0.05; Women: χ2 = 10.34; p > 0.05).
1Other represents responses of "reaching overhead", "bending down", "uncertain", or "other"
2Other/Uncertain represents responses of "other" or "uncertain" as to the cause of the fall.
3Other represents responses of "new eyeglasses" and/or "darkness/dimness of environment" and/or "glare from lights"
Figure 1 Frequency comparison of injury groups in fallers by age and gender. The X-axis displays the breakdown of all fallers into age and gender groups. The Y-axis shows the total number of participants. Percentages of each injury type per age/gender groups are displayed parenthetically on the graph bars.
Most frequently, the young group injured their wrist/hand, knees and ankles; the middle-aged injured their knees; and the older group injured their head and knees. Of those who fell, there were no differences in the percentage of injuries across age groups for men and women (Men χ2 = 0.30; p = 0.86; Women: χ2 = 0.75; p = 0.69). Women reported a higher percentage of injuries and women in the youngest age group reporting more injuries (80.6% or 25 of 31 fallers) than any other group. Table 3 displays frequencies of falls and most serious injury type.
Table 3 Frequency of falls and injuries* in fallers by age group and gender over two years
Total Cohort Men Women
Young
(N = 292) Middle
(N = 616) Old
(N = 589) Young
(N = 137) Middle
(N = 261) Old
(N = 359) Young
(N = 155) Middle
(N = 355) Old
(N = 230)
Total Number of Falls (percentages)
0 238(81.5) 485(78.7) 384(65.2) 114(83.2) 220(84.2) 253(70.5) 124(80.0) 265(74.5) 131(57)
1 25(8.6) 74(12.0) 99(16.8) 12(8.8) 22(8.4) 50(13.9) 13(8.4) 52(14.6) 49(21.3)
2 12(4.1) 24(3.9) 53(9.0) 3(2.2) 10(3.8) 31(8.6) 9(5.8) 14(3.9) 22(9.6)
3 3(1.0) 9(1.5) 25(4.2) 0 2(0.8) 12(3.3) 3(1.9) 7(2.0) 13(5.7)
4 3(1.0) 11(1.8) 16(2.7) 1(0.7) 2(0.8) 8(2.2) 2(1.3) 9(2.5) 8(3.5)
5+ 11(3.8) 13(2.1) 12(2.0) 7(5.1) 5(1.9) 5(1.4) 4(2.6) 8(2.2) 7(3.0)
Total Fallers 54(18.5) 131(21.3) 205(34.8) 23(16.8) 41(15.7) 106(29.5) 31(20.0) 90(25.3) 99(43.0)
If Fell, Injured?
No 15(27.8) 40(30.5) 60(29.3) 9(39.1) 16(39.0) 37(34.9) 6(19.4) 24(26.7) 23(23.2)
Yes 39(72.2) 91(69.5) 145(70.7) 14(60.9) 25(61.0) 69(65.1) 25(80.6) 66(73.3) 76(76.8)
Most Serious Injury Sustained from the Fall (percent of injured fallers)
Cut/Laceration1 6(15.4) 11(12.1) 29(20.0) 1(7.1) 2(8.0) 17(24.6) 5(20.0) 9(13.6) 12(15.8)
Bruise/Hematoma2 11(28.2) 25(27.5) 40(27.6) 4(28.6) 7(28.0) 14(20.2) 7(28.0) 18(27.3) 26(34.2)
Sprain or Strain3 18(46.2) 25(27.5) 19(13.1) 8(57.1) 7(28.0) 12(17.4) 10(40.0) 18(27.3) 7(9.2)
Fracture 3(7.7) 15(16.5) 31(21.3) 1(7.1) 6(24.0) 14(20.3) 2(8.0) 9(13.6) 17(22.4)
Other4+ 1(2.6) 15(16.5) 26(17.9) 0 3(12.0) 12(17.4) 1(4.0) 12(18.2) 14(18.4)
Values are frequencies (percentages)
*most serious injury sustained only
1 Out of all Cut/Lacerations, 34 were Untreated and 12 were Treated Injuries
2 Out of all Bruise/Hematomas, 51 were Untreated and 25 were Treated Injuries
3 Out of all Sprain/Strains, 26 were Untreated and 36 were Treated Injuries
4 Out of all Other Injuries, 7 were Untreated and 35 were Treated Injuries
+Other also includes responses of "burn" and not reported
Discussion
This is the first study to compare falls in young, middle and older community-dwelling men and women. The main finding is the high frequency of fallers in all age groups and the frequency of injury. We found significant differences between the three age groups with respect to number of fallers, activities leading to falling, perceived cause and location of the fall. Injuries across the age groups and between genders were not significantly different. Women had a higher percentage of fallers than men regardless of age group consistent with both retrospective [15,16] and prospective studies [17].
Activities causing the fall
Significant age differences were found in activities leading to falling among men (p < 0.001), but not women (p = 0.07) (Table 2). The highest percentage of individuals fell while ambulating regardless of age group with young people falling while participating in sports, exercise or running. Also, vigorous activities contributed to falls in all groups except older women. Our results contrasted with Tideiksaar [18] where the majority of community-dwelling older aged persons (average age of 84 years) fell while transferring (43%) and walking (28%).
In older fallers [19], the majority of falls occurred while walking on a level surface during ordinary daily activities in the absence of hazardous behavior. In a Japanese study, falls occurred most frequently while walking [20]. Speechley and Tinetti [21] found in a one year follow up of community dwellers over 75 years that active individuals were more likely to fall during displacing activities (climbing ladders, engaged in sports, or on stairs) while the frail fell during routine non-displacing daily activities.
Although these comparison studies are of older adults, the functional status varies greatly in such a broad age group. This has led some researchers to compare their results by factors of functional status and/or independence [21,22]. Koski et al. [22] found that while major injuries in the independent older persons were more likely to be a result of running or jumping activities, no activity was related to falls resulting in a major injury in the less functionally independent older people in their sample. This difference in risk factors for falling and being injured in older persons by functional status further complicates comparing studies when there is no or different means to identifying functional status of the study sample.
Perceived causes
Most older fallers reported tripping as the cause for their fall [18,23], with a large percentage unable to give any reason in a one year retrospective study [24]. No information was found comparing young, middle and older fallers' perceived cause of their fall.
In the present study, of those that fell, age group differences were found for their perceived cause of falling (table 3). Gait and balance problems were the most frequently cited cause in all age and gender groups. Accidents/environment reasons were cited more frequently in the young (37%; 20 of 54 fallers) and the middle-aged groups (30%; 38 of 126 fallers) compared to older fallers (15.8%; 32 of 202 fallers).
Environmental factors
As people age, changes in balance, proprioception, muscle strength, attention and vision make compensation to environmental hazards more difficult [25-27]. Environmental hazards are more likely to precipitate the fall in healthy active older individuals than in their frailer counterparts [21,28]. In a Swedish hospital-based study, slipping on ice or snow caused 3.5 injuries per 1000; older women and young men (20–29 years) had the highest incidence of injury with half being fractures [29]. A study of hospital-treated fractures in Norway revealed 90% of fractures in adults over 65 were caused by falls; the fracture risk increased fivefold in months with ice and snow [30]. No other studies were found that examined environmental factors and falls in younger and middle-aged adults, other than studies regarding specific occupational hazards. In our study, the environmental factor contributing to the highest percentage of falls was uneven surfaces/steps (Table 2).
Where fall occurred
Older fallers are more likely to fall indoors than outdoors [24]. The very old (average age 84 years) in Tideiksaar's study [18] fell most often at home (81%). Campbell et al [17] found in a prospective study of participants (over 69 years) from a rural setting no gender difference in fall rate; however men were more likely to fall outside and at more intense levels of activity. A Danish study of older adults using a 24-hour recall method found that one specific component of the home environment had a strong association with falls; vinyl or linoleum on the floors [15]. Active participants tend to fall outside the home and the frail older participants fell inside the home [31]. There was a significant difference in our population with regard to where participants fell by age group (χ2 = 19.699; p < .001). The highest percentage fell outside (62.5%: 244 of 390 fallers). Less than 4% (2 of 54 fallers) of the younger group in our study reported falling in their own home. This compares to the middle-aged and older groups where 17.6% (23 of 131 fallers) and 29.3% (60 of 205 fallers) fell in their homes respectively.
With regard to gender, the percentage of falls occurring inside the home increased with age. Thirty-seven percent (36 of 97 women fallers) of the older women and 23% (24 of 106 men fallers) of the older men reported that they fell inside their own home.
Injury severity
Tinetti et al [32] found in a prospective study that 24% of the falls in community dwelling older aged persons resulted in serious injury and 6 percent had fractures. Gryfe et al [33] found 17.5% of falls in older aged persons living in an apartment setting suffered serious injury (6.1% fractures and 11.4% soft tissue injury). In the Netherlands, community dwellers (> 70 years) who fell within one year previous to a postal survey, 8% reported a major injury and 49% had minor injuries [31]. While in another study (mean age = 84), 18% of fallers sustained an injury, and 14% of the injuries were fractures [18].
In a trauma study, a 6% fall injury rate was found for the frail older people contrasted to a 0.4% rate for 40 year olds [34]. In a trauma registry study in Connecticut, serious injury was seen in 32% of participants > 65 years of age and only 11% for the younger group [1]. No information was found comparing type and severity of injury in young, middle and older age groups of fallers.
Our population reported a higher percentage (70.5%; 275 of 390 fallers) of injuries from their falls than previous prospective [32,33,35] and retrospective studies [20,31]. Part of the difference may relate to different definitions for categorizing the injury. For example, Tinetti et al [32] defined a serious injury as all nonvertebral fractures, injuries resulting in an emergency room stay > 24 hours, hospital admission, bed rest > 48 hours, or activity restriction > 72 hours. This reflects a more restrictive definition than used in our report, which would lead to lower rates. Koski et al [22] defined a major injury by fractures, joint dislocations, lacerations needing sutures, and intracranial injuries. This prospective study found that 32% of all fallers suffered a major injury. However, by their definition, major injury included fractures. Breaking down their major injury category into fractures and other major injuries, a fracture rate of 15.5% is revealed showing it be very similar to our results [22]. Tideiksaar [18] also reported a fracture rate that was comparable to the observations in this study.
Speechley and Tinetti [21] found 22% of falls by vigorous participants resulted in serious injuries compared to 6% of falls by frail participants, suggesting active older people are involved in activities that would cause greater injury if a fall occurred. This higher rate of serious injury in the active individuals could also be a result of increased exposure to risky activities and environments or also a higher speed and force of an activity leading to a fall. Koski et al [22] found among more active participants the primary activity connected to major injurious falls was running or jumping; however, the highest rate of major injury was in the more frail participants.
Previous studies have focused on older populations and not the entire adult lifespan as in this report. We found that severity of injury, as defined by treatment, across age and gender was not significantly different. The percentage of falls resulting in fractures increased from young (5.56%; 3 of 54 young fallers) to middle (11.5%; 15 of 131 fallers) to oldest (15.1%; 31 of 205 fallers) age groups. In the middle and older age groups, the occurrence of treated and untreated injuries resulting from falls was similar. A higher percentage of women reported any injury from the fall. Interestingly, younger women reported a higher percentage of injuries (80.6%; 25 of 31 fallers) than any other age group. Middle-aged men (14.6%; 6 of 25 injured fallers) joined older men (13.2%; 14 of 69 injured fallers) and older women (16.2%; 17 of 76 injured fallers) as the groups reporting the highest percentage of fractures.
Anatomic site of injury
The most common injury sites for older (> 65) adults were face (17%), upper extremity (16%), hand/wrist (14%), and head (12%), with lower extremity and ankle/foot less common [36]. Another study of injury in older adults (51% were falls) reported legs (43%) or arms (39%) most frequently occurring [37].
No study has described common injury sites from a fall across the adult age span. Our study showed that the young group injured their wrist/hand most frequently (20%) while middle-aged adults injured their knees (19%) and older adults injured their head (15%).
Study limitations
Study limitations include the cross-sectional nature of the design, the self-report of injury, and memory recall of falls spanning 2-years. In any study of falls, the incidence of self-reported falls is underestimated and less reliable than observed events with participants more likely to remember falls that resulted in injury. Also, we did not control for the number and type of chronic conditions that make the older population more susceptible to falls. The severity categories used here limit how well we are able to compare our results with other studies that utilized more restrictive criteria.
The BLSA represents a well educated and health conscious group of individuals who are more likely to practice health promoting behaviors and therefore more likely to remember, report falls, and seek medical help than the general population. As a result, one potential source of bias would be a higher rate of reported falls and subsequent injury than in studies with a more diversified population, thus limiting comparisons.
Conclusion
Falling is a health risk meeting all criteria for prevention: high frequency, evidence of preventability and high burden of morbidity. While falls in young adults may be considered a lifestyle issue related to sports and vigorous activity, more attention may need to be directed to the middle-aged adult where approximately one in four reported falling at least once in a two year period. Falling often results from multiple concurrent problems including environmental and behavioral factors as well as disease processes. Middle-aged adults progressively start to show higher incidences of diseases and medication use, along with lower levels of physical activity, and physiological changes that begin to alter postural stability. Physical exercise and balance training have shown some benefit in older people, and may be of benefit in middle-age, but this has not been studied [38]. Events in this middle-aged group are likely to predispose individuals for the higher risks that lead to falls in later years.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
LAT and EJM were the key authors in the conception, design, coordination, and drafting of the manuscript as well as the analysis and interpretation of the data. RJM participated in the design and interpretation of the data and helped in drafting the manuscript. EKW contributed substantively by revising the manuscript critically for intellectual content and participating in the interpretation of data and the revision of the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to express their appreciation to the volunteers in the Baltimore Longitudinal Study of Aging.
Funding source
Intramural Research Program of the National Institute on Aging
Disclaimer
The views expressed are those of the author and do not reflect the official policy or position of USUHS, the Department of Defense, or the United States Government.
==== Refs
Sterling D O'Conner JA Bonadies J Geriatric Falls: Injury Severity is high and disproportionate to mechanism. Journal of Trauma-Injury Infection and Critical Care 2001 50 116 119 10.1097/00005373-200101000-00021
Fried LP Epidemiology of Aging Epidemiologic Reviews 2000 22 95 106 10939013
Verbrugge LM Jette AM The disablement process Soc Sci Med 1994 38 1 14 8146699 10.1016/0277-9536(94)90294-1
Stalenhoef PA Diederiks JP deWitte LP Schiricke KH Crebolder HF Impact of gait problems and fall on functioning in independent living persons of 55 years and over: A community survey Patient Educ Couns 1999 36 23 31 10036557 10.1016/S0738-3991(98)00071-8
Gillespie L Gillespie W Robertson M Lamb S Cumming R Rowe B Interventions for Preventing Falls in Elderly People Cochrane Database of Systematic Reviews 2005 CD00635
McClure R Turner C Peel N Spinks A Eakin E Hughes K Population-Based Interventions for the Prevention of Fall-Related Injuries in Older People Cochrane Database of Systematic Reviews 2005 CD03487
Versteegen GJ Kingma J Meijler WJ Duis HJ Neck sprain not arising from car accidents: a retrospective study covering 25 years. European Spine Journal 1998 7 201 205 9684952 10.1007/s005860050056
Lipscomb HJ Dement JM Li L Nolan J Patterson D Work-related injuries in residential and drywall carpentry. Appl Occup Environ Hyg 2003 18 479 488 12746070 10.1080/10473220301422
Thal LJ Grundman M Golden R Alzheimer's Disease: a correlational anaysis of the Blessed Information-Memory-Concentration Test and the Mini-Mental State Exam Neurology 1986 36 262 264 3945395
Talbot LA Metter EJ Fleg JL Leisure-time physical activities and their relationship to cardiorespiratory fitness in healthy men and women 18-95 years old. Medicine & Science in Sports & Exercise 2000 32 417 10694126 10.1097/00005768-200002000-00024
Radloff LS The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement 1977 1 385 401
Ross CE Mirowsky J Components of depressed mood in married men and women. The Center for Epidemiologic Studies' Depression Scale American Journal of Epidemiology 1984 119 997 1004 6731436
Myers AH Young Y Langlois JA Prevention of falls in the elderly Bone 1996 18 87S 101S 8717552 10.1016/8756-3282(95)00384-3
Cherpitel CJ Alcohol consumption and injury in the general population: from a national sample. Drug and Alcohol Dependence 1994 34 217 224 8033759 10.1016/0376-8716(94)90159-7
Larsen ER Mosekilde L Foldspang A Correlates of falling during 24 h among elderly Danish community residents Preventive Medicine 2004 39 389 398 15226051 10.1016/j.ypmed.2004.02.001
Yasumura S Haga H Nagai H Suzuki T Amano H Shibata H Rate of falls and the correlates among elderly people living in an urban community in Japan. Age Ageing 1994 23 323 327 7976782
Campbell AJ Borrie MJ Spears GF Jackson SL Brown JS Fitzgerald JL Circumstances and consequences of falls experienced by
a community population 70 years and over during a prospective study. Age Ageing 1990 19 136 141 2337010
Tideiksaar R Reducing the risk of falls and injury in older persons: Contribution of a falls and immobility clinic. Facts and Research in Gerontonogy 1996 162 183
Wild D Nyack USL Isaacs B Prognosis of falls in old people at home. Journal of Epidemiology and Community Health 1981 35 200 204 7328380
Niino N Tsuzuku S Ando F Shimokata H Frequencies and circumstances of falls in the National Institute for Longevity Sciences, Longitudinal Study of Aging (NILS-LSA). J Epidemiol 2000 10 S90 S94 10835834
Speechley M Tinetti M Falls and injuries in frail and vigorous community elderly persons. Journal of the American Geriatrics Society 1991 39 46 52 1987256
Koski K Luukinen H Laippala P Kivela SL Risk factors for major injurious falls among the home-dwelling elderly by functional abilities. A prospective population based study Gerontology 1997 44 232 238 9657085 10.1159/000022017
Blake AJ Morgan K Bendall MJ Dallosso H Ebrahim SB Arie TH Fentem PH Bassey EJ Falls by elderly people at home: prevalence and associated factors. Age Ageing 1988 17 365 372 3266440
Prudham D Evans JG Factors associated with falls in the elderly: A community study. Age Ageing 1981 10 141 146 7270321
Rawsky E Review of the literature on falls among the elderly. Image J Nurs Sch 1998 30 47 52 9549941
Adler-Traines M Falls in the senior population. Physical Therapy Forum 1994 9 14 17
Chen HC Schultz AB Ashton-Miller JA Giordani B Alexander NB Guire KE Stepping over obstacles: Dividing attention impairs performance of old more than young adults. J Gerontol A, Biol Sci Med Sci 1996 51 M116 122 8630704
Northridge ME Nevitt MC Kelsey JL Link B Home hazards and falls in the elderly: the role of health and functional status. Am J Public Health 1995 85 509 515 7702114
Bjornstig U Bjornstig J Dahlgren A Slipping on ice and snow--elderly women and young men are typical victims. Accid Anal Prev 1997 29 211 215 9088360 10.1016/S0001-4575(96)00074-7
Ytterstad B The Harstad injury prevention study: the characteristics and distribution of fractures amongst elders, an eight year study International Journal of Circumpolar Health 1999 58 84 95 10429338
Stalenhoef PA Diederiks JP Knottnerus JA Witte LP Crebolder HR The construction of a patient record-based risk model for recurrent falls among elderly people living in the community. Family Practice 2000 17 490 496 11120721 10.1093/fampra/17.6.490
Tinetti ME Speechley M Ginter SH Risk factors for falls among elderly persons living in the community. New England Journal of Medicine 1988 319 1701 1707 3205267
Gryfe CI Amies A Ashley MJ A longitudinal study of falls in an elderly population: I. Incidence and morbidity Age Aging 1977 6 211 220
Fife D Barancik JI Chatterjee BF Northeastern Ohio Trauma Study: II. Injury rates by age, sex, and cause American Journal of Public Health 1984 74 473 478 6711722
Cummings SR Nevitt MC Kidd S Forgetting Falls: The limited accuracy of recall of falls in the elderly. Journal of the American Geriatrics Society 1988 36 613 616 3385114
Shapiro MJ Partridege RA Jenouri I Micalone M Gifford D Functional decline in independent elders after minor traumatic injury. Academic Emergency Medicine 2001 8 78 81 11136156
Carter SE Campbell EM Sanson-Fisher RW Gillespie WJ Accidents in older people living at home: A community-based study assessing prevalence, type, location and injuries. Austr NZ J Public Health 2000 24 633 636
Metter EJ Yim-Chipless PK Musiol R Talbot LA Balance Training: Preventing Falls in Older Adults The American Journal of Medicine and Sports 2003 5 94 99
|
16109159
|
PMC1208908
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Aug 18; 5:86
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-86
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-871611149710.1186/1471-2458-5-87Research ArticleHigh risk of HIV in non-brothel based female sex workers in India Dandona Rakhi [email protected] Lalit [email protected] Juan Pablo [email protected] Anil G [email protected] Sam [email protected] Fiona [email protected] Stefano M [email protected] ASCI FPP Study Team [email protected] Health Studies Area, Centre for Human Development, Administrative Staff College of India, Hyderabad, India2 Division of Health Economics and Policy, National Institute of Public Health, Cuernavaca, Mexico3 Research and Evaluation Unit, International HIV/AIDS Alliance, Brighton, UK4 CIDE, Mexico City, Mexico2005 20 8 2005 5 87 87 22 12 2004 20 8 2005 Copyright © 2005 Dandona et al; licensee BioMed Central Ltd.2005Dandona et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Heterosexual contact is the most common mode of HIV transmission in India that is largely linked to sex work. We assessed the non-use of condoms in sex work and with regular sex partners by female sex workers (FSWs), and identified its associations that could assist in planning HIV prevention programmes.
Methods
Detailed documentation of various aspects of sex work, and sexual behaviour with regular sex partners, was done through confidential interviews for 6648 FSWs in 13 districts in the Indian state of Andhra Pradesh. Multivariate analysis was done to understand condom non-use with clients.
Results
5010 (75.4%), 1499 (22.5%), and 139 (2.1%) FSWs were street-, home-, and brothel-based, respectively. Of the total 6648 FSWs, 6165 (92.7%) had penetrative vaginal/anal sex with at least one client in the last 15 days, and of these 2907 (47.2%; 95% CI 41.2–53.2%) reported non-use of condom with at least one of her last three clients. Lack of knowledge that HIV could be prevented (odds ratio 5.01; 95% CI 4.38–5.73), no access to free condoms (odds ratio 3.45; 95% CI 2.99–3.98), being street-based as compared with brothel-based (odds ratio 3.36; 95% CI 1.87–6.04), and no participation in FSW support groups (odds ratio 2.02; 95% CI 1.50–2.70) were the most significant predictors of condom non-use with clients. Other associations included lower social support, lower income, age >24 years, illiteracy, and living in medium-size urban or rural areas. Of the 2582 who had penetrative sex with regular sex partner within the last 7 days, 2428 (94%; 95% CI 92.1–95.9%) had not used condom at last sex, and 1032 (41.8%) had neither used condom consistently with clients nor with regular sex partner.
Conclusion
About half the FSWs do not use condom consistently with their clients in this Indian state putting them at high risk of HIV infection. Non-brothel-based FSWs, who form the majority of sex workers in India, were at a significantly higher risk of HIV infection as compared with brothel-based FSWs. With their high vulnerability, the success of expansion of HIV prevention efforts will depend on achieving and sustaining an environment that enables HIV prevention with the non-brothel based FSWs.
==== Body
Background
Around the world, the number of people living with HIV continues to rise despite the fact that effective prevention strategies exist. India has the largest number of people living with HIV, an estimated 5.1 million in the year 2003, after South Africa [1,2]. Heterosexual contact has been estimated to be the most common mode of transmission of infection in India, and six Indian states have been categorised as high prevalence states because HIV prevalence in these states exceeds 5% among the high-risk individuals and 1% among the women attending antenatal clinics [1]. In these six states, HIV is estimated to be transmitted through heterosexual sex to a large degree and is linked to sex work in four states of Andhra Pradesh, Karnataka, Maharashtra, and Tamil Nadu, and through injecting drug use in the other two states of Manipur and Nagaland [1,2].
Epidemiologically, great majority of new HIV infections in Asia occur in individuals who are at high risk – sex workers and their clients, men who have sex with men, and injecting drug users, and their immediate long-term sex partners [3-5]. Increasing prevalence of HIV in sex workers is an indication of increasing probability of a generalised epidemic [2]. A high prevalence of HIV in female sex workers (FSWs) has been reported recently from some parts of India, including the state of Andhra Pradesh [6,7] This occurrence means that adequately resourced efforts focused on achieving good coverage among those individuals at high risk of acquiring or transmitting HIV may prevent further spread of HIV in broader population. There is evidence that the HIV prevention programmes for FSWs can be highly effective in preventing HIV transmission [8-11]. Recent modelling to assess the impact of four types of interventions on prevention of HIV transmission in India has suggested FSW interventions that promote use of condoms in addition to other safe sex practices to be the most effective in preventing HIV transmission as compared with interventions focussing on treatment of sexually transmitted infections, prevention of mother-to-child transmission and provision of the highly active antiretroviral therapy [12].
Of the 835 government-supported targeted intervention programmes for high-risk individuals in India, 199 (23.8%) target FSWs and 171 (20.5%) target truckers, and the remaining target migrant workers, street children, prisons, men who have sex with men, intravenous drug users, and others [13]. These interventions follow a comprehensive approach to reduce HIV transmission and include behaviour change communication, counseling and provision of health care support, treatment for sexually-transmitted infections, and creation of an enabling environment to facilitate behaviour change [13]. One of the main foci of behaviour change in the HIV prevention efforts for FSWs is encouraging correct and consistent use of condom between them and their clients [2,8-11,13-16], as condoms reduce the risk of HIV transmission [17,18].
In this background, we assessed the non-use of condom for penetrative sex for a large sample of FSWs in the state of Andhra Pradesh, which is one of the high HIV prevalence states in India. This study was carried out as part of an impact assessment study of the Frontiers Prevention Project (FPP). The FPP aims to reduce the spread of HIV within the population through provision of HIV interventions in a geographically defined area that reduce risk behavior and STI prevalence among FSWs, men who have sex with men, and people living with HIV/AIDS by working in close collaboration with these population groups. FPP is being implemented in India (the state of Andhra Pradesh), Ecuador and Cambodia. We report data on condom use in sex work for penetrative sex between FSWs and their clients, and between FSWs and their regular sex partners in Andhra Pradesh, India.
Methods
The objectives of this baseline study were to document the socio-demographic and sex work characteristics of FSWs, analyse these data to identify issues that needed particular attention for prevention of HIV and other sexually transmitted infections, and compare these baseline data later with a follow-up study to assess the impact of the FPP. The methods relevant to this paper are mentioned below.
This study was approved by the Ethics Committees of the Administrative Staff College of India, Mexico's National Institute of Public Health, the International HIV/AIDS Alliance, and by the Indian Health Ministry's Screening Committee, Indian Council of Medical Research, New Delhi. Permissions were obtained from the Andhra Pradesh State AIDS Control Society, the agency coordinating HIV/AIDS control activities in the state, to carry out the study.
Study area
Forty geographic sites in 13 districts of the Telangana and Rayalseema regions of Andhra Pradesh state were identified where access to FSWs was considered feasible through non-governmental organisations having links with them. Each geographic site consisted of one or more close-by locations (cities/towns/villages) where FSWs were accessible. The total number of locations included in the 40 geographic sites were 72, of which 25 were rural and 47 were urban of various sizes, according to the Census of India definitions. [19]. The total sample size required of FSWs at the 40 geographic sites was estimated as 6,500 to detect a significant change in the various aspects of high-risk sexual behaviour between the baseline and follow-up studies. The sample size of the FSWs in each site was aimed to be proportional to their estimated number and type in that site, which was based on enumeration with the help of FSWs.
Data collection
Data collection questionnaire was developed by an international team with multidisciplinary background through review of worldwide literature including previously used questionnaires, focus group discussions and in-depth interviews with FSWs to better understand the local context in Andhra Pradesh, and pre-pilot studies were conducted to capture a variety of issues related to the socio-economic context of FSWs, the sexual practices between FSWs and their clients and with regular sex partners, and awareness about HIV and sexually transmitted infections. An international technical advisory group provided input regarding the refinement of questionnaire. The questionnaire was developed in English, was translated in Telugu, the local language, following which it was back-translated into English in order to ensure accurate and relevant meaning and intent of the questions. Extensive training of the interviewers was done by a variety of survey experts and FSW representatives in order to address technical and ethical issues as well as to promote cultural sensitivity.
Data were collected between July 2003 and April 2004. At each study location, FSW facilitators helped contact and recruit FSW respondents more than 15 years of age for this study. Standardised procedures were followed for contacting respondents, which included approaching them with the help of FSW facilitators (peers) and convincing them of the confidentiality of the interviews. Written informed consent for participation was obtained from each respondent. Trained interviewers did one-to-one interviews confidentially in private settings that were selected in consultation with FSW facilitators. The names of respondents were not recorded and hence cannot be linked to the data. The data collection process in the field involved supervision of the work of interviewers by a Quality Control Supervisor and a Field Manager in each of the two field teams. Data were entered in an LSD (Sistemas Integrales, Santiago, Chile) database, and all data entries made by a data entry operator were fully checked by another operator to minimise errors in data entry.
Data analysis
SPSS software was used for data analyses, and the different types of FSWs were defined as:
• Street-based FSWs if they primarily solicited their clients on streets (such as cinema, park, bus-stand, railway station, hotel / lodge) and provided services at hotel/lodge or a place of client's choice.
• Home-based FSWs if they primarily solicited their clients at home either directly or through a mediator and provided services at their homes.
• Brothel-based FSWs if they primarily solicited clients through an agent (such as pimp, madam) or mediator and provided services at a brothel. Brothel was defined as a place of sex work with at least 2 FSWs working under control of an agent.
The main outcome variable assessed in this analysis was the no or inconsistent use of condom for penetrative vaginal/anal sex between FSWs and their clients. Non-use of condom with all the last 3 clients or at least with one of the last 3 clients was considered as no or inconsistent condom use for this analysis. The 95% confidence intervals of these estimates of condom non-use were adjusted for the design effect of cluster sampling, based on intra cluster variance for these variables [20]. Univariate and multivariate analyses were done to understand the association of no or inconsistent use of condom for penetrative vaginal/anal sex with clients with other characteristics to identify those that may play a significant role in determining the use of condom. In the multiple logistic regression model, the effect of each category of a multi-categorical variable was assessed by keeping the first or the last category as reference. All the variables were introduced simultaneously in the model. Possible interactions between different variables in the model were assessed, where necessary. Use of condom for the last penetrative vaginal/anal sex between FSWs and their regular sex partners was also assessed. Regular sex partner was defined as a man who was not a client and with whom the FSW had sexual contacts.
Results
FSW characteristics
A total of 7251 FSW were contacted of whom 6648 (91.7%) participated in the study. Among these, 5010 (75.4%), 1499 (22.5%), and 139 (2.1%) were street-, home-, and brothel-based FSWs, respectively. The age range of FSWs was 16 to 54 years with mean age of 27.3 years, 2698 (40.6%) were currently married, 2833 (42.6%) were previously married, 1117 (16.8%) were never married, 4966 (74.7%) had no schooling, and 3105 (46.7%) were also involved in work other than sex work.
Sex work
Details of sex work are summarised in Table 1 for the different types of FSWs. Among the 5851 (88.%) FSWs who had worked in the last 7 days from the day of interview, the number of paying clients ranged from 1 to 49 with the mean number of paying clients being 7.5 in those 7 days. The income from sex work in these 7 days ranged from Rs. 10 to 13000, and the mean income per day was Rs. 96 (US$ 2.1). Street-based FSWs had lower income from sex work as compared with the home- and brothel-based FSWs.
Table 1 Distribution of variables related to sex work for the different types of FSWs.
Type of female sex workers*
Variable Variable categories Street-based (N = 5010) Number (% of N) Home-based (N = 1499) Number (% of N) Brothel-based (N = 139) Number (% of N) Total (N = 6648) Number (% of N)
Age at starting sex work (years) 12–15 191 (3.8%) 356 (23.7%) 33 (23.7%) 580 (8.7%)
16–19 955 (19.1%) 394 (26.3%) 43 (30.9%) 1392 (20.9%)
20–24 1643 (32.8%) 334 (22.3%) 40 (28.8%) 2017 (30.3%)
25–29 1475 (29.4%) 273 (18.3%) 15 (10.8%) 1763 (26.5%)
30–34 578 (11.5%) 109 (7.3%) 5 (3.6%) 692 (10.4%)
> = 35 168 (3.4%) 32 (2.1%) 3 (2.2%) 203 (3.1%)
Duration of being in sex work (years) 1 year or less 916 (18.3%) 174 (11.6%) 23 (16.5%) 1113 (16.7%)
1.1 – 2.0 1060 (21.2%) 195 (13.0%) 18 (12.9%) 1273 (19.1%)
2.1 – 3.0 892 (17.8%) 234 (15.6%) 20 (14.4%) 1146 (17.2%)
3.1 – 4.0 629 (12.6%) 186 (12.4%) 16 (11.5%) 831 (12.5%)
4.1 – 5.0 567 (11.3%) 196 (13.1%) 23 (16.5%) 786 (11.8%)
More than 5 946 (18.9%) 514 (34.3%) 39 (28.1%) 1499 (22.5%)
Number of months practiced sex work in the last 12 months 0–6 409 (8.2%) 59 (3.9%) 12 (8.7%) 480 (7.2%)
7–9 583 (11.6%) 127 (8.5%) 8 (5.8%) 718 (10.8%)
10–12 4016 (80.2%) 1312 (87.6%) 118 (85.5%) 5446 (82%)
Number of paying clients in the last 7 days None 638 (12.7%) 151 (10.1%) 8 (5.8%) 797 (12.0%)
1–2 631 (12.6%) 156 (10.4%) 5 (3.6%) 792 (11.9%)
3–7 2159 (43.1%) 495 (33.0%) 28 (20.1%) 2682 (40.3%)
8–14 1322 (26.4%) 471 (31.4%) 46 (33.1%) 1839 (27.7%)
15–28 248 (5.0%) 220 (14.7%) 49 (35.3%) 517 (7.8%)
More than 28 12 (0.2%) 6 (0.4%) 3 (2.2%) 21 (0.3%)
Income in the last 7 days from sex work (in Indian Rupees) None 637 (12.7%) 151 (10.1%) 9 (6.5%) 797 (12.0%)
250 or less (US $5.5 or less) 1688 (33.7%) 292 (19.5%) 14 (10.1%) 1994 (30.0%)
251 – 500 1551 (31.0%) 360 (24.0%) 29 (20.9%) 1940 (29.2%)
More than 500 1132 (22.6%) 696 (46.4%) 87 (62.6%) 1915 (28.8%)
Participation in FSW support group Yes 433 (8.6%) 152 (10.1%) 15 (10.8%) 600 (9.0%)
No 4576 (91.4%) 1347 (89.9%) 124 (89.2%) 6047 (91.0%)
Family aware of sex work Yes 1157 (23.1%) 954 (63.6%) 69 (49.6%) 2180 (32.8%)
No 3836 (76.6%) 542 (36.2%) 69 (49.6%) 4447 (66.9%)
Refused to answer 14 (0.3%) 3 (0.2%) 1 (0.7%) 18 (0.3%)
*The total of sub-categories may not always be 5010, 1499, and 139 due to a few missing values for street-, home-, brothel-based FSWs, respectively.
Of the 6648 FSWs, 1594 (23.9%; 95% CI 19.4–28.6%; design effect 20.2) reported having never used condom. Among the 5010 FSWs who reported having ever used condom, 2942 (58.7%) reported having access to free condoms and 2468 (83.9%) of these had received condoms for free within the last 30 days from the day of interview. Non-governmental organisations were the main source of free condoms (71.6%) followed by clinic/hospital (10.3%). The other sources reported were condom outlet box, another FSW/peer educator, and pimp/madam.
Client characteristics
Data were documented for each FSW on her last three clients (if she had that many clients) within the 15 days from the day of interview. Among the 6648 FSWs, 6,171 (92.8%) had at least one client in the last 15 days and the remaining 477 (7.2%) FSWs did not engage in sex work in those 15 days. Detailed data were documented on the last three clients for 5472 (82.4%) FSWs, on last two for 408 (6.1%), and on one client only for 288 (4.3%). In total, data were available on 17529 clients of 6171 FSWs who had at least one client in the last 15 days. Client characteristics as reported by the FSWs are summarised in Table 2.
Table 2 Distribution of client characteristics as reported by the FSWs.
Variable Variable categories Number (%) (N = 17529)
Type of client* New 12047 (68.7%)
Regular 5459 (31.1%)
Do not remember 23 (0.2%)
Age of client Young 8639 (49.3%)
Middle-aged 8468 (48.3%)
Old 272 (1.6%)
Cannot say 150 (0.8%)
Marital status of client Single 4785 (27.3%)
Married 9358 (53.4%)
Do not know 3386 (19.3%)
Economic status of client Poor 3265 (18.6%)
Average 9649 (55.0%)
Wealthy 2535 (14.5%)
Cannot say 2080 (11.9%)
Truck driving as profession of the client Yes 1461 (8.3%)
No 13217 (75.4%)
Cannot say 2851 (16.3%)
Client highly intoxicated with alcohol/ drugs Yes 1015 (5.8%)
No 16447 (93.8%)
Do not remember 67 (0.4%)
*New client was the one who visited the FSW for the first time, and regular client was the one who visited the FSW repeatedly.
Use of condom between FSWs and their clients
Of the 6171 FSWs who had at least one client in the last 15 days, 6165 (99.9%) had had penetrative vaginal/anal sex with at least one client. This penetrative sex was predominantly vaginal with only 12 (0.2%) and 49 (0.8%) reporting anal and oral sex with clients – the anal/oral sex was in addition to vaginal sex. 2907 (47.2%; 95% CI 41.2–53.2%; design effect 23.0) FSWs had either not used condom at all or not used with all the clients with whom penetrative sex was done (for a maximum of the last 3 clients on whom data were available). The proportion of no or inconsistent use of condom with clients was 53.7%, 30.2% and 13.3% for the street-, home- and brothel-based FSWs, respectively.
With multiple logistic regression analysis (Table 3), the highest odds ratio predicting for no or inconsistent use of condom for penetrative vaginal/anal sex were for FSWs who did not know that HIV could be prevented, followed by those who did not have access to free condoms in the last 30 days, who did not participate in FSW support group, street- and home-based FSWs, and those who had low social support score. The other variables significantly associated with condom non-use are shown in Table 3. We also assessed the interactions between some variables in another logistic regression model. Knowledge that HIV can be prevented interacted significantly with access to free condoms in the last 30 days, education level of FSW, and rural-urban area where the FSW was sampled from (p < 0.001); and there was also significant interaction between participation in FSW support group and social support score (p < 0.001). The level of knowledge that HIV can be prevented varied significantly among the FSWs among the 40 geographic sites ranging from 14.1% to 95.2%, and there was a direct linear correlation between this knowledge and consistent use of condom (p < 0.001) (Figure 1).
Table 3 Association of select variables with no or inconsistent use of condom for penetrative vaginal/anal sex by FSWs with their clients in multiple logistic regression.
Variable* Variable categories† Total who had penetrative sex with at least one client (6,128)‡ Number who reported no or inconsistent use of condom (%) Odds of having no or inconsistent use of condom (95% CI)
Knowledge that HIV can be prevented Yes 3321 840 (25.3%) 1.00
No 2807 2067 (73.6%) 5.01 (4.38–5.73)
Access to free condoms in the last 30 days Yes 2468 501 (20.3%) 1.00
No 3660 2406 (65.7%) 3.45 (2.99–3.98)
Participation in FSW support group Yes 566 77 (13.6%) 1.00
No 5561 2829 (50.9%) 2.02 (1.50–2.70)
Type of sex worker Street-based 4599 2468 (53.7%) 3.36 (1.87–6.04)
Home-based 1394 421 (30.2%) 2.66 (1.46–4.86)
Brothel-based 135 18 (13.3%) 1.00
Social support score§ 1.00 – 2.50 1306 836 (64%) 2.60 (2.17–3.12)
2.51 – 3.50 2518 1387 (55.1%) 2.27 (1.95–2.64)
>3.50 2304 684 (29.7%) 1.00
Income in the last 7 days (Rupees) >501 1903 515 (27.1%) 1.00
251 – 500 1929 946 (49%) 1.31 (1.09–1.57)
250 or less 2295 1445 (63%) 1.66 (1.35–2.04)
Age group (years) 16 – 24 2292 875 (38.2%) 1.00
25 – 34 3034 1533 (50.5%) 1.29 (1.11–1.51)
35 or more 802 499 (62.2%) 1.69 (1.33–2.14)
Family aware of sex work Yes 2064 655 (31.7%) 1.00
No 4061 2250 (55.4%) 1.32 (1.13–1.53)
Rural-urban area where the FSW was sampled from Rural 1345 544 (40.4%) 1.45 (1.19–1.77)
Urban small 680 266 (39.1%) 1.07 (0.84–1.35)
Urban medium 2621 1460 (50.2%) 1.73 (1.47–2.04)
Urban large 1482 637 (43%) 1.00
Number of clients in last 7 days 7 or less 3768 2062 (54.7%) 1.20 (0.88–1.63)
8 – 14 1829 738 (40.3%) 1.25 (0.93–1.67)
> = 15 531 107 (20.2%) 1.00
Marital status Never married 1072 257 (24%) 1.00
Currently married 2483 1327 (53.4%) 1.26 (1.01–1.59)
Other# 2573 1323 (51.4%) 1.07 (0.85–1.35)
Education level of FSW Illiterate 4552 2428 (53.3%) 1.32 (1.13–1.55)
Literate 1576 479 (30.4%) 1.00
Duration of being in sex work (years) 0–5 4773 2336 (48.9%) 1.00
>5 1355 571 (42.1%) 0.98 (0.82–1.17)
*Variables listed in descending order of effect on outcome variable.
†Some categories of variables combined based on initial iterations that showed similar values for outcome variable in order to increase the power of the analysis.
‡Data on condom use was available for 6128 (99.4%) FSWs; the total of sub-categories may not always be 6128 due to a few missing values.
§The social support score for each respondent was averaged for responses to 7 questions used for this score, which documented whether the respondent could count on someone for money, going to doctor, talking about problems, food or place to stay, abuse by anyone, abuse by client, and client's refusal to use condom; this score ranged from 1 to 5, with 1 indicating least social support and 5 indicating maximum social support; the three suggested categories indicate low, medium and high social support, respectively.
¶Urban small were towns with population < 50,000; urban medium were towns/cities with population 50,001 – 200,000; urban large were cities with population more than 200,000; this classification was done based on Census of India data for each sub-site.
#Other includes separated, divorced and widowed.
Figure 1 Relation between FSWs having knowledge that HIV could be prevented and inconsistent or no use of condom with clients for penetrative vaginal/anal sex in the 40 geographic sites.
Dynamics of condom use
A total of 17517 (99.9%) clients had penetrative vaginal/anal sex with FSWs. Of these clients, 10860 (61.9%) used condom, 6418 (36.6%) did not use, and information on condom use was not available for 239 (2.5%) clients. Among the 10860 clients who had used condom – FSWs had suggested using condom to 6816 (62.8%) clients, and of these 2697 (39.6%) clients had to be convinced to use condom; 2070 (19.1%) clients themselves had suggested using condom; and both the FSWs and clients had suggested using condom with 1971 (18.1%) clients. Condom was supplied by FSWs for 7531 (69.3%) clients. Among the 6418 clients who did not use condom for penetrative sex, FSWs had condoms available with them at the time of sex for 830 (12.9%) of these clients.
Considering only the last client for each FSW with whom she had penetrative vaginal/anal sex, the clients who did not use condom were more likely to be middle-aged or old (p < 0.001), married (p < 0.001), and with lower economic status (p < 0.001) as reported by the FSWs.
Use of condom between FSWs and their regular sex partners
A total of 3642 (54.8%) FSWs reported having a regular sex partner (husband, lover, boyfriend) of whom 2582 (70.9%) had had penetrative sex with him within the last 7 days from the day of interview. 2428 (94%; 95% CI 92.1–95.9%; design effect 4.3) FSWs had not used condom for the last penetrative sex with their regular sex partners. The major reason cited for not using condom was do not use because he is my regular partner (1802, 73.9%) followed by not aware of condom (364, 14.9%), he does not like it (178, 7.3%), want to have children (118, 4.8%), and others (not mutually exclusive).
Of the 2464 FSWs who had had penetrative vaginal/anal sex with at least one client in the last 15 days and also had penetrative vaginal/anal sex with their regular sex partner in the last 7 days, 1032 (41.8%) FSWs had neither used condom consistently with clients nor had used with regular sex partner and 1102 (44.7%) had used condom consistently with clients but not used with regular sex partner. The former group tended to be street-based FSW (p < 0.001), having no knowledge about HIV prevention (p < 0.001), more than 24 years of age (p = 0.003), ever married (p < 0.001), and illiterate (p < 0.001) as compared with the FSWs who had used condom consistently with their clients but had not used with their regular sex partner.
Discussion
The national HIV prevalence in India is still estimated to be low but there is a serious HIV epidemic in six states where the majority of the infections are acquired sexually [1,2]. We have reported data on condom non-use for a large sample of FSWs from the state of Andhra Pradesh, where HIV prevalence among FSWs was estimated to be 16%, ranging from 8–41% in seven surveillance sites in 2004 [21]. These data on the different types of FSWs from various urban and rural areas of the state using a standard methodology would allow relatively broader understanding of issues that are relevant for HIV prevention programmes, and for promoting use of condom in particular.
The FSWs who participated in this study may not be representative of all FSWs as they were recruited through FSW facilitators suggesting a bias towards those who are better connected with their peers, and hence the results should be interpreted within this limitation. It is also possible that some respondents would have over-reported use of condom, therefore the actual use of condom may be lower than that reported.
Non-use of condom
Nearly half of the FSWs had not used condom consistently with all the clients in this study. In another assessment in this state in year 2000, only 25.9% and 8.5% FSWs had reported use of condom with all clients in the preceding one-month [7]. From the perspective of HIV prevention programmes, we found substantial differences in use of condom between the different types of FSWs, with the street-based FSWs nearly 3.5 times less likely to use condoms with clients as compared with the brothel-based FSWs. The street-based FSWs are also the highest in proportion among the different types of FSWs in India, including Andhra Pradesh [22].
Knowledge that HIV can be prevented was a strong predictor of consistent use of condom for penetrative sex between FSWs and their clients. Another variable strongly associated with consistent condom use was access to free condoms. These findings reinforces that knowledge about HIV and access to free condoms are vital for promoting increased use of condoms in FSWs in India. In addition, the FSW demographic characteristics that predict inconsistent use of condom with their clients were – age more than 24 years, currently married, illiterate, lower income, poor social support, family unaware of sex work, and no participation in FSW support groups. These characteristics can be used to define target groups for HIV intervention programmes. Some client characteristics associated with non-use of condom were also identified which can be used to promote condom use by these vulnerable clients.
Condom use for penetrative sex with the regular sex partner was negligible, and 41.8% FSWs had neither used condom consistently with clients and nor had used with their regular sex partner in the last sexual act. It may be difficult to promote use of condom between FSWs and their regular sex partners as married and cohabitating couples, in general, use condom less frequently because of various reasons [23]. However, as is highlighted by these data a significant proportion of FSWs have unprotected sex with clients and their regular sex partners, and it is possible that the regular sex partners of FSWs are not necessarily monogamous. Therefore, HIV transmission from regular sex partners may increasingly contribute to the overall spread of HIV as the use of condom increases with the clients. These data have also highlighted the dynamics of condom use ranging from FSW convincing the client to use condom, the client convincing FSW to use condom, to condom being available with FSW but not used at the time of sex with client. Although the overall use of condom was low, when it was used it was primarily at the suggestion of FSW, though a significant proportion of clients also asked for the condom to be used. Further research is necessary to better understand whether the demand from client for use of condom or the ability of FSW to convince the client to use condom is more effective in promoting condom use.
HIV prevention
Context and environment
Effective HIV prevention requires strategies and policies that help reduce vulnerability of FSWs to HIV infection by creating a social, legal and economic environment in which prevention is possible. In India, as elsewhere, creating an enabling environment for behaviour change among the individuals who are at a higher risk of acquiring or transmitting HIV is an integral part of the HIV interventions [13]. We discuss the findings of this study within this context for HIV prevention in sex workers in India.
It is estimated that about 1.1% of the adult women in India could be engaged in sex work [24], most of whom are estimated to be non-brothel based [22]. The non-brothel based FSWs, especially street-based, were at a higher risk of HIV infection as compared with brothel-based FSWs in this study. Because the Indian society discriminates against FSWs as immoral women, not many of them acknowledge that they are sex workers. Only one-third of FSWs in this study reported that their families were aware of their sex work. FSWs with lower social support score were relatively less likely to use condom consistently. These women, for the most part, remain inaccessible to HIV prevention programmes, thereby undermining the efforts of HIV prevention. Acknowledgment of being a sex worker is more of an issue with the non-brothel based FSWs as compared with the brothel-based FSWs because being in a brothel can be interpreted as an acknowledgment that she is a sex worker.
This also makes it difficult to organize non-brothel based FSWs as a group that could be empowered to protect themselves and participate in the HIV prevention efforts. A model from India has been reported to be successful in empowering FSWs in Sonagachi, Kolkatta [11,25]. However, there are also examples of not so successful peer-based HIV interventions in brothel-based FSWs in Mumbai, India who were not interested in conducting education sessions for their peers after being trained, and their madams did not allow them to leave the brothels to conduct education programmes [26]. Only 9% of FSWs in our study reported participation in FSW support group, and these women reported higher use of condom with clients. Even though the number of brothel-based FSWs in this study was small, these data highlight that the risk behaviour for HIV was lower in the brothel-based FSWs as compared with the others, thereby, suggesting that sex workers working together as a group can promote condom use with clients.
Accessibility to and empowerment of non-brothel based FSWs is also very closely linked to the legal environment related to sex work in the country. As recently as June 2004, a participatory intervention programme for HIV prevention among FSWs in Goa, India was put to an abrupt end because the Government of Goa demolished the red-light area of FSWs in its effort to eradicate prostitution and rehabilitate FSWs [27]. The women displaced from this area reported rape, increased violence, reduced ability to negotiate condom use, and multiple partners following this act [27]. The legal context of sex work in India is quite complex, and FSWs are held by police under the Immoral Traffic Prevention Act that deals with human trafficking [28]. Prostitution by itself is not a crime under this Act unless it amounts to nuisance but prostitution-related activities such as running a brothel, making a living on prostitution earnings of another person, or procuring a person for prostitution are a crime. In reality, this Act is more often used to book sex workers and not pimps or clients and is also a source of corruption for the police [29,30], and impedes the provision of HIV prevention for FSWs [27].
In terms of economic vulnerability, the mean income per day for FSWs in this study ranged from Rs. 82, Rs. 135, and Rs. 174 for street-, home- and brothel-based FSWs, respectively. The FSWs with income of Rs. 500 or less in a week were more likely not to use condom consistently with all clients. Therefore, within the context of negligible social empowerment, lack of organized FSW groups, less number of paying clients to earn money, and lower income, it is not always feasible for FSWs to demand the use of condom with the clients.
Approach to prevention
With increased annual budget for the National AIDS Control Programme, expansion of antenatal screening, increased provision of anti-retroviral treatment, and constitution of a National Parliamentarian Forum to generate political support for HIV programmes, the HIV epidemic is one of the top national public health priorities in India [31]. Significant HIV prevention interventions for FSWs are currently on-going in India [13], and would continue to be expanded to increase the coverage of HIV prevention programmes. However, many of the examples of HIV interventions in sex workers available from India are for brothel-based sex workers, and not many for the non-brothel based sex workers who are the majority in the country. Data from this study have indicated significant differences between the brothel- and non-brothel-based FSWs in terms of risk of HIV infection, and therefore, the HIV prevention efforts require strategies to access non-brothel-based sex workers in order to narrow down these differences. The context and environment is also different for the brothel- and non-brothel-based FSWs. The efforts to expand coverage of HIV prevention activities amongst sex workers will depend on achieving and sustaining an environment that enables HIV prevention, which in turn is dependent on the sensitivity of these efforts to the varied contexts of these women. Some examples of such attempts are available globally, including from India [8,25]. There are also lessons to be learnt if the prevention efforts do not involve sex workers as primary stakeholders in these programmes [32,33] or when legal environment disrupts the prevention efforts [27].
Conclusion
It seems necessary that the HIV prevention programmes with female sex workers in India expand beyond generic programmes to be tailored effectively to reach the different types of female sex workers in their local context, especially non-brothel based female sex workers. More comprehensive prevention efforts are needed that include changing the social and legal context of sex work, which would create an environment for sustained reduction of HIV risk in female sex workers.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RD contributed to the study design, data collection, data analysis and interpretation, and drafted the manuscript. LD contributed to the study design, data collection, and data analysis and interpretation. JPG contributed to the study design, data interpretation, and coordination. GAK contributed to the data management, analysis and interpretation. SM, FS and SMB contributed to the study design, data interpretation and coordination. All named authors read, commented on and approved the final version of the manuscript. The ASCI FPP Study Team contributed to the planning of the study logistics, data collection and interpretation, and the members of this Team other than the named authors include (in alphabetical order): G Md Mushtaq Ahmed, Md Akbar, Md Abdul Ameer, Ch Arjun, N Arjun, M Sai Baba, C Satish Babu, J Kishore Babu, I Balasubrahmanyam, V S Udaya Bhaskar, T Gangadhar, P Gopal, Lavanya Gotety, Shaik Omar Hussain, V Indira, S Krishna, P Kiran Kumar, Ch Sri Jaya Lakshmi, T Uma Maheshwar, P Chandra Mouli, S Radhakrishnan, K Raghu, S P Ramgopal, A Srinivas Rao, A Srinivasa Rao, K Hanumantha Rao, N Ananda Rao, P Venkateswara Rao, Parsa V R Rao, D Ravinder, A Srinivas Reddy, G Brahmananda Reddy, S Krishna Reddy, G Uma Sankar, A Satyam, Y S Sivan, P V Sridhar.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank the female sex workers for participating in this study, and the various peer facilitators, non-governmental organisations, the Andhra Pradesh State AIDS Control Society and its Technical Resource Unit, and the International HIV/AIDS Alliance and India HIV/AIDS Alliance for supporting and facilitating this study. The Frontiers Prevention Project, including this study, is supported by a grant from the Bill and Melinda Gates Foundation. The opinions expressed in this paper are those of the authors and do not necessarily reflect the views of the funding agency or the other organisations that supported and facilitated this study.
==== Refs
National AIDS Control Organisation An overview of the spread and prevalence of HIV/AIDS in India
Joint United Nations Programme on HIV/AIDS (UNAIDS) Report on the global AIDS epidemic: 4th global report 2004 Geneva: UNAIDS
Pisani E Garnett GP Grassly NC Brown T Stover J Hankins C Walker N Ghys PD Back to basics in HIV prevention: focus on exposure BMJ 2003 326 1384 1387 12816830 10.1136/bmj.326.7403.1384
Brown T Franklin B MaNeil J Mills S Effective prevention strategies in low HIV prevalence settings 2001 Bangkok: Family Health International
Ainsworth M Beyrer C Soucat A AIDS and public policy: the lessons and challenges of "success" in Thailand Health Policy 2003 64 13 37 12644326 10.1016/S0168-8510(02)00079-9
Desai VK Kosambiya JK Thakor HG Umrigar DD Khandwala BR Bhuyan KK Prevalence of sexually transmitted infections and performance of STI syndromes against aetiological diagnosis, in female sex workers of red light area in Surat, India Sex Transm Infect 2003 79 111 115 12690130 10.1136/sti.79.2.111
Andhra Pradesh State AIDS Control Society (APSACS) Prevalence of sexually transmitted infections and HIV among female sex workers of Kakinada and Peddapuram, Andhra Pradesh, India 2001 Hyderabad: APSACS
Alary M Mukenge-Tshibaka L Bernier F Geraldo N Lowndes CM Meda H Gnintoungbe CA Anagonou S Joly JR Decline in the prevalence of HIV and sexually transmitted diseases among female sex workers in Cotonou, Benin, 1993–1999 AIDS 2002 16 463 470 11834959 10.1097/00002030-200202150-00019
Ghys PD Diallo MO Ettiegne-Traore V Kale K Tawil O Carael M Traore M Mah-Bi G De Cock KM Wiktor SZ Laga M Greenberg AE Increase in condom use and decline in HIV and sexually transmitted diseases among female sex workers in Abidjan, Cote d'Ivoire, 1991–1998 AIDS 2002 16 251 268 11807310 10.1097/00002030-200201250-00015
Hanenberg RS Rojanapithayakorn W Kunasol P Sokal DC Impact of Thailand's HIV-control programme as indicated by the decline of sexually transmitted diseases Lancet 1994 344 243 245 7913163 10.1016/S0140-6736(94)93004-X
Bandyopadhyay N Baneerjee B Sex workers in Calcutta organize themselves to become agents for change Sex Health Exch 1999 2 6 8 12295568
Nagelkerke NJ Jha P de Vlas SJ Korenromp EL Moses S Blanchard JF Plummer FA Modelling HIV/AIDS epidemics in Botswana and India: impact of interventions to prevent transmission Bull World Health Organ 2002 80 89 96 11953786
National AIDS Control Organisation Targeted interventions
Bhave GL Lindan CP Hudes E Desai S Wagle U Tripathi SP Mandal JS Impact of an intervention on HIV, sexually transmitted diseases and condom use among sex workers in Bombay, India AIDS 1995 9 S21 30 8561997
Laga M Alary M Nzila N Manoka AT Tuliza M Behets F Goeman J St Louis M Piot P Condom promotion, sexually transmitted diseases treatment, and declining incidence of HIV-1 infection in female Zairian sex workers Lancet 1994 344 246 248 7913164 10.1016/S0140-6736(94)93005-8
Ngugi EN Plummer FA Simonsen JN Cameron DW Bosire M Waiyaki P Ronald AR Ndinya-Achola JO Prevention of transmission of human immunodeficiency virus in Africa: effectiveness of condom promotion and health education among prostitutes Lancet 1988 2 887 890 2902326 10.1016/S0140-6736(88)92480-4
Global HIV Prevention Working Group Access to HIV prevention: closing the gap
Centres for Disease Control Fact sheet for public health personnel: male latex condoms and sexually transmitted diseases
Census of India Series 29 – Andhra Pradesh; Provisional population totals: rural-urban distribution of population 2001 Hyderabad: Director of Census Operations, Andhra Pradesh
Bennett S Woods T Liyanage WM Smith DL A simplified general method for cluster-sample surveys of health in developing countries World Health Stat Quar 1991 44 98 106
Andhra Pradesh State AIDS Control Society (APSACS) VIIIth Round of National Annual Sentinel Surveillance for HIV, Andhra Pradesh 2004 Hyderabad: APSACS
National AIDS Control Organisation National Baseline High Risk and Bridge Population Behavioural Surveillance Survey-2001 Part-I (FSW and their clients)
Population Report Volume XXVII Number 1, April 1999
National AIDS Control Organisation A note on HIV estimates 2003
Jana S Bandopadhyay N Mukherjee S Dutta N Basu I Saha A STD/HIV intervention with sex workers in West Bengal, India AIDS 1998 12 S101 108 9679635
Gadgil A Peer group education among prostitutes of Kamathipura and Khetwadi areas of Bombay X International Conference on AIDS Yokohama 1994 abstract PD047
Shahmanesh M Wayal S Targeting commercial sex-workers in Goa, India: time for a strategic rethink? Lancet 2004 364 1297 1298 15474122 10.1016/S0140-6736(04)17206-1
The Immoral Traffic Prevention Act 1956
Shyamala A Abuses against sex workers and erosions of HIV prevention efforts resulting from anti-trafficking initiatives
Baneerjee R Asian women meet to demand legal status
Rao JVRP Ganguly NK Mehendale SM Bollinger RC India's response to the HIV epidemic Lancet 2004 364 1296 1297 15474121 10.1016/S0140-6736(04)16982-1
Loff B Overs C Longo P Can health programmes lead to mistreatment of sex workers? Lancet 2003 361 1982 1983 12801756 10.1016/S0140-6736(03)13595-7
Wolffers I van Beelen N Public health and human rights of sex workers Lancet 2003 361 1981 12801755 10.1016/S0140-6736(03)13594-5
|
16111497
|
PMC1208909
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Aug 20; 5:87
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-87
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-901612487110.1186/1471-2458-5-90Research ArticleThe relationship between risk factors for falling and the quality of life in older adults Ozcan Ayse [email protected] Hulya [email protected] Nihal [email protected] Mehtap [email protected] Didem [email protected] Dokuz Eylül University School of Physical Therapy and Rehabilitation, 35340 Inciralti-Izmir, Turkey2 T.C. Emekli Sandigi Narlidere Nursing Home (The Izmir Geriatric Centre), 35340 Narlidere-Izmir, Turkey2005 26 8 2005 5 90 90 10 2 2005 26 8 2005 Copyright © 2005 Ozcan et al; licensee BioMed Central Ltd.2005Ozcan et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Falls are one of the major health problems that effect the quality of life among older adults. The aim of this study was to explore the relationship between quality of life (Short Form-12) and the risk factors of falls (balance, functional mobility, proprioception, muscle strength, flexibility and fear of falling) in older adults.
Methods
One hundred sixteen people aged 65 or older and living in the T.C. Emekli Sandigi Narlidere nursing home participated in the study. Balance (Berg Balance test), functional mobility (Timed Up and Go), proprioception (joint position sense), muscle strength (back/leg dynamometer), flexibility (sit and reach) and fear of falling (Visual Analogue Scale) were assessed as risk factors for falls. The quality of life was measured by Short Form-12 (SF-12).
Results
A strong positive correlation was observed between Physical Health Component Summary of SF-12, General Health Perception and balance, muscle strength. Proprioception and flexibility did not correlated with SF-12 (p > 0.05). There was negative correlation between Physical Health Component Summary of SF-12, General Health Perception and fear of falling, functional mobility (p < 0.05).
Conclusion
We concluded that the risk factors for falls (balance, functional mobility, muscle strength, fear of falling) in older adults are associated with quality of life while flexibility and proprioception are not.
==== Body
Background
Quality of life is a term used in a number of disciplines, and definitions and conceptualization varies from utility of health states to life satisfaction, and from possession of socially desirable characteristics to positive affect [1]. Quality of life has recently become commonly used both as a concept and as a field of research [2-5]. Studies have also demonstrated that older elderly people expressed higher life satisfaction/quality of life than younger ones.
There are many socio-demographic characteristics that may contribute to the quality of life such as age, socio-economic status, and marital status in older adults. Falls are one of the major health problems that effect the quality of life among older adults [2-5]. The Short-Form 36 (SF-36) is a widely used quality of life instrument. However, its length could affect response rates, particularly in older adults. The SF-12 has proved to be suitable for older adults because of the limited number of questions. The 12 items in SF-12 represent one physical component summary score and one mental component summary score and assess a person's perceived health-related quality of life.
Many factors were originally considered as possible risk factors for falls based on a review of currently available literature. These factors include age, number of chronic diseases, body composition, muscle strength, functional mobility and performance measures related to balance function [3,4,6].
Impaired balance and functional mobility are major risk factors for falls. There are many studies investigating the relationship between falling and contributory factors [7-13]. However, no study investigating the correlation between risk factors for falls and quality of life in older adults could be found.
Since falls and its consequences have a major role in quality of life, rehabilitation programs, which aim to decrease the risk of falling by considering all contributing factors such as muscle strength, flexibility and balance, have the potential to both decrease the risk of falling and improve the quality of life. Due to this interaction, the relationship between risk factors for falls and the quality of life becomes significant. Based on a review of literature, this study was designed to explore the relationship between the quality of life (Short Form-12) and risk factors for falls (balance, functional mobility, proprioception, muscle strength, flexibility and fear of falling) in older adults.
Methods
Subjects
A total of 116 (52 men and 64 women) participants aged 65 or older with or without a history of falls were recruited from the 535 registered residents of T.C. Emekli Sandigi Narlidere nursing home for this study. Ambulatory individuals having no disability in self-care formed the population of this study and a report stating sound mental healthy from a psychiatrist of a state hospital was required at the registration of all participants.
The exclusion criteria were as follows: being aged less than 65, being unable to walk less than 10 meters, amputation, having had a stroke recently, unstable medical conditions such as diabetes mellitus, hypertension, 2 or more fractures due to osteoporosis, resting angina, recurrent heart failure or recurrent arrhythmias and uncontrolled seizure disorder. Also the residents who were assessed as mentally oriented by the psychiatrist were included the study. After checking health documents of residents, and considering the inclusion criteria, 404 residents were approached about the study. 112 subjects did not agree to participate, and 53 subjects exercised regularly (more than twice a week during the previous 2 months). 37 potential subjects couldn't be reached. 202 subjects accepted the invitation to participate in the study and 141 of them came and were evaluated. The evaluation of 25 subjects couldn't be completed the assessment because they were not able to take some positions of tests physically. Data were obtained from 116 subjects.
The study was performed according to the principles of the Decleration of Helsinki and was approved by the Ethics Committee of Dokuz Eylül University Medical School (reference number: 29.12.03/156). Informed consent was taken from the patients, immediately prior to the data collection. After giving informed consent, all subjects completed a health status questionnaire which provided information on age, medical history, alcohol consumption, self-reported history of fall, use of devices to assist ambulation and medication. The same physiotherapist did all assessments.
Procedure
The quality of life
The quality of life was measured by Short Form-12 (SF-12) (Ware, Kosinski and Keller 1996). The items in the SF-12 instrument were used to calculate two scales, the Physical Component Scale and the Mental Component Scale. Scores range from 0 to 100, a higher score indicates better mental health, physical health and general health perception [14].
Balance
The Berg Balance Scale (BBS) was used to evaluate balance. The BBS is a 14-item balance assessment tool that is scored on a 5 point ordinal scale (0–4) measuring levels of ability in performing each task (4 = safe and independent, 0 = incapable). The BBS includes tasks such as standing with eyes closed, reaching, standing on one foot and picking up objects from the floor. The highest total possible score on the Berg Balance Scale is 56, indicating excellent balance [15].
Functional mobility
The Timed Up and Go (TUG) test was used to measure basic functional mobility. The time taken to complete rising from a chair, walking 10 ft (3 m), turning, walking back to the chair and sitting was recorded in seconds. The starting position was standardized so that the subjects commenced the test with their feet flat on the floor and their arm resting on the armrests. No physical assistance was given. Each subjects was asked to perform three test trials. The mean score was recorded [16].
Proprioception (joint position sense)
Proprioception was assessed using established and validated lower limb matching tasks. In this test, subjects seated with their eyes closed were asked to align their lower limbs simultaneously on either side of a vertical clear acrylic sheet (60 × 60 × 1 cm) inscribed with a protractor and placed between their legs. To prevent limited motion at the knee joint from confounding the results of this test, the examiner needed to ensure that subjects matched their limbs near the midrange of knee joint motion. Each trail was undertaken relatively quickly, with rests between trials, to avoid weakness unduly influencing the results. Any difference in aligning the lower limbs (indicated by disparities in matching the big toes on either side of acrylic sheet) was measured in degrees for both extremities. After 2 practice trails, an average of 5 experimental trails was recorded [17].
Muscle strength
Back/leg dynamometer was used to measure leg strength. The subject stood on a platform with their feet apart at a comfortable distance of shoulder width for balance. Their hands grasped each end of a bar. The subject was asked to flex at their knees to approximately 135 degrees. The back was kept straight and the hips were positioned directly over the ankle joints. In this way, the activation of back muscles was eliminated. The chest was kept forward and the head was held in an erected position. The subject took in a large breath and slowly exhaled as they attempted to extend their knees smoothly and as forcefully as possible. Three attempts were made and a mean score was recorded [18].
Flexibility
In order to assess flexibility, a sit and reach test was used. A box 32 cm in height and 50 cm in length with a top plate 45 cm in width was used for the test. The length of the top plate was 75 cm, the first 25 cm of which was extended over the front edge of the box towards the subject's feet. Older adults were asked to sit, keeping their knees straight, and reach forward as far as possible from a seated position. The score was determined by the furthest position they reached with their fingertips on a scale. Three trials were perfomed and the mean score was recorded [19,20].
Fear of falling
As an indication of fear of falling in daily life a visual analogue scale (VAS) was used. Subjects were asked to express their overall feelings of fear of falling by drawing a mark on a vertical line of exactly 10 centimetres connecting the two statements: "no fear of falling" (below) and "very afraid of falling"(above). The score was the number of centimetres between "no fear of falling" and the subject's mark [21].
Statistical analysis
Applying SPSS version 10.0 for statistical analyses, we considered differences of two-tailed p < 0.05 as statistically significant. All data were shown as means with standard deviations (means ± SD) and ranges were added. Pearson's correlation coefficient was used to analyse the relationship between the SF-12 and balance, functional mobility, proprioception, muscle strength, flexibility and fear of falling.
Results
Demographic characteristics of the subjects and medical history were summarized in Table 1.
Table 1 Demographic characteristics of subjects (n = 116)
n %
Age (X ± SD) (Range) (years) 76.60 ± 6.19 (65–90)
BMI (X ± SD) (Range)(kg/m2) 27.69 ± 3.46 (20.9–40.0)
Sex (n) Female 64 55
Male 52 45
Assistive device No 96 82
Yes 20 17
Alcohol consumption No 94 81
Yes 22 19
Fall(s) in the previous year No 79 68
1 time 25 22
2 times or more 12 10
Reported medical conditions Osteoarthritis 47
Hypertension 41
Osteoporosis 34
Low back pain 21
Visual and hearing problems 20
Cardiac problems 14
Urinary incontinence 7
Peripheral vascular disease 4
Asthma 4
No of medications (X ± SD) (Range) 4.03 ± 2.42 (0–10)
Means ± SD and range of all measurements performed are given in Table 2.
Table 2 Mean and SD of all measurements
Mean ± SD Range
SF-12 (Mental health component summary score) 57.98 ± 12.54 19–100
SF-12 (Physical health component summary score) 57.67 ± 13.77 17–100
SF-12 (Total-General Health Perception) 58.64 ± 13.65 15–100
Fear of falling (VAS) (cm) 3.60 ± 3.13 0–10
Balance (BBS) 52.54 ± 3.50 35–56
Functional mobility (TUG) (sec) 13.70 ± 5.94 6–63
Proprioception (joint position sense) (°) 4.37 ± 2.84 0.2–15.2
Muscle strength (kg) (Back/leg dynomometer) 47.98 ± 22.91 20–112
Flexibility (sit and reach) (cm) 10.57 ± 13.61 -35–45
When the correlation between age, flexibility, proprioception and SF-12 was evaluated, it was observed that there was no change in quality of life with aging, flexibility and proprioception (p > 0.05).
It was found that as the body mass index (BMI) increases the quality of life decreases. Physical Health Component, Mental Health Components of SF-12 and General Health Perception of SF-12 showed the same results for BMI.
The correlation analyses between fear of falling and SF-12 showed that as the fear of falling increases, the quality of life (with the exception of the mental health component) decreases.
A strong positive correlation was observed between Physical Health Component of SF-12, General Health Perception and results of BBS. The increase in quality of life related to the increase in the balance score.
When the correlation of TUG with SF-12 was evaluated, it was found that there was a strong negative correlation between Physical Health Component of SF-12, General Health Perception and TUG. On the other hand, quality of life increased with improving functional mobility.
The analyses showed that quality of life improved as muscle strength increased since muscle strength was correlated with Physical Health Component of SF-12 and General Health Perception.
Correlation coefficients (r) and levels of significance (p) between risk factors for falls and SF-12 were given in Table 3.
Table 3 Pearson correlation coefficients (r) and levels of significance in a comparison of risk factors for falls and SF-12.
SF-12 Mental Health Component SF-12 Physical Health Component SF-12 General Health Perception
Age NS NS NS
Body Mass Index r = -0.213
p = 0.021* r = -0.262
p = 0.004* r = -0.272
p = 0.003*
Fear of falling (VAS) NS r = -0.248
p = 0.007* r = -0.223
p = 0.016*
Balance (BBS) NS r = 0.381
p = 0.000* r = 0.270
p = 0.003*
Functional mobility (TUG) NS r = -0.354
p = 0.000* r = -0.249
p = 0.007*
Proprioception (Joint position sense) NS NS NS
Muscle strength (Back/leg dynamometer) NS r = 0.338
p = 0.000* r = 0.230
p = 0.016*
Flexibility (Sit and reach test) NS NS NS
NS: Not significant
Discussion
In our study, the relationship between risk factors for falls and quality of life was investigated in older adults. Balance, functional mobility, muscle strength and fear of falling were shown to correlate with General Health Perception (SF-12) but no correlation was seen between proprioception and flexibility in relation to the General Health Survey (SF-12).
In the study, it was found that quality of life was not correlated with age. This result suggested that the quality of life does not change with aging but age affects the risk factors for falls.
While BMI increased, physical, mental and general health perception scores of SF-12 decreased. This relationship is important because increased BMI causes functional limitation and affects physical, mental and general health perception in older adults. In addition, the mental health component only correlated with BMI.
There are many studies which state that BBS and TUG test results are the most important risk factors for falls [10-12,14,15]. Physical and general health perception scores of SF-12 strongly correlated with BBS (positively) and TUG test (negatively). These results demonstrated that poor balance and functional mobility were associated with a decreased quality of life.
A lack of flexibility is associated with problems in executing and sustaining motor activities in daily life and is related to an increased risk of falling in older adults [19]. Proprioception is an important component of balance. Interestingly, it was seen that proprioception and flexibility did not correlated with quality of life in our study. There is a need for some further study concerning the relationship between the risk of falling, proprioception and flexibility.
When older adults worry about falling, it may indicate that their physical condition is affected, possibly due to a lack of balance. Therefore, the fear of falling is an important risk factor for falls in older adults. Hence VAS is a simple, practical and easy method of assessment for this subject and so this method was preferred. Several studies have explored the strong association between muscle strength and the risk of falling in older adults [7,8,13]. Similarly in this study, it was found that muscle strength and factors relating to the risk of falling correlated with physical and general health perception scores of SF-12.
Fear of falling correlated with Physical Health Component and General Health Perception of SF-12. Suzuki stated that fear of falling is increasingly recognized as a factor that may affect activity, function and physical condition in older adults [4].
In general while risk factors for falls were associated with physical component and general health perception of SF-12, only the body mass index correlated with the mental component of SF-12 in this study. This result pointed out that risk factors for falls affect the physical health component and general health perception but not the mental health component in older adults.
In our study balance, muscle strength, proprioception, flexibility, functional mobility were evaluated as parameters of physical function. Some investigators use physical component score of SF-12 rather than physical function. But proprioception and flexibility were not correlated with physical component score of SF-12 while muscle strength, functional mobility and balance were correlated in present study. As it is seen in our study some physical function parameters may or may not related to physical component score of quality of life.
As this study is cross sectional, the associations that were demonstrated may or may not be causal. Various factors can affect the physical or mental component scores and general quality of life in older adults. Assessing correlations only between some risk factors for falling and quality of life is the major limitation of this study.
Older adults who live in the Narlıdere Nursing Home are independent in their daily living activities. Only food and room cleaning services are provided for them. They do not spend all of their time in their own rooms but take part in activities such as visiting relatives, shopping, short holidays, taking walks etc. For this reason, some of them did not wish to participate in this study and some of them could not be reached. As a result of this, the population of this study decreased from 404 subjects to 116.
Prevention of falls and their subsequent injuries is an important goal of geriatric evaluation. Proper physical therapy programs minimizing the risk of falls may increase quality of life in older adults.
Conclusion
The risk factors for falls (balance, functional mobility, muscle strength, fear of falling) in older adults are associated with quality of life while flexibility and proprioception are not. Future studies should focus on other factors that affect quality of life in larger elderly populations and investigate the effect of such programs on quality of life in relation to risk factors for falls.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
OA conceived of the the study and participated in the sequence alignment and drafted the study. DH carried out acquisition of data and desing of the study. GN performed the statistical analysis and drafted the manuscript. OM performed the statistical analysis. KD participated in the desing of the study. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We would like to thank physiotherapists at TC Emekli Sandigi Narlıdere Nursing Home-The Izmir Geriatric Centre for their help in coordinating and scheduling the participants in this study and older adults for their participation.
==== Refs
Dijkers M Measuring quality of life: Methodological issues Am J Phys Med Rehabil 1999 78 286 300 10340429 10.1097/00002060-199905000-00022
Berglung AL Ericsson K Different meanings of quality of life: a comparison between what elderly persons and geriatric staff believe is of importance Int J Nurs Practice 2003 9 112 9 12694480 10.1046/j.1322-7114.2003.00414.x
Tseng SZ Wang RH Quality of life and related factors among elderly nursing home residents in Southern Taiwan Public Health Nurs 2001 18 304 11 11559413 10.1046/j.1525-1446.2001.00304.x
Suzuki M Ohyama N Yamada K Kanamari M The relationship between fear of falling, activities of daily living and quality of life among elderly individuals Nurs Health Sci 2002 4 155 61 12406202 10.1046/j.1442-2018.2002.00123.x
Baker PS Bodner EV Allman RM Measuring life-space mobility in community-dwelling older adults J Am Geriat Soc 2003 51 1610 1614 14687391 10.1046/j.1532-5415.2003.51512.x
Downton JH Why do old people fall? Falls in the elderly 1993 Great Britain: Hodder and Stoughton Limited 1 77
Ringsberg K Gerdhem P Johansson J Obrant KJ Is there a relationship between balance, gait performance and muscular strength in 75-year-old women? Age Aging 1999 28 289 293 10.1093/ageing/28.3.289
Wolfson L Judge J Whipple R King M Strength is major factor in balance, gait, and the occurance of falls J Gerontol A Biol Sci Med Sci 1995 50 64 7 7493221
Maki BE Holliday PJ Topper AK A prospective study of postural balance and risk of falling in a ambulatory and independent elderly population J Gerontol 1994 49 M72 M84 8126355
Hatch J Gill-Body KM Portney LG Determinants of balance confidence in community-dwelling elderly people Phys Ther 2003 83 1072 1079 14640866
Shumway-Cook A Baldwin M Polissar NL Gruber W Predicting the probability for falls in community older adults Phys Ther 1997 77 812 819 9256869
Bogle Thorbahn LD Newton RA Use of the Berg Balance test to predict falls in elderly persons Phys Ther 1996 76 576 83 8650273
Gehlsen GM Whaley MH Falls in elderly: Part II, balance, strength, and flexibility Arch Phys Med Rehabil 1990 71 739 741 2403279
Ware JE Kosinski M Keller SD A twelve-item short-form survey: construction of scales and preliminary tests of reliability and validity Med Care 1996 34 220 233 8628042 10.1097/00005650-199603000-00003
Berg KO Wood Dauphniee SL Williams JI Maki B Measuring balance in the elderly: validation of an instrument Can J Public Health 1992 83 S7 S11 1468055
Podsiadlo D Richardson S The timed "Up &Go": a test of basic functional mobility for frail elderly persons J Am Geriart Soc 1991 39 142 148
Lord SR Menz HB Tiedemann A A Physiological profile approach to falls risk assessment and prevention Phys Ther 2003 83 237 253 12620088
Friedman SM Munoz B West SK Rubin G Fried LP Falls and fear of falling: which comes first? A longitudinal prediction model suggests strategies for primary and secondary prevention J Am Geriat Soc 2002 50 1329 35 12164987 10.1046/j.1532-5415.2002.50352.x
Lemmink KAPM Kemper HCG de Greef MHG Rispens P Stevens M The validity of the sit-and-reach test and the modified sit-and-reach test in middle-aged to older men and women Res Q Exerc Sport 2003 74 331 6 14510299
Jones JC Rikli RE Max J Noffal G The reliability and validity of a chair sit-and-reach test as a measure of hamistring flexibility in older adults Res Q Exerc Sport 1998 69 338 343 9864752
Wolf B Feys H Weerdt WD Van der Meer J Noom M Aufdemkampe G Effect of a physical therapeutic intervention for balance problems in the elderly: a single-blind, randomized, controlled multicentre trial Clin Rehabil 2001 15 624 636 11777093 10.1191/0269215501cr456oa
|
16124871
|
PMC1208910
|
CC BY
|
2021-01-04 16:28:56
|
no
|
BMC Public Health. 2005 Aug 26; 5:90
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-90
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-921613733110.1186/1471-2458-5-92Research ArticlePride and confidence at work: potential predictors of occupational health in a hospital setting Nilsson Kerstin [email protected] Anna [email protected] Inga-Lill [email protected] Töres [email protected] School of Life Sciences, University of Skövde, Skövde, Sweden2 National Institute for Psychosocial Medicine (IPM), Stockholm, Sweden3 Department of Public Health Sciences and Department of Occupational and Environmental Health, Stockholm, Sweden2005 1 9 2005 5 92 92 25 2 2005 1 9 2005 Copyright © 2005 Nilsson et al; licensee BioMed Central Ltd.2005Nilsson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
This study focuses on determinants of a healthy work environment in two departments in a Swedish university hospital. The study is based on previously conducted longitudinal studies at the hospital (1994–2001), concerning working conditions and health outcomes among health care personnel in conjunction with downsizing processes. Overall, there was a general negative trend in relation to mental health, as well as long-term sick leave during the study period. The two departments chosen for the current study differed from the general hospital trend in that they showed stable health development. The aim of the study was to identify and analyse experiential determinants of healthy working conditions.
Methods
Thematic open-ended interviews were carried out with seventeen managers and key informants, representing different groups of co-workers in the two departments. The interviews were transcribed verbatim and an inductive content analysis was made.
Results
In the two studied departments the respondents perceived that it was advantageous to belong to a small department, and to work in cooperation-oriented care. The management approaches described by both managers and co-workers could be interpreted as transformational, due to a strain of visionary, delegating, motivating, confirmative, supportive attitudes and a strongly expressed solution-oriented attitude. The daily work included integrated learning activities. The existing organisational conditions, approaches and attitudes promoted tendencies towards a work climate characterised by trust, team spirit and professionalism. In the description of the themes organisational conditions, approaches and climate, two core determinants, work-pride and confidence, for healthy working conditions were interpreted. Our core determinants augment the well-established concepts: manageability, comprehensiveness and meaningfulness. These favourable conditions seem to function as a buffer against the general negative effects of downsizing observed elsewhere in the hospital, and in the literature.
Conclusion
Research illuminating health-promoting aspects is rather unusual. This study could be seen as explorative. The themes and core dimensions we found could be used as a basis for further intervention studies in similar health-care settings. The result could also be used in future health promotion studies in larger populations. One of the first steps in such a strategy is to formulate relevant questions, and we consider that this study contributes to this.
==== Body
Background
In this study we have focused on potential determinants of healthy working conditions in two departments in a Swedish university hospital. In previously conducted longitudinal studies at the present hospital (1994–2001), Hertting and colleagues studied working conditions and health outcomes among health care personnel in conjunction with downsizing processes during the 1990s. This downsizing period was characterised by personnel reductions; the hospital staff was reduced by 22% (1000 persons) between 1995–1997, and 10% were relocated to other departments. Other structural changes during the study period were mergers of departments and outsourcing of service units, along with continuous demands for cutting costs [1-5].
Biological stress markers, measured on 31 female medical secretaries, registered nurses and assistant nurses, indicated that protective and anabolic functions had suffered, from the adjustment phase (1997) to the reconstruction in 1998. The continuing adaptation process indicated increasing difficulty for the women to mobilise energy [1]. Repeated interviews (1997, 1998, 2000 and 2001) with the same respondents confirmed these difficulties [2-4].
A third study, based on a questionnaire and hospital register data (1994–2001), found evidence for negative trends in mental health and in long-term sick leave at the hospital, corresponding to negative trends in working conditions [5]. In international research, increasing work demands and reduced control [6], combined with job insecurity and loss of trust [7], were identified during periods of downsizing. Further, downsizing has been associated with subsequent high rates of sickness absence [8].
Petterson and her colleagues [5] also found that increasing demands to work hard, conflicting demands, and increasing lack of time to plan work, were strongly associated with overall deteriorating health. It was not possible to identify health-promoting departments in terms of positive health development. However, compared with the generally negative hospital trend, four of the 24 departments studied might be labelled more 'health-promoting', since they showed stable trends both in mental health and in short- and long-term sick-leave rate. Two of these departments were selected for this current interview study.
Health promotion at the workplace
A health-promoting hospital has been considered to be a 'healthy organisation' in the sense of having an effective relation-oriented management and active staff participation [9], which means a workplace where people are able to produce, serve, grow and be evaluated [10]. Accordingly, in the context of working life, health promotion should be a matter of supporting peoples' resources by enhancing job control, encouraging social support, networks and relations at the workplace, and providing a meaningful job.
Health promotion is often based on the salutogenic model [11], where health is seen as being created by a sense of coherence, including comprehensibility, meaningfulness and manageability. When applied to working life, comprehensibility implies that work changes have to be predictable. Work assignments must be meaningful in the sense of being perceived as socially rewarding and stimulating. Manageability means access to the necessary resources for handling work pressure, which includes own ability as well as material and supportive resources provided by the environment. Health promotion at the workplace could be a question of enhancing such resources by strengthening the individual's internal control and ability to act, i.e. room for manoeuvre [12].
Ever since the demand-control model was introduced [13], increased decision latitude has been perceived as an important element in health promotion in the workplace. In a longitudinal study an improved relationship between decision latitude and psychological demands has been shown to be related to a rising serum concentration of testosterone, which is an index of increasing regenerative bodily activity [14]. In different controlled experiments it has been shown that efforts to improve decision latitude for employees are associated with improvements in morning serum cortisol [15], decreased staff turnover and decreased sick leave [16].
Health outcomes could be predicted, as there is a correlation between positive self-rated health and a lower risk of future sickness and death [17]. Research has also found that satisfaction with one's place of work and profession are important predictors for positive good health [18]. In a longitudinal participatory worksite intervention, job control was recognised as the mediator for increased mental health, less sick leave and a higher level of self-rated performance [19]. In another extensive longitudinal study in a health-care setting it was found that workmates or managers freezing out employees, and employees' experiences of negative and threatening changes at the workplace are of importance for long-term sick leave [20].
In a survey study, Noblet [21] confirmed that job control and social support at work had a health-promoting value. Job stressors accounted for as much as 25% of explained variance in mental health. The factor most strongly associated with mental health and job satisfaction was time to do the job as well as you would like. Time-related pressure was strongly associated with job strain.
Skill variety, autonomy, social contact and learning were confirmed as determinants of intrinsic work motivation in a study of specific work-setting determinants for contributing to the motivation and good health of nurses. These factors made their work challenging and meaningful. Emotional exhaustion can be reduced if attention is paid to workload [22].
Leadership and employee health
Transformational leadership contains elements of charismatic leadership, inspirational motivation, intellectual stimulation and individual consideration [23]. Bass and Avolio found that leaders with a transformational style are capable of empowering employees to stretch their limits, and of being innovative to create solutions [24]. Empowering management could contribute to extended decision latitude, as power in the organisation transfers to co-workers [25]. A workplace environment characterised by such empowerment strategies can improve productivity as well as cohesion and job satisfaction for the employees [26,27]. The relation between health outcome and job satisfaction has been confirmed [28], and job satisfaction among nursing staff has been found to be positively correlated with transformational leadership style [29].
The relationship between organisational efficiency and transformational leadership style was confirmed by Marks-Moran [30]. However, transformational leadership was found to be most efficient when management by team is a reality and the organisation is development-oriented [31]. Bass [23] concluded that transformational leaders are more successful, in terms of efficiency, than leaders using a transactional style, which includes management-by-expectation, and contingent reinforcement and rewards. In a study among Belgian head nurses, fewer work stressors and lower levels of emotional exhaustion were found among staff when a transformational leadership style was used [32]. However, in a study of Swedish health centres, Bernin and Theorell were not able to confirm any impact of leadership style on subordinates' health [33]. From the review of research in the field of management, there are studies supporting the view that leadership is one factor that influences job satisfaction. However, there are different and inconclusive findings concerning leadership style and its impact on employees' health.
In the present study the purpose was to understand how employees succeeded in retaining their health and going beyond the stable (positive) health trends in the two selected departments in a period of turbulence [5]. By focusing on experienced health determinants in management at department levels, we expect to acquire knowledge about health-promotive factors for employees working in health care organisations. In this study we identify management as a phenomenon where both managers and co-workers participate.
The aim of the study was to identify and analyse experiential determinants of healthy working conditions from the perspectives of hospital managers and staff in key positions.
Methods
Based on previous studies of this particular hospital and on the purpose of the study, thematic open-ended interviews were chosen as a method for data collection. The outcomes of such interviews depend on the quality of interaction between the respondent and the interviewer, but also upon culturally hidden assumptions about understandings of experiences, feelings and intentions [34]. The interview is a face-to-face meeting, where the researcher attempts to understand the respondent's perspective and experiences. This makes the interviews more like a conversation than an interview with scope simply for asking and responding [35]. It is important to mention that it is the outcome of the interviews that is analysed and not the interaction [36].
Participants and data collection
Interviews with 17 persons were conducted by the first author (KN), in two departments with stable health trends between 1994 and 2001. These particular departments were selected as they had all-round health care, such as nursing wards, assessment units, as well as care and treatment wards for outpatients. One of the departments specialised in neurological care with the district as catchment area, and the other in gynaecological ontological care with an extended region as its catchment area. The respondents were managers and selected key persons among the co-workers, including all managers, both on department and ward level, and co-workers including physicians with medical responsibility, staff with special care responsibility and a trade union representative (see Table 1). Twelve of the interview persons were women and five were men. All of them were middle-aged, and all had experience from the same department during the whole study period (1994–2001).
Table 1 Participants
Participants' function (profession) Number
Head of department (physician) 2
Ward sister (nurse) 4
Assistant ward sister (nurse) 2
Medically responsible (physician) 3
Responsible for medical secretariat (medical secretary) 2
Responsible for occupational therapy (occupational therapist) 1
Responsible for cancer nursing (nurse) 2
Trade union representatives (assistant nurse) 1
Total 17
The heads of the departments, in agreement with the research team, selected the interviewees, who were chosen to reflect varied experiences of working conditions, the leadership in the organisation and professional tasks. All interviews lasted about one hour and were carried out in the autumn of 2003. The thematic interview guide included the following themes: organising the work, handling organisational changes, taking advantage of the co-workers' resources, and the meaning of healthy work. The thematic interviews were unstructured, in accordance with Silverman [34]. This implies that each theme was discussed in every interview, but in different sequences depending on how the interview developed. The interviewer represented the field of nursing research, had many years of experience as a teacher in nursing education, and was trained in interviewing.
Data analysis
All interviews were transcribed verbatim and included about 170 pages in all of single-spaced text. An inductive content analysis based on a thematic coding of the text was made in order to obtain meaning and understanding [34]. Interviews with respondents from the two departments were analysed together, as the phenomenon of interest was to identify potential contributing factors to stable health. To gain a holistic depiction of the material, all protocols were read in their entirety by the first author (KN). Keeping the aim of the study in mind, words and sentences were marked in the text. Thereafter followed a process where words and sentences (with related segments of interview texts) were brought together in terms of content. These groups were given preliminary names. In order to validate the analysis, the second author (AH) participated as a co-analyser and examined the themes (found in the first analysis by KN) in relation to the material as a whole. The two analysers then compared each other's analyses and the analysing process lasted until agreement was reached. During this comparative procedure the groups were reduced in number and expanded in content to finally form three themes with accompanying sub-themes reported in the results.
Methodological considerations
Prerequisites for attaining the respondents' descriptions were supported by the interviewers having no previous experience of these departments. Openness and a friendly atmosphere characterised the interview situation, which made it easier for the interviewees to say what they wanted to say in the present situation [34].
In the analysis phase there is always a risk that preconceptions about the studied field of research could influence the process. Preconceptions can be an advantage, as an intimate knowledge of the field facilitates discovery of nuances in the interviewees' statements. But, on the other hand, there is a risk that a field that is too well-known can blind the researcher, who may thereby miss new angles of incidences on the studied phenomenon. To minimise these methodological pitfalls we reflected on this problem together and tried to be aware of our preconceptions in all research phases [34].
In this study we used a co-analyser, which also demands awareness during the analyses. Having a co-analyser could be an advantage, as two people are always able to discover more than one. But it is necessary to point out that there is a risk that the analysers strive for consensus, which in its turn could reduce variation in the content of the themes [34].
In order to validate our tentative findings we gave feedback to the respondents, to receive their reactions [34]. Hence, the findings were discussed in managerial group meetings, resulting in minor modifications. Good agreement supported high face validity.
Ethical considerations
This study followed the Humanistic-Social Research Council Ethics Rules [37]. Due to ethical considerations, special emphasis was placed on informing the participants about the study, obtaining their consent and treating their statements confidentially. The interviewees have neither before, nor during, nor after the interview occasions refused to participate. The quotations in the results are used to exemplify the statements of the individual respondents and are marked with a figure.
Results
The analyses resulted in three themes with accompanying sub-themes (see Table 2), indicating a work process emanating from the organisational conditions at the departments; as well as the development of organisational approaches, attitudes and activities that were favourable to the creation of a climate characterised by trust, team spirit and professionalism.
Table 2 Themes and sub-themes in the result
Theme Sub-theme
Organisational conditions - The benefits of a small department
- The benefits of the character of care
Organisational approaches - Leadership and followership attitudes
- Learning activities
- Supportive activities
Organisational climate - Trust
- Team spirit
- Professionalism
Organisational conditions
The benefits of a small department
Approximately 60 people are employed in each department, which the respondents consider to be small departments, with regard to both the size of the staff and the number of beds and outpatient capacity. According to the interviewees, this means that the staff get to know each other and are seen by both managers and each other. Accordingly, distance is reduced between co-workers, as well as between subordinates and employees in different managerial positions, i.e. within the hierarchies.
It's so easy to have an overview of everything. It's much worse in a big department with several hundred employees. There are so few of us that we can all fit into a bus. (8)
The fact that the departments are small, facilitates vertical and horizontal communication between occupational functions and levels, and is judged by the respondents to be very good. Regular meetings are held between managers at different levels, in order to draw up guidelines and to discuss solutions to any problems that may arise.
The benefits of the nature of care provided
Another important factor is the nature of the care and treatment given to patients at the departments. Work in the departments is cooperation-oriented. Investigation, diagnosis, treatment and rehabilitation demand the involvement of several professional groups. In many cases the doctor cannot make a medical decision without having heard for example how the assistant nurse or the occupational therapist perceives the patient's functional ability. Planning rounds are mentioned as examples of activities that involve several categories of personnel. In one department the sense of belonging is created in the need for working together in caring for patients.
It's a question of teamwork. You don't work as a doctor in isolation. You've got to have the whole team, all the personnel categories; otherwise the care won't be good. (14)
The sense of belonging can also be strengthened by making comparisons with similar departments at other hospitals. This could be exemplified by a statement from a respondent from one of the departments where highly specialised skill has been developed.
The speciality that we have is a very narrow field and is rather unique, and of course it feels good to be doing something exclusive. (8)
Organisational approaches
Leadership and followership attitudes
The respondents indicate that it is important that the heads of department work for development, for example strengthening and preserving their departments' specialities, particularly in times of cuts and rationalisation.
Our boss has been strong and has shown that he knows what he's talking about, standing up to them, showing that it works and that it's going to work/.../ My head of department is really good at having something up his sleeve that might be useful for the next round of cuts. (16)
The value of a clear division of responsibility and authority between levels and individuals is a recurrent theme in the interviews, and permeates all parts of the organisation. The group of assistant nurses who remained after the cutbacks of the 90s are valued and defended, and their competence is emphasised by for example delegating to them work tasks that are normally those of the registered nurses. The intention is that all personnel should use their competence to take responsibility for and influence their work.
I think it's important that you give the staff quite a lot of freedom, that they feel that they can use their competence, that I trust them. (7)
At the same time the managers feel that there is a risk that people who need clear guidelines become frustrated when they are given freedom. For this reason, they say, it is necessary to have a balance between freedom and control.
Many employees might think that it's a little bit frustrating with too much freedom, as they want rather clear guidelines. (7)
The informants describe a leadership characterised by inspiration and support towards the co-workers. The management encourages the co-workers to talk to each other about difficult meetings with patients, but it is the co-workers' responsibility to use all available opportunities for support. The co-workers say that they have the support of the management, because they are allowed not to be efficient every single minute; they are given time for reflection.
If we feel that it's hard to go in to a patient we must be able to say so. Then two of us go in instead, or someone else goes in. After all, we've got patients with serious illnesses, young people with brain tumours, and that can be tough to deal with. (9)
In terms of financial cuts and savings, the staff in the two present departments have experienced the same demands as those of other departments at the hospital. They have grappled with cultural clashes when wards have been merged and have had to solve different conflicts between personnel. In this respect these departments are no different from others. What characterises them is rather that managers and co-workers are prepared to solve difficulties that arise themselves. When they have not been able to solve the problems themselves, they have sought help from external experts. Now and then they have been forced to make organisational changes in order to deal with conflicts that have arisen at the departments.
The co-workers are willing to do their bit to help out and cover the staff vacancies that arise. According to the respondents there is a risk that they take on too much extra work out of loyalty. The heads of departments try not to overwork the staff, but do not always succeed. It is easy to exceed the limit for what is a reasonable workload when, according to the heads of departments, the co-workers are willing and interested in the work and readily take on extra work tasks. The heads of department endeavour to give the staff the opportunity to recover their strength between their work shifts. The following excerpt illustrates the reciprocity in the leader – co-worker relationship.
Being willing to help, covering for each other, helping each other, and this knowledge of each other and personal chemistry that's sort of part of the atmosphere here.... I think all that has been really good for us when we've been under stress. (13)
Learning activities
Managers encourage the co-workers both to use and deepen their competence, for example by encouraging them to participate in groups which develop new knowledge within a particular area, such as care of brain tumour patients or cytotoxic therapy. These interest groups are responsible for getting information and passing it on to their colleagues.
This multidisciplinary way of working, with everyone focussing on a particular complex of problems, has been rather successful. (13)
The transfer of knowledge has a social value and takes place within and between professional groups. Various opportunities are used for the transfer of knowledge, for example during a break, when a doctor talks about a case: "patient of the week". Another forum where transfer of knowledge takes place is planning rounds, which do not begin until all the professional groups are present. The following excerpt illustrates the importance of everyone's involvement:
We've got to work together with focus on the patient; physiotherapists, occupational therapists, and especially the assistant nurses are there and have their say, because they're the ones who really see the patient. (15)
No matter which professional group or department the respondents represent, the fact that the staff help and learn from each other is a recurrent theme in their statements. There is an open climate, which makes it possible to ask anybody at any time about anything.
They [nurses] really get a lot of help when they're new from those who've been there a while, and of course that's the kind of culture we want to keep. (15)
Apart from these efforts to deepen and use the competence of the co-workers, external networks have been developed among all the occupational groups. These networks are said to contribute to learning and to the development of self-confidence, in that knowledge and skills are acknowledged and anchored outside their own department. Physicians participate in several networks, both nationally and internationally. As far as registered nurses are concerned, it is above all a question of national contacts; something that the assistant nurses have made tentative attempts to establish, while the medical secretaries mainly work in local networks.
Supportive activities
Resources, above all staff resources, are insufficient at both departments, in the same way as at other workplaces. At the same time, the staff at these departments work with patients who are seriously and/or chronically ill. According to the respondents, this means that the staff can get into conflict situations. For this reason routines have been developed, where doctors, registered nurses and assistant nurses make joint decisions about what is to be prioritised for the individual patient if the resources are insufficient. In this way the responsibility for prioritising is removed from the individual caregiver.
We have a written list of priorities. Some days it must be OK that the patients don't get up. We have to set our sights at such a level that the staff can manage, but they shouldn't have to make that decision themselves. (1)
Care of patients who are seriously ill puts a strain on the staff. A situation that is mentioned by several interviewees is the care of patients with incurable diseases, and where the staff can identify themselves with the patient's situation. Apart from being open to getting support from one another in their everyday work there are scheduled routines so that the staff can discuss their experiences, depending on their various needs, with both workmates and other professional actors, e.g. social workers with special training in counselling.
If a patient is in a lot of pain and gets no relief for this pain, no matter what you do, this can lead to serious anxiety in the patient – and of course very often this really gets to you/.../ I think that talking about the care situations we've experienced as difficult makes us understand each other better. (17)
Apart from support in their work there are social activities for staff, both in their leisure time and during working hours, e.g. walks, keep fit classes etc. But according to the respondents, health-promoting activities also include that little gesture of caring about and noticing the co-workers, and of creating a good atmosphere. One of the heads of ward says that such measures to promote well-being are perceived by the staff to mean that the head is involved:
I've been given credit for being a head of ward who cares and is involved. I mean they might say 'We know you're always there for us' (16)
At one of the departments, there has been a tradition right from the start, more than 10 years ago, of going off on a study trip every year, and this is something that all of those who were interviewed at that department valued very highly. The trips are considered to have contributed to a good atmosphere, partly because the staff had got to know each other in a different context, and partly because they had come into contact with other workplaces, and the unique collective knowledge and skills of their department had been acknowledged.
Then we all go on a so-called department trip once a year so that we can be together like 24 hours a day in another environment. I think that's really valuable/.../ and then we usually think that we're really well off compared with other hospitals abroad. (11)
Organisational climate
Trust
In the interviews there is an expression of trust for each other; a trust that the respondents feel depends on the fact that they have confidence in each other's knowledge and skills. Every individual is trusted to carry out her/his duties. This confidence exists between leaders and co-workers, just as it exists between and within different professional groups.
Everyone is tolerant; we don't check up on each other that this and that have been done. We all have our own roles, everyone has their own thing and we know who does what. (16)
The sense of security that has developed makes it easier to be able to deal with difficulties and with individual shortcomings, but it is also the basis for enjoying one's work. Trust, security and the joy of working together means that people are happy at work and there is a good team spirit. This means that they perceive themselves to be equipped to deal with the ups and downs in the burden of health care and staff shortages: situations that can make the small department vulnerable.
Team spirit
At the same time as the team spirit is a distinguishing feature of the departments, many of the work tasks are independent ones. A characteristic feature of the studied departments is that cooperation between all professional groups is well developed. Which groups work together depends on the kind of work and the needs of the patients.
Cooperation between different groups is a very characteristic feature of our everyday work in the ward ... we're very dependent on each other, in fact everything is centred on the patient. (5)
Teamwork strengthens the feeling of participation. In the same way, the fact that the levels in the hierarchy are close together helps to involve the co-workers in the decision-making. The team is a contributing factor for the co-workers to be able to influence both the care of the patients and their own working conditions. Working together with the focus on the patient is another factor that contributes to decreasing the hierarchy at the departments.
R: There's a very open atmosphere, there's no old-fashioned hierarchy
I: How come you don't have that in your department any more?
R: I don't know, because it hasn't really... I can't really remember that we've ever had anything like that in all the years I've worked there... There have sort of been very good doctors, but perhaps it's the fact that you've got to see the patient as a whole person at our department. (4)
Professionalism
The team spirit improves loyalty towards one's own occupational group as well as towards the entire working group. There are expressions of pride regarding both the common knowledge and ability in the present departments, as well as regarding professional competence:
Our head of department is very keen on us developing and learning, and really wants us to know our stuff. (3)
Belonging to a unique group (being specialists) promotes a professional attitude, which includes treating the patient correctly, as well as providing good treatment for the illness. According to the respondents there are many stressful situations. The staff support each other in dealing with these situations in a professional manner, together with workmates in collegial or professionally led counselling.
It can be really hard-going and then we have to help each other and sort of see the whole thing from a professional angle, but at the same time we've got to be allowed to have feelings too. (17)
Discussion
In the results the picture of the departments might be considered too positive. Actually booth departments have gone through several difficulties such as mergers, and never-ending cuts in resources. Other difficulties have been conflicts about the philosophy of caring, as well as ongoing problems with organisational structural reconstruction. Consequently, these departments faced problems in the same way as others did. However, the intention of this study has been to concretise and increase understanding of what contributes to a healthy working life during a period of downsizing and restructuring. The results cannot provide definite answers concerning what are generally seen as determining factors for a healthy workplace, since there are certain given conditions for the two departments being studied with regard to the cooperation-oriented nature of their care and their small scale. Nevertheless, it is necessary to once again point out that these departments have faced the same demands for economic cuts and restructuring processes as the other departments at the hospital [38]. As a consequence of this, the proportion of their original staff has diminished from 1994, which has amongst other things meant an increasing trend to work hard. On the other hand, relatively fewer assistant nurses were laid off at the two selected departments compared with the rest of the hospital. In spite of this, the health of the personnel was more stable than that of other staff at other departments [5]. It is against this unstable organisational structure that the results are to be understood.
The informants comment on the value of belonging to a small department. According to Hodson [39], organisational size is a key determinant for job satisfaction and being able to feel pride in relation to one's work. Due to their size the departments in this study can continuously review, get to know each other and build up a sense of loyalty towards each other. In addition to this they have been able to develop a professional attitude with a clear professional identity. The opportunities for recurring personal interaction at small workplaces seem to increase the social capital at the workplace. A high degree of social capital, i.e. that the work team has a social network, norms and trust that facilitates cooperation, implies greater levels of satisfaction and quality of life at work [40,41]. The loyalty that is developed can also be explained in terms of social cohesion, i.e. a sense of belonging to a friendly, cohesive community [42]. Social contacts are thus a motivating factor at work and a determinant of health [22]. At the hospital in this study as a whole it emerged among other things that low group cohesion and low satisfaction with workmates were significantly associated with short-term sick leave during the period of organisational instability [5]. In other words, social capital and cohesion may have contributed to a stable health development at the departments in question, as it is known that work-related social support contributes to lower levels of work stressors [43]. Another contributing factor to the sense of cohesion may be that those who do not fit into the small department leave the organisation. A recently published study of long-term sick leave among women found that there was a greater risk of women tumbling out of the system in big and hierarchically organised workplaces [44]. From that perspective the small departments studied might be seen as protectors from long-term sick leave.
Both departments are highly specialised, and the staff are united by the clear common goal of providing the patients with care of high quality. Departmental goals are not enough in themselves; they must permeate the work of the department. According to Bass and Avolio [24], the ability of managers to empower their employees contributes to developing a climate where everyone works together to reach the common goals. The managers in this study, both at department level and ward level, appear to have the ability to get their co-workers to stretch their limits. Being given the opportunity to stretch yourself at work is considered stimulating. The managers seem to empower the co-workers to take responsibility for their work. Such an empowering leadership style [23] makes it easier for the employees to experience their competence and power to fulfil common goals [25]. The employees are given a more active role from their managers and thereby an opportunity to expand their decision latitude [13,14]. An empowering leadership style contributes to a more equal manager – employee relation. However, it is easy to exceed the limit for what is a reasonable workload when, as in this study, co-workers are willing and interested in their work, and prepared to take on extra tasks. There is a risk that important and stimulating work tasks become so motivating that they lead to illness instead of healthy motivation [2]. The managers in this study are aware of the danger of over-stimulating, but state that they may nevertheless fail to pay attention to the limits of individual co-workers' capacity.
In order to live up to the ambition of providing patients with specialised care of high quality, professional knowledge and skills are required, as well as well-developed cooperation, where everyone's competences are valuable contributions. The heads of department give their co-workers the opportunity to share the responsibility so that they can develop their competence. In this way the work environment can be characterised as empowerment-oriented, which can develop efficiency and job satisfaction at the workplace [26]. At a small and furthermore reduced work unit, the value and usefulness of each co-worker's efforts appear to be even more obvious.
Learning has high status at the departments. The management has realised the importance of competence development and, despite cuts, invests in integrating learning in the everyday work. Learning in everyday work is a resource that contributes to increasing the co-workers' sense of manageability [11]. In this way the individual is given the opportunity of having control over situations and his/her actions [12]. A learning organisation is also a positive determinant for increasing work motivation and making the work challenging as well as comprehensible and meaningful [11,12]. The results from the departments in this study are reminiscent of those that appeared in the interview studies conducted at the same hospital [2-4], where the female informants expressed a strong desire to work towards the same goal in a cooperating and learning team. For them this highly valued request was not a reality; however, this was the case for the two health-promoting departments. The competence development that the two departments invest in is directed towards the core activities of their respective departments. In order to bring about effective everyday learning there must be an opportunity for dialogue [45]. The management at the departments makes a conscious effort to provide opportunities for dialogue between individuals, within and between groups, and between different levels in the hierarchy. In this way they appear to have managed to break the hierarchical structures. At the same time as the teamwork has been developed, they have managed both to create confidence in each other and to retain the authority of the professional groups without creating jealously guarded territory. A possible interpretation of the result of this study is that the communication between the hierarchies is both top-down and bottom-up, which means that the management appears to have broken the communication pattern which the head nurses in Nilsson's study considered to be mostly top-down [46].
The style of management at the departments can be described as transformational, since the management strives to inspire and motivate co-workers to learn and develop. Intellectual stimulation as one aspect of transformational leadership deals with motivating employees to tax their resources to be innovative and creative, both in developing work tasks and solving problems [23]. The leadership style used in the departments leads in turn to the important components active problem-solving, support for coping, good communication and learning organisation. The managers are prepared to give their co-workers support and recognition. Personal recognition [47] could also define transformational style. In this respect the small size of the studied department could be an advantage, as managers and co-workers have many opportunities to communicate. By applying 'management by walking around' they could more easily pay attention to the individual employee [23].
When the organisation is in the middle of an unstable period, as is the case here, with increasing and conflicting work demands and less time to plan work, the co-workers' need for recognition, will probably be even more apparent. Recognition through feedback from the management can serve as an energy-giving activity [2-4] and thus a health promotion determinant. During the adjustment and restructuring phases, there has been an obvious trend that the personnel have worked harder and harder with a risk of overload [5]. At the present departments there is an awareness of these risks. There is a conscious managerial responsibility to keep a watchful eye on overload and to provide opportunity for recovery and time for reflection. Within this there is also the importance of integrating a certain measure of social activities in order to strengthen the group feeling.
There is a balancing act between giving co-workers responsibility and relieving them of it, as the management of the departments has done in a delicate way. Allowing co-workers to take responsibility for their work and their own development is good for productivity and efficiency [23], for work satisfaction [26,30] for decision-making [46] and as stress prevention [32]. The positive effects of being given, and taking, responsibility must be weighed against the risk of overloading the co-workers. In this balancing act the management uses empowerment strategies to delegate responsibility for carrying out the work parallel to strategies for prioritising work tasks. The support, which the co-worker perceives, helps him/her to gain control over the work, and this may have contributed to the health trend being stable at the departments [19,21].
Conclusion
Taken together the core determinants seem to be work-pride and confidence. There is a pride in what is achieved in the departments; individually and together. This pride also includes professional pride, common knowledge and skills, as well as belonging to a certain department. Confidence is based on the strong support that exists at the workplace from managers and co-workers; everyone is there for the patients and for each other. The specific tasks of the departments contribute to the sense of belonging. The confidence in and respect for each other is strengthened when the co-workers cooperate with each other, and when there is collective responsibility for prioritising and for the development of knowledge and skills. In addition to this, the fact that the value of all personnel categories is emphasised by the managers, not only in words but also in action, appears to contribute to the expressed confidence. Our core determinants augment the well-established concepts manageability, comprehensiveness and meaningfulness. The core determinants interpreted in the themes organisational conditions, approaches and climate seem to be favourable conditions, functioning as a buffer against general negative effects of downsizing observed elsewhere in the hospital, and in the literature.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KN: Study design, data collection, data analysis, writing the manuscript
AH: Study design, co-analyser, participating in writing the manuscript
ILP: Participating in writing the manuscript
TT: Contributed with discussions of consistence in the analysis
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We would like to thank the participants for generously sharing their experiences. This study was supported by grants from the Swedish Council for Working Life and Social Research, and Örebro County Council.
==== Refs
Hertting A Theorell T Physiological changes associated with downsizing of personnel and reorganisation in the health care sector Psychother Psychosom 2002 71 117 22 11844949 10.1159/000049355
Hertting A Nilsson K Theorell T Sätterlund Larsson U Personnel Reductions and Structural Changes in Health Care – Worklife experiences of medical secretaries J Psychosom Res 2003 54 161 70 12573738 10.1016/S0022-3999(02)00474-9
Hertting A Nilsson K Theorell T Sätterlund Larsson U Downsizing and reorganization: demands, challenges and ambiguity for registered nurses J Adv Nurs 2004 45 145 54 14705999 10.1046/j.1365-2648.2003.02876.x
Hertting A Nilsson K Theorell T Sätterlund Larsson U Assistant Nurses in the Swedish Health Care Sector during the 1990s – A Hard-hit Occupational Group with a Tough Job Scand J Public Health 2005 33 107 13 15823971 10.1080/14034940410019154
Petterson I-L Hertting A Hagberg L Theorell T Are trends in work and health conditions interrelated? A study of hospital employees in the Swedish 1990s J Occup Health Psychol 2005 10 110 20 15826222 10.1037/1076-8998.10.2.110
Landsbergis P Cahill J Schnall P The impact of lean production and related new systems of work organization on workers' health J Occup Health Psychol 1999 4 80 130
Ferrie J Is job insecurity harmful to health? J R Soc Med 2001 94 71 6 11234203
Kivimäki M Vahtera J Griffiths A Cox T Thomson L Cooper CL Sickness absence and organisational downsizing The organisation in crisis Downsizing, restructuring and privatisation 2000 Oxford: Blackwell publishers 78 94
Pelikan J Lobnig H Health-promoting hospitals World Health 1997 50 24 6
Quick JC Camara WJ Hurrell JJ Johnson JV Piotrkowski CS Sauter SL Spielberger CD Introduction and historical overview J Occup Health Psychol 1997 2 3 6 9552274 10.1037//1076-8998.2.1.3
Antonovsky A Health, stress and coping New perspectives on mental and physical well-being 1987 San Francisco: Jossey-Bass
Forsberg E Starrin B Frigörande kraft Empowerment som modell i skola, omsorg och arbetsliv [Liberating power Empowerment as a model in school, social care and working life] 1997 Göteborg: Gothia
Karasek RA Theorell T Healthy Work Stress, Productivity and the Reconstruction of Working Life 1990 New York: Basic Books Inc Publishers
Theorell T Karasek RA Eneroth P Job strain variations in relation to plasma testosterone fluctuations in working men – a longitudinal study J Internal Med 1990 227 31 6 2299296
Theorell T Emdad R Arnetz B Weingarten A-M Employee effects of an educational program for managers at an insurance company Psychosom Med 2001 63 724 33 11573019
Jackson SE Participation in decision making as a strategy for reducing job-related strain J Appl Psychol 1983 68 3 19 10.1037//0021-9010.68.1.3
Bue Bjorner J Sondergaard Kristensen T Orth-Gomér K Tibblin G Sullivan M Westerholm P Self-rated Health – A Useful Concept in Research, Prevention and Clinical Medicine 1996 Stockholm: Forskningsrådsnämnden
Aronsson G Lindh T Långtidsfriskas arbetsvillkor: En populationsstudie Arbete och Hälsa 2004:10 [The working conditions of long-term healthy individuals: A population study Work and Health 2004:10] 2004 Stockholm: Arbetslivsinstitutet
Bond FW Bunce D Job control mediates change in a work reorganisation intervention for stress reduction J Occup Health Psychol 2001 6 290 2 11605824 10.1037//1076-8998.6.4.290
Vingård E HAKuL – Hållbar arbetshälsa i kommuner och landsting [Sustainable health at work in municipalities and county councils] 2004 Stockholm: Karolinska institutet
Noblet A Building health-promoting work settings: identifying the relationship between work characteristics and occupational stress in Australia Health Promot Int 2003 18 351 59 14695366 10.1093/heapro/dag407
Janssen PPM deJonge J Bakker AB Specific determinants of intrinsic work motivation, burnout and turnover intentions: a study among nurses J Adv Nurs 1999 29 1360 69 10354230 10.1046/j.1365-2648.1999.01022.x
Bass BM Transformational Leadership: Industrial, Military, and Educational Impact 1998 New Jersey: Lawrence Erlbaum Associates
Bass BM Avolio B Improving organizational effectiveness through transformational leadership 1994 London: Sage
Hollander EP Offerman LR Power and leadership in organizations: Relationships in transition Am Psychol 1990 45 179 189 10.1037//0003-066X.45.2.179
Spence Laschinger HK Finega J Shamian J Promoting Nurses' Health: Effect of Empowerment on Job Strain and Work Satisfaction Nurs Econ 2001 19 42 53
Sparks JR Schenk JA Explaining the effects of transformational leadership: an investigation of the effects of higher-order motives in multilevel marketing organisations Journal of Organizational Behavior 2001 22 849 69 10.1002/job.116
Cass MH Ling SO Faragher EB Cooper CL A meta-analysis of the relationship between job satisfaction and employee health in Hong Kong Stress and Health 2003 19 79 95 10.1002/smi.959
Medley F Larochelle DR Transformational Leadership and Job Satisfaction Nurs Manage 1995 26 64 68
Marks-Moran D Reconstructing nursing: evidence, artistry and the curriculum Nurse Educ Today 1999 19 3 11 10222965
Pawar BA Eastman KK The Nature and Implications of Contextual Influences on Transformational Leadership: A Conceptual Examination Acad Manage Rev 1997 22 80 109
Stordeur S D'hoore W Vandenberghe C Leadership, organizational stress, and emotional exhaustion among hospital nursing staff J Adv Nurs 2001 35 533 42 11529953 10.1046/j.1365-2648.2001.01885.x
Bernin P Theorell T Patterns for successful leadership in health care 2003 Stockholm: IPM National Institute for Psychosocial Factors and Health Retrieved from WWW 2003-04-10
Silverman D Interpreting Qualitative Data Method for Analysing Talk, Text and Interaction 2001 London: Sage
Mishler EG Research interviewing Context and Narrative 1986 Cambridge Ma: Harvard University Press
Edwards D Discourse and cognition 1997 London, Thousands Oaks, New Delhi: Sage
HSFR-Etikregler Ethical Principles for Research in Humanistic and Social Science Research 1990 Retrieved from WWW 2003-04-10
Hertting A The Health Care Sector: A Challenging or Draining Work Environment. Psychosocial Work Experiences during the Swedish 1900s PhD Diss 2003 Stockholm: Karolinska Institutet
Hodson R Dignity at Work 2001 Cambridge: University Press
Requena F Social Capital, Satisfaction and Quality of Life in the Workplace Soc Indic Res 2003 61 331 60 10.1023/A:1021923520951
Langbein L Jorstad C Productivity in the Workplace: Cops, Culture, Communication, Cooperation, and Collusion Polit Res Q 2004 57 65 80
Patrick DL Wickizer TM Amick BC, Levine S, Tarlov AR, Walsh CD Community and health Society and Health 1995 New York: Oxford University Press 46 73
Snow DL Swan SC Raghavan C Conell CM Klein I The relationship of work stressors, coping and social support to psychological symptoms among female secretarial employees Work Stress 2003 17 241 63 10.1080/02678370310001625630
Renstig M Sandmark S Kvinnors sjukskrivningar Intervjuer med 25 långtidssjukskrivna kvinnor i Mellansverige [Women on the sick list Interviews with 25 women in central Sweden who are on the long-term sick list] 2005 Karolinska institutet, Institutionen för folkhälsovetenskap, Avdelningen för yrkesmedicin
Barge JK Little M Dialogical Wisdom, Communicative Practice, and Organisational Life Commun Theory 2002 12 375 97 10.1093/ct/12.4.375
Nilsson K Mandat – makt – management: en studie av hur vårdenhetschefers ledarskap konstrueras. [Mandate – power – management: a study of nursing managers' experience of leadership in the health care sector] PhD Diss 2003 Göteborg: Acta Universitatis Gothoburgensis
Rafferty AE Griffin MA Dimensions of transformational leadership: Conceptual and empirical extensions The Leadership Quarterly 2004 15 329 54 10.1016/j.leaqua.2004.02.009
|
16137331
|
PMC1208911
|
CC BY
|
2021-01-04 16:28:57
|
no
|
BMC Public Health. 2005 Sep 1; 5:92
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-92
|
oa_comm
|
==== Front
BMC PhysiolBMC Physiology1472-6793BioMed Central London 1472-6793-5-131610917510.1186/1472-6793-5-13Research ArticleExpression and subcellular localization of aquaporin water channels in the polarized hepatocyte cell line, WIF-B Gradilone Sergio A [email protected] Pamela S [email protected] Patrick L [email protected] Raúl A [email protected] Nicholas F [email protected] Instituto de Fisiología Experimental, Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Facultad de Ciencias Bioquímicas y Farmacéuticas, Universidad Nacional de Rosario, 2000 Rosario, Santa Fe, Argentina2 Center for Basic Research in Digestive Diseases. Mayo Medical School, Clinic and Foundation, Rochester, MN, USA2005 18 8 2005 5 13 13 13 6 2005 18 8 2005 Copyright © 2005 Gradilone et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent data suggest that canalicular bile secretion involves selective expression and coordinated regulation of aquaporins (AQPs), a family of water channels proteins. In order to further characterize the role of AQPs in this process, an in vitro cell system with retained polarity and expression of AQPs and relevant solute transporters involved in bile formation is highly desirable. The WIF-B cell line is a highly differentiated and polarized rat hepatoma/human fibroblast hybrid, which forms abundant bile canalicular structures. This cell line has been reported to be a good in vitro model for studying hepatocyte polarity.
Results
Using RT-PCR, immunoblotting and confocal immunofluorescence, we showed that WIF-B cells express the aquaporin water channels that facilitate the osmotically driven water movements in the liver, i.e. AQP8, AQP9, and AQP0; as well as the key solute transporters involved in the generation of canalicular osmotic gradients, i.e., the bile salt export pump Bsep, the organic anion transporter Mrp2 and the chloride bicarbonate exchanger AE2. The subcellular localization of the AQPs and the solute transporters in WIF-B cells was similar to that in freshly isolated rat hepatocytes and in intact liver. Immunofluorescent costaining studies showed intracellular colocalization of AQP8 and AE2, suggesting the possibility that these transporters are expressed in the same population of pericanalicular vesicles.
Conclusion
The hepatocyte cell line WIF-B retains the expression and subcellular localization of aquaporin water channels as well as key solute transporters for canalicular bile secretion. Thus, these cells can work as a valuable tool for regulatory and mechanistic studies of the biology of bile formation.
==== Body
Background
Hepatocytes are polarized epithelial cells that possess well defined apical (canalicular) and basolateral (sinusoidal) plasma membrane domains. Bile secretion involves the movement of water across hepatocyte plasma membrane domains in response to transient osmotic gradients generated by active solute transport into the canalicular space of bile acids, glutathione and bicarbonate [1]. The main canalicular solute transporters are the bile salt export pump Bsep, the organic anion transporter Mrp2 and the chloride bicarbonate exchanger AE2 [2,3].
We and others recently reported that hepatocytes express aquaporin (AQP) water channels [4-10], a family of integral membrane proteins that increase cell membrane water permeability facilitating passive osmotically driven water movement [11,12]. In hepatocytes, the water channels aquaporin-8 (AQP8) and aquaporin-0 (AQP0) are primarily located within the cell in a vesicular compartment, and AQP8 redistributes to the canalicular membrane under a choleretic stimulus [5,10,13]. AQP8 it was recently described also localized in mitochondria [14]. Aquaporin-9 (AQP9) is found principally on the basolateral membrane [10]. Our previous studies suggest that water channels play an important role in the transcellular transport of water during primary bile secretion by hepatocytes [5,10,13,15,16].
Unfortunately, the bile formation process is difficult to study in vitro. In primary cultures of rat hepatocytes, downregulation of both basolateral and canalicular solute transporters occurs [17] and bile acid uptake gradually decreases and disappears after 1 to 4 days [18]. Moreover, in freshly isolated rat hepatocytes, canalicular bile acid secretion appears limited [19]. In addition, the vectorial transport of solutes is lost in many hepatoma cell lines such as HepG2 [20,21], HTC [22], and Fao [23].
The WIF-B cell line is a highly differentiated and polarized rat hepatoma/human fibroblast hybrid, which forms bile canaliculi-like structures [24]. This cell line has been shown to be a good model for studying hepatocyte polarity [25,26], protein secretion [24], bile acid transport [18] and protein transport [27]. Furthermore, WIF-B cells express the basolateral Na+-taurocholate cotransporter, Ntcp [28] and the canalicular conjugate export pump, Mrp2 [29].
To further explore the usefulness of the WIF-B cell line as an in vitro model for regulatory and mechanistic studies of bile secretion biology, we investigated the presence and localization of aquaporins and the principal solute transporters involved in canalicular bile formation.
Results
Expression of aquaporins and canalicular solute transporters in WIF-B cells by RT-PCR
RT-PCR was run for each of the three AQPs and the three solute transporters on total RNA derived from WIF-B cells. cDNA from freshly isolated rat hepatocytes was used as positive control. As shown in Figure 1, WIF-B cells express mRNA of rat AQP0, AQP8 and AQP9 (Figure 1A) and AE2, Mrp2 and Bsep (Figure 1B). The PCR products were sequenced and the identities of the amplicons, verified by data base homology searches (BLAST; NCBI, National Institutes of Health), were consistent with the predicted rat genes.
Expression of aquaporins and canalicular solute transporters in WIF-B cells by immunoblotting
As shown in Figure 2A, WIF-B cells express AQP0, AQP8, and AQP9 proteins. WIF-B plasma and intracellular membranes, as well as hepatocyte homogenates showed a 28 kDa band on the immunoblotting for AQP0. The immunoblot for AQP8 shows the presence of a 34 kDa band on WIF-B plasma and intracellular membranes. AQP9 is also positive in the plasma membrane fraction of WIF-B cells, showing the 32 kDa band. Figure 2B shows the immunoblotting for the three solute transporters: AE2 (170 kDa) and Bsep (160 kDa) are present on the plasma membrane of WIF-B and, as described [29], Mrp2 (190 kDa) is also present in the cell line.
Confocal immunofluorescence
Immunofluorescent staining of WIF-B cells further confirmed the expression of the three AQPs (Figure 3). The subcellular localization is similar to that of rat hepatocytes, i.e. AQP0 mainly intracellular, AQP8 mostly in intracellular vesicular structures throughout the cytosol and AQP9 on the basolateral membrane. The immunofluorescence for AE2, Bsep and Mrp2 showed pericanalicular localization (Figure 3). Immunofluorescent costaining was performed for AQP8 and AE2. These molecules showed colocalization (Figure 4), suggesting the presence of a population of pericanalicular vesicles containing both AQP8 and AE2.
Discussion
The major finding reported here relate to the expression and subcellular localization of aquaporin water channels in the WIF-B cell line. Using RT-PCR, immunoblotting and confocal immunofluorescence we showed that (i) WIF-B cells express the aquaporin water channels that facilitate the osmotically driven water movements, i.e. AQP8, AQP9, and AQP0; as well as the key solute transporters involved in the generation of the canalicular osmotic gradients, i.e., the bile salt export pump Bsep, the organic anion transporter Mrp2 and the chloride bicarbonate exchanger AE2. (ii) The subcellular localization of the AQPs and the solute transporters in WIF-B cells was similar to that in rat isolated hepatocytes and in whole liver. (iii) Immunofluorescent costaining studies showed intracellular colocalization of AQP8 and AE2, suggesting the possibility that these transporters are expressed in the same population of pericanalicular vesicles.
Hepatocyte bile secretion is formed by passive movement of water from plasma to the bile canaliculus in response to osmotic gradients established by the active secretion of solutes. The biliary excretion of bile salts, via the bile salt transporter Bsep, glutathione, via the organic anion transporter Mrp2, and HCO3-, via the Cl-/HCO3- exchanger AE2, are thought to be the major osmotic driving forces for canalicular bile flow [1]. Conceptually, the generation of bile flow is ultimately dependent on the molecular and functional expression of these transporters on the canalicular plasma membrane domain. AQPs are present in hepatocytes at both apical and basolateral plasma membrane domains as well as in intracellular vesicle compartments. Two of these AQPs can account for the water permeability of both hepatocyte plasma membrane domains, AQP8 modulating mainly the canalicular transport of water, and AQP9 facilitating its basolateral movement [5,10,13,15,16].
In order to further characterize the role of aquaporins in bile secretion, an in vitro cell system with retained polarity and expression of AQPs and the main solute transporters involved in bile formation is highly desirable. The present work shows that WIF-B cells meet these criteria, i.e. express the key solute transporters involved in the osmotic gradient generation (AE2, Bsep and Mrp2) and the channels that facilitates plasma membrane water movement (AQP8 and AQP9).
The WIF-B hybrid cell line stably retains all rat chromosomes, and only a dozen of human chromosomes [30]. The AQP8, AE2, Bsep, and Mrp2 genes have been mapped on the human chromosomes 16 [31], 7 [32], 2 [33], and 10 [34], respectively, which are not retained by the WIF line [35,36]. Furthermore, it was described that WIF-B cells express only the rat homolog of the bile salt transporter Ntcp [28], consistent with the absence of human chromosome 14, on which Ntcp has been mapped [37]. Therefore, the AQP8, AE2, Bsep, and Mrp2 expressed by these cells should be the rat genes. By contrast, the AQP0 and AQP9 genes are located on human chromosomes 12 and 15, respectively [38,39], which are present in WIF cells [35,36]. Thus, WIF-B cells may also express human AQP0 and AQP9.
The subcellular localization of the AQPs in WIF-B cells is similar to that found in isolated hepatocytes and liver from rat. AQP0 was found mainly intracellular and AQP9 exclusively on basolateral membranes. AQP8 showed mostly an intracellular vesicular structures localization, which could be of potential interest for further studies on the mechanisms involved in the hormone-regulated trafficking of AQP8 to the canalicular plasma membrane domain [5,10,13,40].
The canalicular/pericanalicular localization of the solute transporters is consistent with that described for rat liver or isolated hepatocytes. AE2 has been localized to the canalicular membrane domain [41,42]. Nevertheless, the canalicular activity of AE2 is increased in response to stimulation with cyclic AMP, and this increased activity can be blocked with colchicine, suggesting the microtubule-dependent targeting of pericanalicular vesicles containing this exchanger to the canalicular domain [43]. Rat liver Bsep is responsible for the biliary excretion of bile acids and therefore is key to the elaboration of canalicular bile [2]. Immunogold electron microscopy detection of Bsep revealed that the distribution of Bsep in rat hepatocytes is not restricted to the canalicular membrane, but also detected in vesicles close to the bile canaliculus [44]. Pericanalicular distribution of Bsep was also demonstrated by immunofluorescence staining of isolated rat hepatocyte couplets [45]. We found that Mrp2, responsible for the transport into bile of a variety of amphiphilic organic anions [2], is mainly localized on pericanalicular vesicles. This observation does not fully agree with a previous work showing an exclusively canalicular membrane localization of Mrp2 in WIF-B cells [29]. The fact that Mrp2 showed a different localization in our hands could be explained by different culture conditions. Furthermore, immunogold electron microscopy detection of Mrp2 in rat liver revealed that over 50% of Mrp2 resides in intracellular vesicles close to the canalicular membrane [46]; and pericanalicular vesicular distribution was demonstrated by confocal immunofluorescence in isolated rat hepatocyte couplets [45]. This pericanalicular localization gives rise to the possibility of studying transport trafficking to the apical membrane.
It is known that the hormone glucagon stimulates bile secretion [47,48]. As we previously described, glucagon is able to increase the osmotic water permeability of hepatocytes by triggering the translocation of AQP8 vesicles to the plasma membrane, specifically to the canalicular domain [13]. Although the actual osmotic gradients involved in glucagon-induced choleresis are unknown, these transient gradients are most likely created by the facilitated transport of HCO3- via the canalicular Cl-/HCO3- exchanger AE2 [47,48]. There is evidence to suggest that glucagon is also able to stimulate the vesicle trafficking of AE2 to the hepatocyte plasma membranes [43,48]. Since the immunofluorescence for AQP8 and AE2 showed colocalization, is attractive to speculate that the solute transporter and the water channel could be packaged in the same population of vesicles conforming a bile secretory unit as has been observed for AQP1, AE2 and the cystic fibrosis transmembrane regulator Cl- channels in cholangiocytes [49]. Thus, the existence of a novel transporting organelle containing functionally related proteins, that can account for solute-driven water transport, could be proposed in hepatocytes. This organelle could contain flux proteins playing integral roles in hormone-induced bile secretion.
Conclusion
Our study provides the first evidence of a hepatocyte cell line that retains the expression and subcellular localization of aquaporin water channels, as well as key solute transporters for canalicular bile formation, turning these cells into a valuable tool for regulatory and mechanistic studies of the bile formation biology.
Methods
Cell culture
WIF-B cells, kindly provided by Dr. Ann Hubbard (Johns Hopkins University, Baltimore, MD), were grown at 37°C under 5% CO2 in modified Ham's F12 medium supplemented with 5% fetal calf serum and 10 μmol/L hypoxanthine, 0.04 μmol/L aminopterin, and 1.6 μmol/L thymidine as described [24]. Cells were plated onto plastic dishes or glass coverslips at 3.8 × 104 cells/cm2. We used 10 to 14 day old cultures in all experiments, the time point at which the cells reached their maximal density and polarity [24].
RNA isolation
Total RNA was extracted from WIF-B cells or freshly isolated rat hepatocytes using Tri-Reagent (Sigma). Cells were lysed in 1 ml of Tri-Reagent/10 × 106 cells with 5 μl of Glyco-Blue (Ambion Inc., Austin, TX) added as a co-precipitant and stored at room temperature for 5 min. After addition of 0.1 ml of 1-bromo-3-chloro-propane/1 ml of Tri-Reagent, the samples were vortexed, incubated for 15 min at room temperature, and centrifuged at 12,000 × g for 15 min at 4°C. The upper, aqueous phase was collected and transferred to a new tube; to this, 0.5 ml of isopropanol was added per 1 ml of Tri-Reagent used for the initial lysis. The samples were incubated for 10 min and centrifuged at 12,000 × g for 15 min at 4°C. After removing the supernatant, the RNA pellet was washed with 1 ml of 75% ethanol and repelleted by centrifugation at 12,000 × g for 15 min at 4°C. RNA was resuspended in RNA Secure solution (Ambion), and the concentration and purity were determined by spectrophotometry.
Reverse transcription-polymerase chain reaction
5 μg of total RNA was reverse transcribed using an avian myeloblastosis virus reverse transcriptase system (Promega, Madison, WI). RNA was first incubated for 10 min at 70°C. The reaction mixture included reverse transcription buffer, 25 mM MgCl2, 10 mM deoxynucleotide triphosphates, avian myeloblastosis virus reverse transcriptase, RNasin ribonuclease inhibitor, and random primers in a final volume of 95 μl. This mixture was added to the total RNA and incubated for 10 min at room temperature and then 1 h at 42°C. Heating to 95°C for 5 min stopped the reaction. The AQPs, Mrp2, Bsep and AE2 cDNA were amplified using the polymerase chain reaction with specific primers for rat genes (Table 1). cDNA from freshly isolated rat hepatocytes and H2Od were used as positive and negative controls, respectively. The PCR products were electrophoresed in 1% agarose gels, and the bands were visualised by ethidium bromide staining. Sequencing was performed on all positive PCR products (Mayo Molecular Core Facility, Rochester, MN) to confirm the identity of the amplified genes.
Preparation of plasma and intracellular membranes
Cells were washed and sonicated in 0.3 M sucrose containing 0.1 mM phenylmethylsulfonyl fluoride and 0.1 mM leupeptin. Plasma and intracellular membranes were obtained by differential centrifugation, as previously described [5]. Proteins in the membrane fractions were assayed according to Lowry et al. [50], using bovine serum albumin as standard.
Immunoblotting
Solubilized membrane fractions were subjected to SDS-polyacrylamide gel electrophoresis and transferred to polyvinyldifluoride membranes. After blocking, blots were incubated overnight at 4°C with affinity-purified antibodies against AQP0, AQP8, AQP9 (1 μg/ml; Alpha Diagnostics International), AE2 (5 μg/ml; Alpha Diagnostics International), MRP2 or BSEP (5 μg/ml; Santa Cruz Biotechnology). The blots were then washed and incubated with horseradish peroxidase-conjugated goat anti immunoglobulin, and bands were detected by an enhanced chemiluminescence detection system. Autoradiographs were obtained by exposing polyvinyldifluoride membranes to Kodak XAR film.
Immunofluorescence and confocal microscopy
After culturing, cells were fixed with 2% paraformaldehyde for 10 min at room temperature, permeabilized with 0.2 % Triton X-100 for 2 min, and incubated overnight at 4°C with affinity-purified antibodies (10 μg/ml AQP0, 10 μg/ml AQP8, 10 μg/ml AQP9, 10 μg/ml AE2, 20 μg/ml Mrp2 or 20 μg/ml Bsep). After washing, coverslips were incubated with Alexa Fluor 488 or 594 conjugated secondary antibodies for one hour, and mounted with ProLong. Fluorescence localization was detected by confocal microscopy with a laser scanning microscope (Carl Zeiss LSM-510). Images were obtained with the same confocal settings for each set of experiments. With these settings no autofluorescence was detected. Controls using omission of primary or secondary antibodies revealed no labeling. Images were processed using Adobe Photoshop software.
Authors' contributions
SAG carried out preparation of intracellular and plasma membranes, the immunoblots and the confocal immunofluorescence studies and participated in the RT-PCR and sequence alignment and drafted the manuscript. PST participated in the western blotting and confocal immunofluorescence studies and helped to draft the manuscript. PLS carried out the cell culture and the RT-PCR studies and the sequence alignment. RAM participated in the design of the study and in the draft of the manuscript. NFL conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by the National Institutes of Health Grant DK 24031 and by the Mayo Foundation (NFL), and by Grant PICT 05-10590 (RAM) from Agencia Nacional de Promoción Científica y Tecnológica and by Fundación Antorchas (SAG).
Figures and Tables
Figure 1 RT-PCR for aquaporins and canalicular solute transporters in WIF-B cells. Total RNA from WIF-B cells was reverse transcribed using random primers and then PCR-amplified with primers designed to amplify a nonconserved region of each AQP and solute transporter cDNA. Isolated rat hepatocyte cDNA were used as positive controls. Lane 1, negative control; lane 2, positive control; lane 3, WIF-B cells.
Figure 2 Immunoblots for aquaporins and canalicular solute transporters in WIF-B cells. A, 50 μg of plasma (PM) and intracellular (IM) membranes from WIF-B cells were loaded onto 12% SDS-polyacrylamide gels. Immunoblots were performed using affinity-purified rabbit anti-AQP primary antibodies and goat anti-rabbit secondary antibodies and visualized via chemiluminescence. B, 50 μg of PM and IM from WIF-B cells were loaded onto 8% SDS-pliacrylamide gels. Immunoblots were performed using affinity-purified goat anti-Mrp2 or Bsep and rabbit anti-AE2. Rat hepatocyte homogenate (20 μg) were used as positive controls.
Figure 3 Confocal immunofluorescence for aquaporins and solute transporters in WIF-B cells. WIF-B cells were fixed, permeabilized, and labeled with anti-AQPs, AE2, Mrp2 or Bsep. Fluorescence localization was viewed by laser scanning confocal microscopy (see "Materials and Methods" for details). *, bile canaliculi structures.
Figure 4 Co-immunostaining of AQP8 and AE2 by confocal immunofluorescence. WIF-B cells were fixed, permeabilized, and labelled simultaneously with rabbit anti-AE2 and goat anti-AQP8. Fluorescence localization was viewed by laser scanning confocal microscopy (see "Materials and Methods" for details). *, bile canaliculi structures.
Table 1 Primers used for AQPs and solute transporters expression
Gene Primers (forward; reverse) Amplicon size, bp cDNA sequence location
AQP0 5'-acggctcaagagtgtttctga-3' 189 669-689
5'-tccccacagtctctttcttcat-3' 857-836
AQP8 5'-aagaccatgctgctaattcc-3' 275 423-442
5'-tccacaatgacagagaaacc-3' 697-678
AQP9 5'-tgttgtcattagcctcctgatc-3' 356 736-757
5'-tgaagaaagaactggatgaacg-3' 1091-1070
MRP2 5'-ctggttggaaacttggtcgt-3' 172 3719-3739
5'-caactgccacaatgttggtc-3' 3890-3870
BSEP 5'-cactggccttctggtatggt-3' 225 1275-1294
5'-gcttgtagccgtctcctgac-3' 1499-1479
AE2 5'-tctcgttctgcaagagcaacc-3' 270 2527-2547
5'-ttgttactgctgctgtctgcc-3' 2797-2777
==== Refs
Meier P Steiger B Molecular mechanisms in bile formation News Physiol Sci 2000 15 89 93 11390885
Muller M Jansen P Molecular aspects of hepatobiliary transport Am J Physiol 1997 272 G1285 G1303 9227463
Erlinger S Arias IM, Boyer JL, Fausto N, Jacoby W, Schachter D, Shafritz D Bile Flow The Liver: Biology and phatobiology 1994 New York: Raven 769 786
Elkjaer M Vajda Z Nejsum L Kwon T Jensen U Amiry-Moghaddam M Frokiaer J Nielsen S Immunolocalization of AQP9 in liver, epididymis, testis, spleen, and brain Biochem Biophys Res Commun 2000 276 1118 1128 11027599 10.1006/bbrc.2000.3505
Garcia F Kierbel A Larocca M Gradilone S Splinter P LaRusso N Marinelli R The water channel aquaporin-8 is mainly intracellular in rat hepatocytes and its plasma membrane insertion is stimulated by cyclic AMP J Biol Chem 2001 276 12147 12152 11278499 10.1074/jbc.M009403200
Calamita G Mazzone A Bizzoca A Cavalier A Cassano G Thomas D Svelto M Expression and immunolocalization of the aquaporin-8 water channel in rat gastrointestinal tract Eur J Cell Biol 2001 80 711 719 11824790 10.1078/0171-9335-00210
Nicchia GP Frigeri A Nico B Ribatti D Svelto M Tissue distribution and membrane localization of aquaporin-9 water channel: evidence for sex-linked differences in liver J Histochem Cytochem 2001 49 1547 1556 11724902
Elkjaer M Nejsum L Gresz V Kwon T Jensen U Frokiaer J Nielsen S Immunolocalization of aquaporin-8 in rat kidney, gastrointestinal tract, testis, and airways Am J Physiol Renal Physiol 2001 281 F1047 F1057 11704555
Tani T Koyama Y Nihei K Hatakeyama S Ohshiro K Yoshida Y Yaoita E Sakai Y Hatakeyama K Yamamoto T Immunolocalization of aquaporin-8 in rat digestive organs and testis Arch Histol Cytol 2001 64 159 168 11436986
Huebert R Splinter PL García F Marinelli R LaRusso NF Expression and localization of aquaporin water channels in rat hepatocytes. Evidence for a role in canalicular bile secretion J Biol Chem 2002 277 22710 22717 11932260 10.1074/jbc.M202394200
Agre P Kozono D Aquaporin water channels: molecular mechanisms for human diseases FEBS Lett 2003 555 72 78 14630322 10.1016/S0014-5793(03)01083-4
Masyuk A Marinelli R LaRusso N Water transport by epithelia of the digestive tract Gastroenterology 2002 122 545 562 11832467
Gradilone S García F Huebert RC Tietz P Larocca MC Kierbel A Carreras F LaRusso N Marinelli R Glucagon Induces the Plasma Membrane Insertion of Functional Aquaporin-8 Water Channels in Isolated Rat Hepatocytes Hepatology 2003 37 1435 1441 12774023 10.1053/jhep.2003.50241
Calamita G Ferri D Gena P Liquori GE Cavalier A Thomas D Svelto M The inner mitochondrial membrane has aquaporin-8 water channels and is highly permeable to water J Biol Chem 2005 280 17149 53 15749715 10.1074/jbc.C400595200
Carreras F Gradilone S Mazzone A Garcia F Huang B Ochoa J Tietz P LaRusso N Calamita G Marinelli R Rat hepatocyte aquaporin-8 water channels are down-regulated in extrahepatic cholestasis Hepatology 2003 37 1026 1033 12717383 10.1053/jhep.2003.50170
Marinelli R Tietz P Caride A Huang B LaRusso N Water transporting properties of hepatocyte basolateral and canalicular plasma membrane domains J Biol Chem 2003 278 43157 43162 12939275 10.1074/jbc.M305899200
Rippin S Hagenbuch B Meier P Stieger B Cholestatic expression pattern of sinusoidal and canalicular organic anion transport systems in primary cultured rat hepatocytes Hepatology 2001 33 776 82 11283839 10.1053/jhep.2001.23433
Bravo P Bender V Casio D Efficient in vitro vectorial transport of a fluorescent conjugated bile acid analogue by polarized hepatic hybrid WIF-B and WIF-B9 cells Hepatology 1998 27 576 583 9462660 10.1002/hep.510270236
Schramm U Fricker G Buscher H Gerok W Kurz G Fluorescent derivatives of bile salts. III. Uptake of 7b-NBD-NCT into isolated hepatocytes by the transport systems for cholyltaurine J Lipid Res 1993 34 741 757 8509713
Marchegiano P Carubbi F Tiribelli C Amarri S Stebel M Lunazzi G Levy D Bellentani S Transport of sulfobromophthalein and taurocholate in the HepG2 cell line in relation to the expression of membrane carrier proteins Biochem Biophys Res Com 1992 183 1203 1208 1567398 10.1016/S0006-291X(05)80318-3
Ziegler K Blumrich M Hummelsiep S The transporter for the HMG-CoA reductase inhibitor pravastatin is not present in HepG2 cells. Evidence for the nonidentity of the carrier for pravastatin and certain transport systems for BSP Biochim Biophys Acta 1994 1223 195 201 8086488 10.1016/0167-4889(94)90226-7
Von Dippe P Levy D Expression of the bile acid transport protein during liver development and in hepatoma cells J Biol Chem 1990 265 5942 5945 2318840
Blumrich M Zeyen-Blumrich U Pagels P Petzinger E Immortalization of rat hepatocytes by fusion with hepatoma cells.II.Studies on the transport and synthesis of bile acids in hepatocytoma (HPCT) cells Eur J Cell Biol 1994 64 339 347 7813521
Shanks M Casio D Lecoq O Hubbard A An improved polarized rat hepatoma hybrid cell line. Generation and comparison with its hepatoma relatives and hepatocytes in vivo J Cell Sci 1994 107 813 825 8056838
Ihrke G Neufeld E Meads T Shanks M Cassio D Laurent M Schroer T Pagano R Hubbard A WIF-B cells: an in vitro model for studies of hepatocyte polarity J Cell Biol 1993 123 1761 75 7506266 10.1083/jcb.123.6.1761
Decaens C Rodríguez P Bouchaud C Cassio D Establishment of hepatic cell polarity in the rat hepatoma human fibroblast hybrid WIF-B9; a biphasic phenomenon going from a simple epithelial polarized phenotype to an hepatic polarized one J Cell Sci 1996 109 1623 1635 8799849
Ihrke G Martín G Shanks M Schrader M Schroer T Hubbard A Apical plasma membrane proteins and endolyn-78 travel thrugh a subapical compartment in polarized WIF-B hepatocytes J Cell Biol 1998 141 115 133 9531552 10.1083/jcb.141.1.115
Konieczko E Rakston A Crawford A Karpen S Crawford J Enhanced Na+-dependent bile salt uptake by WIF-B cells, a rat hepatoma hybrid cell line, following growth in the presence of a physiological bile salt Hepatology 1998 27 191 199 9425937 10.1002/hep.510270130
Nies A Cantz T Brom M Leier I Keppler D Expression of the apical conjugate export pump, Mrp2, in the polarized hepatoma cell line, WIF-B Hepatology 1998 28 1332 1340 9794919 10.1002/hep.510280523
Chaumontet C Mazzoleni G Decaens C Bex V Cassio D Martel P The polarized hepatic human/rat hybrid WIF 12-1 and WIF-B cells communicate efficiently in vitro via connexin 32-constituted gap junctions Hepatology 1998 28 164 172 9657109 10.1002/hep.510280122
Viggiano L Rocchi M Svelto M Calamita G Assignment of the aquaporin-8 water channel gene (AQP8) to human chromosome 16p12 Cytogenet Cell Genet 1999 84 208 10 10393433 10.1159/000015260
Palumbo A Isobe M Huebner K Shane S Rovera G Demuth D Curtis PJ Ballantine M Croce C Showe L Chromosomal localization of a human band 3-like gene to region 7q35-7q36 Am J Hum Genet 1986 39 307 16 3020980
Jansen P Strautnieks S Jacquemin E Hadchouel M Sokal E Hooiveld G Koning J De Jager-Krikken A Kuipers F Stellaard F Bijleveld C Gouw A Van Goor H Thompson R Muller M Hepatocanalicular bile salt export pump deficiency in patients with progressive familial intrahepatic cholestasis Gastroenterology 1999 117 1370 9 10579978
van Kuijck M Kool M Merkx G Geurts van Kessel A Bindels R Deen P van Os C Assignment of the canalicular multispecific organic anion transporter gene (CMOAT) to human chromosome 10q24 and mouse chromosome 19D2 by fluorescent in situ hybridization Cytogenet Cell Genet 1997 77 285 7 9284939
Cassio D Hamon-Benais C Guérin M Lecoq O Hybrid Cell Lines Constitute a Potential Reservoir of Polarized Cells: Isolation and Study of Highly Differentiated Hepatoma derived Hybrid Cells Able to Form Functional Bile Canaliculi In Vitro J Cell Biol 1991 115 1397 1408 1955480 10.1083/jcb.115.5.1397
Griffo G Hamon-Benais C Angrand P Fox M West L Lecoq O Povey S Cassio D Weiss M HNF4 and HNFI as well as a Panel of Hepatic Functions Are Extinguishedand Reexpressed in Parallel in Chromosomally Reduced Rat Hepatoma-Human Fibroblast Hybrids J Cell Biol 1993 121 887 898 8491780 10.1083/jcb.121.4.887
Hagenbuch B Meier P Molecular cloning, chromosomal localization, and functional characterization of a human liver Na+/bile acid cotransporter J Clin Invest 1994 93 1326 31 8132774
Ma T Yang B Umenishi F Verkman A Closely spaced tandem arrangement of AQP2, AQP5, and AQP6 genes in a 27-kilobase segment at chromosome locus 12q13 Genomics 1997 43 387 9 9268644 10.1006/geno.1997.4836
Tsukaguchi H Weremowicz S Morton C Hediger M Functional and molecular characterization of the human neutral solute channel aquaporin-9 Am J Physiol 1999 277 F685 96 10564231
Gradilone S Carreras F Lehmann G Marinelli R Phosphatidylinositol-3-kinase is involved in the glucagon-induced translocation of aquaporin-8 to hepatocyte plasma membrane Biol Cell 2005 15859947
Meier P Knickelbein R Moseley R Dobbins J Boyer J Evidence for carrier-mediated chloride/bicarbonate exchange in canalicular rat liver plasma membrane vesicles J Clin Invest 1985 75 1256 1263 2985654
Martinez-Anso E Castillo J Diez J Medina J Prieto J Immunohistochemical detection of chloride/bicarbonate anion exchangers in human liver Hepatology 1994 19 1400 1406 8188169 10.1016/0270-9139(94)90234-8
Benedetti A Strazzabosco M Ng O Boyer J Regulation of activity and apical targeting of the Cl-/HCO3-exchanger in rat hepatocytes Proc Nat Acad Sci USA 1994 91 792 796 8290601
Gerloff T Stieger B Hagenbunch B Madon J Landmann L Roth J Hofmann A Meier P The sister of P-glycoprotein represents the canalicular bile salt export pump of mammalian liver J Biol Chem 1998 273 10046 10050 9545351 10.1074/jbc.273.16.10046
Soroka C Pate M Boyer J Canalicular export pumps traffic with polymeric immunoglobulin A receptor on the same microtubule-associated vesicle in rat liver J Biol Chem 1999 274 26416 26424 10473600 10.1074/jbc.274.37.26416
Dombrowski F Kubits R Chittattu A Wettstein M Saha N Haussinger D Electron-microscopic demonstration of multidrug resistance protein 2 (Mrp2) retrieval from the canalicular membrane in response to hyperosmolarity and lipopolysaccharide Biochem J 2000 348 183 188 10794730 10.1042/0264-6021:3480183
Lenzen R Hruby VJ Tavoloni N Mechanism of glucagon choleresis in guinea pigs Am J Physiol 1990 259 G736 G744 2173415
Alvaro D Della Guardia P Bini A Gigliozzi A Furfaro S La Rosa T Piat C Capocaccia L Effect of glucagon on intracellular pH regulation in isolated rat hepatocyte couplets J Clin Invest 1995 96 665 675 7635959
Tietz P Marinelli R Chen X Kole J McNiven M LaRusso N Agonist-induced coordinated trafficking of functionally-related transport proteins for water and ions in cholangiocytes J Biol Chem 2003 278 20413 20419 12660234 10.1074/jbc.M302108200
Lowry O Rosebrough N Farr A Randall R Protein measured with the Folin phenol reagent J Biol Chem 1951 193 265 275 14907713
|
16109175
|
PMC1208912
|
CC BY
|
2021-01-04 16:03:51
|
no
|
BMC Physiol. 2005 Aug 18; 5:13
|
utf-8
|
BMC Physiol
| 2,005 |
10.1186/1472-6793-5-13
|
oa_comm
|
==== Front
BMC PhysiolBMC Physiology1472-6793BioMed Central London 1472-6793-5-141611148110.1186/1472-6793-5-14Research ArticleEffect of the degree of ischaemic injury and reoxygenation time on the type of myocardial cell death in man: role of caspases Vohra Hunaid A [email protected]ñanes Manuel [email protected] Cardiac Surgery Unit, Department of Cardiovascular Sciences, University of Leicester Leicester, UK2005 19 8 2005 5 14 14 14 4 2005 19 8 2005 Copyright © 2005 Vohra and Galiñanes; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The importance of apoptosis in the injury sustained by the human myocardium during ischaemia and reoxygenation and the underlying mechanisms remain unclear. To quantify apoptosis and necrosis induced by simulated ischaemia/reoxygenation in the human atrial myocardium, free-hand sections of right atrial appendage (n = 8/group) were subjected to 90 minutes simulated ischaemia followed by 2, 8 and 24 hours reoxygenation.
Results
Apoptosis, as assessed by TUNEL, was greater than necrosis after 90 minutes simulated ischaemia and 2 hours reoxygenation (35.32 ± 3.22% vs 13.55 ± 1.3%; p < 0.05) but necrosis was greater than apoptosis by 24 hours reoxygenation (45.20 ± 2.75% vs 4.82 ± 0.79%; p < 0.05). Total caspase activation was similar after 90 minutes simulated ischaemia followed by 2 hours and 24 hours reoxygenation (515270 ± 99570 U vs 542940 ± 95216 U; p = NS). However, caspase-3 like activation was higher at 2 hours than at 24 hours reoxygenation (135900 ± 42200 U vs 54970 ± 19100 U; p < 0.05). Inhibition of caspase-3 by z.DEVD.fmk (70 nM) almost completely abolished apoptosis from 23.26 ± 2.854% to 0.73 ± 0.28 % (p < 0.05), without affecting necrosis.
Conclusion
Cell death by apoptosis and necrosis in the human myocardium subjected to simulated ischaemia/reoxygenation depends on the degree of the ischaemic insult and have a different time-course with apoptosis happening early during reoxygenation and necrosis becoming more important later. Importantly, the apoptosis induced by simulated ischaemia/reoxygenation is mainly mediated by activation of caspase-3 but it does not affect necrosis.
==== Body
Background
Ischaemia/reoxygenation of the heart induces apoptosis and necrosis [1-4]; however, the actual contribution of these two forms of cell death to ischaemia/reoxygenation injury and their time-course have not been established in the human myocardium and remains controversial in experimental animal models. Kajstura et al [3] have reported that apoptosis begins in rat ischaemic myocardium either after a prolonged period of permanent ischaemia or during a much shorter period of ischaemia followed by reperfusion whereas others [1,2] have shown that although apoptosis may be initiated during ischaemia, its detection is increased and may well be accelerated during reperfusion. On the other hand, studies in a dog model of coronary artery ligation have shown that necrosis, quantified histologically, develops rapidly after ischaemia and is directly proportional to the ischaemic time [5], although, there is evidence in the literature suggesting that necrosis may also result secondary to reperfusion injury [4]. Furthermore, it has also been suggested that apoptosis may switch to necrosis below certain critical levels of ATP [6,7].
The mechanisms of ischaemia/reoxygenation injury have been extensively investigated, however its pathophysiology is complex and the involvement and importance of various pathways such as the caspases remains unclear.
The aims of the present studies were: (i) to investigate the degree and the time-course of apoptosis and necrosis sustained during ischaemia and reoxygenation of the human myocardium and (ii) to examine the role of caspase activation. Here we have demonstrated that the type of cell death induced by simulated ischaemia/reoxygenation of the human myocardium depends on the ischaemic insult with apoptosis happening early during reoxygenation and necrosis becoming more important later. In addition, the present studies have shown the apoptosis induced by simulated ischaemia/reoxygenation is mainly mediated by activation of caspace-3 but that it does not affect necrosis.
Results
Study 1: Effect of the intensity of ischaemic injury and the time of reoxygenation
(i) Apoptosis and necrosis
Figure 1 shows a low degree of apoptosis and necrosis in muscles aerobically incubated for 2 hours (3.2 ± 1.3% and 2.8 ± 0.8%, respectively) and further increases of the two forms of cell death with the extension of aerobic incubation to 8 and 24 hours. As expected, the degree of necrosis increased with extension of ischaemia, however apoptosis was greater than necrosis (32.0 ± 3.2% vs 10.7 ± 1.9; p < 0.05) after 90 minutes of ischaemia and two hours of reoxygenation and the reverse was seen after 180 minutes of ischaemia (12.6 ± 1.9% vs 27.1% ± 2.8%; p < 0.05). A similar cell death pattern was observed after 8 hours of reoxygenation although the extent of apoptosis and necrosis was greater than after 2 hours of reoxygenation. After 24 hours of reoxygenation, necrosis also increased with the extension of ischaemia and it was greater than after 2 and 8 hours of reoxygenation, but necrosis was significantly more important than apoptosis. Figures 2 and 3 depict representative images of fluorescent apoptotic and necrotic nuclei.
(ii) MTT reduction
Figure 4 shows that there was a decrease in MTT reduction with increasing periods of ischaemia and that, in contrast with the assessment of apoptosis and necrosis by the TUNEL assay, it was not significantly influenced by increasing the period of reoxygenation.
(iii) CK release
Table 1 demonstrates that the CK release exhibited the lowest values in the aerobic control groups and that there was an increase in CK release with increasing ischaemia. The mean CK release values during the first two hours of reoxygenation were similar in the groups with identical ischaemic time suggesting that ischaemic injury was of the same degree in the three reoxygenation protocols
(iv) Caspase activity
Figure 5A shows that there was no significant difference in the total caspase activation irrespective of the time of ischaemia or reoxygenation amongst the groups and that, unexpectedly, the highest values corresponded to the fresh tissue. This activity decreased after 30 minutes of aerobic equilibration. The physiological meaning of this transient increase in total caspase activity is unclear but was not associated with greater caspase-3 activation (see below) and was not translated into a rise in apoptosis.
Figure 5B demonstrates that caspase-3-like activation was significantly increased after 2 hours of aerobic incubation when compared to the mean values in the fresh muscles and that although activity increased with the extension of the ischaemic time, values were not greater than the ones seen in the aerobic control group. Importantly, by 8 and 24 hours of reoxygenation the levels of caspase-3-like activation had decreased to levels close to the values observed in the fresh muscles irrespective of the periods of ischaemia.
Study 2: The role of caspase activation in cell death
(i) Inhibition of caspase-3 activity
Figure 6 shows that following 90 minutes of simulated ischaemia and 2 hours of reperfusion there was a dose-dependent reduction in caspase-3 activity with increasing concentration of z.DEVD.FMK, and that activity was almost completely abolished with 70 nM concentration of the inhibitor.
(ii) Apoptosis and Necrosis
Figure 7 shows the percentage of apoptosis and necrosis in atrial tissue after SI/R in the absence and presence of the caspase-3 inhibitor z.DEVD.FMK (70 nM). Caspase-3 inhibition significantly reduced apoptosis in the muscles aerobically incubated and resulted in almost complete abolition of apoptosis, from 23.3 ± 2.8% to 0.7 ± 0.3% (p < 0.05), in the muscles subjected to simulated ischaemia/reoxygenation. Interestingly, z.DEVD.FMK did not influence the degree of necrosis.
(iii) MTT Reduction
Table 2 shows that there was no significant change in MTT reduction with the addition of z.DEVD.FMK in the muscles aerobically incubated and those subjected to simulated ischaemia/reoxygenation. Since, as seen above, apoptosis was almost abolished whereas necrosis was unaffected by caspase-3 inhibition, these results suggest that the observed changes in MTT reduction are not a reflection of apoptosis.
(iv) CK release
Table 2 also shows that the addition of z.DEVD.FMK did not affect the CK release of the aerobically incubated muscle and that the increase in CK release induced by 90 minutes of simulated ischaemia and 2 hours reoxygenation was unchanged, this suggesting that, as seen with MTT reduction, changes in CK release are not a reflection of apoptosis.
Discussion
The present studies have demonstrated that, depending on the ischaemic insult, apoptosis may be the predominant form of cell death in the human myocardium during the first 8 hours of reoxygenation, so that apoptosis is more important than necrosis when the ischaemic period is ≤ 90 minutes but the reverse is true after 180 minutes of ischaemia; however, by 24 hours of reperfusion cell death by apoptosis has subsided and necrosis, that also depends on the degree of ischaemia, becomes the leading cause of cell death. The implications of these findings for the understanding of the pathophysiology of ischaemia/reoxygenation injury are discussed below.
Effect of the degree of ischaemic injury and the duration of reoxygenation on the type of cell death
Our study is the first to report on the time-course of cell death by apoptosis and necrosis induced by ischaemia and reoxygenation of the human myocardium. The demonstration that apoptosis follows a bell-shaped profile, increasing with the duration of ischaemia up to 90 minutes and then decreasing by 180 minutes of ischaemia, is novel. Previous experimental studies could not show this response because they used limited time-periods of ischaemia [1,8,9]. The finding that apoptosis increases with the duration of reoxygenation from 2 to 8 hours is supported by observations in the dog heart subjected to 60 minutes of ischaemia and reperfused for 6, 24, 48 and 72 hours [9]. However, whilst our study showed a decline in apoptosis when the myocardium was reoxygenated for 24 hours, the latter study [9] reported a progressive increase in apoptosis over the 72 hours of reperfusion. The reason for the differing results of the two studies is unclear but the use of different species (e.g., man versus dog) and experimental preparations (eg, in vitro versus in vivo) may be, at least in part, responsible.
The present studies have also shown that, as expected, the degree of cell death by necrosis is directly proportional to the severity of ischaemia. But, in addition, they have demonstrated that necrosis gradually increases with the duration of reoxygenation, which is consistent with the observation that necrosis is a dynamic process that continues over a period of at least 24 hours of reperfusion [9]. It should be mentioned, however, that the degree of tissue injury as assessed by the reduction of MTT was not increased with the duration of reoxygenation. This apparent contradiction between the results obtained with the propidium iodide staining and MTT reduction is not clear, but it is possible that while the former reflects the extent of necrosis alone, the latter may represent other forms of cell death. The lack of agreement of these two assays highlights the importance of using more than one index to assess tissue injury.
Role of caspase activation on ischaemia/reoxygenation-induced cell death
The similar mean values of total caspase activity seen after all studied ischaemic and reoxygenation periods may suggest that this pathway may not be an important mechanism of cell death in the human myocardium or, more likely, that only specific caspases may take part in this process, the changes of which may not be of sufficient magnitude to significantly alter the whole pool of caspases. Our finding that caspase-3 activity is increased by the degree of ischaemic insult would support the latter hypo-thesis. The rapid dissipation of caspase-3 activity with the extension of the reoxygenation period suggests that this enzyme may be activated for a limited time period and that, therefore, the time points investigated in our studies do not provide a complete time-course of changes in enzyme activity. A participation of caspases in apoptosis has also been observed in experimental animal studies [10], with elevation of caspase-3 activity during the early reperfusion period [11,12]. It is of interest to note that in the present studies total caspase activity, but not caspase-3 activity, was elevated in fresh tissue, a finding that may be explained by the handling and mechanical injury sustained during the sectioning of the muscle.
The almost complete abolition of apoptosis by the specific caspase-3 inhibitor z.DEVD.FMK seen in our studies demonstrates that the induction of apoptosis by ischaemia/reoxygenation in the human myocardium is caspase-3 dependent. This finding in man is supported by in vitro and in vivo animal experimental studies [13,14] in which caspase-3 inhibition attenuated apoptosis and reduced reperfusion injury, all suggesting that caspase-3 activation is an obligatory step in the signal transduction pathway of apoptosis induced by ischaemia/reoxygenation. However, necrosis was unaffected by caspase-3 inhibition, which agrees with results reported in adult rat ventricular myocytes [12], although it has been observed that in cancer cells caspase inhibitors also retard necrosis in an in vitro model of chemical hypoxia [7].
Although these studies did not address the mechanism of caspase-3 activation, it has been previously shown that caspases can be activated by oxidative stress [15], an important element of ischaemia/reoxygenation injury. Furthermore, the apoptosis elicited by oxidative stress can be reduced by the opening of mitoKATP channels [16] and the apoptosis and caspase-3 activation induced by ischaemia/reoxygenation can be reduced by overexpression of the heat shock proteins HSP-60 and HSP-10 [17]. It is clear that more investigation is required to elucidate the signalling pathway leading to activation of caspase-3 during ischaemia/reoxygenation.
Study limitations
In the present studies the human atrial tissue was used and therefore any extrapolation to the ventricular myocardium should be made with caution. Thus, for example, KATP channels that play a role in ischaemic injury exist in both atrium and ventricle [18], although their distribution differs in the two cardiac chambers [19]. Another potential limitation of these studies is that ischaemia and reperfusion were simulated by artificial means instead of using arterial occlusion and release, however, the avoidance of the confounding effects of collateral flow with our preparation could be advantageous. The model of simulated ischaemia and reoxygenation used in the present studies was characterised in our laboratory [20] and subsequently has been used for the investigation of the mechanism of ischaemic and pharmacological preconditioning in man [21-25].
Conclusion
Here we have shown that cell death by apoptosis and necrosis in the human myocardium subjected to ischaemia and reoxygenation depends on the degree of the ischaemic insult and have a different time-course with apoptosis happening early during reoxygenation and necrosis becoming more important later. They also have shown that caspase-3 activation plays a critical role in apoptosis but that it does not affect necrosis.
Methods
Patients
The right atrial appendage from patients undergoing elective coronary artery bypass graft surgery was retrieved at the time of the right atrial cannulation. For this, local ethical approval and patients' informed consent was obtained. Patients with atrial fibrillation, poor ejection fraction (EF < 30%), and those with diabetes and taking potassium channel openers (nicorandil or diazoxide) were excluded.
Experimental preparation and solutions
The sectioning of the atrial muscle and the preparation of simulated ischaemia/reoxygenation have been previously described [20]. Briefly, the appendage was mounted onto an ice cooled ground glass plate with the epicardial surface face down and then sliced freehand with surgical skin graft blades (Shwann-Morton, UK) to a thickness of between 300 and 500 μm. Muscle sections weighing between 30–50 mg were then transferred to conical flasks (25 ml Erlenmeyer flasks, Duran, Astell Scientific, Sidcup, Kent, UK) containing 10 ml of oxygenated buffered solution. Following this, the flasks were placed in a shaking water bath maintained at 37°C. The oxygenation of the incubation medium was maintained by a continuous flow of 95% O2/5%CO2 gas mixture to obtain a pO2 between 25 and 30 kPa and a pCO2 between 6.0 and 6.5 kPa. These sections were equilibrated for 30 minutes in oxygenated Krebs Henseleit Hepes (KHH) buffer containing (in mM): NaCl (118), KCl (4.8), NaHCO3 (27.2), MgCl2 (1.2), KH2PO4 (1.0), CaCl2 (1.20), glucose.H2O (10), HEPES (20) at a pH of 7.4 and a temperature of 37°C. The buffer was supplemented with 10% foetal calf serum (FCS: Harlanseralabs #S-0001A). Ischaemia was simulated by bubbling the media with 95% N2 and 5% CO2 in the absence of glucose (pH 6.6–6.9).
Experimental protocols
Study 1: Effect of the intensity of ischaemic injury and the time of reoxygenation
After being equilibrated in oxygenated KHH buffer for 30 minutes, the atrial sections were randomly allocated to various protocols: 30, 90 and 180 minutes of ischaemia followed by 2, 8 or 24 hours of reoxygenation (n = 8/group). Some slices were not subjected to ischaemia and were aerobically incubated for 2, 8 or 24 hours to serve as aerobic controls. Creatine kinase (CK) release into the media was measured during the first 2 hours of reoxygenation and during the first 2 hours of aerobic incubation in the controls. The tissues were taken at the end of the protocols for the assessment of tissue viability, cell necrosis and apoptosis, and caspase activity (see below).
Study 2: The role of caspase activation in cell death
Free-hand sections of fresh atrial tissue were allowed to equilibrate in oxygenated KHH buffer for 30 minutes and then subjected to 90 minutes of simulated ischaemia followed by 120 minutes of reoxygenation (n = 4/group) in the absence (controls) and presence of 0.7, 7.0 and 70.0 nM concentrations of the specific caspase-3 inhibitor z.DEVD.FMK. The inhibitor was present in the media throughout the entire experimental protocol. The tissue was taken at the end of protocol and stored at -80°C until assessment of caspase-3 activity.
In an additional study, the atrial muscles (n = 6/group) were subjected to the following experimental protocols: (i) aerobic perfusion for 240 minutes; (ii) aerobic perfusion for 240 minutes with the caspase inhibitor z.DEVD.FMK (70 nM); (iii) 90 minutes simulated ischaemia followed by 120 minutes reoxygenation; and (iv) simulated ischaemia/reoxygenation with 70 nM z.DEVD.FMK (70 nM). The caspase inhibitor was incubated with the muscles for the entire experimental period. As before, the CK release was measured in the incubation media during the 120 minutes of reoxygenation or during the last 120 minutes of aerobic incubation in the controls and the tissue was taken at the end of protocols for the assessment of tissue viability and cell necrosis and apoptosis.
Assessment of tissue injury and viability
CK release into the perfusate during the 2 hours of reoxygenation was measured as an index of tissue injury. The enzyme activity was measured by a linked-enzyme kinetic assay employing a commercial assay kit (DG147-K: Sigma Chemicals, Perth, Australia) and expressed as IU/g wet weight.
Tissue viability was assessed by the mitochondrial reduction of 3-[4,5 dimethylthiazol-2-y1]-2,5 diphenyltetrazolium bromide (MTT) to an insoluble purple formazan dye (M2128-Sigma Chemicals, Perth, Australia) at the end of the reoxygenation period, as previously described [26]. Finally, the absorbance of the blue formazan product was measured on a plate reader (Benchmark, Bio-Rad Laboratories, Hercules, CA, USA) at 550 nm and the results were expressed as mM of formazan/g wet weight.
Assessment of apoptosis and necrosis
The muscles were incubated for 10 minutes on ice with 5 μM propidium iodide (PI) in 0.1 M tri-sodium citrate and 20 mM phosphate buffered saline (PBS) at pH7.4 to identify the necrotic nuclei. Sections were then fixed twice, first for 30 minutes then with 4% paraformaldehyde in 30% sucrose and 20 mM PBS overnight on ice and at pH7.4. Following this, serial sections of 10 μm were cut with a Bright cryomicrotome (model OTF) at -25°C in tissue embedding matrix (Tissue Tek® OCT compound). Mirror sections were labelled at this stage with 20 μM PI to stain the total number of nuclei in each section. The cryopreserved tissue sections were washed with 20 mM PBS at pH 7.4 for 2 minutes, then permeabilised in 0.02 mg/ml proteinase-K for 10 min at 37Cs, and pre-sensitised for 1 minute in a microwave oven at 800 watts in 0.1% Triton X-100 and 0.1 M sodium citrate at pH 6.0. To assess apoptosis, the terminal deoxynucleotidyl transferase (TdT) was used to incorporate fluorescein (FITC) labelled dUTP oligonucleotides to DNA strand breaks at the 3'-OH termini in a template dependent manner (TUNEL technique) using a commercially available kit (Roche: 1684795, Basel, Switzerland). The FITC fluorescence emission (range 600–630 nm) was measured using argon-ion fluorescence excitation at 488 nm and detected by laser confocal epifluorescence microscopy with a ×10 oil immersion objective. The PI labelled nuclei was excited with helium-neon laser light at 543 nm and fluorescence was detected using an emission range of 680–730 nm in order to abolish fluorescence 'bleed-through' from FITC labelled nuclei. Analysis was done using NIH Image software (Scion Corp, Frederick, Maryland, USA) with the Cavalieri-3 macro (G. MacDonald, University of Washington). Fluorescent signals with areas greater than 16 μm2 were counted to ensure that only cardiomyocyte nuclei are taken into account and to avoid the inclusion of artefact.
Quantitation of caspase activity
The muscle sections stored at -80°C until analyses were thawed in 400 μl of cell lysis buffer (in mM: Hepes (100), 10% sucrose, 0.1% Chaps and DTT (10), in the presence of a cocktail of enzyme inhibitors (P2850-Sigma Chemicals, Perth, Australia) at a pH of 7.0 to release the intracellular contents. The sections were diced finely and then homogenised (Ultra-Turrax homogeniser: Janke and Kunkel GmbH, Staufen, Germany) at 13,000 rpm for 1 minute on ice. This was followed by centrifugation (PK121R-ALC International) at 14,000 rpm for 30 minutes. Subsequently, the protein concentration of the soluble supernatant (cell lysate) was measured using a detergent compatible Bio-Rad assay (23225-Pierce, Cheshire, UK). Aliquots of cell lysate were then tested for caspase activity by the addition of the caspase-specific substrate DEVD, that is conjugated to the chromophore (fluorescent reporter molecule) 7-amino-4-trifluoromethyl coumarin (AFC). The cleavage of the peptide DEVD from DEVD.AFC (final concentration 20 μM;Alexis Chemicals, San Diego, CA, USA) releases AFC that when excited by light at 400 nm emits fluorescence at 505 nm. The level of caspase activity in the cell lysate is detected by fluorescence signal obtained with a fluorometer (Fluostar P401, BMG software, Longmont, CO, USA). The amount of caspase-3 like activity was measured by using the effector caspase inhibitor z.DEVD.FMK at a final concentration of 10 μM in the well of the reader plate and by subtracting the fluorescence obtained in the presence of the inhibitor by the total fluorescence measured in the absence of the inhibitor. The inherent fluorescence of the aromatic residues in the protein homogenate was subtracted in the final calculations. The results were expressed as units of activity/g wet weight.
Statistical analyses
Data were expressed as mean ± SEM. ANOVA was used for comparisons of means (Microsoft® Excel analysis tool pak) with the application of a post-hoc Tukey's test. A p value of less than 0.05 was considered statistically significant.
Authors' contributions
HAV participated in the design of the study, carried out all of the studies, collated and analyzed the data and drafted the manuscript. MG participated in the design of the study, analysis and presentation of the data and revisions of the manuscript. Both authors have read and approve the final manuscript.
Acknowledgements
This work was partially supported by a BUPA Surgical Research Fellowship (Hunaid Vohra) and a personal contribution from Professor Manuel Galiñanes. We are grateful to Dr. A. Fowler for his technical advice and to Mrs. N. Harris for secretarial assistance.
Figures and Tables
Figure 1 Percentage of apoptosis (empty columns) and necrosis (solid columns) in right atrial muscles subjected to various periods of simulated ischaemia followed by various periods of reoxygenation. The columns represent the mean of 8 experiments and the bars represent the SEM. *p < 0.05 vs aerobic control. †p < 0.05 vs apoptosis.
Figure 2 Representative images for apoptosis (TUNEL) in the following protocols: aerobic control (A); 30, 90 and 180 minutes of simulated ischaemia followed by 120 minutes of reoxygenation (B, C and D, respectively).
Figure 3 Representative images for necrosis (propidium iodide) in the following protocols: aerobic control (A); 30, 90 and 180 minutes of simulated ischaemia followed by 120 minutes of reoxygenation (B, C and D respectively).
Figure 4 MTT reduction of right atrial muscles subjected to various periods of simulated ischaemia followed by various periods of reoxygenation. The columns represent the mean of 8 experiments and the bars represent the SEM. *p < 0.05 vs aerobic control.
Figure 5 Total caspase activity (A) and caspase-3-like activity (B) in right atrial muscles subjected to various periods of simulated ischaemia followed by various periods of reoxygenation. The columns represent the mean of 8 experiments and the bars represent the SEM. *p < 0.05 vs fresh muscle
Figure 6 Dose-response effect of the caspase-3 inhibitor z.DEVD.FMK on caspase-3 activitiy of atrial muscles subjected to 90 minutes of simulated ischaemia followed by 120 minutes of reoxygenation. Data are expressed as the mean of 4 experiments and the bars represent the SEM. *p < 0.05 vs group without inhibitor.
Figure 7 Effect of caspase-3 inhibition (C3i) with z.DEVD.FMK (70 nM) on apoptosis (empty columns) and necrosis (solid columns) in right atrial muscles subjected to 90 minutes of simulated ischaemia and 2 hours of reoxygenation. The columns represent the mean of 6 experiments and the bars represent the SEM. *p < 0.05 vs corresponding group without caspase-3 inhibitor.
Table 1 CK leakage during the first 2 hours of reoxygenation of right atrial muscles (n = 8/group) subjected to various periods of simulated ischaemia followed by various periods of reoxygenation as compared with the enzyme leakage of muscles for the first 2 hours of aerobic incubation. *p < 0.05 vs the respective aerobic controls.
Reoxygenation group Aerobic control Simulated ischaemia (minutes)
30 90 180
2 hours 2.6 ± 0.2 3.5 ± 0.3* 5.0 ± 0.4* 6.4 ± 0.4*
8 hours 2.8 ± 0.3 4.0 ± 0.5* 5.0 ± 0.4* 6.6 ± 0.8*
24 hours 2.5 ± 0.4 3.7 ± 0.5* 4.6 ± 0.7* 6.1 ± 1.0*
Table 2 CK leakage during the first 2 hours of reoxygenation and MTT reduction at the end of the reoxygenation period of right atrial muscles (n = 6/group) after 90 minutes of simulated ischaemia/reoxygenation (SI/R) in the presence and absence of the caspase-3 inhibitor z.DEVD.FMK (70 nM).
Group MTT P value CK P value
Aerobic Control 290.3 ± 32.3 2.5 ± 0.2
Aerobic Control plus C3i 298.0 ± 30.4 NS 2.7 ± 0.1 NS
SI/R 212.7 ± 23.1 5.4 ± 0.4
SI/R plus C3i 220.3 ± 18.4 NS 5.7 ± 0.5 NS
==== Refs
Fliss H Gattinger D Apoptosis in ischemic and reperfused rat myocardium Circ Res 1996 79 949 956 8888687
Gottlieb RA Burleson KO Kloner RA Babior BM Engler RL Reperfusion injury induces apoptosis in rabbit cardiomyocytes J Clin Invest 1994 94 1621 1628 7929838
Kajstura J Cheng W Reiss K Clark WA Sonnenblinck EH Krajewski S Reed JC Olivetti G Anversa P Apoptotic and necrotic myocyte cell deaths are independent contributing variables of infarct size in rats Lab Invest 1996 74 86 107 8569201
Matsumura K Jeremy RW Schaper J Becker LC Progression of myocardial necrosis during reperfusion of ischemic myocardium Circulation 1998 97 795 804 9498544
Reimer KA Lowe JE Rasmussen MM Jennings RB The wave-front phenomenon of ischemic cell death 1. Myocardial infarct size vs duration of coronary occlusion in dogs Circulation 1977 56 786 794 912839
Allen JW Eldadah BA Huang X Knoblach SM Faden AI Multiple caspases are involved in beta-amyloid-induced neuronal apoptosis J Neurosci Res 2001 65 45 53 11433428 10.1002/jnr.1126
Shimizu S Eguchi Y Kamiike W Waguri S Uchiyama Y Matsuda H Tsujimotot Y Retardation of chemical hypoxia-induced necrotic cell death by Bcl-2 and ICE inhibitors: possible involvement of common mediators in apoptotic and necrotic signal transductions Oncogene 1996 12 2045 2050 8668329
Maulik N Yoshida T Engelman RM Deaton D Flack JE 3rdRousou JA Das DK Ischemic preconditioning attenuates apoptotic cell death associated with ischemia/reperfusion Mol Cell Biochem 1998 186 139 145 9774195 10.1023/A:1006883717174
Zhao ZQ Velez DA Wang NP Hewan-Lowe KO Nakamura M Guyton RA Vinten-Johansen J Progressively developed myocardial apoptotic cell death during late phase of reperfusion Apoptosis 2001 6 279 90 11445670 10.1023/A:1011335525219
Vanden Hoek TL Shao Z Li C Zak R Schumacker PT Becker LB Reperfusion injury on cardiac myocytes after simulated ischemia Am J Physiol 1996 270 H1334 H1341 8967373
Freude B Masters TN Robicsek F Fokin A Kostin S Zimmermann R Ullmann C Lorenz-Meyer S Schaper J Apoptosis is initiated by myocardial ischaemia and executed during reperfusion J Mol Cell Cardiol 2000 32 197 208 10722797 10.1006/jmcc.1999.1066
Kang P Haunstetter A Aoki H Usheva A Izumo S Morphological and molecular characterization of adult cardiomyocyte apoptosis during hypoxia and reoxygenation Circ Res 2000 87 118 125 10903995
Gottlieb R Gruol DL Zhu JY Engler RL Preconditioning rabbit cardiomyocytes: role of pH, vacuolar proton ATPase, and apoptosis J Clin Invest 1996 98 2854 2865 8981934
Yaoita H Ogawa K Maehara K Maruyama Y Attenuation of ischemia/reperfusion injury in rats by a caspase inhibitor Circulation 1998 97 276 281 9462530
Cesseli D Jakoniuk I Barlucchi L Beltrami AP Hintze TH Nadal-Ginard B Kajstura J Leri A Anversa P Oxidative stress-mediated cardiac cell death is a major determinant of ventricular dysfunction and failure in dog dilated cardiomyopathy Circ Res 2001 89 279 286 11485979
Akao M Ohler A O'Rourke B Marban E Mitochondrial ATP-sensitive potassium channels inhibit apoptosis induced by oxidative stress in cardiac cells Circ Res 2001 88 1267 1275 11420303
Lin KM Lin B Lian IY Mestril R Scheffler IE Dillmann WH Combined and individual mitochondrial HSP60 and HSP10 expression in cardiac myocytes protects mitochondrial function and prevents apoptotic cell deaths induced by simulated ischaemia-reoxygenation Circulation 2001 103 1787 1792 11282911
Heidbuchel H Vereecke J Carmeliet E Three different potassium channels in human atrium: contribution to the basal potassium conductance Circ Res 1990 66 1277 1286 2335026
Amos GJ Wettwew E Metsger F Li Q Himmel HM Differences between outward currents of human atrial and subepicardial ventricular myocytes J Physiol 1996 491 31 50 9011620
Zhang JG Ghosh S Ockleford C Galiñanes M Characterization of an in-vitro model for the study of the short and prolonged effects of myocardial ischaemia and reperfusion in man Clin Sci (Lond) 2000 99 439 453 11052925
Ghosh S Standen NB Galiñanes M Preconditioning the human myocardium by simulated ischaemia: studies on the early and delayed protection Cardiovasc Res 2000 45 339 350 10728354 10.1016/S0008-6363(99)00353-3
Loubani M Galiñanes M Long-term administration of nicorandil abolishes ischemic and pharmacologic preconditioning of the human myocardium: role of mitochondrial adenosine triphosphate-dependent potassium channels J Thorac Cardiovasc Surg 2002 124 750 757 12324733 10.1067/mtc.2002.126037
Loubani M Galiñanes M Pharmacological and ischaemic preconditioning of the human myocardium: mitoKATP channels are upstream and P38MAPK is downstream of PKC BMC Physiol 2002 2 10 12123527 10.1186/1472-6793-2-10
Loubani M Hassouna A Galiñanes M Delayed preconditioning of the human myocardium: signal transduction and clinical implications Cardiovasc Res 2004 61 600 609 14962490 10.1016/j.cardiores.2003.10.013
Hassouna A Matata BM Galiñanes M PKC-ε is upstream and PKC-α is downstream of mitoKATP channels in the signal transduction pathway of ischaemic preconditioning of human myocardium Am J Cell Physiol 2004 287 C1418 C1428 10.1152/ajpcell.00144.2004
Vistica DT Tetrazolium-based assays for cellular viability: A critical examination of selected parameters affecting formazan production Cancer Research 1991 51 2515 2526 2021931
|
16111481
|
PMC1208913
|
CC BY
|
2021-01-04 16:03:51
|
no
|
BMC Physiol. 2005 Aug 19; 5:14
|
utf-8
|
BMC Physiol
| 2,005 |
10.1186/1472-6793-5-14
|
oa_comm
|
==== Front
BMC Pulm MedBMC Pulmonary Medicine1471-2466BioMed Central London 1471-2466-5-111613892610.1186/1471-2466-5-11Research ArticleA new paradigm in respiratory hygiene: increasing the cohesivity of airway secretions to improve cough interaction and reduce aerosol dispersion Zayas Gustavo [email protected] John [email protected] Ana [email protected]'Brien Darryl [email protected] Malcolm [email protected] Mucobiology Research Unit, Pulmonary Research Group, University of Alberta, Edmonton, Canada2 Summer students, University of Alberta, Edmonton, Canada2005 2 9 2005 5 11 11 1 2 2005 2 9 2005 Copyright © 2005 Zayas et al; licensee BioMed Central Ltd.2005Zayas et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Infectious respiratory diseases are transmitted to non-infected subjects when an infected person expels pathogenic microorganisms to the surrounding environment when coughing or sneezing. When the airway mucus layer interacts with high-speed airflow, droplets are expelled as aerosol; their concentration and size distribution may each play an important role in disease transmission. Our goal is to reduce the aerosolizability of respiratory secretions while interfering only minimally with normal mucus clearance using agents capable of increasing crosslinking in the mucin glycoprotein network.
Methods
We exposed mucus simulants (MS) to airflow in a simulated cough machine (SCM). The MS ranged from non-viscous, non-elastic substances (water) to MS of varying degrees of viscosity and elasticity. Mucociliary clearance of the MS was assessed on the frog palate, elasticity in the Filancemeter and the aerosol pattern in a "bulls-eye" target. The sample loaded was weighed before and after each cough maneuver. We tested two mucomodulators: sodium tetraborate (XL"B") and calcium chloride (XL "C").
Results
Mucociliary transport was close to normal speed in viscoelastic samples compared to non-elastic, non-viscous or viscous-only samples. Spinnability ranged from 2.5 ± 0.6 to 50.9 ± 6.9 cm, and the amount of MS expelled from the SCM increased from 47 % to 96 % adding 1.5 μL to 150 μL of XL "B". Concurrently, particles were inversely reduced to almost disappear from the aerosolization pattern.
Conclusion
The aerosolizability of MS was modified by increasing its cohesivity, thereby reducing the number of particles expelled from the SCM while interfering minimally with its clearance on the frog palate. An unexpected finding is that MS crosslinking increased "expectoration".
==== Body
Background
Infectious respiratory diseases are a prime cause of morbidity, mortality and health system utilization worldwide, but have a greater impact on the developing and least developed countries. Respiratory infections including tuberculosis (TB) caused 5.5 million deaths during 2001. Mortality rates are only part of the burden, since individuals with infections are often unable to work, becoming trapped in a vicious cycle of ill health and poverty [1].
Viruses and bacteria cause most of the infectious respiratory diseases. These pathogens can travel from country to country in different continents in a matter of hours, as exemplified by the recent outbreak of SARS [1,2]. Children, the elderly, and those with defective immune systems or with underlying chronic diseases are at highest risk of being affected by airborne infectious diseases.
Health care providers are at risk of acquiring transmissible respiratory diseases while providing care or working in contaminated environments. The SARS outbreak has validated these concerns in advance of the feared flu pandemic. SARS has also confirmed that infectious diseases can cause great disruptions in all sectors – economic, educational, recreational, familial – and can bring health care systems to near collapse. Several strategies and devices have been designed to protect various sectors of society whether at peacetime or during conflicts.
Different individual protective barriers, such as surgical masks, face shields and respiratory protection equipment have been used; however, they yield a much lower protection than the theory might indicate [3]. The individual protective equipment must be chosen according to the agent and to the activities performed by the user, and consider the compliance of the user and the adequacy of the fit and the effectiveness of the seal. Wearing well-fitted protective equipment for long periods of time could impair the satisfactory performance of duties of health care providers. Hence adequate and cost-effective strategies are still needed.
Vaccination is the main strategy to control outbreaks of a number of infectious diseases including the flu pandemic, but the WHO states that other advance control measures will definitively be required since vaccines take time to become available [4].
Tuberculosis (TB) caused by Mycobacterium tuberculosis is a curable, transmissible malady known for at least 3,400 years. The WHO and other organizations estimate that two billion people around the world are infected with M. tuberculosis. Eight million new active cases of TB are reported annually, mostly in age-productive adults. Two million people died of TB in 2002, more than in any previous year in history, becoming the leading killer of adults worldwide [5].
M. tuberculosis can develop resistance to multiple, first-line anti-TB drugs (MDR-TB). This type of resistance spirals the costs beyond 100 times per patient, bringing more suffering to countries afflicted with TB [6]. Involuntary quarantine in TB patients with non-adherent attitudes is enforced as a last resource measure [7] to control its spread.
In cystic fibrosis, recurrent and persistent respiratory infection by Pseudomonas aeruginosa is a major cause of morbidity and mortality in CF patients. Jones et al. [8] concluded that cross-infection outbreaks of P. aeruginosa were caused by respiratory aerosol dissemination.
Infectious respiratory diseases, whether viral or bacterial, are transmitted to non-infected subjects when an infected patient expels pathogenic microorganisms to the surrounding environment when coughing, sneezing or even possibly when talking. Duguid [9] determined that the number of particles of 100 μm or less in diameter contained in the aerosol formed during speaking were between 0 – 200 particles, while during coughing there were 0 – 3,500 particles and during sneezing there were 4,500 – 1, 000,000 particles. Fennelly [10], studying the size of the infectious particles generated during cough, estimated that close to 50% of them were fine particles, less than 2 μm range, small enough to reach the alveoli within the lungs.
For the transmission of any respiratory pathogen at least three components are required: a) the transmissor or source (infected person), b) the surrounding environment, and c) the recipient (non-infected person). It is important to additionally consider the concentration of infectious particles determined by the volume of the space and its ventilation, the length of time of exposure, as well as the status of defense mechanisms of the exposed individual.
This prompted a research question: Is it possible to control the transmission of infectious respiratory diseases by modulating the physical characteristics of viscoelasticity, cohesivity and surface tension of the respiratory secretions to minimize the aerosolization that facilitates transmission of airborne diseases – a transmission-blocking approach? We found in our literature search no report in regards to this type of approach or drugs. Before testing the mucomodulators we established a fundamental criterion: Can we modulate the physical characteristics of the respiratory secretions to reduce aerosolization without affecting normal clearance function, or even better by improving it?
Hence, we focused our attention on the mucus layer of the airways. The mucin macromolecule (Figure 1) consists of a protein core surrounded by short oligosaccharide side-chains, held together by different links: O-glycosidic bonds, disulphide bridges, hydrogen bonds and ionic bonds, which are the targets of the existing mucolytic agents [11].
Figure 1 Schematic diagram of the macromolecules of mucus with the different types of crosslinks. Sodium tetraborate would contribute to crosslink formation involving accessible galactose moieties in the oligosaccharide side-chains. Calcium chloride would add crosslinking through divalent ionic interactions.
The mucus along with serous fluid forms the airway surface fluid (ASF) that provides a protective milieu for the airways. The composition and physical characteristics of ASF allow for normal ciliary activity and airway protection [12,13]. When disruption of normal secretory or mucociliary clearance processes occurs, respiratory secretions can accumulate and impair pulmonary function, reduce lung defenses and increase the risk for infection and possibly neoplasia [14].
Ninety-five per cent (95%) of mucus is water and the remaining 5% is made up of proteins, carbohydrates, salts, etc. Therefore, when the airway mucus layer interacts with high-speed airflow during coughing, there is formation of droplets of different sizes that are expelled to the surrounding environment as an aerosol. The concentration of droplets and their size distribution may each play an important role in the transmission of respiratory infectious diseases.
Micron-sized droplets dry quickly, remain airborne for long periods and reach the lower respiratory system when inhaled. These breathable particles might indirectly infect many persons, e.g. via a common ventilation system. In contrast, large droplets are propelled onto the nearest surface – oral, respiratory, or ocular mucosa of a nearby person – or settle readily to the floor. Regarding transmission, larger particles may have an immediate effect – direct but limited – while micron-size particles may have more indirect effects with longer-lasting and widespread consequences.
For this study, our basic goal was to assess a new generation of mucomodulators in a novel approach that has the potential to modulate the physical characteristics of the mucin glycoprotein gel network. This intervention would increases the mucin crosslinking binding sites to reduce the aerosolizability on mucus simulants (surrogate of transmissibility), and/or forming poorly soluble mucin complexes. We aimed to achieve the former while interfering only minimally with normal mucus clearance and hopefully even improving clearability in patients with airway mucus retention.
Our hypothesis is that transmission of airborne diseases can be controlled by implementing a co-adjuvant pharmacological intervention, additional to the indicated standard treatment, aimed at modulating the physical and biochemical characteristics of the respiratory secretions of both the transmissor who carries the infection to minimize expectorated material that carries the pathogens, and in the non-infected potential recipient.
Methods
For this phase we used mucus simulants for testing the effect of the mucomodulators. We also used our ex vivo frog palate exposure model in combination with our simulated cough machine to assess the clearability as well as the aerosol formation capability of the artificial mucus with or without mucomodulators.
Mucus simulants (MS)
The MS ranged from a non-viscous, non-elastic substance (water) to MS of varying degrees of viscosity and elasticity. To prepare viscoelastic mucus simulants, we followed the procedure described by King et al. [15]. Mucus simulants were formulated and developed to be transported within the normal range of velocity in a ciliated epithelium and cleared from a simulated cough machine (SCM) set to have a close to normal human cough. Mucus simulants viscous-only were prepared dissolving different concentrations (0.5 %, 0.75 % and 1 %) of locust bean gum (LBG) in warm Ringer solution, and then adding activated charcoal, bromophenol blue 1% or Coomassie blue to dye the solution for better visualization.
Frog palate preparation
From a bullfrog, Rana catesbiana, the upper portion of the head was removed following the procedures described in previous work [16,17], by cutting with scissors through from the junction of the posterior pharynx and esophagus out to the skin of the back. This procedure was carried out after lowering the body temperature of the frog for 30 – 60 minutes inside a refrigerator to abolish pain sensations. The palate was examined for macroscopic lesions, such as ulcers, petechia or redness as evidence of inflammation; only palates free of inflammatory indicators were used. Any blood remaining in the epithelial surface was carefully washed away, then the excised head was placed palate side facing upwards on a piece of gauze saturated with frog Ringer solution (FRS) in a Petri dish.
The palate was placed inside the frog chamber, a wooden box with a glass top and fitted glass front, and manipulated through glove openings. The humidity inside the box was maintained near 100% using a Pari jet nebulizer, and the temperature was kept between 22° and 24°C by a rheostat-controlled, externally mounted light source. Before carrying out any measurement, the palate was allowed to stabilize inside the box for 15 minutes before testing.
The FRS was prepared by mixing standard Ringer injection with sterile water (2:1). The composition of standard frog Ringer (in mmol/L) is 90 NaCl, 3 KCl, 2 CaCl2, and 15 NaHCO3 (220 mosm/L). The Health Sciences Animal Policy and Welfare Committee, University of Alberta approved the experimental procedures involving animals
Simulated mucus transport velocity (MTV) determination
The palate was placed under a dissecting stereomicroscope provided with a reticulated eyepiece. Mucociliary clearance was determined by observing the movement of particles of charcoal powder gently deposited on a sample of simulated mucus on the palate surface; its clearance was visually monitored and MTV determined. The displacement of 3 – 5 μL of endogenous frog mucus sample was calculated by dividing the distance traveled by the transit time across the 0.3-inch (7.62 mm) segment marked between 0.1 and 0.4 inches in the graduated eyepiece. To obtain control and simulated mucus transport velocities, at least five measurements of the time required for the mucus sample to travel the defined distance were made.
Tissue and mucus sample collection
Samples of tissue and mucus from frog palates exposed to topical application of mucomodulators, were immersed in glutaraldehyde 2.5%, and stored at 4°C for later scanning electron microscope (SEM) studies. The epithelial tissue was carefully dissected and separated from the palate musculature.
Aerosol formation assay
To assess the aerosol pattern formed after coughing, we used a range of fluids of varied physical characteristics from non-viscous, non-elastic substances such as water, to highly viscoelastic substances such as simulated mucus samples of varying degrees of viscosity and elasticity. To better visualize the dispersion pattern of the fluids exposed to high-speed airflow in the SCM, we stained all the fluids.
Spinnability
The capacity of a fluid to form threads when stretched, an expression of elasticity, was assessed in the Filancemeter (SEFAM, France) prior to exposure to airflow in the SCM. A volume of approximately 50 μL of the sample to be studied is introduced in a reservoir that is then sucked in. A low current is applied to the sample from a voltage supply, and when the sample is stretched vertically, the apparatus electronically measures the maximum length of the thread at the moment of the rupture of the thread. For every mucus sample, four measurements of the length of the thread at the moment of rupture were made, and used to compute the spinnability in mm.
Cough aerosolization
The simulated cough machine was used to test the main hypothesis that the modulated mucus will result in less aerosolized (airborne) material following cough. Since cough is the most critical process of mucus aerosolization, the successful reduction of droplet production during cough will likely predict reduced amounts emitted during sneezing and speaking. All fluids were placed at different distances from the "mouth" opening of the simulated airway to examine if distance from the target may be a factor in determining the dispersion pattern. Later, we decided on a fixed distance to place the sample of MS to be exposed to high-speed airflow interaction. The target was a "bullseye-type" (circles within circles). We also placed paper sheets to lateral sides, on top, and below the target sheet to account for any fluid that is dispersed in areas other than on the target sheet.
Cough clearability assay
The modified simulated cough machine system comprises the following elements: A 10-L tank with compressed air, which serves as a pressure reservoir that generates airflow, simulating the lungs during a cough maneuver. The pressure generated prior to a normal adult cough is approximately 8 psi, and we used this pressure value to simulate the airflow pattern of a human cough for each trial.
This modified machine has a model "airway or trachea" consisting of a 28 cm long rigid acrylic tube with a circular 2.5 cm cross-section. It has an 8 cm detachable muzzle end-piece (59.805 g) that enabled us to weigh the sample of MS loaded before and immediately after each cough maneuver and report the mass of MS expelled or retained. A solenoid valve located between the pressure reservoir and the model trachea control the gas released from the tank.
An aliquot of sputum is layered on the bottom of the model trachea and driven forward by the pressurized gas. The distance traveled by sputum samples under standardized cough-simulating airflow is used as a measure of cough clearability. The target was placed at a working distance of 40 cm from the mouth of the SCM. The simulated cough maneuver is repeated four times with each sputum sample.
Airway clearance ("expectoration" of mucus simulant from the "airways"): Measurements of the weight of the MS sample loaded before and after each cough maneuver were carried out in a Mettler AE 163 microbalance.
Scanning electron microscopy
Samples of mucus and tissue were placed in 2.5% glutaraldehyde solution immediately after collection and stored at 4°C until processing. The samples were post-fixed in 1% osmium tetroxide in Millonig's buffer at room temperature for one hour. They were then washed in a series of ethanol (50 – 100%), ten minutes at each step, followed by two additional periods of absolute ethanol (10 minutes each). The samples were further dehydrated by critical point drying at 31°C for 5 – 10 minutes, then mounted on a specimen holder for SEM and dried overnight in vacuum desiccators. In the final stage of preparation for viewing, the samples were sputter coated with gold (Edwards, model S150B Sputter Coater). Samples were viewed using SEM (Hitachi S-2500). Images were scanned directly to a computer and stored as image files for subsequent viewing and analysis.
Mucomodulators
We explored four different approaches that have the potential to accomplish our primary goals of reducing aerosolization of mucus while promoting airway clearance. However, for the purpose of this study, we report only two of them here. Sodium tetraborate, named also as XL"B" or mucomodulator B, and calcium chloride, named also as XL "C" or mucomodulator C. The other approaches, high molecular weight dextran, named also XL "D" or mucomodulator D, and polyarginine, named also XL "A" or mucomodulator A will be reported independently in other publications [18].
Sodium tetraborate (XL "B")
Sodium tetraborate causes reversible crosslink formation between galactose units [19], which are the major neutral sugar components of mucin. Addition of borate to galactose polymers preferentially raises elasticity relative to viscosity. We added three different volumes of 0.1 M borate (1.5 μL, 15 μL and 150 μL) to the MS to produce three levels of crosslinking, and exposed each of the gels to four cough maneuver measurements per trial.
Calcium chloride (XL "C")
Calcium chloride increases the concentration of divalent cations in the mucus. Conceptually, this is the opposite of what occurs in nature during mucin exocytosis, where intracellular mucin granules, which were held tightly through ion crosslinks, give way to much looser interactions as ion exchange occurs during fusion with the apical membrane [20]. A volume of 4 μL of frog Ringer and 4 μL of three different concentrations of calcium chloride (0.001 M, 0.01 M and 0.1 M) were directly applied through a micropipette to a fresh frog palate in order to assess their effect on mucus transport velocity and on the morphology of the tissue and mucus layer.
Statistical analysis
Data are expressed as mean ± standard deviation unless otherwise stated. A paired Student-T test was used for simple comparison. Neuman-Keuls analysis was applied for repeated measurements. The level of significance was set at 5 %.
Results
In this study we report the effect that the mucomodulator XL"B" has on MS on spinnability, clearance by ciliary action and by high-velocity airflow interaction (cough maneuver), as well as some effects that XL "C" has on the physical appearance of frog mucus and on native mucus transport on the frog palate.
Effects of sodium tetraborate
Staining the samples
Bromophenol Blue made the MS solutions very watery and it reduced viscosity of the MS; hence we rejected the use of this dye from further testing. Activated charcoal did not alter the viscoelastic properties of the MS sample and helped to better visualize the gel sample on the frog palate; however charcoal adds mass to the sample that eventually might alter the dispersion pattern. Coomassie Blue did not alter the viscoelastic properties of MS; hence it was used to dye the samples.
Transportability
Samples of MS of non-viscous, non-elastic properties (frog Ringer solution) of approximately 0.2 – 0.5 mL at room temperature were placed on the frog palate; these dispersed on the surface and were not transported by ciliary action.
Viscous-only samples in the low range of concentration (LBG 0.5%) did not maintain their shape and dispersed on the epithelial surface and were not propelled by ciliary action. Viscous-only samples in a higher concentration range (LBG 1.0%) maintained their shape and were transported very slowly by ciliary action, approximately five times slower than control.
Thread formation
Non-viscous, non-elastic samples and the viscous-only samples did not produce any measurable spinnability in the Filancemeter. Working with samples of viscoelastic MS, we attained repeatable lengths of the filament in the low range (1 – 10 mm) and high range (30 – 50 mm) values in the Filancemeter, but found it difficult to consistently obtain medium ranges (11 – 29 mm), even using different combinations of viscosity and elasticity in MS. Elasticity of the crosslinked MS (length of the filament) increased in a direct proportion to the volume (1.5 μL, 15 μL and 150 μL) of 0.1 M added borate (Figure 2).
Figure 2 Effect of the crosslinking agent XL "B" (sodium tetraborate) on spinnability in artificial mucus. * p < 0.05 with respect to 150 μL; ** p < 0.01 with respect to 150 μL.
Paired T-testing showed significant differences in thread formation when comparing the three added volumes. The Newman-Keuls test also indicated a significant difference (p < 0.05) in thread formation of the MS when comparing medium added volume (15 μL) to high volume (150 μL) of sodium tetraborate. Also, the Newman-Keuls test indicated a very significant difference (p < 0.01) in thread formation when comparing low added volume (1.5 μL) to high volume (150 μL) of borate.
The average linear velocity transport rate by ciliary action of crosslinked mucus simulant in the frog palate was approximately 60 percent (46% – 76%) as effective as native mucus transport used as control (100%) when different volumes (1.5 μL – 150 μL) of 0.1 M sodium tetraborate were applied. When different volumes of lower concentration borate (0.01 M) were applied, the transport velocity rate was no more than 30 percent (13% – 30%) as effective as control.
"Cough clearance"
Non-viscous, non-elastic MS samples (water) placed at various distances from the mouth of the artificial airway and exposed to a simulated cough maneuver had a high degree of dispersion onto the bullseye target.
A sample of approximately 0.5 mL of crosslinked MS stained with Coomassie Blue was loaded inside the detachable portion of the artificial airway and then weighed in a microbalance. The cough machine was triggered and the detachable portion with the remaining aliquot of the sample not expelled by the cough maneuver was immediately weighed again. The amount of MS expelled from the artificial airway increased in direct proportion to the volume of 0.1 M sodium tetraborate added (Figure 3). Paired T-test analysis showed there exists a significant difference comparing the volume of MS sample expelled with increasing volume of 0.1 M added borate. The Newman-Keuls test indicated a significant difference (p < 0.01) in the weight of MS expelled from the SCM when comparing low added volume (1.5 μL of XL "B") to high volume (150 μL of XL "B").
Figure 3 Cough clearance ("expectoration") of sodium tetraborate (XL :B") crosslinked mucous gel simulants, i.e. the percentage of initial load clearing the mouth of the artificial trachea during the cough maneuver. ** p < 0.01 with respect to 150 μL.
MS loaded in the simulated airway and exposed to high-speed airflow interaction (simulated cough) broke up into multiple droplets of various sizes that were expelled in the form of a cloud of aerosol that landed on the target. Some portions of the MS expelled did not reach the target but landed on pieces of paper placed below, above and along the lateral sides of the corridor between the mouth and the target. The amount of aerosol striking the target and the amount of fine aerosol decreased progressively with added volume of crosslinking agent (Figure 4).
Figure 4 Aerosolization and dispersion pattern effects for sodium tetraborate crosslinked mucous gel simulants. Aerosol pattern at 40 cm. following a standardized cough maneuver. Both the amount of aerosol striking the target and the amount of fine aerosol decreased progressively with added crosslinking agent.
The volume of XL "B" added to the MS sample was inversely related to the number of droplets formed after a cough maneuver and airflow interaction; the larger the volume added, the fewer droplets were formed. It was also observed that as the added volume of XL "B" (borate) increased, the size of the droplets formed after the airflow interaction increased; hence few, big cohesive droplets were formed, which did not reach the target but did fall in the paper placed immediately below the exit. The proportion of MS sample remaining in the "airway" after a cough maneuver correspondingly decreased with increasing added volume of crosslinking agent (from 53% for 1.5 μL to 40% for 15 μL to just 4% for 150 μL added sodium tetraborate).
Calcium chloride effects
XL "C" (calcium chloride 5 μL) applied topically to our ex vivo frog palate model, causes mucus "clumping" without modifying the normal physical appearance of the ciliated surface when compared with ciliated surface exposed to frog Ringer solution (Figure 5).
Figure 5 Effect of the crosslinking agent calcium chloride (XL "C") on the physical changes of the fresh frog mucus. SE micrograph showing the "clumping" effect of calcium chloride applied topically to the frog palate. The ciliated surface of the palate is seen below the mucus and does not appear different from control views.
Direct application (5 μL) of three concentrations (0.001 M, 0.01 M, 0.1 M) of CaCl2 on the frog palate seems to moderately slow the way that native frog mucus is transported by ciliary action in the epithelial model (Figure 6). The Newman-Keuls analog procedure showed a significant difference (p < 0.05) of MTV of frog mucus samples only when the lowest concentration was applied, compared to the other concentrations and to the Frog Ringer control.
Figure 6 Effect of different concentrations of sodium tetraborate (XL "B'') on MTV of frog mucus. * p < 0.05 with respect to the other concentrations.
Discussion
In this study we have acquired preliminary evidence that the use of mucomodulators yields acceptable mucociliary clearance in an ex vivo epithelial model, increases "expectoration" in a model system, and yet achieves a desired target of a significant reduction in fine aerosol formation. We have also obtained graphic evidence that one of our mucomodulator agents, calcium chloride, which increases the concentration of divalent cations, in fact increases the interaction of the ion crosslinks in the frog mucus, resulting in modification of the physical structure of the mucous layer. This action suggests that the aerosolizability of the mucus or the ability to form tiny droplets may be decreased.
To enhance the validity of the obtained results, we formulated an artificial mucus according to King [19] that displayed physical properties of viscosity, elasticity, reproducible thread formation and transportability in a ciliated epithelium within an acceptable range that most closely simulates airway mucus. The aerosolizability of MS was modified by increasing its cohesivity, thereby reducing the number of particles expelled from the SCM while interfering minimally with its clearance on the frog palate. An unexpected finding is that MS crosslinking increased "expectoration". Rubin & Van der Schans [21] stated, "If an intervention is thought to have a directly measurable effect on sputum properties or on sputum clearability and if these changes cannot be demonstrated in vitro, it is unlikely that they will be observed in clinical trials."
Results presented in this article and obtained thus far from our model system – mucous gel simulant, artificial airway, simulated cough function and ex vivo epithelial model – are limited but striking, and clearly support our mucomodulator approach. We were encouraged to continue with an in vivo mammalian model that will be presented independently (18). We expect that similar results will eventually be observed in clinical trials. Results from this study suggest that the mucomodulators tested appear not to affect the morphology or function of the mucociliary apparatus.
We have designed a new mucomodulator therapy – aiming to change the physical properties of ASF to enhance the clearance of mucus from the respiratory tract, as well as to optimize aspects of lung defense that depend on the mucous layer. Mucomodulators work differently than mucolytics, since the latter were designed to disrupt the structural macromolecules that give respiratory tract mucus its physical characteristics, and from other agents designed to increase mucus flow by stimulating ciliary activity or improving periciliary fluid hydration. As used here, mucomodulators act to increase the cohesivity of airway mucus simulants. Mucomodulator therapy to combat mucus retention may become a major consideration in the treatment of cystic fibrosis and other chronic lung diseases in which mucus hypersecretion and impaired airway clearance produce symptoms [11], and require adequate and novel secretion management. With the mucomodulator approach, we intend to adjust or regulate the mucin macromolecules that give respiratory mucus its physical characteristics in order to minimize the aerosolization that facilitates transmission of airborne diseases. It should be noted that it is not "normal" to cough and aerosolize particles that can be breathed in by others, although it is common. Thus a potential therapy that could reduce this tendency to aerosolize particles when coughing should be considered a form of modulation in the sense of restoring a balance or state of homeostasis.
Cellular debris and other byproducts of infection contribute to make more difficult the clearance of infected mucus in CF patients, requiring antibiotics and mucolytic therapy. These approaches have helped to reduce the mortality rate in CF and other patients; however, there still remains a long way to go to reduce transmission of pathogens in non-infected subjects.
We have invented a novel and non-intuitive approach to improving airway mucus clearance by adding crosslinks in a selective and controlled manner through a specialized new generation of mucomodulators. If further testing bears out our initial findings, it could lead to treatments that would reduce the rate of morbidity and improve the quality of life of individuals who have difficulties with airways secretions management.
Essentially, we have found preliminary evidence that certain mucomodulators that increase cohesive interactions between mucin macromolecules lead to a mucus that is more easily clearable, most likely by airflow-dependent mechanisms, as suggested by results obtained from our SCM The latter is a preliminary answer to a recurring question from lung experts: that using this approach could "gum up" the airway secretions. Such an improvement in mucus quality related to clearance would benefit many, if not most, patients with chronic airway disease, including those suffering from chronic obstructive pulmonary disease, who depend in whole or in part on airflow clearance mechanisms to maintain airway hygiene [13,22]. Further work is needed in clarifying the mechanisms involved in the clearance of the airway secretions.
This novel approach might be of particular importance for those in the ICU or those with spinal cord lesions who require assisted ventilation. At the same time, increasing the cohesiveness in this manner leads to a reduction in fine aerosol formation during expectoration – a clear advantage in helping to control the spread of airborne infections. This new approach may be applicable to a range of infectious disease threats occurring naturally like the feared next flu pandemic or the possible deliberate use of biological agents.
There are some additional critical advantages of this novel technology: Health Canada and the US FDA and other regulatory agencies have already approved the selected compounds for pharmaceutical use in humans. If any of these compounds (or a combination thereof) is judged suitable for clinical trials, established safety data could be utilized to fulfill some of the regulatory requirements for pre-clinical studies.
Some areas of applicability would include transmission prevention of clinically and epidemiologically important pathogens like M. tuberculosis and MDR-TB, in the feared flu pandemic and SARS-CoV, as well as P. aeruginosa in cystic fibrosis. Our mucomodulators would most likely represent containment for transmission of respiratory pathogens especially during outbreaks, allowing strategic sectors of society to be protected and remain functional, like the health system, law enforcement, service providers, as well as the work force.
Some other critical advantages of this novel technology are: Our low cost mucomodulators will not need to undergo constant development and reformulation due to viral or bacteria mutation or resistance. The technology is designed to act on the host rather than the causative organism, which will provide more longevity to it. The compounds under consideration are less perishable than vaccines, amenable to distribution, particularly in warm climates, and simple to manufacture to even allow for rapid response in outbreak situations. In addition, some of the concept testing could involve veterinary trials in airborne animal diseases, an aspect which we are also exploring.
In model studies using vegetable polysaccharides, sodium tetraborate preferentially raises elasticity relative to viscosity, and current knowledge prior to the development of mucomodulators indicated that tetraborate solutions would favour mucociliary clearability at the expense of cough clearability and aerosolizability [12] Now, it appears that cough clearability and aerosolizability are both linked to cohesivity, but in opposite senses, at least within the range of the current study. Thus it appears that sodium tetraborate solutions, by altering mucous gel cohesivity, can decrease aerosolizability (fine aerosol formation during coughing) while increasing cough clearability (bulk clearance of mucus).
In summary, the mucomodulator technology is innovative in that it represents a reversal in current thinking about respiratory disease management. The resulting product will be an unprecedented type of tool in airway hygiene, in secretion management as well as in prevention of droplet-spread illnesses, close person-to-person-contact and airborne transmission – a safe, efficacious, non-vaccine preventative therapy.
Competing interests
The author(s) declare that they have no competing interests.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This project was funded in part by the Canadian Cystic Fibrosis Foundation and the Canadian Institutes of Health Research.
==== Refs
WHO Kindhauser MK Global Defense against the Infectious Disease Threat 2003 World Health Organization, Geneva
WHO Infectious disease report 2002 World Health Organization, Geneva
Somogyi R Vesely AE Azami T Preiss D Fisher J Correia J Fowler RA Dispersal of respiratory droplets with open vs. closed oxygen delivery masks: implications for the transmission of severe acute respiratory syndrome Chest 2004 125 1155 1157 15006983 10.1378/chest.125.3.1155
WHO Influenza pandemic preparedness plan. The role of WHO and guidelines for national or regional planning 1999 World Health Organization, Geneva
WHO Global Plan to Stop TB, Chapter 1, p 27, Stop TB Partnership 2001 WHO, Geneva
WHO AntiTuberculosis Drug Resistance in the World. Report N° 2: Prevalence and Trends The WHO/IUATLD Global Project on Anti-tuberculosis Drug Resistance Surveillance 2000
Centers for Disease Control & Prevention Improving Patient Adherence to Tuberculosis treatment (1994) 2002 National Center for HIV, STD and TB Prevention, Division of Tuberculosis Elimination
Jones AM Govan JR Doherty CJ Dodd ME Isalska BJ Stanbridge TN Webb AK Identification of airborne dissemination of epidemic multiresistant strains of Pseudomonas aeruginosa at a CF centre during a cross infection outbreak Thorax 2003 58 525 527 12775867 10.1136/thorax.58.6.525
Duguid J The numbers and sites of origin of the droplets expelled during respiratory activities Edinburgh Med J 1945 52 385
Fennelly KP Martyny JW Isolation of viable airborne Mycobacterium tuberculosis: a new method to study transmission Am J Respir Crit Care Med 1998 157 A706
King M Rubin BK Pharmacological approaches to discovery and development of new mucolytic agents Advanced Drug Delivery Reviews 2002 54 1475 1490 12458156 10.1016/S0169-409X(02)00156-4
King M Rubin BK Takishima T, Shimura S Rheology of airway mucus: relationship with clearance function Chapter 7 of: Airway Secretion: Physiological Bases for the Control of Mucus Hypersecretion 1994 New York: Marcel Dekker 283 314
Clarke SW King M Gibson CJ Respiratory defences: Physical defences Chapter 31 of: Respiratory Medicine 2002 3 London: Saunders 181 193
Zayas JG Rubin BK York E Lien D King M Bronchial mucus properties in lung cancer: relationship with site of lesion Can Respir J 1999 6 246 252 10393286
King M Brock G Lundell C Clearance of mucus by simulated cough J Appl Physiol 1985 58 1776 1782 4008399
King M Festa E Baum GL Evolution of the frog palate model for mucociliary clearance 1970 – 1997 Cilia, Mucus, and Mucociliary Interactions 1998 New York: Marcel Dekker 191 201
Rubin B Ramirez O King M Mucus-depleted frog palate as a model for the study of mucociliary clearance J Appl Physiol 1990 69 424 429 2228850
Zayas G Valle JC Alonso M Alfaro H Vega D Bonilla G Reyes M King M A new paradigm in respiratory hygiene: modulators of airway secretions to improve clearance Am J Respir Crit Care Med 2005 171 [2005 Am Thoracic Soc Conf (abstract)]
King M Role of mucus viscoelasticity in cough clearance Biorheology 1987 24 589 597 3502760
Verdugo P Takashima T, Shimura S Molecular biophysics of mucin secretion Chapter 3 of: Airway Secretion: Physiological Bases for the Control of Mucous Hypersecretion 1994 New York: Marcel Dekker 101 121
Rubin BK van der Schans CP Rubin BK, van der Schans CP Outcomes for trials of mucoactive therapy Therapy for Mucus-Clearance Disorders 2004 New York: Marcel Dekker 87 103
Puchelle E Zahm JM Duvivier C Spinnability of bronchial mucus: relationship with viscoelasticity and mucus transport properties Biorheology 1983 20 239 249 6871438
|
16138926
|
PMC1208914
|
CC BY
|
2021-01-04 16:30:12
|
no
|
BMC Pulm Med. 2005 Sep 2; 5:11
|
utf-8
|
BMC Pulm Med
| 2,005 |
10.1186/1471-2466-5-11
|
oa_comm
|
==== Front
BMC Pulm MedBMC Pulmonary Medicine1471-2466BioMed Central London 1471-2466-5-91604865210.1186/1471-2466-5-9Research ArticleHeart rate variability in non-apneic snorers and controls before and after continuous positive airway pressure Gates Gregory J [email protected] Susan E [email protected] Jason H [email protected] Departments of Internal Medicine and Physiology, Wayne State University School of Medicine, Detroit, MI, USA2 Research and Development, John D. Dingell Veterans Administration Medical Center, Detroit, MI, USA3 Department of Biobehavioral Sciences, Teachers College, Columbia University, New York, NY, USA2005 27 7 2005 5 9 9 1 4 2005 27 7 2005 Copyright © 2005 Gates et al; licensee BioMed Central Ltd.2005Gates et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
We hypothesized that sympathetic nervous system activity (SNSA) is increased and parasympathetic nervous system activity (PNSA) is decreased during non-rapid eye movement (NREM) sleep in non-apneic, otherwise healthy, snoring individuals compared to control. Moreover, we hypothesized that these alterations in snoring individuals would be more evident during non-snoring than snoring when compared to control.
Methods
To test these hypotheses, heart rate variability was used to measure PNSA and SNSA in 11 normotensive non-apneic snorers and 12 control subjects before and 7-days after adapting to nasal continuous positive airway pressure (nCPAP).
Results
Our results showed that SNSA was increased and PNSA was decreased in non-apneic snorers during NREM compared to control. However, these changes were only evident during the study in which snoring was eliminated with nCPAP. Conversely, during periods of snoring SNSA and PNSA were similar to measures obtained from the control group. Additionally, within the control group, SNSA and PNSA did not vary before and after nCPAP application.
Conclusion
Our findings suggest that long-lasting alterations in autonomic function may exist in snoring subjects that are otherwise healthy. Moreover, we speculate that because of competing inputs (i.e. inhibitory versus excitatory inputs) to the autonomic nervous system during snoring, the full impact of snoring on autonomic function is most evident during non-snoring periods.
==== Body
Background
Epidemiological findings have suggested that snoring is an independent risk factor for the development of daytime hypertension [1-3]. Moreover, studies completed in normotensive individuals suffering from obstructive sleep apnea have shown that increases in sympathetic nervous system activity (SNSA) precede the development of hypertension [4]. Given these findings, we hypothesized that nocturnal increases in SNSA and decreases in parasympathetic nervous system activity (PNSA) may exist in non-apneic normotensive snoring individuals. Additionally, we postulated that the impact of snoring on autonomic nervous system activity might be most evident during periods of non-snoring (i.e. either during wakefulness or during periods of non-snoring during sleep). This latter postulation was based on findings from humans during wakefulness [5-7] or sleep [8,9], which showed that breathing frequency [5-7,9], pattern (i.e. inspiratory and expiratory time) [5-7,9] and tidal volume [5,7,8] are altered in response to either snoring [9], the application of high frequency oscillations (i.e. simulated snoring) [7,8] or breathing against an increased airway resistance [5,6]. Moreover, these alterations in breathing are accompanied by enhanced PNSA [6] or decreased SNSA [10]. Thus, we postulated that snoring elicits a two-fold response from the autonomic nervous system. An acute response that occurs concomitantly with snoring, in which increases in SNSA and decreases in PNSA may not be clearly evident, and a longer lasting potentially permanent response in which increases in SNSA and decreases in PNSA are unmasked in the absence of snoring. To test these hypotheses, we used the technique of heart rate variability to measure PNSA and SNSA in non-apneic snorers and controls during non-rapid eye movement (NREM) sleep before and after the application of nasal continuous positive airway pressure (nCPAP).
Methods
Eleven self-reported snoring subjects with no known medical conditions and 12 non-snoring control subjects were recruited from the community. The snoring group was comprised of 9 males (1 Asian, 1 African-American, 2 Hispanic and 5 Caucasian) and 2 Caucasian females (Table 1). The control group was comprised of 10 males (1 Asian, 1 African-American, 2 Hispanic and 6 Caucasian) and 2 Caucasian females (Table 1). All subjects gave their informed consent to participate in the study, which was approved by the Institutional Review Boards of Teachers College, Columbia University, Wayne State University and John D. Dingell VA Medical Center.
Table 1 Anthropometric and blood pressure measures
Control Snorer
Subject Sex Age BMI MAP Awake MAP NREM Subject Sex Age BMI MAP Awake MAP NREM
1 M 28 26.9 95 77 1 M 25 28.1 84 84
2 M 31 26.7 89 88 2 M 35 23 88 88
3 M 25 27.3 83 81 3 M 27 28 70 65
4 F 43 22.5 75 69 4 F 43 25.6 72 62
5 M 33 28 87 83 5 M 36 30 85 82
6 M 27 31 75 74 6 M 34 30 78 75
7 M 27 28.4 85 74 7 M 36 28 88 80
8 M 29 24.5 89 86 8 M 26 23 95 75
9 F 25 20 85 71 9 F 27 24 79 71
10 M 32 27 80 83 10 M 38 26 91 89
11 M 28 23.2 87 80 11 M 28 24.1 88 82
12 M 26 22.9 81 72
Mean 29.5 25.7 84.3 78.2 Mean 32.3 26.3 83.5 77.5
S.E. 1.4 0.9 1.7 6.2 S.E. 1.8 0.8 2.4 2.7
The snoring and control subjects visited the sleep laboratory on three occasions. Twenty-four hours prior to each occasion the subjects were advised to avoid alcohol and caffeine. During the first visit to the laboratory, which is referred to from hereon as the preliminary study, subjects received a physical examination, which included three separate measures of blood pressure using a standard mercury sphygmomanometer that were separated by 15 minutes. Additionally, subjects completed a general health questionnaire to confirm the absence of pre-existing medical conditions. Subsequently, subjects completed a sleep study in order to familiarize themselves with the laboratory environment and to confirm that the subjects were non-apneic snoring or non-snoring individuals. Moreover, to confirm that subjects were normotensive, beat-to-beat measures of blood pressure were obtained for 1 hour during wakefulness and throughout NREM sleep (see Nocturnal polysomnography for further details). The blood pressure measures were in addition to the values obtained during wakefulness using the sphygmomanometer. If subjects suffered from another sleep disorder (i.e. obstructive sleep apnea, insomnia or upper airway resistance syndrome) they were excluded from the study. Individuals with upper airway resistance syndrome were excluded because we were interested in investigating the impact of snoring on heart rate variability independent of excessive cortical arousal from sleep.
The second study was completed in order to measure the autonomic variables outlined below during NREM sleep. This second study will be referred to as Trial 1 from hereon. Subsequent to the completion of Trial 1, the snoring and control subjects adapted to 5 cmH2O of nasal continuous positive airway pressure (nCPAP) for 7 days at home. The selection of the nCPAP pressure was based on results from our pilot data, which showed that this level of pressure effectively eliminated snoring in non-apneic individuals. The purpose of the adaptation period was to ensure that the subjects were able to tolerate nCPAP for a minimum of 4 hours during completion of a third sleep study (Trial 2). Nasal continuous positive airway pressure was employed initially for 1 hour during the first night at home. Thereafter, the duration of treatment was increased by an additional hour each night until 4 hours of nCPAP was tolerated. During the adaptation period subjects received a phone call on days 3 and 6 to ensure that the protocol was being followed. During all sleep studies subjects were required to sleep in the supine position. The subjects were monitored via an infrared camera to ensure that this position was maintained throughout the sleep period.
Nocturnal polysomnography
The sleep monitoring montage included an electroencephalogram (C3/A2, C4/A1, O1/A2, O2/A1), electrooculogram, submental and tibialis anterior electromyogram and an electrocardiogram. Abdominal movements were monitored using a piezoelectric band (Pro-tech, Woodinville, WA) and nasal pressure was measured using a pressure transducer/airflow sensor (Pro-tech, Woodinville, WA). Thus, breathing frequency was monitored breath-by-breath. Oxygen saturation was measured using a pulse oximeter (Biox 3700, Ohmeda Corp., Boulder, CO). Snoring was measured using a microphone that was mounted on the wall located adjacent to the subjects head. During the preliminary study, blood pressure was monitored continuously and non-invasively from the third finger of the left hand using a digital infrared photoplethysmograph (Finapres 2300, Ohmeda Corp., Madison, WI). The accuracy of the blood pressure monitor was verified during pre-sleep wakefulness and nocturnal awakenings by comparing its values to measurements made with a standard mercury sphygmomanometer. To ensure that the monitoring site of the Finapres was adequately perfused with blood throughout the evening, the operation of the Finapres was discontinued consistently during rapid eye movement sleep (REM) and at times during NREM sleep if necessary. The total number of segments analyzed for each subject represented on average 2.2 hours of data obtained from stage II and SWS recorded over the entire sleep period.
During sleep all physiological variables were analogue to digitally converted at a sampling frequency of 200 Hz/channel and input into a microcomputer using a commercially available software package (Gamma, Version 4.0, Astro-Med Inc., West Warwick, RI).
Data analysis (sleep variables)
Sleep was staged in 30-s epochs according to standard criteria [11,12]. For each subject the total sleep period time as well as the percent of total sleep time spent in a given sleep stage was calculated. The total number of arousals, apneas, hypopneas, snores, and the mean, minimal and maximal oxygen saturation measured was calculated for the total sleep time. An apnea was defined as the absence of inspiratory airflow for a minimum of 10 s. The apnea index was defined as the total number of apneas per hour of sleep. A hypopnea was defined as greater than a 50 % reduction in the flow signal lasting more than 10 s, accompanied by a 2 % decrease in oxygen saturation. We chose to employ a 2 % oxygen desaturation criteria (rather than 3 or 4%) in order to ensure that snoring associated with small changes in oxygen saturation were identified. Based on the use of this criterion we reasoned that if the number of breathing events was similar between snoring and control subjects and less than 5, we could be reasonably confident that the subjects recruited for this study were non-apneic. The hypopnea index was defined as the total number of hypopneas per hour of sleep time. A breath characterized by respiratory noises that registered as an obvious deflection from the baseline of the snoring channel was counted as a snore. In addition, the respiratory noises were subjectively determined to be snores by a polysomnographic technologist monitoring an audio-visual system. We are confident that the sounds recorded were associated with snoring since normal and heavy breathing during wakefulness did not register on the sound system while simulated snoring during wakefulness was detected.
After staging the sleep studies completed during trial 1 and 2, we randomly selected a 15-minute snoring segment from NREM sleep (stage 2 or SWS) recorded between midnight-2 am, 2–4 am and 4–6 am. Thus, 3–15 segments were selected from sleep studies completed during Trials 1 and 2. The segments selected were devoid of apneas, hypopneas and arousals. The snoring segments (Trial 1) were identified as such if 67–100 % of the breaths in a given segment were associated with snoring. Three segments were selected from each study to ensure that our findings reflected the impact of snoring on heart rate variability throughout the night. However, we did not report the results obtained from the snoring segments selected from the beginning, middle and end of the night separately. Although the data was analyzed originally in this fashion, we found that our findings were not time dependent, and thus the data from the three snoring segments were combined.
In addition to the snoring segments, we also analyzed one 15-minute non- snoring segment recorded during Trial 1. In most subjects, only one 15-minute non-snoring segment in deep stage II or SWS was available for analysis. We chose to analyze a non-snoring segment to compare heart rate variability measures in snoring subjects during non-snoring, which was independent of nCPAP application, and snoring.
Data analysis (respiratory, cardiovascular and autonomic variables)
The number of snores and breaths was calculated for each segment. Subsequently, the values calculated were divided by the total segment time and reported as snoring frequency (Sf) and breathing frequency (Bf) (snores/min and breaths/min, respectively). To obtain a measure of SNSA and PNSA the R waves of the electrocardiogram were identified using a threshold detection program. The time intervals between the detected R waves (interbeat interval – IBI) were plotted and the mean IBI was calculated for each segment. Prior to frequency analysis, each 15-minute segment was divided into three 5-minute segments and the interbeat intervals were re-sampled at 20 Hz, which provided 6000 equidistant data points per 5 minute segment [13]. For each 5-min segment, an auto power spectrum was created via a Fast Fourier transform of the interpolated IBI data. The resulting spectrum was integrated and areas associated with discrete frequency bands were determined. Power spectra within the 0.04–0.15 Hz range are considered low frequency components (LF), whereas frequencies of 0.15–0.4 Hz are defined as high frequency components (HF). It is assumed that the power content of the HF component represents PNSA whereas the power content in the LF domain (0.04 – 0.15 Hz) represents SNSA and PNSA at the sinus node [13,14]. The ratio of LF/HF is assumed to be a measure of SNSA.
The absolute values of LF and HF were ln transformed so that the data was normally distributed. Additionally, to minimize the effect that changes in total power have on the absolute values of the low and high frequency power, the LF and HF values were normalized [13]. The presentation of LF and HF values in normalized units represents the relative value of each power component in proportion to the total power minus the very low frequency component. Once the very low frequency component is subtracted, the remaining power equals LF+HF. Thus,
normalized LF = LF ÷ (LF + HF)
normalized HF = HF ÷ (LF + HF)
The normalized units assume a balanced behavior of the sympathetic (normalized LF) and parasympathetic nervous system (normalized HF values). Statistical analysis was completed on both the ln transformed and normalized values (see Statistical analysis below). However, for the sake of clarity and based on the guidelines that LF and HF values should be corrected for total power, the normalized data was used as our point of reference in the discussion (see critique of the methods for further discussion of this point).
Statistical analysis
Unpaired t-tests were used to compare the anthropometric data between groups. A two-way analysis of variance in conjunction with Student-Newman-Keuls post hoc test was used to determine if differences in mean arterial blood pressure existed between groups during wakefulness and NREM sleep during completion of the preliminary study. The two factors in the design were "subject population" (i.e. snorers versus controls) and "arousal level" (i.e. wakefulness vs. sleep). A similar analysis was employed to determine if differences in ln LF, ln HF, LF/HF, normalized LF and normalized HF existed between groups. The two factors in the design were "subject population" (i.e. snorers versus controls) and "treatment" (i.e. before versus after nCPAP treatment). This analysis was also used to determine if differences in sleep architecture existed between and within groups before and after nCPAP treatment. A paired t-test was used to determine whether ln LF, ln HF, LF/HF, normalized LF and normalized HF during non-snoring in trial 1 was different from measures obtained during snoring in trial 1. All values in the tables are presented as means ± SE and the level of significance chosen was p ≤ 0.05.
Results
Age, body mass index and blood pressure measured during wakefulness and sleep throughout the preliminary sleep study, were not significantly different between the snoring and control groups (Table 1). Additionally, apnea index, hypopnea index, arousal index and mean nocturnal oxygen saturation were not different between the snoring and control groups during Trial 1 (Table 2). Similarly, sleep efficiency and the percentage of time spent in a given stage of NREM sleep did not vary significantly between groups during completion of Trial 1 or Trial 2 (Table 2). Moreover, the sleep architecture during Trial 2 did not differ significantly from that reported during Trial 1 within both groups (Table 2).
Table 2 Sleep Measures
Control(n = 12) Snorer (n = 11)
NREM sleep
(without CPAP)
(Trial 1) NREM sleep
(with CPAP)
(Trial 2) NREM sleep
(without CPAP)
(Trial 1) NREM sleep
(with CPAP)
(Trial 2)
Apnea index (apnea/hr) 0.38 ± 0.15 0.67 ± 0.34 1.19 ± 0.56 0.16 ± 0.07
Hypopnea index (hypopnea/hr) 1.80 ± 0.71 0.89 ± 0.29 2.14 ± 0.62 0.34 ± 0.19
Arousal index (arousals/hr) 12.87 ± 1.39 12.87 ± 1.02 14.11 ± 2.20 10.74 ± 1.83
Nocturnal Oxygen Saturation (%) 96.58 ± 0.23 96.7 ± 0.30 96.10 ± 0.39 96.40 ± 0.42
Sleep efficiency (%) 88.76 ± 3.62 88.09 ± 2.02 86.32 ± 2.18 83.73 ± 1.99
% of time in Stage I 11.29 ± 1.83 7.83 ± 0.90 7.22 ± 1.04 8.88 ± 1.57
% of time in Stage II 46.04 ± 1.42 45.53 ± 1.91 48.07 ± 2.58 48.48 ± 2.57
% of time in Slow Wave Sleep 22.45 ± 3.10 20.65 ± 1.42 21.95 ± 1.41 21.27 ± 2.36
Snoring episodes measured from the non-apneic snorers were characterized by an average snoring frequency of 11.60 ± 0.43 snores per minute. In contrast, no snoring was detected from the control group. The application of nCPAP effectively eliminated snoring in all subjects.
Breathing frequency and heart rate were not significantly different between or within groups for all conditions during Trial 1 and 2 (Table 3). The two-way analysis of variance revealed that a significant "subject population" (i.e. snorers versus controls) × "treatment" (i.e. before versus after nCPAP treatment) interaction existed for both ln HF and normalized HF. Post-hoc analysis revealed that ln HF and normalized HF recorded during snoring periods were not significantly different from control during trial 1(i.e. before nCPAP treatment) (ln HF - p = 0.47; normalized HF - p = 0.34, respectively) (Table 3). In contrast, normalized HF recorded during nCPAP treatment (i.e. trial 2) in the snoring population was lower than normalized HF recorded from the control population during nCPAP treatment (p = 0.01) (Table 3). Moreover, ln HF and normalized HF recorded during nCPAP treatment (i.e. trial 2) in the snoring population was significantly less compared to similar measures obtained from snoring periods during trial 1 (ln HF - p < 0.0001; normalized HF - p = 0.001) (Table 3).
Table 3 Autonomic measures obtained from controls and non-apneic snorers during wakefulness and non-rapid eye movement sleep
Control (n = 12) Snorer (n = 11)
NREM sleep
(without CPAP)
(Trial 1) NREM sleep
(with CPAP)
(Trial 2) NREM sleep
(without CPAP)
(snoring)
(Trial 1) NREM sleep
(without CPAP)
(non-snoring)
(Trial 1) NREM sleep
(with CPAP)
(Trial 2)
Heart rate (beats/min) 55.25 ± 1.53 54.58 ± 1.90 57.27 ± 1.45 55.18 ± 1.77 56.27 ± 1.74
Breathing Frequency (breaths/min) 14.89 ± 0.53 14.35 ± 0.46 13.85 ± 0.69 13.91 ± 0.53 14.23 ± 0.62
ln LF 7.24 ± 0.16 7.11 ± 0.13 7.68 ± 0.16 7.99 ± 0.27 7.41 ± 0.19
ln HF 7.99 ± 0.21 7.72 ± 0.19 8.19 ± 0.13 7.83 ± 0.18§ 7.40 ± 0.21§
LF/HF 0.57 ± 0.12 0.70 ± 0.14 0.69 ± 0.08 1.43 ± 0.29§ 1.20 ± 0.18*
HF (nu) 67.51 ± 3.60 64.44 ± 3.95 62.46 ± 2.60 46.71 ± 5.38§ 50.32 ± 4.18*
LF (nu) 34.27 ± 4.37 39.24 ± 5.01 36.61 ± 2.58 52.0 ± 5.26§ 48.79 ± 4.09*
* – significantly different from snorer – without CPAP – snoring; control – with CPAP
§ – significantly different from snorer – without CPAP – snoring
A significant "subject population" (i.e. snorers versus controls) × "treatment" (i.e. before versus after nCPAP treatment) interaction also existed for normalized LF and LF/HF. Post-hoc analysis showed that normalized LF and LF/HF recorded during snoring was not significantly different from control during trial 1 (normalized LF - p = 0.36; LF/HF - p = 0.54, respectively) (Table 3). In contrast, normalized LF and LF/HF recorded during nCPAP treatment (i.e. trial 2) in the snoring population was greater than measures recorded from the control population during treatment (normalized LF - p = 0.01; LF/HF - p = 0.01), and from similar measures recorded from snoring periods during trial 1 (normalized LF - p < 0.001; LF/HF - p < 0.001) (Table 3).
Within the snoring group during trial 1, ln HF (p = 0.03) and normalized HF (p = 0.01) were less and LF/HF (p = 0.02) and normalized LF (p = 0.01) were greater during non-snoring as compared to snoring (Table 3).
Discussion
Summary of findings
Based on the normalized HF and LF values our major findings are that PNSA and SNSA during snoring (i.e. Trial 1) was not significantly different from measures acquired from non-snoring control subjects. However, PNSA was decreased and SNSA was increased in the snoring population compared to control when snoring was abolished with nCPAP (i.e. Trial 2). Similarly, PNSA was less and SNSA was greater when snoring was abolished (i.e. non-snoring periods during trial 1 or during nCPAP application in trial 2) compared to when it was present (trial 1) within the snoring population. Finally, within the control group PNSA and SNSA did not vary before and after treatment with nCPAP.
Critique of the methods
Our subjects were exposed to nCPAP for a 7-day adaptation period prior to completing Trial 2. This adaptation period occurred after a baseline study was completed. Consequently, the baseline and treatment study were not randomized. Nevertheless, if the order of the trials impacted on our measures we would expect that the response to nCPAP application would have been similar between groups since both groups were comparable (e.g. healthy, no complaint of daytime sleepiness) in all aspects with the exception of snoring. This was not the case.
It is also unlikely that any differences observed between groups during Trial 2 were a consequence of discomfort associated with being unfamiliar with nCPAP treatment. Both the snoring and control subjects adapted to the treatment for a similar duration of time prior to participating in Trial 2, and no significant differences in heart rate variability occurred within the control group when comparing measures from Trial 1 and 2. Moreover, it is also improbable that the differences observed were a consequence of discrepancies in nCPAP compliance because, based on subject self-report, this measure was similar between groups. Although it is possible that some subjects did not truthfully report their compliance, we believe it is unlikely that healthy non-snorers and non-apneic snorers would have tolerated nCPAP for at least 4 hours during Trial 2 if they did not adhere to the protocol during the adaptation period. The ability of the subject's to tolerate nCPAP is supported by the measures of sleep architecture, which were similar between trials 1 and 2. Lastly, it is doubtful that the impact of nCPAP on cardiovascular function [15], and consequently autonomic function, was different between groups since the applied pressure was identical (5 cmH2O in all cases) for both groups.
Heart rate variability has been used in numerous studies to obtain non-invasive measures of SNSA (LF/HF and normalized LF) and PNSA (HF and normalized HF). The changes in the HF domain that we observed in our snoring subjects were likely indicative of changes in PNSA since this assumption has been well established by a number of physiological investigations [13]. In contrast, questions remain regarding the validity of LF/HF and normalized LF as a non-invasive indicator of SNSA [14]. One issue is that the LF/HF ratio assumes that HF in the denominator (i.e. PNSA) cancels out the influence of PNSA in the low frequency domain of the power spectrum, leaving the LF/HF ratio as a measure of SNSA. Although we assume that LF/HF and normalized LF are indicative of changes in SNSA we acknowledge that the support for this assumption is equivocal. Nonetheless, we have included this analysis along with the HF measures for those who accept that LF/HF or normalized LF reflects SNSA.
Investigators have shown that absolute values of LF and HF are impacted by changes in total power [13]. Consequently, it has been suggested that absolute LF and HF values should be normalized based on changes in total power [13]. Based on this guideline, and for the sake of clarity, we have used normalized LF and HF as our reference point in the subsequent discussion of the results. Nevertheless, independent of whether one accepts that normalization of our data is necessary, results obtained from the normalized LF (i.e. SNSA) and HF (i.e. PNSA) data are similar to the ratio of absolute LF/HF (i.e. SNSA) and ln HF (i.e. PNSA).
Parasympathetic and sympathetic nervous system activity during snoring and non-snoring
Parasympathetic nervous system activity and SNSA during snoring (i.e. trial 1) were not significantly different from measures obtained from control subjects. This finding on its own implies that snoring does not influence autonomic function. However, our results showed that PNSA was less and SNSA was greater in the snoring group compared to control when snoring was abolished during trial 2. Similarly, PNSA was less and SNSA was greater in the snoring group when snoring was abolished compared to when snoring was present.
These latter two findings have three primary implications. The first implication is that snoring may induce an acute increase in PNSA and a decrease in SNSA, which may mask the more deleterious effect that snoring has on autonomic function. This would account for the similarities in autonomic function observed between groups during trial 1 despite the differences that were present during trial 2. The second implication is that the deleterious long-lasting impact that snoring may have on autonomic function (see Physiological Mechanisms for additional discussion on this issue) is more evident during periods of non-snoring as indicated by the differences observed between the groups during trial 2. The third implication is that short-term nCPAP treatment did not eliminate this long-lasting impact. Arguably, SNSA might have decreased and PNSA increased because of nCPAP treatment. However, the lack of response to treatment is not surprising since the subjects did not receive treatment for a prolonged period with respect to number of days and nighttime hours of exposure.
Physiological mechanisms
We can only speculate on the mechanism responsible for the potential decrease in SNSA and increase in PNSA that occurs during snoring. Snoring is often associated with increased airway resistance and the development of more negative intrathoracic pressures [9]. Additionally, snoring, or the application of high frequency, low amplitude pressure oscillations (which simulates snoring), has been associated with i) decreases in breathing frequency [5,6,9] ii) increases in inspiratory time or inspiratory time/total time ratio [5,6,9] and iii) increases in tidal volume [5,7,8,16]. Moreover, breathing frequency, breathing pattern or tidal volume are known to impact on measures of heart rate variability [13,14]. Thus, it is possible that the increase in HF that we observed during snoring, which mirrors the amplitude of respiratory sinus arrhythmia, was caused by changes in one or more of these variables. Since snoring did not alter breathing frequency in our study, we speculate that alterations in pattern and/or tidal volume may have contributed to the increase in measures of HF. More specifically, prolongation of inspiratory time coupled with an increase in tidal volume may have led to increased activation of pulmonary vagal afferents, which have a profound influence on respiratory sinus arrhythmia, and hence measures of HF. Additionally, activation of pulmonary vagal afferents are known to inhibit SNSA [17], thus the decrease in SNSA during snoring may have been a consequence of this activation.
Since snoring itself was not accompanied by an increase in SNSA or decrease in PNSA compared to control, the question remains as to how snoring may ultimately result in long-lasting increases in SNSA and/or decreases in PNSA during non-snoring periods. There are at least three possibilities. First, it is possible that snoring subjects participating in our study were genetically predisposed to a reduction in PNSA and an increase in SNSA. Given our experimental design, we cannot eliminate this possibility. Future studies designed to examine autonomic nervous system activity in non-apneic snorers, before and after prolonged nCPAP treatment, may resolve whether the changes in PNSA and SNSA we observed are an inheritable trait.
Second, the impact of exposure to nCPAP treatment on the non-apneic snorers compared to controls may have been different (see Critique of methods for detailed discussion on this issue). We consider this possibility highly unlikely since the nCPAP pressure used and the duration of application was similar for both populations. More importantly, even in the absence of nCPAP treatment, PNSA decreased and SNSA increased during non-snoring compared to snoring (i.e. non-snoring vs. snoring in trial 1).
Third, we postulate that inputs activated during snoring concurrently with vagal afferents may ultimately be responsible for the long-lasting changes in PNSA and SNSA activity that we observed. Thus, the influence of pulmonary vagal afferents may predominate during snoring, leading to a reduction in SNSA and an increase in PNSA. However, we speculate that resistive loading and snoring may concurrently enhance brainstem activity (i.e. the reticular activating system), possibly via inputs from mechanoreceptors [18], metabolic receptors [10,19], and upper airway receptors [20,21], that eventually leads to increases in SNSA and decreases in PNSA once snoring is terminated. The possibility that snoring may influence brainstem activation and ultimately autonomic nervous system activity is supported by the findings of Lofaso and colleagues [22]. These investigators showed in non-apneic snorers that graded levels of arousal (based on various changes in EEG characteristics) subsequent to termination of snoring were associated with increases in heart rate and blood pressure. More importantly for our findings, Lofaso and colleagues [22] showed that even in the absence of cortical arousal (no evidence of EEG changes) blood pressure and heart rate were greater immediately after termination of snoring compared to measures obtained during snoring [22]. The possibility that both the acute and long-lasting impact of brainstem activation is masked during snoring, when pulmonary vagal afferents are most influential, is supported by studies which have shown that muscle sympathetic nervous system activity is profoundly inhibited by the activation of pulmonary vagal afferents in the presence of stimuli known to enhance SNSA [23].
Clinical implications
The upper airway resistance syndrome has been described as a form of sleep disordered breathing that is characterized by repetitive increases in resistance in airflow within the upper airway leading to brief cortical arousals and daytime somnolence [24]. In contrast to this clinical description, our subjects did not display excessive numbers of cortical arousal compared to control and did not suffer from daytime somnolence. Nevertheless, despite the absence of cortical arousal we showed that PNSA was decreased and SNSA activity was increased during non-snoring periods in non-apneic snorers compared to control and during non-snoring periods compared to snoring periods within the non-apneic population. The latter finding is similar to results which showed that blood pressure and heart rate increased at the termination of snoring independent of cortical arousal [22,25]. Thus, alterations in autonomic nervous system activity may lead to autonomic and cardiovascular alterations in individuals with increased airflow resistance independent of evidence of cortical arousal. Moreover, our results suggest that alterations in heart rate variability may precede changes in traditional markers of cardiovascular function (i.e. blood pressure) in non-apneic snoring individuals. Given that measures of heart rate variability have been shown to be useful as a prognostic indicator of future cardiovascular events [13,14,26-30] these measures may be useful in determining whether normotensive individuals that snore (or for that matter suffer from obstructive sleep apnea) are at an increased risk for the development of hypertension or cardiovascular disease. Whether the alterations that we observed in our relatively young, healthy and predominantly male snoring population would benefit from long-term treatment with nCPAP, or become more severe with age and/or the presence of other co-morbid conditions remains to be determined.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
GG participated in data acquisition, performed statistical analysis on the data and helped draft the manuscript. SM participated in data acquisition, coordination of the study and helped to draft the manuscript. JM conceived the study, participated in its design and coordination, performed statistical analysis and helped draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by the American Heart Association. ResMed supplied the continuous positive pressure airway machines used in this study.
==== Refs
Bixler EO Vgontzas AN Lin HM Ten Have T Leiby BE Vela-Bueno A Kales A Association of hypertension and sleep-disordered breathing Arch Intern Med 2000 160 2289 2295 10927725 10.1001/archinte.160.15.2289
Silverberg DS Oksenberg A Are sleep-related breathing disorders important contributing factors to the production of essential hypertension? Curr Hypertens Rep 2001 3 209 215 11353571
Young T Finn L Hla KM Morgan B Palta M Snoring as part of a dose-response relationship between sleep-disordered breathing and blood pressure Sleep 1996 19 S202 S205 9085511
Narkiewicz K van de Borne PJ Pesek CA Dyken ME Montano N Somers VK Selective potentiation of peripheral chemoreflex sensitivity in obstructive sleep apnea Circulation 1999 99 1183 1189 10069786
Calabrese P Dinh TP Eberhard A Bachy JP Benchetrit G Effects of resistive loading on the pattern of breathing Respir Physiol 1998 113 167 179 9832235 10.1016/S0034-5687(98)00063-2
Calabrese P Perrault H Dinh TP Eberhard A Benchetrit G Cardiorespiratory interactions during resistive load breathing Am J Physiol Regul Integr Comp Physiol 2000 279 R2208 R2213 11080087
DeWeese EL Sullivan TY Yu PL Ventilatory response to high-frequency airway oscillation in humans J Appl Physiol 1985 58 1099 1106 3988667
Henke KG Sullivan CE Effects of high-frequency oscillating pressures on upper airway muscles in humans J Appl Physiol 1993 75 856 862 8226491
Stoohs R Guilleminault C Snoring during NREM sleep: respiratory timing, esophageal pressure and EEG arousal Respir Physiol 1991 85 151 167 1947456 10.1016/0034-5687(91)90058-Q
Dempsey JA Sheel AW St Croix CM Morgan BJ Respiratory influences on sympathetic vasomotor outflow in humans Respir Physiol Neurobiol 2002 130 3 20 12380012 10.1016/S0034-5687(01)00327-9
Anonymous EEG arousals: scoring rules and examples: a preliminary report from the Sleep Disorders Atlas Task Force of the American Sleep Disorders Association Sleep 1992 15 173 184 11032543
Rechtschaffen A Kales AA A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects 1968 National Institutes of Health, Washington, D.C
Anonymous Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology Circulation 1996 93 1043 1065 8598068
Berntson GG Bigger JT Eckberg DL Grossman P Kaufmann PG Malik M Nagaraja HN Porges SW Saul JP Stone PH van der Molen MW Heart rate variability: origins, methods, and interpretive caveats Psychophysiology 1997 34 623 648 9401419
Montner PK Greene ER Murata GH Stark DM Timms M Chick TW Hemodynamic effects of nasal and face mask continuous positive airway pressure Am J Respir Crit Care Med 1994 149 1614 1618 8004320
Eastwood PR Curran AK Smith CA Dempsey JA Hemodynamic effects of pressures applied to the upper airway during sleep J Appl Physiol 2000 89 537 548 10926636
Gerber U Polosa C Effects of pulmonary stretch receptor afferent stimulation on sympathetic preganglionic neuron firing Can J Physiol Pharmacol 1978 56 191 198 638870
Leuenberger UA Mostoufi-Moab S Herr M Gray K Kunselman A Sinoway LI Control of skin sympathetic nerve activity during intermittent static handgrip exercise Circulation 2003 108 2329 2335 14597585 10.1161/01.CIR.0000093280.40118.30
St Croix CM Morgan BJ Wetter TJ Dempsey JA Fatiguing inspiratory muscle work causes reflex sympathetic activation in humans J Physiol 2000 529 493 504 11101657 10.1111/j.1469-7793.2000.00493.x
Bachoo M Polosa C Properties of a sympatho-inhibitory and vasodilator reflex evoked by superior laryngeal nerve afferents in the cat J Physiol 1985 364 183 198 4032295
Tomori Z Widdicombe JG Muscular, bronchomotor and cardiovascular reflexes elicited by mechanical stimulation of the respiratory tract J Physiol 1969 200 25 49 5761951
Lofaso F Goldenberg F d'Ortho MP Coste A Harf A Arterial blood pressure response to transient arousals from NREM sleep in nonapneic snorers with sleep fragmentation Chest 1998 113 985 91 9554636
Seals DR Suwarno NO Dempsey JA Influence of lung volume on sympathetic nerve discharge in normal humans Circ Res 1990 67 130 141 2364488
Exar EN Collop NA The upper airway resistance syndrome Chest 1999 115 1127 39 10208219 10.1378/chest.115.4.1127
Krieger J Schroder C Erhardt C [Cortical arousal, autonomic arousal. Evaluation techniques and clinical importance] Rev Neurol (Paris) 2003 159 6S107 12 14646811
Bigger JT Fleiss JL Steinman RC Rolnitzky LM Kleiger RE Rottman JN Correlations among time and frequency domain measures of heart period variability two weeks after acute myocardial infarction Am J Cardiol 1992 69 891 898 1550018 10.1016/0002-9149(92)90788-Z
Eckberg DL Physiological basis for human autonomic rhythms Ann Med 2000 32 341 9 10949066
La Rovere MT Pinna GD Hohnloser SH Marcus FI Mortara A Nohara R Bigger JT JrCamm AJ Schwartz PJ Baroreflex sensitivity and heart rate variability in the identification of patients at risk for life-threatening arrhythmias: implications for clinical trials Circulation 2001 103 2072 2077 11319197
Liao D Cai J Barnes RW Tyroler HA Rautaharju P Holme I Heiss G Association of cardiac autonomic function and the development of hypertension: the ARIC study Am J Hypertens 1996 9 1147 1156 8972884 10.1016/S0895-7061(96)00249-X
Molgaard H Sorensen KE Bjerregaard P Attenuated 24-h heart rate variability in apparently healthy subjects, subsequently suffering sudden cardiac death Clin Auton Res 1991 1 233 237 1822256 10.1007/BF01824992
|
16048652
|
PMC1208915
|
CC BY
|
2021-01-04 16:30:12
|
no
|
BMC Pulm Med. 2005 Jul 27; 5:9
|
utf-8
|
BMC Pulm Med
| 2,005 |
10.1186/1471-2466-5-9
|
oa_comm
|
==== Front
BMC PsychiatryBMC Psychiatry1471-244XBioMed Central London 1471-244X-5-311612022410.1186/1471-244X-5-31Research ArticlePharmacological challenge with a serotonin 1D agonist in alcohol dependence Vythilingum Bavanisha [email protected] Charmaine J [email protected] J Stefan [email protected] Willie [email protected] Dan J [email protected] MRC Unit on Anxiety and Stress Disorders, Dept of Psychiatry University of Stellenbosch, South Africa2 Dept of Psychiatry, University of Stellenbosch, South Africa3 Biostatistics Unit, Medical Research Council of South Africa2005 24 8 2005 5 31 31 4 5 2005 24 8 2005 Copyright © 2005 Vythilingum et al; licensee BioMed Central Ltd.2005Vythilingum et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Both animal and clinical studies have implicated serotonergic dysfunction in the pathogenesis of alcohol abuse and dependence. However the exact mechanisms involved remain unknown. Theoretically, low serotonin promotes alcohol seeking behavior. Sumatriptan is a serotonin1D agonist. It is postulated that sumatriptan's agonism at this terminal autoreceptor increases negative feedback, creating a net effect of decreased serotonergic neurotransmission. Administration of sumatriptan should therefore produce a craving for alcohol and the desire to drink.
Methods
Fifteen patients with alcohol dependence who had undergone detoxification were recruited. Sumatriptan (100 mg) and placebo was administered in cross-over fashion on 2 separate days 72 hours apart. Both patients and raters were blind to all treatments.
Patients were assessed on the following scales at -30, 0, 30, 90, 150 and 210 minutes: A 6-item scale designed to rate the patient's intention to drink; The Sensation Scale; a 13-item affect analog scale designed to rate the pattern and extent of emotional changes; and an 8-item scale designed to rate the patient's craving for alcohol
Results
No significant differences were found between the placebo and sumatriptan groups and no significant cross over effects were found.
Conclusion
The general lack of efficacy of sumatriptan in producing alcohol-like symptoms or a desire to drink alcohol may suggest that the 5HT1D receptor plays little role in the pathophysiology of alcoholism.
==== Body
Background
Both animal and clinical studies have implicated serotonergic dysfunction in the pathogenesis of alcohol abuse and dependence [1,2]. However the exact mechanisms involved remain unknown. Pharmacological challenge studies provide one way of identifying the specific receptor systems involved.
Serotonergic dysfunction is thought particularly to be involved in alcohol craving. Verheul et al [3] proposed three types of craving – reward craving, relief craving and obsessive craving. Obsessive craving is defined as the loss of control over intrusive thoughts about alcohol and is believed to be mediated by a deficit of serotonin.
m-chlorophenylpiperazine (m-CPP) is a serotonin agonist with actions at 5-HT1a, 5-HT1d, 5-HT2c receptors. The 5HT1a and 5HT1d receptors are autoreceptors and agonism here has been demonstrated to decrease net serotonergic transmission.[4] m-CPP has been reported to elicit craving for alcohol and feelings similar to intoxication in abstinent alcoholics [5,6]). This relates well to animal studies, where TFMPP administration (a compound structurally similar to mCPP) has discriminant stimulus properties similar to low doses of alcohol [7].
Taken together, this suggests that alcohol may be used to self-medicate in patients with dysequilibrium of the serotonergic system. This hypothesis is given further support by studies showing the efficacy of serotonergic agents in the treatment of alcoholism [8-11]. The serotonin system is, however, complex, and mCPP is a relatively nonselective agent. In order to delineate further the exact mechanisms involved in alcoholism more selective agents must be used.
Sumatriptan is a selective serotonin(5HT)1D agonist. It is postulated that sumatriptan's agonism at this terminal autoreceptor increases negative feedback, creating a net effect of decreased serotonergic neurotransmission [4] In line with Verheul's model of craving [3] sumatriptan administration, by decreasing serotonergic transmission, should theoretically produce a desire to drink.
Studies of sumatriptan administration in alcohol dependent patients suggest involvement of the 5HT1D receptor. Sumatriptan stimulates the release of growth hormone (GH) and inhibits the release of prolactin in normal individuals. In both long term and newly abstinent alcoholics sumatriptan administration did not increase GH secretion, suggesting alteration at the 5HT1D receptor. [12,13]
The 5HT1D receptor may also be associated with repetitive behaviours. In autism, increased GH response to sumatriptan was noted [14], with the severity of repetitive behaviours correlating with response. Addictive behaviours may be viewed as a type of repetitive behaviour.
Furthermore, in OCD, sumatriptan has been suggested to produce similar effects to those of mCPP, and its administration led to exacerbation of symptoms [15], i.e loss of control over obsessive thoughts. We hypothesized that effects of sumatriptan would be similar to those of mCPP in patients with alcoholism, namely that administration of sumatriptan would potentiate obsessive craving for alcohol. Should this hypothesis be bourne out, it would point to specific involvement of the 5HT1d receptor in the pathophysiology of alcohol craving.
Methods
Fifteen patients with alcohol dependence who had undergone detoxification (4 females and 11 males) were recruited from the acute detoxification units at Tygerberg and Stikland hospitals, Cape Town, South Africa. All subjects were medically stable, with no history of cardiac disorder, and with liver enzymes less than 1.5 times the maximum laboratory range at baseline. All subjects had at least seven alcohol and benzodiazepine-free days. For 72 hours prior to the challenge, subjects adhered to a tyramine-free diet.
Informed written consent was obtained from all subjects and the ethics committee of the University of Stellenbosch approved the study. The research was carried out in accordance with the Declaration of Helsinki of the World Medical Association. During the study subjects were inpatients at the Alcohol Rehabilitation Unit at Stikland Hospital, and received ongoing psychotherapy. On completion of a 6 week program, subjects were discharged to a weekly outpatient therapy group. These measures were felt to be sufficiently containing to prevent experimentally induced relapse.
Sumatriptan (100 mg) and placebo was administered orally in cross-over fashion on 2 separate days 72 hours apart. Sumatriptan was chosen as the challenge agent as it has the least side effects of the triptans. A drawback though, is that it has poor penetration of the blood brain barrier. However, previous work [12,13] has showed altered hormonal responses in alcohol dependent patients, suggesting that sufficient amounts do cross over. Furthermore, in studies of OCD and related disorders, sumatriptan is the most commonly used challenge agent. The 100 mg dose was chosen as in previous work on OCD and OCD related disorders, 100 mg had been shown to increase obsessions and repetitive behaviours [14,15] Both patients and raters were blind to all treatments.
Patients were assessed on the following scales at -30, 0 30, 90, 150 and 210 minutes:
1. A 6-item scale designed to rate the patient's intention to drink. (modified from [6]) Items were rated from 0 (not at all) to 4 (extremely), and included questions such as "I would like to drink alcohol" and "I intend to drink alcohol in the near future". A 6th question, "the tablet I took this morning feels similar to alcohol" was also posed.
2. The Sensation Scale [16]. Items included side effects commonly produced by alcohol, such as nausea, facial flushing, dizziness and numbness. Items were rated from 0 (not at all) to 10 (a great deal).
3. A 13-item affect analog scale designed to rate the pattern and extent of emotional changes[17].
4. An 8-item scale designed to rate the patient's craving for alcohol. Items included " I crave a drink right now" and "I want a drink so bad I can taste it". [18]
Statistical analyses were carried on out the total score for each questionnaireas well specific variables of interest. For every variable the six sequential observed scores, y1...y6, of every subject on every occasion were summarized in two informative statistics, the sum S = y1+y2+y3+y4+y5+y6, and the linear trend contrast L = -8y1-5y2-2y3+y4+4y5+10y6. Note that the weights in the L contrast are obtained from the first order orthogonal polynomial for the unequally spaced time intervals of observation.
Thus, for a particular variable every subject has two values of S one for sumatriptan and one for placebo,, say S1 and S2., and the drug effect is measured by S1-S2. The cross-over effect is measured by an estimate of the difference in location of the distribution of S1-S2 in the two groups; it can be mean(S1-S2, Group 1)-mean((S1-S2, Group2), or an estimate based on the Wilcoxon Rank Sum test; there are other possibilities, not used here. In the case of the mean the natural test statistic is of the form z = (difference of means)/(s.e. of the difference); note that the Welch version of the denominator was used in the calculation of z. The distribution of this statistic is "t" if the variable S1-S2 is normally distributed, with appropriate adjustment of the number of degrees of freedom for the Welch version of the ttest. In the present instance it is not, but being a difference of linear functions the distribution is symmetric, and significance levels given by the t-distribution are robust. However, in marginal cases exact significance levels for z was obtained from a permutation distribution. Further, in marginal cases the Wilcoxon Rank Sum test was used for confirmation, or otherwise, of the result.
The linear effect scores, L, were analysed in the same way as the S scores.
Results
All but 2 measures showed no significant changes (See table 1). On the variable "anxious" (item 8 of the affect analog scale) a significant cross over effect was found (z = -2.36, p = 0.046, df = 8; exact permutation P = 0.031; Wilcoxon Rank Sum test P = 0.023). In the group that received sumatriptan first the mean S1-S2 was negative, in the group that received placebo first it was positive. Neither mean differs clearly significantly from zero. In the placebo first group the signficance level attained was 0.08. (See table 2)
Table 1 Group N Mean SE Mean T p
Affect analog Scale Cross-over effect Placebo first 9 0.0 1.3 -.1.25 0.241
Sumatriptan first 6 2.5 1.5
Linear Trend effect Placebo first 9 -3.7 1.3 -0.91 0.391
Sumatriptan first 6 -4.5 1.5
Intent to drink scale Cross Over effect Placebo first 9 -2.3 4.5 -0.63 0.54
Sumatriptan first 6 0.67 1.7
Linear Trend effect Placebo first 9 16.1 7.9 2.12 0.058
Sumatriptan first 6 -2.67 4.0
Sensation Scale Cross Over effect Placebo first 9 -6.6 24 -0.66 0.52
Sumatriptan first 6 11 11
Linear Trend effect Placebo first 9 -6.8 21 0.80 0.44
Sumatriptan first 6 -27.7 15
Craving Scale Cross Over effect Placebo first 9 -7.4 4.2 -1.92 0.084
Sumatriptan first 6 2 2.4
Linear Trend effect Placebo first 9 3.6 5.9 -0.37 0.72
Sumatriptan first 6 6.3 4.2
Table 2 Cross Over Effect – variable "Anxious" on Affect Analog Scale
Group N Mean SE Mean T p
"Anxious" Cross-over effect Placebo first 9 -0.67 0.33 -2.36 0.046
Sumatriptan first 6 0.83 0.54
On the intention to drink scale no significant mean effect was found. In the linear trend scores there is some evidence of a significant crossover effect (t = 2.12, p = 0.058, df = 11; exact permutation P = 0.075; Wilcoxon Rank Sum test P = 0.059). Further analysis showed that no significant difference in linear trend was found for the group that received sumatriptan first, but in the group that received placebo first, a marginally significant linear trend effect was noted (t = 2.039, two-sided p = 0.07, one-sided p = 0.04) (See table 3)
Table 3 Linear Trend – Intent to Drink Scale
Group N Mean SE Mean T P
Intent to drink Linear Trend effect Placebo first 9 16.1 7.9 2.039 0.04
Sumatriptan first 6 -2.67 4.0 -.066 0.53
Discussion
Our results show that overall sumatriptan has no significant ability to provoke alcohol lsymptoms or to cause a desire or craving for alcohol. The significant placebo effect on the variable "anxious" may merely represent normal anxiety when confronted with an unfamiliar task. This is made especially plausible by the fact that it only occurs in the group that took placebo first, suggesting that as patients become more familiar with the trial the anxiety disappears.
The interpretation of the linear trend on the intention to drink scale in patients who received placebo first is more difficult. On its own this suggests that sumatriptan did cause changes not seen with placebo. However, if this was true, one would also expect to find a significant linear trend in the group who received sumatriptan first. Furthermore, a significant overall drug effect should have been found. It is more plausible that the trend found is due either to error introduced by the small group size and multiple tests, or by factors unrelated to sumatriptan within the group.
The general lack of efficacy of sumatriptan in producing alcohol-like symptoms or a desire to drink alcohol may suggest that the 5HT1D receptor plays little role in the pathophysiology of alcoholism. This is in contrast to studies that show decreased GH response to sumatriptan administration in abstinent alcoholics.
However, several limiting factors must be borne in mind. Firstly, the sample size was small and consisted of recently detoxified alcoholics. Chronic use of alcohol may alter the neurochemical milieu and provide a false picture of receptor involvement in alcoholism [1].
Secondly, sumatriptan has poor penetration of the blood-brain barrier, and its failure to provoke a response may be a result of this, rather than lack of involvement of the 5HT1D receptor in the pathogenesis of alcoholism. However, given that previous work has shown hormonal (in alcohol dependence) [12,13]and behavioural (in OCD) responses [15], this is less likely.
Furthermore, it may be that GH response to sumatriptan does not necessarily correlate with behavioural changes. In OCD, there have been conflicting reports about symptom exacerbation following sumatriptan challenges, yet a decreased GH response is observed. It is possible that GH release is mediated via different pathways to behavioural effects. This is suggested by the finding in autism, where increased GH release correlated with severity of repetitive behaviours. As both OCD and alcohol dependence can be conceptualized as repetitive behaviours, this contrasts with findings in these disorders.
Finally, the majority of the sample (11 patients) had late onset alcohol dependence (age of onset > 25) using the von Knorring criteria [19]. George et al [20] found ethanol -like responses to mCPP were only found in the early onset group. It is possible that we did not observe a response in our sample as most had late onset alcohol dependence.
Conclusion
While our study suggests that the 5HT1D receptor does not play a significant role in the pathogenesis of alcoholism, further studies with sumatriptan and with agents that have better blood-brain barrier penetration are needed in both alcohol dependant and non-alcohol dependent samples to consolidate this finding. Stratification of further samples into early and late onset alcohol dependence may also be of use.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BV participated in the statistical analysis and prepared the manuscript for publication. CJH participated in the study design and carried out the patient ratings. WP participated in the preparation of the manuscript. JSM performed the statistical analysis. DJS conceived of the study, participated in the study design and participated in the preparation of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank Mimi Roberts for her help in patient recruitment. Bavanisha Vythilingum, Charmaine Hugo and Dan Stein are supported by the Medical Research Council of South Africa.
==== Refs
LeMarquand D Pihl RO Benkelfat C Serotonin and alcohol intake, abuse, and dependence: clinical evidence Biological Psychiatry 1994 36 326 227 7993959 10.1016/0006-3223(94)90630-0
LeMarquand D Pihl RO Benkelfat C Serotonin and alcohol intake, abuse, and dependence: findings of animal studies Biological psychiatry 1994 36 395 421 7803601 10.1016/0006-3223(94)91215-7
Verheul R van den Brink W Geerlings P A three-pathway psychobiological model of craving for alcohol Alcohol Alcohol 1999 34 197 222 10344781
Davidson C Stamford JA Evidence that 5-hydroxytryptamine release in rat dorsal raphe nucleus is controlled by 5-HT1A, 5-HT1B and 5-HT1D autoreceptors Br J Pharmacol 1995 114 1107 1109 7620698
Benkelfat C Murphy DL Hill JL George DT Nutt D Linnoila M Ethanol-like properties of the serotonergic partial agonist m-chlorophenylpiperazine in chronic alcoholic patients Archives of General Psychiatry 1991 48 333 1672588
Krystal JH Webb E Cooney N Kranzler HR Charney DS Specificity of ethanol-like effects elicited by serotonergic and noradrenergic mechanisms Archives of General Psychiatry 1994 51 898 911 7944878
Signs SA Schechter MD The role of dopamine and serotonin receptors in the mediation of the ethanol interoceptive cue Pharmacol Biochem Behav 1988 30 55 64 3174755 10.1016/0091-3057(88)90424-8
Naranjo CA Sullivan JT Kadlec KE Woodley-Remus DV Kennedy G Sellers EM Differential effects of viqualine on alcohol intake and other consummatory behaviors Clin Pharmacol Ther 1989 46 301 9 2673621
Naranjo CA Kadlec KE Sanhueza P Woodley-Remus D Sellers EM Fluoxetine differentially alters alcohol intake and other consummatory behaviors in problem drinkers Clin Pharmacol Ther 1990 47 490 498 2328557
Naranjo CA Poulos CX Bremner KE Lanctot KL Fluoxetine attenuates alcohol intake and desire to drink Int Clin Psychopharmacol 1994 9 163 72 7814825
Naranjo CA Poulos CX Bremner KE Lanctot KL Citalopram decreases desirability, liking, and consumption of alcohol in alcohol-dependentdrinkers Clin Pharmacol Ther 1992 51 729 39 1535302
Vescovi PP Coiro V Persistence of defective serotonergic and GABAergic controls of growth hormone secretion in long-term abstinent alcoholics Alcohol Alcohol 1997 32 85 90 9131896
Coiro V Vescovi PP Alcoholism abolishes the growth hormone response to sumatriptan administration in man Metabolism 1995 44 1577 80 8786727 10.1016/0026-0495(95)90078-0
Novotny S Hollander E Phillips A Allen A Wasserman S Iyengar R Increased repetitive behaviours and prolactin responsivity to oral m-chlorophenylpiperazine in adults with autism spectrum disorders Int J Neuropsychopharmacol 2004 7 249 54 15132762 10.1017/S146114570400433X
Stein DJ Van Heerden B Wessels CJ Van Kradenburg J Warwick J Wasserman HJ Single photon emission computed tomography of the brain with Tc-99m HMPAO during sumatriptan challenge in obsessive compulsive disorder : investigating the functional role of the serotonin auto-receptor Prog Neuropsychopharmacol Biol Psychiatry 1999 23 1079 1099 10621951 10.1016/S0278-5846(99)00051-2
Maisto SA Connors GJ Tucker JA McCollam JB Adesso VJ Validation of the sensation scale: a measure of subjective physiological responses to alcohol Behav Res Ther 1980 18 37 43 7369986 10.1016/0005-7967(80)90067-4
Hollander E Stein DJ DeCaria CM Cohen L Saoud JB Skodol AE Kellman D Rosnick L Oldham JM Serotonergic sensitivity in borderline personality disorder :preliminary findings American Journal of Psychiatry 1994 151 277 280 8296905
Bohn MJ Krahn DD Staehler BA Development and initial validation of a measure of drinking urges in abstinent alcoholics Alcoholism, Clinical and Experimental Research 1995 19 600 6 7573780
von Knorring A-L Bohman M von Knorring L Oreland L Platelet MAO activity as a biological marker in subgroups of alcoholism Acta Psychiatr Scand 1985 72 51 58 4036659
George DT Benkelfat C Rawlings RR Eckardt MJ Phillips MJ Nutt DJ Wynne D Murphy DL Linnoila M Behavioral and neuroendocrine responsesto mchlorophenylpiperazine in subtypes of alcoholics and in healthy comparisonsubjects Am J Psychiatry 1997 154 81 7 8988963
|
16120224
|
PMC1208916
|
CC BY
|
2021-01-04 16:33:03
|
no
|
BMC Psychiatry. 2005 Aug 24; 5:31
|
utf-8
|
BMC Psychiatry
| 2,005 |
10.1186/1471-244X-5-31
|
oa_comm
|
==== Front
BMC Struct BiolBMC Structural Biology1472-6807BioMed Central London 1472-6807-5-161611148910.1186/1472-6807-5-16Research ArticleComputational identification of residues that modulate voltage sensitivity of voltage-gated potassium channels Li Bin [email protected] Warren J [email protected] Department of Biological Sciences, University of Alberta, Edmonton, Canada T6G 2E92 Department of Cell Biology, University of Alberta, Edmonton, Alberta, Canada3 Partners AIDS Research Center, Massachusetts General Hospital, Harvard Medical School, 149 13th Street 6th floor, Charlestown MA USA 021292005 19 8 2005 5 16 16 19 3 2005 19 8 2005 Copyright © 2005 Li and Gallin; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Studies of the structure-function relationship in proteins for which no 3D structure is available are often based on inspection of multiple sequence alignments. Many functionally important residues of proteins can be identified because they are conserved during evolution. However, residues that vary can also be critically important if their variation is responsible for diversity of protein function and improved phenotypes. If too few sequences are studied, the support for hypotheses on the role of a given residue will be weak, but analysis of large multiple alignments is too complex for simple inspection. When a large body of sequence and functional data are available for a protein family, mature data mining tools, such as machine learning, can be applied to extract information more easily, sensitively and reliably. We have undertaken such an analysis of voltage-gated potassium channels, a transmembrane protein family whose members play indispensable roles in electrically excitable cells.
Results
We applied different learning algorithms, combined in various implementations, to obtain a model that predicts the half activation voltage of a voltage-gated potassium channel based on its amino acid sequence. The best result was obtained with a k-nearest neighbor classifier combined with a wrapper algorithm for feature selection, producing a mean absolute error of prediction of 7.0 mV. The predictor was validated by permutation test and evaluation of independent experimental data. Feature selection identified a number of residues that are predicted to be involved in the voltage sensitive conformation changes; these residues are good target candidates for mutagenesis analysis.
Conclusion
Machine learning analysis can identify new testable hypotheses about the structure/function relationship in the voltage-gated potassium channel family. This approach should be applicable to any protein family if the number of training examples and the sequence diversity of the training set that are necessary for robust prediction are empirically validated. The predictor and datasets can be found at the VKCDB web site [1].
==== Body
Background
During the evolution of proteins there is interplay between selection acting to keep some residue identities constant, thus preserving protein function, and selection acting to accept new sequence variants with altered properties that confer improved survival. Thus, when studying the evolution of the structure-function relationship in a family of proteins, identification of invariant residues within the family identifies parts of the protein that are of central importance to its function. This idea is central to many comparative studies of protein structure/function relationships, and the concept has been extended to studies of pairs of residues whose identities co-vary in an apparently compensatory manner [2].
However, the converse idea, that varying residues are not centrally important to the protein's function, is not necessarily true. Although it is true that residues that do not have a major impact on protein function will show extensive variation over time, it is also true that residues that contribute to the quantitative variation in a protein's properties will also vary.
The problem that arises, then, is how to distinguish the residues whose variation in identity is responsible for functional variations in the protein from those residues whose variation in identity is relatively immaterial to function. These residues will not be detected by evaluating the extent of variation in a given residue or in pairs of residues. Rather, the residues will co-vary with the property of the protein that they affect. To solve this problem it is necessary to use techniques that can detect associations between combinations of residue identities at any position in the protein and the quantitative value of the parameter of interest. We here report an analysis to detect such an association between structure and function in voltage-gated potassium channels (VKCs) using machine learning techniques.
VKCs are membrane proteins that regulate the passage of potassium ions through membranes [3]. When the voltage difference across a membrane reaches a threshold, the probability that VKCs will open begins to become significant, allowing increased potassium ion diffusion through an ion-selective pore in the channel. This voltage-regulated potassium ion permeability is critical in cellular excitability. Mutations in VKC genes have been shown to be associated with cardiac arrhythmias [4], episodic ataxia [5], and other diseases [6,7]. Consequently, VKC proteins have been considered good targets for drug design directed at a number of diseases [8-10].
A functional VKC consists of four subunits, each containing six transmembrane regions, S1 through S6. S4 has been shown to function as the main voltage-sensing domain [3], acting by moving perpendicular to the plane of the membrane upon depolarization [11,12]. This movement causes a conformational change in the region of the pore to open the "gate" and allow potassium ions to pass through. There is currently heated dispute over which of several mechanisms that have been proposed to explain how the sensor movement changes the channel conformation is correct [13].
Elements of the molecular basis of VKC function have been elucidated through structural studies [11,14-18]. The structures of several potassium channels of different types have been determined crystallographically [11,15,17,19]. The structure of the ion selective pore is very similar in all of the models; these studies have clearly identified many aspects of the molecular dynamics of selective ion permeability. Although the structure of one voltage-gated potassium channel has been determined [11], the unusual mobility of the voltage sensor region and the necessity of using a bound antibody to stabilize the crystallized conformation raise serious questions of how similar that determined structure is to the functional conformation of the ion channel [20]. Extensive mutagenesis and biophysical studies of different voltage-gated ion channels have lead to several models of function that are quite different from each other and from the model proposed based on the KvAP crystal structure (KvAP structure ref). Thus, although the molecular basis of ion selectivity and permeability is well supported by the current structural knowledge, the molecular basis of voltage sensing, and in particular the molecular basis of the fine differences in voltage sensitivity between channels, is not well defined by structural studies.
In the absence of three-dimensional structures of various VKCs that unambiguously show different opening/closing stages, mutagenesis of individual residues of different VKCs has been the main method for inferring the structure-function relationship of VKCs. However, it is prohibitively time-consuming and costly to do mutagenesis of all residues individually and in combinations in different VKCs. Computational tools, usually multiple sequence alignment, have been used to identify conserved regions of VKCs and limit the priority in mutagenesis experiments to evolutionarily conserved residues [21-23]. Unfortunately, details of the complex structure-function relationship between individual residues and the electrophysiological properties, which are mostly continuous quantitative parameters [24], are too complicated to understand by simple inspection of aligned VKC sequences. With dozens of VKC sequences of a few hundred residues each and continuous electrophysiological variables, more mature data mining tools, such as machine learning, are necessary.
Machine learning generalizes an underlying data model for a phenomenon by "learning" from existing data from specific examples of that phenomenon, using various classification rules. It yields a mathematical or computational model that can best describe the existing data and predict classifications of new data [25]. Because of its ability to extract complex models from large datasets, machine learning has been successfully applied to many data-rich problems such as marketing reports, weather prediction, automatic genome annotation and microarray data analysis [26-29].
In this report we tested several learning algorithms, as implemented in the WEKA program package [30]. Most of these were categorical learners; they evaluate how features can be used to assign each sequence to a pre-defined discrete category. The OneR classifier identifies a set of rules based on the identity of the amino acid residue at only one of the various aligned amino acid positions that best classifies each sequence in the training set. The Decision Tree classifier identifies a minimal set of amino acid positions, and branching decisions based on residues at these positions, that correctly classify the training data. The Naïve Bayes classifier uses observed frequencies of residues at selected positions to apply Bayes' theorem to make probabilistic predictions of the category to which a sequence belongs. Kernel Density estimation estimates the probability distribution for use in the Bayesian analysis if a non-normal distribution is suspected. The K Nearest Neighbour (KNN) classifier was used for both categorical classification and for classification that treats V50 as a continuously varying quantitative characteristic; this method uses a distance measure to determine which elements of the training set are closest in attribute space to the example being evaluated and assigns the average of the k examples that are most similar to the test example as the predicted value.
Typically, a protein family comprises dozens of members with hundreds of residues in each member. Such datasets present a unique type of problem for machine learning. First, a typical training dataset for machine learning contains distinctively labelled "features" in every instance. With protein sequence datasets, all sequences must be aligned with each other to identify homologous residues (features). Second, dozens of sequences with hundreds of residues each create a dataset with very high dimensionality, which compromises learning performance. Finally, besides generating a classifier with high accuracy, it is pertinent to bench biologists to evaluate the biological importance of individual residues (features) that contribute to a good learning performance during training, so it is desirable to use learning methods that return the basis for their prediction.
In this report, we have mined available VKC sequence and electrophysiological data using machine learning and related feature selection techniques, and derived a model that predicts one of the central electrophysiological parameters, half activation voltage (V50) [24], of a given VKC, based on only its amino acid sequence. Our best result was obtained using a k-nearest neighbor classifier (k = 1) combined with feature selection using a wrapper algorithm [31], yielding a mean absolute error (MAE) between the predicted and published V50 values of 7.0 mV in a repeated ten-fold cross validation. The prediction by our final predictor was validated by permutation test and by comparison of predictions to independently obtained experimental results. The training process also provides a rational basis for identifying residues potentially critical to the activation of VKCs, and several identified key residues are located in regions that have been proposed to modulate VKC activation.
The methods that we have applied to the study of VKCs are general. With appropriate alignment, feature selection and model validation, this analytical approach can be used to generate biological hypotheses in other protein families and these hypotheses can be practically tested using site-directed mutagenesis.
Results
Learning without feature selection
A dataset consisting of 296 aligned positions from 58 VKC sequences (Dataset 1) was initially used to train different learning algorithms to predict the V50 value of a given VKC sequence. Figure 1 illustrates schematically the process that was used in developing the final predictor. V50 values were divided into seven nominal classes for categorical learning. The accuracy of the best categorical learning was below 30% (Figure 2A). The MAE of the best numerical prediction of V50 values with the KNN classifier (Figure 2B) was close to 18 mV. This analysis is equivalent to assigning a predicted V50 to a channel based on the V50 of the channel's nearest neighbor or neighbors in a distance-based phylogenetic tree. Evidently, these learning algorithms alone do not produce an accurate model for prediction if they are trained with such a high dimensional dataset of less than 60 instances. This is likely because a relatively small number of residues affect the V50 of a channel and the majority of residues affect other parameters, which do not co-vary with V50, of the channels.
Learning with data filtering
To improve learning performance with this high dimensional dataset, we added feature selection before learning, using a filtering algorithm to decrease the dimensionality. All residues (features) were ranked based on their information gain scores [32,33]. Different numbers of top-ranked residues were then used for learning. The best learning performance was obtained using only the top five ranked features (residues), and the categorical accuracy improved to 36%. The MAE of the numerical prediction of V50 values with a KNN classifier was now reduced to 15 mV (Figure 2B). While dimension reduction by filtering to select residues with high information gain did appear to yield a better learning performance, the improvement is marginal.
Learning with wrapper
We also applied a wrapper algorithm, a more learning performance-driven feature selection method than filtering [31]. From a large number of sets of residue (feature) combinations, wrapper selected the residue set that yielded the best learning performance. The prediction accuracies with all categorical learning algorithms improved, with the best classification of 60% accuracy using the KNN classifier. When the KNN classifier (k = 1) was combined with wrapper to predict a numerical V50 value based on a VKC sequence, the MAE of prediction improved to 9.5 mV from 17.8 mV (Figure 2B). The best prediction accuracy was obtained with six residues (features).
Effect of scoring matrix choice and distance formula
We used an identity matrix (Formula 1.1) and transformed BLOSUM62 and PAM100 amino acid matrices (Formulas 1.0 and 1.2) for calculating distances in KNN classification. The best MAEs were the same for all three scoring matrices in repeated ten-fold cross validations. The best predictors obtained with each scoring matrix used 6 features (BLOSUM62), 8 features (PAM100) and 9 features (identity matrix). We performed the remainder of our analyses using the BLOSUM62-based matrix because it yielded the best accuracy with the fewest features. Five of the six features identified with the BLOSUM62-based matrix were identified by analyses with at least one of the other two scoring matrices (Table 1).
We also evaluated the result obtained by summing the individual character distances (Manhattan distance) to the result obtained using the Euclidian distance as described in Materials and Methods, using the transformed BLOSUM62 distance matrix. We obtained the same MAE with the same six selected features, indicating that the learning method is robust to the distance calculation method, at least with this data set.
Learning combined with outlier selection
Since the dataset has only 58 VKC sequences, a small number of outliers, or incorrect class labels, might have greatly affected the training process and thus led to poor learning performance. We evaluated the effect of deleting each sequence from the dataset, by training the KNN classifier with each of the 58 possible subsets of 57 sequences. The top 50 subsets with 57 VKC sequences that produced best learning performances using a repeated ten-fold cross validation were kept and the pruning procedure was then repeated with each of the 50 subsets as a starting point (the flow diagram for this process is shown in Figure 3A). The six-feature set that gives the best learning performances using Dataset 1 (MAE = 9.5 mV) was used during outlier selection. In spite of the plateau in Round 1 and 3, there were significant improvement of learning accuracies in Round 2 and Round 4. After four pruning rounds the improvement in accuracy significantly slowed down in the following rounds (Figure 3B). Thus, we believe that Round 1–4 represents informative gains in accuracy from deleting true outliers, whereas the improvement in later rounds is due to over fitting. This outlier selection improves the MAE from 9.5 to 7.0 mV
During the pruning process, four VKC sequences, VKC8 (Kv1.3 mouse), VKC98 (Kv1.4 dog), VKC149 (Kv2 squid), and VKC171 (Kv4.3 mouse) [34], were consistently selected as outliers from Rounds 1–4, although the order by which they were deleted varied. We therefore created a new dataset of 54 sequences (Dataset 2). The new dataset was used to construct the KNN final classifier, for which the best MAE improved to 7.0 mV (Figure 2B). We also re-ran the complete training protocol using the wrapper algorithm with Dataset 2, and exactly the same feature set was again selected, producing the best MAE of 7.0 mV.
We also evaluated two other measures of prediction quality, the R-squared value, which represents the percentage of variance in the training set that is explained by the predictor, and the correlation between the actual and predicted V50 values. For Dataset 1 the R-squared value for all characters is -0.18 and the correlation coefficient is 0.28; the R-squared value for the selected six features is 0.64 and the correlation coefficient is 0.79. For Dataset 2 the R-squared value for all characters is -0.09 and the correlation coefficient is 0.36; for the selected six features the R-squared value is 0.77 and the correlation coefficient is 0.89.
Permutation tests
One hundred permuted datasets were generated by randomly shuffling the V50 values among the VKC sequences in the Dataset 2. With the same parameters and settings with which we obtained the predictor, we applied KNN classification combined with the wrapper algorithm for feature selection to each one of these permuted datasets, yielding 100 different predictors with different feature sets and performance values. These 100 replicates provide an estimate of the probability distribution of the MAEs for the null hypothesis that there is no functional relationship between the sequence and the V50 value. The MAEs with the permuted datasets range from 9.9 mV to 15.4 mV (mean = 13.4 mV, SD = 1.1 mV). The performance of the predictor with the original dataset (MAE = 7.0 mV) is significantly better (P < 10-10) than would be expected if there were no connection between the sequence and V50 value.
Validation of predictor with independent experimental data
Thirteen wild type VKCs that were not part of the training data were evaluated with the final predictor. The V50 values of these "new" VKCs had been determined independently in electrophysiological experiments [34,35] (Salvador-Recala V, Gallin WJ, Abbruzzese J, Ruben PC, Spencer AN: A Kv4 channel cloned from the heart of the tunicate Ciona intestinalis and its modulation by a KChIP subunit. Manuscript submitted). The MAE of these predictions is 9.7 mV (See Additional file 2). Within this test set, two VKCs are from species that are evolutionarily distant from any of the other sequences in the training dataset, Hirudo medicinalis and Ciona intestinalis [36,37], and the prediction errors for these two VKC are both over 27 mV (Table 2). When the most distant sequence (H. intestinalis) is removed, the MAE of the remaining twelve VKCs is 8.3 mV, and if both sequences are removed MAE = 6.6 mV, below the MAE we estimated using a repeated ten-fold cross validation (7.0 mV).
We also evaluated the predictor by comparing the predicted V50 values of a number of VKC mutants with experimental data from an alanine mutagenesis scan of rat Kv2.1 by Li-Smerin et al [38]. The comparison is shown in Table 2. The MAE between our predictions and data obtained experimentally is 7.5 mV, which is reasonably close to our estimated MAE of 7.0 mV using cross validation. Note that in five of the cases the predicted values for V50 are the same; this is because none of the mutations to alanine makes a new set of informative sites that is closer to a new training sequence. In the two cases where one of the informative residues was mutated to a residue that is represented in one of the channels in the learning set (see Table 2), the prediction improves significantly (MAE = 1.35 mV, n = 2) (Table 2).
Identification of informative features (residues)
The wrapper algorithm identifies a relatively small number of residues that are the primary determinants of accurate learning. With both Dataset 1 (58 instances) and Dataset 2 (54 instances), six residues were consistently selected to produce the best learning performances (Table 1), using a KNN classifier and a transformed BLOSUM62 scoring matrix. We reason that the residues that were identified as most informative in learning are more likely involved in modulating the physical activation process of VKCs. The selected residues were mapped onto a schematic of the S1–S6 structure (Figure 4). All of them reside in S1–S3, a region that likely plays a modulating role in VKC functioning [39,40].
Independent evolution of character states in informative characters
A phylogenetic tree was inferred with MrBayes v3.0b4 [41] using the 54 channel data set. The evolution of individual characters was then inferred on this tree using maximum parsimony criteria as implemented in MacClade [42]. In the case of all six of the informative characters, at least one of the character states has arisen multiple times during channel evolution (Table 3). In all cases, the residue identities that have arisen independently during channel evolution have large hydrophobic side chains (F, I, V, L).
Discussion
Learning with high dimensional data
Data with high dimensionality are a "curse" to learning performance. As a rule of thumb, the number of instances should be no less, and preferably more, than the number of features to obtain a reasonable learning accuracy [31]. Even with a large number of instances, a large number of irrelevant features can still compromise the learning performance [31]. For biological data, however, enough examples with relatively small dimension are not always achievable. Without feature selection or some other form of dimension reduction prior to data analysis, learning performance with high dimensional data is poor. Many dimension reduction methods have thus been applied to improve learning performance, including principle component analysis and linear discriminant analysis [43,44].
We faced this problem in our analyses. There were less than 60 VKC sequences with published V50 values, and there are nearly 300 residues in each sequence alignment after trimming poorly aligned regions. Most residues likely have little or no role in determining V50 values, and thus are "irrelevant features". Training without feature selection using several machine learning algorithms yielded categorical prediction accuracies consistently lower than 30% (Figure 2A).
Application of a filtering analysis before learning improved the accuracy marginally (Figure 2B). Filtering is a pre-learning data processing method based on evaluation of the information content of the dataset, and thus is independent of the training process. It has been successfully used in other tasks to obtain better learning performance [32,33]. However, it may or may not select truly relevant features, depending on the datasets and the selection criteria. Considering the number of features and number of instances in our datasets, some irrelevant features may well correlate with the final class labels by chance and display a high information gain potential that does not reflect a functional connection between sequence and V50.
We then applied a wrapper algorithm to select features during the learning process. Wrapper uses a heuristic search to select the subset or subsets of features that yield the best learning performance [31]. It "wraps" around the learner and selects best feature sets based on learning accuracies. In a heuristic search there is no formal guarantee that the search will not become trapped in a local optimum and miss the global optimum. To decrease the chance of missing the global optimum without making the analysis intractable, we selected the top 200 residue combinations at each round and used them all as starting points for the next round of searching. Best-first searching was continued until learning performance stabilized. The learning accuracies with wrapper increased greatly for all learning algorithms we used. The best categorical result was obtained with the KNN classifier (k = 1) with an accuracy of 60%. To generate a numerical predictor for V50, we also trained the KNN classifier combined with wrapper for numerical classification; the mean absolute error of prediction improved to 9.5 mV from 17.8 mV (Figure 2B).
Since wrapper does not use an exhaustive search and does not guarantee optimal feature selection, we applied "residue swapping" to explore the possibility that one or more residues would yield better results in the context of the finally selected residue set than they would in the initial search. However, residue swapping did not produce any feature sets that yielded significantly better predictive performance. Although, as an empirical operation, branch swapping does not guarantee the global optimum in phylogenetic analysis [45], this process is a reasonable heuristic approach for searching in the neighborhood of an optimum for other, better, feature sets.
Outlier selection
Typically, with a sufficient amount of data, classification using machine learning is expected to be insensitive to outliers. However, a dataset with a low number of instances relative to features of structural and functional data increases susceptibility to outlier effects.
The V50 values in our dataset were obtained from publications from dozens of labs. We used averages for three V50 values in our datasets because different investigators have published different V50 values for the same VKC sequences. The difference in V50 values of same VKCs from different labs sometimes for two of those channels exceeds 15 mV [46-49]. Thus, it is almost certain that some VKCs in our datasets compromise learning because they are incorrectly labeled.
We evaluated the prediction accuracies with datasets from which one sequence was pruned at each round of training (Figure 3A). Based on the variations in learning performances, we stopped at Round 4 (Figure 3B). At both Round 2 and 4, learning performances displayed an improvement of MAE of almost 1.5 mV (Figure 3B). The improvement of learning performances after Round 4 decreased significantly (Figure 3B).
One sequence was deleted in each round of pruning. Creating the best learning performance from the remaining data, the top fifty such "remaining" sequence sets were all used as starting points for the next round of searching. Four sequences were consistently selected for "deletion" in the first four rounds, although in different orders. The best learning performance produced a MAE of 7.0 mV with the new dataset of 54 VKC sequences (Figure 2B), after deleting the four potential outliers; with the outliers the MAE was 9.5 mV.
Outliers may arise from experimental errors, or the channels may be activated by different mechanisms so V50 values would be affected by a set of residues that is different from those that affect the non-outliers. In the latter case, the "deleted" outliers become interesting research targets [50,51]. However, we could not rigorously exclude the possibility that they were selected as outliers due to the specific dataset we used and possible data over-fitting in our training.
Among the deleted outliers, VKC149, a squid Kv2 channel, was shown to undergo extensive RNA editing, leading to its functional diversity [52]. Its G-V curve, which was used to obtain its V50 value, had to be fitted with two Boltzmann functions, adding another layer of complexity to its gating mechanism [52]. VKC171 (Kv4.3 mouse) [50] is a fast inactivating channel. Its activation might overlap its inactivation, which would make it difficult to obtain an accurate V50 value [50].
We also tested the possibility channels with the most distant nearest neighbors might be outliers. We identified the nine sequences that were most distant from their nearest neighbors and sequentially removed them from the dataset, evaluating the MAE of the resulting predictor after each deletion using a 10 times 10-fold cross-validation. None of the four channels mentioned above was among this set. The MAEs of the first 8 deletion datasets were greater than that for the full set of 58 sequences (9.5 mV); after the ninth deletion the MAE was 9,2 mV. Thus, an a priori assumption that the most distant sequences will cause poor prediction performance is not valid in this case.
The predictor performance was also evaluated using the R-squared statistic and correlation coefficient of the predicted vs. the actual V50 values. Both of these measures showed a significant improvement of prediction performance with feature selection and outlier removal. The 10-times 10-fold cross validation of best predictor, using the six selected characters and Dataset 2, yielded an R-squared value of .77, indicating that the predictor was accounting for approximately 77% of the variation on V50 between channels. The linear correlation coefficient between the predicted and actual V50 values was 0.89, which also indicates that the predicted V50 values are well correlated with the actual values for the channels.
Statistical evaluation using a permutation test
The machine learning approach that we have implemented searches for an optimal value of MAE, so it will always yield a model that associates the identity of some residues with the V50 value, even if that association is specious. To evaluate whether the association of residue identity and V50 is significant, a statistical test of significance is necessary.
A permutation test is a special case of randomization tests. With a small sample of data, a permutation test generates an approximate probability distribution for a null hypothesis. Permutation tests have been widely used in biomedical and other areas including microarray analysis, SNP research, and clinical studies [53-55]. Compared with other statistical analyses, a permutation test works well with small sample sets and it does not require a normal distribution, which many small samples do not have. Some researchers have even proposed that permutation test should be used in all cases [56].
We evaluated whether there is significant information linking the sequence of a VKC to its V50 value using one hundred permutations of the original dataset, where V50 values were randomly reassigned to sequences. One hundred different computational models were generated with one hundred different sets of features (residues). The best and worst MAEs among these permutation learning are 9.9 mV and 15.4 mV, respectively, with a mean MAE of 13.4 mV and standard deviation of 1.1 mV. The mean MAE, 13.4 mV, differs significantly from that of the predictor that was generated with the original, non-permuted, dataset, 7.0 mV (P < 10-10). Since both KNN classification and the feature selection process were involved in the permutation test, each test yielded a best model that mathematically correlates a set of residues with a permuted set of V50 values. The fact that the original model significantly outperforms any of the "permuted" models strongly supports the conclusion that the original learning has detected a valid association between sequence elements of VKCs and their V50 values.
Evaluation of predictor using independent experimental data
Due to the limited number of data, we did not retain a portion of data as an independent test set when constructing the predictor. Instead, we used a repeated ten fold cross validation to estimate prediction errors on unseen data. To obtain an independent objective assessment of predictor performance, we located another thirteen VKCs that have been functional characterized, including VKCs that were recently cloned [36], as an independent test set. Using the predictor, the MAE of predictions of all thirteen new VKC instances is 9.7 mV (Additional file 2), which is higher than what we estimated using a repeated ten-fold cross validation (7.0 mV).
We also tested an alternative model, that the V50 could be predicted by assigning the average V50 value of the sister group of each channel on a phylogenetic tree that was constructed based on the full set of 296 aligned amino acid positions. We tested both a distance tree and a Bayesian maximum likelihood tree (Figure 6). The phylogeny-based predictions were significantly poorer than those we obtained with the fully optimized predictor (Additional file 2).
The optimal predictor was built with a KNN classifier. In KNN classification, a close "neighbor" from the training set will be used as a template to classify a new instance. If the training data are not evenly distributed in the instance space, some areas contain fewer instances with larger empty space than others, as is shown in the distance tree of the training data (Figure 5). Evidently, instances that are in these sparse areas will likely be less accurately classified, since they do not have close neighbors. In fact, superposition of the test VKC data on the distance tree of the training data clearly showed an unequal distribution in the sequence space (Figure 5). Among the thirteen test VKC data, all but two are from species that exist in the training set. One VKC is from Hirudo medicinalis [37] and the other is from Ciona intestinalis [36]. The difference between the H. medicinalis channel sequence and its nearest neighbor in the training set is much greater than for any other sequence and its nearest neighbor. The prediction errors of these two VKCs using the predictor are 27.3 mV and 27.4 mV, respectively. When the H. medicinalis sequence is removed from the analysis the MAE for the test set is 8.3 mV, and if both of these sequences are removed the MAE is 6.6 mV. This analysis indicates that a more phylogenetically diverse selection of channels in the training set should improve prediction performance.
In the training set, all V50 values were determined from channels expressed in Xenopus oocytes. In the test set, however, we also included VKCs that have V50 values determined in other cells, such as HEK293 and CHO cells [35,57]. Although it is known that the experimental V50 values of VKCs can vary if they are measured in different cells, the difference is often not significant, as shown by experimental data of several VKCs that have been characterized in both Xenopus oocytes and other cells. Therefore, we believe that the test set serves as a valid independent test set. In fact, it is likely that a better estimate would be obtained if all test instances were measured in Xenopus oocytes, which would remove variation due to differences in expression systems.
We also compared experimental data from a mutagenesis scanning study by Li-Smerin et al with predictions by our predictor (Table 2) [38]. Despite using a test set comprising results from VKC mutants, and the presumably drastic difference between data distributions of the test set and our training sets, prediction results are consistent with results from the published mutagenesis study of VKCs (Table 2). This result supports the conclusion that our estimated prediction error is close to the true error.
Although all of the mutations have changed one of the six informative residues, only two of them have mutated to an amino acid that is represented at that position in one of the training set sequences. The V50 predictions for these two mutants, marked with asterisks, are much closer to the observed V50 values than those for the other three mutants. This suggests that as more channels are cloned and characterized and the variety of the training set increases the performance of this kind of machine learning and prediction will improve.
Little variation of voltage sensitivity from the wild type (Kv2.1 rat) was predicted for most VKC mutants (Table 2). These mutants were shown to have little impact on voltage sensitivity if they were mutated to Ala in Kv2.1 [38]. Consistently, these mutants were predicted to have a V50 value close to that of the wild type (Table 2). One mutant, A154Y of Kv2.1 rat, displayed a large shift of V50 of over 10 mV [38], and our predictor also predicted a large positive shift in V50 value (Table 2). Although this is the largest margin between the predicted V50 and the experimental data, the correct prediction of direction in V50 shift by our predictor is encouraging.
Identification of biologically important residues (features)
One goal of building a model that can predict V50 value of a given VKC sequence with reasonable accuracy is to identify residues that are involved in modulating voltage sensitivity. In our analysis, different feature (residue) sets were selected by wrapper and screened to identify the feature set that yielded the best learning performance. In a forward selection approach, one feature (residue) was added at each round. Although different features were sequentially selected in different orders during the first five rounds, the feature set that produced the best learning performance converged to six residues. These six residues are the best features in predicting the V50 value of VKCs based on their amino acid sequences. The prediction was also validated by independent experimental data (Additional file 2 and 2). Likely, these residues are central to setting the voltage sensitivity of VKCs.
Some functionally important residues may not be identified using our approach. If a group of residues co-vary because they interact with each other to affect V50 values of VKCs, for example, after one residue is identified, the addition of the other residues may not further improve learning performances, and thus they would not be selected. However, no feature co-varied precisely with the six selected features in a covariation analysis (not shown). Also, our datasets contain a tiny subset of all the VKCs in nature that may not be an unbiased representation of all VKCs, so the residues that are selected may be only pertinent to these specific datasets. The quality of the experimental data are also a factor, indicated by the different V50 values obtained by different research labs for a same VKC [46-49]. Outlier selection may have helped alleviate the problem, but it is still a potential error source. Nevertheless, the combination of the selected residues should be a good indication of potentially functionally important structure elements.
The positions of the six features selected with a modified BLOSUM62 matrix are shown mapped onto a schematic of the VKC channel in the membrane in Figure 4A. All six selected residues are on one of the transmembrane helices S1, S2 or S3, none are found on the primary voltage sensor, S4, or on the pore/gate complex S5–S6. Although the S1–S3 region is not the primary voltage sensor, there is extensive experimental evidence that interactions between the S4 helix and the S1–S3 region are important in setting the value of V50. Tiwari-Woodruff et al. [58,59] and Papazian et al. [60] have demonstrated that three acidic residues in S2 and S3 (E283, E293 and D316) interact with basic residues in S4 during the process of voltage response, and that altering these residues in the D. melanogaster Shak channel will alter the value of V50 by more than 40 mV [61]. Those results suggest that one face of both S2 and S3 face the charged surface of S4 at least during the process of activation, and possibly constitutively. There is also a highly conserved acidic residue at the C-terminal limit of S1. Of the six residues that have been identified as relevant features in the current study, four are exactly two residues before or after one of these highly conserved acidic residues in the sequence. As illustrated in Figure 4B, this spacing along an alpha helix places the variable residues that we have identified approximately on the opposite face of the helix from the charged residues that have a demonstrable role in setting V50.
The variation of the six selected residues are mostly limited to nonpolar, hydrophobic residues including Ile, Leu, Val, Phe, and Ala, although there are several incidences of polar residues and one charged residue, the His at position 117 of the data in all Kv3 (Shaw) channels)[62] (Table 3). Thus, it appears that the observed variations in these six residues are responsible for many of the small energetic differences between channels that are responsible for differences in V50 values; whether these energy differences involve intramolecular interactions between several helices of a single channel protein or interactions between the helices of the protein and the surrounding hydrophobic layer of the lipid bilayer is unknown.
Many residues of VKCs that are responsible for voltage sensing and selective ion permeation are charged or polar amino acids, which generate relatively strong ionic interactions [40]. Variation of residues involved in strong ionic and bonding interactions often lead to, if not inactive channels, drastic variation in function [40]. Hydrophobic interactions among nonpolar residues or between nonpolar residues and lipids, on the other hand, are often energetically of lower magnitude and thus variations in these interactions will cause quantitatively small variations in functional parameters. These hydrophobic interactions can yield a practically continuous range of interaction energy magnitudes. Small variations in these kinds of residues are less likely to cause functional disruption, but are more likely to play "secondary" roles in VKC functioning and help tuning and shaping the sensitivity of different functional properties. The nonpolar hydrophobic features of the predicted voltage modulating residues are consistent with a role in modulating the targeted functional feature, the voltage sensitivity of VKCs.
The extant models for voltage sensing all focus on the movement of the voltage sensing S4 transmembrane helix and how it may alter the conformation of the pore and the S5 and S6 helices to open and close the ion channel. In most cases, the consequent movements of other parts of the molecule are not explicitly addressed. The "paddle" model based on the structure of the KvAp channel [11,63] predicts that the C-terminal half of the S3 helix moves in concert with the S4 helix during gating. However, both of the informative residues in the S3 helix are in the N-terminal half of S3 and so would not be directly involved in the primary gating movement.
The results of the current study indicate that as the interactions of S4 with the other transmembrane helices change, the interactions of those helices with each other or the lipid bilayer, or both, are also changing. These changes in interaction are in turn responsible for the small incremental differences in the V50 value for the channel. This analysis neither confirms nor rejects any of the current models of voltage response. It does predict that that when these models are formulated more precisely, the energy differences between the open and closed states must take into account hydrophobic interaction energies.
One possible confounding effect in a comparative study is the possibility of evolutionary hitchhiking, the possibility that the residue identities are correlated with functional variation by virtue of having been independently fixed in an ancestral population at the same point as a voltage variation evolved. However, if that were the case, then it would be expected that evolutionary reconstruction of the individual features would not show multiple independent origins of specific residues. This is not the case for the features selected in this analysis. Each of the six selected residue positions has evolved independently to specific residue identities (Table 3).
In a recent study pairs of residues that co-vary during evolution and are presumably involved in the essential functions of VKCs, were computationally identified. Most of these residues are located in the so-called core functional elements (S4–S6) [2], the pore region and the voltage sensor. Our approach to structure/function analysis is aimed at identifying structural elements that modulate the voltage sensitivity, not those that are essential for voltage sensitivity. While S4 is considered the main voltage-sensing unit, S1–S3 is thought to play a modulating role in the voltage sensitivity of VKCs [39,40]. Consistent with their modulating roles, all residues selected in our study are indeed located in S1–S3 region (Figure 3). It is not surprising that other residues in the S1–S3 region that were not selected in the present study have been shown to modulate the voltage sensitivity of VKCs [59,60]. Most of these highly conserved residues appear to interact directly with the positively charged key residues in the voltage sensor (S4 helix) through charge interactions. These residues are highly conserved among VKCs and so do not co-vary significantly with V50 values in most channels.
Conclusion
Machine learning methods have been widely used in biological analyses because of their capacity for dealing with data-rich tasks. Using a dataset of 58 VKC sequences with their V50 values, we built a predictor that predicts the V50 value of a given VKC based on its amino acid sequence. Despite the limited number of training data, and the uncertain quality, an MAE of prediction of 7.0 mV was obtained using a KNN classifier combined with a wrapper for feature selection (Figure 2D). The prediction accuracy was evaluated by a repeated (ten times) ten-fold cross validation. It is also validated by V50 prediction from independent experimental data (Table 2 and Additional file 2). As more data become available from ongoing isolation and characterization of novel VKCs, better prediction is expected. During training, four possible outliers were singled out and removed from the training set to improve the learning performance (Figure 3). Several residues with potential biological implications were identified for further study (Figure 4).
The analysis presented in this report demonstrates how machine learning methods can be productively applied to structure-functional study with datasets of limited size. These analyses can predict certain biological functions with a reasonable accuracy and can identify potentially functionally important residues for experimental testing of specific hypotheses of the structure/function relationship in a family of proteins.
Methods
Dataset
Data used in this project were drawn from VKCDB, a voltage-gated potassium channel database [34]. Although VKCDB has over 350 channel sequence entries, only 58 VKC sequences have associated half activation voltage (V50) values. Sequence and V50 values for these channels were extracted from VKCDB; most of the sequences have more than 500 amino acid residues. The published V50 values have been experimentally determined under similar experimental conditions, using a two-electrode voltage clamp in Xenopus oocytes [24]. Averages were used for those VKCs for which different V50 values have been published by different groups [46-49].
All sequences were aligned with PepTool [64], followed by manual adjustment. Because there is large sequence variation at both termini and some loop regions of the VKCs, only blocks of residues that contained relatively few gaps were kept for analysis (Dataset 1, see Additional file 1).
Independent test dataset
Thirteen wild type VKCs with experimentally determined V50 values were used to obtain an objective assessment of the predictor (see Additional File 1). The V50 values of these VKCs had been determined in several different cell hosts including Xenopus oocytes, HEK293 cells, and CHO cells [34-36]. Another six VKC mutants with V50 values determined in an Ala-scanning mutagenesis experiment [38] were also used to evaluate the predictor.
Problem formulation
To formulate our problem into a typical supervised learning task, the dataset was considered as a training set with 58 instances. Each of the alignment positions was taken as one nominal attribute (feature), and all attributes were assumed to be independent of each other. In numerical prediction analyses, the classes were the real V50 numerical values. In categorical prediction analyses, V50 values were divided into seven nominal classes based on their values; -50 > V50 ≥ -30 mV, -30 > V50 ≥ -20 mV, -20 > V50 ≥ -10 mV, -10 > V50 ≥ 0 mV, 0 > V50 ≥ 10 mV, 10 > V50 ≥ 20 mV and 20 > V50 ≥ 65 mV. The goal is to extract the data model that can best describe the relationship between the (attributes) features and the labeled classes of these data, and correctly predict the class or the numerical value of V50 of any given VKC sequence.
Basic learning algorithms
The KNN (k-nearest neighbor) classifier was used in both numerical prediction and categorical prediction analysis. All KNN classifications were tested with k values of 1 to 5. Decision Tree, Naïve Bayes Learner, Kernel Density Classifier and OneR Classifier algorithms were also used in categorical predictions. The algorithms used are implemented in the WEKA package 3.2.3 [30].
The prediction accuracies were used to evaluate the learning performance in categorical prediction. The mean absolute errors (MAEs), the average absolute difference between the predicted values and the published values, were used to assess the numerical prediction. All learning performances were evaluated using a repeated ten times ten-fold cross validation.
Feature selection
Filtering and wrapper algorithms were used to select a subset of features with the best prediction performance, to decrease the dimensionality of the learning problem.
For filtering, features were ranked by information gain [32,33,65], then different numbers of top-ranked features were selected for learning, and the sets that produced the best learning performance were considered the best feature sets using filter.
The wrapper approach to feature selection screens subsets of features in a dataset and selects the "relevant" features based on learning performances [31]. Forward selection was used in this approach. In the first round of analysis each individual residue was evaluated for its prediction quality (lowest MAE in 10 times 10-fold cross validation) and the 200 residues with best predictive performance were retained. In subsequent rounds additional features (residues) were added at each round, the prediction quality of the resulting subsets of residues were evaluated, and the 200 subsets with best prediction quality were retained for the next round of feature addition. This process was repeated until learning performances stopped improving [31]. Despite the existence of redundant feature sets at each round, the number of non-redundant feature sets was well above 100 at each round. The search was continued for five rounds after the learning performance stopped improving to ensure that performance had reached a plateau.
Residue swapping
We also applied a "residue swap" heuristic, similar to the branch-swapping step during the construction of phylogenetic trees [45], to try to further improve the prediction accuracy. For the best feature set selected by the wrapper algorithm, each residue was sequentially replaced with every other residue that was not in the final set, and the new feature combination was evaluated for prediction accuracy using a repeated ten-fold cross validation.
Distance matrices in k-nearest neighbor classification (KNN)
A KNN classifier is a set of n-dimensional vectors (where n = the number of features) to which new instances are compared [25]. It classifies a new instance by evaluating its distance from each of the classifier instances and chooses the class label of the classifier instance that is closest to the new instance as the predicted class of the new instance. For more than one classifier instance with an identical distance to the new instance, one of the class labels of these classifier instances is randomly picked and assigned in categorical predictions; averages of equidistant classifier instances are calculated for numerical prediction.
The Euclidean distance between any two vectors is obtained by taking the sum of the square of the distances between all pairs of attributes (dimensions), on the assumption that the sites are independent and therefore their dimensions are orthogonal. For nominal attributes, such as amino acid residues, the KNN algorithm can simply takes 1 and 0 as the distance between a pair of different and same residues, respectively. We also implemented the KNN algorithm to incorporate PAM [66] and BLOSUM [67] matrices as a measure of distance between pairs of features (residues) of two VKC sequences (Formulas 1.1 and 1.2). Since the scores in amino acid comparison matrices go up when two amino acid residues are more similar to each other, which is the opposite of distance measurement in KNN classification, we converted amino acid comparison scores accordingly (Formula 1.0). In all cases any gap was scored as the maximum distance for the relevant scoring matrix.
In BLOSUM62 or PAM100 matrix:
converted scorei = matrix_maximum_value - original_scorei (1.0)
D: Distance between two instances.
n: Number of features.
Other matrices: scorei = converted scorei from pairwise comparison (1.2)
Outlier selection
To minimize the effect of possible outliers, a best-first search was performed. One VKC sequence was deleted from the training set at each round, and the learning was carried out with the remaining VKC sequences. The deleted sequence was considered an outlier if the remaining dataset yielded better learning performance than the full dataset. The search stopped if the learning performance no longer improved after a further round of deletion. Due to computational complexity, the outlier selection was not combined with full feature selection of wrapper [31]. Instead, the best feature set selected by the wrapper algorithm was applied to outlier selection.
Permutation test
We randomly shuffled the classes of each instance in Data Set 2 (the 54 sequences remaining after outlier selection) to produce 100 permuted datasets. With each of these datasets, training was repeated using KNN classification combined with wrapper, with identical parameters and settings as in the original training. This process was repeated one hundred times and the MAE of each of the predictors was collected.
Final predictor construction
After removal of four outliers from the original dataset, the remaining 54 sequences formed a new dataset (Dataset 2) that was used to develop the final predictor. During the training process using Dataset 2, one best feature (residue) set was selected by wrapper to predict the V50 values with an MAE of 7.0 mV. One predictor was then constructed, using Dataset 2, the best feature set, the BLOSUM 62 scoring matrix and the KNN classification (k = 1).
To predict the V50 value of a new query sequence, the query sequence is first aligned with the profile alignment of Dataset 2 using ClustalW [68]. The residues at the aligned selected positions are extracted to produce a data file for V50 prediction.
Phylogenetic reconstruction and distance tree construction
The training data were used to construct a phylogenetic tree using MrBayes v. 3 [41]. Reconstruction of the evolution of the character states in the six selected features was then evaluated on this tree using maximum parsimony as implemented in MacClade [42].
The relative distances of the 54 VKCs in the training set (Dataset 2) and the 13 independently characterized test sequences were evaluated by summing the distances for each of the six selected features. A distance tree was then constructed using PAUP*4 [69] (Figure 5). The names of the 54 sequences in the training set are not shown, for clarity. The branches connecting the 13 test sequences to the tree formed by the training set are highlighted in red, and the names of those 13 sequences are shown.
Authors' contributions
BL did the data collection and performed the computational analyses. WJG initiated the project, supervised the execution of the analyses and provided regular evaluation of the outcomes. Both of authors shared in the preparation of the manuscript.
Supplementary Material
Additional File 1
TrainingandTest.nexus is a plain text file in NEXUS format, containing the sequence data matrix that was used for the machine learning analysis and the 13 channel sequences that were used for validation of the predictor performance.
Click here for file
Additional File 2
Table 2.
Click here for file
Acknowledgements
Authors would like to thank Dr. Andrew Spencer, Dr. Russ Greiner, and Dr. David Wishart for their comments on the manuscript. This study was supported by a grant from CIHR (MOP-62685) to Andrew N. Spencer and WJG.
Figures and Tables
Figure 1 Flow chart of procedures followed to develop the optimal V50 predictor. The data set was subjected to several different learning algorithms, either alone or in combination with two types of feature selection. The KNN learning algorithm and the wrapper feature selection algorithm (highlighted in red) were found to yield the best results. These algorithms were then used to evaluate the effect of removing sequences as outliers to yield the data set used for construction of the final predictor. The individual processes that were used to construct the final predictor are highlighted in red.
Figure 2 Learning performances with different algorithms and different implementations. All bars represent results of a repeated (ten times) ten fold cross validation. A: Categorical learning with different learning algorithms without feature selection. The V50 values were divided into seven classes based on their values. The learning was done without feature selection. B: Improvement of KNN prediction accuracies in different implementations. Results of KNN classification without feature selection, with the filter algorithm, with the wrapper algorithm, and with outlier selection in combination with the wrapper algorithm are shown. Both feature selection algorithms improved learning performance. The best learning accuracy was obtained using the KNN classifier combined with wrapper and removal of four outlier sequences. It yields a mean absolute errors of 7.0 mV with the new dataset (Dataset 2) of 54 VKC sequences.
Figure 3 Schematic of the process of outlier selection, and the variations in MAEs during outlier selection using KNN classifier. A: Each sequence of Dataset 1 was individually deleted to select the resulting datasets that produce improved learning performances. The top 50 new subsets were kept at each round, and individual deletions were repeated. Due to computational complexity, the best feature set selected by wrapper as described in the paper was used in training. B: Variation of learning performance using KNN classifier during outlier selection. The mean absolute errors of prediction improved with selective removal of putative outlier instances. There was a significant improvement of learning accuracies at Round 2 and 4 (thick lines). After Round 4, the improvement of learning performances slowed down significantly.
Figure 4 Residues selected as informative features by wrapper and mapped onto VKC structures. A. Schematic structure of the VKCs. Six transmembranehelices, labeled S1 through S6, traverse the plasma membrane. The N-terminus and C-terminus are both cytoplasmic. The loop connecting S5 and S6 folds into the channel to form the ion-selective pore. The S4 region, in which every third residue is positively charged, is the central voltage-sensing element of the VKC. The six residues selected by the wrapper algorithm are approximately mapped onto the schematic diagram (black dots). Four highly conserved acidic residues are indicated by black stars. The numbers assigned to each marked residue are positions within the dataset alignment, not positions in any actual VKC protein sequence.B. Schematic end-on view of helices S1, S2 and S3 with the radial positions of conserved acidic residues (numbered in black) and four of the selected residues (numbered in red) indicated. Residue 117, 125, 135 and 154 are each exactly two residues away from a highly conserved negatively charged residue, 115, 127, 137 and 152 respectively.
Figure 5 Distance tree of training data and independent test data. Distance tree illustrating the relative similarities of the 54 sets of six selected informative residues from the training set and the corresponding residues in the 13 sequences that were used for independent testing of the prediction performance. The names of the training set channels have been removed for clarity. Each of the four subfamilies form well supported clades, and are labeled as Kv1, Kv2, Kv3, and Kv4. The branches and names of the wild type VKC channels from the independent test set are in red. Most test VKC sequences are clustered closely with one or more channels from the training data. Two VKCs from Hirudo medicinalis (LeechKv1) and Ciona intestinalis (CionaKv4), have long branch lengths.
Figure 6 Bayesian maximum likelihood tree of training data and independent test data. The set of 54 VKC sequences used for the final predictor and the 13 VKC sequences from the independent test set were aligned using MUSCLE [70] and the alignment was trimmed to 296 residues. This data set was used to generate a maximum likelihood tree using MrBayes [41], using a total of 500,000 cycles with a 250 cycle burn-in. The training set channels are labelled with their VKCDB identification number and their V50 value, the 13 test sequences are labelled with the name of the species and the channel, and were highlighted in red (Figure 6).
Table 1 "Best" feature (residue) sets selected by wrapper with different distance matrices.
Distance matrices Selected residue sets
BLOSUM62 97, 100, 117, 125, 135, 154
PAM100 83, 95, 97, 100, 117, 131, 141, 154
Identity matrix 83, 92, 95, 100, 103, 123, 135, 154, 273
Wrapper feature selection identified different, but overlapping, sets of informative residues, depending on the distance matrix that was used. Residues that were selected with more than one matrix are underlid, and residue 100 and 154 (in bold) were selected with all three distance matrices.
Table 2 Published mutant data and predicted V50 values.
Ala scan (Kv2.1) Published V50 (mV) Predicted V50 (mV) Wild type (mV)
L97A 0.6 -7.23 -4.9
*I100A -7.3 -7.23 -4.9
L117A -1.6 -7.23 -4.9
*V125A -4.6 -7.23 -4.9
L135A 1.5 -7.23 -4.9
A154Y 7 27.5 -4.9
MAE 7.5
The V50 predictions for a set of published alanine scanning mutants are compared to the published measured V50 values. The residues marked with * are residues in which an alanine exists at this position in at least one of the channels in the training dataset.
Table 3 Amino acid residues at the six positions selected as most informative in our dataset.
Position Residue (number of independent origins)
97 *C (1), F (2), I (4), L (A), V (5), *Y (1)
100 A (1), *F (1), *G (1), I (A), *L (1), *T (1), V (4)
117 H (1), I (4), L (A), V (1)
125 A (A), *C (1), *F (1), I (3), L (2), T (1), V (3)
135 I (4), L (A), *T (1), V (2)
154 A (1), *C (1), *F (1), I (A), L (2), *M (1), V (5)
The amino acid residues that are present at each of the six informative positions of the 56 channels used for construction of the final predictor are presented. Residues marked with an asterisk only occurred once at that position in the training data set. The number in brackets represents the number of independent origins of that residue, that is the minimum number of times that the residue evolved independently at that position. An A in brackets indicates that this is the inferred ancestral residue identity.
==== Refs
Li B Gallin W VKCDB
Fleishman SJ Yifrach O Ben-Tal N An evolutionarily conserved network of amino acids mediates gating in voltage-dependent potassium channels J Mol Biol 2004 340 307 318 15201054 10.1016/j.jmb.2004.04.064
Yellen G The voltage-gated potassium channels and their relatives Nature 2002 419 35 42 12214225 10.1038/nature00978
Jentsch TJ Neuronal KCNQ potassium channels: physiology and role in disease Nat Rev Neurosci 2000 1 21 30 11252765 10.1038/35036198
Comu S Giuliani M Narayanan V Episodic ataxia and myokymia syndrome: a new mutation of potassium channel gene Kv1.1 Ann Neurol 1996 40 684 687 8871592 10.1002/ana.410400422
Abdul M Hoosein N Voltage-gated potassium ion channels in colon cancer Oncol Rep 2002 9 961 964 12168055
Koni PA Khanna R Chang MC Tang MD Kaczmarek LK Schlichter LC Flavella RA Compensatory anion currents in Kv1.3 channel-deficient thymocytes Journal of Biological Chemistry 2003 278 39443 39451 12878608 10.1074/jbc.M304879200
Cooper EC Potassium channels: how genetic studies of epileptic syndromes open paths to new therapeutic targets and drugs Epilepsia 2001 42 Suppl 5 49 54 11887968 10.1046/j.1528-1157.2001.0420s5049.x
Ford JW Stevens EB Treherne JM Packer J Bushfield M Potassium channels: gene family, therapeutic relevance, high-throughput screening technologies and drug discovery Progress in Drug Research 2002 58 133 168 12079199
Lerche H Jurkat-Rott K Lehmann-Horn F Ion channels and epilepsy American Journal of Medical Genetics 2001 106 146 159 11579435 10.1002/ajmg.1582
Jiang Y Lee A Chen J Ruta V Cadene M Chait BT MacKinnon R X-ray structure of a voltage-dependent K+ channel Nature 2003 423 33 41 12721618 10.1038/nature01580
Larsson HP Baker OS Dhillon DS Isacoff EY Transmembrane movement of the shaker K+ channel S4 Neuron 1996 16 387 397 8789953 10.1016/S0896-6273(00)80056-2
Sands Z Grottesi A Sansom MS Voltage-gated ion channels Curr Biol 2005 15 R44 7 15668152 10.1016/j.cub.2004.12.050
Bixby KA Nanao MH Shen NV Kreusch A Bellamy H Pfaffinger PJ Choe S Zn2+-binding and molecular determinants of tetramerization in voltage-gated K+ channels Nature Structural Biology 1999 6 38 43 9886290 10.1038/4911
Doyle DA Cabral JM Pfuetzner RA Kuo A Gulbis JM Cohen SL Chait BT MacKinnon R The structure of the potassium channel: molecular basis of K+ conduction and selectivity Science 1998 280 69 77 9525859 10.1126/science.280.5360.69
Jiang Y Lee A Chen J Cadene M Chait BT MacKinnon R The open pore conformation of potassium channels Nature 2002 417 523 526 12037560 10.1038/417523a
Kuo A Gulbis JM Antcliff JF Rahman T Lowe ED Zimmer J Cuthbertson J Ashcroft FM Ezaki T Doyle DA Crystal structure of the potassium channel KirBac1.1 in the closed state Science 2003 300 1922 1926 12738871 10.1126/science.1085028
Sokolova O Kolmakova-Partensky L Grigorieff N Three-dimensional structure of a voltage-gated potassium channel at 2.5 nm resolution Structure (Camb) 2001 9 215 220 11286888 10.1016/S0969-2126(01)00578-0
Jiang Y Lee A Chen J Cadene M Chait BT MacKinnon R Crystal structure and mechanism of a calcium-gated potassium channel Nature 2002 417 515 522 12037559 10.1038/417515a
Laine M Papazian DM Roux B Critical assessment of a proposed model of Shaker FEBS Lett 2004 564 257 263 15111106 10.1016/S0014-5793(04)00273-X
Heginbotham L Abramson T MacKinnon R A functional connection between the pores of distantly related ion channels as revealed by mutant K+ channels Science 1992 258 1152 1155 1279807
MacKinnon R New insights into the structure and function of potassium channels Current Opinion in Neurobiology 1991 1 14 19 1822292 10.1016/0959-4388(91)90005-R
Miller C 1990: annus mirabilis of potassium channels Science 1991 252 1092 1096 2031183
Hille B Ionic Channels of Excitable Membranes 2001 3rd Sunderland, Mass. , Sinauer Associates Inc. 814
Mitchell TM Machine learning 1997 New York, NY , McGraw-Hill xvii, 414 p.
Bose I Mahapatra RK Business data mining - a machine learning perspective Inform Manage 2001 39 211 225 10.1016/S0378-7206(01)00091-X
Hayes WS Borodovsky M How to interpret an anonymous bacterial genome: machine learning approach to gene identification Genome Research 1998 8 1154 1171 9847079
Tag PM Peak JE Machine learning of maritime fog forecast rules J Appl Meteorol 1996 35 714 724 10.1175/1520-0450(1996)035<0714:MLOMFF>2.0.CO;2
Ringner M Peterson C Microarray-based cancer diagnosis with artificial neural networks Biotechniques 2003 Suppl 30 35 12664682
Witten IH Frank E Data mining : practical machine learning tools and techniques with Java implementations The Morgan Kaufmann series in data management systems 2000 San Francisco, CA , Morgan Kaufmann xxv, 371 p.
Kohavi R John GH Wrappers for feature subset detection Artificial Intelligence 1997 97 273 324 10.1016/S0004-3702(97)00043-X
Almuallim H Dietterich TG Learning with many irrelevant features.: Anaheim, CA.
1991 MIT Press 547 552
Kira K Rendell LA The feature selection problem: Traditional methods and a new algorithm.: San Jose, CA.
1992 MIT Press 129 134
Li B Gallin WJ VKCDB: voltage-gated potassium channel database BMC Bioinformatics 2004 5 3 14715090 10.1186/1471-2105-5-3
Fry M Maue RA Moody-Corbett F Properties of Xenopus Kv1.10 channels expressed in HEK293 cells J Neurobiol 2004 60 227 235 15266653 10.1002/neu.20024
Salvador-Recatala V Gallin WJ Abbruzzese J Ruben PC Spencer AN The Structure and Function of a Kv4-like Potassium Channel Expressed in the Myocardium of the Tunicate, Ciona intestinalis Submitted 2004
Weiss JL Yang J Jie C Walker DL Ahmed S Zhu Y Huang Y Johansen KM Johansen J Molecular cloning and characterization of LKv1, a novel voltage-gated potassium channel in leech J Neurobiol 1999 38 287 299 10022573 10.1002/(SICI)1097-4695(19990205)38:2<287::AID-NEU10>3.0.CO;2-U
Li-Smerin Y Hackos DH Swartz KJ alpha-helical structural elements within the voltage-sensing domains of a K(+) channel Journal of General Physiology 2000 115 33 50 10613917 10.1085/jgp.115.1.33
Treptow W Maigret B Chipot C Tarek M Coupled Motions between Pore and Voltage-Sensor Domains: A Model for Shaker B, a Voltage-Gated Potassium Channel Biophys J 2004 87 2365 2379 15454436 10.1529/biophysj.104.039628
Yellen G The moving parts of voltage-gated ion channels Q Rev Biophys 1998 31 239 295 10384687 10.1017/S0033583598003448
Ronquist F Huelsenbeck JP MrBayes 3: Bayesian phylogenetic inference under mixed models Bioinformatics 2003 19 1572 1574 12912839 10.1093/bioinformatics/btg180
Maddison DR Maddison WP MacClade 4: Analysis of phylogeny and character evolution 2003 4.06 Sinauer Associates, Sunderland, Massachusetts
Mendez MA Hodar C Vulpe C Gonzalez M Cambiazo V Discriminant analysis to evaluate clustering of gene expression data FEBS Letters 2002 522 24 28 12095613 10.1016/S0014-5793(02)02873-9
Nguyen DV Rocke DM Tumor classification by partial least squares using microarray gene expression data Bioinformatics 2002 18 39 50 11836210 10.1093/bioinformatics/18.1.39
Adams ENIII Consensus techniques and the comparison of taxonomic trees. Systematic Zoology 1972 21 390 397
Rettig J Wunder F Stocker M Lichtinghagen R Mastiaux F Beckh S Kues W Pedarzani P Schroter KH Ruppersberg JP Characterization of a Shaw-related potassium channel family in rat brain EMBO Journal 1992 11 2473 2486 1378392
Scholle A Koopmann R Leicher T Ludwig J Pongs O Benndorf K Structural elements determining activation kinetics in Kv2.1 Receptors Channels 2000 7 65 75 10800777
Schroter KH Ruppersberg JP Wunder F Rettig J Stocker M Pongs O Cloning and functional expression of a TEA-sensitive A-type potassium channel from rat brain FEBS Letters 1991 278 211 216 1840526 10.1016/0014-5793(91)80119-N
Stuhmer W Conti F Suzuki H Wang X Noda M Yahagi N Kubo H Numa S Structural parts involved in activation and inactivation of the sodium channel Nature 1989 339 597 603 2543931 10.1038/339597a0
Holmqvist MH Cao J Hernandez-Pineda R Jacobson MD Carroll KI Sung MA Betty M Ge P Gilbride KJ Brown ME Jurman ME Lawson D Silos-Santiago I Xie Y Covarrubias M Rhodes KJ Distefano PS An WF Elimination of fast inactivation in Kv4 A-type potassium channels by an auxiliary subunit domain Proceedings of the National Academy of Sciences of the United States of America 2002 99 1035 1040 11805342 10.1073/pnas.022509299
Overturf KE Russell SN Carl A Vogalis F Hart PJ Hume JR Sanders KM Horowitz B Cloning and characterization of a Kv1.5 delayed rectifier K+ channel from vascular and visceral smooth muscles American Journal of Physiology 1994 267 C1231 8 7977686
Patton DE Silva T Bezanilla F RNA editing generates a diverse array of transcripts encoding squid Kv2 K+ channels with altered functional properties Neuron 1997 19 711 722 9331360 10.1016/S0896-6273(00)80383-9
de Lichtenberg U Jensen LJ Fausboll A Jensen TS Bork P Brunak S Comparison of computational methods for the identification of cell cycle regulated genes Bioinformatics 2005 21 1164 1171 15513999 10.1093/bioinformatics/bti093
Listgarten J Damaraju S Poulin B Cook L Dufour J Driga A Mackey J Wishart D Greiner R Zanke B Predictive models for breast cancer susceptibility from multiple single nucleotide polymorphisms Clin Cancer Res 2004 10 2725 2737 15102677
Potter DM A permutation test for inference in logistic regression with small- and moderate-sized data sets Stat Med 2004
Routledge RD P-values from permutation and F-tests Comput Stat Data An Comput Stat Data An 1997 24 379 386 10.1016/S0167-9473(96)00073-4
Rae JL Shepard AR Kv3.3 potassium channels in lens epithelium and corneal endothelium Exp Eye Res 2000 70 339 348 10712820 10.1006/exer.1999.0796
Tiwari-Woodruff SK Lin MA Schulteis CT Papazian DM Voltage-dependent structural interactions in the Shaker K(+) channel J Gen Physiol 2000 115 123 138 10653892 10.1085/jgp.115.2.123
Tiwari-Woodruff SK Schulteis CT Mock AF Papazian DM Electrostatic interactions between transmembrane segments mediate folding of Shaker K+ channel subunits Biophys J 1997 72 1489 1500 9083655
Papazian DM Shao XM Seoh SA Mock AF Huang Y Wainstock DH Electrostatic interactions of S4 voltage sensor in Shaker K+ channel Neuron 1995 14 1293 1301 7605638 10.1016/0896-6273(95)90276-7
Seoh SA Sigg D Papazian DM Bezanilla F Voltage-sensing residues in the S2 and S4 segments of the Shaker K+ channel Neuron 1996 16 1159 1167 8663992 10.1016/S0896-6273(00)80142-7
Jan LY Jan YN Voltage-gated and inwardly rectifying potassium channels J Physiol 1997 505 ( Pt 2) 267 282 9423171 10.1111/j.1469-7793.1997.267bb.x
Jiang Y Ruta V Chen J Lee A MacKinnon R The principle of gating charge movement in a voltage-dependent K+ channel Nature 2003 423 42 48 12721619 10.1038/nature01581
Wishart DS Boyko RF Sykes BD Constrained multiple sequence alignment using XALIGN Computer Applications in the Biosciences 1994 10 687 688 7704670
Cardie C Using decision trees to improve case-based learning.: San Meteo, CA.
1993 Morgan Kaufmann Publishers, Inc. 25 32
Schwartz RM Dayhoff MO Dayhoff MO Matrices for detecting distant relationships. Atlas of Protein Sequence and Structure 1978 5, suppl. 3 Washington, D.C. , Natl. Biolmed. Res. Found. 345 352
Henikoff S Henikoff JG Amino acid substitution matrices from protein blocks Proceedings of the National Academy of Sciences of the United States of America 1992 89 10915 10919 1438297
Higgins DG Thompson JD Gibson TJ Using CLUSTAL for multiple sequence alignments Methods in Enzymology 1996 266 383 402 8743695
Swofford DL PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). 2000 4 Sunderland, Massachusetts , Sinauer Associates
Edgar RC MUSCLE: a multiple sequence alignment method with reduced time and space complexity BMC Bioinformatics 2004 5 113 15318951 10.1186/1471-2105-5-113
|
16111489
|
PMC1208917
|
CC BY
|
2021-01-04 16:03:52
|
no
|
BMC Struct Biol. 2005 Aug 19; 5:16
|
utf-8
|
BMC Struct Biol
| 2,005 |
10.1186/1472-6807-5-16
|
oa_comm
|
==== Front
BMC SurgBMC Surgery1471-2482BioMed Central London 1471-2482-5-181610518310.1186/1471-2482-5-18Study ProtocolHEGPOL: Randomized, placebo controlled, multicenter, double-blind clinical trial to investigate hepatoprotective effects of glycine in the postoperative phase of liver transplantation [ISRCTN69350312] Luntz Steffen P [email protected] Kristina [email protected] Monika [email protected] Hartwig [email protected] Thomas W [email protected]üchler Markus W [email protected] Ernst [email protected] Peter [email protected] Coordination Centre for Clinical Trials (KKS), University of Heidelberg, Germany2 Department of Surgery, University of Heidelberg, Germany3 Department of Surgery, University of North Carolina (UNC) at Chapel Hill, USA4 Department of Surgery, University of Rostock, Germany2005 17 8 2005 5 18 18 4 7 2005 17 8 2005 Copyright © 2005 Luntz et al; licensee BioMed Central Ltd.2005Luntz et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Kupffer cell-dependent ischemia / reperfusion (I/R) injury after liver transplantation is still of high clinical relevance, as it is strongly associated with primary dysfunction and primary nonfunction of the graft. Glycine, a non-toxic, non-essential amino acid has been conclusively shown in various experiments to prevent both activation of Kupffer cells and reperfusion injury. Based on both experimental and preliminary clinical data this study protocol was designed to further evaluate the early effect of glycine after liver transplantation.
Methods / design
A prospective double-blinded randomized placebo-controlled multicenter study with two parallel groups in a total of 130 liver transplant recipients was designed to assess the effect of multiple intravenous doses of glycine after transplantation. Primary endpoints in hierarchical order are: peak levels of both aspartat-amino-transaminase (AST) and alanine-amino-transaminase (ALT) as surrogates for the progression of liver related injury, as well as both graft and patient survival up to 2 years after transplantation. Furthermore, the effect of glycine on cyclosporine A-induced nephrotoxicity is evaluated.
Discussion
The ongoing clinical trial represents an advanced element of the research chain, along which a scientific hypothesis has to go by, in order to reach the highest level of evidence; a randomized, prospective, controlled double-blinded clinical trial. If the data of this ongoing research project confirm prior findings, glycine would improve the general outcome after liver transplantation.
==== Body
Background
Pathophysiology of liver grafts
The cause of graft failure after transplantation is complex and includes many factors involving organ retrieval, preservation, and the transplantation procedure itself. Important factors include general condition and nutritional status of the donor, cold and warm ischemic times of the graft, operative complications in the recipient, immune status of the recipient and the experience of the surgeon [1,2]. Thus, primary non function (PNF), which is initially determined by reperfusion injury, continues to challenge liver transplantation.
Once activated, Kupffer cells, the resident macrophages in the liver, play a pivotal role for the development of both PNF and primary dysfunction (PDF) [2,3]. Kupffer cell-activation is characterized by an intracellular increase of Ca2+ [4] with a subsequent release of toxic mediators such as proteases, tumor necrosis factor alpha (TNFα), and arachidonic acid derivatives [5,6]. These mediators potentially impair liver function via mechanisms including disturbance of the intrahepatic microcirculation, hypoxia, increased oxygen consumption, and depletion of hepatic glycogen reserves [2].
Fusaoka et al. [7] showed that activated Kupffer cells increase oxygen uptake of the liver after cold storage. This effect is most likely due to Kupffer cell-derived prostaglandine E2 (PGE2), which stimulates oxygen uptake in hepatocytes and could be involved in early dysfunction of the graft [8]. Indeed, graft survival is impaired after liver transplantation most likely via mechanisms including both hypoxia and increased oxygen consumption of hepatocytes, creating a hypermetabolic state [8]. Activation of Kupffer cells occurs early during organ harvest for transplantation due to in situ organ manipulation, which is inevitable with standard harvesting techniques [1,8,9].
Glycine
Glycine, a non-toxic, non-essential amino acid is important for the synthesis of many proteins, i.e. creatinine, uric acid, and heme. Under physiological conditions, blood levels of glycine range between 200–400 μmol/L in humans [10,11].
Clinical use of glycine
To date various indications for supplementation with glycine have been established, i.e. for total parenteral nutrition, local irrigation during transurethral prostate or urine bladder resections, being a hypotonic solution and having the capacity as an antacidotic agent.
Cytoprotective effects of glycine
Addition of amino acids during renal perfusion can protect tubular integrity and can prolong renal function [12]. Weinberg et al. were the first to connect this protective effect with the amino acid glycine [13]. Glycine protects tissue against damage via mechanisims involving proinflammatory mediators, hypoxia reduction, reperfusion enhancement and toxin attenuation in various animal species [10-15].
Glycine inhibits nonlysosomal calcium-dependent proteases and protects hepatocytes against anoxic damage. Ozaki et al. demonstrated that glycine could protect livers in situ from reperfusion damage by minimizing lipid peroxidation [16]. Glycine could stabilize the cell membrane by inhibiting phospholipase A2 leading to a reduction of arachidonic acid and eicosanoids which influence hepatic microcirculation [17]. Carolina rinse solution which contains glycine, prevents reperfusion injury to livers in both experimental and human liver transplantation [18]. Intravenous glycine application also prevents Kupffer cell-dependent reperfusion injury in rats [9,19].
Kupffer cells
Most recently, a glycine-gated chloride channel (GlyR) has been identified within the membranes of Kupffer cells [10,11,20]. Glycine specifically binds to its receptor. Subsequently, chloride ions enter the cell resulting in the hyperpolarization of the cell membrane and making a Ca2+ influx via voltage dependent Ca2+-channels more difficult, which effectively reduces the increase of intracellular Ca2+ [4,9,14,20-22]. As a result, glycine reliably prevents Kupffer cell-dependent reperfusion injury and initial dysfunction of grafts after experimental liver transplantation [1,2,9].
Calcineurin inhibitor (CNI)-induced nephrotoxicity
Cyclosporin A (CyA), a calcineurin inhibitor, is widely used as an immunosuppressive agent. Since its introduction in solid organ transplantation, CyA has significantly improved the overall graft survival; however, patients have to maintain therapy for the rest of their lives. Further, this drug is used to treat a variety of autoimmune diseases. Unfortunately, one of the typical side effects of CNIs is dose dependent nephrotoxicity, which is of clinical relevance in up to 30% of patients [23]. Underlying mechanisms most likely include the CyA significant inhibition of respiration in mitochondria isolated from the kidney [24], and thus causing cell damage. Further, CNI-induced nephrotoxicity is characterized by vasoconstriction in kidneys [25], reduced glomerular filtration rate (GFR), and pathological changes such as proximal tubular cell swelling, necrosis, infiltration of macrophages, and interstitial fibrosis [23]. These changes potentially lead to hypoxia-reoxygenation injury involving free radicals. Indeed, a previous study showed that binding of a 2-nitroimidazole hypoxia marker, pimonidazole, in the kidney was increased nearly 3-fold by CyA, indicating marked tissue hypoxia [26].
Most recently, glycine prevented hypoxic and ischemic injury to kidney in rats [26]. This can be explained by the fact that glycine acts as a neurotransmitter with inhibitory effects to the autonomous nervous system, i.e. sympathic nerves [27], decreasing renal nerve firing. This mechanism prevents injury due to CyA [28]. As a result, glycine dilates efferent arterioles and protects cultured proximal tubules from hypoxic injury [29]. Moreover, dietary glycine totally blocked CyA-induced alterations in renal function, such as decreased GFR and pathological changes including cell necrosis and infiltration of macrophages [26,30,31].
Study rationale
Both experimental studies and clinical trials [8,9,32-34] have shown that glycine is safe for use in patients [21,22,35-37] and would potentially be beneficial for the treatment of various diseases [10,11]. There is preliminary evidence for beneficial effects of glycine in human liver transplantation [21,22]. However, there is still a lack of solid clinical evidence for these effects of glycine in liver transplantation. The first clinical results with intravenous glycine are very promising [21,22]. Thus to date there is no routine indication for glycine application to liver transplant recipients and no intravenous infusion containing exclusively glycine is commercially available yet.
Based on the beneficial effects of glycine on both liver grafts and kidney function during CyA therapy [26,30,31,38] and its potential application in humans without toxicity, this clinical trial protocol was designed to assess the effects of intravenous glycine in a prospective, double-blind, randomized, placebo-controlled, multicenter study with two parallel groups of liver transplant recipients for the first time in detail.
Methods / design
After the positive vote of the ethics committee of all involved study sites, enrollment of 130 subjects scheduled for liver transplantation was started in this multicenter, prospective, placebo-controlled, double-blind, randomized clinical trial with two parallel treatment groups (verum / placebo). Only patients who meet all inclusion and exclusion criteria (Table 1) are considered for uptake into the study. Since its initiation in May 2003, about 69 patients have been included until June 2005.
Table 1 Criteria of inclusion and exclusion of patients.
Inclusion Criteria Exclusion Criteria
Patients meeting all of the following criteria are considered for inclusion in the study: Patients presenting with any of the following are not be included in the trial:
- men and women between 18 and 65 years of age, - pregnant or nursing women,
- scheduled for first liver transplantation and graft (dead body) already available, - history of hypersensitivity to glycine or to drugs with a similar chemical structure (amino acids, e.g. serine, threonine, or methionine),
- written informed consent. - mental conditions rendering the subject incapable to understand the nature, scope, and consequences of the trial,
- participation in another clinical trial.
No subject will be enrolled in this study more than once.
Patients are treated for a total of 8 consecutive days (day of transplantation and the seven following days). Subsequently, all patients are observed for an additional follow-up period of at least 23 days. Thus every patient is observed for at least one month (Figure 1). In detail, to assess the incidence of a late onset of graft failure based on patients' death or patients' announcement for re-transplantation all patients will be followed-up until one month after the last patient is randomized for this trial. Patients' status will be obtained via phone interview of the patients' general practitioner and / or the responsible transplant consulting office. The Schedule for all study related activities and data collection is listed in Table 2.
Figure 1 Scheme depicting the work flow of the study. After giving informed consent liver transplant recipients are randomized to two parallel groups treated for eight days (day of surgery and the following seven days) using glycine solution or placebo. Follow-up period will be at least 31 days after transplantation.
Table 2 Schedule of study related activities and data collection. The following study related activities are planned for each subject. Blood samples already taken in the routine process at the planned time do not have to be taken twice.
Clinical period Follow-up period
Study related activity pre-OP OP ICU-entry POD 1 POD 2–7 POD 8
Informed consent x
Medical history x
Physical examination x
Donor and graft history x
Randomisation x
Treatment x x x
Safety parameters (measured after 5 a.m. before additional treatment, unless otherwise stipulated)
12-lead ECG x x
Vital signs x x x x x
Blood chemistry x x x x x
Haematological parameters x x x x x
Coagulation parameters x x x x x
CyA (trough concentration) x x x
Pregnancy testing x
Efficacy Parameters (measured after 5 a.m. before additional treatment, unless otherwise stipulated)
Biopsy x (immediately after rearterialisation)
Blood flow in portal vein and common hepatic artery x (1 hour after reperfusion)
AST, ALT x x# x# x x
Biltotal, Bildir, Quick, AT III x x x x x
Gly* x x (POD3)
KreaS, HS x x x x x
24-hours-urine (VolU, KreaU) start x x
Indicators for early onset of graft failure x x x
Occurrence of late onset of graft failure X
#; AST and ALT will be measured every 4 to 6 hours during the first 24 hours after ICU entry.
*; Samples for glycine plasma level has to be collected immediately after first study drug application during surgery and on POD 3 (after study drug application).
Objectives and endpoints
The primary objective of this trial is to demonstrate both efficacy and safety of glycine treatment compared to placebo in the postoperative period during the first eight days after liver transplantation and during the long-term follow-up. Secondary objectives are reperfusion injury to the graft and mortality. Furthermore, it will be investigated whether CyA-induced nephrotoxicity can be reduced by glycine.
Both AST and ALT peaks are expected to be surrogates of the extent of reperfusion injury. AST and ALT are the most common parameters for progression of liver related disease. An increase of these transaminases correlates with both primary failure of the organ and graft injury with subsequent organ dysfunction. Thus peak serum level of AST, measured at intensive care unit (ICU) entry and within the first 8 days after transplantation, has been chosen as the most important primary endpoint of this trial. Further primary endpoints ordered hierarchically are ALT peak during the same period and graft survival based on patients' death or announcement for re-transplantation (minimum observation period 31 days). The last reflects the potential clinical benefit for the patient.
Secondary endpoints are the effect of glycine on liver injury based on liver biopsy immediately after re-arterialisation (according to pathological report), total blood flow in portal vein and common hepatic artery 1 hour after reperfusion, graft injury based on both AST and ALT serum levels (area under the curve (AUC)), incidence of early graft failure based on peak of transaminases or clotting factor support, early onset of graft dysfunction based on Quick's value, serum bilirubin (AUC), and CyA-induced nephrotoxicity based on retention parameters during the first eight days after transplantation (AUC).
Sample size calculation
The sample size calculation is based on the AST peak, the most important primary endpoint in this trial. According to historic data of about 450 liver transplant recipients at the Department of Surgery, University of Heidelberg, a log-normal distribution of the AST peaks is plausible with a standard deviation of 1.01 for the log (AST peak)-values (details are available on request). These data served for the sample size calculation of this trial. A decreased AST-peak by 300 U/L is considered clinically relevant. Thus a sample size of 65 patients per group, i.e. a total of 130 patients, is sufficient to detect differences with a power of 80%, taking into account the planned interim analysis (two-sided t-test for the log (AST peaks), overall level of significance α = 0.05) (nQuery® 4.0, EaSt®-2000). Details on transformation of the clinically relevant difference on the original scale to a clinically relevant difference on the log scale used for sample size calculation are available on request.
Randomization and treatment
Here a 1:1 randomization ratio has been chosen. Randomization is stratified for each center and for the duration of cold ischemia (≤ 10 hrs or >10 hrs). The study medication is produced, labeled, and packed by a Clinical Pharmacy Department. The guideline for good manufacturing practice (GMP) is adhered to. The intravenous medication for the verum group contains 250 ml glycine solution (4.4%; 11 g glycine, dissolved in aqua ad injectione). In contrast the placebo consisted of 250 mL of the vehicle (aqua ad injectionem) have to be infused to patients.
All patients receive their first study medication via a central venous line during liver transplantation one hour prior to reperfusion. During the following week, on post operative days (POD) 1 to 7, the study medication is infused once per day in the morning, after taking blood and urine samples for safety and efficacy parameters. A last visit is scheduled for POD 8 to investigate the patients' clinical status and to collect all parameters for efficacy and safety. To detect late graft failures all patients are observed after the initial first eight days until the last patient is enrolled in this trial. These patients will then be observed for at least 31 days after transplantation. A late graft failure is defined as patients' death or announcement for re-transplantation.
If it is medically imperative to know the patient's treatment, emergency envelopes contain the information on the subject's study medication. Those are to be opened only under emergency circumstances. During the trial a concomitant treatment may be given at the discretion of the investigator, if these are considered necessary for the subject's welfare.
All patients routinely receive standard immunosuppressive therapy. The initial dose of CyA is 2 × 2 mg/kg body weight during the first 24 hrs. Subsequently, daily doses of CyA are adapted to the actual CyA serum level which is measured in the blood samples taken in the morning of POD 1 to 8. The trough concentration should be between 200–250 μg/L during the first month after liver transplantation. In case of renal insufficiency FK506 can be used alternatively.
The amount of coagulation factors and AT III, red blood cells, and fresh frozen plasma given after transplantation are documented until the morning of POD 8.
Adverse events
All adverse events (AE) are recorded. Events related to the initial diagnosis for liver transplantation, to the transplantation procedure itself, or problems associated with routine procedures after transplantation, i.e. liver biopsy, are not to be noted as AE or serious adverse event (SAE) unless the investigator deems the events to be a cause of the study drug. All SAE potentially associated with the application of study medication must be documented on a "Serious Adverse Event" form which has to sent to the principal investigator (LKP, according to German Drug Law) within 24 hrs or latest on the following working day. The LKP ensures that SAE are reported to the safety board, ethics committees, and to further investigators, if applicable.
Quality assurance
The study is performed according to the principles of the ICH-GCP guidelines [39] and the ethical principles according to the current revision of the Declaration of Helsinki [40] and local legal and regulatory requirements. The trial is monitored by the KKS Heidelberg according to Standard Operation Procedures (SOP) that are based on ICH-GCP guidelines.
An independent safety board monitors closely the proper conduct of the trial and all SAE reports to ensure the safety of the subjects during the course of the study.
Statistics and data management
All findings including both the clinical and laboratory findings are documented in the subject's case report form (CRF). All data are entered in a database as recorded. To ensure highest data quality a double data entry is performed. All missing data or inconsistencies are reported back to the center(s) and clarified by the responsible investigator. If no further corrections of the database are to be made it will be declared closed and used for statistical analysis.
The primary endpoints of the trial are peak serum levels of both AST and ALT measured within the first 8 days after transplantation, and graft survival based on patients death or announcement for re-transplantation.
Therefore the statistical hypotheses to be tested are as follows:
H0: ASTglycine = ASTplacebo vs. H1: ASTglycine ≠ ASTplacebo, (1)
H0: ALTglycine = ALTplacebo vs. H1: ALTglycine ≠ ALTplacebo, (2)
H0: Sglycine = Splacebo vs. H1: Sglycine ≠ Splacebo, (3)
While ASTglycine and ASTplacebo represent the distribution of AST peaks in the glycine and placebo group, respectively, ALTglycine and ALTplacebo represent the distribution of ALT peaks in the two groups, and Sglycine and Splacebo represent the survival functions of the two groups regarding graft survival.
The statistical hypotheses formulated above will be tested, each at the level of significance α (see below), in the strict order (1), (2), (3), i.e. a statistical testing procedure with a priori ordered hypotheses is applied [41].
All analyses are done according to the principle of intention-to-treat. Additional analyses (per protocol population, sensitivity analyses) will be described in the statistical analysis plan in more detail before closure of the database.
In the interim analysis only safety parameters and the primary endpoints AST peak and ALT peak will be evaluated. Data for the third primary endpoint (graft survival) are collected only at the end of the trial. The interim analysis is performed according to the group sequential design of O'Brien and Fleming [42]. The overall level of significance is α = 0.05, i.e. the level of significance for the interim analysis is α 1 = 0.0035, the level of significance for the final analysis is α2 = 0.0488.
If a patient dies or a re-transplantation has to be performed during the clinical period, serum measurements are not available. As this is the worst possible outcome, patients will be allocated the worst ranks for the respective (non-parametric) analyses.
Sensitivity analyses regarding the methods for dealing with missing values will be performed. Further details regarding the handling of missing data will be laid down in the statistical analysis plan that will be completed before unblinding of the data.
Discussion
This ongoing clinical trial particularly demonstrates how a scientific hypothesis has been developed from bench to bedside, with classical investigations along the value chain of translational research from early in vitro work to in vivo experiments and finally from single case observations to the highest level of evidence, a randomized, prospective, controlled, double-blinded clinical trial. If previous findings are confirmed by the data of this ongoing clinical trial, glycine would improve the overall outcome after liver transplantation.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SPL participated in the design of the study, developed essential study documents and acts as coordinating investigator (Leiter der Klinischen Prüfung (LKP) according to German drug law). KU participated in the design of the study, developed the statistical part of the study protocol and the statistical analysis plan. MSG performed quality review to assure adherence to current guidelines and laws. HB, TWK, MWB, and EK supported the design of the study with their knowledge and experience. PS conceived and designed the study based on his own preclinical and clinical results. Further he designed and conducts the study as the main investigator. PS and SPL wrote the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors want to acknowledge the members of the safety board (Hartwig Bunzendahl, Rolf Holle, and Peter Sauer) and the site staff for supporting and conducting the study. Further we thank Genevieve Dei-Anane for editing the manuscript.
This study is supported in parts by an unrestricted grant from Novartis Pharma GmbH, Germany.
==== Refs
Williams DA Fogarty KE Tsien RY Fay FS Calcium gradients in single smooth muscle cells revealed by the digital imaging microscope using Fura-2 Nature 1985 318 558 561 3934562 10.1038/318558a0
Schemmer P Mehrabi A Kraus T Sauer P Gutt CN Uhl W Büchler MW New aspects on reperfusion injury to liver – impact of organ harvest Nephrol Dial Transplant 2004 19 26 35 14671034
Lemasters JJ Thurman RG Reperfusion injury after liver preservation for transplantation Annu Rev Pharmacol Toxicol 1997 37 327 338 9131256 10.1146/annurev.pharmtox.37.1.327
Ikejima K Qu W Stachlewitz RF Thurman RG Kupffer cells contain a glycine-gated chloride channel Am J Physiol 1997 272 G1581 G1586 9227496
Bouwens L Structural and functional aspects of Kupffer cells Revis Biol Celular 1988 16 69 94 3067272
Cerra FB West MA Keller G Mazuski J Simmons RL Hypermetabolism/organ failure: the role of the activated macrophage as a metabolic regulator Prog Clin Biol Res 1988 264 27 42 3289026
Fusaoka T Hunt KJ Lemasters JJ Thurman RG Evidence that activation of Kupffer cells increases oxygen uptake after cold storage Transplantation 1994 58 1067 1071 7974712
Schemmer P Enomoto N Bradford BU Bunzendahl H Raleigh JA Lemasters JJ Thurman RG Activated Kupffer cells cause a hypermetabolic state after gentle in situ manipulation of liver in rats Am J Physiol 2001 280 G1076 G1082 11352799
Schemmer P Bradford BU Bunzendahl H Raleigh JA Lemasters JJ Thurman RG Intravenous glycine improves survival in rat liver transplantation Am J Physiol 1999 276 G924 G932 10198336
Wheeler MD Ikejima K Enomoto N Stachlewitz RF Seabra V Zhong Z Yin M Schemmer P Rose ML Rusyn I Bradford BU Thurman RG Glycine: a new anti-inflammatory immunonutrient Cell Mol Life Sci 1999 56 843 856 11212343 10.1007/s000180050030
Zhong Z Wheeler MD Li X Froh M Schemmer P Yin M Bunzendahl H Bradford BU Lemasters JJ L-Glycine: a novel antiinflammatory, immunomodulatory, and cytoprotective agent Curr Opin Clin Nutr Metab Care 2003 6 229 240 12589194 10.1097/00075197-200303000-00013
Epstein FH Brosnan JT Tange JD Ross BD Improved function with amino acids in the isolated perfused kidney Am J Physiol 1982 243 F284 F292 7114258
Weinberg JM Davis JA Abarzua M Rajan T Cytoprotective effects of glycine and glutathione against hypoxic injury to renal tubules J Clin Invest 1987 80 1446 1454 2890657
Zhong Z Jones S Thurman RG Glycine minimizes reperfusion injury in a low-flow, reflow liver perfusion model in the rat Am J Physiol 1996 270 G332 G338 8779976
Marsh DC Vreugdenhil PK Mack VE Belzer FO Southard JH Glycine protects hepatocytes from injury caused by anoxia, cold ischemia and mitochondrial inhibitors, but not injury caused by calcium ionophores or oxidative stress Hepatology 1993 17 91 98 8423046 10.1016/0270-9139(93)90197-U
Ozaki M Ozasa H Fuchinoue S Teraoka S Ota K Protective effects of glycine and esterified gamma-glutamylcysteine on ischemia/reoxygenation injury of rat liver Transplantation 1994 58 753 755 7940706
Schilling M den Butter G Marsh DC Belzer FO Southard JH Glycine inhibits phospholipolysis of hypoxic membranes Tx Med 1994 6 140 143
Bachmann S Peng X-X Currin RT Thurman RG Lemasters JJ Glycine in Carolina rinse solution reduces reperfusion injury, improves graft function, and increases graft survival after rat liver transplantation Transplant Proc 1995 27 741 742 7879166
Schemmer P Jost F Conzelmann LO Pöschl U Mehrabi A Gebhard MM Büchler MW Klar E Glycin minimiert Leukozyten-Endothel-Interaktion nach warmer Ischämie durch Inaktivierung der Kupfferzellen Chirurgisches Forum 2003 32 455 458
Takei Y Marzi I Kauffman FC Currin RT Lemasters JJ Thurman RG Increase in survival time of liver transplants by protease inhibitors and a calcium channel blocker, nisoldipine Transplantation 1990 50 14 20 2368133
Schemmer P Golling M Kraus T Mayatepek E Herfarth C Klar E Glycine reduces reperfusion injury in human liver transplantation: our first patients Transplant Proc 2001 33 3750 3752 11750598 10.1016/S0041-1345(01)02588-X
Schemmer P Golling M Kraus T Mehrabi A Mayatepek E Herfarth C Klar E Extended experience with glycine for prevention of reperfusion injury after human liver transplantation Transplant Proc 2002 34 2307 2309 12270410 10.1016/S0041-1345(02)03247-5
Farthing MJ Clark ML Nature of the toxicity of cyclosporin A in the rat Biochem Pharmacol 1981 24 3311 3316 7326039 10.1016/0006-2952(81)90604-3
Jung K Pergande M Influence of cyclosporin A on the respiration of isolated rat kidney mitochondria FEBS Lett 1985 183 167 171 2858411 10.1016/0014-5793(85)80977-7
Mehring N Neumann K-H Rahn K-H Zidek W Mechanisms of ciclosporin A-induced vasoconstriction in the isolated perfused rat kidney Nephron 1992 60 477 481 1584325
Zhong Z Arteel GE Connor H Yin M Frankenberg Mv Stachlewitz RF Raleigh JA Mason RP Thurman RG Cyclosporin A increases hypoxia and free radical production in the rat kidney: prevention by dietary glycine Am J Physiol 1998 275 F595 F604 9755131
Ito S Cherubini E Strychnine-sensitive glycine responses of neonatal rat hippocampal neurons J Physiol 1991 440 67 83 1804982
Heyman SN Brezis M Epstein FH Spokes K Rosen S Effects of Glycine and Hypertrophy on Renal Outer Medullary Hypoxic Injury in Ischemia Reflow and Contrast Nephropathy Am J Kidney Dis 1992 19 578 586 1595707
Venkatachalam MA Weinberg JM Mechanisms of cell injury in ATP-depleted proximal tubules. Role of glycine, calcium, and polyphosphoinositides Nephrol Dial Transplant 1994 9 15 21 7800249
Yin M Zhong Z Connor HD Bunzendahl H Finn WF Rusyn I Li X Raleigh JA Mason RP Thurman RG Protective effect of glycine on renal injury induced by ischemia-reperfusion in vivo Am J Physiol 2002 282 F417 F423
Zhong Z Connor HD Yin M Moss N Mason RP Bunzendahl H Forman DT Thurman RG Dietary glycine and renal denervation prevents cyclosporin A-induced hydroxyl radical production in rat kidney Mol Pharmacol 1999 56 455 463 10462532
Schemmer P Schoonhoven R Swenberg JA Bunzendahl H Thurman RG Gentle in situ liver manipulation during organ harvest decreases survival after rat liver transplantation: Role of Kupffer cells Transplantation 1998 65 1015 1020 9583858 10.1097/00007890-199804270-00001
Schemmer P Connor HD Arteel GE Raleigh JA Bunzendahl H Mason RP Thurman RG Reperfusion injury in livers due to gentle in situ organ manipulation during harvest involves hypoxia and free radicals J Pharmacol Exp Ther 1999 290 235 240 10381781
Schemmer P Enomoto N Bradford BU Bunzendahl H Raleigh JA Thurman RG Autonomic nervous system and gut-derived endotoxin: involvement in activation of Kupffer cells after in situ organ manipulation World J Surg 2001 25 399 406 11344388 10.1007/s002680020070
D'Souza DC Gil R Cassello K Morrissey K Abi-Saab D White J Sturworld R Bennett A Karper LP Zuzarte E Charney DS Krystal JH I.v. glycine and oral D-cycloserine effects on plasma and CSF amino acids in healthy humans Biol Psychiat 2000 47 450 462 10704956 10.1016/S0006-3223(99)00133-X
Evins E Fitzgerald SM Wine L Rosselli R Goff DC Placebo-controlled trial of glycine added to clozapine in schizophrenia Am J Psychiatry 2000 157 826 828 10784481 10.1176/appi.ajp.157.5.826
Heresco-Levy U Javitt DC Ermilov M Mordel C Silipo G Lichtenstein M Efficacy of high-dose glycine in the treatment of enduring negative symptoms of schizophrenia Arch Gen Psychiatry 1999 56 29 36 9892253 10.1001/archpsyc.56.1.29
Zhong Z Li X Yamashina S von Frankenberg M Enomoto A Ikejima K Kolinsky M Raleigh JA Thurman RG Cyclosporine A causes a hypermetabolic state and hypoxia in the liver: prevention by dietary glycine J Pharmacol Exp Ther 2001 299 858 865 11714869
International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH): Guideline E6 Note for Guidance on good clinical practice (GCP)
World Medical Association Declaration of Helsinki Recommendations guiding physicians in biomedical research involving human subjects Adopted by the 18th World Medical Assembly Helsinki, Finland, June 1964, amended by 48th General Assembly, Somerset West, South Africa 1996
Kieser M A confirmatory strategy for therapeutic equivalence trials Int J Clin Pharmacol Ther 1995 33 388 390 7582393
O'Brian PC Fleming TR A multiple testing procedure for clinical trial Biometrics 1979 35 549 556 497341
|
16105183
|
PMC1208918
|
CC BY
|
2021-01-04 16:28:04
|
no
|
BMC Surg. 2005 Aug 17; 5:18
|
utf-8
|
BMC Surg
| 2,005 |
10.1186/1471-2482-5-18
|
oa_comm
|
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-401599241110.1186/1475-925X-4-40ResearchA two-dimensional mathematical model of non-linear dual-sorption of percutaneous drug absorption George K [email protected] School of Information Systems, Computing and Mathematics; Department of mathematical Sciences, Brunel University, Uxbridge, Middlesex, , UB8 3PH, UK2 Aditi College, University of Delhi, Bawana, Delhi, 110039, India3 Civil and Computational Engineering Centre, School of Engineering, Singleton Park, Swansea, SA2 8PP Wales, UK2005 3 7 2005 4 40 40 6 12 2004 3 7 2005 Copyright © 2005 George; licensee BioMed Central Ltd.2005George; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Certain drugs, for example scopolamine and timolol, show non-linear kinetic behavior during permeation process. This non-linear kinetic behavior is due to two mechanisms; the first mechanism being a simple dissolution producing mobile and freely diffusible molecules and the second being an adsorption process producing non-mobile molecules that do not participate in the diffusion process. When such a drug is applied on the skin surface, the concentration of the drug accumulated in the skin and the amount of the drug eliminated into the blood vessel depend on the value of a parameter, C, the donor concentration. The present paper studies the effect of the parameter value, C, when the region of the contact of the skin with drug, is a line segment on the skin surface. To confirm that dual-sorption process gives an explanation to non-linear kinetic behavior, the characteristic features that are used in one-dimensional models are (1) prolongation of half-life if the plot of flux versus time are straight lines soon after the vehicle removal, (2) the decrease in half-life with increase in donor concentration. This paper introduces another feature as a characteristic to confirm that dual-sorption model gives an explanation to the non-linear kinetic behavior of the drug. This new feature is "the prolongation of half-life is not a necessary feature if the plots of drug flux versus time is a non-linear curve, soon after the vehicle removal".
Methods
From biological point of view, a drug absorption model is said to be nonlinear if the sorption isotherm is non-linear. When a model is non-linear the relationship between lag-time and donor concentration is non-linear and the lag time decreases with increase in donor concentration. A two-dimensional dual-sorption model is developed for percutaneous absorption of a drug, which shows non-linear kinetic behavior in the permeation process. This model may be used when the diffusion of the drug in the direction parallel to the skin surface must be examined, as well as in the direction into the skin, examined in one-dimensional models. The dual-sorption model is an initial/boundary value problem which consists of (1) one non-linear, two-dimensional, second-order parabolic equation, (2) boundary conditions, (3) one initial condition. Note that, the number of boundary conditions are, six and four, respectively, if the permeation process under consideration is, during the application of the vehicle and during the removal of the vehicle. Adopting the approach of method of lines, the initial/boundary value problem is transformed into an initial-value problem, which consists of (1) a system of non-linear ordinary differential equations, (2) one initial condition. The system of non-linear ordinary differential equations contains time-dependent non-homogeneous terms, if the permeation process under consideration is, during the application of the vehicle. To solve this initial-value problem, an eight-stage sequential algorithm which is second-order accurate, and requires only tri-diagonal solvers, is developed.
Results
Simulation of the numerical methods described is carried out with various values of the parameter C. The illustrations are given in the form of figures. The concentration profiles are viewed as parabolas along the mesh lines parallel to x-axis or y-axis. The flow rates in different subregions of the skin-region are studied. The shapes of the concentration profiles are examined before and after the steady-state concentration is reached. The concentration reaches steady-state when the flux reaches the steady state. The plots of flux versus time and cumulative amount of drug eliminated into the receptor cell versus time are given.
Conclusion
Based on the various values of the parameter, C, conclusions are drawn about (1) flow rate of the drug in different regions of the skin, (2) shape of the concentration profiles, (3) the time required to reach the steady-state value of the concentration, (4) concentration of the drug in different regions of the skin, when steady-state value of the concentration is reached, (5) the time required to reach the steady-state value of the flux, (6) time required to reach the steady-state value of the concentration of the drug, (7) half-life of the concentration of the drug and (8) lag-time.
A comparison, between this two-dimensional model and the one-dimensional non-linear dual-sorption model that exists in the literature, is done based on (1) the shape of the concentration profiles at various time levels, (2) the time required to reach the steady-state value of the concentration, (3) lag-time and (4) half-life.
==== Body
Background
A non-linear model takes into account the property of the skin to sorb and bind substances during the process of permeation in addition to the ordinary dissolution. Such a non-linear model can be used to explain the disparity between the steady-state diffusivity of the drug in the skin and the unsteady-state value computed from transient (time-lag) permeation experiments, see [1]. A non-linear dual-sorption model of percutaneous drug absorption is described in [2]. Another mathematical model of percutaneous drug absorption and its exact solution is described in [3]. In [4], the non-linear percutaneous permeation kinetics of timolol is studied in vitro with human cadaver skin. A model for a suspension with a finite dissolution rate is solved numerically in [5]. All these non-linear models are one-dimensional and the region of contact of the skin with the drug, is a single point, say x = 0, where x measures the distance into the skin.
To the authors' knowledge, no two-dimensional non-linear mathematical model of percutaneous drug absorption exists in the literature, where the region of contact is a line segment, say x = 0, 0 ≤ y ≤ Lc. The purpose of this paper is (1) to model the above situation mathematically (2) to obtain an efficient numerical method which can use a large time step compared to spacial discretization step; and also handles discontinuities between initial and boundary conditions in the mathematical model.
Methods
Model equations
(a) Concentration before the steady state is reached (Concentration during the application of the vehicle)
Let the drug be applied as an ointment on the skin-surface, say, {(0, y) : 0 ≤ y ≤ Lc}, at time t = 0. The two-dimensional model developed in [6] takes into account the drug kinetics in the skin where the ointment is not directly applied. The one-dimensional non-linear dual-sorption model given in [2] postulates the total concentration of the drug, CT, in the skin is composed of two parts, (1) the mobile solute concentration CD, which is due to the mechanism of simple dissolution and is expressed as CD = KDC, in which KD is skin/receptor cell partition coefficient and C is donor-cell concentration, (2) the immobile solute concentration CI, which is due to the adsorption process, and is expressed as , in which is Langmuir's saturation constant and b is Langmuir's affinity constant. Hence the total concentration CT, in the skin is given by
CT = CD + CI.
Based on [6] and [2], to describe the drug kinetics at any time t > 0, a two-dimensional non-linear dual-sorption model is developed. Let the thickness (distance between the skin-surface and skin-capillary boundary) of the skin be Ls. Assuming that skin is an isotropic medium, that is, the diffusivity κ, is the same in the x and y directions, the drug concentration CD = CD(x, y, t) in the skin is governed by
where Ls, Ld, Lu and Lc are positive real numbers. Note that, (1) is non-singular as b > 0, > 0 and CD ≥ 0. Following [2], the concentration along the skin-receptor cell boundary may be considered to have the value zero. Assuming that the donor-cell concentration at the uppermost epidermis {(0, y, t) : 0 ≤ y ≤ Lc, t > 0} is maintained at the value C, the boundary conditions for the model are given by
CD (0, y, t) = KDC, 0 ≤ y ≤ Lc, t > 0, (2b)
CD (Ls, y, t) = 0, -Ld ≤ y ≤ Lu, t > 0, (2f)
Assuming that there is no drug in the skin before the application, the initial distribution associated with the PDE is
CD(x, y, 0) = 0,0 ≤ × ≤ Ls, -Ld ≤ y ≤ Lu (3)
The flux J = J(t), through the skin to the receptor site, per unit area, is given by
The cumulative amount of drug eliminated into the receptor cell per unit area at time τ (see [2]) is
The initial/boundary-value problem given by (1) to (3) is named as "problem (PA)".
(b) Concentration after the steady state is reached (Concentration after the vehicle removal)
Let Ts denote the time at which the steady-state is reached by the problem (PA). Suppose that at time t = Ts, the vehicle is removed instantaneously from the skin. Then (1) remains unchanged, but the initial conditions and boundary conditions changes and consequently, (2a) to (2c) disappear from the mathematical model, leaving the PDE's
CD (Ls, y, t) = 0, -Ld ≤ y ≤ Lu, t > Ts (7d)
The initial distribution associated with PDE (6) is
CD = CD (x, y, Ts), 0 ≤ x ≤ Ls, -Ld ≤ y ≤ Lu, (8)
where CD (x, y, Ts) is computed by solving problem (PA). The initial/ boundary-value problem given by (6) to (8) is named as "problem (PB)".
Numerical methods
The approach to be adopted in obtaining a numerical solution, is the method of lines in which the initial/ boundary-value problem to be solved is transformed into a first-order initial-value problem. At any time t > 0, the region associated with the skin is given by Rs = {(x, y) : 0 ≤ x ≤ Ls, - Ld ≤ y ≤ Lu}. The region of the contact Rc, of the drug and the skin surface is defined as Rc = {(0, y) : 0 ≤ y ≤ Lc}. Superimpose on Rs a rectangular grid with mesh lengths, h > 0 and H > 0, respectively, in the x and y directions. Let k > 0 be a constant time step. Let N be a positive integer, and h = 1/(N + 1). Also it is assumed that Lc = QcH, Lu = QuH, Ld = QdH, where Qc, Qu and Qd are positive integers. The grid points are given by (xl, ym, tj); xl = lh, l = 0(1)(N + 1), ym = mH, m = -Qd(1)Qu and tj= jk, j = 0, 1, 2, ...s, ...etc. with sk = Ts. The values tj represent the time-levels for the problems (PA) and (PB) for j ≤ s and j ≥ s respectively. The finite-difference solution which approximates the solution CD (x, y, t) of the two-dimensional parabolic equation (1), is sought at each mesh point (xl, ym, tj) in the region [Rs- Rc - {(Ls, y), -Ld ≤ y ≤ Lu}] × t > 0. Note that, for the initial-value problem (PB), Rc = Φ (the empty set). For the problem (PA), the differential equation (1), subject to the boundary conditions (2a) to (2f), is discretized at all grid points in the region [Rs - Rc -{(Ls, y), -Ld ≤ y ≤ Lu}] × [0 ≤ t ≤ Ts]. For the problem (PB), the differential equation (6), subject to the boundary conditions (7a) to (7d), is discretized at all grid points in the region [Rs -{(Ls, y), -Ld ≤ y ≤ Lu}] × [t > Ts]. For notational simplicity, denote
Discretization for problem (PA)
At any time-level t = tj, or , the {(Qd + Qu + 1) (N + 1) - (Qc + 1)} elements will be ordered in rows parallel to the x-axis and in vector form, and will be denoted by CD(t) and Q(t), where Q(t) is due to the boundary condition (2b). For W = CD and Q, and = or , denote
In the above notation n takes the value 0 if m ≠ 0(1)Qc and takes the value 1 otherwise. The vector CD(t) is a vector of unknowns and the vector Q(t) is given by
All other are zero. In the following discretizations, (xl, ym), pl,m, ql,m and CDl,m, respectively, denote (xl, ym, tj), , and . At any time-level t = tj, the differential equation (1) can be discretized as
If and are respectively, suitable approximations to CDxx;l;m and CDyy;l;m, the above discretization can be written as
At an interior grid point (xl, ym), and are given by
If (xl, ym) is a boundary point, using the boundary conditions (2a) to (2f), and are given by
where
A1 = 0, B1 = 2, if l = 0, m = -Qd(1)(-1), (Qc + 1)(1)Qu,
A1 = 0, B1 = 1, if l = 1, m = 0(1)Qc,
A1 = 1, B1 = 1, if l = 1(1)N - 1, m = -Qd(1)Qu,
A1 = 1, B1 = 0, if l = N, m = -Qd(1)Qu,
A2 = 0, B2 = 2, if l = 0(1)N, m = -Qd,
A2 = 2, B2 = 0, if l = 0(1)N, m = Qu,
A2 = 1, B2 = 1, if l = 0, m = (-Qd + 1)(1)(-2), (Qc + 2)(1)(Qu - 1),
A2 = 1, B2 = 0, if l = 0, m = -1
A2 = 0, B2 = 1, if l = 0, m = (Qc + 1).
System of ordinary differential equations
Combining the discretizations (9) together with expressions for and given by (10) and (11) respectively, a system of ordinary differential equations is formed as
where
The value of m in Ed and Fd is -Qd, and that in Eu and Fu is Qc + 1. The abbreviations m1, m2, ... etc. represent m + 1, m + 2, ... etc. For m ≠ 0, 1, 2, ...Qc, Em and Fm are square matrices of order N + 1 and are given by
For m = 0, 1, 2, ..., Qc, Em and Fm are square matrices of order N and are given by
For the values of m = -1, Qc + 1, and m = 0, Qc, the matrices Gm are of orders (N + 1) × N and N × (N + 1), respectively, and are given by
For the values of m = 0, 1, ..., Qc, m ≠ 0, 1, 2, ...Qc, the matrices Om are the zero matrices of orders N and N + 1 respectively. The matrices Od and Oc are zero matrices of orders (N + 1) × N and N × (N + 1), respectively.
Recurrence relation and its implementation via sequential algorithm
In [7], an eight stage sequential algorithm is described to solve two space linear parabolic equations. In [2], a sequential algorithm of two tridiagonal solvers is described to solve one-space non-linear dual-sorption model. To the author's knowledge, there is no sequential algorithm in the literature to solve the two-dimensional non-linear parabolic equation (1). The aim of this section is to develop an eight-stage sequential algorithm of tridiagonal solvers, to solve the system of non-linear ordinary differential equations given by (12), which gives the solution of two-space non-linear parabolic equation (1). The main idea used is, to rewrite the system of equations (12) in such a way that, the numerical techniques described in [7] and [2] can be extended to two-space non-linear equations. To achieve this, we proceed as follows.
Let Q(t) = R(t) + S(t), where the elements of the vectors R(t) and S(t) are, respectively, denoted by and , and are defined by
Note that the vector, R(t), is the contribution of CDxx to the vector Q(t). The vector, S(t), is the contribution of CDyy to the vector Q(t). The system of non-linear ordinary differential equations given by (12) can be written as
It is known (see [2,8]) that the system of ordinary differential equations given by (12), subject to the initial condition (3), satisfies the recurrence relation
where D* = diag{d/dt}. The exponential term in the recurrence relation (16) will be approximated by its (2, 0) Padé approximant to give
which, following pre-multiplication, gives
Using the split form (15), and following [7] and [2], a four-stage sequential algorithm is developed to obtain , which is an approximation to CD(t + k) in (17). It is given by
(I - r1kE) Z = CD(t) + r1kR(t), (18a)
(I - r2kE) V = Z + r2kR(t), (18b)
(I - r1kF) Z = V + r1kS(t), (18c)
(I - r2kF) (t + k) = Z + r2kS (t). (18d)
where (see [8])
Another solution, , which is an approximation to CD(t + k) in (17), is obtained by interchanging the matrices E and F; and the vectors R and S; in (18a) to (18d). It is given by
(I - r1kF) Z = CD(t) + r1kS(t), (18e)
(I - r2kF) V = Z + r2kS(t), (18f)
(I - r1kE) Z = V + r1kR(t), (18g)
(I - r2kE) (t + k) = Z+ r2kR (t). (18h)
Note that the approximations, and , given by (18d) and (18h), are first-order accurate in time and their linear combination defined by
is second order accurate in time. The eight-stage algorithm, defined by (18a) to (18i) is, L0 stable and uses only tridiagonal solver.
Initial/boundary value problem (PB)
The "problem (PB)" is modelled by the equations (6) to (8). At any time-level t = tj, j ≥ s, , the (Qd + Qu + 1) × (N + 1) elements, will be ordered in rows parallel to the x-axis and in vector form, and will be denoted as CD(t), with
At a grid point xl, ym, the required discretization is
For l ≠ 0, m ≠ (-Qd + 1) (1) (Qu - 1); and are given by (10) and for l = 0, m = (-Qd + 1)(1)(Qu - 1),
and
Combining the discretizations given by (19), a system of differential equations is formed as
in which
In the above matrices, the value of m is -Qd and m1 = m + 1, m2 = m + 2,... The matrices Ei and Fi, i = -Qd(1)Qu, are square matrices of order N +1, and are given by (13). Proceeding as in problem (PA), the eight-stage algorithm to solve the system of differential equations (20), subject to the initial conditions (8), is
(I - r1kE) Z = CD(t), (21a)
(I - r2kE) V = Z , (21b)
(I - r1kF) Z = V, (21c)
(I - r2kF) (t + k) = Z, (21d)
(I - r1kF) Z = CD(t), (21e)
(I - r2kF) V = Z, (21f)
(I - r1kE) Z = V, (21g)
(I - r2kE) (t + k) = Z, (21h)
Note that the sequential algorithm defined by (21a) to (21i) is obtained from the equations (18a) to (18i), replacing the vectors R(t) and S(t), by the zero vector of length (Qd + Qu+ 1) × (N + 1). This shows the efficiency of the splitting of the vector Q(t), and the eight-stage algorithm developed in problem (PA).
Stability Analysis
The stability analysis of two-stage sequential algorithm for the solution of one-space non-linear second-order parabolic equation is described in [2]. The stability analysis of four-stage sequential algorithm for the solution of two-space second-order parabolic equation is described in [7]. In this section, following [7] and [2], it is to prove that the eight-stage sequential algorithms described by (18a) to (18i) and (21a) to (21i) are L0 stable.
The amplification matrix R*(k(E + F)) of the method defined by (18a) to (18d) is given by
neglecting the higher order terms O (-k3).
Hence the symbol R*(-z), of the method defined by (18a) to (18d), to evaluate (t + k) is given by
Similarly the symbol R+(--z), of the method defined by (18e) to (18h), to evaluate (t + k) is given by
In (22) and (23), z = -kλ where λ is an eigen value of the matrix E + F. Hence using (22) and (23) the symbol, S(-z), of the method (18i) is
The method is L0 stable if
| S(-z) | ≤ 1 (24)
S(-z) → 0, as z → ∞
• Proof of (24):- Applying Brauer's theorem, it can be shown that all the eigen values λ of the matrix E + F are negative. Hence, z = kλ > 0.
• Proof of (25):-
From (24) and (25), it is concluded that the numerical method defined by (18a) to (18i) is L0 stable. Similarly the numerical method defined by (21a) to (21i) is L0 stable. Since the method is L0 stable, the discontinuities around y = 0 and y = Lc are not propagated.
Results and discussion
In order to examine the behavior of the recurrence relations (18a) to (18i) and (21a) to (21i), a series of five numerical experiments, similar to those described in [2], are carried out for two space dimensions. In these experiments the parameter-values used are those used in [2]. For the experiments numbered n = 1, 2, 3, 4, and 5, the parameter C was given the values 4.1, 19.5, 43.1, 51.4 and 64.0 respectively. The rest of the parameter values are given by
Ls = 0.004 cm, Ld = 0.0320 cm, Lc = 0.128 cm, Lu = 0.160 cm, k = 0.1 h, H = 0.0004, κ = 0.0000018 cm2/h, = 5.0 mg/ml, KD = 1.1, h = 0.0002, N = 19, Qc = 320, Qu = 400, Qd = 80, b = 0.5599 ml/mg.
It was assumed in all numerical experiments that the drug was applied until steady-state concentration profile is reached. Let denotes the time in hours, at which the concentration profiles for the nth (n = 1, 2, 3, 4 and 5) experiment reaches the steady-state. Suppose that at time t = , the vehicle is removed instantaneously from the skin. The pattern of the concentration profile is observed for fifty hours more, after the vehicle is removed at the steady-state. For the nth (n = 1(1)5) experiment, the values of CD are computed (1) using the sequential algorithm (18a) to (18i), in the time interval 0 <t ≤ , (2) using the sequential algorithm (21a) to (21i), in the time interval <t ≤ + 50. Following [2], for the value of n = 1 (C = 4.4), the profiles of concentration CD at t = 0.5. 1.0, 2.0, 3.0, 4.5 and 6.0 are given in figure 1. In figure 2, the profiles of concentration CD are given at t = 24.0, 26.0, 28.0, 30.0, 32.0 and 34.0. In figure 3 the profile of concentration CD, are given at time t = = 128.7, the time at which concentration, CD, reaches the steady state for the first experiment. In figures 1, 2 and 3, the concentration profiles are viewed as parabolas along mesh lines x = xl, l = 0(1) N. The figures 1 and 2 show the flow rate in different subregions of the skin region Rs, before the concentration reaches the steady state. Figure 3 shows the shape of the concentration profile when the concentration reaches steady-state.
Figure 1 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines x = xl, l = 0(1)N; t = 0.5, 1.0, 2.0, 3.0, 4.5, 6.0.
Figure 2 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines x = xl, l = 0(1)N; t = 24.0, 26.0, 28.0, 30.0, 32.0, 34.0.
Figure 3 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines at x = xl, l = 0(1)N; t = 128.7.
In figure 4 and figure 5, for the value of C = 4.4, the concentration profiles at t = 0.5 and 3.0 are viewed as parabolas along the mesh lines y = ym, m = -Qd(1)Qu. Figure 4 and figure 5 show the change in shape of the parabolas in the region away from the region of contact of drug and skin. In figure 6, corresponding to the value of C = 4.4, the concentration profiles at t = = 128.7 are shown along the mesh lines y = ym, m = Qc(1)Qu. The figure 6 shows the importance of two-dimensional modelling.
Figure 4 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines at y = ym, m = -Qd(1)Qu; t = 0.5.
Figure 5 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines at y = ym, m = -Qd(1)Qu; t = 3.0.
Figure 6 Concentration profiles for C = 4.4, viewed as parabolas along mesh lines at y = ym, m = 0(1)Qc; t = 128.7.
Corresponding to each value of n = 1(1)5, the concentration profiles at t = are given in figures 7, 8, 9, 10 and 11, respectively. The above figures show the effect of the value of C on the steady-state concentration CD in different regions of Rs.
Figure 7 Concentration profiles for C = 4.4, when steady-state is reached.
Figure 8 Concentration profiles for C = 19.5, when steady-state is reached.
Figure 9 Concentration profiles for C = 43.1, when steady-state is reached.
Figure 10 Concentration profiles for C = 51.4, when steady-state is reached.
Figure 11 Concentration profiles for C = 64.0, when steady-state is reached.
In figure 12, the concentration profiles for the value of C = 4.4 are given at t = 128.7, 130.7, 132.7, 134.7, 136.7, and 138.7. This graph show the rate of decreasing of the concentration in different regions of Rs after the removal of the vehicle.
Figure 12 Concentration profiles for C = 4.4 viewed as parabolas along mesh lines x = xl, l = 0(1)N; t = 128.7, 130.7, 132.7, 134.7, 136.7, 138.7.
The Ae versus time and J versus time profiles are monitored for the time interval 0 ≤ t ≤ + 50 in each experiment.
The values J and Ae are computed at each time step using the trape-zoidal rule to approximate the integrals in (4) and (5). Thus, to second-order accuracy, for j = 1, 2, 3,... etc,
where
and
In literature see [2], the drug absorption experiments are monitored for a large time interval. Hence it is desirable to use a numerical method which gives acceptable results with larger time steps compared to spacial discretization. Since the numerical method used in this section are L0 stable, it is expected to give accurate results when a large time step is used. Note that the conclusions obtained from all the five experiments are independent of the time step k. As an illustration, with k = 0.1 and k = 0.01, J versus time profiles are monitored for the first experiment (C = 4.4) and are given in figure 15. The time required to reach the steady-state value of the flux with k = 0.1 and k = 0.01 respectively are t = 128.7 and t = 130.37. When the numerical computation is done with k = 0.01, the difference in flux during the time interval [128.7 130.37] is 10-10, which is negligible. It is also noted that even if the drug is removed at t = 130.37 instead of t = 128.7, the conclusions drawn in next section remain valid.
Figure 15 J versus time profiles for C = 4.4 with k = 0.1 and k = 0.01.
Conclusion
Conclusion is presented in five parts.
Part 1:- Conclusions based on concentration profiles till steady-state concentration, CD, is reached
Concentration profiles at various time levels are examined until steady state is reached. Graphically these concentration profiles are represented as parabolas along mesh lines x = xl, l = 0(1) N or parabolas along mesh lines y = ym, m = -Qd(1)Qu. Divide the region of skin, Rs, into two mutually exclusive regions R1 and R2 as follows.
R1 = Rc × Ls and R2 = Rs - R1.
The following conclusions are drawn based on drug kinetics in the regions R1, R2 and Rs.
(1) In figure 1 the concentration profiles are viewed as parabolas along mesh lines x = xl, l = 0(1)N. Consider the flow rate of concentration of the drug in the region R1. From the subplots (a) to (c) of figure 1, it is concluded that during the initial time levels, flow rate towards the skin surface is larger than flow rate towards the skin-capillary boundary. Later-on time levels (from subplots (c) to (f) of figure 1), flow rate towards the skin-capillary boundary becomes larger than flow rate near to the skin surface.
(2) In figure 2 the concentration profiles are viewed as parabolas along mesh
lines x = xl, l = 0(1)N.
From figure 1 and figure 2 it is concluded that in the region R2, the flow rate near to the skin surface is larger than flow rate towards the skin-capillary boundary.
(3) Consider the flow rate in the whole region Rs. From figure 1 and figure 2 it is concluded that, during the initial time levels, flow rate in the region R1 is larger than the flow rate in the region R2. But later-on time levels, flow rate in the region R2 becomes larger than the flow rate in the region R1, leading to a steady-state concentration profile as given in figure 3. In figure 3 the concentration profiles are viewed as parabolas along mesh lines x = xl, l = 0(1)N.
(4) At any time level t = tj, consider the concentration of the drug along the mesh line x = xl, l = 0(1)N. From figure 1, 2 and 3, it is concluded that the concentration of the drug at a mesh point (xl, ym, tj) ∈ R1, m = 0(1)Qc is larger than the concentration of the drug at a mesh point (xl, ym, tj) ∈ R2, m ≠ 0(1)Qc.
(5) In figure 4 and figure 5, the concentration profiles are viewed as parabolas along mesh lines y = ym, m = -Qd(1)Qu. From figure 4 it is concluded that, during the initial time-levels, the shape of the parabolas in the whole region Rs retains the same shape as the shape of the concentration profiles given in page 97 of [2]. If a one-dimensional model of drug absorption was sufficient, the following results are expected from [2].
• When steady-state concentration reaches, the shape of the concentration profiles viewed along the mesh lines y = ym, m = -Qd(1)Qu is same as the shape of the parabolas given in page 97 of [2].
• When steady-state concentration reaches, concentration profiles along the mesh lines y = ym, m = 0(1)Qc are straight lines.
Contrary to this result, the following results are obtained from two-dimensional modelling.
5.1 From figure 4 and figure 5 it is concluded that, as time increases, the parabolas in the region R2 have a convex shape, whereas the parabolas given in page 97 of [2] have a concave shape.
5.2 The concentration profiles along the mesh lines, y = ym, m = 5(1) Qc - 5 are straight lines. These straight lines are shown in figure 6.
5.3 The concentration profiles along the mesh lines, y = ym, m = 0, 1, 2, 3, 4, Qc - 4, Qc - 3, Qc - 2, Qc - 1 and Qc are not straight lines, but they have the shape as shown in figure 6.
Even though the illustration of figures is done only for one experiment (with n = 1, C = 4.4), that is only for the first experiment, the conclusions drawn are true for all other experiments (that is, for all values of C).
Part 2:-Conclusions based on concentration profiles when steady-state value of concentration is reached
The steady-state concentration profiles for all the five experiments are given in figure 7, figure 8, figure 9, figure 10 and figure 11. From these steady-state concentration profiles, the following conclusions are drawn.
(1) As the value of C increases, the concentration at any point (xl, ym, tj) increases.
(2) As the value of C increases, the difference between the concentrations of the drug at mesh points (xl, ym, tj) ∈ R1, m = 0(1)Qc and (xl, ym, tj) ∈ R2, m ≠ 0(1)Qc increases.
(3) As the value of C increases, the difference between the concentrations of the drug at mesh points (x0, ym, tj) and (xN-1, ym, tj) increases for m = -Qd(1)Qc. This increase in difference is more prominent in the region R1 than in the region R2.
Part 3:-Conclusions based on concentration profiles after the removal of the vehicle when steady-state is reached
The vehicle is removed when the concentration reaches steady-state. The concentration profiles are examined for ten more hours, after the vehicle-removal, in an interval of two hours. In figure 12, these concentration profiles are given for n = 1 (C = 4.4), at time levels t = = 128.7, t = 130.7, t = 132.7, t = 134.7, t = 136.7, and 138.7. From figure 12 the following conclusions are made.
(1) After the removal of the vehicle, rate of decrease in the region R1 is larger than that in R2.
(2) After the removal of the vehicle, rate of decrease near to the region of skin-surface is larger than that near to region of skin-capillary boundary. Even though the illustration of figures is done only for one experiment (with n = 1, C = 4.4), that is only for the first experiment, the conclusions drawn are true for all other experiments also.
Part 4:-Conclusions based on figure 13 (flux versus time) and figure 14 (Ae versus time)
Figure 13 J versus time profiles for various values of C.
Figure 14 Ae versus time profiles for various values of C.
The graphs of J versus time and Ae versus time for the time interval 0 ≤ t ≤ ( + 50) are shown in figure 13. The following conclusions are drawn based on figure 13.
(1) As the value of C increases, the steady-state value of flux increases.
(2) As the value of C increases half-life decreases, during the time levels, soon after the vehicle-removal. There is no prolongation of half-life during later-on time, as stated in [2]. It is evident from figure 13 that, the plots of a drug flux versus time is a non-linear curve, soon after the vehicle removal. Hence, in experimental studies even when the prolongation of half-life is not seen, the nonlinearity of drug flux versus time can be introduced as a characteristic to confirm that dual-sorption model gives an explanation to non-linear kinetic behavior of the drug.
The graphs of Ae versus time for the time interval 0 ≤ t ≤ + 50 are shown in figure 14.
(3) As the value of C increases, at any time level t = tj, the value of Ae(t) increases.
(4) As the value of C increases, lag-time decreases. The lag-time is computed as t-intercepts of the linear portion of graphs of Ae versus time. For various values of C, the lag-times are compared with the lag-times given in [2]. Let TL1 denotes the lag-times quoted in [2] as a result of one-dimensional modelling of drug-absorption and TL2 denotes the lag-times obtained as a result of two-dimensional modelling. Table 1 gives the values of TL1 and TL2, corresponding to each value of C.
C 4.4 19.5 43.1 51.4 64.0
TL1 3.44 2.29 1.99 1.93 1.84
TL2 3.67 2.52 2.16 2.11 2.04
From Table 1 it is concluded that, for a particular value of C, the lag-time obtained as a result of two-dimensional modelling is greater than the lag-time obtained as a result of one-dimensional modelling. For a linear model, lag time is independent of donor concentration C. The linear model corresponding to (1) is obtained by substituting = 0. Its lag time is computed as 1.56 which is greater than 1.44, the lag-time obtained from the one-dimensional linear model (substitute = 0 in (8) of [2]).
It is an established fact that the drug permeation profiles of a nonlinear model is different from that of a linear model, as a non-linear model assumes that the drug molecules are either dissolved or immobile inside the skin. In this article it is shown that the drug permeation described by the two-dimensional non-linear model is different from one-dimensional non-linear model, as the two-dimensional model takes into account the drug kinetics at the site distant from the area where the ointment is directly applied.
Acknowledgements
The author thanks DST, Government of India, for the award of a BOYSCAST fellowship.
==== Refs
Chandrasekaran SK Michaels AS Campbell PS Shaw JE Scopolamine permeation through human skin in vito Am Inst Chem Eng J 1976 22 828 832
Gumel AB Kubota K Twizell EH A sequential algorithm for the nonlinear dual-sorption model of percutaneous drug absorption Mathematical Biosciences 1998 152 87 103 9727298 10.1016/S0025-5564(98)10021-4
Kubota K Ishizaki T A calculation of percutaneous drug absorption-I. Theoretical Comput Biol Med 1986 16 7 19 3948495 10.1016/0010-4825(86)90058-2
Kubota K Koyama E Twizell EH Dual sorption model for the nonlinear percutaneous permeation kinetics of timolol J Pharmaceutical Sciences 1993 82 1205 1208
Kubota K Twizell EH Maibach HI Drug release from a suspension with a finite dissolution rate: Theory and its application to a betamethasone 17-valerate patch J Pharmaceutical Sciences 1994 83 1593 1599
George K Kubota K Twizell EH A two-dimensional mathematical model of percutaneous drug absorption BioMedical Engineering OnLine 2004 3 18 15202943 10.1186/1475-925X-3-18
Gumel AB Parellel and sequential algorithms for second-order parabolic equations with applications PhD thesis 1993 Brunel University
Twizell EH Numerical Methods, with applications in the Biomedical Sciences 1998 Ellis Horwood, Chichester and Wiley, Newyork
|
15992411
|
PMC1208919
|
CC BY
|
2021-01-04 16:37:35
|
no
|
Biomed Eng Online. 2005 Jul 3; 4:40
|
utf-8
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-40
|
oa_comm
|
==== Front
Biomed Eng OnlineBioMedical Engineering OnLine1475-925XBioMed Central London 1475-925X-4-511613892110.1186/1475-925X-4-51ResearchMining for diagnostic information in body surface potential maps: A comparison of feature selection techniques Finlay Dewar D [email protected] Chris D [email protected] Paul J [email protected] Norman D [email protected] School of Computing and Mathematics, Faculty of Engineering, University of Ulster, Shore Road, Belfast, UK2005 2 9 2005 4 51 51 11 7 2005 2 9 2005 Copyright © 2005 Finlay et al; licensee BioMed Central Ltd.2005Finlay et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In body surface potential mapping, increased spatial sampling is used to allow more accurate detection of a cardiac abnormality. Although diagnostically superior to more conventional electrocardiographic techniques, the perceived complexity of the Body Surface Potential Map (BSPM) acquisition process has prohibited its acceptance in clinical practice. For this reason there is an interest in striking a compromise between the minimum number of electrocardiographic recording sites required to sample the maximum electrocardiographic information.
Methods
In the current study, several techniques widely used in the domains of data mining and knowledge discovery have been employed to mine for diagnostic information in 192 lead BSPMs. In particular, the Single Variable Classifier (SVC) based filter and Sequential Forward Selection (SFS) based wrapper approaches to feature selection have been implemented and evaluated. Using a set of recordings from 116 subjects, the diagnostic ability of subsets of 3, 6, 9, 12, 24 and 32 electrocardiographic recording sites have been evaluated based on their ability to correctly asses the presence or absence of Myocardial Infarction (MI).
Results
It was observed that the wrapper approach, using sequential forward selection and a 5 nearest neighbour classifier, was capable of choosing a set of 24 recording sites that could correctly classify 82.8% of BSPMs. Although the filter method performed slightly less favourably, the performance was comparable with a classification accuracy of 79.3%. In addition, experiments were conducted to show how (a) features chosen using the wrapper approach were specific to the classifier used in the selection model, and (b) lead subsets chosen were not necessarily unique.
Conclusion
It was concluded that both the filter and wrapper approaches adopted were suitable for guiding the choice of recording sites useful for determining the presence of MI. It should be noted however that in this study recording sites have been suggested on their ability to detect disease and such sites may not be optimal for estimating body surface potential distributions.
==== Body
Background
Although extensively utilised, the limitations of the conventional 12-lead ECG for optimal detection of cardiac abnormalities are widely appreciated [1]. The main deficiency in the 12-lead approach is the fact that only 6 chest electrodes are incorporated which cover a relatively constrained area of the precordium. The main reason for the choice of the location of the conventional precordial electrodes, suggested by Wilson over 70 years ago [2,3], was the need to adopt some standard which to this day has remained relatively unchallenged. In the years since then, the growing appreciation for the limitations of the conventional precordial electrode positions and the increase in understanding of the localisation of various cardiac abnormalities on the body surface has led to the suggestion of various alternatives. One of the most widely studied alternatives to the 12-Lead ECG in both clinical and experimental electrocardiology has been the BSPM. In this approach, anything between 32 and 219 electrodes [4] are used in an attempt to sample all electrocardiographic information as projected onto the body's surface. The merits of this enhanced spatial sampling are obvious, in that, localised abnormalities that are perhaps difficult to detect using the 12-lead approach can readily be picked up with the additional electrodes. As well as this ability to provide more diagnostic information, BSPMs facilitate an alternative method for visualisation as recorded data can be displayed as a sequence of contour maps, allowing isolation of significant electrocardiographic events in both space and time.
In body surface potential mapping, many of the inadequacies associated with the conventional 12-lead approach are addressed, but despite this, clinical utilisation outside the research laboratory is close to negligible. Several reasons exist for this lack of clinical uptake. Kors et al. [5] suggests the opposing interests of two groups as the main motivating factor: on one hand researchers are keen to further the diagnostic potential of the ECG using all information on the body surface, on the other hand, clinicians satisfied with the standard 12-lead ECG are reluctant to replace it with one of perceived technological complexity. This complexity stems from the requirement to sample dozens of channels of ECG information simultaneously, with the application of the associated number of chest electrodes viewed as highly impractical, particularly in acute care.
To address the impracticalities associated with high density spatial sampling procedures, investigators are interested in exploiting the redundancy in BSPMs to suggest 'limited lead' systems. This process can be succinctly described as locating the minimum number of recording sites required to capture the maximum amount of ECG information [6]. Although in all studies the number and location of sites is the focus, there are two main ways in which 'optimality' can been quantified. These are:
1. Sites that provide the maximum diagnostic information allowing enhanced discrimination between abnormalities.
2. Sites that allow the most accurate estimation of the heart's activity at other sites where information has not been recorded.
The first significant work on this problem of selecting limited leads was conducted by Barr et al. [7] who proposed a technique based on principal component analysis to locate 24 recording sites that allowed consistent representation of the total body surface potential during depolarisation (over the QRS). Subsequent to this, and based on a more representative dataset, Lux et al. [8,9] analysed correlation and covariance in 192 lead BSPMs to suggest 32 optimal recording sites that could be used to reconstruct the original BSPM frames with a level of error that was consistent with the estimated system noise, which, in the studied dataset was estimated to be 20 μV. In both sets of studies [7-9], the recording sites were chosen based on the ability to estimate potentials at sites that were not recorded. Kornreich et al. [10-12] on the other hand conducted several studies that suggested electrode configurations that were most suitable for diagnosing a range of abnormalities; in each study the objective was to find the best discriminating recording sites. More recently Kors et al. [5] conducted a study where the positions of the 6 precordial electrodes in the conventional ECG were altered intuitively to provide greater information capture. In this study, it was found that two of the standard precordial leads could be re-positioned to provide greater information capture whilst the remaining precordial electrodes could accurately reconstruct the 12-lead ECG. Although well validated in the research literature, the lead systems proposed by all of these studies have never been widely accepted in clinical practice.
Regardless of the rationale for the choice of recording sites, early investigators were limited to mathematical and statistical techniques in the selection process, exploiting phenomena such as correlation and variance in the recorded signals. Although yielding acceptable results, new techniques that have emerged through the proliferation of domains such as data mining and knowledge discovery [13] may provide greater insight into the process of lead selection. The process of lead selection is itself analogous to that of 'feature selection' which is a term commonly used in the aforementioned domains of data mining and knowledge discovery to describe the elimination of redundant variables in a dataset [14]. Ideally, feature selection would involve exhaustively evaluating all possible combinations of input features and choosing the best subset, but in reality the computational cost of this is prohibitive. For this reason, much research effort has been directed at developing algorithms and strategies that locate optimal features at low computational cost. In the current study commonly used feature selection methodology has been applied to the BSPM domain in order to select electrode subsets that are best for discriminating between normal subjects and those with MI.
In many domains, determining the features or variables that are relevant can give a useful insight into the nature of the prediction problem at hand [15]. This is particularly the case in the BSPM domain as information on the location of the best recording sites may complement information or understanding for those interested in the underlying cardiac behaviour. To provide both information that could be used to better understand cardiac function as well as information that is useful for computerised classification, two common feature selection techniques, 'filters' and 'wrappers', were employed in the current study.
Filters
In the filter approach, the predictive performance of each individual feature is assessed and features that are deemed unnecessary are 'filtered' out [16]. The resulting measure of performance for each feature is used in a ranking process and the desired number of features n, with the highest scores, are selected. Although in this description, and throughout the current study, consideration is given to features on an individual basis only, subsets of features can also be considered for inclusion or elimination. The main evaluation techniques include correlation analysis (determining how each variable correlates with the target variable) and discrimination analysis (determining how each variable acts as a classifier). This process is often described as classifier independent as although a single variable classifier may be used in the selection of each feature, the selected feature subset as a whole is not chosen to suit any classifier.
Wrappers
In this approach, the aim is to find the best subset of features by testing them with the classifier that they are intended for use with. As well as picking features to suit the classifier, this approach is advantageous in that features are selected based on how well they work together with other features. The process begins with the suggestion of a feature subset which is evaluated using the desired classifier, this feature subset is repeatedly modified (features are added and/or removed) and evaluated until the desired level of performance is attained. The two most common methods used to generate the feature subsets are SFS and Sequential Backward Elimination (SBE) [17]. In SFS the process begins with an empty set of features. In the first iteration, all feature subsets containing only one feature are evaluated i.e. the performance of each individual feature is evaluated. The feature with the highest accuracy is used as the basis for the next iteration, where, each remaining feature is evaluated in conjunction with the previously selected feature. The feature that performs best in conjunction with the first feature is selected forming the basis for the next iteration. This process is repeated until all features have been incorporated in the subset, or the desired level of accuracy has been reached. In SBE the process begins with a subset that contains all available features, from which features are removed one by one based on the performance of the remaining features. In theory, going backwards from the full set of features in the data may capture interacting features more easily; however this is at greater computational expense as building classifiers where there are few features in the dataset is much faster [18]. At this point it is worth underlining that the main difference between the filter and wrapper approach is that, in the filter approach, no feedback is used from the classification model.
Methods
From a practical perspective, the most widely used technique in locating recording sites or lead subsets is to start off with a dataset recorded using a full BSPM electrode array and try to find the best recording sites in that array that satisfy the desired criteria. The locations of these 'best sites', which are a subset of the original BSPM array, are then proposed as the new limited lead set. In the current study, BSPMs recorded using an electrode array of 192 electrodes were used. This dataset is of similar genre in terms of acquisition format as that used in [8,9]. The layout of the electrodes is depicted in the schematic in Figure 1. In this configuration electrodes are placed in equally-spaced columns, each consisting of 12 electrodes, around the thoracic circumference. For each subject, the 192 channels of ECG information were sampled simultaneously at 1000 Hz for a duration of several seconds. Subsequent to recording, this data was reduced to represent just one cardiac cycle which on average consisted of 600 milliseconds of information. This was achieved through RMS averaging. In all, maps recorded from 116 subjects were studied. Within this group of subjects, 59 were considered as normal exhibiting no disease symptoms, the remaining 57 had electrocardiographic evidence of old MI with lesions at various locations. Table 1 describes the composition of the dataset with a breakdown of the various infarct locations. Following recording, the map sequences from each patient were processed to provide QRS, STT, and QRST isointegral features. This is common practice in the BSPM domain and involves reducing a sequence of map frames to one single map. Often referred to as isoarea mapping, the process involves summing all the potentials under a specific portion of the ECG trace at each recording site, and plotting a contour map using the resulting values. In this study, QRS and STT isointegrals were calculated to illustrate mean potential distributions during depolarisation (activation) and repolarisation (recovery) respectively, and QRST isointegrals were calculated to illustrate mean potential distributions during depolarisation and repolarisation combined. Each isointegral consists of 192 values which are used to generate a contour map. Figure 2 depicts an example of such a map, in this case a QRS isointegral map for a subject with Inferior wall MI (IMI). The pattern of extrema, maxima and minima of such a map are studied by the clinician in order to provide diagnosis and because the pattern is characterised by the 192 calculated values, these values can be used as inputs (features) to a computerised classifier. As we have calculated 3 such maps, we effectively end up with 576 features for each subject (3 × 192). This also translates to having 3 features per recording site per patient, e.g. for each recording site we have one QRS, one STT, and one QRST value. In all experiments, we combine the three set of isointegral measurements for each patient, resulting in 576 features per patient.
Figure 1 192 Electrode Array. Schematic representation of the 192 electrode array, depicted as an unrolled cylindrical matrix. The middle region correspond with the anterior torso and the left and right regions correspond with the posterior.
Table 1 Composition of dataset describing infarct locations.
Normals 59
Myocardial Infarction 57
Inferior 30
Anterior 14
Posterior 2
Aterolateral 8
Inferolateral 2
Inferior-posterior 1
Total 116
Figure 2 Example QRS isointegral map. Example of a QRS isointegral map derived from subset with Inferior wall MI (IMI). The areas of positive and positive and negative polarity are indicated with red and blue circles respectively.
Single variable classifier approach
In this study, the prediction accuracy of each individual feature was assessed using a Nearest Neighbour (NN) classifier [19]. From this point onwards we shall refer to this as the SVC approach as this is effectively a special case of the filter approach. Using Leave One Out (LOO) cross validation, the SVC was used to evaluate the performance of each individual feature on a randomly selected subset of 87 of the 116 original recordings. In LOO cross validation, the performance of the given feature is obtained by using all records bar one (86 recordings) to train the classifier and using the remaining record to evaluate the classifier. This process is repeated until every subset of one record is evaluated, the final performance being obtained through taking an aggregate of each individual classification outcome. The remaining 29 records were treated as a hold out set and were used to test the classification accuracy of various sets of the highest ranked features again in conjunction with a NN classifier. The same features were also tested using classifiers based on 5 Nearest Neighbour (5NN) and Logistic Regression (LR), this was performed to assess the chosen features on a range of classifiers.
Sequential forward selection wrapper approach
In the current study, the high ratio of features to cases (observations) also becomes an issue in SBE, as most classifiers will exhibit instability when the number of features greatly exceeds the number of variables. For this reason the SFS approach was adopted, and applied using three separate classifiers (LR, NN, 5NN) which resulted in three feature sets each of which were classifier specific. In order to reduce the computational time associated with the procedure the SFS process was invoked for 100 iterations, resulting in selection of the 100 best features. To validate the results, the data set of 116 records was again randomly partitioned into a training set of 87 records and a test set of 29 records (same as above). The training set was then used to conduct the SFS process, and LOO cross validation was used on each iteration of the feature subset selection process. When the selection process was completed, the training set was used to develop a classifier using the selected features, which was then validated using the unseen test set. This approach was taken to alleviate optimistic biasing of the results as opposed to some studies, as pointed out by [18] that use the classification accuracy of the selection runs as a measure of final performance. A flow diagram illustrating the wrapper approach implemented in these experiments is shown in Figure 3.
Figure 3 Wrapper Based Feature Selection. Illustration of the wrapper based feature selection approach where the feature subset is generated by iterative evaluation of available features.
In addition to performing the experiments outlined above, a simple procedure was conducted to illustrate the localisation of difference between normal and MI populations in the training set. This procedure involved subtracting the mean of the normal population from the mean of the MI population and presenting the resulting difference as a contour map of the absolute values at each recording site. For each isointegral (QRS, STT, QRST) 'difference' maps were calculated and the results are illustrated in Figure 4. These difference maps have been generated merely for comparison with later results.
Figure 4 Difference Maps. Spatial representation of the absolute difference between the means of the normal and MI population for (a) QRS, (b) STT, and (c) QRST isointegrals. Magnitude of difference increases as the colour changes from dark blue through to dark red.
Results
SVC approach
The classification accuracy obtained for each individual feature was used to rank the features in descending order of significance. In cases where two or more features had an identical score these features were ordered arbitrarily. A method of Lead Performance Maps (LPMs) similar to that adopted in [20] was used to graphically illustrate the spatial distribution of feature performance. In this approach a contour map is produced using the feature performance values positioned with the same layout of the original electrode array, the contour map is then shaded through linear interpolation between the 192 data points. In this study this resulted in three separate LPMs (for QRS, STT, QRST isointegrals) which are illustrated in Figure 5.
Figure 5 Lead Performance Maps. Spatial distributions of the prediction accuracy at each of the 192 recording sites for (a) QRS, (b) STT, and (c) QRST isointegrals. Magnitude of difference increases as the colour changes from dark blue through to dark red.
As the overarching goal of this study was to assess the performance of limited electrode arrays the classification performance of the best 3, 6, 9, 12, 24, and 32 recording sites, deduced from the top ranked features, were evaluated. The choice of an upper limit of 32 is synonymous with the number of leads in the 'clinically practical' lead system proposed by Lux in [9]. It is worth pointing out that for each of the lead subsets it was possible to have more than the corresponding number of features as in some cases multiple features were selected for each lead. A schematic illustrating the positions of the top 32 recording sites is illustrated in Figure 6. For the sake of comparison with more conventional lead systems, the best 6 electrodes out of the 32 have also been highlighted. The classification accuracy of the various lead subsets is listed in Table 2. As well as classifications accuracies, measures for sensitivity and specificity have been included with all results.
Figure 6 Electrode Locations (SVC Approach). Locations of the 32 top recording sites as chosen using the variable ranking method. The first 6 sites chosen are highlighted using the filled circles; the filled squares show the remaining 26 sites.
Table 2 Subset Performance-SVC Approach. The performance of subsets of leads chosen using the SVC approach, evaluated using three classifiers.
No. Recording Sites NN 5NN LR
ACC (%) SEN (%) SPE (%) ACC (%) SEN (%) SPE (%) ACC (%) SEN (%) SPE (%)
3 72.4 85.7 80.0 69.0 64.3 73.3 72.4 78.6 66.7
6 82.8 71.4 93.3 72.4 64.3 80.0 75.9 78.6 73.3
9 79.3 85.7 73.3 69.0 71.4 66.7 72.4 78.6 66.7
12 79.3 78.6 80.0 75.9 78.6 73.3 65.5 78.6 53.3
24 79.3 78.6 80.0 79.3 71.4 86.7 79.3 85.7 73.3
32 79.3 78.6 80.0 79.3 71.4 86.7 79.3 85.7 73.3
SFS approach
Because of the nature of the SFS approach, where the classification accuracy of each feature is determined accumulatively, it is not possible to display the individual classification accuracy of each feature using LPMs. What is possible, and probably of more significance from a raw computer science perspective, is the display of the performance of the SFS algorithm as features are accumulatively selected. This is depicted in Figure 7 where a graph for each wrapper based on a different classifier is shown. In these three graphs, the performance of the classifier in the feature evaluation and selection process is illustrated with the solid line, and the performance of the same classifier and feature set with unseen data is illustrated with the dashed line. In Figure 8, the best 32 leads chosen by each wrapper have been schematically represented as previously described, and as conducted for the filter approach the classification accuracy of the various lead subsets was also evaluated as presented in Table 3.
Figure 7 Wrapper Performance. Performance of the three SFS based wrappers using (a) NN classifier, (b) 5NN classifier, and (c) LR classifier. Each plot shows the classification accuracy as features are added during the selection process (train): And, the performance of selected features on an unseen test dataset (test).
Figure 8 Electrode Locations (SFS Approach). Recording sites chosen by SFS algorithms using (a) NN, (b) 5NN, and (c) LR. On each layout the top 32 recording sites have been illustrated. Additionally the top 6 sites which are a subset of the 32 are also shown (filled circles).
Table 3 Subset Performance-SFS Approach. Respective classification accuracy of recording sites chosen using NN, 5NN and LR wrappers.
No. Recording Sites NN 5NN LR
ACC (%) SEN (%) SPE (%) ACC (%) SEN (%) SPE (%) ACC (%) SEN (%) SPE (%)
3 79.3 71.4 86.7 72.4 71.4 73.3 79.3 85.7 73.3
6 72.4 71.4 73.3 79.3 78.6 80.0 82.8 92.9 73.3
9 72.4 50.0 93.3 79.3 78.6 80.0 75.9 92.9 60.0
12 75.9 64.3 86.7 75.9 78.6 73.3 72.4 85.7 60.0
24 79.3 78.6 80.0 82.8 85.7 80.0 69.0 78.6 60.0
32 79.3 71.4 86.7 82.8 85.7 80.0 72.4 71.4 73.3
Discussion
The main observation from the preliminary difference map calculation illustrated in Figure 4 is that for all features per three maps there is significant localisation in the precordial region, indicating that on average these recording sites exhibit the most difference between normal and subjects with MI. Although this is useful in illustrating the significant localisation of difference between populations, this technique has limitations in lead selection as pointed out by Kornreich et al. [10], who suggested that the discriminant ability of each recording site is not truly reflected as intra-group variability is not taken into consideration.
The variable ranking procedure was primarily implemented to allow the selection of recording sites for evaluation in simple classification problems, and in this study the classification accuracy of 3, 6, 9, 12, 24, and 32 recording sites has been evaluated using three classifiers. From the results presented in Table 2, it can be observed that when using 24 recording sites, each classifier is capable of a predication accuracy of almost 80% which remained the same when the number of recording sites was increased to 32. Indeed, for all the configurations of recording sites evaluated, with the exception of one (12 sites with the LR classifier), a reasonable level of classification accuracy was observed. One characteristic of the results that deserves further explanation is the fact that the classification accuracy does not always increase, and sometimes decreases when the number of recording sites is increased. This is possibly due to the fact that in the selection procedure no consideration is given to the selection of variables that are correlated. In some classifiers correlation between variables can lead to deterioration in performance. It should also be noted that in some cases the seemingly coarse variations in performance, for example a change in 82.8% to 79.3% between 6 and 9 recording sites for the NN based SVC classifier, are largely attributable to the relatively modest size of the validation dataset. Evident in this particular example with a deterioration in performance of 3.5% through the misclassification of just one subject.
As well as providing useful information for classifier development this variable ranking also lends itself well to graphical display of the spatial distribution of the diagnostic features as is demonstrated in the LPMs in Figure 5. These LPMs show the areas of the torso where the classification accuracy of each individual feature is greatest. It can also be observed that the LPMs bear some resemblance to the difference maps that are displayed in Figure 4, in particular the corresponding maps for STT and QRST isointegrals bear significant resemblance with respect to the locations of areas of high and low significance.
In terms of the actual recording sites that were chosen using the SVC approach it can be seen that the upper right chest and back, and the lower left abdominal regions exhibit the highest concentrations of suggested sites. This of course is indicative of the information presented in the difference maps and the LPMs, however one interesting point is the fact that the six best electrodes are located in the upper right chest and back as opposed to the left precordial region as is the case with conventional electrocardiographic leads.
The performance of the three SFS algorithms depicted in Figure 7 gives a good indication of the general performance of this type of algorithm, as in general there is an increase in classification accuracy as features are initially added which is followed by a fluctuation, or decrease in accuracy as subsequent features are added. This is similar to the responses observed in other studies where non-electrocardiographic data has been used [15,21]. It can be observed that for all three wrappers, there is a consistent increase in performance on the train data as features are added, and in the NN and 5NN based approaches there is no significant deterioration in performance on the training data within the scope of the conducted experiments (up to 100 features). In the case of the LR based wrapper there is a less consistent response as performance deteriorates at approximately 20 features. This can be attributed to the fact that LR based classifiers exhibit significant instability when the number of variables is excessive compared with the number of cases. Although there is no definitive measure for predicting when this instability might occur, it can be seen here that 20 variables with 86 cases, a ratio of 10/43 induced this phenomenon. With regard to the performance associated with the test data, it can be seen that in some cases, particularly the NN based wrapper, there is not always consistent increase in performance as features are added. Again this is similar to observations in other studies [15,21] and appears to be a trait of the SFS approach
The inclusion of the sensitivity and specificity figures, as defined in [22], also provides some insight into how each set of features perform. These figures, which give an indication of how well subjects from each group (MI and Normal) have been correctly classified, were not taken into consideration during the selection process. The first observation is that the SFS selection approach using the 5NN classifier seems to provide the best balance between the two measures, as for each lead subset there is typically just a few percent difference between the measures. This is in contrast to some of the other selection/classifier combinations, for example the SFS/NN combination at 9 recording sites, where the sensitivity was 50% but the specificity was 93.3%. The reason for such an imbalance is likely to be attributable to the fact that these measures were not taken into consideration as features were being selected. If it were necessary to give preference to either sensitivity of specificity then consideration would have to be given to these variables during the selection process.
The classification accuracies obtained using the SFS wrappers (Table 3) show that this technique is as good as and in some cases better than the variable ranking procedure discussed above. In particular the performance of the 5NN classifier using features obtained from 24 and 32 recording sites is consistently better than the corresponding numbers of recording sites selected using the variable ranking approach. This can be attributed to two factors that make the wrapper approach superior in most applications. Firstly, as features are selected, their evaluation is based on how well they work together as a sub set, thus reducing the likelihood of selecting multiple features that are highly correlated and effectively measuring the same thing. Secondly, the characteristics of the final classifier are considered in the feature selection process therefore the chosen features are classifier specific. This second point accounts for the fact that although measuring sites were located on the same general areas of the torso as can be seen in Figure 8, there is significant discrepancy between the exact measuring sites chosen for each of the three wrappers. To illustrate just how specific the chosen recording sites are to each individual classifier a simple experiment was conducted. In this experiment the recording sites and hence features that were chosen by the 5NN based wrapper were tested on a LR classifier. The results are listed in Table 4. Here it can be seen that there is significant deterioration in the classification accuracy as the number of recording sites increases, illustrating that the features chosen for the 5NN classifier do not work anywhere as well for the LR classifier.
Table 4 Subset Performance-Cross Comparison. Classification accuracy of recording sites selected for 5NN classifier, evaluated on LR classifier.
No. Recording Sites ACC (%) SEN (%) SPE (%)
3 75.9 85.7 66.7
6 75.9 85.7 66.7
9 72.4 85.7 60.0
12 69.0 85.7 53.3
24 69.0 85.7 53.3
32 65.5 78.6 53.3
A further issue that we wished to investigate was the notion of 'uniqueness' introduced in [8]. In the study the authors suggested that there was no unique set of 'optimal leads' for reconstruction of BSPMs. Fuelled by the diversity of recording sites suggested in each of the techniques investigated in this study we wanted to find out if the same were true for recording sites that were selected for classification. To test for this condition, the experiment using the SFS wrapper based on the 5NN classifier was repeated. Under normal conditions this algorithm starts off by choosing the best individual feature as it makes its first pass through the list of available features. This time, however, the algorithm was forced to pick a non-optimal feature the first time around; subsequently the algorithm reverted back to normal operation choosing the best combinations from that point on. The 32 measuring sites chosen in this approach are illustrated in Figure 9, and the number of correctly classified cases are shown in Table 5. It can be seen that the pattern of chosen recording sites is not dissimilar to those for the original 5NN classifier based approach (Figure 8b), however, the exact location of each electrode differs in many cases, particularly within the case of the first 6 chosen electrodes as only one of these occupies the same position as in the original experiment. Consulting the table of classification accuracies it can be observed this rearrangement of electrodes does not necessarily come at a cost, indeed in the case of 9 and 12 recording sites a superior rate of classification is in fact achieved. These results would tend to suggest that like in the best recording sites chosen for reconstruction [8], there is no unique best set of electrodes for classification.
Figure 9 Electrode Locations (Forced SFS Approach). Recording sites chosen by SFS algorithms using 5NN when the first selected feature is forced.
Table 5 Subset Performance-Forced SFS Approach. Classification accuracy of recording sites chosen using the 5NN based wrapper approach with a forced initial selection.
No. Recording Sites ACC (%) SEN (%) SPE (%)
3 79.3 78.6 80.0
6 75.9 71.4 80.0
9 86.2 85.7 86.7
12 86.2 85.7 86.7
24 82.8 78.6 86.7
32 75.9 71.4 80.0
Conclusion
In this study previously unutilised techniques have been applied to the perennial problem of selecting optimal recording sites from BSPMs. The filter and wrapper approaches to feature selection have been studied and have been found useful in locating useful diagnostic information. As well as evaluating the diagnostic capabilities of selected lead subsets, the study demonstrated how features chosen using the wrapper approach are specific to the given classifier. In addition, the study also introduced the notion of the non-uniqueness of selected lead subsets by suggesting two relatively different lead subsets using the same method, which exhibited comparable performance. Although the relatively small data set does limit the findings of the study, it is believed that the results obtained show that these techniques could be applied to a larger population of recordings to suggest limited lead systems that could be used in clinical practice.
A subtle limitation to the study is the fact that features used mainly in the BSPM domain are used to evaluate the classification performance of limited lead sets. These features are the isointegral measurements already mentioned. In reality, if a limited number of leads were used there may be a more effective way for representing the underlying ECG information, as isointegral maps were adopted mainly to counter the effect of excessive amounts of spatial data. This limitation does not prohibit the comparison of results within the study, however, it makes benchmarking with studies conducted by other investigators more difficult. To this end, if the study were to be extended it may be useful to investigate the use of temporal features such as wave amplitudes and durations as would be more common in conventional 12-lead electrocardiography.
Finally, the study has investigated the application of knowledge discovery to the problem of locating diagnostic information in BSPMs and the tools studied could be used to guide the development of recording system configurations. A clear distinction exists however between this approach and that where lead systems are developed to allow reconstruction of BSPMs. There is no evidence to suggest that recording sites selected based on their ability to discriminate between normal and abnormal are optimal for estimating potential distributions on the entire body surface, hence this area warrants further investigation.
Authors' contributions
DDF was responsible for designing the study, conducting all experimental procedures, and along with CDN, interpreting and presenting the results. Both DDF and CDN participated equally in drafting the manuscript. PJM and NDB participated in the design of the study and also helped to draft the manuscript.
Acknowledgements
The authors would like thank Prof. R. Lux of the Nora Eccles Harrison Cardiovascular Research and Training Institute, University of Utah, USA, for providing the clinical data and guidance required to realise this study.
==== Refs
Carley SD Beyond the 12 Lead: Review of the use of additional leads for the early electrocardiographic diagnosis of acute myocardial infarction Emergency Medicine 2003 15 143 154 12675624
Wilson FN Johnston FD Macleod AG Barker PS Electrocardiograms that represent the potential variations of a single electrode American Heart Journal 1934 9 447 458 10.1016/S0002-8703(34)90093-4
Wilson FN MacLeod AG Barker PS Electrocardiographic leads which record potential variations produced by the heartbeat at a single point Proceedings of the Society of Experimental Medicine 1932 29 1011 1012
Hoekema R Uijen GJH van Oosterom A On selecting a Body Surface Mapping Procedure Journal of Electrocardiology 1999 32 93 101 10338028 10.1016/S0022-0736(99)90088-2
Kors JA van Herpen G How many electrodes and Where: A "Poldermodel" for Electrocardiography Journal of Electrocardiology 2002 7 12 12539094 10.1054/jelc.2002.37149
Lux RL Leads: How Many and Where? Journal of Eletrocardiology 2002 213 214
Barr RC Spach MS Herman-Giddens S Selection of the number and position of measuring locations for electrocardiography IEEE Transactions on Biomedical Engineering 1971 18 125 138 4930411
Lux RL Smith CR Wyatt RF Abildskov JA Limited Lead Selection for the Estimation of Body Surface Potential Maps in Electrocardiography IEEE Transactions on Biomedical Engineering 1978 25 270 276 680756
Lux RL Burgess MJ Wyatt RF Evans AK Vincent GM Abildskov JA Clinically Practical Lead Systems for Improved Electrocardiography: Comparison with Precordial Grids and Conventional Lead Systems Circulation 1979 59 356 363 759003
Kornreich F Rautaharju PM Warren J Montague TJ Horacek BM Identification of Best Electrocardiographic Leads for Diagnosing Myocardial Infarction by Statistical Analysis of Body Surface Potential Maps American Journal of Cardiology 1985 56 852 856 4061325 10.1016/0002-9149(85)90768-4
Kornreich F Montague TJ Rautaharju PM Block P Warren JW Horacek MB Identification of Best Electrocardiographic Leads for Diagnosing Anterior and Inferior Myocardial Infarction by Statistical Analysis of Body Surface Potential Maps American Journal of Cardiology 1986 58 863 871 3776844
Kornreich F Montague TJ Kavadias M Segers J Rautaharju PM Horacek MB Taccardi B Qualitative and Quantitative Analysis of Characteristic Body Surface Potential Map Features in Anterior and Inferior Myocardial Infarction American Journal of Cardiology 1987 60 1230 1238 3687774 10.1016/0002-9149(87)90600-X
Fayyad U Piatetsky-Shapiro G Smyth P From data mining to knowledge discovery in databases (a survey) AI Magazine 1996 17 37 54
Dash M Liu H Feature Selection for Classification Intelligent Data Analysis 1997 1 131 156 10.1016/S1088-467X(97)00008-5
Reunanen J Overfitting in making comparisons between variable selection methods Journal of Machine Learning Research 2003 3 1371 1382 10.1162/153244303322753715
Hall M Smith L Kasabov N Feature Subset Selection: a Correlation Based Filter Approach Proceedings of the 1997 International Conference on Neural Information Processing and Intelligent Information Systems: New Zealand 1998 Springer 855 858
Cantú-Paz E Deb K Feature subset selection, class separability, and genetic algorithms Proceedings of the Genetic and Evolutionary Computation Conference (GECCO): 26–30 June 2004; Seattle 2004 Springer Verlag 959 970
Kohavi R John G Wrappers for Feature Subset Selection Artificial Intelligence Journal 1996 97 273 324
Cover TM Hart PE Nearest neighbour pattern classification IEEE Transactions on Information Theory 1967 13 21 27 10.1109/TIT.1967.1053964
Lopez JA Nugent CD van Herpen G Kors JA Finlay DD Black ND Visualisation of body surface electrocardiographic features in Myocardial Infacrtion Journal of Electrocardiology 2004 149 15286927 10.1016/j.jelectrocard.2004.08.041
Loughrey J Cunningham P Bramer M, Coenen F, Allen T Overfittingur in Wrapper-Based Feature Subset Selection: The Harder You Try the Worse it Gets Proceedings of the 24th SGAI International Conference on Innovative Techniques and Applications of Artificial Intelligence: 13–15 December 2004; Cambridge 2004 Springer 33 43
Rautaharju PM Blackburn HW Warren JW The concepts of sensitivity, specificity and accuracy in evaluation of electrocardiographic, vectorcardiographic and polarcardiographic criteria Journal of Electrocardiology 1976 9 275 281 781162
|
16138921
|
PMC1208920
|
CC BY
|
2021-01-04 16:37:35
|
no
|
Biomed Eng Online. 2005 Sep 2; 4:51
|
utf-8
|
Biomed Eng Online
| 2,005 |
10.1186/1475-925X-4-51
|
oa_comm
|
==== Front
Cancer Cell IntCancer Cell International1475-2867BioMed Central London 1475-2867-5-261607899410.1186/1475-2867-5-26Primary ResearchRapid in vivo Taxotere quantitative chemosensitivity response by 4.23 Tesla sodium MRI and histo-immunostaining features in N-Methyl-N-Nitrosourea induced breast tumors in rats Sharma Rakesh [email protected] Richard P [email protected] Ed X [email protected] Jose K [email protected] Department of Medicine, W168th Street, Columbia University, New York, NY 10032 USA2 Department of Radiology, W168th Street, Columbia University, New York, NY 10032 USA2005 3 8 2005 5 26 26 21 8 2004 3 8 2005 Copyright © 2005 Sharma et al; licensee BioMed Central Ltd.2005Sharma et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Sodium weighted images can indicate sodium signal intensities from different features in the tumor before and 24 hours following administration of Taxotere.
Aim
To evaluate the association of in vivo intracellular sodium magnetic resonance image intensities with immuno-biomarkers and histopathological features to monitor the early tumor response to Taxotere chemotherapy in Methyl-Nitroso-Urea induced rat xenograft breast tumors.
Methods and Materials
Methyl-Nitroso-Urea (MNU) induced rat xenograft breast tumors were imaged for sodium MRI and compared with tumor histology, immunostaining after 24 hours chemotherapy.
Results
Sodium MRI signal intensities represented sodium concentrations. Excised tumor histological sections showed different in vitro histological end points i.e. single strand DNA content of cell nuclei during cell cycle (G1/S-G2/M), distinct S or M histograms (Feulgen labeling to nuclear DNA content by CAS 200), mitotic figures and apoptosis at different locations of breast tumors. Necrosis and cystic fluid appeared gray on intracellular (IC) sodium images while apoptosis rich regions appeared brighter on IC sodium images. After 24 hours Taxotere-treated tumors showed lower 'IC/EC ratio' of viable cells (65–76%) with higher mitotic index; apoptotic tumor cells at high risk due to cytotoxicity (>70% with high apoptotic index); reduced proliferation index (270 vs 120 per high power field) associated with enhanced IC sodium in vivo MR image intensities and decreased tumor size (3%; p < 0.001; n = 16) than that of pre-treated tumors. IC-Na MR signal intensities possibly indicated Taxotere chemosensitivity response in vivo associated with apoptosis and different pre-malignant features within 24 hours of exposure of cancer cells to anti-neoplastic Taxotere drug.
Conclusion
Sodium MRI imaging may be used as in vivo rapid drug monitoring method to evaluate Taxotere chemosensitivity response associated with neoplasia, apoptosis and tumor histology features.
Apoptosisantineoplasticsnon-invasivetherapeutic efficacyTaxotere
==== Body
Introduction
In tissue, sodium exists as total extracellular sodium (EC-Na) ~ 139 mM occupying ~0.15 extracellular water spaces and bound intracellular sodium (IC-Na) ~ 15 mM concentrations. The sodium nuclei exhibit slower and faster relaxation times respectively [1]. Tissue sodium concentrations may change during early tumor stages without membrane damage. Heterogeneous tumors showed the enhanced sodium MR image signal and sodium concentration in tumors [2,3]. Increased intracellular sodium along with histopathology and apoptosis related cytomorphological changes can provide the real time information of antineoplastic effects to optimize chemotherapeutic efficacy at the onset [2,4,5].
Docetaxal has gained attention for its cytotoxic effect in cancer prevention. Rapid time-dependent monitoring of docetaxal chemosensitivity by sodium MRI is emerging as clinical tool in cancer therapy [2,5].
Recently, MNU induced breast tumors in rats showed distinct pre-malignant stages such as lymph node invasion, ductal carcinoma, cell proliferation, apoptosis, cyclin D1/p27 expression, codon 12 mutation frequencies in Ha-ras, microsatellite instability and apoptosis [6]. Cytopathological response and in vivo sodium MRI measured the chemosensitivity of primary tumor to drug as first estimate of neoplasm and metastatic sensitivity [7].
For in vivo sodium imaging, 4.23 T clinical MRI was developed and evaluated to achieve high resolution of animal tumor and drug monitoring in our laboratory [5]. The present study reports the calibration of the in vivo sodium MRI for both EC-Na and IC-Na sodium signal intensities of phantoms consisting varying concentrations of free sodium and bound NaCl dissolved in agarose gel. The novelty of high resolution in vivo sodium MRI images with high signal-noise-ratio was the choice of using inversion recovery pulse sequence to measure IC-Na measurement in tumor; and use of single quantum filter for tumor size measurement to compare with histological features, immunostaining maps of tumor to suggest active apoptosis and neoplasia with different tumor stages before and 24 hours post-injection of Taxotere in MNU induced rat xenograft breast tumors. Primary hypothesis was that IC-Na sodium in tumors was enhanced after MNU induced cell destruction and associated with apoptosis. Secondary hypothesis was that Taxotere chemosensitivity effect can be quantitated by biomarkers using sodium imaging, histopathology and histo-immunostaining features. To accomplish it, sodium MRI signal intensities (as rapid quantitative method) were compared with different in vitro staining end points to assess tumor response to Taxotere therapy. These results may have implications on utility of in vivo drug monitoring by sodium MRI in human breast tumors.
Sodium MRI depends on sodium nuclei having 3/2 spin states and a quadruple moment. Sodium nuclei exhibit four transitions and two different longitudinal and transverse relaxation constants, a long T2 = 16–30 ms and a short T2 = 0.7–3.0 ms. Sodium MRI utilizes the multiple quantum Na-MRI in a single quantum (SQ) framework as shown in Figure 1 (left panel). Single quantum sodium MRI acquires EC-Na sodium signal-to-noise ratio quickly without the need of paramagnetic shift reagents. We believe that quantum filter interactions may display their effects on nuclear spin transitions as multiple transverse relaxation constants and characteristic IC-Na signal of sodium nuclei in heterogeneous tumor. IC-Na sodium content can be measured by multiple quantum filter (MQF) using shift reagents. Other way is triple quantum filter method that takes long acquisition time. In tumors, sodium nuclei of the outer cell membrane surfaces display an extracellular MQ sodium signal. We propose a rapid and reliable approach of inversion recovery pulse sequence in spin echo sequence to acquire IC-Na sodium high resolution images as shown in Figure 1 (right panel) and previously described elsewhere [8].
Figure 1 The figure shows a spin echo pulse sequence with single quantum mode (A on left) and inversion recovery pulse sequence (B on right) using two inversion pulses for sodium imaging. The exponential return of the Mz after inversion is shown in graph in the center in relation to the RF pulses. The FID signal, produced by the 90° pulse after TE, has initial amplitude related to the value of Mz at the end of interval TI1 and TI2 due to applied selective π pulse. The insert highlights the unique optimum TI value in distinguishing tumor density associated with IC sodium in tumor.
Materials and Methods
Experimental MNU induced Sprague Dawley rat breast tumor model was used as described elsewhere [4]. Breast tumors were imaged in vivo using intracellular sodium weighted MRI approach both at baseline and following administration of Taxotere. All rats were used in accordance with animal care IACUC and Columbia University rules for humane treatment of animals. Subsequently, tumors were excised and in vitro cell DNA phase staging, single strand-DNA antibody (SS-DNA) staining and histopathological features were measured [9-11]. Tumor areas were delineated on SQ and IR sodium MR images using Optimas 6.5 software and their intracellular sodium was measured by relative signal intensity of known sodium concentration i.e. 1 M in 4% agarose phantom for determination of intracellular [Na]i.
Validation of intracellular and extracellular sodium Magnetic Resonance Imaging of phantoms in vivo
Two reference phantoms containing 0.1 M NaCl and 10 mM NaCl in 4% agarose-Na were glued to a plastic animal platform. The 4% agarose-Na phantom was the brightest phantom on the IR image and the NaCl phantom was brightest on the SQ image. For image quality purposes, IR images and SQ images were normalized to the brightest phantom within each slice for comparison using NIH Image J software.
For sensitivity of intracellular sodium MR image intensity, a series of intracellular sodium phantoms was created by dissolving NaCl into four concentrations of agarose (10, 20, 30, and 40% w/v in saline water) for six different NaCl concentrations (5, 20, 50, 100, 120, 150 mM). KCl was added with NaCl to maintain a constant cationic molarity of 150 mM and constant response to radiofrequency coil. The samples were kept at a constant temperature (25°C) and pH (3.54).
Sodium concentrations in tumors were measured by relative sodium concentrations of phantoms placed next to tumor in animals. Mean sodium signal intensity Ik and sodium concentration per kilogram wet weight in region of interest can be represented as:
Ik = A[1 - B exp(-TR/T1)] (1)
where a and b are constants and depend upon signal intensities I1 and I2 and sodium concentrations C1 and C2 in both intracellar and extracellular sodium phantoms as following:
where R1 and R2 are sensitivity factors. SF1 and SF2 are saturation factors [2].
Methyl-Nitrosourea (MNU) induced breast tumor propagation
To propagate the tumor in rats, 42 to 50 days old 30 Sprague Dawley rats were obtained from Harlem Sprague-Dawley (Indianapolis, IN) and the 15 rats were injected i.p. with N-methyl-N-Nitrosourea (MNU), at a dose of 50 mg/kg body wt., purchased from (Ash Stevens Inc., Detroit). After injection, the animals were weekly monitored for tumor propagation and tumor development. First palpable tumor was observed and recorded. The animals were fed 4% Purina Chow diet ad libitum and free access to water. The lesions varied in the 0.2 – 1.0 cm range (stage I), presence of cyst, scar, or solid, benign tumor or lesions in the range of 1.0–3.0 cm. (stage II), and malignant or lesions more than 3.0 cm (stage III) [4]. For comparison, 2 animals were injected placebo water in similar way instead of MNU at sham site. No injection was given to 3 animals and these served as control.
Administration of Chemotherapy
Taxotere (40 mg/ml) was dissolved in the 20% ethanol diluent to a working solution concentration of 6 mg/ml. Taxotere dose levels were 6.0 mg/kg body weight. For it, Taxotere was dissolved in supplied diluent to prepare working solution 100–150 μL and slowly injected into a femoral vein of rats anesthetized by intradermal injection of 1.25 cc of a mixture (Ketamine 10 mg/ml + Xylazine 0.5 mg/ml) at the rate of 1.25 ml/hr (15 mg/hr Ketamine; 0.75 mg/hr Xylazine). Each animal was injected 6 mg/ml Taxotere (i.e. ~1.5 mg/250 gram body weight) calculated from standard equations which convert weight to surface area for small mammals [5].
Chemosensitive effect of Taxotere on breast tumor
When the tumors were developed approximately 2.0 – 2.5 cm in diameter, the tumor bearing, sham and control animals were treated with Taxotere and sodium imaging was done and repeated following 24 hours after injection. For MRI imaging, animals were anesthetized as mentioned above. In units of animal weight, it was a loading dose of 37.5 mg/kg wt for Ketamine and 1.875 mg/kg for Xylazine. This i.p. injecting dose was sufficient to maintain anesthesia in animals for 1 hour.
Magnetic resonance imaging and analysis
All imaging experiments were performed on a high field (4.23 Tesla) whole body clinical MRI system at the Columbia University Hatch NMR Center. A small quadrature birdcage radiofrequency coil (50 mm internal diameter, Morris Inc.) and a high strength gradient insert coil (30 mT/m, Bruker model G-33) were employed in this study to achieve 0.5 × 0.5 × 1.0 cubic mm spatial resolution and 1.0 mm in plane resolution.
The anesthetized rats were aligned in a deep groove with the tumor positioned through a small elliptical opening in the plastic holder, thus maintaining the same relative rat eye position to the phantoms from experiment to experiment. 24 slices were acquired for 3D gradient-echo single quantum image (acquisition time 15 min) and subsequently, an inversion recovery image (acquisition time 45 min). Acquisition parameters were: TR = 100 msec, TE = 5.6 ms, FOV = 40 mm, slice thickness = 2.5 mm, inversion time = 25 msec, and flip angle = 90°, acquisition matrix was 64 × 64 × 8. Both inversion and excitation pulses were non-selective pulses [5].
Analysis of tumor sodium MRI signal intensity change and sodium concentration in all pre- and post- drug treated tumors were measured (in millimoles) relative with corresponding sodium signal intensity in phantom image at its brightest region using NIH J software (see Figure 2). The software puts high quality sodium MRI image sets into common geometrical frame by rotate-translate transform to define region of interest.
Figure 2 Different 6 concentrations of Na (mM) and their SQ MRI signal intensities are shown in images (left panel) with their relationship as histogram bars and graph (right panels) between sodium concentrations of NaCl solutions and SQ MRI intensities. In second row at bottom, different sodium concentrations and their visibility is shown on left. Histogram bars and graph shows IR sodium MRI signal intensities with different sodium concentrations in 4% agarose.
High resolution images with high sensitivity
Minimum duration of 90 degree Rf pulse with minimum saturation time of preamplifier was used to minimize magnetic inhomogeneity and FID. The image signal intensity depends on both Rf receiver and transmitter coils as:
Ik= Rk.M0k.|sin (Φk)| (5)
where Ik is signal intensity of kth pixel, Rk is sensitivity at the coil center, M0k is the equilibrium magnetization. The flip angle Φk is function of transmitter gain setting TG as:
Φk = (C.Rk).10TG/20 (6)
(C.Rk) represents relative sensitivity Rk for the kth pixel. The constant C is independent of spatial position. The receiver sensitivity is directly proportional to transmit field distribution for the coil.
Post-imaging tumor histo-immunostaining end points
The entire neoplastic explant tumor was removed and sliced into 2 mm thick serial segments starting from the cranial aspect marked by suture and proceeding caudally. A touch preparation of the caudal aspect of each slide was prepared and slides were air dried and fixed with 10% phosphate buffered formalin for Feulgen staining to determine nuclear DNA content. Thereafter slices were fixed by immersion in 10% buffered formalin and embedded in paraffin. For histology, 4 micrometer thick sections were prepared of each segment. The MRI co-registered histology sections were stained for hematoxylin and Eosin (H & E) to determine cytomorphology. Different tumor cytomorphic features were counted as number of micrometer squares per mm2 under high power field (HPF) at magnification (× 400) microscope equipped with micrometer squares based on: mitotic rate (number of mitotic figures in micrometer) as mitotic index (MI), amount of apoptosis (20–40 apoptotic nuclei in micrometer square) as apoptotic index (AI), viability (<60% viable cells in micrometer square), cyst space (<100 μm size) and necrosis (<25% necrosis cells in micrometer square) in double-blind manner. The distribution of % extracellular (EC) and intracellular (IC) space was determined with the trichrome stain, where EC appeared light and IC as darker. Malignant cells were characterized according to criteria described elsewhere [4].
Cell cycle analysis
The amount of s-DNA content was determined morphometrically using the Feulgen stain to measure the amount of s-DNA per cell nucleus stained using a computer aided image analysis (CAS 200) [11]. The S phase was defined as cells with s-DNA content immediate between diploid and tetraploid. On an average, 200 cancer cells were evaluated and their s-DNA content was plotted in histogram form. The histograms were compared to a standard histogram from non-neoplastic cells of the same tumor specimen to determine the standard DNA content for diploid cells (G0/G1) phase of cell cycle and the range of s-DNA content for tetraploid (G2 phase) cells. The proliferation index was expressed as the number of cancer cells in S/G2 phase vs total number of cells (G0/G1 + S/G2) per HPF for each tumor.
Immunostaining and TUNEL labeling
The immunoperoxidase enzyme staining technique was used to demonstrate proliferating cells by labeling with KI 67 antigen. Formalin-fixed paraffin-embedded sections were processed for heat retrieval for the KI-67 antigen and then incubated with the primary anti-KI 67 antibody. This antibody was linked to a secondary biotinylated antibody. The binding of the target antigen was demonstrated with avidin-biotin-peroxidase technique using 3, 3' diaminobenzidine as substrate for the peroxidase. It generates a brown reaction product in proliferating nuclei. The non-proliferating nuclei was counterstained green (Methyl-green) and the proliferation index (PI) was evaluated using colorimetric difference on computer aided analysis (CAS 200) as reported earlier [12].
FACS and apoptosis index
The apoptotic nuclei were visualized on formalin-fixed, paraffin-embedded sections using the antibody MAB3299 single-strand DNA(ss-DNA MAb) using immunogen F7–26 and peroxidase conjugated isotype IgM by Flourescence Activated Cell Sorter (FACS) analysis (Chemicon International Inc. Temecula, CA)[13]. The formalin-fixed frozen sections were treated with increasing ethanol dilutions. These sections were treated with 0.1 mg/ml saponin-PBS, Protinase K (20 μg/ml PBS) for 20 minutes. Later slices were washed thrice with distilled water and 50% formamide at 55°C. Soon after slices were quenched with endogenous peroxidase (in 3% hydrogen peroxide for 5 minutes) and treated with 100 μl of monoclonal antibody F7–26(1:10) for 15 minutes. After antibody is fixed on slide, these were applied 100 μl peroxidase-conjugated antimouse IgM (1: 200), incubated for 15 minutes and rinsed with PBS and applied DAB chromogen with hematoxylin-counterstaining. These slides were mounted slides for microscopy of MAB 3299 as apoptosis death marker that was independent of internucleosomal DNA fragmentation [13].
Sodium MR imaging and histological comparison
Edge detection co-registration method was used to delineate boundaries and locate center point of tumor on MRI images and histologic sections with accuracy of 0.5 mm pixel resolution [14]. Semi-automated segmentation was used to distinguish location of different IC-Na signal intensities and corresponding tumor features [14]. Different cytomorphology features as histology end points were compared with the total sodium by SQ MRI and [Na]i by IR MRI signal intensities as following : 1) Ratio of intracellular and extracellular space by trichrome stain; 2) Tumor viability (intracellular space) using proliferation and apoptosis index expressed as number of proliferating and/or apoptotic cells per total number of cancer cells; 3) Cell cycle analysis based on DNA content (Feulgen stain); and active necrosis.
Correlation of histological features with intracellular sodium MRI image signal intensity
The correlation was analyzed to co-register digitized histology maps with matched sodium MR images at different regions in the tissue to test null hypothesis. The hypothesis was that altered sodium signal intensities do correlate with immunohistology and cytopathologic biomarkers indicating different tumor features. Step down multivariate statistical analysis was performed to determine the order of immunohistolgical and pathologic parameters that correlate with the intracellular sodium (IC-Na) MRI signal [15].
Results
Animals
Out of total 30 animals, only 9 MNU induced animals (out of 15 animals) showed total 16 breast cancer tumors. Six animals (out of 15 animals) treated with Taxotere only and three animals served as untreated control group. These control untreated animals did not show any tumor. Other animals served as 3 sham and 3 water placebo treated second control animals. These six pre- and post-treated tumors were analyzed for IR intracellular sodium MRI and SQ total sodium MRI imaging. MRI data were correlated with histopathologic features and immunostaining parameters. Tumors were grouped for quantitative observations as control, obtained from MNU induced animals and compared with 16 tumor tissues obtained from Taxotere treated MNU induced animals.
Validation of intracellular and extracellular sodium MRI signal intensity
A series of sodium agarose phantoms, showed distinct sodium MRI signal intensities at varying sodium concentrations as shown in Figure 2 top panel on left. T2 values were measured between the phantom and the extracellular fluid as shown in Table 1 and Figure 2. Figure 2 (panel at bottom on left) represents a plots of visibility as MoSQ vs. sodium concentration for all concentrations of agarose used. The linear relationship between visibility as MoSQ and MoIR using relaxation constants and sodium concentration was observed with change in agarose concentration. The visibility of MoIR increases more with % agarose concentration. Sodium MRI signal intensity showed linear relationship with sodium concentration as shown in Figure 2 on right panels.
Table 1 A comparison of slow and fast transverse relaxation times for different agarose concentrations and dependence of visibility VisibilitySQ (ΔMo/Δ[Na]) and visibilityIR (ΔMo/Δ [Na]) on agarose concentration. VisibilitySQ and visibilityIR for all concentrations of agarose were obtained using SQ and IR techniques. All transverse relaxation time constants T2sSE, T2sSE and T2sIR, T2sIR were determined by single quantum Hahn spin echo and IR pulse sequence on NMR.
% w/v Agarose T2sSE T2sSE MRI VisibilitySQ T2sIR T2sIR MRI VisibilityIR
10 12.4 ± 1.2 8.1 ± 0.48 1.75 13.7 ± 1.2 7.4 ± 0.5 3.45
20 7.5 ± 0.35 4.1 ± 0.31 1.65 8.4 ± 0.3 3.4 ± 0.3 4.50
30 7.1 ± 1.61 1.8 ± 0.04 1.5 6.4 ± 0.2 1.8 ± 0.04 4.70
40 3.4 ± 0.1 1.2 ± 0.04 1.4 4.6 ± 0.2 1.2 ± 0.1 7.22
Sodium Inversion Recovery nulls signal from selected T1 range
The inversion recovery (IR) pulse sequence suppressed the signal from Na nuclei with long T1 relaxation constants (free EC Na). For it, the inversion time (the time between the 180° and 90° pulses in the IR pulse sequence) was calculated as (ln 2)(T1ex), where T1ex was composite longitudinal relaxation time. It generated an intracellular (IC) weighted sodium signal and MRI image. The inversion time selectively suppressed the signal from either long T1 (EC-Na) or short T1 (bound IC-Na) in agarose phantoms at lesser or more than the optimal TI choice as shown in Figure 3. At inversion time TI = 30 msec, the long T1 signal from free EC-Na in NaCl phantom was suppressed completely. Two 1 M NaCl phantoms (with 4% agarose shown as ○ open circles; and, without 4% agarose shown as ● closed circles) were examined during both single quantum MRI acquisition and inversion recovery MRI acquisition for nine different inversion recovery times. Data points represented image intensities while the lines showed theoretical relationships and two null points assuming TI = 27.4 and 43.3 msec for sodium images [5].
Figure 3 (Top row) The phantom consists of two tubes, one is 1 M NaCl in solution and the other is 1 M NaCl in 4% agarose (Panel A). They simulate extracellular and intracellular sodium, respectively without use of inversion recovery pulse. It can be seen that an inversion time of TI = 20 ms completely suppressed intracellular sodium (Panel B) and TI = 30 ms completely suppressed the extracellular sodium signal from the tube containing NaCl in solution (Panel C). At TI = 40 ms partial suppression of extracellular sodium is highlighted (Panel D). (Middle row) Two 1 M NaCl phantom image intensities (with 4% agarose, open circles; and, without 4% agarose, closed circle) were plotted during single quantum acquisition and for 9 different inversion recovery times (TI) showed two distinct null points shown at right.
Application of Sodium Inversion Recovery method to breast Tumors
Single quantum (SQ) sodium images (Figure 4, panels in rows 1 and 2) showed distinct bright tumor on the SQ images. IR pulse sequence perhaps suppressed the extracellular sodium at optimal inversion time (TI) applied on rat model of breast cancer. Optimum inversion time (TIopt) was chosen 25–30 msec based on quantitative difference in spin-lattice relaxation constants (T1). At this inversion time, the signal was totally nulled for cyst regions e.g. a tumor sac that was predominantly fluid, showed up darker shown as blue arrows in Figure 4, right panels D, E in rows 3 and 4. It enhanced the visibility of semi-solid or proliferative tumor regions that appeared relatively brighter in comparison to the rest of rat body shown as red arrows in Figure 4, panels F, H, I. It may be presumably due to the higher intracellular (IC) sodium in proliferative regions in tumor vs normal breast tissue. IR sodium images (Figure 4, image panels in rows 3 and 4) showed distinct tumor heterogeneity i.e. cyst, solid tumors but poor tumor boundaries.
Figure 4 Taxotere chemosensitivity effect is compared on SQ images of pre- and post Taxotere treatment (first vs second rows on top). Panel on second row on rightmost image, represents the tumor SQ image (left arrow) and phantom (right arrow). Third and fourth rows represent the panels showing contrast (arrows) on contiguous intracellular sodium MRI images as "IR" images of pre- and post Taxotere treatment respectively by inversion recovery method. Panels in fifth and sixth rows represent the intracellular sodium signal intensity increases (arrows) in some areas representing active proliferation or resting viable cells. Loss of signal in late apoptosis rich regions and increased tumor areas can be seen as hypointense areas (panels in fifth row vs sixth row for pre- and post-injection of taxotere). An enlarged "IR" image (panel in sixth row) of tumor (on rightmost insert A) after segmentation image processing is shown with different bright, gray and darker pixel signal intensities.
Breast tumors overall showed around 2 – 3 times higher intracellular(IC) sodium MRI signal over normal body IC sodium MRI signal intensity in baseline control animals shown in Table 2 and Figure 4 (panels in rows 5 and 6).
Table 2 Comparison of tumor characteristics in different regions is shown in pre-drug treated tumor@ and post-treated tumor# by sodium MRI image intensity and histology, histo-immunology parameters as shown in Figures 5 and 6. By using eyepiece-micrometer square counter (100 squares per high power field), necrosis*(<25% cells in micrometer square), viable cells**(<60% cells in micrometer square) and apoptosis*** (20–40 apoptotic nuclei in HPF) and cyst space****(<100 μm) per HPF were premalignant histology characteristics. IC/EC space (% space in HPF), necrosis, viable cells are shown as number of micrometer squares with <25% necrosis area in HPF by histology. Apoptotic index (A.I.) and proliferation index (P.I.) are shown as average number of apoptotic nuclei per HPF and number of mitotic figures per HPF. S-DNA histogram area was measured by CAS 200 system in arbitrary units. Single strand-DNA mAb area was measured in digital images by Optimas 6.5 and ss-DNA mAb density was measured in arbitrary units of photomultiplier scanner.
Pre-drug treated@ Post-drug treated# Sodium MRI signal intensityd Histology (in HPF) A.I. (KI-67) P.I. (in HPF) S-DNA (histogram) ss-DNA mAb (density units)
Tumor feature IR SQ (CAS)
Control (n = 41) 4.25 ± 1.0 5.27 ± 2.0 - - - - -
Sham (n = 40) 4.30 ± 0.2 5.38 ± 1.0
Tumor area(mm2;n = 16)@ 4.56 ± 0.30 4.67 ± 0.30 4.49 ± 0.2 - - 3.6 ± 0.2 4.2 ± 0.2
Tumor area(mm2;n = 3)# 4.60 ± 0.29 4.61 ± 0.26 4.42 ± 0.3 - - 3.5 ± 0.3 4.3 ± 0.3
IC/EC space@ 60–70 65–76 - - - -
IC/EC space# 80–95 60–70 - - - -
Necrosis*(squares)@ gray 56 ± 32 - 270 M-DNA -
Necrosis*(squares)# bright 50 ± 23 - 120 M-DNA -
Viable**(squares)@ dark 65 ± 21 - - - -
Viable**(squares)# dark 76 ± 11 - - - -
Apoptosis***(nuclei)@ bright 40 ± 12 150 - S-DNA 159 ± 12
Apoptosis***(nuclei)# bright 50 ± 13 160 - M or S-DNA 129 ± 20
Cyst****(size in μm)@ gray 120 ± 29 - - - -
Cyst****(size in μm)# gray 80 ± 22 - - - -
IR-Na signal intensity changes and immunostaining features in post Taxotere treated tumors
Comparison between pre- and post-treatment single quantum (SQ) sodium images (Figure 4, panels in rows 1 vs 2) showed increased size as tumor boundaries but poor tumor heterogeneity. However, the tumor showed up as bright on the post-treatment SQ image. Viable, active proliferative necrosis and apoptosis regions showed up hyperintense on post-treated IR images but unchanged or hypointense on post-treated SQ images. Cysts, extracellular space showed hypointense regions on post-treatment IR images but hyperintense on IR images as shown with arrows in Figure 4 (rows 5 and 6). These enhanced IR signal intensities after Taxotere 24 hour post-treatment were associated with both enhanced apoptotic index (9%) and reduced proliferation index (8.6 vs 2.2; 4 fold decrease over pre-treated tumors) in these regions based on NIH Image J pixel reader counts at different colored locations in post-segmented tumor as shown on Figure 4 (rightmost insert A in row 6). Using calibration graph, the high sodium signal intensities in active regions were measured relative to phantom (P) as shown in Figure 4 and Table 3 showing distribution of sodium in different tumor regions.
Table 3 Different SQ and IR sodium MRI signal intensities and sodium concentrations (mean ± SD) are shown by using NIH Image J pixel density reader at different x, y pixel positions in tumor different regions of active necrosis, viable, apoptosis, cyst regions shown in a representative tumor in Figure 6. Untreated 3 control, 3 sham animals, and 3 water placebo injected animals; counterpart pre-treated 3 tumor bearing animals on day 1 and post-treated 6 tumor bearing animals on day 2 after taxotere injection showed distinct SQ and IR MRI sodium signal intensities in excised tumors (3 pretreated tumors and 16 post-treated tumors) at their different n = 160 tumor sites in pre-treated and n = 160 post-treated tumor sites showed comparable changes in SQ and IR sodium signal intensities. For validation, sodium MRI signal intensities changes were compared with histology and immunostaining methods in different tumor x, y positions (n = 160) showing different features in 24 hours post-treated animal tumors. % difference of sodium signal intensity represents comparison between control vs. sham or water placebo; pre-treated vs post-treated tumor sites. *Apoptosis represents single large nuclear bead inside the cytoplasm during terminal stage without nuclear fragmentation. Mitotic figures represent number of visible mitotic spindles observed per high field under microscope.
Tumors (n = 16) features SQ Sodium MRI IR Sodium MRI % difference
signal intensity A.U. concentration (mM) signal intensity A.U. concentration (mM) SQ IR Mitotic figures (per HPF)
Contol Phantom: 9.5 ± 0.3 100 4.2 ± 0.5 10
Day 1(0 hour):
Control(n = 41)sites 5.6 ± 0.2 58.9 ± 2.1 4.5 ± 0.1 10.7 ± 1.0
Sham(n = 40)sites 5.5 ± 0.6 57.8 ± 6.3 4.6 ± 0.2 10.9 ± 2.1 1.7 ± 0.5
Water placebo(n = 40) 5.9 ± 0.8 62.0 ± 8.4 3.3 ± 0.3 7.8 ± 0.7 5.3 ± 2
Pre-treatment
Tumor regions (n = 160 sites) 2.4 ± 0.6
Viable 6.8 ± 1.1 71.5 ± 11.6 6.1 ± 0.8 14.5 ± 1.9
Active necrosis 7.6 ± 1.4 79.9 ± 9.4 8.6 ± 1.0 20.5 ± 2.4
Apoptosis 6.2 ± 0.9 65.2 ± 9.2 10.8 ± 1.2 26 ± 3
Cyst 19.5 ± 2.3 205.1 ± 24.2 7.2 ± 0.7 17.1 ± 1.6
*Apoptosis 6.2 ± 1.6 65.2 ± 16.8 10.3 ± 1.4 24.5 ± 3.3
Day 2(24 hours):
Control(n = 41)sites 4.7 ± 0.2 49.4 ± 2.1 3.5 ± 0.4 8.3 ± 0.9 16.0 ± 0.2 22 ± 0.6
Sham(n = 40)sites 6.8 ± 0.8 71.5 ± 8.4 4.2 ± 1.1 10 ± 2.6 23.6 ± 3.3 8.6 ± 0.8
Taxotere injected (n = 40 sites) 7.9 ± 0.4 83.1 ± 4.2 5.0 ± 0.4 11.9 ± 0.9 33.8 ± 0.5 51.5 ± 3.3
24 hr Post treated
Tumor regions (n = 160 sites) 8.5 ± 0.5
Viable 6.2 ± 1.6 65.2 ± 16.8 7.6 ± 1.2 18.0 ± 2.8 8.8 ± 0.5 24.5 ± 0.4
Active necrosis 6.8 ± 1.2 71.5 ± 12.6 9.1 ± 1.0 21.6 ± 2.4 10.5 ± 1.4 5.8 ± 0.5
Apoptosis 6.4 ± 0.9 67.3 ± 9.2 18.8 ± 1.7 44.7 ± 4.0 3.2 ± 0.0 74.0 ± 4.1
Cyst 22.6 ± 2.6 237.7 ± 27.3 6.2 ± 1 14.7 ± 4.3 15.9 ± 1.3 13.8 ± 0.58
Co-registration indicated the possibility of edge detection and point-by-point (equal square size) stereotactic matched quantitation of sodium MRI images with histology and immunostaining maps as shown by squares 1–5 on × axis and squares a-g on y axis in Figure 5. The sodium MRI signal intensities were measured relative with phantom (P) at different tumor locations under high power field shown as apoptosis (A), high EC volume (B), cyst (C), proliferation and viable cells (D) and necrosis (E) in Figure 5 and Table 3. However, DNA cycle maps indicated different features but not conclusive while sodium MRI signal intensities differed by SQ and IR methods mainly for apoptosis (A) and cyst (C) along with other tumor features as shown in Figure 5 (rightmost panel) and Table 3.
Figure 5 The figure shows the stereotactic co-registration method of IR sodium tumor image, its counterpart histology and DNA ploidy map at different x- and y- co-ordinate locations shown as 1–5 and a-g for different areas. Notice the matched tumor delineated areas on three digitized images and corresponding features shown with arrows for high % apoptosis (A), high % EC volume (B), cyst (C), % viable cells and proliferation (D), % necrosis (E) in high power fields. The tumor IR sodium MR image appears hypointense for cyst and high EC regions and hyperintense for active proliferation while isointense for necrosis. On right panel, sodium signal intensities are shown for 100 mM SQ sodium phantom and 10 mM IR sodium phantom along with relative MRI signal intensities in the different tumor regions.
MRI images showed isointense IC sodium signal intensities in 3 sham rats following placebo water injections for comparison with 3 pre-treatment MNU breast tumors. In control animals, histology and sodium MRI images were normal as shown in left panels of two rows in Figure 6. MNU induced breast tumor sites showed increased IC sodium signal intensity (2–5 fold increase, p < 0.01; n = 16) compared with matched water placebo-sham control sites shown in Figure 6 (left panels in second row). In 3/9 tumor bearing pre-treated animals (second column) showed main features as following: 1. Increased IR-Na signal intensities (yellow arrows) with corresponding immunostained areas of late S-phase in their excised tumors; 2. Tetraploid nuclei or actively dividing nuclei in G2 phase with M and S histogram indicating the presence of more necrotic cells with late S phase completion in viable cells; 3. Neoplasia stage appeared as gray with isointense IR-Na signal during their late S-phase shown as distinct S-DNA histogram in Figure 6 (rightmost panel in second row) with corresponding S-DNA cycle map in Figure 6 (rightmost panel in fourth row); 4. Distinct excised tumor histology features in panels under high power fields as active viable cells (a), proliferation (b), necrosis (c), apoptosis (d), mitosis (e), fibrous cyst (f), and infiltrating ductile carcinonoma (g) in different x- and y- coordinate locations after point-by-point coregistration with IR sodium image (arrows a-g). In 24 hours Taxotere treated animals showed main features as following: 1. Increased IR-Na signal intensities (yellow arrows) were associated with tumor cells arrested in G0/G1 or early S phase (diploid or anoploid cells), neoplasia stage with S-DNA histogram patterns as shown in Table 2 and Figure 6 (rightmost panel in fourth row); 2. Hyperintense, isointense, and gray-green features on color-coded segmented IC-Na image (C panel in second row) with tumor histology features as apoptosis (A), necrosis (B) and neoplasia (C); 3. Reduced gray tumor area on sodium MRI image perhaps due to more bright apoptosis rich areas (yellow arrows on panels A and B) with corresponding histology features (delineated tumor area); 4. Longer early S-phase of resting tumor cells; 5. Reduced diploidy or aneuploidy in dividing tumor cells undergoing cell division (in 80% tumors); 6. Decreased tumor viability as mean proliferation index (PI); and 7. Reduced proliferation index.
Figure 6 The sham control, pre-treatment and Taxotere post-treated animals (top panels on left) show tumors as SQ sodium (A) and IR sodium (B) images at 0 hr pre-Taxotere and 24 hours post-Taxotere treatment (panels A and B on second row). On third row on left, control tumor histology shows normal vesicles. Pre- and post treated excised tumor histology by trichrome staining is shown with delineated area. On fourth row on left, the excised tumor histology features in high power fields are shown with arrows (active viable cells (a), proliferation (b), necrosis (c), apoptosis (d), mitosis (e), fibrous cyst (f), and infiltrating ductile carcinonoma (g) in different x- and y- coordinate locations after coregistration with IR sodium images. On right, panels on top show a IR sodium MR image before (C) and after non-parametric segmentation by Optimas 6.5 to highlight the different signal intensities that appeared hyperintense, isointense, and gray-green colored on segmented image and histology features showed them as apoptosis (A), necrosis (B) and neoplasia (C). On right, second row shows corresponding S DNA histograms of neoplasia features by CAS 200 (panels on top), apoptosis staining (panel with green stain). On right, third row shows a post-Taxotere treated tumor histology by pentachrome stain to highlight mitotic figures (M) with active PMN cells (P) and high EC volume (EC) and corresponding digitized map of DNA cycle, with neoplasia shown as arrow.
In excised pre-treated tumor histology sections, number of cells with distinct histology features per HPF (in mm2) were: mitotic figures (MI) (2.4 ± 0.6 in mm2), necrosis (56 ± 32 in mm2), viable cells (65 ± 21 in mm2) with early phase of apoptosis (fragmented nuclei with smooth call membrane) (40 ± 12 in mm2) in 100 small squares per high power field (HPF), at magnification × 400 including pre-malignant stages of ductal hyperplacia, intraductal proliferation in each specimen as shown in Figure 7 with Tables 2 and 4. After Taxotere treatment, excised tumor cells showed the number of cells with histology features per HPF (in mm2) were: mitotic figures (MI) (8.5 ± 0.5 in mm2), necrosis (50 ± 23 in mm2), viable cells (76 ± 11 in mm2) and apoptosis (50 ± 13 in mm2) in 100 small squares per high power field (HPF), at magnification × 400 showing pre-malignant stages, hyperplacia, proliferation cells with PI decrease 4-fold from 270 per mm2 (per HPF) to 120 per mm2 (per HPF) as shown in Tables 2 and 4. However, the biological significance could not be assessed with these observations. Different tumor malignant regions showed similar features in pre-treatment and post-treatment tumors rich with papillary or invasive comado and cribriform carcinoma, sarcoma or neoplacia and late apoptosis (tiny beaded nuclei with distorted cell membrane) and lower mean PI in Taxotere treated breast tumors as shown in Figure 7 and Table 4.
Table 4 Histological characteristics of MNU induced rat breast pre-treatment tumors under magnification × 400 microscopy equipped with micrometer. Different premalignancy and malignant characteristics and cytomorphology characteristics were co-registered with IR sodium signal intensities. Level of brightness is shown as + sign on MRI gray scale.
Sodium MRI signal Intensity Histomorphology Tumor characteristics
SQ Na MRI IR Na MRI Tumor features Tumor stage
Pre-malignant stage:
isointense dark/gray ++ + Viable Active necrosis Intraductal proliferation, ductal hyperplacia, apoptosis
Malignant stage:
gray ++++ apoptosis Transformation stage
+++ bright/gray apoptosis Carcinoma (papillary; invasive comedo and cribriform)
++++ dark Cyst sarcoma, neoplasia
Figure 7 The figure represents a comparison of different tumor pre-malignant and malignant features in coregistered sodium IR images (panels in first row with phantom P) and their counterpart histology in low power (second row). Notice the heterogeneous regions of tumor and corresponding distinct histopathology. At some selected locations, relative IR sodium MR signal intensities* (A.U.) with reference to sodium phantom P (50 A.U. in top panels) are shown as numbers and histograms (see Figure 8) as distinct values for different tumor stages with different tumor features by NIH Image J 1.63. Panels (1–22) show high power optical microscopy (images 1–8) in control 0 hour (on left) and post-Taxotere 24 hours treated tumors (images 9–22 on right) with distinct features shown by arrows.
Main findings were: 1. The number of apoptotic cells per mm2 (HPF) i.e. apoptotic index (AI) was increased 10%; ~150 per mm2 (HPF) vs ~-160 per mm2 (HPF) in pre- and post-Taxotere treated tumors associated with enhanced intracellular IR sodium signal intensity by MRI in highly viable cell areas rich with pre-malignant stages; 2. Active single cell necrosis regions as isointense IC-Na signal intensity showing chemotherapeutic effect while advanced necrosis showed gray IC-Na signal intensity (in 6/6 animals; on 9/54 tissue slices). This was suggestive of increased IC sodium in 6 post-treatment animals as shown in Figures 6 and 7.
The tumors exhibited decreased IC sodium intensity (22 ± 4%) following chemotherapy (p < 0.001; n = 40), a response significantly different (p < 0.002) than that in control as shown in Figure 6. The reduced IC sodium intensity change (-5 ± 4%; n = 41 sites) in control animals vs. placebo-water injection perhaps was due to its water-insensitive characteristic (as shown in Figure 3 on panel B and Figure 6) but significant in control vs. placebo water injected animals (28 ± 2%; n = 40 sites) after 24 hrs perhaps due to inflammatory processes at sham sites. The average IC sodium signal intensity at Taxotere injected plain tissue sites showed elevated IC-Na signal compared to water placebo-sham control sites (30 ± 4%, p < 0.05; n = 40 sites) as shown in Table 3. Taxotere treatment showed a 34 ± 3% IC sodium signal intensity increase over pre-Taxotere treated tumor (p < 0.001; n = 40; Figure 4 panels right vs left on fifth vs sixth row).
The sham control, pre-treated and Taxotere post-treated animals bearing MNU induced tumor are shown on top row in Figure 6. Their SQ and IR sodium images and histology with immunostaining features showed characteristic signal intensities as shown in Table 3. SQ sodium images displayed better tumor area (panels in 1 and 2 rows) while IR sodium images displayed tumor heterogeneous features as shown in Figure 4 (panels in 3 to 6 rows) and Figure 7 (panels on top row).
IR-Na signal intensity changes and extracellular vs. intracellular space
Histology and the pentachrome staining measured the ratio of intracellular vs. extracellular space (EC) as shown (red arrow) in the panel on third row with polymorph cells (P) rich with mitotic figures (M) in Figure 6. Pentachrome staining further demonstrated the distribution of extracellular space (EC) shown as lighter and intracellular space (IC) as darker in 54/58 (in 90%) histology slides. The % IC/EC space ratio indicated viable cells and cells at risk or dying cells. Viable cells showed the average ratio '% IC/EC' ratio as 60–80%. It represented the number of viable cells and higher mitotic Index. The dying cells showed higher '% EC/IC ratio' 70% or above with higher Apoptotic Index (AI) of cells possibly in late apoptosis phase. The extracellular space showed cell debris, dying cells with fluids. It contributed minimum to intracellular space. Gross morphology further determined the intracellular contents including collagen, extracellular matrix, cyst fluid (color ranging lighter for cyst to darker for collagen) as shown with arrows in IR sodium MRI image. The comparison of IC/EC ratio with the MRI images is shown in Table 2. Areas comprising less than 30% IC space showed up as dark to hypointense regions whereas areas comprising 30–50% intracellular space showed up hypointense to isointense on IC-Na MRI images. More than 50% IC space (i.e. less than 50% EC) showed up brightest or hyperintense on IR-Na MRI images. Intracellular sodium images (IR-Na) showed medium to bright MRI areas in (36/54 tumor image slices) representing more than 50% IC space. Low intracellular sodium signal intensity in MRI images represented less than 40% IC space in the tumor regions. Control tumors showed the more viable cells and more intracellular space than the post-Taxotere treated tumors by H&E staining.
The apoptotic index was increased marginally in post-treated breast tumors (160 per mm2) vs. control tumors (150 per mm2) that indicated the negative correlation between intracellular sodium signal intensity by MRI and cell viability as shown in Tables 2 and 4. Single cell necrosis showed enhanced IC-Na sodium signal intensity depending on the degree of necrosis while extensive late necrosis showed low SQ Na MRI signal intensity as shown in Table 4 and Figure 6, perhaps due to chemosensitivity effect, ischemia or increased tumor vascular supply. Large areas of late necrosis (low signal intensity) were not observed within 24 hours after initiation of Taxotere chemotherapy in breast cancer.
Histology and Immunostaining features
A "touch slide preparation" of 4 micron thick serial sections by 10% Fuelgen staining (41/48 tumor sites) determined the s-DNA content (in 3 control tumor tissues and 16 Texotere treated tumor tissues) as shown in rightmost panel on third row in Figure 6. The 4 micron thick histology section staining by H&E showed histology features of necrosis, apoptosis, proliferation, mitotic rate (number of mitotic figures per mm2) and amount of apoptosis (number of apoptotic nuclei per mm2) in 36/48 (75%) tumor sites.
Chemosensitivity assay of Taxotere effect using cell proliferation
Ten explanted Taxotere-treated tumors showed enhanced IR-Na MRI signal intensities by microimaging of tumors at 0 hour and after 24 hours. Central necrosis appeared dark in center and bright peripherypresentation on the IR image as shown in second row on Figure 6. These tumors were uniformly bright on the IR image. Based on histology, these post-treatment tumors showed neoplastic features rich with cyst or absence of cell nuclei, apoptotic cells and proliferation rich necrosis zones as shown in bottom row in Figure 6. An outermost peripheral zone appeared viable tumor tissue. Apoptosis cells were distributed in unspecific manner but apoptosis rich zones were common between viable (V) zones and necrosis (N) rich zones seen in HPF shown in Figure 6 bottom row (see panel a) and Figure 7 (see panels 13).
Under HPFs, the untreated tumors showed high number of mitotic figures (see Figure 6, panel b and Figure 7 panel 1). The mitotic figures measured up to >20 per HPF in the viable zones of tumor. Mitotic figures were altered significantly (p < 0.0001) in the Taxotere-treated 9 tumors (8.5 ± 0.5 per field; n = 160 HPFs) as shown in Figure 7 (see panel 9) and Table 2. All post-treatment tumors showed a significant inverse correlation (p < 0.02) between the reduced mitotic figures and the relative increase in IR sodium signal intensity on IR images.
Correlation of Image Configuration with Histology and Immunostaining
Pre-Taxotere and post-Taxotere treated tumors showed enhanced SQ- and IR-MRI signal intensities (panels A and B) over the untreated matched tissue (panel A) and sham-control tissue (panel B) on their sodium MRI images as shown in Figure 6. In tumors with small or absent non-viable centers, the IR-MRI images showed uniformly bright regions. The non-viable centers were comparatively smaller than necrosis rich regions. The chemosensitivity effect of Taxotere (enhanced sodium MR image intensity) was evident at histology matched locations of apoptosis (A), neoplastic (B) tumor features rich with cyst or absence of cell nuclei (C), and proliferation rich (D) and necrosis (E) zones as shown in Figures 5 and 6. Taxotere 24 hour post-treatment showed the increased bright area on the IR MRI image as shown in Figure 6 (panels A and B on second row). The IR MRI image in panel C showed different tumor features as dark center indicating late necrosis region (A), hyperintense region indicating apoptosis (B) and lower hypointense region indicating viable cells (C) as shown with red arrows in corresponding post-segmented IR MRI image (A) and its histology matched section. The 24 hours post Taxotere treated tumors showed unchanged IR MRI signal intensities at histology matched necrosis but increased IR MRI signal intensities at histology matched apoptosis showed as yellow arrows on panels A and B (see pre Taxotere and post taxotere images) following the Taxotere drug administration as shown in Figure 6. However, neoplasia and apoptosis were identified by CAS 200 and fluorescent ss-DNA monoclonal antibody staining showed apoptosis cells (next to the viable region) as brightest and the cells closer to the smaller nonviable center were less bright as shown in Figure 6 (red arrows on rightmost panel in third row). The delineation reflected measurement accuracy on histology and matched IR MRI image is shown with red arrows in Figure 6.
Microimaging and tumor characterization
Sodium MRI signal intensities showed heterogeneous regions of pre-treated and post-treated tumors corresponding with characteristic pre-malignant and malignant different stages by cytomorphology as shown in Table 4 and Figure 7. However, sodium distinct IR MRI signal intensities and relative IC sodium concentrations using standard sodium phantom data of different tumor stages were evident but not conclusive. Main evident cytomorphic features were apoptosis, necrosis and cysts with characteristic sodium MRI signal intensities. Typical tumor histology features in both pre- and post Taxotere treatment appeared as intraductal proliferation, intraductal carcinoma with cribriform or comedo patterns as shown in Figure 7 (see panels 1 – 8 for pre-treatment tumors and panels 9–22 for post-treatment tumors). Some tumors showed papillary carcinoma, adenoma and neoplasia as shown in Figure 7 and Table 4. IC sodium MRI signal intensities and IC sodium concentrations showed marginal insignificant differences for tumor premalignant and malignant locations confirmed by histology as shown in Table 5 and Figures 7 and 8. However, relationship of drug treatment and tumor histology pre-malignant features was evident but non-specific and limited to one time point.
Figure 8 The sodium IR MRI signal intensities and calculated IR sodium concentrations are shown as histograms for pre-malignant tumor features shown in Figure 7. In left panel A, different intracellular sodium MRI signal intensities and concentrations indicated for [intraductal proliferation 11.5(2); ductal hyperplacia 8.8 (3); apoptosis with intraductal carcinoma 12.4* (6,13) and active proliferation 6.2* (2); ductal carcinoma 14.2* (17); cribriform ductal carcinoma 12.8* (7); ductal comedo carcinoma 12.2* (4)]; Malignant features [papillary carcinoma 13.5* (9,11); invasive cribriform carcinoma 12.8* (18); invasive comedo carcinoma 13.5* (22); tubular carcinoma 13.8*; adenoma 12.5* (2,8)]. Tumor different histology features under high power fields are shown in brackets correspond with histology regions in Figure 7. The panel B, in center represents the match of IC sodium concentrations with IC sodium MRI signal intensities. In different tumor regions, IC sodium MRI concentrations were calculated relative to the phantom MRI signal intensities shown as linear curve in Figure 2. In the right panel C, IC sodium concentrations and MRI signal intensities at different tumor premalignant or malignant locations represent the possibility of quick assessment of tumor characterization with possible staging.
Table 5 Different IC Na MRI signal intensities represent the matched locations of different tumor premalignant and malignant stages on tumor histology sites as shown in Figure 7 in different panels for: pre-malignant features [intraductal proliferation 11.5(2); ductal hyperplacia 8.8 (3); apoptosis with intraductal carcinoma 12.4* (6,13) and active proliferation 6.2* (2); ductal carcinoma 14.2* (17); cribriform ductal carcinoma 12.8* (7); ductal comedo carcinoma 12.2* (4)]; Malignant features [papillary carcinoma 13.5* (9,11); invasive cribriform carcinoma 12.8* (18); invasive comedo carcinoma 13.5*(22); tubular carcinoma 13.8*; adenoma 12.5* (2,8)]. The IC Na concentrations were measured by using IC Na MRI phantom standard graph as shown in Figure 2.
Tumor feature IC Na MRI signal intensity IC Na concentration (mM)
intraductal proliferation 11.50 13.8
ductal hyperplacia 8.80 10.6
intraductal carcinoma 12.40 15.0
ductal carcinoma 14.20 17.0
cribriform ductal carcinoma 12.80 15.4
ductal comedo carcinoma 12.20 14.6
papillary carcinoma 13.50 16.2
invasive cribriform carcinoma 12.80 15.4
active proliferation 6.20 7.4
invasive comedo carcinoma 13.50 16.2
tubular carcinoma 13.80 16.6
adenoma 12.50 15.0
10 mM IC Na phantom 8.5 10.0
100 mM IC Na phantom 80.0 100.0
Co-registration and comparison of histological and IR-Na sodium images
In co-registered histological digital images and IR-Na sodium images showed % difference ~5% for tumor delineated area as shown in Table 6 and Figure 9 at the top row. Bar diagrams visualized the quick comparison of tumor areas. MRI imaging methods showed correlation with histology (r2 = 0.34; n = 6) and s-DNA maps (r2 = 0.39; n = 6) as shown in Figure 8 at bottom on right.
Table 6 Comparison of delineated feature areas on intracellular sodium [Na]i images and histology digital images is shown at different levels for contiguous IR sodium MR slices (top row) and corresponding histology digital maps next to phantom (P). At bottom, each left bar represents histological area and each right bar represents delineated tumor area on both [Na]i image and histology digital images at different slice levels as shown in Table 5. Quantification of tumor area was done by Optimas 6.5 using sodium MRI and histology digital images. The tumor area was measured in mm2 in one representative tumor shown in Figure 8. Tumor area was measured and correlated in contiguous 6 histology sections with matched IR images (r2 = 0.3487) and scanned ss-DNA mAb digital images in mm2 (r2 = 0.3987). ss-DNA densities in arbitrary units were measured at different locations of the tumor digital images as shown in Figure 5.
Tumor Area Measurement_in images
MRI slice Sodium MRI Histology ss-DNA mAb
(in mm2) (in mm2) (in mm2) (arbitrary units)
(a) (b) (c)
Slice 1 4.68 4.54 4.51 102.2
Slice 2 4.37 3.57 4.21 118.9
Slice 3 5.10 5.20 5.10 114.6
Slice 4 5.02 4.47 4.44 97.9
Slice 5 3.53 4.97 4.50 106.9
Slice 6 3.61 4.48 4.12 142.9
Figure 9 Comparison of delineated feature areas on intracellular sodium [Na]i images and histology digital images is shown at different levels for contiguous IR sodium MR slices (top row) and corresponding histology digital maps next to phantom (P). At bottom, each left bar represents histological area and each right bar represents delineated tumor area on both [Na]i image and histology digital images at different slice levels as shown in Table 5. Quantification of tumor area was done by Optimas 6.5 using sodium MRI and histology digital images. The tumor area was measured in mm2 in one representative tumor shown in Figure 8. Tumor area was measured and correlated in contiguous 6 histology sections with matched IR images (r2 = 0.3487) and scanned ss-DNA mAb digital images in mm2 (r2 = 0.3987). ss-DNA densities in arbitrary units were measured at different locations of the tumor digital images as shown in Figure 5.
Statistical analysis suggested main determinants in the following order: tumor area by histology > tumor area s-DNA map > apoptosis index (immunostaining) > cell viability parameters associated with increased sodium MR signal intensity.
Discussion
The study reports an application of a novel sodium MRI imaging method to stage breast cancer and monitoring Taxotere treatment response. This approach has significant potential as a tool for exploring breast cancer biologic features and as a clinical imaging approach. It describes sodium MRI inversion recovery pulse sequence and single quantum Hahn spin-echo method, their validation and application in tumor characterization in rapid sodium MRI in vivo monitoring of Taxotere chemotherapeutic response and breast tumor cytomorphology features.
Sodium concentration vs MRI MQIR signal intensity relationship was reliable in phantom experiments [8]. However, at sub-physiological levels, such relationship in tumor propogation is still not known. We believe that tumor cell destruction may cause intracellular sodium increase up to 2–4 times based on sodium extracellular space ~0.15 of total water space. Previous reports suggested selective suppression of extracellular sodium signal due to its high concentration by use of multiple-quantum, triple-quantum filter with little success to suppress selectively extracellular or intracellular sodium with or without use of contrast agent [13-18]. However, noninvasive sodium MRI methods without contrast agent for imaging cancer tumors and other tissues are emerging as new methods.
Present study focused on quantitative approach to demonstrate possible correlation of single quantum and in vivo sodium MRI inversion recovery (IR) image signal intensities with MNU induced breast cancer histology and immunostaining features to monitor Taxotere chemosensitivity. Sodium MRI and tumor cytomorphology features appeared relevant as previously reported MNU rat model closely simulated with human breast cancer [19-22].
The present study indicated the reliability and possibility of correlation between sodium heterogeneity and pre-malignancy with improved both spatial resolution and in-plane resolution up to some extent. Main problems were identified as the poor signal-to-noise ratio and molecular diffusion during the measurement or encoding. Sensitivity was enhanced by using small high temperature sensitive micro-coil receivers. The problem of molecular diffusion was minimized with strong magnetic field 4.23 T imager suitable for clinical imaging equipped gradients to encode spatial information quickly. Previously, different transverse relaxation constants and MR signal intensities of different sodium concentrations in model solutions indicated the possibility of sodium measurements in breast tumor at different pre-malignant or malignant locations [23,24].
The higher total sodium signal-to-noise ratio (SNR) of SQ pulse sequence in Na-MRI was an advantage over the IR pulse sequence. Intracellular Na weighting was accomplished by applying this inversion recovery pulse sequence to null the signal from sodium nuclei with long T1 in tumors [24]. Both phantoms and tumor MRI experiments suggested global suppression by use of non-selective 180° pulse and TI = 25–30 msec and it confirmed our previous report [5]. Earlier maximum suppression was reported at a TI = 50 msec [8]. In present study, short T1 intracellular Na nuclei were used as in vivo marker of antineoplastic and chemosensitive effects against apoptosis in MNU induced breast tumors in rats.
The inversion recovery technique minimized the contribution of fat. However, it suffers from the limitation of maximum inversion recovery after applying 180° pulse due to sodium diffusion in molecular domains with different T1 values during the recovery from inversion pulse. Choice of short inversion time is not advantageous as it reduces the differences of null points of two sodium components at short repetition time (TR).
Premalignant and malignant lesions in rats developed by injection of MNU at 50 days old animals have shown several advantages [4]. The present study supports earlier reports of elevated intracellular sodium [Na]i in both benign and malignant breast tumors, neoplastic, well-differentiated and poorly differentiated tumors than in their normal tissue counterparts [24-26]. Ionic alterations are important events in malignant transformation, apoptosis and necrosis, and progression through the cell cycle. Antineoplastic agents increase the intracellular sodium ions. Antineoplastic drugs change cell cycle distribution and often lead to apoptosis and altered intracellular sodium [Na]i [27-30].
Taxotere injected doses were at the higher end of the maximum tolerable dose (MTD) range (40 nmol/L) for rat above the clinical levels (10 nmol/L) for 24 hours[21,22]. Present study exhibited comparison with placebo effect and control. These in vivo breast tissue studies and in vitro literature reports both corroborate on increased intracellular sodium MR image intensity following chemotherapy at subphysiological levels [20,21]. However, intervention of hypoxia or ouabain poisoning further increases [Na]i and IC-Na longitudinal T1 constants [20,21].
The assessment of Taxotere chemosensitivity effect was based on assumption that different tumor cell characteristics such as cell cycle during apoptosis, subcellular intracellular or extracellular (EC or IC) space, necrosis, cell viability, compartmentalization all affect the sodium diffusion and Taxotere restores intracellular sodium reserve. Earlier reports on correlation of short T1 constants and sodium signal intensities of interstitial intracellular sodium with apoptosis support our view significant in comparing the histology and MR images in tumors [25,27,28]. Different tumor features in different tumor regions were visualized on the 3-D IR image by setting the pixel intensity threshold to a point about 50% below the tumor peak intensity (see Figure 3, rightmost insert in bottom row).
The present study signified tumor features with cystic fluid spaces or cellular debris appeared as dark on the IR weighted images. Control tumor showed low IR sodium signal and larger non-viable center. These tumors served as positive control pre-Taxotere tumor. Post-Taxotere treated excised tumors showed different pre-malignancy and malignancy stages with apoptosis, necrosis and cystic fluid. Tumors with bright centers on MRI images appeared rich in early apoptosis nuclei (fragmented nuclei but smooth cell membranes) and central region with non-viable tissue. These results were suggestive of neoplasia having brighter apoptosis-rich regions and/or early necrosis regions as shown in Figure 6. Moreover, ploidy CAS 200 analysis and H & E stained slices also supported these observations.
Single-strand DNA monoclonal based method enhanced the power of detection for different stages of apoptosis in tumor that was consistent with other reports [16,25,27,28]. Other available histo-immunostaining methods may speculate better such as caspase based methods for different apoptosis staging in tumors [31,32]. The pixel intensity histogram pattern showed distinct main peak of S-DNA predominantly neoplasia. CAS 200 method was efficient in distinguishing cell cycle phases as S, M peaks [12].
Stereo-micrometric divisions on co-registered digital histology and IR images further enhanced dimensional accuracy with sufficient resolution to analyze the cellular details at different locations in the tumors. The % difference <5% between sodium MRI and histology observations suggested the possibility of diagnostic accuracy and correlation of IR sodium signal intensity with histology features. Spatial resolution of the image (~1 mm within each slice) enhanced the sensitivity and sodium MR signal intensity on images.
The present study has limitations. The relationship of sodium signal intensities and sodium concentrations in phantoms appears linear but in vivo tissues indicate poor sensitivity and need further improvement. Other limitation was very close sodium calculated values of different tumor features. It raises doubt of accuracy in predicting tumor stages. Since the tumor was excised after the final sodium MR imaging session that was not always the point of peak IR-Na image enhancement and tumor immunostaining methods were limited in postmortem analysis. The Taxotere treatment was limited to 24 hours. Origin of the bright entities on the IR images seems to be affected by several subphysiological factors and partial volume averaging (PVA) as possible association. Comparison between pre- and post treated tumor images and histology pre-malignant and malignant features were not conclusive. It may suffer due to different animal tumors and semi-automated or operator bias in the applied methods. A more detailed quantitative analysis is required to speculate biologic significance of different in IC-Na MRI signals for different specific tumor cell populations and specific sodium ion responses with molecular events induced by the chemotherapy.
The interaction of chemotherapy, cellular apoptosis and sodium MRI reflects the possibility of Na-MRI as clinical imaging possibility. Other purpose of present study was to analyze the association of sodium MR intensities with apoptosis and tumor histology features. In near future, sodium images can be co-registered with proton high resolution MRI images using double tuned probes. It seems a further advantage of IR technique that it does not require potentially toxic shift reagents as other sodium Na MRI methods use shift reagents [2,3,17,33-36].
Conclusion
The present study showed the intracellular sodium weighted inversion recovery MR pulse scheme for breast tumor sodium MR imaging and comparison with total sodium weighted single quantum MRI imaging and histo-immunostaining chemotherapeutic assessment. The intracellular sodium in breast tumor appears as associated with neoplasia, apoptosis and tumor histology features.
Acknowledgements
This manuscript in part was presented at peer-reviewed AFLAC award at AACR meet 2001, ISMRM workshop 2001 and ISMRM annual meet 2002. RS carried out the imaging, tumor propagation, histology, immunostaining, data analysis and manuscript writing. RPK participated in the tumor propagation, imaging. EXW standardized imaging sequence. JKK approved the manuscript and study protocol. All authors read and approved the final manuscript. Kenny Hess at Radiology department and Matthias Schbolcs at Pathology department provided control rat breast tumor histology/immunostaining expertise in these imaging and tumor histology analysis experiments.
==== Refs
Boada FE Shen GX Chang SY Thulborn KR Spectrally weighted twisted projection imaging: reducing T2 signal attenuation effects in fast three-dimensional sodium imaging Magn Reson Med 1997 38 1022 1028 9402205
Ouwerkerk R Bleich KB Gillen JS Pomper MG Bottomley PA Tissue sodium concentration in human brain tumors as measured with 23Na MR imaging Radiology 2003 227 529 537 12663825
Clayton DB Lenkinski RE MR imaging of sodium in the human brain with a fast three-dimensional gradient-recalled-echo sequence at 4 T Acad Radiol 2003 10 358 365 12678174 10.1016/S1076-6332(03)80023-5
Thompson HJ Singh M McGinley J Classification of Premalignant and malignant lesions developing in the rat mammary gland after injection of sexually immature rats with 1-Methyl 1 Nitrosourea J Mammary Gland Biology and Neoplasia 2000 5 201 210 10.1023/A:1026495322596
Kline R Wu EX Petrylak DP Szabolcs M Alderson PO Weisfeldt ML Cannon PJ Katz J Rapid in vivo monitoring of chemotherapeutic response using weighted sodium magnetic resonance imaging Clin Cancer Res 2000 6 2146 2156 10873063
Teh BS Vlachaki MT Aguilar LK Miles B Ayala G Lattime EC, Gerson SL Case study of combined gene and radiation therapy as an approach in the treatment of cancer Gene therapy of cancer 2000 Chapter 34 second Academic Press 513 524
Weissleder R Mahmood U Molecular imaging Radiology 2001 219 316 333 11323453
Song HK Wright AC Wolf RL Worli FW Multislice double inversion recovery pulse sequence for efficient black blood MRI Magn Reson Med 2002 47 616 620 11870851 10.1002/mrm.10094
Liska J Galbavy S Macejova D Zlatos J Brtko J Histopathology of mammary tumors in female rats treated with 1-methyl-1-nitrosourea Endocr Regul 2000 34 91 96 10911410
Shilkaitis A Green A Steele V Lubet R Kelloff G Christov K Neoplastic transformation of mammary epithelial cells in rats is associated with decreased apoptotic cell death Carcinogenesis 2000 21 227 233 10657962 10.1093/carcin/21.2.227
Kimi K Kumamoto H Ooya K Motegi K Immunohistochemical analysis of cell-cycle- and apoptosis-related factors in lining epithelium of odontogenic keratocysts J Oral Pathol Med 2001 30 434 42 11488422 10.1034/j.1600-0714.2001.300709.x
Lee S Tolmachoff T Marchevsky AM Marchevsky AM, Bartels PH DNA content analysis Image Analysis: A Primer for Pathologists 1994 Chapter 9 Raven Press, Ltd., New York 261 307
Kohler S Preibisch C Nittka A Hasse A Fast three-dimentional sodium imaging of human brain MAGMA 2001 13 63 69 11502419
Uemura K Toyama H Baba S Kimura Y Senda M Uchiyama A Generation of fractal dimension images and its application to automatic edge detection in brain MRI Comput Med Imaging Graph 2000 24 73 85 10767587 10.1016/S0895-6111(99)00045-2
Lin SP Song SK Miller JP Ackerman JJ Neil JJ Direct, longitudinal comparison of (1)H and (23) Na MRI after transition focal cerebral ischemia Stroke 2001 32 925 32 11283393
Frankfurt OS Krishan A Identification of apoptotic cells by formamide-induced DNA denaturation in condensed chromatin J Histochem Cytochem 2001 49 363 378
Duvvuri U Leigh JS Reddy R Detection of residual quadruple interaction in the human breast in vivo using sodium-23 multiple quantum spectroscopy J Magn Reson Imaging 1999 9 391 394 10194708 10.1002/(SICI)1522-2586(199903)9:3<391::AID-JMRI5>3.0.CO;2-4
Miller JR Zhang K Ma QY Mun IK Jung KJ Katz J Face DW Kountz DJ Superconducting receiver coils for sodium magnetic resonance imaging IEEE Trans Biomed Eng 1996 43 1197 9 9214839 10.1109/10.544344
Lu J Jiang C Mitrenga T Cutter G Thompson HJ Pathogenic characterization of 1-methyl-1-nitrosourea-induced mammary carcinoma in the rat Carcinogenesis 1996 19 223 227 9472716 10.1093/carcin/19.1.223
Thordarson G Lee AV MacCarty M Van Horn K Chu O Chou YC Growth and characterization of N-methyl-N-nitrosourea-induced tumors in intact and ovariectomized rats Carcinogenesis 2001 22 2039 2047 11751437 10.1093/carcin/22.12.2039
McGinley JN Knott KK Thompson HJ Semi-automated method of quantifying vasculature of 1-methyl-1-nitrosourea-induced rat mammary carcinomas using immunohistochemical detection J Histochem Cytochem 2002 50 213 222 11799140
Knott KK McGinley JN Lubet RA Steele VE Thompson HJ Effect of aromatase inhibitor vorozole on esterogen and progesterone receptor content of rat mammary carcinoma induced by 1-methyl-1-nitrosourea Breast Cancer Res Treat 2001 70 171 183 11804181 10.1023/A:1013051107535
Rooney WD Springer CS Jr The molecular environment of intracellular sodium: 23Na NMR relaxation NMR in Biomedicine 1991 4 227 245 1751346
Liebling MS Gupta RK A comparison of intracellular sodium ion concentrations in neoplastic and nonneoplastic human tissue using 23Na NMR spectroscopy Ann NY Acad Sci 1987 508 149 3439700
Barbiero G Duranti F Bonelli G Amenta JS Baccino FM Intracellular ionic variations in the apoptotic death of L cells by inhibitors of cell cycle progression Experimental Cell Res 1995 217 410 418 10.1006/excr.1995.1104
Zs.-Nagy I Lustyik V Lukacs G Zs.-Nagy V Balazs G Correlation of malignancy with intracellular Na:K ratio in human thyroid tumors Cancer Res 1983 43 5395 5402 6616471
Elledge SJ Cell Cycle Checkpoints: preventing an identity crisis Science 1996 274 1664 1671 8939848 10.1126/science.274.5293.1664
King KL Cidlowski JA Cell cycle and apoptosis: common pathways to life and death J Cell Biochem 1995 58 175 180 7673325 10.1002/jcb.240580206
Bruno R Hille D Riva A Vivier N ten Bokkel Huinnink WW van Oosterom AT Kaye SB Verweij J Fossella FV Valero V Rigas R Seidman AD Chevallier B Fumoleau P Burris HA Ravdin PM Sheiner LB Population pharmacokinetics/pharmacodynamics of docetaxel in phase II studies of patients with cancer J Clin Oncol 1998 16 187 196 9440742
Gorczyca W Gong J Darzynkiewicz Z Detection of DNA strand breaks in Individual Apoptotic cells by the in situ terminal deoxynucleotidyl transferase and nick translation assays Cancer Research 1993 53 1945 1951 8467513
Harootunian AT Kao JPY Eckert BK Tsien RY Fluorescence ratio imaging of cytosolic free Na in individual fibroblasts and lymphocytes J Biol Chem 1991 264 19458 19467 2478559
Charriaut-Marlangue C Margaill I Represa A Popovici T Plotkine M Ben-Ari Y Apoptosis and necrosis after reversible focal ischemia: an in situ DNA fragmentation analysis J Cereb Blood Flow Metab 1996 16 186 94 8594049 10.1097/00004647-199603000-00002
Summers RM Joseph PM Renshaw PF Kundel HL Dextran-magnetite: a contrast agent for sodium-23 MRI Magn Reson Med 1988 8 427 39 2466190
Hashimoto T Ikehira H Fukuda H Yamaura A Watanabe O Tateno Y Tanaka R Simon HE In vivo sodium-23 MRI in brain tumors: evaluation of preliminary clinical experience Am J Physiol Imaging 1991 6 74 80 1867865
Turski PA Houston LW Perman WH Hald JK Turski D Strother CM Sackett JF Experimental and human brain neoplasms: detection with in vivo sodium MR imaging Radiology 1987 163 245 9 3029803
Winkler SS Thomasson DM Sherwood K Perman WH Regional T2 and sodium concentration estimates in the normal human brain by sodium-23 MR imaging at 1.5 T J Comput Assist Tomogr 1989 13 561 66 2745773
|
16078994
|
PMC1208921
|
CC BY
|
2021-01-04 16:24:16
|
no
|
Cancer Cell Int. 2005 Aug 3; 5:26
|
utf-8
|
Cancer Cell Int
| 2,005 |
10.1186/1475-2867-5-26
|
oa_comm
|
==== Front
J CarcinogJournal of Carcinogenesis1477-3163BioMed Central London 1477-3163-4-161613892710.1186/1477-3163-4-16ResearchSulindac metabolites inhibit epidermal growth factor receptor activation and expression Pangburn Heather A [email protected] Hanna [email protected] Dennis J [email protected] Pamela L [email protected] Molecular Toxicology and Environmental Health Sciences Program, Department of Pharmaceutical Sciences, University of Colorado Health Sciences Center, Denver, USA2 Veterans Administration Medical Center, Denver, USA3 Department of Medicine, University of Colorado Health Sciences Center, Denver, USA4 University of Colorado Comprehensive Cancer Center, Denver, USA2005 2 9 2005 4 16 16 20 6 2005 2 9 2005 Copyright © 2005 Pangburn et al; licensee BioMed Central Ltd.2005Pangburn et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Regular use of nonsteroidal anti-inflammatory drugs (NSAIDs) is associated with a decreased mortality from colorectal cancer (CRC). NSAIDs induce apoptotic cell death in colon cancer cells in vitro and inhibit growth of neoplastic colonic mucosa in vivo however, the biochemical mechanisms required for these growth inhibitory effects are not well defined. We previously reported that metabolites of the NSAID sulindac downregulate extracellular-signal regulated kinase 1/2 (ERK1/2) signaling and that this effect is both necessary and sufficient for the apoptotic effects of these drugs. The goal of this project was to specifically test the hypothesis that sulindac metabolites block activation and/or expression of the epidermal growth factor (EGF) receptor (EGFR).
Methods
HT29 human colon cancer cells were treated with EGF, alone, or in the presence of sulindac sulfide or sulindac sulfone. Cells lysates were assayed by immunoblotting for phosphorylated EGFR (pEGFR, pY1068), total EGFR, phosphorylated ERK1/2 (pERK1/2), total ERK1/2, activated caspase-3, and α-tubulin.
Results
EGF treatment rapidly induced phosphorylation of both EGFR and ERK1/2 in HT29 colon cancer cells. Pretreatment with sulindac metabolites for 24 h blocked EGF-induced phosphorylation of both EGFR and ERK1/2 and decreased total EGFR protein expression. Under basal conditions, downregulation of pEGFR and total EGFR was detected as early as 12 h following sulindac sulfide treatment and persisted through at least 48 h. Sulindac sulfone induced downregulation of pEGFR and total EGFR was detected as early as 1 h and 24 h, respectively, following drug treatment, and persisted through at least 72 h. EGFR downregulation by sulindac metabolites was observed in three different CRC cell lines, occurred prior to the observed downregulation of pERK1/2 and induction of apoptosis by these drugs, and was not dependent of caspase activation.
Conclusion
These results suggest that downregulation of EGFR signaling by sulindac metabolites may occur, at least in part, by inhibiting activation and expression of EGFR. Inhibition of EGFR signaling may account for part of the growth inhibitory and chemopreventive effects of these compounds.
==== Body
Background
CRC is the second most common cause of cancer death in the USA, with an estimated annual incidence of 104,950 and mortality of 56,290 in 2005 [1]. The lifetime risk of developing CRC in the general US population is almost 6% [1]. Effective preventive measures could substantially reduce both the incidence and mortality from CRC. NSAIDs are one of the most widely studied and promising groups of compounds for CRC prevention.
NSAIDs mediate their anti-inflammatory effects by inhibiting the enzymatic activity of cyclooxygenase-1 (COX-1) and/or COX-2. Sulindac is a non-selective NSAID that inhibits both COX-1 and COX-2. Sulindac is rapidly metabolized in the liver to two major metabolites; 1) sulindac sulfide, which is an active NSAID, and 2) sulindac sulfone, which does not inhibit COX enzymatic activities and thus is not an NSAID. A large body of evidence from cell culture, animal model, human epidemiologic and clinical studies indicates that NSAIDs, including sulindac and aspirin, have potent chemopreventive and chemoregressive properties against colon cancer [2,3]. Although substantial evidence indicates that NSAIDs inhibit the growth of neoplastic colonic mucosa, the biological and biochemical mechanisms responsible for the growth inhibitory effects of these drugs are not well defined. The chemopreventive and chemoregressive effects of NSAIDs may not be due to inhibition of COX alone as we have shown that sulindac sulfide and sulindac sulfone both inhibit the phosphorylation of ERK1/2 in HCT15 cells that do not express COX-1 or COX-2 [4]. Additionally, sulindac sulfone, the non-NSAID metabolite of sulindac, has been shown, by our lab and others, to induce apoptosis of cancer cells in vitro, prevent tumor formation in animal models, and cause regression of adenomas in familial adenomatous polyposis (FAP) [2,5].
Several biological mechanisms for the chemopreventive effects of NSAIDs have been proposed including inhibition of proliferation, induction of apoptosis, and inhibition of angiogenesis. Our laboratory [5,6] and others [2] have reported that NSAIDs inhibit growth of CRC cells in culture primarily by the induction of apoptotic cell death. An apoptotic mechanism was also suggested in human adenomas treated with sulindac sulfone [7]. We have reported that the apoptotic effect of sulindac is not dependent on COX inhibition [4,5,8], but does appear to be dependent on the downregulation of ERK1/2 [8].
Work from several laboratories has demonstrated substantial interactions between the biochemical effects of NSAIDs and EGFR signaling. It is well established that COX-2 protein expression is stimulated by EGF and decreased by EGFR inhibitors [9]. Sulindac sulfide and indomethacin inhibit TGFα induced prostaglandin production and thymidine incorporation in RIE-1 cells [10], and indomethacin, ibuprofen, and aspirin all block EGF-induced Ca++ influx in Caco-2 colon cancer cells [11]. Furthermore, combination therapy of NSAIDs and EGFR antagonists display an additive effect against colon tumor development in vivo [12] and two previous studies suggest that NSAIDs might inhibit EGFR signaling [13,14]. Finally, sulindac has been shown to inhibit expression of ErbB2/HER2 protein expression in rectal mucosa of FAP patients [15]. The biochemical mechanisms by which NSAIDs might modulate EGFR signaling has not been addressed in any of these studies.
We have previously shown that sulindac sulfide and sulindac sulfone inhibit ERK1/2 phosphorylation both basally and in response to EGF [6], and that this inhibition is sufficient [5] and necessary [8] for the apoptotic effect of these drugs. The goal of this study was to specifically test the hypothesis that sulindac metabolites block activation and/or expression of EGFR (Figure 1). We report, for the first time, that sulindac metabolites block both basal and EGF induced phosphorylation of EGFR and downregulate total EGFR protein expression. Together these results identify a new target for the actions of sulindac and suggest that inhibition of EGFR activation may be a major biochemical mechanism for both the downregulation of ERK1/2 signaling and the chemopreventive activity of NSAIDs.
Figure 1 The EGFR/MAPK/ERK1/2 signaling pathway. We have previously shown that sulindac metabolites inhibit both MEK1/2 and ERK1/2 activation. The goal of this study was to determine if this inhibition is due to downregulation of the EGFR.
Methods
Reagents
Cell culture media and fetal bovine serum (FBS) were purchased from Mediatech (Herndon, VA), antibiotic/antimycotic solution (penicillin/streptomycin/fungizone) from Life Technologies, Inc. (Grand Island, NY), and tissue culture plates from Falcon (Franklin Lakes, NJ). Primary antibodies raised against phosphorylated ERK1/2 (Thr202/Tyr204), total ERK1/2, and total EGFR were purchased from Santa Cruz Biotechnology (Santa Cruz, CA); primary antibody against phosphorylated EGFR(Tyr1068), horseradish peroxidase-conjugated anti-mouse and anti-rabbit secondary antibodies, and EGF were purchased from Biosource (Camarillo, CA). Primary antibody against caspase-3 (recognizes both full length caspase-3 and cleavage products) and cleaved caspase-3 (recognizes only cleavage products of activated caspase-3 resulting from cleavage adjacent to Asp175) were obtained from Cell Signaling Technology (Beverly, MA). Immobilon-P membranes were purchased from Millipore (Bedford, MA), chemiluminescent visualization reagents from NEN (Boston, MA), and X-ray film from Pierce (Rockford, IL). Sulindac sulfide was from LKT Laboratories Inc. (St. Paul, MN). ZVAD-fmk (caspase inhibitor I) and primary antibody against α-tubulin were purchased from Calbiochem (San Diego, CA). Sulindac sulfone was a generous gift from Cell Pathways Inc. (Horsham, PA).
Cell culture
HT29, Caco-2, and HCT116 human colon cancer cells were purchased from American Type Culture Collection (Manassas, VA). HT29 and HCT116 cells were maintained in RPMI 1640 media supplemented with 10% FBS and 1% penicillin/streptomycin/fungizone solution; Caco-2 cells were maintained in DMEM media with 4 mM L-glutamine adjusted to contain 1.5 g/L glucose and supplemented with 0.01 mg/ml human transferring, 10% FBS, and 1% penicillin/streptomycin/fungizone solution. Cells were plated and grown on tissue culture plates as specified in the results section.
Morphologic quantitation of apoptotic cell death
An ethidium bromide/acridine orange double-dye morphological assay [16] was used to quantitate apoptotic cell death induced by sulindac metabolites. Cells were grown in multi-well plates and given the appropriate drug treatment. Three days after treatment, cells were harvested and stained with a mixture of 100 μg/ml ethidium bromide and 100 μg/ml acridine orange and then examined by fluorescence microscopy. Three fields of 100 cells each were counted and the number of apoptotic cells (as determined by nuclear morphology) as well as the number of living and necrotic cells were expressed as a percentage of total cells counted.
Cell lysates and Western immunoblotting
At time of harvest, cells were scraped from the plates, washed with ice-cold phosphate buffered saline (PBS), and pelleted at 2,400 × g for 5 min. Care was taken to keep cells as cold as possible during the entire procedure. After aspirating the supernatant, the cells were washed twice with ice cold PBS and centrifuged. The cell pellet was lysed in cell extraction buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 20 mM Na4P2O7, 2 mM Na3VO4, 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate, 1 mM PMSF, 60 μg/mL aprotinin, 10 μg/mL leupeptin, 1 μg/mL pepstatin) for 30 min, on ice, with vortexing at 10 min intervals. Lysates were then centrifuged at 18,000 × g for 10 min at 4°C, and supernatants collected. Protein concentrations of lysates were determined by the method of Lowry [17].
Lysate samples (50 μg total protein) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electrotransferred overnight onto Immobilon-P polyvinylidene difluoride membranes. Nonspecific binding was blocked by immersing membranes in Tris-neutral saline with 1% (w/v) dry milk and 0.05% Tween 20 for 30 min at room temperature, then blots were incubated with pEGFR, pERK1/2, caspase 3, cleaved caspase-3, or α-tubulin primary antibody while rocking at 37°C for 1 h. Immunoreactive protein was detected by incubating blots with horseradish peroxidase-conjugated secondary antibody for 1 h followed by chemiluminescent substrate for 1 min. Membranes probed for phospho-proteins were then stripped and reprobed with total EGFR or total ERK1/2 primary antibody followed by horseradish peroxidase-conjugated secondary antibody as described above. Non-stripped control blots were run to show that this stripping procedure gave the same results as primary blots with each antibody. Immunoreactive proteins were visualized by exposure to radiographic film. Quantitation of protein levels was determined by densitometry using a visual imaging system (UN-SCAN-IT gel, Silk Scientific, Inc., Orem, UT).
Statistical Analysis
Data was analyzed by one way ANOVA using a Newman-Keuls multiple comparison test and statistical significance accepted at p < 0.05 (GraphPad Prism 4.0, GraphPad Software Inc., San Diego, CA).
Results
EGF induces phosphorylation of EGFR and ERK1/2 in HT29 human colon cancer cells
In order to verify that EGFR signaling was intact in our cells, the effects of EGF treatment on the expression and activation of EGFR and ERK1/2 in HT29 human colon cancer cells was examined. It was observed that EGF rapidly induced a > 7 fold increase of pEGFR that was maximal at 10 min (Figure 2A and 2B). Similarly, EGF induced a rapid, > 30 fold, increase of pERK1/2 that was also maximal at 10 min (Figure 2A and 2C). EGF-induced phosphorylation of EGFR and ERK1/2 persisted through 60 min, and was no longer detected at 24 h (data not shown). Similar results were obtained in HCT116 and Caco-2 human colon cancer cell lines (data not shown).
Figure 2 EGF induces EGFR and ERK1/2 phosphorylation. HT29 human colon cancer cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 48 h before treatment with vehicle (water), 10, or 100 ng/ml EGF. Cells were harvested 10 min after EGF treatment and lysates prepared for (A) Immunoblotting with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, and total ERK1/2; α-tubulin immunoblots of the same lysates served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and pERK1/2 bands (C) normalized for the loading controls. Data represent mean of triplicate samples ± SD; statistical significance is denoted by ***p < 0.001 versus vehicle. Results shown in the figure are representative of 2 separate experiments each with triplicate samples.
Sulindac metabolites block EGF-induced phosphorylation of EGFR and ERK1/2, downregulate total EGFR protein, and induce apoptotic cell death
We next determined the effect of sulindac sulfide on EGF-induced phosphorylation and expression of EGFR and ERK1/2. Pretreatment of HT29 cells for 24 h with 40–160 μM sulindac sulfide blocked the ability of EGF to induce phosphorylation of EGFR in a dose dependent fashion. Total EGFR expression and EGF-induced phosphorylation of ERK1/2 proteins was also downregulated following pretreatment with sulindac sulfide (Figure 3). There was no effect of drug or EGF treatment on total ERK1/2 expression.
Figure 3 Dose response of sulindac sulfide inhibition of EGFR. HT29 cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 24 h before addition of drug. Cells were then treated with vehicle (0.1% DMSO), 40, 80, 120, or 160 μM sulindac sulfide, drug doses previously shown to induce apoptotic cell death in these cells. Twenty four hours after drug treatment, vehicle or 10 ng/ml EGF was added and cells were harvested 10 min later. (A) Immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and caspase-3; total ERK1/2 immunoblots served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Results shown in figure are representative of 3 separate experiments.
The effect of sulindac sulfone on EGF-induced phosphorylation of EGFR and ERK1/2 was also examined. As seen with sulindac sulfide, pretreatment of HT29 cells for 24 h with 200–800 μM sulindac sulfone markedly decreased the ability of EGF to induce phosphorylation of EGFR; total EGFR expression levels and ERK1/2 phosphorylation were also downregulated (Figure 4), albeit to a lesser extent than with sulindac sulfide.
Figure 4 Dose response of sulindac sulfone inhibition of EGFR. HT29 cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 24 h before addition of drug. Cells were then treated with vehicle (0.2% DMSO), 200, 400, 600, or 800 μM sulindac sulfone, drug doses previously shown to induce apoptotic cell death in these cells. Twenty four hours after drug treatment, vehicle or 10 ng/ml EGF was added and cells were harvested 10 min later. (A) Immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and caspase-3; total ERK1/2 immunoblots served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Results shown in figure are representative of 3 separate experiments.
The finding that total EGFR protein expression was decreased by sulindac metabolites (Figures 3 and 4) was unexpected; we therefore confirmed these results using a second antibody (Biosource catalog #AHR5062). The confirmatory EGFR antibody is an EGFR antibody cocktail (composed of four monoclonal antibodies) directed against both the extracellular and cytoplasmic domains of EGFR. The same downregulation of total EGFR elicited by sulindac sulfide and sulfone was found using this antibody (data not shown), confirming that these results are not peculiar to a single EGFR epitope or a result of an artifact of the use of a single antibody.
EGF had no effect on morphological apoptosis or activation of caspase-3 but treatment with 40–160 μM sulindac sulfide (Figure 3) or 600–800 μM sulindac sulfone (Figure 4) induced substantial apoptotic cell death, as determined by cleavage of caspase-3 and confirmed with the acridine orange and ethidium bromide morphologic assay (data not shown).
Sulindac metabolites downregulate basal EGFR phosphorylation and expression
Next we examined the time course of sulindac's effects on EGFR expression and phosphorylation in cells grown in 10% FBS (basal conditions). Phosphorylation of EGFR and total EGFR protein expression were downregulated as early as 12 h and persisted at 24 h following sulindac sulfide treatment (Figure 5). Significant downregulation of pERK1/2 was not detected prior to 24 h in these experiments. Evidence of apoptosis, by presence of caspase-3 cleavage products (Figure 5A) was also first detected at 24 h; examination of nuclear morphology after staining with acridine orange and ethidium bromide at 48 h confirmed cleaved caspase-3 immunoblot results (data not shown). Sulindac sulfide treatment did not consistently affect levels of total ERK1/2 or α-tubulin proteins. For this experiment immunoblots for each time point were run on individual gels thus, only samples within a single time point, assayed for one particular protein, can be directly compared.
Figure 5 Dose response and time course of sulindac sulfide inhibition of EGFR. HT29 cells were grown to confluence in medium containing 10% FBS and treated with vehicle (0.1% DMSO), 160, or 180 μM sulindac sulfide for 1 h, 12 h, and 24 h. Cells were then harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and cleaved caspase-3; α-tubulin immunoblots of the same lysates served as loading controls. (A) 1 h, 12 h, and 24 h immunoblot results. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Data represent mean of triplicate samples ± SD; statistical significance is denoted by **p < 0.01 and ***p < 0.001 versus respective time point vehicle. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
The effect of sulindac sulfone on basal EGFR activation and expression was also examined. Sulindac sulfone treatment resulted in a modest but significant decrease in EGFR phosphorylation as early as 1 h after treatment, and this effect increased and persisted through 24 h (Figure 6A and 6B). Downregulation of total EGFR (Figure 6A and 6C) and pERK1/2 (Figure 6A) proteins was first seen at 24 h; caspase-3 cleavage (Figure 6A) was also first detected at 24 h. Morphologic evidence of apoptosis was apparent at 48 h, the earliest time point examined in this manner. Sulfone treatment did not consistently affect levels of total ERK1/2 or α-tubulin proteins.
Figure 6 Dose response and time course of sulindac sulfone inhibition of EGFR. HT29 cells were grown to confluence in medium containing 10% FBS followed by treatment with vehicle (0.2% DMSO), 400, or 600 μM sulindac sulfone for 1 h, 12 h, and 24 h. Cells were then harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and cleaved caspase-3; α-tubulin immunoblots of the same lysates served as loading controls. (A) 1 h, 12 h, and 24 h Western blot results. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Data represent mean of triplicate samples ± SD; statistical significance is denoted by **p < 0.01 and ***p < 0.001 versus respective time point vehicle. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
To exclude the possibility that downregulation of EGFR by sulindac metabolites was unique to HT29 cells, similar experiments were carried out in Caco-2 and HCT116 cells. Sulindac sulfide and sulfone downregulated expression of both pEGFR and total EGFR in these cell lines as well (data not shown).
ZVAD blocks sulindac-induced apoptosis but not EGFR downregulation
Although the downregulation of EGFR observed with sulindac metabolites occurred prior to either biochemical or morphologic evidence of apoptosis, it remains possible that some of the decrease in EGFR protein expression could be a consequence of caspase activation by sulindac metabolites. To examine this possibility, HT29 cells were pretreated with vehicle or 25 μM of the broad specificity caspase inhibitor ZVAD-fmk (ZVAD) for 60 minutes prior to treatment with 600 μM sulindac sulfone or vehicle for 48 h. ZVAD pretreatment completely blocked sulindac sulfone induced apoptosis as measured by examination of nuclear morphology (Figure 7A) and it prevented cleavage of caspase 3 to the active 17 and 19 kDa fragments (Figure 7B; the 20 kDa caspase 3 fragment seen with the combination of sulfone and ZVAD has been shown to be an inactive cleavage product [26-29]. While 25 uM ZVAD prevented both caspase activation and morphologic apoptosis induced by sulindac sulfone, it did not prevent sulindac sulfone induced decreases in either total or pEGFR (Figure 7B–D). Combination experiments with ZVAD and sulindac sulfide showed similar results as those shown in Figure 7 for sulindac sulfone. These results demonstrate that downregulation of EGFR by sulindac metabolites is not a consequence of the apoptotic effect of these drugs.
Figure 7 Effect of the caspase inhibitor, ZVAD, on apoptosis and inhibition of EGFR. HT29 colon cancer cells were grown to confluence in medium containing 10% FBS followed by pretreatment with or without 25 μM zvad for 1 h. Cells were then treated with vehicle (0.2% DMSO) or 600 μM sulfone for 48 h. Cells were harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, and cleaved caspase 3; α-tubulin immunoblots of the same lysates served as loading controls. The graphs show morphological apoptosis results (A) 48 h Western immunoblot results (B) and densitometry of the pEGFR bands (C) and total EGFR bands (D). Data represent mean of triplicate samples ± SD; statistical significance is denoted by *p < 0.05, **p < 0.01 and ***p < 0.001. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
Discussion
Our laboratory originally reported that sulindac metabolites inhibit the EGFR/mitogen activated protein kinase (MAPK)/ERK1/2 signaling pathway at the level of MEK or higher [5] (Figure 1) and that the downregulation of mitogen-activated protein kinase kinase (MEK)/ERK1/2 is both necessary and sufficient for the apoptotic effect of these drugs [8]. We now report, for the first time, that sulindac metabolites inhibit EGFR phosphorylation and downregulate the expression of total EGFR protein in human colon cancer cells. Downregulation of EGFR occurred at doses and within a time course that is consistent with the possibility that the effect of drug on the EGFR could be the mechanism of the inhibition of MEK1/2 and ERK1/2 activity and apoptotic cell death induced by these drugs.
Both sulindac sulfide and sulindac sulfone downregulate EGFR expression and phosphorylation at comparable doses to those that inhibit ERK1/2 phosphorylation and induce apoptotic cell death (Figures 5 and 6). The time course experiments suggest that the kinetics of the EGFR effect may differ between the two sulindac metabolites. With sulindac sulfide, the effect on EGFR and ERK1/2 occurred concurrently (first seen at 12 h) and before any detectable effect on apoptosis. This suggests that the downregulation of EGFR may be responsible for some or all of the ERK1/2 inhibitory effects of the drug. The relationship between sulindac sulfone's effect on EGFR and ERK1/2 is not as clear since an effect on pEGFR was detected substantially earlier (1 hour) than the effect on pERK1/2 and apoptosis (24 hours). These differences in the kinetics of the two sulindac metabolites on pEGFR and pERK1/2 downregulation could represent differences in potency, duration of effect, the spectrum of EGFR phosphorylation sites that are affected by the different drugs, differential effects on phosphatases, and/or differences in the molecular mechanism of action. Similar to our results, Jimeno et al. reported that treatment of HuCCT cells with the EGFR tyrosine kinase inhibitor (TKI) erlotinib or the mononclonal antibody (MAb) cetuximab led to significant downregulation of pEGFR as early as 1 h following treatment whereas major downstream effectors (pERK1/2 and pAKT) were not coordinately affected [18]. Despite these differences in kinetics, the finding that EGFR downregulation was observed consistently with both sulindac metabolites and in three different CRC cell lines (HT29, HCT116, and Caco-2) establishes the EGFR as a new target of sulindac.
Activation of the EGFR pathway is particularly common in colorectal cancers [19-22] and overexpression of EGFR correlates with more aggressive disease and poor prognosis [23-26]. Both EGFR MAbs and TKIs that block activation of EGFR are now being used for the treatment of colorectal cancer [27]. In light of our evidence that NSAIDs downregulate EGFR signaling and induce apoptosis, it is interesting to note that both EGFR MAbs and TKIs have been shown to stimulate important pro-apoptotic genes and downregulate anti-apoptotic genes, particularly those of the Bcl-2 family [28-30]. Furthermore, treatment with combinations of agents that inhibit EGFR by different mechanisms is more effective than treatment with a single agent. For example, the combination of erlotinib plus cetuximab caused a synergistic induction of apoptosis in vitro, and an additive effect on tumor growth arrest in vivo [18]. Thus, downregulation of EGFR could mediate the apoptotic and chemopreventive effects of NSAIDs and could explain some of the clinical evidence of additive growth inhibitory effects of EGFR inhibitors and NSAIDs. These data suggest a biochemical rationale for the use of combinations of classic EGFR inhibitors and NSAIDs in the treatment or prevention of CRC.
If inhibition of EGFR signaling is a mechanism of NSAIDs as our data indicates, it will be of great interest to examine the full spectrum of possible consequences of downregulation of EGFR. Thus far, we have only examined the ras-raf-MEK1/2-ERK1/2 arm of EGFR signaling but several other signaling pathways are known to be activated by the EGFR including the phosphatidylinositol-3-kinase (PI3-K) and Janus kinase (JAK)-signal transducers and activators of transcription (STAT) pathways [31]. Sulindac metabolites may also affect other members of the ErbB family of receptor tyrosine kinases (RTK) including Erb2, ErbB3, ErbB4, non-ErbB RTKs including PDGFR and VEGFR, and/or receptors lacking tyrosine kinase activity such as the adenosine receptor. Studies to examine the effect of NSAIDs on these pathways and receptors are underway.
We have not systematically examined the mechanism by which sulindac metabolites downregulate EGFR but the relatively short time course (as early as 1 hour) suggests that the effect is more likely due to increased degradation of EGFR rather than inhibition of new protein synthesis. Our data demonstrate that EGFR inhibition is not a consequence of caspase activation or due to the apoptotic effects of the drugs in that apoptosis was first detected 24 h after sulindac metabolite treatment, whereas downregulation of EGFR was seen as early as 1 to 12 hours after drug treatment. In addition, inhibition of apoptosis with the broad specificity caspase inhibitor ZVAD did not prevent EGFR downregulation by sulindac metabolites. Thus, we hypothesize that EGFR downregulation by sulindac metabolites is not due to activation of caspases, but rather is a result of increased proteasomal and/or lysosomal degradation. Further studies are warranted to determine the precise mechanism of EGFR downregulation by sulindac metabolites.
Sulindac metabolites decreased expression of both total and pEGFR proteins. The time course experiments, particularly those with sulindac sulfone (Figure 6), suggest that downregulation of these two forms of the receptor could be occurring by distinct mechanisms. Although total and pEGFR downregulation with sulindac sulfide occurred concurrently, the sulfone metabolite decreased pEGFR as early as 1 h whereas downregulation of total EGFR protein was first observed at the 24 h time point. This sequence indicates that the downregulation of pEGFR is not likely to be a consequence of the downregulation of the total receptor, at least with sulindac sulfone.
Conclusion
In summary, we have described for the first time that sulindac metabolites downregulate EGFR and prevent EGFR phosphorylation by EGF. This effect may explain, at least in part, our previous observation that these drugs inhibit basal and EGF-induced activation of the MEK/ERK arm of the EGFR signaling pathway. Downregulation of EGFR was shown to be independent of COX inhibitory activity as it occurred with sulindac sulfone. These results add to the increasing amount of data that suggest interactions exist between the effects of NSAIDs and EGFR signaling [13,14]. Our results are the first to suggest that these interactions are due to a direct drug effect on the EGFR itself.
Abbreviations
NSAID, non-steroidal anti-inflammatory drug; CRC, colorectal cancer; ERK1/2, extracellular-signal regulated kinase; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; pEGFR, phosphorylated epidermal growth factor receptor; pERK1/2, phosphorylated extracellular-signal regulated kinase 1/2; COX, cyclooxygenase; FAP, familial adenomatous polyposis; FBS, fetal bovine serum; PBS, phosphate buffered saline; MAPK, mitogen-activated protein kinase; MEK1/2, mitogen-activated protein kinase kinase; MAb, monoclonal antibody; TKI, tyrosine kinase inhibitor; PI3-K, phosphatidylinositol-3-kinase; JAK/STAT, Janus kinase/signal transducers and activators of transcription; RTK, receptor tyrosine kinase
Acknowledgements
Funding provided by Department of Veterans Affairs Merit Review Program and Cancer Research and Prevention Foundation
==== Refs
Jemal A Murray T Ward E Samuels A Tiwari RC Ghafoor A Feuer EJ Thun MJ Cancer statistics, 2005 CA Cancer J Clin 2005 55 10 30 15661684
Gupta RA Dubois RN Colorectal cancer prevention and treatment by inhibition of cyclooxygenase-2 Nat Rev Cancer 2001 1 11 21 11900248 10.1038/35094017
Thun MJ Henley SJ Patrono C Nonsteroidal anti-inflammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues J Natl Cancer Inst 2002 94 252 66 11854387
Rice PL Kelloff J Sullivan H Driggers LJ Beard KS Kuwada S Piazza G Ahnen DJ Sulindac metabolites induce caspase- and proteasome-dependent degradation of beta-catenin protein in human colon cancer cells Mol Cancer Ther 2003 2 885 92 14555707
Rice PL Goldberg RJ Ray EC Driggers LJ Ahnen DJ Inhibition of extracellular signal-regulated kinase 1/2 phosphorylation and induction of apoptosis by sulindac metabolites Cancer Res 2001 61 1541 7 11245463
Rice PL Washington M Schleman S Beard KS Driggers LJ Ahnen DJ Sulindac sulfide inhibits epidermal growth factor-induced phosphorylation of extracellular-regulated kinase 1/2 and Bad in human colon cancer cells Cancer Res 2003 63 616 20 12566304
Piazza GA Alberts DS Hixson LJ Paranka NS Li H Finn T Bogert C Guillen JM Brendel K Gross PH Sperl G Ritchie J Burt RW Ellsworth L Ahnen DJ Pamukcu R Sulindac sulfone inhibits azoxymethane-induced colon carcinogenesis in rats without reducing prostaglandin levels Cancer Res 1997 57 2909 15 9230200
Rice PL Beard KS Driggers LJ Ahnen DJ Inhibition of extracellular-signal regulated kinases 1/2 is required for apoptosis of human colon cancer cells in vitro by sulindac metabolites Cancer Res 2004 64 8148 51 15548677
Coffey RJ Hawkey CJ Damstrup L Graves-Deal R Daniel VC Dempsey PJ Chinery R Kirkland SC DuBois RN Jetton TL Morrow JD Epidermal growth factor receptor activation induces nuclear targeting of cyclooxygenase-2, basolateral release of prostaglandins, and mitogenesis in polarizing colon cancer cells Proc Natl Acad Sci U S A 1997 94 657 62 9012840 10.1073/pnas.94.2.657
DuBois RN Awad J Morrow J Roberts LJ 2ndBishop PR Regulation of eicosanoid production and mitogenesis in rat intestinal epithelial cells by transforming growth factor-alpha and phorbol ester J Clin Invest 1994 93 493 8 8113389
Kokoska ER Smith GS Miller TA Nonsteroidal anti-inflammatory drugs attenuate proliferation of colonic carcinoma cells by blocking epidermal growth factor-induced Ca++ mobilization J Gastrointest Surg 2000 4 150 61 10675238 10.1016/S1091-255X(00)80051-7
Torrance CJ Jackson PE Montgomery E Kinzler KW Vogelstein B Wissner A Nunes M Frost P Discafani CM Combinatorial chemoprevention of intestinal neoplasia Nature Med 2000 6 1024 1028 10973323 10.1038/79534
Kokoska ER Smith GS Wolff AB Deshpande Y Miller TA Nonsteroidal anti-inflammatory drugs attenuate epidermal growth factor-induced proliferation independent of prostaglandin synthesis inhibition J Surg Res 1999 84 186 92 10357918 10.1006/jsre.1999.5640
Kinoshita T Takahashi Y Sakashita T Inoue H Tanabe T Yoshimoto T Growth stimulation and induction of epidermal growth factor receptor by overexpression of cyclooxygenases 1 and 2 in human colon carcinoma cells Biochim Biophys Acta 1999 1438 120 30 10216286
Winde G Lugering N Glodny B Schmid KW Muller O Senninger N Osswald H Decreased HER-2 tyrosine kinase expression in rectal mucosa of FAP patients following low-dose sulindac chemoprevention Cancer Lett 1998 134 201 7 10025882 10.1016/S0304-3835(98)00254-7
Duke R Cohen JJ K Coligan JE, AM, Margulies DJ, Shevach EM, Strober W Morphological and biochemical assays of apoptosis Current Protocols in Immunology 1992 Wiley-Interscience: New York 3.17.1 3.17.16
Lowry OHRN Farr AL Randall RJ Protein measurement with the folin phenol reagent J Biol Chem 1951 193 265 275 14907713
Jimeno A Rubio-Viqueira B Amador ML Oppenheimer D Bouraoud N Kulesza P Sebastiani V Maitra A Hidalgo M Epidermal growth factor receptor dynamics influences response to epidermal growth factor receptor targeted agents Cancer Res 2005 65 3003 10 15833824
Messa C Russo F Caruso MG Di Leo A EGF, TGF-alpha, and EGF-R in human colorectal adenocarcinoma Acta Oncol 1998 37 285 9 9677101 10.1080/028418698429595
Salomon DS Brandt R Ciardiello F Normanno N Epidermal growth factor-related peptides and their receptors in human malignancies Crit Rev Oncol Hematol 1995 19 183 232 7612182
Kirschbaum MH Yarden Y The ErbB/HER family of receptor tyrosine kinases: A potential target for chemoprevention of epithelial neoplasms J Cell Biochem Suppl 2000 34 52 60 10762015 10.1002/(SICI)1097-4644(2000)77:34+<52::AID-JCB10>3.0.CO;2-X
O'Dwyer PJ Benson AB 3rd Epidermal growth factor receptor-targeted therapy in colorectal cancer Semin Oncol 2002 29 10 7 12422309 10.1053/sonc.2002.35643
Yasui W Sumiyoshi H Hata J Kameda T Ochiai A Ito H Tahara E Expression of epidermal growth factor receptor in human gastric and colonic carcinomas Cancer Res 1988 48 137 41 2446740
Skibber J Minsky BD Hoff P HS DeVita VT, Rosenberg SA Cancer of the colon, in Cancer: Principles and Practice of Oncology 2001 Lippincott Williams & Wilkins: Philadelphia 1216 1270
Mayer A Takimoto M Fritz E Schellander G Kofler K Ludwig H The prognostic significance of proliferating cell nuclear antigen, epidermal growth factor receptor, and mdr gene expression in colorectal cancer Cancer 1993 71 2454 60 8095852
Radinsky R Modulation of tumor cell gene expression and phenotype by the organ-specific metastatic environment Cancer Metastasis Rev 1995 14 323 38 8821093 10.1007/BF00690601
Jimeno A Hidalgo M Blockade of epidermal growth factor receptor (EGFR) activity Crit Rev Oncol Hematol 2005 53 179 92 15718144
Jost M Class R Kari C Jensen PJ Rodeck U A central role of Bcl-X(L) in the regulation of keratinocyte survival by autocrine EGFR ligands J Invest Dermatol 1999 112 443 9 10201527 10.1046/j.1523-1747.1999.00543.x
Kari C Chan TO Rocha de Quadros M Rodeck U Targeting the epidermal growth factor receptor in cancer: apoptosis takes center stage Cancer Res 2003 63 1 5 12517767
Wu X Rubin M Fan Z DeBlasio T Soos T Koff A Mendelsohn J Involvement of p27KIP1 in G1 arrest mediated by an anti-epidermal growth factor receptor monoclonal antibody Oncogene 1996 12 1397 403 8622855
Jorissen RN Walker F Pouliot N Garrett TP Ward CW Burgess AW Epidermal growth factor receptor: mechanisms of activation and signalling Exp Cell Res 2003 284 31 53 12648464 10.1016/S0014-4827(02)00098-8
|
16138927
|
PMC1208922
|
CC BY
|
2021-01-04 16:39:20
|
no
|
J Carcinog. 2005 Sep 2; 4:16
|
utf-8
|
J Carcinog
| 2,005 |
10.1186/1477-3163-4-16
|
oa_comm
|
==== Front
J CarcinogJournal of Carcinogenesis1477-3163BioMed Central London 1477-3163-4-161613892710.1186/1477-3163-4-16ResearchSulindac metabolites inhibit epidermal growth factor receptor activation and expression Pangburn Heather A [email protected] Hanna [email protected] Dennis J [email protected] Pamela L [email protected] Molecular Toxicology and Environmental Health Sciences Program, Department of Pharmaceutical Sciences, University of Colorado Health Sciences Center, Denver, USA2 Veterans Administration Medical Center, Denver, USA3 Department of Medicine, University of Colorado Health Sciences Center, Denver, USA4 University of Colorado Comprehensive Cancer Center, Denver, USA2005 2 9 2005 4 16 16 20 6 2005 2 9 2005 Copyright © 2005 Pangburn et al; licensee BioMed Central Ltd.2005Pangburn et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Regular use of nonsteroidal anti-inflammatory drugs (NSAIDs) is associated with a decreased mortality from colorectal cancer (CRC). NSAIDs induce apoptotic cell death in colon cancer cells in vitro and inhibit growth of neoplastic colonic mucosa in vivo however, the biochemical mechanisms required for these growth inhibitory effects are not well defined. We previously reported that metabolites of the NSAID sulindac downregulate extracellular-signal regulated kinase 1/2 (ERK1/2) signaling and that this effect is both necessary and sufficient for the apoptotic effects of these drugs. The goal of this project was to specifically test the hypothesis that sulindac metabolites block activation and/or expression of the epidermal growth factor (EGF) receptor (EGFR).
Methods
HT29 human colon cancer cells were treated with EGF, alone, or in the presence of sulindac sulfide or sulindac sulfone. Cells lysates were assayed by immunoblotting for phosphorylated EGFR (pEGFR, pY1068), total EGFR, phosphorylated ERK1/2 (pERK1/2), total ERK1/2, activated caspase-3, and α-tubulin.
Results
EGF treatment rapidly induced phosphorylation of both EGFR and ERK1/2 in HT29 colon cancer cells. Pretreatment with sulindac metabolites for 24 h blocked EGF-induced phosphorylation of both EGFR and ERK1/2 and decreased total EGFR protein expression. Under basal conditions, downregulation of pEGFR and total EGFR was detected as early as 12 h following sulindac sulfide treatment and persisted through at least 48 h. Sulindac sulfone induced downregulation of pEGFR and total EGFR was detected as early as 1 h and 24 h, respectively, following drug treatment, and persisted through at least 72 h. EGFR downregulation by sulindac metabolites was observed in three different CRC cell lines, occurred prior to the observed downregulation of pERK1/2 and induction of apoptosis by these drugs, and was not dependent of caspase activation.
Conclusion
These results suggest that downregulation of EGFR signaling by sulindac metabolites may occur, at least in part, by inhibiting activation and expression of EGFR. Inhibition of EGFR signaling may account for part of the growth inhibitory and chemopreventive effects of these compounds.
==== Body
Background
CRC is the second most common cause of cancer death in the USA, with an estimated annual incidence of 104,950 and mortality of 56,290 in 2005 [1]. The lifetime risk of developing CRC in the general US population is almost 6% [1]. Effective preventive measures could substantially reduce both the incidence and mortality from CRC. NSAIDs are one of the most widely studied and promising groups of compounds for CRC prevention.
NSAIDs mediate their anti-inflammatory effects by inhibiting the enzymatic activity of cyclooxygenase-1 (COX-1) and/or COX-2. Sulindac is a non-selective NSAID that inhibits both COX-1 and COX-2. Sulindac is rapidly metabolized in the liver to two major metabolites; 1) sulindac sulfide, which is an active NSAID, and 2) sulindac sulfone, which does not inhibit COX enzymatic activities and thus is not an NSAID. A large body of evidence from cell culture, animal model, human epidemiologic and clinical studies indicates that NSAIDs, including sulindac and aspirin, have potent chemopreventive and chemoregressive properties against colon cancer [2,3]. Although substantial evidence indicates that NSAIDs inhibit the growth of neoplastic colonic mucosa, the biological and biochemical mechanisms responsible for the growth inhibitory effects of these drugs are not well defined. The chemopreventive and chemoregressive effects of NSAIDs may not be due to inhibition of COX alone as we have shown that sulindac sulfide and sulindac sulfone both inhibit the phosphorylation of ERK1/2 in HCT15 cells that do not express COX-1 or COX-2 [4]. Additionally, sulindac sulfone, the non-NSAID metabolite of sulindac, has been shown, by our lab and others, to induce apoptosis of cancer cells in vitro, prevent tumor formation in animal models, and cause regression of adenomas in familial adenomatous polyposis (FAP) [2,5].
Several biological mechanisms for the chemopreventive effects of NSAIDs have been proposed including inhibition of proliferation, induction of apoptosis, and inhibition of angiogenesis. Our laboratory [5,6] and others [2] have reported that NSAIDs inhibit growth of CRC cells in culture primarily by the induction of apoptotic cell death. An apoptotic mechanism was also suggested in human adenomas treated with sulindac sulfone [7]. We have reported that the apoptotic effect of sulindac is not dependent on COX inhibition [4,5,8], but does appear to be dependent on the downregulation of ERK1/2 [8].
Work from several laboratories has demonstrated substantial interactions between the biochemical effects of NSAIDs and EGFR signaling. It is well established that COX-2 protein expression is stimulated by EGF and decreased by EGFR inhibitors [9]. Sulindac sulfide and indomethacin inhibit TGFα induced prostaglandin production and thymidine incorporation in RIE-1 cells [10], and indomethacin, ibuprofen, and aspirin all block EGF-induced Ca++ influx in Caco-2 colon cancer cells [11]. Furthermore, combination therapy of NSAIDs and EGFR antagonists display an additive effect against colon tumor development in vivo [12] and two previous studies suggest that NSAIDs might inhibit EGFR signaling [13,14]. Finally, sulindac has been shown to inhibit expression of ErbB2/HER2 protein expression in rectal mucosa of FAP patients [15]. The biochemical mechanisms by which NSAIDs might modulate EGFR signaling has not been addressed in any of these studies.
We have previously shown that sulindac sulfide and sulindac sulfone inhibit ERK1/2 phosphorylation both basally and in response to EGF [6], and that this inhibition is sufficient [5] and necessary [8] for the apoptotic effect of these drugs. The goal of this study was to specifically test the hypothesis that sulindac metabolites block activation and/or expression of EGFR (Figure 1). We report, for the first time, that sulindac metabolites block both basal and EGF induced phosphorylation of EGFR and downregulate total EGFR protein expression. Together these results identify a new target for the actions of sulindac and suggest that inhibition of EGFR activation may be a major biochemical mechanism for both the downregulation of ERK1/2 signaling and the chemopreventive activity of NSAIDs.
Figure 1 The EGFR/MAPK/ERK1/2 signaling pathway. We have previously shown that sulindac metabolites inhibit both MEK1/2 and ERK1/2 activation. The goal of this study was to determine if this inhibition is due to downregulation of the EGFR.
Methods
Reagents
Cell culture media and fetal bovine serum (FBS) were purchased from Mediatech (Herndon, VA), antibiotic/antimycotic solution (penicillin/streptomycin/fungizone) from Life Technologies, Inc. (Grand Island, NY), and tissue culture plates from Falcon (Franklin Lakes, NJ). Primary antibodies raised against phosphorylated ERK1/2 (Thr202/Tyr204), total ERK1/2, and total EGFR were purchased from Santa Cruz Biotechnology (Santa Cruz, CA); primary antibody against phosphorylated EGFR(Tyr1068), horseradish peroxidase-conjugated anti-mouse and anti-rabbit secondary antibodies, and EGF were purchased from Biosource (Camarillo, CA). Primary antibody against caspase-3 (recognizes both full length caspase-3 and cleavage products) and cleaved caspase-3 (recognizes only cleavage products of activated caspase-3 resulting from cleavage adjacent to Asp175) were obtained from Cell Signaling Technology (Beverly, MA). Immobilon-P membranes were purchased from Millipore (Bedford, MA), chemiluminescent visualization reagents from NEN (Boston, MA), and X-ray film from Pierce (Rockford, IL). Sulindac sulfide was from LKT Laboratories Inc. (St. Paul, MN). ZVAD-fmk (caspase inhibitor I) and primary antibody against α-tubulin were purchased from Calbiochem (San Diego, CA). Sulindac sulfone was a generous gift from Cell Pathways Inc. (Horsham, PA).
Cell culture
HT29, Caco-2, and HCT116 human colon cancer cells were purchased from American Type Culture Collection (Manassas, VA). HT29 and HCT116 cells were maintained in RPMI 1640 media supplemented with 10% FBS and 1% penicillin/streptomycin/fungizone solution; Caco-2 cells were maintained in DMEM media with 4 mM L-glutamine adjusted to contain 1.5 g/L glucose and supplemented with 0.01 mg/ml human transferring, 10% FBS, and 1% penicillin/streptomycin/fungizone solution. Cells were plated and grown on tissue culture plates as specified in the results section.
Morphologic quantitation of apoptotic cell death
An ethidium bromide/acridine orange double-dye morphological assay [16] was used to quantitate apoptotic cell death induced by sulindac metabolites. Cells were grown in multi-well plates and given the appropriate drug treatment. Three days after treatment, cells were harvested and stained with a mixture of 100 μg/ml ethidium bromide and 100 μg/ml acridine orange and then examined by fluorescence microscopy. Three fields of 100 cells each were counted and the number of apoptotic cells (as determined by nuclear morphology) as well as the number of living and necrotic cells were expressed as a percentage of total cells counted.
Cell lysates and Western immunoblotting
At time of harvest, cells were scraped from the plates, washed with ice-cold phosphate buffered saline (PBS), and pelleted at 2,400 × g for 5 min. Care was taken to keep cells as cold as possible during the entire procedure. After aspirating the supernatant, the cells were washed twice with ice cold PBS and centrifuged. The cell pellet was lysed in cell extraction buffer (10 mM Tris, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM NaF, 20 mM Na4P2O7, 2 mM Na3VO4, 1% Triton X-100, 10% glycerol, 0.1% SDS, 0.5% deoxycholate, 1 mM PMSF, 60 μg/mL aprotinin, 10 μg/mL leupeptin, 1 μg/mL pepstatin) for 30 min, on ice, with vortexing at 10 min intervals. Lysates were then centrifuged at 18,000 × g for 10 min at 4°C, and supernatants collected. Protein concentrations of lysates were determined by the method of Lowry [17].
Lysate samples (50 μg total protein) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and electrotransferred overnight onto Immobilon-P polyvinylidene difluoride membranes. Nonspecific binding was blocked by immersing membranes in Tris-neutral saline with 1% (w/v) dry milk and 0.05% Tween 20 for 30 min at room temperature, then blots were incubated with pEGFR, pERK1/2, caspase 3, cleaved caspase-3, or α-tubulin primary antibody while rocking at 37°C for 1 h. Immunoreactive protein was detected by incubating blots with horseradish peroxidase-conjugated secondary antibody for 1 h followed by chemiluminescent substrate for 1 min. Membranes probed for phospho-proteins were then stripped and reprobed with total EGFR or total ERK1/2 primary antibody followed by horseradish peroxidase-conjugated secondary antibody as described above. Non-stripped control blots were run to show that this stripping procedure gave the same results as primary blots with each antibody. Immunoreactive proteins were visualized by exposure to radiographic film. Quantitation of protein levels was determined by densitometry using a visual imaging system (UN-SCAN-IT gel, Silk Scientific, Inc., Orem, UT).
Statistical Analysis
Data was analyzed by one way ANOVA using a Newman-Keuls multiple comparison test and statistical significance accepted at p < 0.05 (GraphPad Prism 4.0, GraphPad Software Inc., San Diego, CA).
Results
EGF induces phosphorylation of EGFR and ERK1/2 in HT29 human colon cancer cells
In order to verify that EGFR signaling was intact in our cells, the effects of EGF treatment on the expression and activation of EGFR and ERK1/2 in HT29 human colon cancer cells was examined. It was observed that EGF rapidly induced a > 7 fold increase of pEGFR that was maximal at 10 min (Figure 2A and 2B). Similarly, EGF induced a rapid, > 30 fold, increase of pERK1/2 that was also maximal at 10 min (Figure 2A and 2C). EGF-induced phosphorylation of EGFR and ERK1/2 persisted through 60 min, and was no longer detected at 24 h (data not shown). Similar results were obtained in HCT116 and Caco-2 human colon cancer cell lines (data not shown).
Figure 2 EGF induces EGFR and ERK1/2 phosphorylation. HT29 human colon cancer cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 48 h before treatment with vehicle (water), 10, or 100 ng/ml EGF. Cells were harvested 10 min after EGF treatment and lysates prepared for (A) Immunoblotting with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, and total ERK1/2; α-tubulin immunoblots of the same lysates served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and pERK1/2 bands (C) normalized for the loading controls. Data represent mean of triplicate samples ± SD; statistical significance is denoted by ***p < 0.001 versus vehicle. Results shown in the figure are representative of 2 separate experiments each with triplicate samples.
Sulindac metabolites block EGF-induced phosphorylation of EGFR and ERK1/2, downregulate total EGFR protein, and induce apoptotic cell death
We next determined the effect of sulindac sulfide on EGF-induced phosphorylation and expression of EGFR and ERK1/2. Pretreatment of HT29 cells for 24 h with 40–160 μM sulindac sulfide blocked the ability of EGF to induce phosphorylation of EGFR in a dose dependent fashion. Total EGFR expression and EGF-induced phosphorylation of ERK1/2 proteins was also downregulated following pretreatment with sulindac sulfide (Figure 3). There was no effect of drug or EGF treatment on total ERK1/2 expression.
Figure 3 Dose response of sulindac sulfide inhibition of EGFR. HT29 cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 24 h before addition of drug. Cells were then treated with vehicle (0.1% DMSO), 40, 80, 120, or 160 μM sulindac sulfide, drug doses previously shown to induce apoptotic cell death in these cells. Twenty four hours after drug treatment, vehicle or 10 ng/ml EGF was added and cells were harvested 10 min later. (A) Immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and caspase-3; total ERK1/2 immunoblots served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Results shown in figure are representative of 3 separate experiments.
The effect of sulindac sulfone on EGF-induced phosphorylation of EGFR and ERK1/2 was also examined. As seen with sulindac sulfide, pretreatment of HT29 cells for 24 h with 200–800 μM sulindac sulfone markedly decreased the ability of EGF to induce phosphorylation of EGFR; total EGFR expression levels and ERK1/2 phosphorylation were also downregulated (Figure 4), albeit to a lesser extent than with sulindac sulfide.
Figure 4 Dose response of sulindac sulfone inhibition of EGFR. HT29 cells were grown to 80% confluence in medium containing 10% FBS then serum deprived for 24 h before addition of drug. Cells were then treated with vehicle (0.2% DMSO), 200, 400, 600, or 800 μM sulindac sulfone, drug doses previously shown to induce apoptotic cell death in these cells. Twenty four hours after drug treatment, vehicle or 10 ng/ml EGF was added and cells were harvested 10 min later. (A) Immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and caspase-3; total ERK1/2 immunoblots served as loading controls. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Results shown in figure are representative of 3 separate experiments.
The finding that total EGFR protein expression was decreased by sulindac metabolites (Figures 3 and 4) was unexpected; we therefore confirmed these results using a second antibody (Biosource catalog #AHR5062). The confirmatory EGFR antibody is an EGFR antibody cocktail (composed of four monoclonal antibodies) directed against both the extracellular and cytoplasmic domains of EGFR. The same downregulation of total EGFR elicited by sulindac sulfide and sulfone was found using this antibody (data not shown), confirming that these results are not peculiar to a single EGFR epitope or a result of an artifact of the use of a single antibody.
EGF had no effect on morphological apoptosis or activation of caspase-3 but treatment with 40–160 μM sulindac sulfide (Figure 3) or 600–800 μM sulindac sulfone (Figure 4) induced substantial apoptotic cell death, as determined by cleavage of caspase-3 and confirmed with the acridine orange and ethidium bromide morphologic assay (data not shown).
Sulindac metabolites downregulate basal EGFR phosphorylation and expression
Next we examined the time course of sulindac's effects on EGFR expression and phosphorylation in cells grown in 10% FBS (basal conditions). Phosphorylation of EGFR and total EGFR protein expression were downregulated as early as 12 h and persisted at 24 h following sulindac sulfide treatment (Figure 5). Significant downregulation of pERK1/2 was not detected prior to 24 h in these experiments. Evidence of apoptosis, by presence of caspase-3 cleavage products (Figure 5A) was also first detected at 24 h; examination of nuclear morphology after staining with acridine orange and ethidium bromide at 48 h confirmed cleaved caspase-3 immunoblot results (data not shown). Sulindac sulfide treatment did not consistently affect levels of total ERK1/2 or α-tubulin proteins. For this experiment immunoblots for each time point were run on individual gels thus, only samples within a single time point, assayed for one particular protein, can be directly compared.
Figure 5 Dose response and time course of sulindac sulfide inhibition of EGFR. HT29 cells were grown to confluence in medium containing 10% FBS and treated with vehicle (0.1% DMSO), 160, or 180 μM sulindac sulfide for 1 h, 12 h, and 24 h. Cells were then harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and cleaved caspase-3; α-tubulin immunoblots of the same lysates served as loading controls. (A) 1 h, 12 h, and 24 h immunoblot results. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Data represent mean of triplicate samples ± SD; statistical significance is denoted by **p < 0.01 and ***p < 0.001 versus respective time point vehicle. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
The effect of sulindac sulfone on basal EGFR activation and expression was also examined. Sulindac sulfone treatment resulted in a modest but significant decrease in EGFR phosphorylation as early as 1 h after treatment, and this effect increased and persisted through 24 h (Figure 6A and 6B). Downregulation of total EGFR (Figure 6A and 6C) and pERK1/2 (Figure 6A) proteins was first seen at 24 h; caspase-3 cleavage (Figure 6A) was also first detected at 24 h. Morphologic evidence of apoptosis was apparent at 48 h, the earliest time point examined in this manner. Sulfone treatment did not consistently affect levels of total ERK1/2 or α-tubulin proteins.
Figure 6 Dose response and time course of sulindac sulfone inhibition of EGFR. HT29 cells were grown to confluence in medium containing 10% FBS followed by treatment with vehicle (0.2% DMSO), 400, or 600 μM sulindac sulfone for 1 h, 12 h, and 24 h. Cells were then harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, pERK1/2, total ERK1/2, and cleaved caspase-3; α-tubulin immunoblots of the same lysates served as loading controls. (A) 1 h, 12 h, and 24 h Western blot results. The graphs show the densitometry results of the pEGFR bands (B) and total EGFR bands (C). Data represent mean of triplicate samples ± SD; statistical significance is denoted by **p < 0.01 and ***p < 0.001 versus respective time point vehicle. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
To exclude the possibility that downregulation of EGFR by sulindac metabolites was unique to HT29 cells, similar experiments were carried out in Caco-2 and HCT116 cells. Sulindac sulfide and sulfone downregulated expression of both pEGFR and total EGFR in these cell lines as well (data not shown).
ZVAD blocks sulindac-induced apoptosis but not EGFR downregulation
Although the downregulation of EGFR observed with sulindac metabolites occurred prior to either biochemical or morphologic evidence of apoptosis, it remains possible that some of the decrease in EGFR protein expression could be a consequence of caspase activation by sulindac metabolites. To examine this possibility, HT29 cells were pretreated with vehicle or 25 μM of the broad specificity caspase inhibitor ZVAD-fmk (ZVAD) for 60 minutes prior to treatment with 600 μM sulindac sulfone or vehicle for 48 h. ZVAD pretreatment completely blocked sulindac sulfone induced apoptosis as measured by examination of nuclear morphology (Figure 7A) and it prevented cleavage of caspase 3 to the active 17 and 19 kDa fragments (Figure 7B; the 20 kDa caspase 3 fragment seen with the combination of sulfone and ZVAD has been shown to be an inactive cleavage product [26-29]. While 25 uM ZVAD prevented both caspase activation and morphologic apoptosis induced by sulindac sulfone, it did not prevent sulindac sulfone induced decreases in either total or pEGFR (Figure 7B–D). Combination experiments with ZVAD and sulindac sulfide showed similar results as those shown in Figure 7 for sulindac sulfone. These results demonstrate that downregulation of EGFR by sulindac metabolites is not a consequence of the apoptotic effect of these drugs.
Figure 7 Effect of the caspase inhibitor, ZVAD, on apoptosis and inhibition of EGFR. HT29 colon cancer cells were grown to confluence in medium containing 10% FBS followed by pretreatment with or without 25 μM zvad for 1 h. Cells were then treated with vehicle (0.2% DMSO) or 600 μM sulfone for 48 h. Cells were harvested and immunoblots were performed on cell lysates with antibodies raised against pEGFR (pY1068), total EGFR, and cleaved caspase 3; α-tubulin immunoblots of the same lysates served as loading controls. The graphs show morphological apoptosis results (A) 48 h Western immunoblot results (B) and densitometry of the pEGFR bands (C) and total EGFR bands (D). Data represent mean of triplicate samples ± SD; statistical significance is denoted by *p < 0.05, **p < 0.01 and ***p < 0.001. Results shown in figure are representative of 2 separate experiments each with triplicate samples.
Discussion
Our laboratory originally reported that sulindac metabolites inhibit the EGFR/mitogen activated protein kinase (MAPK)/ERK1/2 signaling pathway at the level of MEK or higher [5] (Figure 1) and that the downregulation of mitogen-activated protein kinase kinase (MEK)/ERK1/2 is both necessary and sufficient for the apoptotic effect of these drugs [8]. We now report, for the first time, that sulindac metabolites inhibit EGFR phosphorylation and downregulate the expression of total EGFR protein in human colon cancer cells. Downregulation of EGFR occurred at doses and within a time course that is consistent with the possibility that the effect of drug on the EGFR could be the mechanism of the inhibition of MEK1/2 and ERK1/2 activity and apoptotic cell death induced by these drugs.
Both sulindac sulfide and sulindac sulfone downregulate EGFR expression and phosphorylation at comparable doses to those that inhibit ERK1/2 phosphorylation and induce apoptotic cell death (Figures 5 and 6). The time course experiments suggest that the kinetics of the EGFR effect may differ between the two sulindac metabolites. With sulindac sulfide, the effect on EGFR and ERK1/2 occurred concurrently (first seen at 12 h) and before any detectable effect on apoptosis. This suggests that the downregulation of EGFR may be responsible for some or all of the ERK1/2 inhibitory effects of the drug. The relationship between sulindac sulfone's effect on EGFR and ERK1/2 is not as clear since an effect on pEGFR was detected substantially earlier (1 hour) than the effect on pERK1/2 and apoptosis (24 hours). These differences in the kinetics of the two sulindac metabolites on pEGFR and pERK1/2 downregulation could represent differences in potency, duration of effect, the spectrum of EGFR phosphorylation sites that are affected by the different drugs, differential effects on phosphatases, and/or differences in the molecular mechanism of action. Similar to our results, Jimeno et al. reported that treatment of HuCCT cells with the EGFR tyrosine kinase inhibitor (TKI) erlotinib or the mononclonal antibody (MAb) cetuximab led to significant downregulation of pEGFR as early as 1 h following treatment whereas major downstream effectors (pERK1/2 and pAKT) were not coordinately affected [18]. Despite these differences in kinetics, the finding that EGFR downregulation was observed consistently with both sulindac metabolites and in three different CRC cell lines (HT29, HCT116, and Caco-2) establishes the EGFR as a new target of sulindac.
Activation of the EGFR pathway is particularly common in colorectal cancers [19-22] and overexpression of EGFR correlates with more aggressive disease and poor prognosis [23-26]. Both EGFR MAbs and TKIs that block activation of EGFR are now being used for the treatment of colorectal cancer [27]. In light of our evidence that NSAIDs downregulate EGFR signaling and induce apoptosis, it is interesting to note that both EGFR MAbs and TKIs have been shown to stimulate important pro-apoptotic genes and downregulate anti-apoptotic genes, particularly those of the Bcl-2 family [28-30]. Furthermore, treatment with combinations of agents that inhibit EGFR by different mechanisms is more effective than treatment with a single agent. For example, the combination of erlotinib plus cetuximab caused a synergistic induction of apoptosis in vitro, and an additive effect on tumor growth arrest in vivo [18]. Thus, downregulation of EGFR could mediate the apoptotic and chemopreventive effects of NSAIDs and could explain some of the clinical evidence of additive growth inhibitory effects of EGFR inhibitors and NSAIDs. These data suggest a biochemical rationale for the use of combinations of classic EGFR inhibitors and NSAIDs in the treatment or prevention of CRC.
If inhibition of EGFR signaling is a mechanism of NSAIDs as our data indicates, it will be of great interest to examine the full spectrum of possible consequences of downregulation of EGFR. Thus far, we have only examined the ras-raf-MEK1/2-ERK1/2 arm of EGFR signaling but several other signaling pathways are known to be activated by the EGFR including the phosphatidylinositol-3-kinase (PI3-K) and Janus kinase (JAK)-signal transducers and activators of transcription (STAT) pathways [31]. Sulindac metabolites may also affect other members of the ErbB family of receptor tyrosine kinases (RTK) including Erb2, ErbB3, ErbB4, non-ErbB RTKs including PDGFR and VEGFR, and/or receptors lacking tyrosine kinase activity such as the adenosine receptor. Studies to examine the effect of NSAIDs on these pathways and receptors are underway.
We have not systematically examined the mechanism by which sulindac metabolites downregulate EGFR but the relatively short time course (as early as 1 hour) suggests that the effect is more likely due to increased degradation of EGFR rather than inhibition of new protein synthesis. Our data demonstrate that EGFR inhibition is not a consequence of caspase activation or due to the apoptotic effects of the drugs in that apoptosis was first detected 24 h after sulindac metabolite treatment, whereas downregulation of EGFR was seen as early as 1 to 12 hours after drug treatment. In addition, inhibition of apoptosis with the broad specificity caspase inhibitor ZVAD did not prevent EGFR downregulation by sulindac metabolites. Thus, we hypothesize that EGFR downregulation by sulindac metabolites is not due to activation of caspases, but rather is a result of increased proteasomal and/or lysosomal degradation. Further studies are warranted to determine the precise mechanism of EGFR downregulation by sulindac metabolites.
Sulindac metabolites decreased expression of both total and pEGFR proteins. The time course experiments, particularly those with sulindac sulfone (Figure 6), suggest that downregulation of these two forms of the receptor could be occurring by distinct mechanisms. Although total and pEGFR downregulation with sulindac sulfide occurred concurrently, the sulfone metabolite decreased pEGFR as early as 1 h whereas downregulation of total EGFR protein was first observed at the 24 h time point. This sequence indicates that the downregulation of pEGFR is not likely to be a consequence of the downregulation of the total receptor, at least with sulindac sulfone.
Conclusion
In summary, we have described for the first time that sulindac metabolites downregulate EGFR and prevent EGFR phosphorylation by EGF. This effect may explain, at least in part, our previous observation that these drugs inhibit basal and EGF-induced activation of the MEK/ERK arm of the EGFR signaling pathway. Downregulation of EGFR was shown to be independent of COX inhibitory activity as it occurred with sulindac sulfone. These results add to the increasing amount of data that suggest interactions exist between the effects of NSAIDs and EGFR signaling [13,14]. Our results are the first to suggest that these interactions are due to a direct drug effect on the EGFR itself.
Abbreviations
NSAID, non-steroidal anti-inflammatory drug; CRC, colorectal cancer; ERK1/2, extracellular-signal regulated kinase; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; pEGFR, phosphorylated epidermal growth factor receptor; pERK1/2, phosphorylated extracellular-signal regulated kinase 1/2; COX, cyclooxygenase; FAP, familial adenomatous polyposis; FBS, fetal bovine serum; PBS, phosphate buffered saline; MAPK, mitogen-activated protein kinase; MEK1/2, mitogen-activated protein kinase kinase; MAb, monoclonal antibody; TKI, tyrosine kinase inhibitor; PI3-K, phosphatidylinositol-3-kinase; JAK/STAT, Janus kinase/signal transducers and activators of transcription; RTK, receptor tyrosine kinase
Acknowledgements
Funding provided by Department of Veterans Affairs Merit Review Program and Cancer Research and Prevention Foundation
==== Refs
Jemal A Murray T Ward E Samuels A Tiwari RC Ghafoor A Feuer EJ Thun MJ Cancer statistics, 2005 CA Cancer J Clin 2005 55 10 30 15661684
Gupta RA Dubois RN Colorectal cancer prevention and treatment by inhibition of cyclooxygenase-2 Nat Rev Cancer 2001 1 11 21 11900248 10.1038/35094017
Thun MJ Henley SJ Patrono C Nonsteroidal anti-inflammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues J Natl Cancer Inst 2002 94 252 66 11854387
Rice PL Kelloff J Sullivan H Driggers LJ Beard KS Kuwada S Piazza G Ahnen DJ Sulindac metabolites induce caspase- and proteasome-dependent degradation of beta-catenin protein in human colon cancer cells Mol Cancer Ther 2003 2 885 92 14555707
Rice PL Goldberg RJ Ray EC Driggers LJ Ahnen DJ Inhibition of extracellular signal-regulated kinase 1/2 phosphorylation and induction of apoptosis by sulindac metabolites Cancer Res 2001 61 1541 7 11245463
Rice PL Washington M Schleman S Beard KS Driggers LJ Ahnen DJ Sulindac sulfide inhibits epidermal growth factor-induced phosphorylation of extracellular-regulated kinase 1/2 and Bad in human colon cancer cells Cancer Res 2003 63 616 20 12566304
Piazza GA Alberts DS Hixson LJ Paranka NS Li H Finn T Bogert C Guillen JM Brendel K Gross PH Sperl G Ritchie J Burt RW Ellsworth L Ahnen DJ Pamukcu R Sulindac sulfone inhibits azoxymethane-induced colon carcinogenesis in rats without reducing prostaglandin levels Cancer Res 1997 57 2909 15 9230200
Rice PL Beard KS Driggers LJ Ahnen DJ Inhibition of extracellular-signal regulated kinases 1/2 is required for apoptosis of human colon cancer cells in vitro by sulindac metabolites Cancer Res 2004 64 8148 51 15548677
Coffey RJ Hawkey CJ Damstrup L Graves-Deal R Daniel VC Dempsey PJ Chinery R Kirkland SC DuBois RN Jetton TL Morrow JD Epidermal growth factor receptor activation induces nuclear targeting of cyclooxygenase-2, basolateral release of prostaglandins, and mitogenesis in polarizing colon cancer cells Proc Natl Acad Sci U S A 1997 94 657 62 9012840 10.1073/pnas.94.2.657
DuBois RN Awad J Morrow J Roberts LJ 2ndBishop PR Regulation of eicosanoid production and mitogenesis in rat intestinal epithelial cells by transforming growth factor-alpha and phorbol ester J Clin Invest 1994 93 493 8 8113389
Kokoska ER Smith GS Miller TA Nonsteroidal anti-inflammatory drugs attenuate proliferation of colonic carcinoma cells by blocking epidermal growth factor-induced Ca++ mobilization J Gastrointest Surg 2000 4 150 61 10675238 10.1016/S1091-255X(00)80051-7
Torrance CJ Jackson PE Montgomery E Kinzler KW Vogelstein B Wissner A Nunes M Frost P Discafani CM Combinatorial chemoprevention of intestinal neoplasia Nature Med 2000 6 1024 1028 10973323 10.1038/79534
Kokoska ER Smith GS Wolff AB Deshpande Y Miller TA Nonsteroidal anti-inflammatory drugs attenuate epidermal growth factor-induced proliferation independent of prostaglandin synthesis inhibition J Surg Res 1999 84 186 92 10357918 10.1006/jsre.1999.5640
Kinoshita T Takahashi Y Sakashita T Inoue H Tanabe T Yoshimoto T Growth stimulation and induction of epidermal growth factor receptor by overexpression of cyclooxygenases 1 and 2 in human colon carcinoma cells Biochim Biophys Acta 1999 1438 120 30 10216286
Winde G Lugering N Glodny B Schmid KW Muller O Senninger N Osswald H Decreased HER-2 tyrosine kinase expression in rectal mucosa of FAP patients following low-dose sulindac chemoprevention Cancer Lett 1998 134 201 7 10025882 10.1016/S0304-3835(98)00254-7
Duke R Cohen JJ K Coligan JE, AM, Margulies DJ, Shevach EM, Strober W Morphological and biochemical assays of apoptosis Current Protocols in Immunology 1992 Wiley-Interscience: New York 3.17.1 3.17.16
Lowry OHRN Farr AL Randall RJ Protein measurement with the folin phenol reagent J Biol Chem 1951 193 265 275 14907713
Jimeno A Rubio-Viqueira B Amador ML Oppenheimer D Bouraoud N Kulesza P Sebastiani V Maitra A Hidalgo M Epidermal growth factor receptor dynamics influences response to epidermal growth factor receptor targeted agents Cancer Res 2005 65 3003 10 15833824
Messa C Russo F Caruso MG Di Leo A EGF, TGF-alpha, and EGF-R in human colorectal adenocarcinoma Acta Oncol 1998 37 285 9 9677101 10.1080/028418698429595
Salomon DS Brandt R Ciardiello F Normanno N Epidermal growth factor-related peptides and their receptors in human malignancies Crit Rev Oncol Hematol 1995 19 183 232 7612182
Kirschbaum MH Yarden Y The ErbB/HER family of receptor tyrosine kinases: A potential target for chemoprevention of epithelial neoplasms J Cell Biochem Suppl 2000 34 52 60 10762015 10.1002/(SICI)1097-4644(2000)77:34+<52::AID-JCB10>3.0.CO;2-X
O'Dwyer PJ Benson AB 3rd Epidermal growth factor receptor-targeted therapy in colorectal cancer Semin Oncol 2002 29 10 7 12422309 10.1053/sonc.2002.35643
Yasui W Sumiyoshi H Hata J Kameda T Ochiai A Ito H Tahara E Expression of epidermal growth factor receptor in human gastric and colonic carcinomas Cancer Res 1988 48 137 41 2446740
Skibber J Minsky BD Hoff P HS DeVita VT, Rosenberg SA Cancer of the colon, in Cancer: Principles and Practice of Oncology 2001 Lippincott Williams & Wilkins: Philadelphia 1216 1270
Mayer A Takimoto M Fritz E Schellander G Kofler K Ludwig H The prognostic significance of proliferating cell nuclear antigen, epidermal growth factor receptor, and mdr gene expression in colorectal cancer Cancer 1993 71 2454 60 8095852
Radinsky R Modulation of tumor cell gene expression and phenotype by the organ-specific metastatic environment Cancer Metastasis Rev 1995 14 323 38 8821093 10.1007/BF00690601
Jimeno A Hidalgo M Blockade of epidermal growth factor receptor (EGFR) activity Crit Rev Oncol Hematol 2005 53 179 92 15718144
Jost M Class R Kari C Jensen PJ Rodeck U A central role of Bcl-X(L) in the regulation of keratinocyte survival by autocrine EGFR ligands J Invest Dermatol 1999 112 443 9 10201527 10.1046/j.1523-1747.1999.00543.x
Kari C Chan TO Rocha de Quadros M Rodeck U Targeting the epidermal growth factor receptor in cancer: apoptosis takes center stage Cancer Res 2003 63 1 5 12517767
Wu X Rubin M Fan Z DeBlasio T Soos T Koff A Mendelsohn J Involvement of p27KIP1 in G1 arrest mediated by an anti-epidermal growth factor receptor monoclonal antibody Oncogene 1996 12 1397 403 8622855
Jorissen RN Walker F Pouliot N Garrett TP Ward CW Burgess AW Epidermal growth factor receptor: mechanisms of activation and signalling Exp Cell Res 2003 284 31 53 12648464 10.1016/S0014-4827(02)00098-8
|
16095540
|
PMC1208923
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Aug 11; 3:19
|
latin-1
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-19
|
oa_comm
|
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-211610720910.1186/1476-7120-3-21ResearchDoes contrast echocardiography induce increases in markers of myocardial necrosis, inflammation and oxidative stress suggesting myocardial injury? Knebel Fabian [email protected] Ingolf [email protected] Stephan [email protected] Torsten [email protected] Reinhard [email protected] Sebastian [email protected] Gert [email protected] Adrian Constantin [email protected] Universitätsmedizin Berlin, Medical Clinic for Cardiology, Angiology, Pulmology, Charité Campus Mitte, Germany2 Paritätisches Krankenhaus Lichtenberg, Fanningerstraße 32, 10365 Berlin, Lichtenberg, Germany3 Universitätsmedizin Berlin, Institute for Laboratory Medicine and Pathobiochemistry, Charité Campus Mitte, Germany2005 17 8 2005 3 21 21 14 6 2005 17 8 2005 Copyright © 2005 Fabian et al; licensee BioMed Central Ltd.2005Fabian et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Contrast echocardiography is a precise tool for the non-invasive assessment of myocardial function and perfusion. Side effects of contrast echocardiography resulting from contrast-agent induced myocardial micro-lesions have been found in animals. The goal of this study is to measure markers of myocardial necrosis, inflammation and oxidative stress in humans to evaluate potential side-effects of contrast echocardiography.
Methods
20 patients who underwent contrast echocardiography with Optison as the contrast medium were investigated. To evaluate myocardial micro-necrosis, inflammation and oxidative stress, cardiac troponin I (cTnI), tumor necrosis factor-α (TNF-α), interleukin (IL)-6, -8 and thiobarbituric acid reactive substances (TBARS) were measured at baseline and at 2, 4, 8 and 24 hours after contrast echocardiography.
Results
At baseline, 50% of the patients had cTnI and TBARS values outside the reference range. TNF-α, IL-6, IL-8 levels were within the reference range. Patients with cTnI above the RR clustered to significantly higher levels of TNF-α and IL-6. After contrast echocardiography, no statistically significant increase of cTnI, cytokines and TBARS was found. However, for nearly 50% of the patients, the intra-individual cTnI kinetics crossed the critical difference (threefold of methodical variation) which indicates a marker increase. This was neither predicted by the baseline levels of the cytokines nor the markers of oxidative stress.
Conclusion
There are no clinically relevant increases in serum markers for micro-necrosis, inflammation and oxidative stress in humans after contrast echocardiography. Future studies have to address whether cTnI increase in some patients represent a subset with increased risk for side effects after contrast echocardiography.
==== Body
Background
Contrast echocardiography is a precise and study-proven tool for the non-invasive assessment of left ventricular systolic function, myocardial perfusion, endocardial border detection and coronary flow reserve [1,2].
Myocardial trauma, e.g. occlusion of a coronary artery during angioplasty, can potentially stimulate inflammatory processes with subsequent activation and infiltration of inflammatory cells in the myocardium. This can induce the release of cytokines and reactive oxygen species [3]. Microbubbles in their use as ultrasound contrast agents, which are destroyed by ultrasound, potentially also pose a risk for myocardial micro-lesions. This was demonstrated by in vitro and in vivo studies [4-7]. Therefore, in addition to markers of myocardial necrosis (cTnI), the measurement of markers for inflammation (TNF-α, IL-6, IL-8) and oxidative stress (TBARS) in peripheral blood is an appropriate tool to evaluate myocardial injury in humans undergoing contrast echocardiography.
These markers are not specific for myocardial damage but the degree of oxidative stress is correlated to the extent and progression of myocardial damage. [8-11]. It has been demonstrated in previous studies that levels of thiobarbituric acid reactive substances (TBARS) are elevated in patients with cardiovascular disease [12]. These patients might have a higher individual risk to develop side-effects from contrast myocardial echocardiography.
In a former study, we analyzed markers for myocardial necrosis (myoglobin, cardiac troponin I, creatine kinase MB activity and mass) in patients undergoing contrast echocardiography. None of the markers crossed the cut-offs generally accepted to indicate myocardial necrosis [13]. Possibly, one reason to these negative findings is that the cut-off values deducted from ROC-curves which were used at that time were too high for the detection of micro-lesions after contrast echocardiography. Based on strongly improved assay performance, it is now generally agreed that the elevation of cTnI without crossing the former cut-off is indicative of myocardial micro-necrosis [14]. Consequently, myocardial infarction without clinical symptoms – also termed "minor myocardial infarction" – is now re-defined as a troponin increase above the 99% percentile of the reference population or a troponin increase of more than the level that can be measured with an imprecision of = 10%, respectively [15-17].
Furthermore, there is increasing evidence that a low level troponin increases are pathologically relevant. Specifically, cTnI values below the ROC cut-off are clearly associated with mild and subclinical myocardial damage [18].
In this study, we performed additional measurements on the serum samples of the former study. cTnI concentrations were measured at baseline and at 2, 4, 8 and 24 hours after contrast echocardiography. Furthermore, we measured the markers of inflammation and oxidative stress (TNF-α, IL-6, IL-8 and TBARS).
Materials and methods
20 consecutive patients with a clinical indication for contrast echocardiography were included in this study. The indications for contrast echocardiography are listed in table 1. The patients were clinically stable, older than 18 years of age and had signed an informed consent. Exclusion criteria included acute coronary syndrome, hemodynamic instability and cardiogenic shock. The study protocol was approved by the ethics committee of the Charité University Hospital. The venous blood sampling was performed immediately before (baseline) and 2, 4, 8 and 24 hours after contrast echocardiography. Due to limited availability of the plasma samples, in only 16 of the 20 patients all cytokines (TNF-α, IL-6, IL-8) could be measured.
Table 1 Baseline characteristics of the patients and indication for contrast echocardiography
Diagnosis n Contrast imaging indication
Ischemic n = 6
Coronary heart disease 6 Endocardial border detection / myocardial perfusion
Non-ischemic n = 14
Dilated cardiomyopathy 6 Endocardial border detection
Aortic valve replacement 2 Doppler flow enhancement / Endocardial border detection
Myocarditis 1 Endocardial border detection
Chronic obstructive lung disease 1 Endocardial border detection
Hypertensive heart disease 2 Endocardial border detection
Hypertrophic obstructive cardiomyopathy 2 Endocardial border detection / Doppler flow enhancement
Total 20
The critical difference indicating a marker increase after echocardiography was calculated based on the data for the within-run variation. There is general agreement that two values measured within-run are significantly different (p < 0.05) if the difference between the values is greater than the three-fold of the within-run variation of the method.
Contrast echocardiography was performed with the Vivid Five System (GE Vingmed Ultrasound, Horten, Norway). Optison (perfluoro-propane-filled albumin microspheres, Mallinckrodt, St. Louis, Mo, USA) was used as the contrast medium. 20 ml of contrast microbubble solution (3 ml Optison and 17 ml 0.9% saline solution) were infused manually over three minutes into a peripheral vein followed by a saline flush. Contrast echo was performed 5 minutes after the beginning of the application of the contrast solution.
Cardiac troponin I was measured with the heterogeneous immunoassay module cardiac troponin-I flex™ reagent cartridge for Dimension (Date Behring, Marburg, Germany). For this assay, the lowest limit of detection (LoD) is 0.04 μg/l. The 99%-percentile of the reference group is 0.07 μg/l. In the range below the ROC cut off (0.6 μg/l), the analytical variation is 10% at a cTnI level of 0.24 μg/l and nearly 20 % at the 99 % percentile [14].
For TNF-α, IL-6 and IL-8, IMMULITE immunoassays obtained from DPC Biermann GmbH (Bad Nauheim, Germany) were used. According to the manufacturer information, reference ranges (95% percentile) and their analytical variation for TNF-α, IL-6 and IL-8 are 12% at 8.1 ng/l, 7% at 9.7 ng/l and 3.5% at 62 ng/l respectively. The LoD and the analytical variation at this level are 18% at 4 ng/l; 10% at 2 ng/l and 6% at 5 ng/l.
TBARS was measured according to TAKEDA et al. [19]. To avoid unspecific TBARS changes caused by in vitro oxidation and to guarantee high precision in the analytical procedures, adequate conditions for sampling, storage and measurement were ascertained in pre-study experiments [20]. Therefore, samples for TBARS were immediately frozen after sampling, stored in liquid nitrogen until measurement and analysed in parallel for high precision. Under these circumstances, the within-run variation was 3.8 %. There is no standardized test for TBARS, so that measurements have to be compared with the laboratory specific reference range. Based on former investigations, the reference group of 20 age-matched healthy controls had a mean value of 3.82 ± 0.51 μmol/l (min 2.62 – max 5.06 μmol/l). The 95%-percentile of the reference group is 4.79 μmol/l.
Statistical Analysis
Data are expressed as mean ± SD. Statistical analysis was performed with the SPSS 12 software package (SPSS; Chicago, Ill, USA). Using bootstrapping simulation, the U-test of Mann-Whitney for group comparison and the Wilcoxon test for comparison of paired values (pre-study value vs. maximum value following contrast echocardiography) were used. In the figures, the measurements below the LoD are displayed by the values that correspond to half of the LoD.
Results
A) Marker levels at baseline
cTnI
7 of the 20 patients had baseline cTnI levels below the LoD (0.04 ug/l). 3 patients had a cTnI level between the LoD and the 99%-percentile of the reference population (≤ 0.07μg/l). Altogether, 10 patients were within the reference range. The other 10 patients had baseline cTnI levels above the reference range: 8 patients with values between the 99%-percentile and the ROC cut-off of 0.6 μg/l and 2 patients with values which were even above the ROC cut-off. (see table 2, figure 1). As shown in figure 2, one patient had a much higher baseline cTnI level (1.99μg/l) than all the others. The patient was diagnosed with chronic dilated cardiomyopathy and intermittent ventricular bigeminus and couplets. In contrast to elevated levels of cTnI and TBARS, he had normal levels of the other markers at baseline.
Table 2 Classification of patients according to baseline levels of cTnI, TNF-α, IL-6, IL-8 and TBARS with respect to inside or outside the reference range
Within reference range Outside reference range
cTnI (n = 20) 10 10
TNF-α (n = 16) 11 3
IL-6 (n = 16) 14 2
IL-8 (n = 16) 16 0
TBARS (n = 20) 6 14
Figure 1 Boxplot analysis of the baseline concentration of TNF-α, IL-6, IL-8 and TBARS in the plasma of patients with plasma cTnI inside or outside of the reference range. Despite the significant differences in the mean values of TNF-α and IL-6 at baseline, the intra-individual changes were not statistically significant.
Figure 2 Increase of cTnI after contrast echocardiography: Baseline value and the maximum value within 24 hours after contrast echocardiography. The dotted lines indicate patients with no significant increase. The black lines represent patients with an increase greater than the three-fold of the within-run variation of the method.
Cytokines
For TNF-α (reference range ≤ 8.1 ng/l), 3 of the 16 patients had baseline values above the reference range. Two patients belonged to the group with cTnI in the reference range, one to the group with cTnI above the reference range.
However, comparing all baseline TNF-α levels to the cTnI levels (inside or outside the reference range), TNF-α was significantly lower in the patient group with cTnI in the reference range (3.8 ± 2.6 vs. 9.4 ± 5.7 ng/l ; 95 % confidence interval of the p value 0.030 – 0.037) than the patients with cTnI above the reference range. The correlation of cTn-I and TNF-α is also documented by the regression analysis (r = 0.57; p = 0.043).
For IL-6 (reference range of ≤ 9.7 ng/l), two patients both from the cTnI group outside of the reference range had IL-6 values outside of the reference range. In analogy to the TNF-α levels, there was a significantly higher IL-6 level in the patient group outside of the cTnI reference range compared with that inside (vs. 2.0 ± 1.0 vs.13.3 ± 16.9 ng/l; 95 % confidence interval of the P value 0.009 to 0.014). The linear regression of baseline IL-6 and cTnI was r = 0.68 (p = 0.01). A significant correlation of baseline TNF-α and IL-6 could be calculated (r = 0.58; p < 0.05).
For IL-8, the baseline values of all patients were all within the reference range. There was no significant difference in IL-8 values between the patient groups inside and outside of the cTnI reference interval (13.8 ± 22.6 vs. 10.0 ± 7.2 ng/l).
Oxidative stress
Based on the calculated reference range (95%-percentile) the TBARS level of 6 of the patients was inside the reference range, while in 14 patients it was outside of the reference range at baseline. TBARS levels were significantly higher in the patient group compared to the healthy controls: mean 5.82 ± 1.30 μmol/l (min 4.18 μmol/l, max 9.23 μmol/l, 95 % confidence interval of the P value: 0.001 to 0.0001). However, no significantly different TBARS levels (5.98 ± 1.65 vs 5.61+0.81 μmol/l) were found in the cTnI groups inside or outside the reference range. There was no significant correlation of baseline TBARS to cTnI, TNF-α and IL-6, however, the TBARS correlated significantly to the baseline IL-8 (r = 0.82; p < 0.01).
Overall, if patients were classified based on their underlying heart disease in "ischemic" (n = 6) and "non-ischemic" (n = 14), no significantly different pre-study levels for all parameters are present.
B) Marker levels after contrast echocardiography
cTnI
As shown in figure 1 the mean cTnI levels were 0.26 ± 0.47 ng/l (baseline), 0.47 ± 1.37 ng/l (2 hours), 0.24 ± 0.48 ng/l (4 hours), 0.24 ± 0.45 ng/l (8 hours), 0.18 ± 0.33 (24 hours). Based on the definition of marker increase as described in material and methods, the intra-individual kinetics showed a cTnI increase for 9 patients after contrast echocardiography at one or more time points compared to baseline (figure 2). The strongest increase (to 6.35 μg/l) was found for the patient with the highest baseline cTnI. This high value was only found 2 hours after contrast echocardiography and returned to baseline thereafter. This patient had no significant changes of the cytokine and TBARS levels. Despite a cTnI increase in some patients, statistical analysis using the Wilcoxon test for comparison of paired values did not reveal a significant increase for the whole group of patients (figure 3).
Figure 3 Time kinetics of cTnI, TNF-α, IL-6, IL-8 and TBARS levels at baseline and at 2, 4, 8 24 hours and maximum value after contrast echocardiography.
There was no predictive power of baseline cTnI levels for an increase after contrast echocardiography. In the group with baseline cTnI within the reference range, we found an increase based on our definition in 5 of the 10 patients. In the group with baseline cTnI above the reference range, the increase was seen in 4 of the 10 patients.
Neither the etiology of the heart disease (ischemic and non-ischemic) nor the baseline cTnI level predicted an increase in cTnI.
TNF-α and Cytokines
In the case of TNF-α and cytokines, the critical difference was crossed 24 hours after contrast echocardiography for TNF-α in 3, for IL-6 in 6 and for IL-8 in 7 of the analyzed patients. The marker increase was neither related to the baseline cTnI nor to the ischemic or non-ischemic background. The correlation of cytokine and cTnI increase was not statistically significant.
Oxidative stress
The oxidative stress marker TBARS exceeded the critical difference after contrast echocardiography in 7 patients. The increase of TBARS was neither related to the baseline cTnI nor to the ischemic or non-ischemic background of the underlying myocardial pathology of the patient.
Discussion
This study was designed to investigate safety of contrast echocardiography in humans. By determination of serum markers for myocardial necrosis, inflammation and oxidative stress, no statistically significant side-effects could be detected. In the most of the patients, the baseline marker levels were within or near to the reference range and there were no sustained increases 24 hours after contrast echocardiography.
As marker for myocardial necrosis, cTnI was chosen because the analytical sensitivity of the test has improved and it is now accepted as an indicator of myocardial micro-necrosis. Furthermore, there is increasing evidence that cTnI can indicate myocardial damage that is not associated with necrosis [21-24].
To examine inflammatory activation after contrast echocardiography, TNF-α and IL-6 and IL-8 were analyzed. These markers are not specific for a myocardial inflammatory response, but a marker increase with temporal association to contrast echocardiography more likely results from the proposed effects of the contrast agent on the heart than from systemic inflammation by other causes [25].
TBARS, a metabolite and end product of lipid peroxidation, is a more static and global parameter. TBARS is, despite the lack in standardization, currently the most frequently used marker to quantify oxidative stress in clinical studies. It was measured in several studies in patients with heart diseases and increased oxidative stress [12,26]. The patients in our study had significantly higher baseline TBARS levels than the control group of age-matched healthy subjects. This is in agreement with the data demonstrating that the induction of oxidative stress is a sequelae in the pathogenesis of heart diseases.
Our results after contrast echocardiograph in humans are be in agreement with a recently published study. A new second-generation myocardial contrast agent (LK565) has been evaluated in healthy individuals undergoing contrast echocardiography. Neither an increase in TNF-α secretion, nor antibody development and activation of leukocyte surface markers was induced by the procedure [27]. However, these results found in healthy subjects cannot be simply transferred to studies such ours since patients with heart disease could develop side-effects which are more pronounced in quality and quantity.
Consequently and in critical view of our study results, we hesitate to negate a potential myocardial damage after contrast echocardiography. An intra-individual increase of cTnI above the pre-determined limit was seen in nearly half of the patients. Furthermore, IL-6 and IL-8 increased after contrast echocardiography based on the pre-defined limit in nearly half of the patients analysed for cytokines. The increases of cTnI, IL-6 and IL-8 were not correlated to a particular group of patients. Further larger studies have to clarify whether subgroups of patients are prone to myocardial damage after contrast echocardiography. In present study, we cannot exclude that the findings in some of our patients resulted from such a subset enrolled in our patient cohort. Due to the sample size we were not able to clearly define subgroups. In our view, patients with cTnI or TBARS above the reference range at baseline could form such a subgroup more susceptible to side- effects.
Possibly, the one patient suffering from dilated cardiomyopathy and ventricular arrhythmia who had the highest baseline and largest increase in cTnI levels after contrast echocardiography could belong to such a subgroup. High troponin levels are found in patients with stable heart failure without an indication for myocardial ischemia [20-22,28]. Furthermore, patients with tachycardia can have persistently high troponin levels [29] which might explain the elevated baseline level of cTnI in our patient affected by intermittent ventricular bigeminus and ventricular couplets. However, the cTnI increase after contrast echocardiography in this patient was neither reflected by a rise in cytokines nor in oxidative stress and was not accompanied by clinical signs and symptoms of myocardial ischemia. Therefore, it is unclear whether this cTnI increase is linked to the contrast echocardiography procedure alone or if it primarily reflects the underlying heart disease.
We have found a significant correlation of the baseline levels of cTnI, TNF-α and IL6. This is in accordance with the findings of [30], who found an association of troponin and cytokine levels in heart disease.
Limitations of this study include the low number of patients, the lack of a control group and the use of just one contrast agent. Increasing the number of patients studied would facilitate the identification of patients who are more sensitive to the side-effects of contrast echocardiography. Controls in future studies would have to receive contrast agent only without echocardiography. We cannot draw comparisons between the effects of different contrast agents because only Optison was used in this study.
In summary, there is no conclusive evidence for the induction of myocardial micro-necrosis, inflammation and oxidative stress by contrast echocardiography. The baseline cTnI levels do not correlate to the oxidative stress and inflammation markers within 24 hours after contrast echocardiography. In some of the patients however, cTnI levels increase slightly after contrast echocardiography. These patients may belong to a subgroup more prone to side effects of contrast echocardiography than others. Hence, this study – although limited in size – suggests that contrast echocardiography does not induce oxidative stress and inflammation.
Abbreviations
TNF-α Tumor necrosis factor alpha
TBARS thiobarbituric acid reactive substances
IL Interleukin
ROC Reviever operating characteristic curve
cTnI cardiac Troponin I
LoD lowest limit of detection
MI myocardial infarction
Authors' contributions
FK, IS and ACB have designed and performed the study and have written the manuscript. IS, FK, RZ and ACB have performed the laboratory measurements. TW, SE, SS and GB have participated in the study design and coordination. All authors read and approved the final manuscript.
==== Refs
Main ML Kusnetzky LL Dillon D Daniel WC Reperfusion assessment using myocardial contrast echocardiography in patients with ST-segment elevation acute myocardial infarction Am J Cardiol 2004 93 1401 1403 15165924 10.1016/j.amjcard.2004.02.040
Janardhanan R Swinburn JM Greaves K Senior R Usefulness of myocardial contrast echocardiography using low-power continuous imaging early after acute myocardial infarction to predict late functional left ventricular recovery Am J Cardiol 2003 92 493 497 12943865 10.1016/S0002-9149(03)00713-6
Sullivan GW Sarembock IJ Linden J The role of inflammation in vascular diseases J Leukoc Biol 2000 67 591 602 10810997
Skyba DM Price RJ Linka AZ Skalak TC Kaul S Direct in vivo visualization of intravascular destruction of microbubbles by ultrasound and its local effects on tissue Circulation 1998 98 290 293 9711932
Lindner JR Ismail S Spotnitz WD Skyba DM Jayaweera AR Kaul S Albumin microbubble persistence during myocardial contrast echocardiography is associated with microvascular endothelial glycocalyx damage Circulation 1998 98 2187 2194 9815874
Miller DL Quddus J Diagnostic ultrasound activation of contrast agent gas bodies induces capillary rupture in mice Proc Natl Acad Sci USA 2000 97 10179 10184 10954753 10.1073/pnas.180294397
Basta G Venneri L Lazzerini G Pasanisi E Pianelli M Vesentini N Del Turco S Kusmic C Picano E In vitro modulation of intracellular oxidative stress of endothelial cells by diagnostic cardiac ultrasound Cardiovasc Res 2003 58 156 161 12667957 10.1016/S0008-6363(02)00665-X
McMurray J McLay J Chopra M Bridges A Belch JJ Evidence for enhanced free radical activity in chronic congestive heart failure secondary to coronary artery disease Am J Cardiol 1990 65 1261 1262 2337039 10.1016/0002-9149(90)90985-A
Belch JJ Bridges AB Scott N Chopra M Oxygen free radicals and congestive heart failure Br Heart J 1991 65 245 248 2039668
Bertuglia S Giusti A Picano E Effects of diagnostic cardiac ultrasound on oxygen free radical production and microvascular perfusion during ischemia reperfusion Ultrasound Med Biol 2004 4 549 557 15121258 10.1016/j.ultrasmedbio.2003.12.008
Nian M Lee P Khaper N Liu P Inflammatory cytokines and postmyocardial infarction remodelling: Circ Res 2004 94 1543 1553 15217919 10.1161/01.RES.0000130526.20854.fa
Walter MF Jacob RF Jeffers B Ghadanfar MM Preston GM Buch J Mason RP PREVENT study. Serum levels of thiobarbituric acid reactive substances predict cardiovascular events in patients with stable coronary artery disease: a longitudinal analysis of the PREVENT study J Am Coll Cardiol 2004 44 1996 2002 15542282 10.1016/j.jacc.2004.08.029
Borges AC Walde T Reibis RK Grohmann A Ziebig R Rutsch W Schimke I Baumann G Does contrast echocardiography with Optison induce myocardial necrosis in humans? J Am Soc Echocardiogr 2002 10 1080 1086 12373250
Panteghini M Pagani F Yeo KT Apple FS Christenson RH Dati F Mair J Ravkilde J Wu AH Committee on Standardization of Markers of Cardiac Damage of the IFCC. Evaluation of imprecision for cardiac troponin assays at low-range concentrations Clin Chem 2004 2 327 332 14656904 10.1373/clinchem.2003.026815
Alpert JS Thygesen K Antman E Bassand JP Myocardial infarction redefined – a consensus document of The Joint European Society of Cardiology/American College of Cardiology Committee for the redefinition of myocardial infarction J Am Coll Cardiol 2000 36 959 69 10987628 10.1016/S0735-1097(00)00804-4
Apple FS Quist HE Doyale PJ Otto AP Murakami MM Plasma 99th Percentile Reference Limits for Cardiac Troponin and Creatinine Kinase MB Mass for Use with European Society of Cardiology / American College of Cardiology Consensus Recommendations Clinical Chemistry 2003 49 1331 1336 12881449 10.1373/49.8.1331
Apple FS Wu AH Jaffe AS European Society of Cardiology and American College of Cardiology guidelines for redefinition of myocardial infarction: how to use existing assays clinically and for clinical trials Am Heart J 2002 144 981 986 12486421 10.1067/mhj.2002.124048
Schulz O Sigusch HH Impact of an exercise-induced increase in cardiac troponin I in chronic heart failure secondary to ischemic or idiopathic dilated cardiomyopathy Am J Cardiol 2002 90 547 550 12208423 10.1016/S0002-9149(02)02535-3
Takeda K Shimada Y Amano M Sakai T Okada T Yoshiya I Plasma lipid peroxides and alpha-tocopherol in critically ill patients Crit Care Med 1984 11 957 959 6499481
Schimke I Papies B [Einige methodische Aspekte der Bestimmung thiobarbitursäurereaktiver Substanzen im Plasma] Some methodological aspects of the determination of thiobarbituric reactive substances in plasma Z Med Lab Diagn 1986 27 71 75 3705704
Angheloiu GO Dickerson RP Ravakhah K Etiology of troponin I elevation in patients with congestive heart failure and low clinical suspicion of myocardial infarction Resuscitation 2004 63 195 201 15531072 10.1016/j.resuscitation.2004.05.018
Hudson MP O'Connor CM Gattis WA Tasissa G Hasselblad V Holleman CM Gaulden LH Sedor F Ohman EM Implications of elevated cardiac troponin T in ambulatory patients with heart failure: a prospective analysis Am Heart J 2004 147 546 552 14999208 10.1016/j.ahj.2003.10.014
Potluri S Ventura HO Mulumudi M Mehra MR Cardiac troponin levels in heart failure Cardiol Rev 2004 12 21 25 14667259 10.1097/01.crd.0000089981.53961.cf
Bakshi TK Choo MK Edwards CC Scott AG Hart HH Armstrong GP Causes of elevated troponin I with a normal coronary angiogram Intern Med J 2002 32 520 525 12412934 10.1046/j.1445-5994.2002.00270.x
Arguelles S Garcia S Maldonado M Machado A Ayala A Do the serum oxidative stress biomarkers provide a reasonable index of the general oxidative stress status? Biochim Biophys Acta 2004 1674 251 259 15541294
Schimke I Muller J Priem F Kruse I Schon B Stein J Kunze R Wallukat G Hetzer R Decreased oxidative stress in patients with idiopathic dilated cardiomyopathy one year after immunoglobulin adsorption J Am Coll Cardiol 2001 38 178 183 11451270 10.1016/S0735-1097(01)01309-2
Funke B Maerz HK Okorokow S Polata S Lehmann I Sack U Wild P Geisler T Zotz RJ Immunological evaluation of the new stable ultrasound contrast agent LK565: a phase one clinical trial Cardiovascular Ultrasound 2004 2 16 15357870 10.1186/1476-7120-2-16
Xue C Yu H Li R Wo J Cui J Cheng H Wang H Guan Q Suo X Jia R Clinical significance of serum cardiac troponin T in patients with congestive heart failure Chin Med J (Engl) 2003 116 469 471 12781062
Zeng L Chen Y Wu M [Cardiac troponin I: a marker for detecting non-ischemic cardiac injury] Zhonghua Yi Xue Za Zhi 2001 81 393 395 11798903
Cusack MR Marber MS Lambiase PD Bucknall CA Redwood SR Systemic inflammation in unstable angina is the result of myocardial necrosis J Am Coll Cardiol 2002 39 1917 1923 12084588 10.1016/S0735-1097(02)01899-5
|
16107209
|
PMC1208924
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Aug 17; 3:21
|
utf-8
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-21
|
oa_comm
|
==== Front
Cardiovasc UltrasoundCardiovascular Ultrasound1476-7120BioMed Central London 1476-7120-3-261613892310.1186/1476-7120-3-26Case ReportLong-term therapy of interferon-alpha induced pulmonary arterial hypertension with different PDE-5 inhibitors: a case report Jochmann Nicoline [email protected] Felix [email protected] Adrian C [email protected] Maja A [email protected] Stephan [email protected] Wolfram [email protected] Gert [email protected] Uwe [email protected] Department of Cardiology, Charité – Universitätsmedizin Berlin, Campus Mitte, Berlin, Germany, Schumannstrasse 20/21, 10117 Berlin, Germany2 Department of Dermatology and Allergy, Skin Cancer Centre, Charité – Universitätsmedizin Berlin, Schumannstrasse 20/21, 10117 Berlin, Germany2005 2 9 2005 3 26 26 12 8 2005 2 9 2005 Copyright © 2005 Jochmann et al; licensee BioMed Central Ltd.2005Jochmann et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
background
Interferon alpha2 is widely used in hepatitis and high-risk melanoma. Interferon-induced pulmonary arterial hypertension as a side effect is rare.
Case presentation
We describe a melanoma patient who developed severe pulmonary arterial hypertension 30 months after initiation of adjuvant interferon alpha2b therapy. Discontinuation of interferon did not improve pulmonary arterial hypertension. This patient could be treated successfully with phosphodiesterase-5 inhibitor therapy.
Conclusion
This is only the 5th case of interferon-induced pulmonary arterial hypertension and the first documented case where pulmonary arterial hypertension was not reversible after termination of interferon alpha2 therapy. If interferon alpha2 treated patients develop respiratory symptoms, pulmonary arterial hypertension should be considered in the differential diagnosis. For these patients phosphodiesterase-5 inhibitors, e.g. sildenafil or vardenafil, could be an effective therapeutic approach.
pulmonary arterial hypertensionPAHPDE-5 inhibitorinterferon-alpha
==== Body
Background
The interferons (IFN) are a group of glycoproteins produced by a wide range of cells in response to viruses, mitogens, double-stranded RNA and other substances. They perform immunoregulatory, as well as antiviral and antineoplastic functions, with the latter being the result of inhibition of cell proliferation, enhanced MHC expression and tumor-associated antigen expression. The alpha interferon's (IFN α2a and IFN α2b) act as immunomodulators by enhancing natural killer cells, macrophages and T-lymphocyte function, as well as having antiangiogenic properties. Various forms of IFNs have been evaluated as therapy in a variety of malignant and non-malignant diseases. The major oncologic indications for IFNs include malignant melanoma, renal cell carcinoma (RCC), AIDS-related or HHV-8 associated Kaposi's sarcoma, cutaneous T-cell lymphoma, hairy cell leukemia, and chronic myelogenous leukemia (CML), whereas the non-oncologic indications include viral infections (including hepatitis C and HPV-associated lesions such as condylomata acuminata), multiple sclerosis, keloids, keratoacanthoma, Behcet's disease or hemangioma [1]. IFN α2 is approved in the US and Europe for adjuvant therapy of melanoma and is considered the standard therapy for high-risk melanoma [2]. Among the side effects are flu-like symptoms such as fever, chills and anorexia, myalgia, as well as neuropathies and neuropsychiatric side effects, bone marrow depression, liver and renal failure, heart failure, cardiac arrhythmias, peripheral hypo- and hypertension and vascular side effects like Raynaud's phenomena, digital ulceration and gangrene [2,3]. Pulmonary arterial hypertension (PAH) and interstitial pneumonitis are described as rare side effects [3-8]. We describe a female patient with high risk melanoma who developed severe PAH 30 months after initiation of adjuvant IFN therapy and who could be treated successfully with PDE-5 inhibitor therapy.
Case Presentation
A 40-year-old woman received excision of a superficial spreading melanoma from the rima ani with a safety margin of 3 cm (Clark-Level IV, tumor thickness 1,82 mm). Lymphatic drainage was detected to both inguinal basins and both excised sentinel lymph nodes were unaffected. None of the staging examinations including computer tomography (CT) of the brain, chest, abdomen and pelvis, as well as lymph node sonography revealed any signs of tumor manifestation. The medical history of the patient was otherwise unremarkable and she was not on any medication. There was no family history of hypertension, heart disease or pulmonary disease. Because of the high-risk nature of the melanoma, the patient started long-term adjuvant therapy with IFN α2b (5 × 10 million U. s.c. per week for 4 weeks followed by 3 × 10 million U. s.c. per week).
After 30 months of IFN α2b treatment the patient reported increasing dyspnea on exertion and afebrile non-productive coughing accompanied by sudden malaise and edema of the lower legs. Electrocardiography showed sinus tachycardia (120 /min) and right axis deviation. A chest x-ray showed signs of right ventricular dilatation and pleural effusion on the right side; no pneumonic infiltrates were seen. Abdominal sonography revealed a significant amount of ascites. The patient was diagnosed with decompensated right heart failure and was therefore hospitalized. Initial investigations with transthoracic echocardiography showed right ventricular hypertrophy and dilatation (Figure 1), PAH with a calculated systolic pulmonary artery pressure (PAPsyst) of 80 mmHg and tricuspid insufficiency grade II-III with morphologically normal valves (Figure 2), a reduced right ventricular ejection fraction of 40%, a hypokinetic right ventricle and pericardial effusion without signs of tamponade. Laboratory work-up showed slightly increased levels of d-dimers and liver enzymes, while inflammatory markers were within the normal range. There were no signs of vasculitis, hypercoagulability or rheumatologic disorders. A high-resolution CT of the chest revealed no signs of pulmonary embolism, alveolar or interstitial lung diseases, but signs of PAH with a widened central pulmonary artery (40 mm), right ventricular dilatation (> 80 mm), regurgitation of contrast medium into liver veins, a circular pericardial effusion and a 300–400 ml pleural effusion of the right side.
Figure 1 Right ventricular hypertrophy and dilatation at initial investigation with transthoracic echocardiography.
Figure 2 Tricuspid insufficiency grade II–III with a morphological normal valve at initial investigation with transthoracic echocardiography.
Diagnostic right heart catheter revealed a PAPmean of 56 mmHg (PAPsyst 87 mmHg), a pulmonary vascular resistance (PVR) of 1.128 dyn × sec × cm-5, an impaired cardiac index and a 3 fold increased total peripheral resistance. Testing of pulmonary vasoreactivity showed a reduction of PAPmean from 56 mmHg to 26 mmHg with the PDE-5-inhibitor sildenafil (Table 1).
Table 1 Testing of pulmonary vasoreactivity at initial investigation and after six months of PDE-5 inhibitor therapy with right heart catheter
HR RR mmHg syst/diast (mean) PAP mmHg syst/diast (mean) RA mmHg PCWP mmHg CI l/min/m2 SV ml/m2 TPR dyn × sec × cm-5 PVR dyn × sec × cm-5
baseline 103 140/89 (104) 87/37 (56) 28 8 1,9 33 3.178 1.128
baseline sildenafil 96 106/58 (76) 74/14 (26) 13 3 2,2 41 1.290 760
six months sildenafil 96 105/71 (88) 64/27 (43) 8 4 2,6 46 1.453 708
six months vardenafil 102 114/69 (83) 63/25 (40) 6 6 2,6 44 1.403 604
baseline = initial investigation; baseline sildenafil = 90 minutes after first dose of 25 mg sildenafil p.o.; six months sildenafil = after 6 months of therapy with sildenafil; six months vardenafil = 120 minutes after first dose of 10 mg vardenafil p.o. HR = heart rate; RR = blood pressure; PAP = pulmonary arterial pressure; RA = right atrial pressure; PCWP = pulmonary capillary wedge pressure; CI = cardiac index; SV = stroke volume; TPR = total peripheral resistance; PVR = pulmonary vascular resistance.
Therefore, treatment with 3 × 25 mg sildenafil per day was initiated. No side effects occurred. After one month, tricuspid insufficiency improved to grade I-II. After six months right ventricular hypertrophy and dilatation were reduced. Because of the 2 documented cases of patients with PAH due to therapy with IFNα2 where PAH was reversible half a year after termination of IFNα2 [3,4] we attempted to terminate sildenafil under control of hemodynamic monitoring. PAPmean increased promptly to 57 mmHg. In this context we could verify the same effectiveness for therapy with vardenafil in comparison to sildenafil. Reduction of PVR was higher in vardenafil vs. sildenafil (Table 1); therefore the therapy was switched from sildenafil to vardenafil 10 mg twice daily. 24 months after onset of PAH the patient felt fine and resumed working. Vardenafil is still necessary to lower her PAP as demonstrated with transthoracic echocardiography: after pausing vardenafil for two days the PAPsyst increased about 40 mmHg, but decreased again 120 minutes after administration of vardenafil. The heart diameter is steadily decreasing (right ventricular outflow tract from 39 to 27 mm and right atrium from 60–65 to 36–44 mm). Regarding the melanoma she remains relapse-free.
Conclusion
IFN α2 is an accepted adjuvant treatment for patients with high risk melanoma [2]. Among the vascular complications reported for IFN are retinopathy [9], cutaneous vasculitis [10], gangrene requiring amputation and biopsy proven pulmonary vasculitis [3]. PAH includes various forms with different etiologies, but similar clinical presentation and functional derangement. Although PAH remains a rare disease, in recent times PAH related to other diseases has been better recognized. These forms are related to systemic connective tissue diseases, thromboembolic disease, congenital heart disease, portal hypertension, HIV infection, or are secondary to the use of drugs. They all result in an indistinguishable histological picture [11]. In addition to common hypertrophy of the tunica media, other proliferative lesions such as intimal thickening or plexiform lesions can be found. Moreover, in situ thrombosis and rarely isolated pulmonary arteritis can be observed in lungs of patients displaying PAH. Different pathomechanisms explaining these morphological changes of pulmonary vasculature have been discussed in the past, including endothelial and thrombocytic dysfunction, vasoconstriction, coagulation abnormalities, or cancer-like growth [12]. It is therefore of interest that IFNα can cause thrombotic microangiopathy which might contribute to the development of PAH [13].
Various cellular pathway abnormalities have been described that may play important roles in the development and progression of PAH [14,15]. These include altered synthesis of nitric oxide (NO), prostacyclin and endothelin, impaired potassium channel and growth factor receptor function, altered serotonin transporter regulation, increased oxidant stress, and enhanced matrix production [14-17]. However, the relative importance of each of these processes remains unclear. It has been reported, that in sheep IFNα is able to increase the PVR and the PAP by activation of the thromboxane-cascade with elevated levels of thromboxane-B2 in plasma and lung lymph [18]. Studies on the importance of inflammatory mediators, such as chemokines, in the lungs of PAH patients have led to a possible inflammatory component in the development of PAH [19]. This might be of relevance in IFN induced PAH since IFNα is known to induce expression of various chemokines [20].
All PAH result in similar histological remodelling of pulmonary arteries with thickening of the intima, proliferation of the media and plexogenic lesions. Today the pathophysiology of these lesions is much better understood and has resulted in new therapies involving substances such as prostacyclins, endothelin receptor antagonists or PDE-5 inhibitors, aimed not only at dilating arteries but also at preventing their remodelling.
Sildenafil and vardenafil inhibit PDE-5, an enzyme that is abundantly expressed in pulmonary vasculature [21]. PDE-5 is the major guanosine 3',5'-cyclic monophosphate (cGMP) degrading phosphodiesterase. PDE-5 gene expression and activity are increased in PAH [22]. Sildenafil binds to the catalytic site of PDE-5 approximately a thousand times more avidly than the natural substrate, cGMP [23]. cGMP is the second messenger of prostacyclin and NO and due to stabilization of this second messenger, therapy with PDE-5 inhibitors leads to prolongation of prostanoid- and NO-related vascular effects [24]. In patients with PAH, short-term application of sildenafil during right heart catheterization showed the potential to reduce PVR in a dose-dependent manner. Interestingly, the vasodilator effects were significantly stronger than with inhaled NO [25]. Advantages of a potential therapy with PDE-5-inhibitors for PAH are p.o. administration, an excellent safety profile and relatively low treatment costs [26]. A number of studies have shown the effectiveness of a therapy with PDE-5 inhibitors for PAH [25-32]. Because of the low incidence of this condition the number of patients studied is small. One study could show a long-lasting benefit with sildenafil with respect to hemodynamic and clinical parameters over 3, 6 and 12 months [26]. A recently published placebo-controlled and randomized phase III study revealed a significant reduction of PAP, improvement of 6-minute-walk test and reduction of hospitalization due to PAH [32].
Data are rare for the use of different PDE-5 inhibitors other than sildenafil in PAH. The in vitro biochemical potentials of sildenafil and vardenafil are comparable, whereas vardenafil has a binding affinity to PDE-5 more than ten times higher than sildenafil [33]. One could speculate that therapy with vardenafil in PAH could be more effective than with sildenafil due to its higher binding affinity to PDE-5. In a small study with different PDE-5 inhibitors in PAH, the reduction of PAP with sildenafil and vardenafil was comparable, while sildenafil showed surprisingly more selectivity for the pulmonary circulation and a better arterial oxygenation compared to vardenafil [34]. In the case presented our patient responded favorably to both sildenafil and vardenafil.
This is only the 5th case of IFN-induced PAH [3-5,7]. The previous patients received IFN for CML (three patients) or RCC (one patient). Our patient developed PAH after 2.5 years of adjuvant IFNα treatment, in the other cases the range was from several months up to 22 months after initiation of IFN therapy. In two cases PAH was reversible within 6 months after termination of IFNα therapy. In one case, the patient died 2 years after withdrawal from IFN therapy. An autopsy was not performed, but the authors suggested a systemic microangiopathy caused by IFNα as underlying reason [5]. To the best of our knowledge this is the first documented case where PAH was not reversible after termination of IFNα therapy and there was a need for continuous vasodilator therapy. Treatment with PDE-5-inhibitors had, in this case, a long-lasting beneficial effect. If IFNα treated patients develop respiratory symptoms, PAH should be considered in the differential diagnosis. For these patients PDE-5 inhibitors, e.g. sildenafil or vardenafil, could be an effective therapeutic approach.
Abbreviations
cGMP = guanosine 3'5'-cyclic monophosphate
CML = chronic myelogenous leukemia
CT = computer tomopgraphy
IFN = Interferon
PAH = pulmonary arterial hypertension
PAP = pulmonary arterial pressure
PDE-5 = phosphodiesterase-5
PVR = pulmonary vasculature resistance
Competing interest statement
The author(s) declare that they have no competing interests.
Authors' contributions
NJ, ACB and GB treated the patient at the cardiology department. ACB, GB and SE performed the cardiological examinations. MAH, FK, WS and UT treated the patient at the dermatology department. NF, FK and UT drafted the manuscript. GB and WS revised the manuscript critically. All authors approved the final version of the manuscript.
==== Refs
Jonasch E Haluska FG Interferon in oncological practice: review of interferon biologic clinical applications, and toxicities Oncologist 2001 6 34 55 11161227 10.1634/theoncologist.6-1-34
Kirkwood J Cancer immunotherapy: the interferon-alpha experience Semin Oncol 2002 29 18 26 12068384
Al-Zahrani H Gupta V Minden MD Messner HA Lipton JH Vascular events associated with alpha interferon therapy Leuk Lymphoma 2003 44 471 475 12688317 10.1080/1042819021000055066
Anderson P Hoglund M Rodjer S Pulmonary side effects of interferon-alpha therapy in patients with hematological malignancies Am J Hematol 2003 73 54 58 12701122 10.1002/ajh.10319
Fruehauf S Steider S Topaly J Ho AD Pulmonary artery hypertension during interferon-alpha therapy for chronic myelogenous leukemia Ann Hematol 2001 80 308 310 11446736 10.1007/s002770100298
Kamisako T Adachi Y Chihara J Yamamoto T Interstitial pneumonitis and interferon-alpha BMJ 1993 306 896 8490417
Kramers C de Mulder PH Barth JD Wagener DJ Acute right ventricular heart failure in a patient with renal cell carcinoma after interferon therapy Neth J Med 1993 42 65 68 8446228
Nakamura F Andoh A Minamigushi H Hodohara K Fujiyama Y Bamba TA A case of interstitial pneumonitis associated with natural alpha-interferon therapy for myelofibrosis Acta Haematol 1997 97 222 224 9158665
Kawano T Shigehira M Uto H Nakama T Kato J Hayashi K Maruyama T Kuribayashi T Chuman T Futami T Tsubouchi H Retinal complications during interferon therapy for chronic hepatitis C Am J Gastroenterol 1996 91 309 313 8607498
Sangster G Kaye SB Calman KC Toy JL Cutaneous vasculitis associated with interferon Eur J Cancer Clin Oncol 1983 19 1647 1649 6685647 10.1016/0277-5379(83)90099-8
Nicod LP Pulmonary hypertension Swiss Med Wkly 2003 1338 103 110 12644956
Pietra GG Capron F Stewart S Leone O Humbert M Robbins IM Reid LM Tuder RM Pathologic assessment of vasculopathies in pulmonary hypertension J Am Coll Cardiol 2004 43 25 32 10.1016/j.jacc.2004.02.033
Politou M Tsaftarides P Vassiliades J Siakantaris MP Michail S Nakopoulou L Pangalis GA Variopoulos G Thrombotic microangiopathy in a patient with Sezary syndrome treated with interferon-alpha Nephrol Dial Transplant 2004 19 733 735 14767035 10.1093/ndt/gfg537
Christman BW McPherson CD Newman JH King GA Bernard GR Groves BM Loyd JE An imbalance between the excretion of thromboxane and prostacyclin metabolites in pulmonary hypertension N Engl J Med 1992 327 70 75 1603138
Tuder RM Cool CD Geraci MW Wang J Abman SH Wright L Badesch D Voelkel NF Prostacyclin synthase expression is decreased in lungs from patients with severe pulmonary hypertension Am J Respir Crit Care Med 1999 159 1925 1932 10351941
Mandegar M Remillard CV Yuan JX Ion channels in pulmonary arterial hypertension Prog Cardiovasc Dis 2002 45 81 114 12411972 10.1053/pcad.2002.127491
Eddhaibi S Humbert M Fadel E Raffestin B Darmon M Capron F Simonneau G Dartevelle P Hamon M Adnot S Serotonin transporter overexpression is responsible for pulmonary artery smooth muscle hyperplasia in primary pulmonary hypertension J Clin Invest 2001 108 1141 1150 11602621 10.1172/JCI200112805
Hanaoka M Kubo K Hayano T Koizumi T Kobayashi T Interferon-alpha elevates pulmonary blood pressure in sheep – the role of thromboxane cascade Eur J Pharmacol 1999 370 145 151 10323263 10.1016/S0014-2999(99)00107-7
Dorfmuller P Humbert M Capron F Muller KM Pathology and aspects of pathogenesis in pulmonary arterial hypertension Sarcoidosis Vasc Diffuse Lung Dis 2003 20 9 19 12737275
Parlato S Santini SM Lapenta C Di Pucchio T Logozzi M Spada M Giammarioli AM Malorni W Fais S Belardelli F Expression of CCR-7, MIP-3beta, and Th-1 chemokines in type I IFN-induced monocyte-derived dendritic cells: importance for the rapid acquisition of potent migratory and functional activities Blood 2001 98 3022 3029 11698286 10.1182/blood.V98.10.3022
Beavo JA Cyclic nucleotide phosphodiesterases: functional implications of multiple isoforms Physiol Rev 1995 75 725 748 7480160
Hanson KA Burns F Rybalkin SD Miller JW Beavo J Clarke WR Developmental changes in lung cGMP phosphodiesterase-5 activity, protein, and message Am J Respir Crit Care Med 1998 158 279 288 9655741
Corbin JD Blount MA Weeks JL Beasley A Kuhn KP Ho YS Saidi LF Hurley JH Kotera J Francis SH (3H)sildenafil binding to phosphodiesterase-5 is specific, kinetically heterogeneous, and stimulated by cGMP Mol Pharmacol 2003 63 1364 1372 12761347 10.1124/mol.63.6.1364
Ghofrani HA Pepke-Zaba J Barbera JA Channick R Keogh AM Gomez-Sanchez MA Kneussl M Grimminger F Nitric oxide pathway and phosphodiesterase inhibitors in pulmonary arterial hypertension J Am Coll Cardiol 2004 43 68 72 14715186 10.1016/j.jacc.2004.02.031
Ghofrani HA Wiedemann R Rose F Olschewski H Schermuly RT Weissmann N Seeger W Grimminger F Combination therapy with oral sildenafil and inhaled iloprost for severe pulmonary hypertension Ann Intern Med 2002 136 515 522 11926786
Michelakis ED Tymchak W Noga M Webster L Wu XC Lien D Wang SH Modry D Archer SL Long-term treatment with oral sildenafil is safe and improves functional capacity and hemodynamics in patients with pulmonary arterial hypertension Circulation 2003 108 2066 2069 14568893 10.1161/01.CIR.0000099502.17776.C2
Bhatia S Frantz RP Severson CJ Durst LA McGoon MD Immediate and long-term hemodynamic and clinical effects of sildenafil in patients with pulmonary arterial hypertension receiving vasodilator therapy Mayo Clin Proc 2003 78 1207 1213 14531479
Ghofrani HA Rose F Schermuly RT Olschewski H Wiedemann R Kreckel A Weissmann N Ghofrani S Enke B Seeger W Grimminger F Oral sildenafil as long-term adjunct therapy to inhaled iloprost in severe pulmonary arterial hypertension J Am Coll Cardiol 2003 42 158 164 12849677 10.1016/S0735-1097(03)00555-2
Ghofrani HA Schermuly RT Rose F Wiedemann R Kohstall MG Kreckel A Olschewski H Weissmann N Enke B Ghofrani S Seeger W Grimminger F Sildenafil for long-term treatment of nonoperable chronic thromboembolic pulmonary hypertension Am J Respir Crit Care Med 2003 167 1139 1141 12684251 10.1164/rccm.200210-1157BC
Ghofrani HA Wiedemann R Rose F Schermuly RT Olschewski H Weissmann N Gunther A Walmrath D Seeger W Grimminger F Sildenafil for treatment of lung fibrosis and pulmonary hypertension: a randomised controlled trial Lancet 2003 360 895 900 12354470 10.1016/S0140-6736(02)11024-5
Prasad S Wilkinson J Gatzoulis MA Sildenafil in primary pulmonary hypertension N Engl J Med 2000 343 1342 11183578 10.1056/NEJM200011023431814
Ghofrani HA for the Sildenafil 1140 Study Group Efficiacy and safety of sildenafil citrate in pulmonary arterial hypertension: results of a multinational, randomized, double-blind, placebo-controlled trial Am Coll Chest Phys 2004
Corbin JD Beasley A Blount MA Francis SH Vardenafil: structural basis for higher potency over sildenafil in inhibiting cGMP-specific phosphodiesterase-5 (PDE5) Neurochemistry International 2004 45 859 863 15312980 10.1016/j.neuint.2004.03.016
Ghofrani HA Voswinckel R Reichenberger F Olschewski H Haredza P Karadas B Schermuly RT Weissmann N Seeger W Grimminger F Differences in hemodynamic and oxygenation responses to three different phosphodiesterase-5 inhibitors in patients with pulmonary arterial hypertension: a randomized prospective study J Am Coll Cardiol 2004 44 1488 1496 15464333 10.1016/j.jacc.2004.06.060
|
16138923
|
PMC1208925
|
CC BY
|
2021-01-04 16:38:31
|
no
|
Cardiovasc Ultrasound. 2005 Sep 2; 3:26
|
utf-8
|
Cardiovasc Ultrasound
| 2,005 |
10.1186/1476-7120-3-26
|
oa_comm
|
==== Front
Chiropr OsteopatChiropractic & Osteopathy1746-1340BioMed Central London 1746-1340-13-131604579510.1186/1746-1340-13-13ReviewThe epidemiology of low back pain in primary care Kent Peter M [email protected] Jennifer L [email protected] School of Physiotherapy, La Trobe University, Melbourne, Victoria, Australia2 Physiotherapy, Monash University, Melbourne, Victoria, Australia2005 26 7 2005 13 13 13 6 5 2005 26 7 2005 Copyright © 2005 Kent and Keating; licensee BioMed Central Ltd.2005Kent and Keating; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This descriptive review provides a summary of the prevalence, activity limitation (disability), care-seeking, natural history and clinical course, treatment outcome, and costs of low back pain (LBP) in primary care.
LBP is a common problem affecting both genders and most ages, for which about one in four adults seeks care in a six-month period. It results in considerable direct and indirect costs, and these costs are financial, workforce and social. Care-seeking behaviour varies depending on cultural factors, the intensity of the pain, the extent of activity limitation and the presence of co-morbidity. Care-seeking for LBP is a significant proportion of caseload for some primary-contact disciplines. Most recent-onset LBP episodes settle but only about one in three resolves completely over a 12-month period. About three in five will recur in an on-going relapsing pattern and about one in 10 do not resolve at all. The cases that do not resolve at all form a persistent LBP group that consume the bulk of LBP compensable care resources and for whom positive outcomes are possible but not frequent or substantial.
==== Body
Review
This descriptive review summarises current knowledge on prevalence, activity limitation (disability), care-seeking, natural history and clinical course, treatment outcome, and costs of low back pain (LBP). Reports of the epidemiology of LBP in primary care were identified through electronic searches of Medline, Cinhahl, Embase, Psychlit, and AMED from inception until October 2004. An example of the search strategies used is attached as Additional file 1. The search also included checking the reference lists of retrieved papers.
Prevalence
Reviews of the literature describing LBP point prevalence in the developed world have produced variable estimates of prevalence rates [1,2]. In the studies deemed by Looney and Stratford to be methodologically superior, the LBP point prevalence was estimated to be 6.8% in North America, 12% in Sweden, 13.7% in Denmark, 14% in the United Kingdom, 28.4% in Canada, and 33% in Belgium [2]. The size of the difference between the North America LBP point prevalence estimated by Deyo and Tsui-Wu at 6.8% [3] and that of Canada at 28.4% [4] illustrates the variability attributable, in unknown proportion, to sample and sampling differences. In a review of world prevalence data, Volinn [5] suggested that there were lower rates of prevalence in developing countries than in developed countries, but did not determine whether differences reflect demographic, cultural or research method factors.
Walker [6] conducted a systematic review of the Australian LBP prevalence literature 1966–1998, and also concluded that the true prevalence of LBP in Australia remained confounded by methodological flaws in previous studies. Walker [7], subsequently surveyed 3000 Australian adults using contemporary epidemiological methods, and estimated the point prevalence of LBP at 25.5%, six-month period prevalence at 64.6% and lifetime prevalence at 79.2%. The retrospective one-year first incidence of LBP in the sample was 8.0%. These data suggest that LBP is common in the Australian population, with four out of five adults experiencing LBP in their life and approximately one in 12 experiencing a new episode of LBP over a 12-month period. A large difference between the point prevalence and the six-month prevalence of LBP in Walker's data is also seen in other epidemiological studies [8] and probably reflects the fluctuating, episodic nature of most LBP. This review did not uncover evidence of gender differences in LBP prevalence in adults sampled from the USA [3] Canada [4], Nordic countries [9] and Australia [7], nor in a Finish sample of children and adolescents [10].
The prevalence of LBP in children is low (1%-6%) [10] but increases rapidly (18%–50%) in the adolescent population [10-12]. The prevalence of LBP peaks around the end of the sixth decade of life. For example, in a prospective 12-month study of 4501 adults in the South Manchester region of the United Kingdom [8], the age distribution of LBP was unimodal, with the peak prevalence occurring in those aged 45 to 59 years old. This is similar to USA epidemiological data describing the peak point prevalence, period prevalence and lifetime prevalence all within ages 55 to 64 years [3]. Though some age-specific back pain cost data show a bimodal distribution with a peak for women over 75 years of age [13], it is likely that this does not represent an increase in the prevalence of non-specific back pain but the prevalence of serious pathology (including compression fracture).
Though LBP treatment and compensation costs have risen markedly over the last three decades [14-16], this may be more the product of case management and cultural attitudes regarding liability and compensation, than changes in either LBP prevalence or LBP activity limitation. There is no compelling biological argument as to why LBP should be increasing in prevalence. Prevalence rates, when measured annually using consistent methods, have shown no change in a Nordic population over a 15-year period [17]. There also is evidence that claim rates for occupational LBP appear to be decreasing in the USA [18], though the relationship of this to prevalence rates is not clear and may also represent an attitudinal change to compensation. Temporal variation in LBP reporting, medical investigation, litigation and compensation may reflect change in societal responses to this common condition rather than any change in LBP prevalence.
Activity limitation (Disability)
In the USA, for people aged 45 years or less, LBP is the most frequent cause of activity limitation [19]. In Walker's data [7], over the previous 6-month period 42.6% of a sample of the Australian adult population reported experiencing low intensity LBP and low associated limitations of activity. A further 10.9% reported experiencing high intensity LBP, but also with low activity limitation. In contrast, an additional 10.5% reported experiencing high intensity LBP with high activity limitation. Though a common problem, it would appear that most LBP in Australia is of low intensity and results in low activity limitation. However, about one in 10 Australian adults have had activity limitation as a result of LBP in the past six months severe enough to result in significant time off from usual activities (Mean time off work = 1.6 months, median 18 days). These data are very similar to the 6-month LBP intensity and activity limitation data of a Canadian adult sample [4]. Though there was no gender difference in prevalence of activity limitation or participation restriction in an Australian LBP sample [7], women were twice as likely to report severe activity limitation in a Canadian sample [4].
Care-seeking
In Walker's data [20], of those Australian adults who experienced LBP over the previous 6-month period, 44.3% sought health care for this condition. This was 28.6% of the total sample. Those seeking care had a greater fear that LBP could impair their life in the future and had higher pain levels than those who did not seek care. Carey et al [21] found that in a sample from North Carolina USA, 61% of recent-onset (<12 weeks) LBP sufferers sought care during their most recent episode. Those seeking care were likely to have more intense pain, leg pain, or a pain onset at work, than those who did not seek care. In a 1995 Australian survey, of those reporting back problems, 46% sought treatment [22]. In summary, about one in two people who experience LBP seek health care during an episode, and they tend to be those experiencing more severe pain, more distal pain, work-related pain or who are more fearful about what the pain might mean.
This review of the LBP epidemiologic evidence found only two studies examining gender differences in care-seeking by those with LBP. In a South Manchester study [8] there was a small gender difference in the frequency of general medical practice consultation for LBP, (mean 7.0% for women, 5.5% for men), but it is unclear whether real gender differences exist or reflect sampling error as the statistical significance of this difference was not reported. However, reinforcing the common perception that women display a greater willingness to seek care for health issues, in an Australian study Walker [20] found women more likely to seek care for LBP (adjusted odds ratio 1.7, 95%CI 1.3 to 2.2).
The most common clinicians consulted for back pain in North America are chiropractors, general medical practitioners and orthopaedists [3,23-25]. In Australia, the most common clinicians consulted for LBP are chiropractors, general medical practitioners, massage therapists, and physiotherapists [20]. People experiencing more severe pain [21,24], who have co-morbidity [24], and women [21] are more likely to consult medical practitioners rather than practitioners in other disciplines.
LBP is a sizeable proportion of casemix for some primary-contact disciplines. Physiotherapy LBP casemix has been estimated to be 25% [26] and 45% [27], depending on the clinical and cultural setting. Chiropractic LBP casemix has been estimated to be 41% in two Australian studies [28,29]. Back pain is the ninth most common presentation in Australian general medical practice [30], contributing between 3.8% [30] and 7.1% [31] of presenting complaints.
Clinicians may choose from a plethora of treatment options, and there are a number of quality evidence-based LBP practice guidelines that can inform those choices [19,32-35]. The extent to which primary-contact practice mirrors recommended practice is unknown [36]. The six most common types of treatment received by Australian adults when seeking care for LBP are back exercises/stretching, massage, spinal manipulation, prescribed medication, non-prescription medication, and bed rest [20]. The lack of knowledge regarding the etiology of most LBP and the lack of a coherent LBP treatment model with cross-discipline acceptance, results in highly varied LBP management strategies being implemented across and within primary-contact disciplines [37-39]. This can result in patient confusion and dissatisfaction [39].
Natural history and clinical course
Von Korff [40] defined natural history as the development of a condition in the absence of treatment, and defines clinical course as its development in the presence of treatment. Studies of the 'natural history' of LBP are potentially compromised by the health care received by any study population, as it is not ethical to prohibit treatment to patients in order to observe the natural history. As there is evidence that specific conservative therapy, (for example, exercise or manipulation [19,33,41,42]) changes the course of an episode of LBP, it is not clear whether studies of the clinical course of people with LBP receiving treatment gives a trustworthy indication of the natural history.
Data describing the clinical course of LBP are also affected by variations in data collection methods, with higher quality studies including independent follow-up for at least 12 months after the onset of a LBP episode. Some reports describe a lack of patient care-seeking from a particular primary-contact practitioner as synonymous with recovery [43], but this approach suffers because people may cease seeking help for a number of reasons. Furthermore, reports of compensation patients, where return-to-work or the ceasing of wage supplementation is the only outcome measure, may not accurately describe the clinical course of LBP in the broader community due to factors affecting reporting, population bias, the complexity of factors that affect return-to-work, and the insensitivity of these outcome measures to LBP recurrence, residual pain and residual activity limitation. Given these considerations, it is reasonable to propose that complete recovery is not synonymous with return-to-work. In addition, up to 60% of injured workers are unable to sustain their initial return-to-work [44], which limits the information about the clinical course of LBP when data collection is confined to initial return-to-work. It is likely that a perspective of LBP derived from research that focuses on the outcome measures of return-to-work and claims management, will be different from a perspective derived from the study of symptom resolution and restoration of all activity (both vocational and non-vocational).
Recent systematic reviews of the clinical course of LBP [45,46] indicate that rapid improvements occur in the first three months post-onset, but that improvements are gradual thereafter. At 6 months post-onset, 16% (range 3–40%) of patients initially off-work remain off-work, and at 12 months post-onset, 62% (range 42–75%) still have pain. Within 12 months of onset, recurrences of both pain (60%, range 44–73%), and recurrences of work absence (33%, range 26–37%) [45] are common.
Ninety percent of the patients who experienced LBP in the South Manchester study [47] ceased consulting their general medical practitioner regarding these symptoms within three months. However, when subsequently interviewed, 79% at three-month follow-up and 75% at 12-month follow-up had not fully recovered (defined as VAS pain score < 2, Hanover Disability Score > 90%). Croft et al [48] recommend revising the view of recent-onset LBP as being self-limiting with only a small proportion that becomes persistent (>12 weeks), to a model of LBP as an essentially persistent condition, characterised by frequent episodes of symptoms interspersed with periods of relative freedom from pain and activity limitation. This recommendation has also been made in other reviews of the clinical course of LBP [34,49,50].
The group of recent-onset LBP patients who remain in intense pain and have substantial activity limitation at 12-months post-onset tend to be the cohort who also remain off-work at that time. However, Watson et al. [51] found that 12-months post-onset, whereas only 0.65% of those experiencing first-onset LBP were still off-work, 4.5% of those who were experiencing recurrences of pre-existing LBP still remained off-work. Recurrence therefore appears to increase the risk of not returning to work (relative risk 6.9). Studies from a number of national and vocational settings indicate that the longer workers remain off-work the lower the probability of them ever returning to work [50].
Although patients with persistent LBP are commonly thought to have a poor prognosis, there are few data describing their long-term outcomes. A Dutch group of patients with persistent LBP were followed for seven years and measures of pain, activity limitation, spinal mobility, and movement-related pain were repeatedly recorded. At the beginning of the study, the mean duration of back pain for the group was 5.4 years (SD 3.6). At three years post-initial measurement (n = 31), statistically significant improvements were found in pain and activity limitation scores, while lumbar spine mobility decreased [52]. At seven years post-initial measurement (n = 22), spinal mobility was unchanged from the three-year level, but further statistically significant improvements in activity limitation and movement-related pain had occurred [53]. These data suggest that once established, persistent LBP does not lead to progressive increases in pain and progressive increases in activity limitation. However, the mean scores for the variables measured were around 50% at the beginning of the study and did not improve over the study period by more than 15%. These data encourage the hypothesis that persistent LBP tends to stabilise and improve a little and slowly in the long-term. Data were obtained from a small sample and the hypothesis warrants testing on a larger sample.
A clinical feature of LBP and a dilemma for LBP research measurement is the recurrent, episodic nature of LBP, as it confounds conclusions based on measurements taken at a set point in time. This has led to recommendations that instead of data indicating numbers remaining off-work at a set point in time, such as 12-months after onset, measures such as total number of days off-work over a 12-month period may be more informative. The same principle can be applied to other dimensions of the LBP experience, for example, measuring the number of days in pain over a period, instead of those still in pain at the end of the period [54]. This fluctuating clinical course of LBP with incomplete resolution has led some authors to suggest that the distinction between acute (recent-onset) and chronic (persistent) LBP is clinically irrelevant [55]. In summary, the clinical course of recent-onset LBP is that patients are likely to recover from their presenting episode, most will still have some symptoms at 12 months, many will experience relapses, and a few will not improve much at all despite treatment.
Treatment outcomes
There are now many randomised controlled trials (RCT) of interventions in both recent-onset and persistent LBP. These trials vary greatly in subject inclusion/exclusion criteria, outcome measures, blinding, concealment, analysis techniques and other research design features. This diversity, combined with the poor quality of many RCTs, has made data synthesis difficult, and resulted in few meta-analyses. Most synthesis of LBP intervention data has been via systematic review. Systematic reviews also vary in methodological quality and in the papers selected for inclusion. Furthermore, even reviews that broadly cover the same literature are subject to author interpretation, and many reach conflicting conclusions regarding intervention effectiveness [56,57]. Reviews with higher methodological rigour tend to report more negative or uncertain conclusions about the effects of interventions for LBP [58].
There are a number of exhaustive reviews of the efficacy of interventions in recent-onset LBP [19,33,34,42,59]. There are also a number of national clinical guidelines for the management of LBP that have been based on comprehensive literature searches [19,33,34,59-66]. Their recommendations regarding positive interventions for recent-onset LBP can be summarised as: patient education and reassurance, medication (Paracetomol, NSAIDs, muscle relaxants, opioids), some forms of exercise, manual therapy (manipulation, mobilisation), and discouragement of bed rest [36].
In a study of reviews of conservative treatment for persistent LBP, Furlan et. al. [57], summarised the results of 109 systematic reviews. The interventions included medication (analgesics, antidepressants, epidural and facet injections, muscle relaxants, NSAIDs, and opioids), education/behavioural (back schools, bed rest, cognitive/behaviour, couple therapy, multidisciplinary teams), and physical treatments (acupuncture, exercise, laser, orthoses, spinal manipulation, TENS, traction). The summaries produced mostly negative or conflicting findings. They concluded that the only interventions associated with positive patient outcomes were muscle relaxants, opioids, and interventions provided by multidisciplinary teams.
LBP costs
The direct financial costs of back pain are health care costs, and indirect costs are production losses to industry and injury impact on insurance costs. Estimates of the indirect costs vary depending on the econometric model chosen. Annual back pain costs have been estimated for Australia [67], the United Kingdom [68] and USA [14], and are summarised in Table 1. Across these countries, the direct costs of back pain represent between 0.19% and 0.42% of GDP, and between 1.65% and 3.22% of all health expenditure.
Table 1 Annual back pain financial costs.
Annual back pain costs
Direct costs (millons) Allied health costs (millons) Indirect costs# (millons) GDP* Health expenditure
Australia (1991) AU$1,020 0.22% 1.65%
AU$724 0.16% 1.17%
AU$2,000–$8,000 0.43%–1.72%
United Kingdom (1998) £1,632,000 0.19% 2.78%
£1,068,000–£5,018,000 0.12%–0.58%
USA (1990) US$24,300 0.42% 3.22%
#Estimates of indirect costs vary depending on the econometric models used.
*GDP = Gross Domestic Product (for year of study).
During 1993/4, in an Australian population of 19.5 million people, there were 3.6 million medical consultations and 2.9 million prescriptions for back pain [13]. However, across the countries in which it has been studied, the majority of compensable LBP costs are generated by a small proportion of claimants. For example, data from the Quebec Workers Compensation System showed that the 8% of claimants who were absent from work for more than six months were responsible for 73% of the medical costs, and 76% of the compensation costs [69].
Direct costs to the health care and compensation systems, and indirect costs to industry do not include the non-financial costs to the patient and his/her family. These non-financial costs include lost participation in domestic, family, and social activities.
Conclusion
LBP is a common problem affecting both genders and most ages, for which about one in four adults seeks care in a six-month period. It results in considerable direct and indirect costs, and these costs are financial, workforce and social. Care-seeking behaviour varies depending on cultural factors, the intensity of the pain, the extent of activity limitation and the presence of co-morbidity. Care-seeking for LBP is a significant proportion of caseload for some primary-contact disciplines. Most recent-onset LBP episodes settle but only about one in three resolves completely over a 12-month period. About three in five will recur in an on-going relapsing pattern and about one in 10 does not resolve at all. The cases that do not resolve at all form a persistent LBP group that consume the bulk of LBP compensable care resources and for whom positive outcomes are possible but not frequent or substantial.
Authors' contributions
PMK conceived of the study, participated in its design, located and selected studies, extracted and interpreted the data, wrote the paper, and approved the final manuscript. JLK conceived of the study, participated in its design, interpreted the data, and revised and approved the final manuscript.
Supplementary Material
Additional File 1
It contains the strategy used for the OVID Medline Search.
Click here for file
Acknowledgements
Supported by Faculty of Health Sciences (La Trobe University), Joint Coal Board Health & Safety Trust (Australia), Musculoskeletal Physiotherapy Association (Victoria).
==== Refs
Leboeuf-Yde C Lauritsen J The prevalence of low back pain in the literature - a structured review of 26 Nordic studies from 1954-1993 Spine 1995 20 2112 2118 8588168
Loney P Stratford P The prevalence of low back pain in adults: A methodological review of the literature Physical Therapy 1999 79 384 396 10201544
Deyo R Tsui-Wu Y Descriptive epidemiology of low-back pain and its related medical care in the United States Spine 1987 12 264 268 2954221
Cassidy D Carroll L Cote P The Saskatchewan Health and Back Pain Survey - The prevalence of low back pain and related disability in Saskatchewan adults Spine 1998 23 1860 1867 9762743 10.1097/00007632-199809010-00012
Volinn E The epidemiology of low back pain in the rest of the world. A review of surveys in low- and middle-income countries Spine 1997 22 1747 1754 9259786 10.1097/00007632-199708010-00013
Walker BF The prevalence of low back pain: a systematic review of the literature from 1966 to 1998 Journal of Spinal Disorders 2000 13 205 217 10872758 10.1097/00002517-200006000-00003
Walker BF Muller R Grant WD Low back pain in Australian adults. Prevalence and associated disability Journal of Manipulative & Physiological Therapeutics 2004 27 238 244 15148462 10.1016/j.jmpt.2004.02.002
Papageorgiou A Croft P Ferry S Jayson M Silman A Estimating the prevalence of low back pain in the general population - Evidence from the South Mancester back pain survey Spine 1995 20 1889 1894 8560337
Leboeuf-Yde C Kloughart N Lauritzen T How common is low back pain in the Nordic population? - Data from a recent study on a middle-aged general Danish population and four surveys previously conducted in Nordic countries Spine 1996 21 1518 1526 8817778 10.1097/00007632-199607010-00005
Taimela S Kujala U Salminen J Viljanen T The prevalence of low back pain among children and adolescents - A nation-wide, cohort based questionnaire survey in Finland Spine 1997 22 1132 1136 9160472 10.1097/00007632-199705150-00013
McMeeken J Tully E Stillman B Nattrass CL Bygott IL Story I The experience of back pain in young Australians Manual Therapy 2001 6 213 220 11673931 10.1054/math.2001.0410
Leboeuf-Yde C Kyvik K At what age does low back pain become a common problem? Spine 1998 23 228 234 9474731 10.1097/00007632-199801150-00015
Mathers C Penm R Health system costs of injury, poisoning and musculoskeletal disorders in Australia 1993-94 1999 Canberra , University of Sydney and the Australian Institute of Health and Welfare
Frymoyer J Cats-Baril W An over-view of the incidences and costs of low back pain Orthop Clin North Am 1991 22 263 271 1826550
Nachemson A White A The natural course of low back pain American Academy of Orthopaedic Surgeons Symposium on Idiopathic Low Back Pain 1982 St.Louis , CV Mosby 46 51
Waddell G A new clinical model for the treatment of low-back pain Spine 1987 12 632 644 2961080
Leino PL Berg MA Puscha P Is back pain increasing? Results from national surveys in Finland. Scand J Rheumatol 1994 23 269 276 7973482
Murphy PL Volinn E Is occupational low back pain on the rise? Spine 1999 24 691 697 10209800 10.1097/00007632-199904010-00015
Bigos S Bowyer O Braen G Acute low back problems in adults. Clinical practice guidelines No. 14. 1994 AHCPR Publication No.95-0642. Rockville, MD:Agency for Health Care Policy and Research, Public Health Service, U.S. Department of Health and Human Servies
Walker BF Muller R Grant WD Low back pain in Australian adults. Health provider utilisation and care seeking Journal of Manipulative & Physiological Therapeutics 2004 27 327 325 15195040 10.1016/j.jmpt.2004.04.006
Carey TS Evans AT Hadler NM Lieberman G Kalsbeek WD Jackman AM Fryer JG NcNutt RA Acute severe low back pain: a population-based study of prevalence and care-seeking Spine 1996 21 339 344 8742211 10.1097/00007632-199602010-00018
Britt H Miller GJ Knox S Imaging orders by General Practitioners in Australia 1999-2000 2001 Canberra , University of Sydney and the Australian Institute of Health and Welfare
Cote P Cassidy JD Carroll L The treatment of neck and low back pain: Who seeks care? Who goes where? Medical Care 2001 39 956 967 11502953 10.1097/00005650-200109000-00006
Hurwitz EL Morgenstern H The effects of comorbidity and other factors on medical versus chiropractic care for back problems Spine 1997 22 2254 2264 9346146 10.1097/00007632-199710010-00011
Shekelle PG Markovich M Louie R Comparing the costs between provider types of episodes of pain care Spine 1995 20 221 226 7716629
Jette AM Smith K Haley SM Davis KD Physical therapy episodes of care for patients with low back pain Physical Therapy 1994 74 101 115 8290616
Battié MC Cherkin DC Dunn R Ciol MA Wheeler KJ Managing low back pain: attitudes and treatment preferences of physical therapists Phys Ther 1994 74 219 226 8115455
Ebrall PS A descriptive report of the case-mix within Australian Chiropractic practice, 1992 Chiropractic Journal of Australia 1993 23 92 97
Walsh MJ Jamison JR A comparison of patients and patient complaints at Chiropractic teaching clinics and private clinics Chiropractic Journal of Australia 1992 22 87 91
Britt H Miller GJ Knox S Charles J Valenti L Henderson J Kelly Z Pan Y General practice activity in Australia 2000-01 2001 Canberra , University of Sydney and the Australian Institute of Health and Welfare
Sayer G Britt H Horn F Bhasale A McGreechan K Charles J Miller GJ Hull B Scahill S Measures of health and health care delivery in general practice in Australia 2000 Canberra , University of Sydney and the Australian Institute of Health and Welfare
Australian Acute Musculoskeletal Pain Guidelines Group NHMRC Evidence-based management of acute musculoskeletal pain: A guide for clinicians 2004 1st Bowen Hills , Australian Academic Press 66
RCGP Clinical guidelines for the management of acute low back pain 1999 London, England , Royal College of General Practitioners
Bogduk N Evidence-based guidelines for the management of acute low back pain.
Accident Compensation Commission New Zealand Acute Low Back Pain Guide 1999 Wellington
Koes B van Tulder M Ostelo R Burton A Waddel G Clinical guidelines for the management of low back pain in primary care Spine 2001 26 2504 2514 11707719 10.1097/00007632-200111150-00022
Cherkin DC Deyo RA Wheeler K Ciol MA Carey T Physician views about treating low back pain: The results of a national survey Spine 1995 20 1 10 7709266
Deyo RA Practice variations, treatment fads, rising disability. Do we need a new clinical research paradigm? Spine 1993 18 2153 2162 8278825
Waddell G Low back pain: a twentieth century health care enigma Spine 1996 21 2820 2825 9112705 10.1097/00007632-199612150-00002
Von Korff M Studying the natural history of back pain Spine 1994 19 2041S 2046S 7801181
Loisel P Abenhaim L Durand P Esdaile JM Suissa S Gosselin L Simard R Turcotte J Lemaire J A population-based, randomized clinical trial on back pain management Spine 1997 22 2911 2918 9431627 10.1097/00007632-199712150-00014
Maher C Latimer J Refshauge K Efficacy of conservative treatments for acute, sub acute and chronic non-specific low back pain and for the prevention of non-specific low back pain. 1998 Melbourne , Australian Physiotherapy Association and the Manipulative Physiotherapy Association of Australia
Dillane JB Fry J Kalton G Acute back syndrome: A study from general practice British Medical Journal 1966 2 82 84
Baldwin ML Johnson WG Butler RJ The error of using returns-to-work to measure the outcomes of health care American Journal of Industrial Medicine 1996 29 632 641 8773723 10.1002/(SICI)1097-0274(199606)29:6<632::AID-AJIM7>3.0.CO;2-L
Hestbaek L Leboeuf C Manniche C Low back pain: what is the long-term course? A review of studies of general patient populations European Spine Journal 2003 12 149 165 12709853
Pengel LH Acute low back pain: systematic review of its prognosis BMJ 2003 327 9 10.1136/bmj.327.7410.323
Croft P Outcome of low back pain in general practice: a prospective study BMJ 1998 316 1356 1359 9563990
Croft P Macfallane GJ Papageorgiou AC Thomas E Silman AJ Outcome of low back pain in general practice: a prospective study. BMJ 1998 316 1356 1359 9563990
Von Korff M Saunders K The course of back pain in primary care Spine 1996 21 2833 2837 9112707 10.1097/00007632-199612150-00004
Waddell G The Back pain revolution 1998 Edinburgh ; New York , Churchill Livingstone xi, 438
Watson PJ Main CJ Waddell G Gales TF Purcell Jones G Medically certified work loss, recurrence and costs of wage compensation for back pain: a follow-up study of the working population of Jersey Br J Rheumatol 1998 37 82 86 9487255 10.1093/rheumatology/37.1.82
Lankhorst G Van de Stadt R Van der Korst J The natural history of idiopathic low back pain - A three-year follow-up study of spinal motion, pain and functional capacity Scandinavian Journal of Rehabilitation Medicine 1985 17 1 4 3159080
Lankhorst GJ Van de Stadt RJ Van Arkel EC The long-term natural history of idiopathic low back pain. A seven-year prospective study of pain, functional abilities and spinal motion Journal of Rehabilitation Sciences 1991 4 7 8
Von Korff M Effects of practice style in managing back pain Annals of Internal Medicine 1994 121 187 195 8017745
Leboeuf-Yde C Grønstvedt A Borge J Lothe J Magnesen E Nilsson O Røsok G Stig L Larsen K The Nordic Back Pain Subpopulation Program: Demographic and clinical predictors for outcome in patients receiving chiropractic treatment for persistent low-back pain J Manipulative Physiol Ther 2004 27 493 502 15510092 10.1016/j.jmpt.2004.08.001
Ferreira PH Ferreira ML Maher CG Latimer J Herbert RD Refshauge K Magarey ME Assessing clinical evidence in pain: How robust are conclusions of low back systematic reviews?: 21-24 November 2001; Adelaide Convention Centre, Adelaide, South Australia.
2001
Furlan AD Clarke J Esmail R Sinclair S Irvin E Bombardier C A critical review of reviews on the treatment of chronic low back pain Spine 2001 26 E155 E162 11295917 10.1097/00007632-200104010-00018
Petticrew M Why certain systematic reveiws reach uncertain conclusions British Medical Journal 2003 326 756 758 12676848
Waddell G Burton AK Occupational health guidelines for the management of low back pain at work - evidence review. 2000 London. , Faculty of Occupational Medicine.
ACC Accident Compensation Commission; 1999 New Zealand Acute Low Back Pain Guide. New Zealand Acute Low Back Pain Guide Wellington, New Zealand
Borkan J Reis S Werner S al. Guidelines for treating low back pain in primary care 1996 Harfuah , The Israeli Low Back Pain Guideline Group 145 151 8682388
Faas A Chavannes AW Koes BW al. NHG-Standaard Lage-Rugpijn. Huisarts Wet 1996 39 18 31
Handlungssleitlinie-Ruckenschmerzen Treatment guideline - backache. - quoted in Koes et al 2001 Clinical guidelines for the management of low back pain in primary care. Spine 26 (22) 2504-2514 Z Artztl Fortbild Qualitatssich 1997 91 457 460
Keel P Weber M Roux EB al. Kreuzschmerzen: Hintergrunde, pravention, behandlung. Basisdokumentation. - quoted in Koes et al 2001 Clinical guidelines for the management of low back pain in primary care. Spine 26 (22) 2504-2514 1998 Bern , Verbingdung der Schweizer Arzte (FMK)
Malmivaara A Kotilainen E Laasonen EM al. Clinical practice guidelines of the Finnish Medical Association Duodecim. Diseases of the low back 1999 Finnish Medical Association
Manniche C Low back pain: frequency, management and prevention from HTA perspective 1999 Danish Institute for Health technology Assessment
Walker BF Muller R Grant WD Low back pain in Australian adults: The economic burden Asia Pac J Public Health 2003 15 79 87 15038680
Maniadakis N Gray A The economic burden of back pain in the UK Pain 2000 84 95 103 10601677 10.1016/S0304-3959(99)00187-6
Abenhaim L Suissa S Importance and economic burden of occupational back pain: A study of 2,500 cases representative of Quebec Journal of Occupational Medicine 1987 29 670 674 2958608
|
16045795
|
PMC1208926
|
CC BY
|
2021-01-04 16:38:23
|
no
|
Chiropr Osteopat. 2005 Jul 26; 13:13
|
utf-8
|
Chiropr Osteopat
| 2,005 |
10.1186/1746-1340-13-13
|
oa_comm
|
==== Front
Chiropr OsteopatChiropractic & Osteopathy1746-1340BioMed Central London 1746-1340-13-171609295510.1186/1746-1340-13-17DebateSubluxation: dogma or science? Keating Joseph C [email protected] Keith H [email protected] Jaroslaw P [email protected] Stephen M [email protected] David [email protected] James F [email protected] 6135 North Central Avenue, Phoenix, AZ, 85012, USA2 School of Medicine, Mayne Medical School, University of Queensland, Herston, Queensland 4006, Australia3 Department of Graduate Education and Research, Canadian Memorial Chiropractic College, 6100 Leslie Street, Toronto ON, M2H 3J1, Canada4 Department of Clinical Sciences, College of Chiropractic, University of Bridgeport, 225 Myrtle Ave., Bridgeport, CT 06604, USA5 Department of Chiropractic Procedures, Southern California University of Health Sciences, 16200 E. Amber Valley Drive, Whittier, CA 90604, USA6 President, National University of Health Sciences, 200 East Roosevelt Road, Lombard, IL 60148, USA2005 10 8 2005 13 17 17 25 5 2005 10 8 2005 Copyright © 2005 Keating et al; licensee BioMed Central Ltd.2005Keating et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Subluxation syndrome is a legitimate, potentially testable, theoretical construct for which there is little experimental evidence. Acceptable as hypothesis, the widespread assertion of the clinical meaningfulness of this notion brings ridicule from the scientific and health care communities and confusion within the chiropractic profession. We believe that an evidence-orientation among chiropractors requires that we distinguish between subluxation dogma vs. subluxation as the potential focus of clinical research. We lament efforts to generate unity within the profession through consensus statements concerning subluxation dogma, and believe that cultural authority will continue to elude us so long as we assert dogma as though it were validated clinical theory.
==== Body
Background
Status of a Construct
More than twenty years ago Donald K. Moon, D.C. wrote of a "flight from the subluxation" among chiropractors [1]. Dr. Moon, a firm believer in the validity of the traditional chiropractic lesion, bemoaned the dearth of scientific data to substantiate the construct, and warned of the possibility that medical researchers would step in to fill the void created by chiropractors' indolence. He decried the tendency among many chiropractors to pit diagnosis against spinal analysis (i.e., subluxation-detection), as though the two were mutually exclusive.
In the years since, some members of the profession have developed scientific skills, and a literature bearing on the usefulness of spinal manipulation, generated by chiropractors and others, has evolved [e.g., [2-5]]. In the United States several chiropractic colleges have been the recipients of federal funds for scientific investigations, and a consortial center for investigations has been established at Palmer College of Chiropractic with federal money. University-based chiropractic schools have been established in several nations [6], and the scholarly works of chiropractors are now much more widely disseminated in chiropractic and non-chiropractic periodicals. The profession may look upon these developments and say with some pride that, indeed, there is a small but meaningful scientific literature in chiropractic [7,8].
Despite these accomplishments, many chiropractors preeminent theoretical construct remains unsubstantiated [9-11], and largely untested [12]. This lack of evidence may reflect a lack of interest among those with research skills; Nelson [11] observed that "clinical studies of the effectiveness of spinal manipulation are conducted and reported without reference to the presence or absence or even the existence of subluxations". The chiropractic subluxation stands pretty much today as it did at the dawn of the 20th century: an interesting notion without validation. And, as it has throughout the past century, D.D. Palmer's mediating variable remains a "bone of contention" between many chiropractors and the scientific community, as well as among chiropractors themselves.
Although books and monographs have been written about the presumed entity [e.g., [13-16]], and intra-professional political consensuses [17-19] have been reached on fuzzy conceptual definitions and unjustified claims (Table 1), little if any substantive experimental evidence for any operational definition of the chiropractic lesion has been offered in clinical trials. Notwithstanding strong intra-professional commitment to the subluxation construct [20,21] and reimbursement strategies that are legally based upon subluxation [22], there is today no scientific "gold standard" (10) for detecting these reputedly ubiquitous and supposedly significant clinical entities, and inadequate basic science data to illuminate the phenomenon [11,23]. The chiropractic subluxation continues to have as much or more political than scientific meaning [24].
Table 1 Assertions about subluxation offered by several chiropractic organizations [17-19]
Association of Chiropractic Colleges
4.0 The Subluxation
Chiropractic is concerned with the preservation and restoration of health, and focuses particular attention on the subluxation.
A subluxation is a complex of functional and/or structural and/or pathological articular changes that compromise neural integrity and may influence organ system function and general health.
A subluxation is evaluated, diagnosed, and managed through the use of chiropractic procedures based on the best available rational and empirical evidence.
Chiropractic Association of Australia
...We recognise and respect a universal intelligence (or order) in all matter and an innate intelligence within a living organism that strives to preserve life and, if uninhibited, will express optimal well being.
We recognise that the practice of chiropractic focuses on the relationship between structure (primarily the spine) and function (as coordinated by the nervous system) and how that relationship affects the preservation and restoration of health.
We recognise that subluxations compromise the expression of innate intelligence, and that prevention and removal of subluxations will facilitate the expression of optimal health.
We respect, care about and are committed to the individual's holistic well being and emphasise the inherent recuperative power of the body to heal itself without the use of drugs or surgery.
We respect and value the importance of intellectual honesty, scientific and academic excellence and the maintenance of integrity in serving the individual, the community and the profession.
New Zealand Chiropractors' Association
Chiropractors use a technique of correcting vertebral subluxations called an adjustment. An adjustment is a carefully executed manoeuvre that usually results in a joint clicking as a sticky joint is released. Adjustments are usually painless, and enjoyable, because the improved mobility is usually immediately noticeable, and the health benefits are noticed soon after.
Benefits of Chiropractic Care
Feel Great Relief from Pain Improves Immunity Restores Nerve Supply More Energy Restores Mobility Improves Athletic Performance More efficient Body Function Allows Better Sleep Back to Work Faster Improves Posture No Drugs Slows the Aging Process No Surgery Quicker Recovery No Needles Add Life to Years Add Years to Life
How It Works
Chiropractic is based on the scientific fact that your nervous system controls the function of virtually every cell, tissue, organ and system of your body. While the brain is protected by the skull, the spinal cord is more vulnerable, covered by 24 moving vertebrae. When these bones lose their normal motion or position, they can irritate the nervous system. This disrupts the function of the tissues or organs these nerves control, and this is called vertebral subluxation complex.
Chiropractic is the science of locating these areas of spinal malfunction and the art of correcting them to allow the body to heal itself. As we all know, regardless of which type of doctor you consult, only the body can heal itself.
We believe that Dr. Moon's concerns were only partly justified. All in all, there has been no flight from the subluxation on the part of the field or its leaders [e.g., [25,26]], nor much move towards it either (on the part of the profession's scholars) [12]. The profession – its rank-and-file and political leadership (see Figure 1) – has not abandoned the subluxation as an a priori principle guiding many of its activities. The chiropractic subluxation and subluxation-related beliefs permeate the practice of chiropractic, the marketing rhetoric offered by many chiropractors, the legal and political strategies pursued by various trade associations, and the sense of identity for many in the profession... all this for a hypothetical construct whose relevance for health and illness has yet to be established.
Figure 1 Political statement rendered on a button by the American Chiropractic Association, 2003.
The traditional chiropractic lesion has not been the focus of systematic clinical research for the purpose of determining its meaningfulness (or lack thereof). In the absence of scientific validation, the propagation of unsubstantiated claims for many chiropractors favorite mediating variable is an obstacle to scientific credibility and cultural authority for the profession. It is our purpose to remind the profession of the implications and consequences of offering subluxation dogmatically, rather than as a plausible and testable proposition.
Discussion
The Dogma of Subluxation
The spinal subluxation, though we have been correcting it with spinal adjustment for 100 years, is not fully understood. Scientific research presently is not sophisticated enough to determine the neurophysiological impact that spinal subluxation has on our patients. Does that mean that we do not adjust our patients because it has not been proven? Absolutely not. I treat my patients as if each spinal adjustment has a virtually unlimited potential in improving their health... [27].
So wrote a member of the American Chiropractic Association's (ACA's) governing board in the centennial year of the profession. We might applaud the good doctor for acknowledging the inadequacy of basic research bearing on the subluxation; on the other hand, no recognition is given that the clinical meaningfulness of subluxation has yet to be established. One can only speculate what it means to treat every patient "as if each spinal adjustment has a virtually unlimited potential."
The dogmatic character of subluxation beliefs is exemplified by several assertions offered by the Association of Chiropractic Colleges (ACC) (see Table 1). Intended as a means of fostering greater unity among the chiropractic colleges, the ACC's "Paradigm" statement on subluxations has since been widely endorsed by national and international membership societies [28]. In effect, the ACC Paradigm has become the standard (if not official) position of a broad segment of the profession. There are several problems with the Paradigm.
First, the hypothesis that subluxation is some "complex of functional and/or structural and/or pathological articular changes that compromise neural integrity" is offered without qualification, that is, without mention of the tentative, largely untested quality of this claim. (As well, a stubbed toe would seem to meet the fuzzy criteria provided by the ACC.) The nature of the supposed compromise of "neural integrity" is unmentioned.
Secondly, the dogmatism of the ACC's unsubstantiated claim that subluxations "may influence organ system function and general health" is not spared by the qualifier "may." The phrase could mean that subluxations influence "organ system function and general health" in some but not all cases, or that subluxation may not have any health consequences. Although the latter interpretation is tantamount to acknowledging the hypothetical status of subluxation's putative effects, this meaning seems unlikely in light of the ACC's statement that chiropractic addresses the "preservation and restoration of health" through its focus on subluxation. Both interpretations beg the scientific questions: do subluxation and its correction "influence organ system function and general health"?
Lastly, the ACC claims that chiropractors use the "best available rational and empirical evidence" to detect and correct subluxations. This strikes us as pseudoscience, since the ACC does not offer any evidence for the assertions they make, and since the sum of all the evidence that we are aware of does not permit a conclusion about the clinical meaningfulness of subluxation. To the best of our knowledge, the available literature does not point to any preferred method of subluxation detection and correction, nor to any clinically practical method of quantifying compromised "neural integrity," nor to any health benefit likely to result from subluxation correction.
All in all, the ambiguities that permeate the ACC's statements on subluxation render it inadequate as a guide to clinical research. Although Wenban [29] proposes that the ACC statements on subluxation might be construed as "a very simplified map, for starting to find the future practice-relevant research priorities for chiropractic," he offers no suggestion that ACC's "map" is any improvement upon existing proposals for subluxation research strategies [e.g., [10,23,30]]. Owens [31] suggests that consensus models of subluxation are "useless for research purposes." Concerning the ACC's statements about subluxation, a signatory to the document asserts, "This paradigm was never intended to be a testable research hypothesis. It was constructed by a process of consensus to serve as a collective political statement, not a research hypothesis" [32]. More to the point of research need is a validated operational definition of subluxation [31]. Nelson [33] advises that "Whether chiropractors are actually treating lesions, or not, is a question of immense clinical and professional consequence. Resolution of the controversy will not be found through consensus panels nor through semantic tinkering, but through proposing and testing relevant hypotheses."
Whether the ACC's subluxation claims have succeeded as a political statement is beyond our concern here. These assertions were published as a priori truths (what many chiropractors have traditionally referred to as "principle"), and are exemplary of scientifically unjustified assertions made in many corners of the profession [34-36]. It matters not whether unsubstantiated assertions are offered for clinical, political, scientific, educational, marketing or other purposes; when offered without acknowledgment of their tentative character, they amount to dogmatism.
We contend that attempts to foster unity (among the schools or in the wider profession) at the expense of scientific integrity is ultimately self-defeating. To be sure, the profession's lack of cultural authority is based in part upon our characteristic disunity. However, attempts to generate unity by adoption of a common dogma can only bring scorn and continued alienation from the wider health care community and the public we all serve.
Subluxation Semantics
The subluxation is identified by a great many names [37], but neither the abundance of labels nor efforts to reach consensus on terminology tell us anything about the validity of the construct. Nelson [11] points out that "...framing the subluxation debate as a semantic issue, resolvable by consensus, is precisely the same as asking whether we should refer to the spaceships used by aliens as flying saucers or UFOs." Neither adoption nor rejection of the term subluxation or any of its myriad synonyms will resolve the problem created by assuming a priori that subluxation is clinically meaningful. If and when we demonstrate that there are alien spaceships hovering over us, we suspect an appropriate terminology will develop on its own.
The clinical meaningfulness, if any, of subluxation cannot be established by definition. The notion that subluxation is inherently pathological, perhaps because some dictionary equates subluxation with ligamentous sprain, does not mean that joint dysfunction merits clinical intervention. Skin tags too might be considered pathological, but the mere presence of aberration or abnormality does not indicate a serious or treatment-worthy health problem. (The unfortunate lesson of decades of surgical intervention for bulging discs, performed in the hope of relieving back pain, seems all too frequently lost on many chiropractors.) We cannot establish the clinical meaningfulness of subluxation merely by branding it pathological; such would be word magic.
This is not to say that efforts to develop a standardized lexicon among chiropractors [e.g., [38]] are without merit. We think it important and useful, for example, to distinguish between the "orthopedic subluxation" [39] vs. "subluxation syndrome" [38]. The former is a more or less observable phenomenon recognized within and beyond chiropractic's borders. The latter is a theoretical notion, which relates subluxation of joints to deleterious health consequences, and is a testable, but largely untested proposition. This is no small distinction.
Subluxation in Practice
As a pragmatic matter, subluxation refers to the target of many chiropractors manual interventions, and the individual practitioner may select from a range of theories, techniques and supposed clinical implications of the traditional chiropractic lesion. The latter include subluxation as a cause of musculoskeletal problems, as an etiological factor in various internal disorders and behavioral/psychological problems, and as a strategic intervention site for disease prevention and wellness enhancement. Hundreds of brand-name techniques have been offered for the purpose of correcting subluxations [13], but the clinical usefulness of subluxation correction has yet to be experimentally demonstrated.
The diversity of altered function attributed to subluxation and "nerve interference" parallels in some respects the "nervism" [40] and "spinal irritation" [41] of nineteenth century neurology and physiology. When coupled with vitalistic concepts of "Innate Intelligence," subluxation theories expand upon the "nature-trusting heresy" [42] of those earlier times. Unlike the therapeutic nihilism recommended by some nineteenth century physicians, many chiropractors' faith in nature gives rise to extensive regimens of subluxation correction [43]. The breadth of contemporary, uncritical speculations bearing on subluxation is captured in the boast of a chiropractic leader: "Rigor mortis is the only thing we can't help" [44]. Seaman [45] argues that "many chiropractic practices are guided by dogmatism instead of philosophy and science." In short, many chiropractors practice as though subluxation is clinically relevant, but seemingly without recognition that maybe it's not. When challenged, many chiropractors respond not with data, but by avowing "the chiropractic principle": subluxation.
The National Board of Chiropractic Examiners offers that: "By manually manipulating vertebrae into their normal physiological relationship, chiropractic practitioners relieve interference with the nervous system along with accompanying symptoms. This correction of joint dysfunction reestablishes normal mobility and comfort... Chiropractors see patients with spinal subluxations and joint dysfunction on a daily basis..." [[46], pp. 2, 53]. Chiropractors list "spinal subluxation/joint dysfunction" as the most frequent of all "conditions" they encounter [[46], pp. 53, 84, 101].
The magic and mystery of subluxation theories all too frequently direct the chiropractor's attention away from the legitimate question of whether subluxation (or any other rationale for manipulation) may be relevant in a patient's health problem, to a search for the "right" vertebra. Individual clinicians derive subluxation theories about particular spinal regions as "keys" to better health or to the resolution of particular disorders. For example, the subluxation sites for which adjustment has been suggested to relieve enuresis range from heads to tails [47-56]. Disciples of B.J. Palmer often restrict themselves to the upper cervical spine, while adherents to Logan's Basic Technique tend to focus on the sacrum. Sacro-occipital technique practitioners work at both ends of the spine. The problem is not the fertile diversity of subluxation hypotheses, but rather that the possible irrelevance of subluxation and adjustment is so infrequently addressed [e.g., [55,56]]. Many chiropractors (and others) have often been more disposed to ask where the subluxation is rather than whether subluxation correction is relevant or warranted.
The popularity of the subluxation construct is reflected in the variety of brand-name clinical techniques vended in the profession [e.g., [57-60]], many of which concern methods of subluxation detection and correction (see Table 2). We propose that the ubiquity and commercial success of these clinical procedures speak to the credence those doctors of chiropractic place in the various iterations of subluxation theories. Comparable claims for the clinical meaningfulness of subluxation may be found at the websites of several chiropractic colleges [36] and in the patient brochures distributed by major provincial, state and national membership societies of chiropractors in Canada and the United States (34). Many chiropractors bombard themselves and the public with subluxation rhetoric, but rarely hint at the investigational status of this cherished idea.
Table 2 Assertions about subluxation made by several brand-name technique organizations of chiropractic
...The mirror image adjustment resets the proprioceptive reflexes, inhibits the nocioceptive impulses and corrects the abnormal loading setting up the subluxation. In so doing, the reflex response of vasoconstriction to the viscera is removed and improved vascular tone to the smooth muscle, cardiac muscle and glands, results. The history of chiropractic success with patients experiencing such conditions as asthma, angina, visual disturbances, and other visceral conditions, is now clearly understood [57]
...D.N.F.T. utilizes a diagnostic system for subluxation analysis consisting of gentle challenging and a unique leg check. This testing allows the body itself to indicate the directions of misalignment of structures that are producing nerve interference. A gentle but directionally specific thumb impulse provides a long lasting correction to bony and soft tissue structures. D.N.F.T. is able to achieve structural corrections without torqueing, strong thrusts, and associated articular sounds that are often associated with traditional chiropractic...
The goals of Directional Non-Force Technique are very much in line with the roots of traditional chiropractic: analyze and correct subluxations wherever they occur in the body, and allow the body to heal itself. Subluxations, as defined in Directional Non-Force Technique, are misalignments of tissue, osseous or soft, which result in nerve interference... [58].
...Minor displacements of the spinal bones, known as vertebral subluxations, can cause endangering stress to the spinal cord which acts as the main line of intelligence for the whole body. These displacements, or subluxations, are the cause of many of the unwanted health conditions that people suffer from every day. Although there have been many valuable techniques that have been developed in the chiropractic profession, the Gonstead System is considered a "gold standard" for chiropractic techniques because of its record of safety and effectiveness in correcting vertebral subluxation... [59].
...When the spinal column is in proper alignment, the "Brain Stem" can pass unimpinged through this foramen. But when one or both of the top two vertebrae become misaligned, the "Brain Stem" is impinged and normal nerve supply is reduced to parts of the body served by that nerve tract, hence sickness and disease... [60].
It has been our informal experience that subluxation is an unchallenged notion for many in the profession; Clum [39] concurs. Among the likely consequences of this unskeptical acceptance are evaluations and interventions that fail to address outcomes (in favor of focus on the presumed mediator: subluxation), excessive treatment (to correct something that may not be relevant: subluxation), unnecessary hazards (e.g., x-ray exposure in the quest for subluxation correction), and delay of appropriate care (through failure to diagnose and/or failure to seek alternative care). Subluxation, a construct that might be a source of guidance to chiropractors (were it to be rigorously investigated and validated), instead functions to distract us from the profession's prime directive: patient benefit.
Subluxation in Marketing
The widespread use of unsubstantiated claims for subluxation and their adjustive correction in marketing to patients [e.g., [57-60]] and to prospective chiropractic students has been noted elsewhere [34,36]. Seaman [45] observes that:
...chiropractors [are] chastised as being "unscientific quacks"... Mostly, it has to do with claims that chiropractors make in marketing their services. Chiropractors are notorious for making treatment claims about chiropractic care that go well beyond the limits of our supportive data, whereas other professionals do not. Consequently, it is the chiropractor who looks like, and subsequently deserves to be called, an amateurish, unscientific huckster.
Some chiropractic suppliers are quite willing to jump on the unsubstantiated bandwagon of the subluxation, as the following promotion for nutritional products suggests:
The practice of Chiropractic is based upon the detection, correction and prevention of the Vertebral Subluxation Complex (VSC)...
The goal of chiropractic care is to restore function to the damaged spine as quickly as possible to minimize the damaging effects of the VSC and the consequential degenerative changes... Current medical literature indicates that specific nutrients can also play an essential and integral role in the support of VSC... [61].
Suffice it to say that the marketing assertions for the value of chiropractic care, frequently offered without acknowledgment of their non-validated status, are commonplace in the profession. The deleterious consequences attributed to subluxation and the clinical outcomes predicted for subluxation correction range from the dread of "killer subluxations" [62] to predictions of "optimal well being" [18] and attainment of maximum human potential. An advertisement that one chiropractor considers in poor taste may profess sacred truth for the next. Since substantiation of assertions may not be considered important to marketers, there are often no scientific boundaries to non-evidence-based chiropractic. Anything goes.
Subluxation as Legal & Political Strategy
The chiropractic subluxation began its legal relevance when the term was included in the wording of various statutes governing the practice of the chiropractic healing art. This trend was continued in the profession's quest for inclusion in the USA Medicare program more than 30 years ago. American chiropractors were chagrined for many years that payment for services in this federal program required radiographic "evidence" of subluxation, but did not compensate the chiropractor for the x-ray films; this stipulation has been eliminated. Many chiropractors now seek to secure their participation in Medicare (despite a skeptical medical community and the availability of manipulative services from non-chiropractor providers) by challenging the federal bureaucracy's interpretation of the Medicare statute.
In their recent "Memorandum of Points and Authorities in Support of Its Cross-motion for Summary Judgment" to the U.S. District Court for the District of Columbia in a suit against the U.S. Department of Health to establish chiropractors' exclusive right to reimbursement for "manipulation to correct a subluxation" in the Medicare program, attorneys for the ACA argue that:
The ACA has presented substantial evidence that Congress did not intend that the services of medical doctors and osteopaths would overlap with the services of chiropractors. In fact, the ACA has clearly demonstrated the illogical paradox of the Secretary's interpretation, namely, that Congress would have had to intended that medical doctors and osteopaths were going to engage in a form of treatment that they believed to be cultist, in order to treat a condition that they did not believe existed, via a treatment method that they did not believe was possible. Surely this type of reasoning would have been absurd, and Congress could not have had that intention when it passed the amendments to the Social Security Act... [63].
The irony here is extreme. Having established the legal meaningfulness of a hypothetical construct whose clinical relevance has yet (if ever) to be scientifically demonstrated, chiropractors now find themselves competing with physical therapists and others over the right to correct subluxations. The greatest absurdity of the situation appears to be missed by all parties concerned: subluxation is "real" because Congress has said so. Data seem irrelevant in this context. Monetary concerns clearly outweigh the issue of scientific validation, and the dogma of subluxation has now spread beyond the chiropractic profession.
Subluxation as Identity
Chiropractors since the Palmers have defined the profession by its focus on finding and adjusting subluxations. Intra-professional feuds have raged over just how exclusive this focus should be, but with few exceptions [e.g., [11,62,64,65]], allegiance is widely pledged to the traditional chiropractic lesion (e.g., Table 1). Clum [39] observes that for some chiropractors "the concept of vertebral subluxation is synonymous with chiropractic and its role has never been questioned." The subluxation is viewed by some chiropractors as a matter of "honor" [66]; anyone who questions the subluxation construct risks vilification as a heretic [66,67]. "Subluxation goes beyond metaphor; it is at the heart of chiropractic" [68]. The International Chiropractors' Association's (ICA's) president seeks a public relations campaign to make subluxation a "household word," and sees the ACC's paradigm as "a really good start" [69]. Edwards [25] insists that the American Chiropractic Association, the world's largest membership society of chiropractors, is no less committed to subluxation than is the Palmer-founded ICA. Gelardi [70] would define the chiropractic profession by its "mission"; his preferred mission is "to contribute to health through the correction of vertebral subluxation." Rome [37] argues that chiropractors' unique subluxation terminology is essential to the preservation of a unique identity. The endorsement of the ACC's statements on subluxation by national membership societies [28] constitutes additional affirmation of the sense among many chiropractic leaders of what a chiropractor is: a subluxation doctor.
Chiropractors' insistence upon defining the profession in terms of a hypothetical (and largely untested) construct is foolish at best: subluxation may or may not be a meaningful notion. This commitment also augurs against the conduct of clinical research to confirm or refute the utility of the subluxation construct, firstly because the presumption of validity undermines the motivation to investigate, and secondly because such research has the potential of undermining this proposed identity (i.e., subluxation doctor). The erosion of reimbursement for chiropractic services is also a possibility if subluxation research fails to measure up to expectations.
Ironically, there is an image of the chiropractor, which seems reasonably well-accepted by many members of the public and whose basis has already garnered some substantial research support [2,3]: the chiropractor as provider of manipulative/adjustive services. Whether the profession can loosen its self-imposed shackle to subluxation dogma is unclear.
Subluxation as Hypotheses
Chiropractors' reluctance to construe subluxation as hypothesis may derive in part from the limited consideration given to epistemology. Epistemology is that branch of philosophy, which deals with the nature of knowledge. Within the context of a clinical discipline such as chiropractic, epistemology addresses the means by which we may gain understanding about the nature of patients' problems, determine optimal methods of resolving or alleviating these problems, and appreciate the mechanisms by which successful interventions are accomplished. Chiropractors have traditionally offered a wide range of epistemological and reasoning strategies [7,71-80], including divine or spiritual inspiration, uncritical empiricism, uncritical rationalism (also referred to as "deductive science" [79]), truth by fiat (e.g., "the chiropractic principle": subluxation), and the critical rationalism and empiricism of the scientific method.
The confusion and incompatibility of these many epistemologies has arisen within a profession, which evolved outside of mainstream higher education and in its early years had little or no sophistication in the realm of scientific investigation [81,82]. Although scholarly and scientific sophistication has emerged in recent decades [83], it appears to be limited to a minority segment of the profession [e.g., [84]]. Inter-professional political pressures may offer a partial explanation for this [85]. Resistance to including chiropractic training within public universities may be more symptomatic than explanatory of the profession's scientific ennui, but the dearth of formal training programs for chiropractor-scientists at chiropractic colleges certainly suggests inadequate concern for the epistemological (i.e., scientific) bases for theories and practice in the profession.
For whatever the reasons, many in the chiropractic profession in the North American continent and in Australia and New Zealand remain committed to a dogmatic orientation to subluxation, its supposed health consequences and the putative benefits to be derived from subluxation-correction [17-19]. Although the percentage of chiropractors who adhere to dogmatism is not known, a 1994 sample of Canadian chiropractors was intriguing [86]. While 86% believed that chiropractors' methods should be validated, 74% disagreed that controlled trials are the best way to accomplish this. And though most (52%) disagreed that "The subluxation is the cause of many diseases," 68% agreed with the notion that "most diseases are caused by spinal malalignment" and most believed that subluxation was detectable by x-ray. Unfortunately, the survey methodology does not allow one to determine the tentative (hypothetical) vs. dogmatic quality of these beliefs.
The traditional chiropractic lesion is often seen as a "philosophical" truth or principle, something that must be defended rather than investigated [87]. This unfortunate pitting of "chiropractic principle" [67] against research scrutiny is often couched in terms of a conflict between philosophy and science:
...It is my contention that a battle between philosophy and science does not and cannot exist within the chiropractic profession or any other discipline. I contend that the real battle is between the great majority of chiropractors who unknowingly allow dogmatism to guide the practice of chiropractic and the extremely rare variety of chiropractor who's practice of chiropractic is guided by philosophy and science [45].
There is nothing inherently dogmatic or anti-scientific in the notion that an articular lesion may have health consequences, or that correction of joint dysfunction may relieve symptoms and/or improve health. Neither does our current inability to predict the effects (if any) of subluxation [88] and/or the benefits of subluxation-correction relegate this hypothetical construct to the dustbin of clinical theories. Indeed, it would be just as inappropriate to dispose of this largely untested theory without data as it is to proclaim its meaningfulness without adequate evidence. On the other hand, as Carl Sagan suggested, extraordinary claims will require extraordinary evidence. With respect to the supposed mechanisms of adjusting, Haldeman [23] reminds us that "What must be avoided... is the unreasonable extrapolation of current knowledge into speculation and presentation of theory as fact." Given the current deficiency of empirical data, the only sound scientific-epistemological position that we can conceive of is to acknowledge our ignorance: we don't know if subluxation is clinically meaningful or not. We suggest that this is a requisite first step toward greater wisdom concerning subluxation.
A Simple Alternative
Speculations and tentative assertions are the stuff from which rigorous science emerges [71]. Indeed, there are those rare scientists whose enduring contributions have derived as much or more from what they theorized than from what they actually tested experimentally (e.g., Isaac Newton and the motions of the planets; Albert Einstein and relativity; Linus Pauling and the role of the hemoglobin molecule in sickle-cell anemia). Hypothetical constructs such as the chiropractic lesion, emotional stress and the neurotic syndromes may or may not have important implications for human biology, but it is entirely appropriate to offer such ideas as tentative assertions.
We could, as C.O. Watkins, D.C. urged decades ago, resolve to be bold in what we hypothesize but cautious and humble in what we claim. In discussing subluxation, all chiropractors should learn to use language that denotes the tentative character of many of our beliefs (hypotheses). Those chiropractors who suspect that subluxation has significant health implications could resolve to investigate scientifically (e.g., through meticulous case reporting), or at least to financially support rigorous investigations, of the meaningfulness of subluxation and its correction. The leaders of our colleges, membership societies and agencies could qualify their statements about subluxation by admitting up front that subluxation is hypothesis(es), not an experimentally demonstrated reality. Those who speak for the profession and who operate in the political, legal and legislative arenas could advance the cultural authority of the profession by becoming credible, balanced, evidence-based sources of information about the chiropractic art. The chiropractic rank-and-file could be encouraged to recognize that responding to charges of quackery with unsubstantiated claims for subluxation and for the outcomes of chiropractic care is self-defeating. Marketers could eliminate the spizzerinctum and hype in their advertisements and concentrate on those aspects of chiropractic for which good data already exist. Speculations could be identified as such, so as not to violate the public's trust and enfeeble the profession's best efforts to progress.
How can such profound change in the profession come about? A century of criticisms by political medicine, many of them not unlike those we offer, has only hardened many chiropractors' attitudes [85]. However, the purpose here is not to contain and eliminate the chiropractic profession, but rather to challenge dogmatic adherence to a hypothetical construct and to help to remedy the many problems that dogmatism has cost the profession. We believe that chiropractic should proceed as a first-class clinical science and art, a profession whose members appreciate and acknowledge what is known and what is not, provide patients with the best care possible given current knowledge, and resolve to extend the borders of scientific understanding in the interest of the public we serve.
The metamorphosis we seek begins with the individual chiropractor who is willing to challenge tradition and peers in the interest of greater integrity for the profession and greater benefit for patients. There is a silent minority who recognize the inappropriateness of the prevailing consensus of dogma concerning subluxation. We recommend that individuals and small groups speak out, educate peers about the distinction between subluxation as hypothesis versus subluxation as dogma, and assert their dissatisfaction with unsubstantiated claims made for the traditional chiropractic lesion. "Silence is not golden: it's consent" [89].
We ask that those who guide the profession and who understand the dilemma that subluxation dogma causes the profession, lead by word and example. Whether one is college faculty or administrator, association official or appointee to a licensing authority, a willingness to reframe subluxation as something tentative rather than something certain is essential. Silence can only serve to sustain our century-long, epistemological misunderstanding of the subluxation construct and corrupt the fullest expression of a worthy future.
Summary
Hypothetical constructs involve tentative assertions about physical reality. They serve as essential tools in the development of science, and permit the empirical testing of the non-obvious. However, when the speculative nature of an hypothesis or hypothetical construct is not made obvious, an otherwise acceptable proposition becomes a dogmatic claim. Such is the history of subluxation in chiropractic.
This brief review of the role of subluxation dogma in clinical practice, in marketing, in the legal and political arenas, as a basis for professional identity, and in the rhetoric of leading chiropractic organizations and agencies, is not a statement about subluxation's validity or lack thereof. Only focused clinical research will enable us to determine whether the traditional chiropractic lesion merits clinicians' attention. We don't know whether subluxation is meaningful or not.
The dogma of subluxation is perhaps the greatest single barrier to professional development for chiropractors. It skews the practice of the art in directions that bring ridicule from the scientific community and uncertainty among the public. Failure to challenge subluxation dogma perpetuates a marketing tradition that inevitably prompts charges of quackery. Subluxation dogma leads to legal and political strategies that may amount to a house of cards and warp the profession's sense of self and of mission. Commitment to this dogma undermines the motivation for scientific investigation of subluxation as hypothesis, and so perpetuates the cycle.
The simple expedient of amending dogmatic assertions to note their tentative, hypothetical character could do much to improve the image of the profession, to re-orient it to the challenge of testing its cherished hypotheses and to establishing the cultural authority of chiropractors in our unique realm of health care. The task of reorienting the profession to a credible science and art belongs to all who understand the scourge of dogma, and who seek a brighter future for the chiropractic profession and its patients.
Authors' contributions
All authors contributed to the writing and re-writing of this paper.
Acknowledgements
None
==== Refs
Moon DK The flight from the subluxation ACA J Chiropr 1981 18 22 3
Bigos S Bowyer O Braen G Acute low back problems in adults 1994 Clinical Practice Guideline No. 14. Rockville, MD: AHCPR Publication No. 95-0642
Bronfort G Spinal manipulation: current state of research and its indications Neurological Clinics of North America 1999 17 91 111
Coulter ID Hurwitz EL Adams AH Meeker WC Hansen DT Mootz RD Aker PD Genovese BJ Shekelle PG The appropriateness of manipulation and mobilization of the cervical spine 1996 Santa Monica CA: RAND Corporation [RAND MR-781-CCR]
Shekelle PG Adams AH Chassin MR Hurwitz EL Phillips RB Brook RH The appropriateness of spinal manipulation for low-back pain: project overview and literature review 1991 Santa Monica, California: RAND Corporation (Document #R-4025/1-CCR/FCER)
Chapman-Smith DA The chiropractic profession: its education, practice, research and future directions 2000 West Des Moines IA: NCMIC Group, Inc
Keating JC Chiropractic: science and antiscience and pseudoscience, side by side Skeptical Inquirer 1997 21 37 43
Keating JC Callender AK Cleveland CS A history of chiropractic education in North America: report to the Council on Chiropractic Education 1998 Davenport IA: Association for the History of Chiropractic
Charlton KH Approaches to the demonstration of vertebral subluxation: 1. introduction and manual diagnosis: a review J Australian Chiropr Assoc 1988 18 9 13
Keating JC To hunt the subluxation: clinical research considerations J Manipulative Physiol Ther 1996 19 613 9 8976481
Nelson CF The subluxation question J Chiropr Humanities 1997 7 46 55
Wenban AB Subluxation research: a survey of peer-reviewed chiropractic scientific journals Chiropr J Australia 2003 33 122 30
Gatterman MI (ed) Foundations of chiropractic: subluxation 1995 St. Louis: Mosby
Leach RA The chiropractic theories: a textbook of scientific research 2004 Fourth Philadelphia: Lippincott, Williams & Wilkins
Masarsky C Todres-Masarsky M Somatovisceral aspects of chiropractic: an evidence-based approach 2001 New York: Churchill Livingstone
Rosner AL The role of subluxation in chiropractic 1997 Arlington VA: Foundation for Chiropractic Education & Research
Association of Chiropractic Colleges Position paper #1 J Manipulative Physiol Ther 1996 19 634 7
Chiropractic Association of Australia, CoreValues
New Zealand Chiropractors' Association 4 May 2004
McDonald WP Durkin K Iseman S Pfefer M Randall B Smoke L How chiropractors think and practice: the survey of North American chiropractors 2003 Ada OH: Institute for Social Research, Ohio North University
New study find unity in chiropractic: surprising agreement among DCs on issues of philosophy, practice Dynamic Chiropractic 21 1, 8, 10 2003 (June 2)
Jensen GA Mootz RD Shekelle PG Cherkin DC Insurance coverage of chiropractic services Chapter 6 in Chiropractic in the United States: training, practice and research 1997 Agency for Health Care Policy and Research, AHCPR Publication No. 98-N002
Haldeman S Neurologic effects of the adjustment J Manipulative Physiol Ther 2000 23 112 4 10714538 10.1016/S0161-4754(00)90078-2
Keating JC Science and politics and the subluxation Amer J Chiropr Med 1988 1 107 10
Edwards J The four biggest lies in chiropractic
Edwards J Golfing with C.J Dynamic Chiropractic 22 34 2004b (May 20)
Lynch RP Passion: where has it gone? J Amer Chiropr Assoc 1995 32 5 6
Barge FH Commitment to a paradigm: the solution to chiropractic's problems The Chiropractic Choice 2003 2 3 4
Wenban AB Commentary: Subluxation-related research: is it time to call it a day? Chiropr J Australia 2003 33 131 7
Oakley PA Harrison DE Letter to the editor J Manipulative Physiol Ther 2004 27 72 3 14739881 10.1016/j.jmpt.2003.11.018
Owens EF Chiropractic subluxation assessment: what the research tells us J Can Chiropr Assoc 2002 46 215 20
Phillips RB Campus Connection 2003 2
Nelson CF Chiropractic scope of practice J Manipulative Physiol Ther 1993 16 488 97 7848386
Grod J Sikorski D Keating JC The unsubstantiated claims of the largest state, provincial and national chiropractic associations and research agencies J Manipulative Physiol Ther 2001 24 514 9 11677551 10.1067/mmt.2001.118205
Keating JC Hansen DT Quackery vs. accountability in the marketing of chiropractic J Manipulative Physiol Ther 1992 15 459 70 1431634
Sikorski DM Grod JP The unsubstantiated web site claims of chiropractic colleges in Canada and the United States J Chiropr Education 2003 17 113 9
Rome PL Usage of chiropractic terminology in the literature: 296 ways to say "subluxation": complex issues of the vertebral subluxation Chiropr Tech 1996 8 49 60
Gatterman MI Hansen DT Development of chiropractic nomenclature through consensus J Manipulative Physiol Ther 1994 17 302 9 7930963
Clum GW Rosner AL Introduction II The role of subluxation in chiropractic 1997 Arlington VA: Foundation for Chiropractic Education & Research
Davenport HW Pavlov's physiology factory: a physiologist as entrepreneur: an essay review J Hist Med & Allied Sciences 2004 59 273 89 10.1093/jhmas/59.2.273
Schiller F Spinal irritation and osteopathy Bull Hist Med 1971 45 250 66 4255365
Starr P The social transformation of American medicine 1982 New York: Basic Books
Long PH The naked chiropractor: insiders' guide to combating quackery and winning the war against pain 2002 Tempe AZ: Evidence-Based Health Services, Inc.,
Sid E Williams DC quoted by the American Chiropractic Association Statement to Associated Press 1 April 1994
Seaman D Philosophy and science versus dogmatism in the practice of chiropractic J Chiropr Humanities 1998 8 55 66
Christensen MG Kerkhoff D Kollasch MW Cohn L Job analysis of chiropractic: A project report, survey analysis, and summary of the practice of chiropractic within the United States 2000 Greeley CO: National Board of Chiropractic Examiners
Peet J Enuresis (bedwetting): restoring normal function Baby Talk: the Baby Adjuster's Newsletter 2 1 4 1993 (April 15)
Parra AK Bonci MA Etiology, treatment and management of enuresis: a review ACA J Chiropr 1989 26 25 8
Wells BF Enuresis and the spine National College J Chiropr 1939 12 21 2
Robuck SV The interosseous lesion as a causative factor in enuresis and other bladder disturbances in children J Amer Osteopathic Assoc 1936 36 73 4
Gemmell HA Jacobson BH Chiropractic management of enuresis: time-series descriptive design J Manipulative Physiol Ther 1989 12 386 9 2607228
Fysh PN Chiropractic management of enuresis Dynamic Chiropractic 12 3 July 30, 1993
Borregard PE Neurogenic bladder and spina bifida occulta: a case report J Manipulative Physiol Ther 1987 10 122 3 3302087
McKellow BM Enuresis and encopresis New Zealand Med J 1984 26 8
Leboeuf C Brown P Herman A Leembruggen K Walton D Crisp TC Chiropractic care of children with nocturnal enuresis: a prospective outcome study J Manipulative Physiol Ther 1991 14 110 5 2019820
Reed WR Beavers S Reddy SK Kern G Chiropractic management of primary nocturnal enuresis J Manipulative Physiol Ther 1994 17 596 600 7884329
Chiropractic BioPhysics
Directional Non-Force Technique
Gonstead Clinical Studies Society
Kale Clinics of Chiropractic
Advertisement for Nutri-West. Nutrition and the vertebral subluxation complex Activator Update 1998 13 4
Carter R Subluxation – the silent killer J Can Chiropr Assoc 2000 44 9 18
American Chiropractic Association, Inc. v. Tommy Thompson, Secretary of the Department of Health and Human Services Plaintiff's reply memorandum of points and authorities in support of its cross-motion for summary judgment, in the United States District Court for the District of Columbia, Case No 98-2762 16 April 2004
Homola S Bonesetting, chiropractic and cultism 1963 Panama City FL: Critique Books
Morgan L Letter to the editor J Can Chiropr Assoc 2000 44 182 3
Lawrence DJ Sacred cows and shibboleths J Chiropr Humanities 1997 7 56 60
Good CJ Traditional chiropractic philosophy: irrelevant or irreplaceable? J Chiropr Humanities 1996 6 50 4
Lawrence DJ Rosner AL Introduction I The role of subluxation in chiropractic 1997 Arlington VA: Foundation for Chiropractic Education & Research
ICA under new management: reform candidates win Dynamic Chiropractic 21 1, 57 2003 (June 16)
Gelardi TA The science of identifying professions as applied to chiropractic J Chiropr Humanities 1996 6 11 7
Charlton KH Data and dogma: the use and abuse of information J Australian Chiropr Assoc 1987 17 46 8
Keating JC Letter to the editor J Manip Physiol Ther 1987 10 80 1
Keating JC A survey of philosophical barriers to technique research in chiropractic J Can Chiropr Assoc 1989 33 184 6
Keating JC Rationalism, empiricism and the philosophy of science in chiropractic Chiropractic History 1990 10 23 30 11622498
Keating JC Toward a philosophy of the science of chiropractic: a primer for clinicians 1992 Stockton CA: Stockton Foundation for Chiropractic Research
Keating JC Scientific epistemology and the status of chiropractic: we are what we do European J Chiropr 1993 41 81 8
Marks D The philosophy of science for chiropractic – simply stated J Chiropr Humanities 1994 4 56 60
Palmer BJ Induction vs deduction 1915 Davenport IA: Palmer School of Chiropractic (pamphlet)
Stephenson RW Chiropractic textbook 1927 Davenport IA: the author
Watkins CO Sidney MT The basic principles of chiropractic government 1944 (reprinted as Appendix A in Keating JC. Toward a philosophy of the science of chiropractic: a primer for clinicians. Stockton CA: Stockton Foundation for Chiropractic Research, 1992)
Keating JC Green BN Johnson CD "Research" and "science" in the first half of the chiropractic century J Manipulative Physiol Ther 1995 18 357 78 7595110
Keating JC Siordia L Green BN Henry G. Higley, pioneer chiropractic scientist Chiropractic History 2004 24 35 60
Keating JC Caldwell S Nguyen H Saljooghi S Smith B A descriptive analysis of the Journal of Manipulative & Physiological Therapeutics, 1989–1996 J Manipulative Physiol Ther 1998 21 539 52 9798183
Keating JC Young MA Who is the chiropractic scientific community? J Australian Chiropr Assoc 1987 17 84 6
Keating JC Mootz RD The influence of political medicine on chiropractic dogma: implications for scientific development J Manipulative Physiol Ther 1989 12 393 8 2691602
Biggs L Mierau D Hay D Measuring philosophy: a philosophy index Journal of the Canadian Chiropractic Association 2002 46 173 84
Keating JC Philosophy and science in chiropractic: essential, inseparable and misunderstood European J Chiropr 2001 46 51 60
Clusserath MT Predictability of the effects of nerve interference J Chiropr Humanities 1997 7 26 31
Charlton KH Guest editorial: Silence is not golden: it's consent Chiropr J Australia 2003 33 81 2
|
16092955
|
PMC1208927
|
CC BY
|
2021-01-04 16:38:23
|
no
|
Chiropr Osteopat. 2005 Aug 10; 13:17
|
utf-8
|
Chiropr Osteopat
| 2,005 |
10.1186/1746-1340-13-17
|
oa_comm
|
==== Front
Clin Mol AllergyClinical and molecular allergy : CMA1476-7961BioMed Central London 1476-7961-3-121611148210.1186/1476-7961-3-12ResearchInsect sting allergy. A study from 1980 to 2003 of patients who started treatment with venom immunotherapy between 1980 and 1998 Haye Rolf [email protected]øsen Liv Kari [email protected] Department of otolaryngology, Rikshospitalet-Radiumhospitalet HF University of Oslo 0027 Oslo, Norway2005 19 8 2005 3 12 12 29 3 2005 19 8 2005 Copyright © 2005 Haye and Døsen; licensee BioMed Central Ltd.2005Haye and Døsen; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Previously we treated patients with insect sting allergy with venom immunotherapy (IT) using whole body insect extracts. From 1980 we changed to insect venoms. The purpose of this study was to analyse data from the patients in order to improve our treatment.
Methods
This is an open, single centre study on patients treated with venom IT 14 years or older with a history of a systemic allergic reaction to an insect sting, a positive skin prick test (SPT) or a positive RAST and willingness to comply with five years of IT. Clinical and laboratory data were registered prospectively at the start of IT and after five years of treatment until 2003 on patients who started IT between 1980 and 1998. Questionnaires were answered in 1989, 1993 and 2003. Statistical analysis was done with Pearson's chi square, Fisher's exact or the t-test.
Results
Of 315 patients treated, 44 were given bee, 248 common wasp and 23 both venoms. Of the common wasp sting incidents 5.5 % resulted in a severe allergic reaction (SAR) during adequate IT and 22% after cessation. Seventy-one per cent of the patients carried epinephrine. Precautionary steps were taken by 77% of the patients during or after inadequate IT. On or after adequate IT 83% felt completely or substantially safe. Surprisingly 29 % of those inadequately treated felt safer and 50% were satisfied with having had the opportunity to be treated. The SPT became negative in 68% of the wasp allergic patients after five years of adequate IT. Increased risk of experiencing SAR to a future sting in wasp allergic patients after cessation of adequate IT was significantly associated with a SAR due to IT during the rush regimen. SAR due to IT occurred very rarely during maintenance dosing.
Conclusion
Adequate venom IT is very effective while ongoing but somewhat less effective after cessation, while inadequate treatment gives poor results. More of our patients should complete five years of IT and some should continue IT. The type of reaction to IT during incremental dosing may be of help in deciding who should continue beyond five years. Maintenance IT may be taken over by the general physician.
==== Body
Background
After the study of Hunt et al [1] was published in 1978 showing the ineffectiveness of whole body insect extracts, we stopped using whole body insect extracts in the treatment of patients with insect sting (hymenoptera) allergy. When venom from bee and common wasp (vespula sp) became available in Norway in 1979, we started skin testing patients allergic to insect stings and thereafter treating those who had a positive skin prick test (SPT) or positive specific IgE reaction (RAST) with immunotherapy (IT) using insect venoms.
The purpose of this study was to examine the short and long term effectiveness; side effects; causes for cessation of IT; the serological data and the SPT results and quality of life of our insect sting allergic patients, in order to help us improve our treatment. The patients started IT between 1980 and 1998. The results from the first eight years of treatment were published in 1988 [2], and are incorporated in the present study.
Materials and methods
In Norway there are about 20 different species of wasps, of which vespula vulgaris and germanica (common wasps) are frequently seen. Vespa crabro has not been observed in Norway [3]. From a previous study we found that we could use venom from common wasp (vespula sp) as the sole wasp venom to our wasp allergic patients [3]. In 1979 we established the concentration that discriminates insect sting allergic patients from normal individuals and found this to be 100 microgram pr. ml for the SPT and 1 microgram pr ml for the intracutaneous test (IC) [4] the latter corresponding to data from the study of Hunt et al [5]. We found the sensitivity to be almost equal for both tests [4] and as SPT is the preferred method for skin testing in Norway, we chose SPT as the standard method
Patients who had a history of an immediate systemic allergic reaction, with vascular and/or respiratory symptoms, angioedema of the head and neck or symptoms from two organ systems, to a sting from bee and/or wasp and who were willing to comply with a minimum of five years of treatment were offered IT. The SPT or RAST reaction had to be positive to the respective venom. For SPT we used Pharmalgen (ALK) venoms at 100 microgram pr. ml concentration and Soluprick (ALK) for inhalant allergens. Histamine chloride served as reference, 1 mg/ml for insect venoms during the whole study. For the first five years of the study the commercially available SPT test for inhalant allergens used 1 mg/ml histamine as reference and later 10 mg/ml. The concentration of the inhalant allergens was adjusted accordingly. If the size of the wheal of the SPT area was 50% or larger than that caused by histamine and had a minimum diameter of 3 mm larger than the control, it was recorded as positive. Serum analyses of total and specific IgE were done at a laboratory using Pharmacia RAST from 1980 to1987, CAP from1988 to1998 and DCP Alastat from 1999 to 2003. In addition analyses of haemoglobin, leukocytes, CRP, creatinine, ASAT, ALAT and serum electrophoresis were undertaken. Serum tryptase was not analysed, however none of the patients had urticaria pigmentosa. The IT was performed with Pharmalgen (ALK) venoms. The injections were given subcutaneously on the lateral part of the upper arm. Aspirations were done at the beginning and during the injection. We used a rush regimen of five days duration giving a starting dose of 0.1 microgram venom, doubling this every 2 hours for three days. The dose was increased more slowly on day four and five aiming for 50 micrograms or more for the last injection. Thereafter the venom was given at weekly intervals gradually increasing the dose to 0.1 mg and the interval to six weeks. Seven patients received Alutard (ALK) venoms. Many patients were referred back to their general practitioner to continue the IT after having reached 0.1 mg as their maintenance dose. A few patients who only tolerated 50 micrograms venom used this as their maintenance dose even if this is less effective. If they did not tolerate 50 micrograms of venom after up to three years of IT, the treatment was ended. Patients who tested positive to both bee and common wasp venoms were treated with both venoms. Beekeepers and their relatives who were allergic to bee stings were not given IT if they abandoned bee keeping and there was no bee keeping within 10 km. Patients were instructed to keep epinephrine at hand and for those who had not reached the dose of 0.1 mg venom, to inject it immediately after a sting. Having reached this dose they should only inject if symptoms occurred.
Five years after the start of IT the patients were asked to return for clinical examination, SPT and serological tests, regardless of status of completion of IT. Many patients also met for an extra consultation at a later date for the same procedures. Information was also obtained at the time of renewal for prescription of venom. Questionnaires asking for the patients reactions to stings, and reaction to and length of IT were answered by mail and/or by telephone interview in 1989, 1993 and 2003. The 2003 questionnaire also included questions on precautionary steps and safety feeling. We divided our patients into two groups according to the duration and dosage of the IT. IT is considered adequate if the maintenance dose is 0.1 mg venom and at cessation has lasted a minimum of five years. It is inadequate when the dose is less than 0.1 mg or at cessation of a shorter duration than five years. SAR (serious allergic reaction) is defined as a respiratory and/or vascular reaction with an additional feeling of impending disaster or other serious symptom or requiring treatment with epinephrine.
Statistical analysis was done with Pearson's chi square or Fisher's exact test. For the duration of IT two samples t-test was used.
Results
Three hundred and fifteen patients were included in the study, 151 males (48%) and 164 females (52%). The age varied from 14 to 73 with a mean of 41.1 years. Forty four patients (14%) were given bee, 248 (79%) common wasp (vespula sp.) and 23 (7%) both venoms. Table 1 shows the most serious symptom the patients had experienced from a sting.
According to the classification of Mueller, 306 were classified as having a grade three or four and nine a grade two or lower reaction. We have grouped together grades three and four as we in many cases found it difficult to adequately classify the patients from the case histories. In stead we tried to differentiate whether the reaction was vascular, respiratory or combined vascular-respiratory (Table 1). The SPT for inhalant allergy was positive in 32.5 % of the patients, but only 21% reported having a respiratory allergy.
Table 1 Main Symptom
Insect\Symptom Vascular and respiratory Vascular Respiratory Angioedema head/neck Skin + gastrointest. Total
Bee 12 18 13 1 44
Common wasp 87 110 46 3 2 248
Bee/common wasp 6 8 6 3 23
Total 105 136 65 7 2 315
Up to 2003 twenty one patients have died of causes unrelated to insect allergy. Information regarding these patients obtained prior to their deaths is included in the results. At the five years recall 92.2% met for consultation and an additional 43.9% at a later date. The response rate to the questionnaires was 98.3% in 1989, 90.7% in 1993 and 88.6% in 2003.
Several patients have experienced a reaction to the venom injections. Their most serious reaction during incremental dosing and during maintenance is recorded in Table 2. Nearly half of the patients had a reaction and 23.5% a systemic allergic one. SAR occurred more than once in a few patients although we adjusted further dosing carefully. SAR occurred in 45 patients 14.2% (12 males, 33 females) (33 allergic to wasp, and 12 to bee) during increasing dosing.
Table 2 Number of patients reacting to immunotherapy
Type of reaction During incremental dosing During maintenance dosing
Large local reaction 14 5
Itching in throat or nose / vomiting 18 1
Urticaria 11 2
Sedation 51 25
SAR 45 5
Joint / muscle pain 21 26
Tachycardia 10 1
Total 170 65
Several patients experienced joint or muscle pain. In five cases swelling of joints was related to the injections disappearing later but sometimes steroids were given to alleviate the symptoms. In other cases rheumatoid arthritis or osteoarthritis proved to be the cause of these symptoms. During maintenance dosing SAR in one bee allergic patient occurred when an attempt was made to increase the dose above 0.1 mg, in one wasp allergic patient due to a higher than planned dose after a prolonged interval of IT, in one patient who received both venoms at the second maintenance dose, in one wasp allergic patient after a few maintenance doses and in one case there was an unexplained syncope one day later. In three of these cases there was a change in allergic sensitivity so that it was impossible later to reach an adequate maintenance dose again and IT was abandoned.
Seventy-seven patients (24%) did not complete the minimum of five years of IT. The reasons for stopping treatment before five years are given in Table 3. The most serious reaction causing cessation occurred in a female patient who had a cardiac arrest without prior warning at day four, 20 minutes after receiving 30 microgram common wasp venom as the second dose that day. It was difficult to resuscitate her, but she has recovered completely. Another patient received the dose accidentally in an intravenous drip. In the group other causes for stopping IT one was a patient afraid of acquiring AIDS from the venom injections and three patients afraid of receiving injections during pregnancy although we recommended continuation.
Table 3 Causes for cessation of immunotherapy before 5 years. Number of patients
Problems due to work or change of domicile 15
Reactions to immunotherapy 36
Other serious illness 7
Death of other causes 1
Own initiative 11
Tolerated multiple stings 1
Others 6
Total 77
Eighteen patients are still treated.
Seventy one per cent of the patients keep epinephrine at hand. No difference was seen whether or not the patients had completed five years of treatment. Of those inadequately treated 76 % take precautionary steps to avoid being stung, whereas only 57.4% of those adequately treated do so. Table 4 lists the different precautionary steps taken.
Table 4 Precautionary steps taken. Number of patients
Total
Never barefoot 126
No perfume 84
Never drinking/eating outdoors 73
Never hiking alone 46
Always window screen 17
Mobile phone present 6
Others 16
Unknown 47
In Table 5 the feeling of safety as experienced by the patients according to the status of their IT is shown. Eighty-three per cent of the adequately treated patients feel completely or substantially safe and surprisingly 29% of the inadequately treated. Those who feel unsafe although having completed adequate IT had either experienced a SAR to a sting during treatment or were discouraged by having a positive SPT reaction at the fifth year, even if informed that this is not uncommon. Ninety-four per cent of the adequately treated patients found the investment in time, expenses and effort worthwhile (Table 6). Those who did not benefit from the IT regarded the inconvenience of IT as bothersome.
Table 5 Feeling of safety. Number of patients
Adequate treatment Inadequate treatment Total
Complete 54 4 58
Substantially 118 14 132
Somewhat 21 19 40
None 14 24 38
Unknown 28 19 47
Total 235 80 315
Table 6 Worth the effort. Number of patients
Adequate treatment Inadequate treatment Total
Yes 192 31 223
No 11 29 40
Unknown 32 20 52
Total 235 80 315
Reactions to field stings during treatment in patients allergic to common wasp as reported by the patients are presented in Table 7. The number of stings per incident was low, usually only one and never more than four. We have recorded the reactions to each incident regardless of the number of stings. SAR occurred in 55% of the incidents when the venom dose was below 0.1 mg. At full maintenance dosing (0.1 mg venom) only 5.5% of the incidents resulted in a SAR. Epinephrine was self administered in most cases of SAR and also in some cases where no reaction occurred (table 7). A few times the reaction occurred so rapidly that the patients did not have time enough to administer it. In bee allergic patients six of 13 patients had a SAR while inadequately treated, but only one of 14 in adequately treated ones. In addition two beekeepers had multiple stings without reaction.
Table 7 Reaction to sting in common wasp allergic patients during IT. Number of times stung
Type of reaction Inadequate dosage Adequate dosage
No / local swelling 7 (1) 132 (23)
Sedation 2 (1) 25 (10)
Itching / urticaria etc. 0 7
SAR 14 (14) 9 (7)
Joint / muscle pain 0 3 (3)
Tachycardia 0 2 (1)
Total 23 (16) 178 (44)
In parenthesis number of times adrenaline was administered.
We have also registered the reactions to stings after cessation of IT. In inadequately treated patients SAR occurred in 19 of 45 (42%) of the sting incidents in wasp allergic patients and in four of 19 (21%) in bee allergic ones. After cessation of adequate IT in common wasp allergic patients (Table 8) SAR occurred in 22% of all the sting incidents. The reaction rate was higher after five years of cessation. None of the bee allergic patients experienced a SAR after cessation of IT.
Table 8 Reaction to sting after adequate immunotherapy of 5 or more years' duration in common wasp. Number of times stung
Type of reaction 0–5 years after cessation More than 5 years after cessation Total
No / local swelling 61 (3) 14 75 (3)
Itching / urticaria etc. 5 (2) 0 5 (2)
Sedation 7 (3) 1 8 (3)
SAR 19 (11) 7 (5) 26 (16)
Joint pain 0 0 0
Tachycardia 4 (1) 0 4 (1)
Total 96 (20) 22 (5) 118 (25)
In parenthesis: Number of times adrenaline was administered.
The SPT became negative in 68% of the patients allergic to common wasp having completed adequate IT, and only in 42% of inadequately treated ones (Table 9). For bee allergic patients the SPT became negative in 74.3% of the adequately treated ones.
Table 9 SPT at start and after 5 years in bee and wasp allergic patients
Bee-allergic SPT grading
Year Neg. Pos. Total
0 0+0 40+23 40+23 = 63
5 29+11 10+11 39+22 = 61
Wasp-allergic
Year Neg. Pos. Total
0 1 208+43 209+43 = 252
5 142+18 67+25 209+43= 252
Italics: Number of patients incompletely treated.
Total IgE did not change appreciably during the five years, see Table 10.
Table 10 Total IgE. Patients allergic to common wasp
kU/ml < 122 123 – 300 > 300 Total
Year: 0 132 + 26 29 + 8 15 + 7 176 + 41 = 217
Year: 5 131 + 27 24 + 8 21 + 5 176 + 40 = 216
In italics: Number of patients incompletely treated.
No changes that could be attributed to the IT were seen in haemoglobin, creatinine, ASAT, ALAT, CRP or serum electrophoresis.
Table 11 lists a comparison between those who did not experience any reaction to a sting after cessation of adequate IT (group A) and those who experienced a SAR (group B) in patients allergic to common wasp. A statistical analysis was made on the following data, age, gender, symptom at inclusion, SPT to inhalant allergens, SPT, RAST and total IgE at five years of treatment, reaction due to incremental IT and duration of IT, to assess the prognostic value for a SAR to future stings after cessation of IT. Only the reaction to IT during incremental dosing had a prognostic value in assessing the reaction to a future sting.
Table 11 Comparison between 2 different groups of patients with allergy to common wasp according to reaction to sting after cessation of it
A
N = 50 B
N = 17 p values
No. % No %
Age ≤ 29 10 20 1 6 0.434
30–59 34 68 14 82
≥ 60 6 12 2 12
Gender Male 33 66 9 53 0.391
Female 17 34 8 47
Inclusion symptom Vascular 20 40 6 35 0.731
Respiratory 13 26 3 17 0.485
Inclusion symptom Vascular + respiratory 16 32 8 47 0.38
SPT pos. inhalation allergens 15 30 6 35 0.684
Reaction to rush regimen No / local swelling 29 58 7 41 0.020
SAR 4 8 6 35 0.013
Pos. SPT to venom 5 th year 16 32 9 53 0.123
Venom RAST, pos. 5 th year 26/48 54 11/14 79 0.129
Total IgE ≥ 123, 5 th year 9/44 20 6/14 43 0.158
Duration of IT in years 7.1 (5–15) 6.5 (5–10) 0.156
A: Patients allergic to common wasp who have had a minimum of 5 years of immunotherapy at full dose and have not had any reaction to subsequent stings.
B: Patients allergic to common wasp, who have had at least 5 years of immunotherapy at full dose, but after cessation have had serious reactions to stings.
Discussion
This study was started in 1980 and was ongoing for many years. However the main author has been in charge during the whole period. The clinical information and laboratory data have been continuously registered. Some of the test procedures have changed. At the start the SPT kits contained histamine 1 mg/ml as reference and inhalant allergens corresponding to that (1 HEP). In 1985 histamine 10 mg/ml and allergens corresponding to this (10 HEP) became standard. As we have graded the reaction according to the corresponding reference, we believe that the results are comparable for inhalant allergens. As we chose to use 0.1 mg venom as basis for the SPT, we continued to use histamine 1 mg/ml as reference and we have therefore been able to compare the test results throughout the years.
The RAST results however may not be directly comparable as the laboratory changed their methods. The CAP system is more sensitive than the original RAST [6] so that the conversion factor from this study was used. The DPC AlaStat method has in unpublished data in-house shown to be comparably sensitive to CAP.
It is estimated that in the Nordic countries a patient in average gets a sting every tenth year [4,6]. That means that the untreated patient would expect SAR to several future stings.
This study has shown that ongoing IT for common wasp is very efficient when the maintenance dose is 0.1 mg, as also seen in other studies [1,8] and for bee even better than others [9]. The results may have been slightly different as some patients who did not experience any reaction to a sting still injected epinephrine.
We strive to get as high a percentage of patients to complete five years of IT as seen in other studies [10] because the results of our inadequately treated patients are poor and do not seem to differ from untreated ones [11]. After cessation of IT the results for inadequately treated patients are unfavourable compared to adequately treated ones.
However, the adequately treated wasp allergic patients experienced a SAR to sting after cessation of IT more frequently than has been found in other studies [12]. We have no explanation for this. Some of the patients should therefore preferable continue IT indefinitely. Such selection is difficult. According to other studies the following symptoms and signs have been found to be risk factors for experiencing SAR at a future sting after cessation of adequate IT [13]: severity of pre-treatment reaction, safety and efficacy of IT, duration of IT, and high sensitivity with the IC test. Specific IgE has in some but not all studies [13,14] been found predictive. We, however, have only found SAR due to IT during incremental dosing to be significantly predictive. Using this criterion we may be able to offer continued IT to some patients and still keep the total number treated low. The rest of the patients will have to rely on epinephrine.
Epinephrine offered as the only treatment instead of IT seems to reduce quality of life [15]. This has also been found when recommended to patients given IT [16]. We believe that carrying epinephrine is a sound safety precaution as our data show that the patients were able to use it when needed. We are satisfied that a majority of our patients carry epinephrine and feel completely or substantially safe even carrying epinephrine. This has fortunately also been seen in other studies [17,18].
We have encountered several side effects during IT. The percentage of patients with a systemic reaction during incremental dosing is comparable to the EAACI multicenter study [19] although we did not pretreat our patients with antihistamines. The percentage of patients experiencing a SAR during incremental dosing is higher. One reason for this may due to the higher percentage of females in this study as the reaction rate to IT in females was higher than in males in this and the other study [19]. Our rush regime leads to a high accumulated dose on the fourth or fifth day. This was in another study [20] found to be a cause for serious reactions. We therefore planned to modify our IT schedule on day four and five to reduce the number of reactions. However, venoms are presently only available in depot extracts in Norway. The very few reactions during normal adequate venom maintenance dosing allows us to continue our practice of having the patients' local physicians continue the IT if they take the necessary precautions.
Besides the allergic reactions we found sedation (fatigue) and particularly joint and muscle pain to be an obstacle to further IT. Headache and fatigue is well known [21] but did not lead to cessation of IT in our study, although some patients had to abstain from work on the day of injection and sometimes also the next day. Little attention has been paid to joint/muscle pain which in this study led to cessation of IT in some cases. Although we suspect this to be an immune disease, we were unable to confirm it.
A large majority of the patients felt that the investment in time, effort and expenses was worthwhile if they completed IT. This is comparable to reports by Oude Elbrink et al. [18] and Røsjø et al. [17]. Surprisingly 50% of those that did not complete adequate IT also felt so. We believe that these patients were grateful for having had the opportunity to be treated even though it was not adequate.
Conclusion
Ongoing venom IT is very effective when adequately performed and the majority of patients feel substantially safer. Most of them found the effort very worthwhile. We should endeavour to get more of the patients to complete five years of IT as the results both while ongoing and after IT is much better for the adequately than for the inadequately treated patients. Still after cessation of adequate IT 22% of the wasp sting incidents resulted in a SAR. Fortunately the majority of patients keep epinephrine at hand and were able to use it when needed. Maintenance IT of 0.1 mg of single venom may be taken over by the general physician. Only the following parameter, the reaction to IT during incremental dosing was statistically significant for predicting the reaction to a future sting after cessation of adequate IT. This may help in deciding whether a patient should continue IT after five years or not.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RH performed the consultations, registrations and follow ups of patients from 1980 to 2003, drafted and took care of the questionnaires from 1989 and 1993, compiled the data, drafted the tables and is the main author. LKD participated in the consultations and the follow up from 2001–2003, designed and took care of the 2003 questionnaire, participated in compiling the data and in drafting and revising the manuscript.
Acknowledgements
We are grateful to Marte Olstad, Marijke Veenstra and Geir Aamodt, Section of Biostatistics, Rikshospitalet -Radiumhospitalet HF for doing the statistical analysis
==== Refs
Hunt KJ Valentine MD Sobotka AK Benton AW Amodio FJ Lichtenstein LM A controlled trial of immunotherapy in insect hypersensitivity N Engl J Med 1987 299 157 161 78446
Haye R Hyposensibilisering med insektgift 1980–87 Tidsskr Nor Lægeforen 1988 108 2476 2478
Haye R Søhoel P Olsen E Insektallergi. De forskjellige insekters betydning Tidsskr Nor Lægeforen 1985 105 1321 1323
Brøndbo K Haye R Insektallergi Tidsskr Nor Lægeforen 1980 100 1344 1346
Hunt KJ Valentine MD Sobotka AK Lichtenstein LM Diagnosis of allergy to stinging insects by skin testing with Hymenoptera venoms Ann Intern Med 1976 85 56 59 59564
Jeep S Kirchhof E Kunkel G Comparison of the Phadebas RAST with the Pharmacia CAP system for insect venom Allergy 1992 47 212 217 1510233
Mosbech H Anafylaxis to insect venom Novartis Found Symp 2004 257 177 192 15025398
Golden DBK Kagey-Sobotka A Gadde J Valentine MD Lichtenstein LM Dose dependence of Hymenoptera venom immunotherapy J Allergy Clin Immunol 1981 67 370 374 7229226 10.1016/0091-6749(81)90082-8
Mueller U Helbling A Berchtold E Immunotherapy with honey bee venom and yellow jacket venom is different regarding efficacy and safety J Allergy Clin Immunol 1992 89 529 535 1740583 10.1016/0091-6749(92)90319-W
Mellerup MT Hahn GW Poulsen LK Malling H-J Safety of allergen-specific immunotherapy. Relation between dosage regimen, allergen extract, disease and systemic side-effects during induction treatment Clin exp allergy 2000 30 1423 1429 10998019 10.1046/j.1365-2222.2000.00910.x
Settipane GA Charfee FH Natural history of allergy to Hymenoptera Clin Allergy 1979 9 385 390 476909
Golden DBK Discontinuing venom immunotherapy Curr Opin Allergy Clin Immunol 2001 1 353 356 11964712 10.1097/01.all.0000011038.45505.c6
Mueller UR Golden DBK Demarco PJ Lockey RF Immunotherapy for Hymenoptera Venom and Biting Insect Hypersensitivity Clinical Allergy and Immunology 2004 18 541 559 15042934
van Halteren HK van der Linden PW Burgers JA Bartelink AK Discontinuation of yellow jacket venom immunotherapy: follow-up of 75 patients by means of deliberate challenge J Allergy Clin Immunol 1997 100 767 770 9438484
Goldberg A Confino-Cohen R Insect sting-inflicted systemic reactions: Attitudes of patients with insect venom allergy regarding after-sting behavior and proper administration of epinephrine J Allergy Clin Immunol 2000 106 1184 1189 11112904 10.1067/mai.2000.110927
Confino-Cohen R Melamed S Goldberg A Debilitating beliefs, emotional distress and quality of life in patients given immunotherapy for insect sting allergy Clin exp allergy 1999 29 1626 1631 10594538 10.1046/j.1365-2222.1999.00656.x
Røsjø HR Hov JEJ Borchsenius F Skjønsberg OH Vaksinasjon mot vepseallergi Tidsskr Nor Lægefor 2003 123 1075 1077
Oude Elberink JNG Dubois AEJ Quality of life in insect venom allergic patients Curr Opin Allergy Clin Immunol 2003 3 287 293 12865773 10.1097/00130832-200308000-00009
Mosbech H Mueller U Side effects of insect venom immunotherapy: results from an EAACI multicenter study Allergy 2000 55 1005 1010 11097308 10.1034/j.1398-9995.2000.00587.x
Birnbaum J Charpin D Vervloet D Rapid Hymenoptera venom immunotherapy: Comparative safety of three protocols Clin exp allergy 1993 23 226 230 8472191
Rueff F Przybilla B Venom immunotherapy: adverse reactions and treatment failure Curr Opin Allergy Clin Immunol 2004 4 307 311 15238797 10.1097/01.all.0000136754.13077.fc
|
16111482
|
PMC1208928
|
CC BY
|
2021-01-04 16:36:24
|
no
|
Clin Mol Allergy. 2005 Aug 19; 3:12
|
utf-8
|
Clin Mol Allergy
| 2,005 |
10.1186/1476-7961-3-12
|
oa_comm
|
==== Front
Curr Control Trials Cardiovasc MedCurrent Controlled Trials in Cardiovascular Medicine1468-67081468-6694BioMed Central 1468-6708-6-111605352510.1186/1468-6708-6-11ResearchThe SPHERE Study. Secondary prevention of heart disease in general practice: protocol of a randomised controlled trial of tailored practice and patient care plans with parallel qualitative, economic and policy analyses. [ISRCTN24081411] Murphy Andrew W [email protected] Margaret E [email protected] Susan M [email protected] Molly [email protected] Claire [email protected] Mary C [email protected] Department of General Practice, Clinical Sciences Institute, National University of Ireland, Galway, Ireland2 Department of General Practice, Queen's University, Dunluce Health Centre, 1 Dunluce Avenue, Belfast BT9 7HR, Northern Ireland3 Department of Public Health and Primary Care, Trinity College Centre for Health Sciences, Adelaide and Meath Hospital, Tallaght, Dublin 24, Ireland4 Department of Psychology, National University of Ireland, Galway, Ireland2005 29 7 2005 6 1 11 11 2 6 2005 29 7 2005 Copyright © 2005 Murphy et al; licensee BioMed Central Ltd.2005Murphy et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The aim of the SPHERE study is to design, implement and evaluate tailored practice and personal care plans to improve the process of care and objective clinical outcomes for patients with established coronary heart disease (CHD) in general practice across two different health systems on the island of Ireland.
CHD is a common cause of death and a significant cause of morbidity in Ireland. Secondary prevention has been recommended as a key strategy for reducing levels of CHD mortality and general practice has been highlighted as an ideal setting for secondary prevention initiatives. Current indications suggest that there is considerable room for improvement in the provision of secondary prevention for patients with established heart disease on the island of Ireland. The review literature recommends structured programmes with continued support and follow-up of patients; the provision of training, tailored to practice needs of access to evidence of effectiveness of secondary prevention; structured recall programmes that also take account of individual practice needs; and patient-centred consultations accompanied by attention to disease management guidelines.
Methods
SPHERE is a cluster randomised controlled trial, with practice-level randomisation to intervention and control groups, recruiting 960 patients from 48 practices in three study centres (Belfast, Dublin and Galway). Primary outcomes are blood pressure, total cholesterol, physical and mental health status (SF-12) and hospital re-admissions.
The intervention takes place over two years and data is collected at baseline, one-year and two-year follow-up. Data is obtained from medical charts, consultations with practitioners, and patient postal questionnaires.
The SPHERE intervention involves the implementation of a structured systematic programme of care for patients with CHD attending general practice. It is a multi-faceted intervention that has been developed to respond to barriers and solutions to optimal secondary prevention identified in preliminary qualitative research with practitioners and patients. General practitioners and practice nurses attend training sessions in facilitating behaviour change and medication prescribing guidelines for secondary prevention of CHD. Patients are invited to attend regular four-monthly consultations over two years, during which targets and goals for secondary prevention are set and reviewed. The analysis will be strengthened by economic, policy and qualitative components.
==== Body
Background
Coronary heart disease (CHD) is a common cause of death and an important cause of morbidity in Ireland. Secondary prevention of heart disease involves long-term management of risk factors among people who have been diagnosed with established CHD. The Scottish Intercollegiate Guidelines Network has defined secondary prevention as the 'identification and modification of risk factors by the introduction of lifestyle measures and pharmacological therapy and cardiac rehabilitation' [1]. Secondary prevention has been recommended as a key strategy for reducing levels of CHD [2,3]. Secondary prevention can be achieved by stopping smoking, making healthier food choices (including reducing fat intake and increasing intake of fruit, vegetables and fibre), becoming physically active, achieving an ideal weight, consuming alcohol in moderation, appropriate prescription of, and adherence to, pharmacological therapy, achieving blood pressure level at or under 140/90 mmHg and achieving a total cholesterol level at or below 5 mmol/l [3].
Most people with CHD regularly attend their general practitioner and general practice has been highlighted as an ideal setting for secondary prevention initiatives [2]. Previous studies of provision of secondary prevention in general practice have identified sub-optimal levels [4-6]. Current indications suggest that there is considerable room for improvement in the provision of secondary prevention for patients with established heart disease on the island of Ireland [7]. Recent data from the national Heartwatch programme suggest that 44% of Irish patients with established heart disease have a systolic blood pressure above the recommended guidelines of 140 mmHg, and that 36% have a baseline cholesterol level of greater than 5 mmol/l (Leahy J, personal communication). In a survey of secondary preventive care of 1,600 patients, from 35 randomly selected general practices in the west of Ireland, results were as follows: 23% of patients were regular smokers, 45% had cholesterol readings greater than 5 mmol/l and 34% had blood pressure readings greater than 140/90 mmHg. In addition, GP records were found to be incomplete, with 38% of patients with no record of smoking status and 25% with no cholesterol reading. Information from Northern Ireland suggests a similar situation: a recent survey found that among patients with a confirmed diagnosis of CHD 18% smoked, 25% had a body mass index (BMI) greater than 30, 51% had a systolic blood pressure greater than 140 mmHg and approximately 50% had cholesterol levels greater than 5 mmol/l [8].
Randomised controlled trials have investigated the effectiveness of secondary prevention interventions including health visitor advice [9], nurse-led secondary prevention clinics [10,11], outreach visits to practices by nurses trained in cardiovascular risk factor recording [12], provision of specialist liaison nurses to bridge the gap between secondary and primary care [13], postal prompts to patients to encourage lifestyle changes [14], audit and systematic recall of patients from disease registers [15], combined training to all practice staff in information systems and evidence-based medicine [16] and a cognitive-behavioural disease management programme [17].
These trials have reported a number of successful outcomes, including increased physical activity [9,10,17], improved diet [9,10,17], increased health functioning [9,10,17], increased patient assessment and recording of risk factors [12,14-16], increased rate of consultation for coronary heart disease [14], improvements in medication prescribing [10,15], improved lipid management [10,14,16], reduction in cholesterol level [16], improved blood pressure management [10], increased lifestyle advice provision [14] and reduced anxiety and depression [17].
However, with the exception of the PIER trial [16] which achieved significant reductions in cholesterol levels among intervention group patients, trials have not achieved significant improvements in objective biophysical risk factors, such as blood pressure or cholesterol levels.
A systematic review identified twelve randomised trials of disease management programmes for patients with established heart disease [18]. It concluded that such programmes do improve processes of care, reduce hospital admissions and enhance quality of life and functional status in patients with CHD. However, the reviewers noted that studies included imprecise descriptions of both the interventions and the usual care provided to the control groups, resulting in an inability to determine the incremental benefits of the various components of each intervention. Reviewers concluded that several important issues require further clarification, including the optimal mix of interventions and the cost-effectiveness or the economic impact of such interventions. Another earlier systematic review [19] also noted both the lack of detail provided of complex health service interventions, especially in the description of the usual care provided to control groups, and the need for the careful design and evaluation of different implementation models of secondary prevention.
Further possible explanations for the lower than expected clinical impact of secondary prevention interventions to date include inadequate consideration of doctors' and patients' perspectives on heart disease, lack of patient-centredness of the intervention and a failure to tailor practice interventions to individual patients and practices [13]. Wiles [20], utilising a qualitative approach as part of the SHIP study, especially emphasised the failure of 'official accounts' to acknowledge patient perceptions as a key feature in people's reluctance to adopt lifestyle change.
There is substantial evidence that structured systematic care is important to improve levels of secondary prevention of coronary heart disease, and an effective register, recall system and routine audit of care are essential components of such a system [13,21]. There is also some evidence to suggest that nurse-led clinics [10], cognitive behavioural interventions [17,22,23] and tailored training identified by practice staff [16] are effective in improving secondary prevention in general practice and facilitating patients to make lifestyle changes.
A recent Cochrane review [24] of interventions to implement prevention in primary care highlighted the need to tailor interventions to address specific barriers to change in a particular setting. The review also concluded that multi-faceted interventions may be more effective than single interventions, as more barriers to change can be addressed. The authors of the review recommend that future research should analyse barriers to change and report interventions to implement preventive services in more detail, to clarify how interventions relate to specific barriers. Since such complex interventions are likely to be more costly than single interventions, these reviewers also advise that economic evaluations should be included in future studies.
The SPHERE Study aims to take these recommendations on board in developing an intervention to improve secondary prevention of CHD in general practice in Ireland. The SPHERE study involves the implementation of a structured systematic programme of care for patients with CHD attending general practice. The SPHERE intervention is multi-faceted and has been developed to respond to the barriers and solutions to optimal secondary prevention identified in preliminary qualitative research with practitioners and patients. The delivery of the intervention is tailored as much as possible to the needs of practices and patients. The process of the intervention will be documented and described in sufficient detail to allow replication of the intervention and to enable identification of effective intervention components. The intervention will be implemented over a 24 month period, with data collection at baseline, 12 months and 24 months. Data collection at these time points will enable identification of any initial and/or sustained changes effected by the intervention. An economic evaluation of the intervention will be carried out. The implementation of the intervention will be monitored throughout the duration of the trial and documented.
In summary, SPHERE will add to what is already known in respect of secondary prevention in primary care, i.e.
• that structured programmes, with continued support and follow-up of patients, help improve recording of process measures of preventive care, patients' functional status, quality of life and lifestyle change [9-11,15,17,25].
• prompts to patients improve their attendance but there is no associated change in measures of risk [13,15].
• structured recall programmes which do not take account of individual practice needs improve recording but not prescribing [15]; provision of training, tailored to practice needs of access to evidence of effectiveness of secondary prevention is associated with improved prescribing [16]
• patient centred consultations impact positively on health outcomes [26] but must be accompanied by attention to disease management guidelines to achieve clinical objectives [27].
SPHERE will:
• record organisational detail of care, in both intervention and control practices
• provide information and training tailored to practices' needs for delivery of secondary prevention
• support patient centred consultations and practitioners' adherence to disease management guidelines.
Aim
To design, implement and test tailored practice and personal care plans to improve the process of care and objective clinical outcomes for patients with established coronary heart disease in general practice in Ireland.
Design
Cluster randomised controlled trial, with practice-level randomisation to intervention and control groups.
Random Selection of Practices and Patients
Practice identification (48 practices, 16 at each centre)
The same method is followed by each centre in Belfast, Dublin and Galway for their respective regions of the Eastern Health and Social Services Board, the South Western Area Health Board and the Western and North Western Health Boards.
Lists of all practices meeting eligibility criteria are generated by the research nurses in each centre (see 'practice eligibility' section below for criteria). Practices are then allocated a number and randomly selected from the lists in each region by an individual independent of the research team using computer-generated random numbers. The practice manager or lead general practitioner in each practice is contacted by telephone by the research nurse to determine their interest in receiving information about the trial. With their agreement, information is posted to the practice (see additional file 1: Appendix A) and arrangements are made for a member of the research team to visit the practice to discuss their participation in the study.
During this meeting the research nurse explains the study to practice staff and gets commitment to the study, indicated by a signature, from practice staff (at least one GP, one practice nurse and one administrator). A Practice Implementation Plan is completed (see additional file 1: Appendix A) and all practice staff are invited to review and approve it. Practices that agree to participate are offered an honorarium of €1000 or the Sterling equivalent upon completion of the study.
If selected practices decline to participate or confirm that they do not meet the eligibility criteria, the next practice on the list is selected. Relevant details (number of whole-time equivalent (WTE) practitioners, list size and reason for not being recruited or not agreeing to participate) are recorded for those practices which are not recruited.
Patient identification (20 patients in each practice)
Participating practices compile a list of all eligible patients (see 'patient eligibility' section below for detail) from their practice and allocate a number to each patient from 1 to Y (where Y is the total number of patients on the list). Research nurses assist the practices in using a list of random numbers to select patients for participation (see additional file 1: Appendix A). A confidentiality agreement will be signed between research nurses and practice staff, to ensure that normal ethical procedures governing each practice are adhered to. Selected patients are posted an invitation letter, information sheet and reply slip (see additional file 2: Appendix B) as well as a questionnaire (see additional file 3: Appendix C). Patients who do not respond within ten days are contacted by a member of the practice staff to ascertain interest in participation. If selected patients do not confirm that they fulfil the eligibility criteria or decline to participate, the next patient on the list is selected until the quota of 20 is reached. The age, gender and diagnostic inclusion criteria for those who do not participate are recorded. Patients who agree to participate are invited to attend the practice for a baseline data collection consultation.
Once sixteen practices have been recruited and the relevant patient baseline data collected, at each centre an individual independent of the research team using computer-generated random numbers will allocate, with prior stratification, practices to intervention or control groups. Stratification will be according to the WTE for practitioners (less-than-two and two-or-more). Patient identification and baseline data collection occur prior to practice randomisation to minimize potential bias introduced by researchers and practitioners being aware of their allocation to either the intervention or control group. For intervention practices, meetings will be arranged to develop tailored practice care plans, deliver appropriate training, and support practitioners in reviewing patients four monthly for two years. Control practices will be visited again after one year and after two years, at which time points follow-up data will be collected.
Practice and Patient Selection: Inclusion and Exclusion Criteria
Practice eligibility
Practices are eligible to take part if they:
• Have a practice nurse, involved in general patient care (to be confirmed at the initial practice visit).
• Did not participate in the pilot phase of the study.
For practices in the Republic of Ireland
• Are not participating in Heartwatch (a national pilot programme in the Republic of Ireland on the provision of secondary care in general practice, currently ongoing).
• Have a minimum General Medical Scheme (GMS) list size of 700 patients. Based on an Irish prevalence rate of 3.16% for CVD [7], this will ensure at least 22 eligible GMS patients per practice, a figure which will be supplemented by non-GMS patients.
For practices in the North of Ireland
• Have a minimum NHS list size of 1800 patients. Based on a UK prevalence rate of 2.6% for CVD [4], this will ensure at least 40 eligible patients per practice.
Patient eligibility
Inclusion criteria
Patients with existing cardiovascular disease (defined as: documented MI, CABG or angioplasty, or a diagnosis of angina – confirmed by exercise stress test, isotope test or coronary angiogram) are eligible for inclusion.
Exclusion criteria
Patients with significant mental or physical illness (based on the recorded judgement of practice staff), which is likely to impair capacity to change lifestyle behaviour or assimilate new information, are not included.
Sample Size Calculation
The study aims to achieve a final sample of 635 patients from at least 42 practices (see Figure 1).
Figure 1 Flow of Practices and Patients Throughout Trial.
Sample size calculations are based on the following:
• Power of 80% and alpha of 5%.
• Only patients agreeing to participate in the study and completing baseline data collection will be entered into the study. There will, therefore, be 100% uptake at the start of the study. Reasons for non-participation will be documented and recorded, and participation rates will provide a measure of the acceptability of this type of intervention to patients.
• A final completion rate of 70% is anticipated based on similar studies in the West of Ireland 7 (completion rate = 68.7%) and Dublin [28] (completion rate = 93%).
• Practice attrition of 12.5% has been allowed for (though practice attrition in another similar study was only 4% [28]).
• Improvements in the control group as a result of participation in a research study have been anticipated [25].
• Design effects are calculated using intracluster correlation coefficients from data from similar populations in Galway and Scotland (Campbell N, personal communication).
A total sample size of 907 patients recruited from 46 practices gives a varying but minimum power of 80% to detect significant changes in proportions with poor blood pressure and cholesterol control, physical and mental well being (SF-12) and hospital readmissions at two years. (Detailed power calculations are given below). This allows for a final patient follow-up and retention rate of 70% and a practice attrition rate of 10%. The number of practices is rounded up from 46 to 48, to allow equal numbers of practices to be recruited in each of the three study centres. Sixteen practices will therefore be recruited in each of the three centres, eight intervention practices and eight control practices. Twenty patients are recruited in each practice. Final data analysis will incorporate modelling techniques that are more statistically efficient and will ensure greater power to detect significant changes in study outcomes.
Power Calculations – Further Detail
Blood pressure
44% of Irish patients with established heart disease have a systolic blood pressure above recommended guidelines (SBP >140 mmHg) (Leahy J, personal communication). Previous research indicates that an improvement of 20% can be anticipated in the control group [10]. Taking this into consideration, a sample size of 408 gives 80% power to detect a 50% reduction in the proportion of patients with a SBP above 140 mmHg (i.e. giving a proportion of 22%) in the intervention group with a corresponding 20% improvement in the control group. Other studies have indicated that it is possible to achieve similar targets in general practice populations with established coronary heart disease [15]. To take account of cluster randomisation, based on recruiting 15 patients per cluster, a design effect size of 1.15 and an intracluster correlation coefficient of 0.011 [4,10] suggests a sample size of 470 patients from 31 practices. Allowing for a final response rate of 70% and a practice attrition rate of 10% requires recruitment of 670 patients from 34 practices.
Cholesterol
36% of Irish patients have a baseline cholesterol level above the upper recommended limit, i.e. greater than 5 mmol/l (Leahy J, personal communication). To demonstrate an improvement of 50% in the intervention group and 20% in the control group requires a sample size of 500 patients. To ensure a final sample of 15 patients per practice, this is inflated by a design effect size of 1.27 (ICC = 0.019; Campbell N, personal communication) to give a sample of 635 patients from 42 practices. This is increased to 907 patients from 46 practices to allow for final follow up and response rate of 70% and practice attrition of 10%.
SF-12
A 5 point improvement in the SF-36 has been found to be clinically significant [29]. Based on a mean at baseline of 53.98 (SD 8.39) [7], a sample size of 120 patients is sufficient to detect a clinically significant improvement of 5 points in the SF-12 measure of physical and mental health status in the intervention group. The intracluster correlation coefficient from previous data [7] is <0.0001, indicating that there is no clustering effect for this variable and that the design effect size is 0. Allowing for a patient retention rate of 70% and a 10% practice attrition rate, 170 patients from 10 practices are needed.
Hospital re-admissions
Northern data (Cupples M, personal communication) indicates that 43% of control patients had hospital admissions in two years. To reduce this by 50% in the intervention group and 20% in the control group needs a sample size of 356 patients. To ensure a final sample of 15 patients per practice, this is inflated by a design effect size of 1.08 (ICC = 0.006) to give a sample of 406 patients from 27 practices, increased to 580 patients from 30 practices to allow for final follow up and response rate of 70% and practice attrition of 10%.
The study acknowledges that smoking is a significant risk factor for cardiovascular disease. For pragmatic reasons smoking has not been used to power the study, as levels of smoking have become low among this population and its measurement would require a sample size too large for the resources available.
Intervention
Level 1: Tailored practice care plans
(1) Training for practice staff
Training will be delivered independently in each of the three regional study centres. All trainers will adhere to a single training protocol to ensure standardised delivery of the training across centres. Training delivery will be planned and rehearsed jointly by all trainers using role-play and peer review techniques. In addition, the project manager will act as an observer during the first two training sessions in each centre and will provide feedback to trainers with a view to further standardizing the training. The Irish College of General Practitioners (in the Republic of Ireland) and the Northern Ireland Medical and Dental Training Agency (in Northern Ireland) have approved the proposed training sessions for their educational contribution to practitioners' personal development plans.
First training session: Medication training
GPs and practice nurses are invited to attend a ninety-minute training session on medication guidelines, delivered by the study GP in each region:
• Belfast: Dr. Margaret Cupples, Department of General Practice, Queen's University Belfast.
• Dublin: Dr. Susan Smith, Department of Public Health and Primary Care, Trinity College Dublin.
• Galway: Professor Andrew Murphy, Department of General Practice, NUI Galway.
This is an interactive session in which practice staff are invited to review and discuss the most recent medication prescribing guidelines for secondary prevention. The session uses case-based scenarios to enable the practitioners to reflect on their own practice. Practitioners are offered a summary sheet of prescribing guidelines and a summary 'prompt' showing commonly used drugs and dosages within the various medication categories. If practices wish their practice nurse to make prescribing decisions the research nurse will help to prepare appropriate specific protocols.
The objectives of this session are:
• To increase confidence and competence regarding prescribing and the secondary prevention of heart disease.
• To discuss the role of the SIGN guidelines and apply to real clinical situations.
• To explore practitioner attitudes to secondary prevention and to discuss points raised specifically by each practice.
• To address issues of patient adherence to medication.
Second training session: Behaviour change training
GPs and practice nurses are invited to attend a ninety-minute training session on facilitating patients with behaviour change, delivered by the research nurse in each region. This is an interactive session in which practice staff are invited to reflect on their views around lifestyle change and practice new techniques (through role play) to improve their ability to facilitate patients with behaviour change. Strategies to increase patient motivation are discussed, following guidelines from brief motivational interviewing literature [30,31]. Additional techniques, based on principles of behaviour modification and social learning theory [32], include: setting small achievable goals, action planning, using prompts, self-monitoring, offering rewards, habit reinforcement, and relapse prevention.
The objectives of this session are:
• To enable practitioners to develop skills based on leading behaviour change theories, which may help them with behaviour change consultations.
• To introduce the SPHERE patient-held booklet and the patient care plan (see 'Level 2' below).
(2) Responding to individual practice needs
Additional practice needs in relation to the delivery of preventive care for patients with established cardiovascular disease, as well as the day-to day running of the study, are recorded on a tailored practice care plan and followed up by the SPHERE research nurse.
(3) Ongoing support from the SPHERE research nurse
The SPHERE research nurse maintains regular contact with the practices, and is easily contactable by phone if practice staff have any queries about SPHERE.
(4) SPHERE newsletter
Intervention practices receive a study newsletter which is published every four months. The newsletter contains the latest news and updates relating to the study.
Level 2: Tailored patient care plans
(1) Initial target setting consultations
The first patient intervention consultation with the GP/practice nurse takes place as soon as possible after baseline data collection, once training of practice staff has been completed. BP and BMI are measured, diet, smoking and exercise habits reviewed and the result of the baseline cholesterol assay discussed. The patient and practitioner together identify areas of management which could be improved and the patient is invited to prioritise one particular aspect of their lifestyle for change. Possible ways of achieving targets reflecting optimal management (BP <140/90 mmHg, cholesterol <5 mmol/l, BMI <25, non-smoking, taking exercise for 30 minutes 5 days per week, eating fruit/vegetables daily and avoiding saturated fat intake, avoidance of stress) are identified. Plans for action are recorded on a personal care plan tailored for each patient and retained in the practice. The GP may also see the patient at this visit or at a further appointment within 2 weeks to determine if any change in medication is appropriate. A further follow up telephone call is made by the practice nurse to give support and address any questions 2 weeks later.
(2) The patient-held booklet
This booklet contains information on all the key risk factors for CHD. It is used by the practitioners in all initial target setting discussions with the patient and is given to the patient to be kept by them. The booklet can be referred to at subsequent consultations with the GP or practice nurse for consolidation of information relevant to secondary prevention and cardiovascular risk. There are 6 sections in the book:
a) Medications
b) Smoking
c) Exercise
d) Healthy eating
e) Stress
f) Community support
By providing a summary of lifestyle advice for secondary prevention of CHD, the booklet serves not only as an information resource for patients but also as a reference guide to help practitioners in their lifestyle-related consultations with patients.
(3) Regular consultations with patients for two years
Patients are invited to attend for an appointment with the GP/nurse every 4 months. At each visit a review of factors relating to targets and goals for optimal secondary prevention are made as appropriate. Goals relating to diet, smoking and exercise habits are routinely reviewed. Cholesterol is checked if it has been found to be raised at baseline, but if normal at baseline cholesterol is checked annually. Relevant measurements taken at each review visit are recorded and held in the practice and reviewed with the patient at subsequent review visits. Care which would be given according to current clinical guidelines will also be provided, as required.
Control Practices
Data are collected for patients in intervention and control practices in identical ways. In control practices, data are collected at baseline, one and two years post baseline. Otherwise, patients in control practices continue to receive health care as usual and the nature of this in each participating practice will be clearly described.
Informed Consent
An information sheet is sent to all potentially eligible patients (see additional file 2: Appendix B). Patients are asked to sign and return, with their postal questionnaire, a reply slip (see additional file 2: Appendix B) to allow further contact regarding the research. At their first consultation for baseline data collection, the GP/practice nurse ensures their understanding of the study, allows them the opportunity to ask questions and confirms their willingness to participate in the study before asking them to sign a consent form (see additional file 2: Appendix B).
Ethical Approval
Ethical approval has been granted by the Ethics Committee of the Irish College of General Practitioners and the Queen's University Research Ethics Committee.
Data Collection
Data collection will involve three components:
• Questionnaire posted to patient from the practice with administrative assistance from the SPHERE research nurse. Questionnaire is posted back to the practice in a pre-paid envelope. The research nurses liaise with practice staff to follow up non-responders.
• Consultation data gathered by practice staff and inserted into one page SPHERE study baseline measurement record.
• Chart search to establish process of care and collect patient outcomes recorded in GP records. This is carried out by research nurse (after obtaining patient consent), and recorded onto SPHERE study database.
The above data will be collected at three time points:
1) Baseline
2) One year
3) Two years.
Data are collected at one year to assess the initial effects of the intervention. No results of any interim analyses will be revealed to any member of the research team unless the Data Monitoring Committee (see below) determine that it is unethical to continue the study.
Data to be collected are described below.
Questionnaire components (see additional file 3: Appendix C)
A. Physical and mental well-being: SF-12.
B. Exercise: Godin Leisure-Time Exercise Questionnaire.
C. Smoking status: adapted from SLÁN study [33].
D. Diet: DINE, including question regarding portions of fresh fruit and vegetables.
E. Personal and demographic variables:
a. Demographics: name, date of birth, marital status, educational level, occupation, GMS eligibility (for practices in the Republic of Ireland).
b. Economic analysis questions: journey time to GP and hospital.
c. Health service attendance in previous 12 months: including hospital outpatient services, admission to hospital as an inpatient, visits to Accident and Emergency department and days spent as inpatient.
F. Adherence to medication: Medication Adherence Report Scale MARS-5.
Consultation/physical assessment components (taken by practice nurse/GP) (see additional file 4: Appendix D)
1. Blood pressure.
2. Total serum cholesterol concentration (HDL and LDL).
3. Current medications, dose and contraindications.
4. Height in cm/weight in kg/BMI/waist circumference in cm/waist hip ratio.
Chart search data (collected by SPHERE research nurse) (see additional file 4: Appendix D)
1. Cardiac history: months since diagnosed with CHD, record of myocardial infarction and time since myocardial infarction, presence of angina, investigation confirming diagnosis, previous PTCA or CABG, presence of diabetes.
2. Number of GP and practice nurse consultations in last 12 months.
3. Number of attendances at outpatients/inpatient services in last 12 months.
4. Recording of risk factors in medical records: smoking, cholesterol and blood pressure.
Data Analysis
Description of study participants
Practices and patients participating in the SPHERE study will be heterogeneous. Participating practices will be described in terms of:
• Number and gender of full time equivalent GPs
• Number of practice nurses
• Numbers of support staff
• Practice list size – in the Republic of Ireland this will be presented as GMS list size and estimated non-GMS list.
Participating patients will be described in terms of:
• Age
• Gender
• Disease duration
• Nature of cardiovascular disease:
Angina
MI
• Procedures:
Angioplasty
CABG
• Socioeconomic status (according to occupation).
Randomised controlled trial analysis
Primary outcomes include:
• Blood pressure
• Total cholesterol
• Physical and mental status as measured by the SF-12
• Hospital re-admissions.
Secondary outcomes include:
• Measure of the process of care
SPHERE visits
Total number of GP visits
Total number of hospital OPD visits
Recording of risk factors in GP records
• Body Mass Index/Waist-Hip Ratio
• Exercise: Godin Leisure-Time Exercise Questionnaire
• Smoking status
• Diet: DINE, including question regarding portions of fresh fruit and vegetables
• Adherence to medication (Medication Adherence Report Scale MARS-5).
All results are analysed using SPSS and Stata statistical software. Statistical modelling is carried out using Stata statistical software which has a facility for complex survey sample analysis that allows for adjustments in data analysis based on design effects, planned or unplanned. Statistical significance will be taken at the 5% level for both primary and secondary outcomes. The analysis examines intervention vs control groups on study completion, as opposed to looking at changes in the intervention and control groups from baseline.
Sub-group analyses
Sub-group analyses will be carried out on the following groups:
• Angina only vs other diagnoses
• CABG vs angioplasty
• Diabetes vs non-diabetes
• Men vs women
• Northern Ireland vs Republic of Ireland
• Patients < 70 years vs older.
Analyses will be carried out using two approaches:
• By intention to treat i.e. including all randomised patients, regardless of their participation in the intervention.
• Sensitivity analyses will explore whether adherence to the intervention influences the effect of the intervention on primary outcomes.
Data management
The project manager has primary responsibility for the maintenance and management of the study database, which will be stored on a password-protected computer in a locked office. Research nurses have responsibility for ensuring the completeness, accuracy and confidential storage of data collected in their respective region from patient questionnaires, during the data collection consultations and from patient charts. Each practice and patient participating in the study has a unique identifying code.
Data will be collected and processed in the following ways:
1. Questionnaire data: once received in the practice, the research nurse collects the questionnaires and inputs the data into a master SPSS study database. Questionnaires are stored securely in their regional research centre.
2. Consultation data: once complete, the research nurse collects the data collection sheets from the practice and inputs the data into the master SPSS database. Data collection sheets are stored securely in their regional research centre.
3. Chart search data: the research nurse enters this data directly from patient notes onto an electronic study database created in Filemaker Pro software. Data can subsequently be exported directly into the master SPSS database using the unique patient identifier as the link field.
In the interests of data accuracy, numerical fields have range limiters to ensure that values outside a particular range cannot be entered; also, random data checks are performed on a regular basis to confirm completeness and accuracy of data. Inter-researcher data entry reliability is ensured by a process of trial data entry by the three research nurses of a sample of 20 questionnaires and consultation data sheets and 10 patient charts data early in the study data collection process. Where discrepancies in data entry are observed between the researchers, discussion will ensue to resolve any ambiguities or disagreement and clarify data collection and entry guidelines. Once all data has been completely entered and checked in the master SPSS database, preliminary analysis can be performed. The data will be sent in this format to the study statistician, who will import the data into Stata software to perform the necessary statistical modelling procedures.
Data monitoring
Interim results after one year will be presented to an independent data monitoring committee (DMC), made up of a statistician, an epidemiologist, and a clinician with statistical expertise. The role of the DMC will be to:
a) Ensure that no harm is being conducted to study participants
b) Act in an advisory capacity with respect to research governance.
Qualitative Evaluation
Introduction
Qualitative research will be conducted after years 1 and 2 of the intervention. It will facilitate evaluation of the intervention based on the experiences of practice staff and patients in delivering and receiving the intervention respectively. It also aims to compare their experiences with those of patients and practitioners in the control practices. The qualitative research will be integrated with the economic and policy evaluations.
Aims for intervention group (Year 1)
The aims of the qualitative evaluation study with the intervention group at the end of year 1 are to explore:
• How the intervention is implemented
• How the intervention is integrated with other practice activities
• How well practitioners and patients understand the intervention
• Whether elements of the intervention are particularly important or problematic
• Attitudes to the cost of implementing the intervention for practices and patients.
Aims for control group (Year 1)
The aims of the qualitative evaluation study with the control group at the end of year 1 are to explore:
• The nature of secondary prevention care implemented during the study period
• The impact of any policy changes during that time.
Aims for intervention group (Year 2)
The aims of the qualitative evaluation study with the intervention group at the end of year 2 are to explore:
• Changes in how the intervention is implemented and the reasons why
• Whether the aims of the intervention have been achieved
• The feasibility of the intervention being continued after the study ends
• Whether patients would be willing to pay for such a service.
Aims for control group (Year 2)
The aims of the qualitative evaluation study with the control group at the end of year 2 are to explore:
• Changes in how secondary prevention care is implemented and the reasons why
• The impact of any policy changes during that time.
Data collection and analysis
The qualitative research will be guided by the principles of the methodology of grounded theory. Practitioners and patients will be purposively sampled using theoretical and maximum variation sampling methods to ensure a variety of characteristics and experiences. Small and large practices that reflect a mixture of rural, urban, deprived and affluent locations will be selected.
The data will be collected using focus groups and individual semi-structured interviews. The focus groups aim to explore the diverse experiences of patients receiving the intervention. They will include men and women who vary in the nature of their heart disease, length of time since diagnosis, age and ethnic group (if possible). Participants will be invited to participate by letter which will explain the purpose of the interviews, followed by a telephone call from the practice nurse. Their reasons for not wishing to participate in the interviews will be recorded. The research nurses will act as observers of the focus group and will take notes of the discussion and map the interaction of the participants. Individual interviews with some patients will be used to clarify and to explore in more detail issues discussed in the focus groups. Free text comments in the patient questionnaire that forms part of the economic evaluation will also form part of the qualitative evaluation.
Individual interviews will also be conducted with GPs and practice nurses who are involved in implementing the intervention, with more interviews being conducted with the nurses, who tend to be responsible for the clinics. The interviews and focus groups will be audio-taped with the participants' consent and transcribed verbatim.
The interviews and focus groups will be divided between two qualitative researchers. The development of an interview schedule and role plays of the interviews by the researchers will be used to standardise the interviews and minimise the effects of the researchers on the data.
Data on the actual process of care will also be collected using overt observation methods by the research nurses. This will strengthen the validity of the interview data by comparing what practitioners actually do with what they say they do and will act as a quality control measure. Field notes of the research nurses' observations of the intervention based on their regular contacts with the practices will also form part of the qualitative data. These will be explored in more detail in a focus group with the research nurses.
The control practices will act as negative cases and will test whether the data from the intervention practices are unique to them or are shared with non-intervention practices. This will strengthen the reliability of the data from the intervention practices. The characteristics of the participants will be matched to the characteristics of participants in the intervention practices.
The data collection and analysis will be iterative and will continue until data saturation is achieved. Analysis will be conducted using the constant comparative method and will be facilitated by importing the interview transcripts into the computer software programme Nudist. A summary of the main findings will be sent to a sub-sample of patients and practitioners from the intervention and control groups in each of the 3 centres who were interviewed to validate our findings.
The number of interviews will be divided equally among the 3 centres. The number of interviews to be conducted among patients and practitioners is shown in Tables 1 and 2.
Table 1 Qualitative Research with Intervention Groups
Focus Groups Individual Interviews
Patient Focus Groups Research Nurse Focus Groups Total Focus Groups Patient Interviews Practice Nurse Interviews GP Interviews Total Interviews
Year 1 3 1 4 6 6 3 15
Year 2 3 1 4 6 6 3 15
Total 6 2 8 12 12 6 30
Table 2 Qualitative Research with Control Groups
Focus Groups Individual Interviews
Patient Focus Groups Total Focus Groups Practice Nurse Interviews GP Interviews Total Interviews
Year 1 3 3 3 3 6
Year 2 3 3 3 3 6
Total 6 6 6 6 12
Economic Evaluation
Introduction
The economic analysis incorporates both cost minimization analysis and cost effectiveness analysis for the various interventions for the two groups. The basic tasks of the evaluation are to identify, measure, value and compare the costs and outcomes of the alternatives being considered. Costs are likely to fall on patients, their families and the state. The health care resources consumed will reflect the costs of organising and operating the two programmes: intervention and control. These costs will reflect the time input of health professionals and fixed or overhead costs, including any new equipment and capital expenditure. All contacts with the health services are recorded and valued, including general practice contacts, hospital attendances, admissions and drug prescriptions. The patient and family costs include any out-of-pocket expenses and any own time input into the treatment process. Estimates will be made of work and leisure time foregone as part of the treatment alternatives, together with quality of life estimates. Cost-effectiveness analysis is the primary method of economic evaluation with measures of effectiveness taken from the range of primary and secondary outcomes being considered for the two groups.
In the Republic of Ireland, free healthcare is available to patients who qualify under the means-tested General Medical Scheme (GMS), although chronic disease management is not fully supported by this system. Non-GMS patients are required to pay for all of their own healthcare costs. In order that the findings of the SPHERE study will be translatable into practice within the existing healthcare systems, non-GMS patients in the Republic of Ireland will be required to pay for SPHERE-related visits to their general practice as normal. The cost of extra visits by GMS patients will be borne by the practices. The research team acknowledges that this may have an impact on recruitment of both practices and patients, although in a similar study the impact on patient recruitment was not substantial [28].
Aim
The aim of the economic analysis is to provide both cost minimization analysis and cost-effectiveness analysis of secondary prevention of heart disease in general practice. The cost effectiveness analysis provides information on the marginal costs and effects of the intervention relative to the alternative through the calculation of an incremental cost-effectiveness ratio. In situations where there is no significant difference in effects, the use of cost minimization analysis allows the reporting of cost differences only.
Data Collection
The following items of data are recorded to allow reporting of health services costs:
• Number of visits to GP in last 12 months. It was decided not to record length of consultation, but to use average length from other research in calculations.
• Number of visits to practice nurse. As length of practice nurse consultations vary, an average consultation length will be estimated by research nurses during quality control visits.
• Use of hospital services: (1) Number of attendances to outpatients in last year; (2) Number of days spent as inpatient in hospital in last year; (3) Total number of days spent in hospital in last year; (4) Number of attendances to A&E in last year.
• Current prescription of medications: type and dose, using Defined Daily Dose methodology.
• The study is not in a position to record patients' use of community services, such as physiotherapy or occupational therapy.
The following items of data are recorded to allow reporting of patient costs:
• Occupation data.
• Travel distance, length, mode and waiting time for visits to the practice and the hospital.
• In the Republic of Ireland, GMS status.
• Patient-estimated additional expenditure or savings made in the process of lifestyle change as a result of the intervention, e.g. changes to shopping bill with changed diet, paying for exercise facilities, giving up smoking.
The following items of data are recorded to allow reporting of capital costs:
• Any new equipment purchased by/for the practice to carry out the intervention.
The following items of data are recorded to allow reporting of intervention costs:
• All expenses associated with the intervention, such as cost of training, information and education materials.
Outcome measures
Both primary and secondary outcome measures are collected as part of the randomised controlled trial analysis. The cost-effectiveness analysis will be based primarily on intermediate endpoints. However, some intermediate endpoints, such as blood pressure, are uni-dimensional measures and may or may not have a high correlation with a final endpoint, such as life years gained. Hence the need to explore the estimation of final endpoints in the analysis, where possible, thereby allowing much greater scope to make comparisons of cost-effectiveness with other studies.
Analysis
The analysis will use standard economic analysis for the calculation of costs and incremental cost-effectiveness ratios. Unit costs will be applied to the resource use data to calculate the various costs of care. Comparisons between the intervention and control groups will be made on the various primary and secondary outcomes and costs will be assigned accordingly. The analysis will allow us to consider whether significant differences emerge between intervention and control groups in terms of the various outcome measures. We will examine whether the intervention is associated with overall cost increases or cost decreases and link the cost changes to incremental gains in effectiveness, where they exist.
Policy Evaluation
A policy analysis will be carried out in the context of the economic analysis described above. There are relatively few comprehensive evaluations of health policies and even fewer that use or incorporate an experimental design as part of the appraisal process. It is expected increasingly that rigorous, replicable, relevant and independent research should be available to make a contribution to evidence-based policy [34].
A policy analysis is the process through which we identify and evaluate policies and programmes that are intended to lessen or resolve health, social, economic and physical problems relative to secondary prevention of cardiovascular disease. The proposed analysis is specific to policy circumstances in Northern Ireland and the Republic of Ireland, but its outcomes will be generalisable since these policies are examples of a wider policy movement that has affected other industrialised countries in similar ways. General conclusions can be drawn from case studies as long as data about context, processes and outcomes are collected and this study will collect such data.
The best way to assess the implications of any policy is to identify, quantify and value systematically the costs and benefits of the proposed policy. In brief, the policy appraisal process involves the following stages: defining the policy objectives; identifying the policies or programme options; identifying and measuring the costs and benefits associated with each option; identifying and assessing uncertainties; and assessing the balance between options [35].
The 5-year research programme will provide a comparative, cross-health system analytical framework with which to investigate and to evaluate the major policies relating to heart disease as articulated in Building Healthier Hearts [2] and in the National Service Framework [36], respectively. The two systems differ with respect to, for example, funding and organisational structures yet experience similar problems such as waiting lists and staff shortages. Using this framework, the policy analysis will begin by detailing and critically appraising the context, nature and distribution of heart disease in each system as well as the consequential policies and service responses. The analysis of documentary material and secondary data will be supplemented with interviews with key 'stakeholders' in the policy process. This part of the policy analysis will lead to, among other things, the identification and specification of policy goals and evaluative criteria as well as an account of the way in which the research programme will provide a 'test' and progress marker of policy objectives and policy implementation. The next step in accomplishing a thorough policy analysis is to describe and evaluate how the policy in the form of the proposed secondary prevention services in each system benefits the previously established criteria. The experimental, quantitative data relating to outcomes together with the conceptual, qualitative data about organisational and implementation processes plus the results of the economic appraisal will be analysed in an integrative way so as to evaluate and compare comprehensively the policies in each system – at least at a micro-level. This incremental, step-by-step approach plus the iterative nature of the analytical process are important because they improve the chances of producing a rigorous, context sensitive policy analysis. The research team individually and collectively will revise and deepen earlier levels of analysis as new data become available and new perspectives or interpretations emerge.
In conclusion, the comparative analytical framework will provide important insights and lessons about the process of policy development and implementation as well as about the pace and degree to which the policy in practice impacts in terms of the costs and benefits of producing "healthier hearts".
Publications Planned
Qualitative
• A qualitative study of the provision of secondary prevention care for established heart disease in general practice in two different health systems.
• A descriptive analysis of the development of a multidisciplinary intervention to improve secondary prevention care for established heart disease in general practice.
• A qualitative study of practitioners' experiences of a randomised controlled trial of a multidisciplinary intervention to improve secondary prevention care for established heart disease in general practice in two different health systems.
• A qualitative study of patients' responses to a multidisciplinary intervention to improve secondary prevention care for established heart disease in general practice in two different health systems.
Main trial
• A baseline study of the provision of secondary prevention care for established heart disease in general practice in two different health systems.
• A randomised controlled trial of a multi-faceted intervention to improve secondary prevention of cardiovascular disease in general practice.
• Differences between large and small practices in a randomised controlled trial of an intervention to improve secondary prevention of cardiovascular disease in general practice.
Economic analysis
• An economic analysis of an intervention to improve secondary prevention of heart disease in general practice.
Policy analysis
• A policy analysis of an intervention to improve secondary prevention of heart disease in general practice in two different health systems.
Other
• The impact of the process of care on outcomes of an intervention to improve secondary prevention of heart disease in general practice.
• The experience of implementing a randomised controlled trial to improve secondary prevention of heart disease in general practice in two different health systems.
Competing interests
Pfizer Healthcare funds an annual teaching conference at the Department of General Practice, NUI Galway. MC is a member of Ezetrol and Inegy Primary Care Advisory Boards, funded by the Merck Sharpe Dohme /Schering Plough Partnership.
Authors' contributions
AWM, MC, SS and MB conceived of the study and coordinated the application for funding. AWM, MC, SS, MB and CL contributed to the design of the study, drafted sections of the manuscript, and read and approved the final manuscript. SS calculated the sample size. MCB coordinated the final stages of manuscript preparation and revision.
Supplementary Material
Additional File 1
Appendix A: Practice Recruitment Documentation
Click here for file
Additional File 2
Appendix B: Patient Recruitment Documentation
Click here for file
Additional File 3
Appendix C: Patient Questionnaire
Click here for file
Additional File 4
Appendix D: Other Baseline Data Collection Documentation
Click here for file
Acknowledgements
The SPHERE Study is funded by a Programme Grant from the Health Research Board (Ireland).
Our thanks to the other members of the SPHERE Steering Committee: Dr. Mairead Corrigan, Dr. Michael Donnelly, Mr. Paddy Gillespie, Ms. Ailish Houlihan, Dr. Alan Kelly, Ms. Mary O'Malley, Dr. Eamon O'Shea.
==== Refs
Scottish Intercollegiate Guideline Network Secondary Prevention of Coronary Heart Disease Following Myocardial Infarction 2000 Edinburgh: SIGN
Cardiovascular Health Strategy Group Building Healthier Hearts 1999 Dublin: The Stationery Office
De Backer G Ambrosioni E Borch-Johnsen K Brotons C Cifkova R Dallongeville J European guidelines on cardiovascular disease prevention in clinical practice European Heart Journal 2003 24 17 1601 1610 12964575 10.1016/S0195-668X(03)00347-6
Campbell NC Thain J Deans HG Ritchie LD Rawles JM Secondary prevention in coronary heart disease: baseline survey of provision in general practice British Medical Journal (Clinical Research Ed) 1998 316 7142 1430 4 9572757
Svilaas N Thoresen M Kristoffersen JE Hjartaaker J Westheim A How well are patients with atherosclerotic disease treated? Secondary prevention in primary care Scandinavian Journal of Primary Health Care 2000 18 232 6 11205092 10.1080/028134300448805
Kahan T Wandell P Risk factors in established coronary heart disease: evaluation of a secondary prevention programme Journal of Cardiovascular Risk 2001 8 2 73 80 11324373 10.1097/00043798-200104000-00003
Byrne M Murphy AW Secondary prevention of heart disease: a baseline survey of patients' lifestyles and service provision in the North Western and Western Health Boards Research and Development Report Number 2 2002 Galway: Department of General Practice
Connolly P Cupples M Cuene-Grandidier H Johnston D Passmore P The importance of validating the diagnosis of coronary heart disease when measuring secondary prevention: a cross sectional study in general practice Pharmacoepidemiology and Drug Safety 2002 11 311 7 12138599 10.1002/pds.709
Cupples ME McKnight A Randomised controlled trial of health promotion in general practice for patients at high cardiovascular risk British Medical Journal 1994 309 6960 993 6 7950723
Campbell NC Ritchie LD Thain J Deans HG Rawles JM Squair JL Secondary prevention in coronary heart disease: a randomised trial of nurse led clinics in primary care Heart (British Cardiac Society) 1998 80 5 447 52 9930042
Murchie P Campbell NC Ritchie LD Simpson JA Thain J Secondary prevention clinics for coronary heart disease: four year follow up of a randomised controlled trial in primary care British Medical Journal 2003 326 7380 84 12521974
VanDrenth BB Hulscher MEJL Nokkink HGA van de Lisdonk EH van der Wouden JC Grol RPTM Effects of outreach visits by trained nurses on cardiovascular risk factor recording in general practice European Journal of General Practice 1997 3 90 5
Jolly K Bradley F Sharp S Randomised controlled trial of follow up care in general practice of patients with myocardial infarction and angina: final results of the Southampton heart integrated care project (SHIP). The SHIP Collaborative Group British Medical Journal 1999 318 7185 706 11 10074017
Feder G Griffiths C Eldridge S Spence M Effect of postal prompts to patients and general practitioners on the quality of primary care after a coronary event (POST): randomised controlled trial British Medical Journal 1999 318 7197 1522 6 10356008
Moher M Yudkin P Wright L Cluster randomised controlled trial to compare three methods of promoting secondary prevention of coronary heart disease in primary care British Medical Journal 2001 322 7298 1338 11387182
Langham J Tucker H Sloan D Pettifer J Thom S Hemingway H Secondary prevention of cardiovascular disease: a randomised controlled trial of training in information management, evidence-based medicine, both or neither: the PIER trial British Journal of General Practice 2002 52 818 24 12392122
Lewin RJP Furze G Robinson J A randomised controlled trial of a self-management plan for patients with newly diagnosed angina British Journal of General Practice 2002 52 194 201 12030661
McAlister FA Lawson FME Teo KK Armstrong PW Randomised trials of secondary prevention programmes in coronary heart disease: Systematic review British Medical Journal 2001 323 7319 957 62 11679383
Ebrahim S Smith GD Systematic review of randomised controlled trials of multiple risk factor interventions for preventing coronary heart disease British Medical Journal 1997 314 7095 1666 74 9193292
Wiles R Patients' perceptions of their heart attack and recovery: the influence of epidemiological "evidence" and personal experience Social Science and Medicine 1998 46 11 1477 86 9665577 10.1016/S0277-9536(97)10140-X
Begg A Tackling the clinical indicators: secondary prevention of CHD Guidelines in Practice 2003 6 9 45 51
Steptoe A Doherty S Rink E Kerry S Kendrick T Hilton S Behavioural counselling in general practice for the promotion of healthy behaviour among adults at increased risk of coronary heart disease: randomised trial British Medical Journal 1999 319 7215 943 7 10514155
Steptoe A Perkins-Porras L McKay C Rink E Hilton S Cappucio FP Behavioural counselling to increase consumption of fruit and vegetables in low income adults: randomised trial British Medical Journal 2003 326 855 61 12702620
Hulscher M Wensing M van der Weijen T Grol R Interventions to implement prevention in primary care (Cochrane review) The Cochrane Library 2001 1 John Wiley and Sons, Ltd 11279688
Cupples ME McKnight A Five year follow up of patients at high cardiovascular risk who took part in a randomised controlled trial of health promotion British Medical Journal 1999 319 7211 687 8 10480826
Stewart M Effective physician-patient communication and health outcomes: a review Canadian Medical Association Journal 1995 152 1423 33 7728691
Kinmonth AL Woodcock A Griffin S Spiegal N Campbell MJ Randomised controlled trial of patient centred care of diabetes in general practice: impact on current wellbeing and future disease risk. The Diabetes Care From Diagnosis Research Team British Medical Journal 1998 317 7167 1202 8 9794859
Smith S Byrne G Shannon WF Thompson C O'Leary M Staines A Tynan A The North Dublin Diabetes Shared Care (DiSC) Project: A profile of current diabetes care in Ireland Irish Medical Journal 2001 94 8 240 243 11758626
Ware JE SF-36 health survey – manual and interpretation guide 1993 Boston: Nimrod Press
Emmons KM Rollnick S Motivational interviewing in health care settings: Opportunities and limitations American Journal of Preventive Medicine 2001 20 1 68 74 11137778 10.1016/S0749-3797(00)00254-3
Rollnick S Kinnersley P Stott N Methods of helping patients with behaviour change British Medical Journal 1993 307 6897 188 90 8343750
Bandura A Social Learning Theory 1977 London: Prentice Hall
Kelleher C Nic Gabhainn S Friel S Corrigan H Nolan G Sixsmith J The National Health and Lifestyle Surveys 2003 Dublin: Health Promotion Unit, Department of Health and Children
Economic and Social Research Council Centre for Evidence Based Policy and Practice
Policy Planning Research Unit Policy Evaluation: Occasional Paper no 32 1996 Belfast: Department of Health and Social Services
National Health Service National Service Framework for Coronary Heart Disease 2000 London
|
16053525
|
PMC1208929
|
CC BY
|
2021-01-04 16:47:32
|
no
|
Curr Control Trials Cardiovasc Med. 2005 Jul 29; 6(1):11
|
utf-8
|
Curr Control Trials Cardiovasc Med
| 2,005 |
10.1186/1468-6708-6-11
|
oa_comm
|
==== Front
Filaria JFilaria Journal1475-2883BioMed Central London 1475-2883-4-61602272810.1186/1475-2883-4-6Short PaperMass ivermectin treatment for Onchocerciasis: Lack of evidence for collateral impact on transmission of Wuchereria bancrofti in areas of co-endemicity Richards FO [email protected] A [email protected] D [email protected] A [email protected] A [email protected] JOA [email protected] MY [email protected] ES [email protected] The Carter Center, One Copenhill, Atlanta GA 30307, USA2 Department of Zoology, University of Jos, Jos, Plateau State, Nigeria3 Room 913, Phase II, Federal Secretariat, Federal Ministry of Health, Lagos, Nigeria2005 15 7 2005 4 6 6 16 8 2004 15 7 2005 Copyright © 2005 Richards et al; licensee BioMed Central Ltd.2005Richards et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
There has long been interest in determining if mass ivermectin administration for onchocerciasis has 'unknowingly' interrupted lymphatic filariasis (LF) transmission where the endemicity of the two diseases' overlaps. We studied 11 communities in central Nigeria entomologically for LF by performing mosquito dissections on Anopheline LF vectors. Six of the communities studied were located within an onchocerciasis treatment zone, and five were located outside of that zone. Communities inside the treatment zone had been offered ivermectin treatment for two-five years, with a mean coverage of 81% of the eligible population (range 58–95%). We found 4.9% of mosquitoes were infected with any larval stage of W. bancrofti in the head or thorax in 362 dissections in the untreated villages compared to 4.7% infected in 549 dissections in the ivermectin treated villages (Mantel-Haenszel ChiSquare 0.02, P = 0.9). We concluded that ivermectin annual therapy for onchocerciasis has not interrupted transmission of Wuchereria bancrofti (the causative agent of LF in Nigeria).
==== Body
Findings
Ivermectin is an effective microfilaricidal oral medication that is being distributed in mass drug administration programmes for two filarial diseases, onchocerciasis [1] and lymphatic filariasis (LF) [2,3]. Both onchocerciasis and LF are vector borne, with onchocerciasis transmitted by Simulium black flies, and LF by Anopheline mosquitoes in rural Africa. Merck and Co. donates ivermectin (Mectizan®) to global control programmes for both these parasitic diseases, although annual ivermectin in combination with albendazole (donated by GlaxoSmithKline) is recommended by WHO for the treatment of LF in Africa, because of the presumed synergy [4,5], although this remains in debate [6].
Of the two initiatives, the oldest is that for onchocerciasis and ivermectin has been distributed in annual ivermectin monotherapy (150 micrograms/kg) programmes in Africa for over 16 years [1]. There has long been interest in determining if such ivermectin distribution for onchocerciasis has 'unknowingly' interrupted LF transmission where the endemicity of the two diseases' overlaps [7]. We had occasion to address this question in central Nigeria in 1999 while conducting Anopheline entomological sampling for LF in and outside of onchocerciasis programme zones.
The study was performed in Plateau and Nasarawa States, Nigeria, as part of an integrated onchocerciasis, schistosomiasis, and lymphatic filariaisis programme described by Hopkins et al. [8]. Twelve of the 30 local government areas (LGA) in these two states are onchocerciasis treatment zones and have been offered annual ivermectin monotherapy since 1993. LF mapping in 1998 designated all 30 LGA for combined ivermectin and albendazole mass treatment for LF. Prior to launching the larger LF treatment programme, we sought (in 1999) entomology sentinel sites for a longitudinal study of treatment impact on LF transmission [9]. To identify villages with high baseline infection rates, our team captured and dissected resting Anopheles gambiae sl and An. funestus in randomly selected households in 11 villages, 5 of which were outside of the onchocerciasis ivermectin treatment zone, and 6 were inside the treatment zone. Treatment coverage for those six ivermectin treatment villages during the years 1995–1999 ranged from 58–95% of the eligible population (Table 1) with a mean of 81%.
Table Ivermectin treatment coverage of the eligible population (1995–1999) and LF antigenemia (1999) among male residents in five ivermectin treated villages, with 1999 LF antigenemia in one untreated village (Gwamlar)
Village Angwan Lemu Apanda Bakin-Kogi Lankan Mungkohot Gwamlar
Ivermectin rounds 2 2 2 5 5 -
Mean coverage (range) 91.4% (91–92) 66.9% (58–76) 85.3% (82–89) 85.9% (65–95) 77% (73–80) -
1999 coverage 92% 58% 89% 90% 80% -
% LF antigenemia in males (n) 40% (30) 43% (30) 27% (30) 47% (30) 47% (30) 58% (50)
After obtaining permission from local village chiefs and residents of the selected household, trained collectors used aspirators and torches to capture indoor resting Anopheline mosquitoes; 75% of these were Anopheles gambiae sl, the remainder were An.funestus. The mosquitoes, most of which were blood fed, were immediately transferred to screened paper cups and kept alive in an ice chest containing wet towels until dissected later that same day. At that time the mosquitoes were killed, placed on a glass slide, separated into head, abdomen and thorax, teased apart in normal saline, and examined under a binocular microscope. Infection rates were based on the finding of any larval stage of W. bancrofti in head or thorax. Microfilaria in the abdomen where not considered in the infection rate calculations.
LF antigenemia testing occurred on a separate occasion using the rapid ICT card test [10] (AMRAD Corporation Ltd., North South Wales, Australia). The test was performed as described by Eigege [11] on finger stick blood samples of 30 randomly selected adult male residents from five of the six treated villages and in 50 individuals in one of the six untreated villages (Gwamlar).
We found that the untreated village of Gwamlar, had both the highest mosquito infection rate (20%) and the highest antigenemia rate (58%). However, no statistically significant entomological differences could be demonstrated between the villages in treated and untreated zones (Figure 1): 4.9% of mosquitoes were infected in 362 dissections in the untreated villages compared to 4.7% infected in 549 dissections in the ivermectin treated villages (Mantel-Haenszel ChiSquare 0.02, P = 0.9).
In contrast LF antigenemia (Table 1) was less common among the 150 adult residents examined in the ivermectin treated villages (mean 41%, village range 27–47%) compared to the untreated village of Gwamlar, having the forementioned 58% antigenemia prevalence (ChiSquare 4.5, P = 0.03).
We conclude therefore, that ivermectin monotherapy for onchocerciasis has not been sufficient to interrupt transmission of LF in central Nigeria. Among treated villages, mosquito infection rates in treated and untreated areas were statistically equivalent, and antigenemia rates in treated villages were unacceptably high (although lower than those in Gwamlar). Mosquito infection rates were indeed highest in the two villages (Lankan and Mungkohot) with the longest treatment history (5 years) with adequate coverage. Our conclusion is in support of the findings of Kyelem et al., [7] who, working in Burkina Faso, demonstrated that ivermectin monotherapy given twice per year for onchocerciasis reduced but did not interrupt LF transmission there.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Drs. Richards, Eigege, Jinadu, and Miri and Professor Oneyka were involved in the design, supervision, analysis and preparation of the manuscript. Mr. Pam and Mr. Kal supervised the fieldwork and performed the dissections, under the field supervision of Professor Oneyka. Ms Lenhart played a major role in data analysis.
Acknowledgements
We would like to thank Dr. T. Lehman for his assistance in launching the fieldwork for this study. This project was carried out with support from SmithKline Beecham (now GlaxoSmithKline). Mectizan® was donated by Merck & Co., and treatment activities in the onchocerciasis zones have been supported with grants from the River Blindness Foundation, Lions Clubs SightFirst Program, and the African Programme for Onchocerciasis Control.
==== Refs
Richards FO Boatin B Sauerbrey M Sékétéli A Control of Onchocerciasis Today: Status and Challenges Trends Parasitol 2001 17 558 563 11756018 10.1016/S1471-4922(01)02112-2
Brown KR Ricci FM Ottesen EA Ivermectin: effectiveness in lymphatic filariasis Parasitology 2000 121 S133 146 11386685 10.1017/S0031182000006570
Molyneux DH Zagaria N Lymphatic filariasis elimination: progress in global programme development Annals Trop Med Parasitol 2002 96 S15 40 10.1179/000349802125002374
Addiss DG Beach MJ Streit TG Lutwick S LeConte FH Lafontant JG Hightower AW Lammie PJ Randomised placebo-controlled comparison of ivermectin and albendazole alone and in combination for Wuchereria bancrofti microfilaraemia in Haitian children Lancet 1997 350 480 484 9274584 10.1016/S0140-6736(97)02231-9
Ottesen EA Ismail MM Horton J The role of albendazole in programmes to eliminate lymphatic filariasis Parasitol Today 1999 15 382 386 10461168 10.1016/S0169-4758(99)01486-6
Dunyo SK Nkrumah FK Simonsen PE Single-dose treatment of Wuchereria bancrofti infections with ivermectin and albendazole alone or in combination: evaluation of the potential for control at 12 months after treatment Trans R Soc Trop Med Hyg 2000 94 437 443 11127253 10.1016/S0035-9203(00)90135-4
Kyelem D Sanou S Boatin B Medlock J Coulibaly S Molyneux DH Impact of long-term ivermectin (Mectizan) on Wuchereria bancrofti and Mansonella perstans infections in Burkina Faso: strategic and policy implications Ann Trop Med Parasitol 2003 97 827 838 14754495 10.1179/000349803225002462
Hopkins DR Eigege A Miri ES Gontor I Ogah G Umaru J Gwomkudu CC Mathai W Jinadu MY Amadiegwu S Oyenekan OK Korve K Richards FO Lymphatic filariasis elimination and schistosomiasis control in combination with onchocerciasis control in Nigeria Am J Trop Med Hyg 2002 67 266 272 12408665
Richards FO JrPam DD Kal A Gerlong GY Onyeka J Sambo Y Danboyi J Ibrahim B Terranella A Kumbak D Dakul A Lenhart A Rakers L Umaru J Amadiegwu S Withers PC JrMafuyai H Jinadu MY Miri ES Eigege A Significant decrease in the prevalence of Wuchereria bancrofti infection in anopheline mosquitoes following the addition of albendazole to annual, ivermectin-based, mass treatments in Nigeria Ann Trop Med Parasitol 2005 99 155 64 15814034 10.1179/136485905X19838
Weil GJ Lammie PJ Weiss N The ICT Filariasis Test: A rapid-format antigen test for diagnosis of bancroftian filariasis Parasitology Today 1997 13 401 404 15275155 10.1016/S0169-4758(97)01130-7
Eigege A Richards F Blaney D Miri E Umaru J Jinadu M Mathai W Hopkins D Rapid assessment for lymphatic filariasis in central Nigeria: a comparison of the immunochromatographic card test and hydrocele rates in an area of high endemicity Am J Trop Med Hyg 2003 68 643 646 12887020
|
16022728
|
PMC1208930
|
CC BY
|
2021-01-04 16:38:25
|
no
|
Filaria J. 2005 Jul 15; 4:6
|
utf-8
|
Filaria J
| 2,005 |
10.1186/1475-2883-4-6
|
oa_comm
|
==== Front
Global HealthGlobalization and Health1744-8603BioMed Central London 1744-8603-1-141607898910.1186/1744-8603-1-14ResearchThe health impacts of globalisation: a conceptual framework Huynen Maud MTE [email protected] Pim [email protected] Henk BM [email protected] International Centre for Integrative Studies (ICIS), Maastricht University, Maasticht, The Netherlands2 Faculty of Natural Sciences, Open University, Heerlen, The Netherlands3 Zuyd University, Heerlen, The Netherlands4 Netherlands Environmental Assessment Agency (MNP), Bilthoven, the Netherlands2005 3 8 2005 1 14 14 31 1 2005 3 8 2005 Copyright © 2005 Huynen et al; licensee BioMed Central Ltd.2005Huynen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper describes a conceptual framework for the health implications of globalisation. The framework is developed by first identifying the main determinants of population health and the main features of the globalisation process. The resulting conceptual model explicitly visualises that globalisation affects the institutional, economic, social-cultural and ecological determinants of population health, and that the globalisation process mainly operates at the contextual level, while influencing health through its more distal and proximal determinants. The developed framework provides valuable insights in how to organise the complexity involved in studying the health effects resulting from globalisation. It could, therefore, give a meaningful contribution to further empirical research by serving as a 'think-model' and provides a basis for the development of future scenarios on health.
==== Body
Introduction
Good health for all populations has become an accepted international goal and we can state that there have been broad gains in life expectancy over the past century. But health inequalities between rich and poor persist, while the prospects for future health depend increasingly on the relative new processes of globalisation. In the past globalisation has often been seen as a more or less economic process. Nowadays it is increasingly perceived as a more comprehensive phenomenon, which is shaped by a multitude of factors and events that are reshaping our society rapidly. This paper describes a conceptual framework for the effects of globalisation on population health. The framework has two functions: serving as 'think-model', and providing a basis for the development of future scenarios on health.
Two recent and comprehensive frameworks concerning globalisation and health are the ones developed by Woodward et al. [1], and by Labonte and Togerson [2]. The effects that are identified by Woodward et al. [1] as most critical for health are mainly mediated by economic factors. Labonte and Torgerson [2] primarily focus on the effects of economic globalisation and international governance. In our view, however, the pathways from globalisation to health are more complex. Therefore, a conceptual framework for the health effects of the globalisation process requires a more holistic approach and should be rooted in a broad conception of both population health and globalisation. The presented framework is developed in the following three steps: 1) defining the concept of population health and identifying its main determinants, 2) defining the concept of globalisation and identifying its main features and 3) constructing the conceptual model for globalisation and population health.
Population health
As the world around us is becoming progressively interconnected and complex, human health is increasingly perceived as the integrated outcome of its ecological, social-cultural, economic and institutional determinants. Therefore, it can be seen as an important high-level integrating index that reflects the state-and, in the long term, the sustainability-of our natural and socio-economic environments [3]. This paper primarily focuses on the physical aspects of population health like mortality and physical morbidity.
Our identification of the most important factors influencing health is primarily based on a comprehensive analysis of a diverse selection of existing health models (see Huynen et al [4] for more details). We argue that the nature of the determinants and their level of causality can be combined into a basic framework that conceptualises the complex multi-causality of population health. In order to differentiate between health determinants of different nature, we will make the traditional distinction between social-cultural, economic, environmental and institutional factors. These factors operate at different hierarchical levels of causality, because they have different positions in the causal chain. The chain of events leading to a certain health outcome includes both proximal and distal causes; proximal factors act directly to cause disease or health gains, and distal determinants are further back in the causal chain and act via (a number of) intermediary causes [5]. In addition, we also distinguish contextual determinants. These can be seen as the macro-level conditions shaping the distal and proximal health determinants; they form the context in which the distal and proximal factors operate and develop.
Subsequently, a further analysis of the selected health models and an intensive literature study resulted in a wide-ranging overview of the health determinants that can be fitted within this framework (Figure 1 and Table 1). We must keep in mind, however, that determinants within and between different domains and levels interact along complex and dynamic pathways to 'produce' health at the population level. Additionally, health in itself can also influence its multi-level, multi-nature determinants; for example, ill health can have a negative impact on economic development.
Figure 1 Multi-nature and multi-level framework for population health.
Table 1 Determinants of population health
Level/ Nature General determinants More detailed determinants
Contextual level
Institutional Institutional infrastructure Governance structure
Political environment
System of law
Regulation
Economic Economic infrastructure Occupational structure
Tax system
Markets
Social-cultural Culture Religion
Ideology
Customs
Population Population size
Structure
Geographical distribution
Social infrastructure Social organisation
Knowledge development
Social security
Insurance system
Mobility and communication
Environmental Ecological settings Ecosystems
Climate
Distal level
Institutional Health policy Effective public health policy
Sufficient public health budget
Health-related policies Effective food policy
Effective water policy
Effective social policy
Effective environmental policy
Economic Economic development Income/wealth
Economic equity
Trade Trade in goods and services
Marketing
Social-cultural Knowledge Education and literacy
Health education
Technology
Social interactions Social equity
Conflicts
Travel and migration
Environmental Ecosystem goods and services Habitat
Information
Production
Regulation
Proximal level
Institutional Health services Provision of and access to health services
Economic - -
Social-cultural Lifestyle Healthy food consumption patterns
Alcohol and tobacco use
Drug abuse
Unsafe sexual behaviour
Physical activity
Stress coping
Child care
Lifestyle related endogen factors (blood pressure, obesity, cholesterol levels)
Social environment Social support and informal care
Intended injuries and abuse/violence
Environmental Food and water Sufficient quality
Sufficient quantity
Sanitation
Physical environment Quality of the living environment (biotic, physical and chemical factors)
Unintended injuries
Globalisation
There is more and more agreement on the fact that globalisation is an extremely complex phenomenon; it is the interactive co-evolution of multiple technological, cultural, economic, institutional, social and environmental trends at all conceivable spatiotemporal scales. Hence, Rennen and Martens [6] define contemporary globalisation as an intensification of cross-national cultural, economic, political, social and technological interactions that lead to the establishment of transnational structures and the global integration of cultural, economic, environmental, political and social processes on global, supranational, national, regional and local levels. Although somewhat complex, this definition is in line with the view on globalisation in terms of deterritorialisation and explicitly acknowledges the multiple dimensions involved.
However, the identification of all possible health effects of the globalisation process goes far beyond the current capacity of our mental ability to capture the dynamics of our global system; due to our ignorance and interdeterminacy of the global system that may be out of reach forever [7]. In order to focus our conceptual framework, we distinguish-with the broader definition of globalisation in mind-the following important features of the globalisation process: (the need for) new global governance structures, global markets, global communication and diffusion of information, global mobility, cross-cultural interaction, and global environmental changes (Table 2) (see Huynen et al. [4] for more details).
Table 2 Features of globalisation
New global governance structure Globalisation influences the interdependence among nations as well as the nation state's sovereignty leading to (a need for) new global governance structures.
Global markets Globalisation is characterised by worldwide changes in economic infrastructures and the emergence of global markets and a global trading system.
Global communication and diffusion of information Globalisation makes the sharing of information and the exchange of experiences around common problems possible.
Global mobility Global mobility is characterised by a major increase in the extensity, intensity and velocity of movement and by a wide variety in 'types' of mobility.
Cross-cultural interaction Globalising cultural flows result in interactions between global and local cultural elements.
Global environmental changes Global environmental threats to ecosystems include global climate change, loss of biodiversity, global ozone depletion and the global decline in natural areas.
Conceptual model for globalisation and health
We have identified (the need for) global governance structures, global markets, global communication and the diffusion of information, global mobility, cross-cultural interaction, and global environmental changes as important features of globalisation. Based on Figure 1 and Table 1, it can be concluded that these features all operate at the contextual level of health determination and influence distal factors such as health(-related) policies, economic development, trade, social interactions, knowledge, and the provision of ecosystem goods and services. In turn, these changes in distal factors have the potential to affect the proximal health determinants and, consequently, health. Our conceptual framework for globalisation and health links the above-mentioned features of the globalisation process with the identified health determinants. This exercise results in Figure 2.
Figure 2 Conceptual framework for globalisation and population health.
Figure 3, subsequently, shows that within the developed framework, several links between the specific features of globalisation and health can be derived. These important links between globalisation and health are discussed in the following sections. It is important to note that Figure 3 primarily focuses on the relationships in the direction from globalisation to health. This does not mean, however, that globalisation is an autonomous process: globalisation is influenced by many developments at the other levels, although these associations are not included in the Figure for reasons of simplification. In addition, the only feedback that is included in Figure 3 concerns the institutional response. One also has to keep in mind that determinants within the distal level and within the proximal level also interact with each other, adding complexity to our model (see Huynen et al. 4 for more details and examples of important intralevel relationships).
Figure 3 Conceptual model for globalisation and population health.
Globalisation and distal health determinants
Figure 3 shows that the processes of globalisation can have an impact on all identified distal determinants (Figure 3; arrows 1–4). Below, the implications of the globalisation process on these distal determinants will be discussed in more detail.
Health(-related) policies
Global governance structures are gaining more and more importance in formulating health(-related) policies (Figure 3; arrow 1). According to Dodgson et al. [8], the most important organisations in global health governance are the World Health Organization (WHO) and the World Bank (WB). The latter plays an important role in the field of global health governance as it acknowledges the importance of good health for economic development and focuses on reaching the Millennium Development Goals [9]. The WB also influenced health(-related) policies together with the International Monetary Funds (IMF) through the Structural Adjustment Programmes (SAPs) (e.g. see Hong [10]). In order to give a more central role to pro-poor growth considerations in providing assistance to low-income countries, the IMF and WB introduced the Poverty Reduction Strategy approach in 1999 [11]. In addition, the policies of the World Trade Organization (WTO) are also increasingly influencing population health [10,12-14]. Fidler [15] argues that 'from the international legal perspective, the centre of power for global health governance has shifted from WHO to the WTO'. Opinions differ with regard to whether the WTO agreements provide sufficient possibilities to protect the population from the adverse (health) effects of free trade or not [16]. In 2002, the WTO ruled that the French ban on the import of all products containing asbestos was legal on health grounds, despite protests from Canada [17,18]. However, protecting citizens against health risks remains difficult, as health standards often need to be supported by sound scientific evidence before trade can be restricted (see e.g. the WTO ruling against the European trade barrier concerning hormone-treated meat [19,20]).
Another important development is the growing number of public-private partnerships for health, as governments increasingly attract private sector companies to undertake tasks that were formerly the responsibility of the public sector. At the global level, public-private partnerships are more and more perceived as a possible new form of global governance [12] and could have important implications for health polices, but also for health-related policies.
Economic development
Opinions differ with regard to the economic benefits of economic globalisation (Figure 3; arrow 2). On the one side, 'optimists' argue that global markets facilitate economic growth and economic security, which would benefit health. They base themselves on the results of several studies that argue that inequities between and within countries have decreased due to globalisation (e.g. see Frankel [21], Ben David [22], Dollar and Kraay [23]). Additionally, it is argued that although other nations or households might become richer, absolute poverty is reduced and that this is beneficial for the health of the poor [24]. On the other side, 'pessimists' are worried about the health effects of the exclusion of nations and persons from the global market. They argue that the risk of exclusion from the growth dynamics of economic globalisation is significant in the developing world [25]. In fact, notwithstanding some spectacular growth rates in the 1980's, especially in east Asia, incomes per capita declined in almost 70 countries during the same period [26]. Many worry about what will happen to the countries that cannot participate in the global market as successful as others.
Trade
Due to the establishment of global markets and a global trading system, there has been a continuing increase in world trade (Figure 3; arrow 2). According to the WTO, total trade multiplied by a factor 14 between 1950 and 1997 [27]. Today all countries trade internationally and they trade significant proportions of their national income; around 20 percent of world output is being traded. The array of products being traded is wide-ranging; from primary commodities to manufactured goods. Besides goods, services are increasingly being traded as well [28]. In addition to legal trade transactions, illegal drug trade is also globalising, as it circumvents national and international authority and takes advantage of the global finance systems, new information technologies and transportation.
Social interactions: migration
Due to the changes in the infrastructures of transportation and communication, human migration has increased at unprecedented rates (Figure 3; arrow 3) [28]. According to Held et al. [28] tourism is one of the most obvious forms of cultural globalisation and it illustrates the increasing time-space compression of current societies. However, travel for business and pleasure constitutes only a fraction of total human movement. Other examples of people migrating are missionaries, merchant marines, students, pilgrims, militaries, migrant workers and Peace Corps workers [28,29]. Besides these forms of voluntary migration, resettlement by refugees is also an important issue. However, since the late 1970s, the concerns regarding the economic, political, social and environmental consequences of migration has been growing and many governments are moving towards more restrictive immigration policies [30].
Social interactions: conflicts
The tragic terrorist attacks in New York and Washington D.C. in September 2001 fuelled the already ongoing discussions on the link between globalisation and conflicts. Globalisation can decrease the risk on tensions and conflicts, as societies become more and more dependent on each other due the worldwide increase in global communication, global mobility and cross-cultural interactions (Figure 3; arrow 3). Others argue that the resistance to globalisation has resulted in religious fundamentalism and to worldwide tensions and intolerance [31]. In addition, the intralevel relationships at the distal level play a very important role, because many developments in other distal factors that have been associated with the globalisation process are also believed to increase the risk on conflicts. In other words, the globalisation-induced risk on conflict is often mediated by changes in other factors at the distal level [4].
Social interactions: social equity and social networks
Cultural globalisation (global communication, global mobility, cross-cultural interaction) can also influence cultural norms and values about social solidarity and social equity (Figure 3; arrow 3). It is feared that the self-interested individualism of the marketplace spills over into cultural norms and values resulting in increasing social exclusion and social inequity. Exclusion involves disintegration from common cultural processes, lack of participation in social activities, alienation from decision-making and civic participation and barriers to employment and material sources [32]. Alternatively, a socially integrated individual has many social connections, in the form of both intimate social contacts as well as more distal connections [33]. On the other hand, however, the geographical scale of social networks is increasing due to global communications and global media. The women's movement, the peace movement, organized religion and the environmental movement are good examples of such transnational social networks. Besides these more formal networks, informal social networks are also gaining importance, as like-minded people are now able to interact at distance through, for example, the Internet. In addition, the global diffusion of radio and television plays an important role in establishing such global networks [28]. The digital divide between poor and rich, however, can result in social exclusion from the global civil society.
Knowledge
The knowledge capital within a population is increasingly affected by developments in global communication and global mobility (Figure 3: arrow 3). The term 'globalisation of education' suggests getting education into every nook and cranny of the globe. Millions of people now acquire part of their knowledge from transworld textbooks, due to the supraterritoriality in publishing. Because of new technologies, most colleges and universities are able to work together with academics from different countries, students have ample opportunities to study abroad and 'virtual campuses' have been developed. The diffusion of new technologies has enabled researchers to gather and process data in no time resulting in increased amounts of empirical data [34]. New technologies have even broadened the character of literacy. Scholte [34] argues that 'in many line of work the ability to use computer applications has become as important as the ability to read and write with pen and paper. In addition, television, film and computer graphics have greatly enlarged the visual dimensions of communication. Many people today 'read' the globalised world without a book'. Overall, it is expected that the above-discussed developments will also improve health training and health education (e.g. see Feachem [24] and Lee [35]).
Ecosystem goods and services
Global environmental changes can have profound effects on the provision of ecosystem goods and services to mankind (Figure 3; arrow 4). The Intergovernmental Panel on Climate Change (IPCC) [36] concludes that it is expected that climate change can result in significant ecosystem disruptions and threatens substantial damage to the earth's natural systems. In addition, several authors have addressed the link between biodiversity and ecosystem functioning and it is agued that maintaining a certain level of biodiversity is necessary for the proper provision of ecosystem goods and services [37-40]. However, it is still unclear which ecosystem functions are primarily important to sustain our physical health. Basically, the following types of 'health functions' can be distinguished. First, ecosystems provide us with basic human needs like food, clean air, clean water and clean soils. Second, they prevent the spread of diseases through biological control. Finally, ecosystems provide us with medical and genetic resources, which are necessary to prevent or cure diseases [41].
Globalisation and proximal health determinants
Figure 3 shows that the impact of globalisation on each proximal health determinant is mediated by changes in several distal factors (Figure 3; arrows 5–12). The most important relationships will be discussed in more detail below. It is important to note that health policies and health-related policies can have an influence on all proximal factors (Figure 3; arrow 5).
Health services
Health services are increasingly influenced by globalisation-induced changes in health care policy (Figure 3; arrow 5), economic development and trade (Figure 3: arrow 6), and knowledge (Figure 3; arrow 7), but also by migration (3: arrow 7). Although the WHO aims to assist governments to strengthen health services, government involvement in health care policies has been decreasing and, subsequently, medical institutions are more and more confronted with the neoliberal economic model. Health is increasingly perceived as a private good leaving the law of the market to determine whose health is profitable for investment and whose health is not [10]. According to Collins [42] populations of transitional economies are no longer protected by a centralized health sector that provides universal access to everyone and some groups are even denied the most basic medical services. The U.S. and several Latin American countries have witnessed a decline in the accessibility of health care following the privatisation of health services [43].
The increasing trade in health services can have profound implications for provision of proper health care. Although it is perceived as to improve the consumer's choice, some developments are believed to have long-term dangers, such as establishing a two-tier health system, movement of health professionals from the public sector to the private sector, inequitable access to health care and the undermining of national health systems [10,12]. The illegal trading of drugs and the provision of access to controlled drugs via the Internet are potential health risks [44]. In addition, the globalisation process can also result in a 'brain-drain' in the health sector as a result of labour migration from developing to developed regions.
However, increased economic growth is generally believed to enhance improvements in health care. Increased (technological) knowledge resulting from the diffusion of information can further improve the treatment and prevention of all kinds of illnesses and diseases.
Social environment
The central mechanism that links personal affiliations to health is 'social support,' the transfer from one person to another of instrumental, emotional and informational assistance [45]. Social networks and social integration are closely related to social support [46] and, as a result, globalisation-induced changes in social cohesion, integration and interaction can influence the degree of social support in a population (Figure 3; arrow 9). This link is, for example, demonstrated by Reeves [47], who discussed that social interactions through the Internet influenced the coping ability of HIV-positive individuals through promoting empowerment, augmenting social support and facilitating helping others. Alternatively, social exclusion is negatively associated with social support.
Another important factor in the social environment is violence, which often is the result of the complex interplay of many factors (Figure 3; arrows 5, 8 and 9). The WHO [48] argues that globalisation gives rise to obstacles as well as benefits for violence prevention. It induces changes in protective factors like social cohesion and solidarity, knowledge and education levels, and global violence prevention activities such as the implementation of international law and treaties designed to reduce violence (e.g. social protection). On the other hand, it also influences important risk factors associated with violence such as social exclusion, income inequality, collective conflict, and trade in alcohol, drugs or firearms.
Lifestyle
Due to the widespread flow of people, information and ideas, lifestyles also spread throughout the world. It is already widely acknowledged and demonstrated that several modern behavioural factors such as an unhealthy diet, physical inactivity, smoking, alcohol misuse and the use of illicit drugs are having a profound impact on human health [49-52] (Table 3). Individuals respond to the range of healthy as well as unhealthy lifestyle options and choices available in a community [53], which are in turn determined by global trade (Figure 3; arrow 8), economic development (Figure 3; arrow 8) and social interactions (Figure 3; arrow 9).
Table 3 Lifestyle and health
Lifestyle factor Health effects
Diet Excess energy intake results, together with physical activity, in obesity. Obesity is an increasing health problem and has several co-morbidities such as non-insulin dependent diabetes and cardiovascular diseases [49]. The nutritional quality of the diet (e.g. fruit and vegetable intake, saturated versus unsaturated fats) is also very important for good health.
Inactivity Physical inactivity has been linked to obesity, coronary hearth disease, hypertension, strokes, diabetes, colon cancer, breast cancer and osteoporotic fractures [49].
Smoking Tobacco is predicted to be the leading health risk factor by 2030 [50]. It causes, for example, cancer of the trachea, bronchus and lung [49], and cardiovascular diseases.
Alcohol use The consumption of alcoholic beverages increases to risk on liver cirrhosis, raised blood pressure, heart disease, stroke, pancreatitis and cancers of the oropharnix, larynx, oesophagus, stomach, liver and rectum [49]. The role of alcohol consumption in non-communicable disease epidemiology is, however, complex. For example, small amounts of alcohol reduce the risk on cardiovascular diseases, while drinking larger amounts is an important cause of these very same diseases [51].
Illicit drugs According to the World Health Report 2001 [52], 0,4 % of the total disease burden is attributable to illicit drugs (heroin and cocaine). Opiate users can have overall mortality rate up to 20 percent higher than those in the general population of the same age, due to not only overdoses but also to accidents, suicides, AIDS and other infectious diseases [49].
Although the major chronic diseases are not transmittable via an infectious agent, the behaviours that predispose to these diseases can be communicated by advertising, product marketing and social interactions [54]. Global trade and marketing developments drive, for example, the nutrition transition towards diets with high proportions of salt, saturated fat and sugars [51,53]. Another example is the worldwide spread of tobacco consumption as transnational tobacco companies take advantage of the potential for growth in developing countries [51,55]. Additionally, the scale of cigarette smuggling poses a considerable global threat to the efforts to control tobacco consumption [44]. Illegal trade in illicit drugs poses similar problems. At the same time, the alcohol industry is almost as globalised as the tobacco industry [56].
However, health education can play a role in promoting healthy lifestyles by improving an individual's knowledge about the health effects of different lifestyle options (Figure 3; arrow 9). Besides health education, (global) policies can also directly discourage unhealthy behaviour by means of economic incentives (e.g. charging excise on tobacco) or other legislation (Figure 3; arrow 5).
Physical environment: infectious diseases pathogens
The spread of infectious diseases is probably one of the most mentioned health effects of globalisation and past disease outbreaks have been linked to factors that are related to the globalisation process (see e.g. Newcomb [57]). The recent outbreak of the Severe Acute Respiratory Syndrome (SARS) demonstrates the potential of new infectious diseases to spread rapidly in today's world, increasing the risk of a global pandemic. The combination of movement of goods (Figure 3; arrow 10) and people (Figure 3; arrow 11), and profound changes affecting ecosystem goods and services (Figure 3; arrow 12) all contribute to increased risk of disease spread [57]. For example, the globalisation of food production, trade and consumption has been associated with the increased spread and transmission of food born diseases [57,58]. Diseases like HIV/AIDS or hepatitis B can also spread through trade in infected biological products (e.g. blood) [44].
Enhanced knowledge and new technologies will improve the surveillance of infectious diseases and monitoring of antibiotic resistance [24,35] (Figure 3; arrow 11). Globalisation potentially increases the speed of responses in some cases. Wilson [29] states that responding to disease emergence requires a global perspective-both conceptually and geographically-as the current global situation favours the outbreak and rapid spread of infectious disease. As a result, the policies and actions undertaking by the WHO are becoming increasingly important in controlling infectious diseases at a global level (Figure 3; arrow 5). For instance, the WHO played a critical role in controlling SARS by means of global alerts, geographically specific travel advisories and monitoring [59].
Food
Food trade has become an increasingly important factor with regard to food security worldwide (Figure 3; arrow 10). At present, however, the developed countries usually subsidise their agricultural sectors. Current liberalisation policies are expected to have profound implications on food trade and, subsequently food security [60]. Some argue that the resulting free trade will create access to better and cheaper food supplies via food imports and can stimulate more efficient use of the world's resources as well as the production of food in regions that are more suitable to do so [60,61]. Free trade permits food consumption to grow faster than domestic food production in countries where there are constraints on increasing the latter. Accelerated economic growth can also contribute to food security (Figure 3; arrow 10) [60]. Others, however, argue that the forces of globalisation in fact endanger food security (e.g. see Lang [62]) and that countries should strive to become more self-sufficient [60]. For many countries the increasing dependence on food imports goes hand in hand with a higher vulnerability to shocks arising in global markets, which can affect import capacity and access to food imports [60]. Many food insecure countries are not able to earn enough with exporting goods in order to pay for the needed food imports [63].
At the global level, there are increasing international efforts to achieve widespread food security (Figure 3; arrow 5). For instance, the right to adequate food is directly addressed in the 1966 International Covenant on Economic, Social and Cultural Rights. In 1996, the World Food Summit reaffirmed the right of everyone to have access to safe and nutritious food. In case of extreme food-insecurity and insufficient import capacity, food aid may be provided in order to supplement the scarce food imports. Globalisation can affect food security by enhancing the knowledge of foreign nations about the usefulness of food aid (Figure 3; arrow 11) [60].
Besides food trade, one can also deal with the mismatch between demand and supply by increasing food production in food-short regions. The globalisation process can increase food security by facilitating the worldwide implementation of better technologies and improved knowledge (e.g. irrigation technologies, research on genetically modified food) (Figure 3; arrow 11). At the same time, the natural resource base for food production is increasingly threatened (Figure 3; arrow 12). Finally conflicts are, of course, a threat to food security and it is expected that food security in sub-Saharan Africa, for example, will not increase without the establishment of political instability (Figure 3; arrow 11) [64].
Water
The effects of globalisation are also raising concerns over water security. The current globalisation process is accompanied by privatisation policies affecting the provision of water [65] (Figure 3; arrow 5). Governments and international financial institutions promote privatisation, as they believe it will promote market competition and efficiency. However, others are less optimistic about the effects of privatisation. In fact, some cases show that prices and inequalities in access even rise [66]. It is also argued that water, with vital importance socially, culturally, and ecologically, 'cannot be protected by purely market forces' [65]. On a global scale, there are increasing efforts to set up global guidelines or policies with regard to fresh water (Figure 3; arrow 5), however none of the international declarations and conference statements requires states to actual meet individual's water requirements [67].
The virtual trade of water is also believed to be of increasing importance (Figure 3; arrow 10). The water that is used in the production process of a commodity is called the 'virtual water' contained in that commodity. Therefore, the increasing global trade of commodities is accompanied by an increasing global trade in virtual water. The global volume of virtual water embedded in crop and livestock products traded between nations is estimated to be 1400 billion cubic metres per year [68].
In addition, the globalisation process can increase water security by facilitating the worldwide implementation of better technologies and improved knowledge (Figure 3; arrow 11). At the same time, the natural resource base is increasingly threatened as, for example, global climate change and deforestation profoundly affect our ecosystems ability to provide us with sufficient and adequate fresh water (Figure 3; arrow 12).
Conclusion
Globalisation is causing profound and complex changes in the very nature of our society, bringing new opportunities as well as risks. In addition, the effects of globalisation are causing a growing concern for our health, and the intergenerational equity implied by 'sustainable development' forces us to think about the right of future generations to a healthy environment and a healthy life.
Despite some empirical research efforts indicating the links between the globalisation process and specific health impacts, the present weakness in empirical evidence on the multiple links between globalisation and health is still a problem [44]. The described conceptual framework could give a meaningful contribution to further empirical research by serving as a well-structured 'think-model' or 'concept map'. It clearly demonstrates that an interdisciplinary approach towards globalisation and health is required, which draws upon the knowledge from relevant fields such as, for example, medicine, epidemiology, sociology, political sciences, (health) education, environmental sciences and economics.
In addition, the exploration of possible future health impacts of different globalisation pathways by means of scenarios analysis could provide a useful contribution to the ongoing discussions on globalisation and health [4]. Scenarios can be described as 'plausible but simplified descriptions of how the future may develop, according to a coherent and internally consistent set of assumptions about key driving forces and relationships' [69]. Recent research showed, however, that the health dimension is largely missing in existing global scenarios [70]. The developed framework for globalisation and population health has contributed to the understanding of future health implications and the model is, therefore, considered to be a useful tool to structure future scenario studies on the health implications of the globalisation process.
To conclude, the framework provides valuable insights in how to organise the complexity involved in studying the health effects resulting from globalisation. We claim that our approach has several beneficial characteristics. First, it is embedded in a holistic approach towards globalisation; in this paper we perceive globalisation as an overarching process in which simultaneously many different processes take place in many societal domains. In addition, the conceptual framework is embedded in a holistic approach towards population health. As a result, our model explicitly visualises that globalisation affects the institutional, economic, social-cultural and ecological determinants of population health and that the globalisation process mainly operates at the contextual level, while influencing health through the more distal and proximal determinants.
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
We would like to thank all colleagues at the International Centre for Integrative Studies (ICIS) and the Netherlands Environmental Assessment Agency (MNP-RIVM) for the fruitful discussions leading to this paper. This work is financially supported by MNP-RIVM within the project 'Population & Health'.
==== Refs
Woodward D Drager N Beaglehole R Lipson D Globalization and health: a framework for analysis and action Bulletin of the World Health Organization 2001 79 875 881 11584737
Labonte R Torgerson R Frameworks for analyzing the links between globalization and health. Draft report to the World Health Organization. 2002 Saskatoon, SPHERU, University of Saskatchewan
Martens P McMichael AJ Patz J Globalisation, Environmental Change and Health Global Change and Human Health 2000 1 4 8 10.1023/A:1011572212445
Huynen MMTE Martens P Hilderink H The health impacts of globalisation: a conceptual framework 2005 Bilthoven, Netherlands Environmental Assessment Agency (MNP)
WHO The world health report 2002: reducing risks, promoting healthy life 2002 Geneva, World Health Organization
Rennen W Martens P The globalisation timeline Integrated Assessment 2003 4 137 144 10.1076/iaij.4.3.137.23768
Martens P Rotmans J Transitions in a globalising world 2002 Lisse, Swets & Zeitlinger
Dodgson R Lee K Drager N Global health governance: a conceptual review 2002 London, Centre on Global Change and Health, London School of Hygiene and Tropical Medicine
UN UN Millennium Development Goals , United Nations. http://www.un.org/millenniumgoals/. Accessed 10-01-05
Hong E Globalisation and the impact on health: a third world view 2000 Issue paper prepared for The Peoples' Assembly, December 4-8, 2000, Savar Bangladesh,
IMF Evaluation of the IMF's role in Poverty Reduction Strategy Papers and the Poverty Reduction and Growth Facility 2004 Washington D.C., International Monetary Fund
Walt G Globalization and health 2000 Paper presented at the Medact Meeting,
WHO WTO WTO agreements and public health 2002 Geneva, World Health Organization and the World Bank
Commission on the Future of Health Care in Canada Globalization and Canada's health care system. 2002 Vancouver, University of British Colombia
Fidler D Global health governance: overview of the role of international law in protecting and promoting global public health 2002 London, Centre on Global Change and Health, London School of Hygiene and Tropical Medicine
Singer P One world 2002 New Haven, Yale University Press
WTO European Communities measures affecting asbestos and asbestos-containing products. WT/DS135R. Panel Report. 2000 Geneva, World Trade Organization
WTO European Communities measures affecting asbestos and asbestos-containing products. WT/DS135/AB/R. Appellate Body Report. 2001 Geneva, World Trade Organization
WTO European Communities measures concerning meat and meat products (hormones). WT/DS26/R/USA and WT/DS48/R/CAN. Panel report. 1997 Geneva, World Trade Organization
WTO European Communities measures concerning meat and meat products (hormones). WT/DS26/AB/R and WT/DS48/AB/R. Appellate Body report. 1998 Geneva, World Trade Organization
Frankel JA Romer D Does trade cause growth? American Economic Review 1999 379 399
Ben-David D WTO Trade, growth and disparity among nation Income Disparity and Poverty, World Trade Organization Special Study 5 2000 Geneva, WTO publications
Dollar D Kraay A Growth is good for the poor. 2001 Washington, DC, World Bank
Feachem RGA Globalisation is good for your health, mostly BMJ 2001 323 504 506 11532848 10.1136/bmj.323.7311.504
Oman C The policy challenges of globalisation and regionalisation 1996 Paris, OECD Development Centre
Reinicke WH Global public policy: governing without government? 1998 Washington D.C., Brookings Institution Press
WTO The World Trade Organization in brief 2003 Geneva, World Trade Organization
Held D McGrew AG Goldblatt D Perraton J Global transformations: politics, economics and culture 1999 Stanford, Stanford University Press
Wilson ME Travel and the emergence of infectious diseases Emerging infectious diseases 1995 1 39 46 8903157
UN International migration report 2002 2002 New York, United Nations, Department of Economic and Social Affairs, Population Division
Frenk J Sepulveda J Gomez-Dantes O McGuinnes MJ Knaul F The future of world health: the new world order and international health BMJ 1997 314 1404 1407 9161318
Reid C Wounds of exclusion: poverty, women's health and social justices 2004 Edmonton, Qualitative Institute Press
Kawachi I Kennedy B Wilkinson RG Income inequality and health. 1999 New York, The New Press
Scholte JA Globalization: a critical introduction 2000 New York, Palgrave
Lee K The global context: a review of priority global health issues for the UK 1999 London, Nuffield Trust
IPCC Climate change 2001: impacts, adaptation and vulnerability 2001 Cambridge, Cambridge University Press
UNEP Global biodiversity assessment. 1995 Cambridge, Cambridge University Press
Schulze ED Mooney HA Biodiversity and ecosystem function 1994 Berlin, Springer
Schwartz MW Brigham CA Hoeksema JD Lyons KG Mills MH Mantgem van PJ Linking biodiversity to ecosystem function: implications for conservation ecology Oecologica 2000 122 297 305 10.1007/s004420050035
Chapin FS Zavaleta ES Eviners VT Naylor RL Vitousek PM Reynplds H Hooper DU Lavorel S Sala OE Hobbie SE Mack MC Diaz S Consequences of changing biodiversity Nature 2000 405 234 242 10821284 10.1038/35012241
Huynen MMTE Martens P De Groot RS Linkages between biodiversity loss and human health: a global indicator analysis International Journal Of Environmental Health Research 2004 14 13 30 14660115 10.1080/09603120310001633895
Collins T Globalization, global health and access to health care Int J Health Plann Manege 2003 18 97 104 10.1002/hpm.698
Anomynous Trading health care away? GATS, public services and privatisation 2001 Dorset, The Corner House
Lee K Collin J Review of existing empirical research on globalization and health. 2001 Geneva, World Health Organization
House JS Landis KR Umberson D Social relations and health Science 1988 241 540 545 3399889
Berkman LF Glass T Brisette I Seeman TE From social integration to health: Durkheim in the new millennium Social Science and Medicine 2000 51 843 857 10972429 10.1016/S0277-9536(00)00065-4
Reeves PM Coping in cyberspace: the impact of Internet use on the ability of HIV-positive individuals to deal with their illness J Health Commun 2000 5 (Suppl) 47 59 11010356 10.1080/10810730050019555
WHO World report on violence and health 2002 Geneva, The World Health Organization
WHO The European health report 2002 2002 Copenhagen, World Health Organization, Regional Office for Europe
WHO The world health report 1999 1999 Geneva, World Health Organization
Beaglehole R Yach D Globalisation and the prevention and control of non-communicable diseases: the neglected chronic diseases of adults The Lancet 2003 362 903 908 13678979 10.1016/S0140-6736(03)14335-8
WHO The world health report 2001 2001 Geneva, World Health Organization
Murray CJL Smith R Diseases of globalisation 2001 London, Earthscan Publication Ltd.
Marks JS McQueen DV Koop CE, Pearson CE and Schwartz MR Chronic disease Critical issues in global health 2001 San Francisco, Jossey-Bass
Cunningham R Smoke and mirrors: the Canadian tobacco war 1996 Canada, International Development Research Centre
Jernigan DH Thirsting for markets: the global impact of corporate alcohol 1997 Marin County, Marin Institute for the prevention of alcohol and other drug problems
Newcomb J Biology and borders: SARS and the new economics of bio-security 2003 Cambridge, Bio Economic Research Associates
Lee K Globalization, communicable disease and equity Development 1999 42 35 39 10.1057/palgrave.development.1110080
Fidler D Germs, governance, and global public health in the wake of SARS J Clin Invest 2004 113 799 804 15067309 10.1172/JCI200421328
FAO Food and international trade 1996 Rome, Food and Agricultural Organisation of the United Nations
FAO The state of food insecurity in the world 2003 2003 Rome, Food and Agricultural Organisation of the United Nations
Lang T Food security: does it conflict with globalisation? Development 1996 4 45 50
FAO Trade reforms and food security: conceptualizing the linkages 2003 Rome, Food and Agricultural Organisation of the United Nations
Meade B Rosen S Shapouri S Andrews M Trueblood M Nord M Persaud P Food security assessment 2002-2003 2003 Washington D.C., United States Department of Agriculture
Gleick PH The world's water 2002 - 2003: the biennial report on freshwater resources 2002 Washington, D.C., Island Press
Olivera O Lewis T Cochabamba! Water war in Bolivia 2004 Cambridge, South End Press
Gleick PH The world's water 2000 - 2001: the biennial report on freshwater resources 2000 Washington, D.C., Island Press
Hoekstra AY Globalisation of water 2004 Presented at Aqua: European citizenship through water. Collegno, Italy, June 9,
Swart RJ Raskin P Robinson J The problem of the future: sustainability science and scenario analysis Global Environmental Change 2004 14 137 146 10.1016/j.gloenvcha.2003.10.002
Martens P Huynen MMTE A future without health: health dimension in global scenario studies Bulletin of the World Health Organization 2003 81 896 8901 14997242
|
16078989
|
PMC1208931
|
CC BY
|
2021-01-04 16:39:02
|
no
|
Global Health. 2005 Aug 3; 1:14
|
utf-8
|
Global Health
| 2,005 |
10.1186/1744-8603-1-14
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-511613732910.1186/1477-7525-3-51ResearchSubjective assessments of comorbidity correlate with quality of life health outcomes: Initial validation of a comorbidity assessment instrument Bayliss Elizabeth A [email protected] Jennifer L [email protected] John F [email protected] Kaiser Permanente, PO Box 378066, 80237-8066 Denver, CO, USA2 Department of Family Medicine, University of Colorado Health Sciences Center, Denver, CO, USA3 Colorado Health Outcomes Program, University of Colorado Health Sciences Center, Denver, CO, USA2005 1 9 2005 3 51 51 8 7 2005 1 9 2005 Copyright © 2005 Bayliss et al; licensee BioMed Central Ltd.2005Bayliss et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Interventions to improve care for persons with chronic medical conditions often use quality of life (QOL) outcomes. These outcomes may be affected by coexisting (comorbid) chronic conditions as well as the index condition of interest. A subjective measure of comorbidity that incorporates an assessment of disease severity may be particularly useful for assessing comorbidity for these investigations.
Methods
A survey including a list of 25 common chronic conditions was administered to a population of HMO members age 65 or older. Disease burden (comorbidity) was defined as the number of self-identified comorbid conditions weighted by the degree (from 1 to 5) to which each interfered with their daily activities. We calculated sensitivities and specificities relative to chart review for each condition. We correlated self-reported disease burden, relative to two other well-known comorbidity measures (the Charlson Comorbidity Index and the RxRisk score) and chart review, with our primary and secondary QOL outcomes of interest: general health status, physical functioning, depression screen and self-efficacy.
Results
156 respondents reported an average of 5.9 chronic conditions. Median sensitivity and specificity relative to chart review were 75% and 92% respectively. QOL outcomes correlated most strongly with disease burden, followed by number of conditions by chart review, the Charlson Comorbidity Index and the RxRisk score.
Conclusion
Self-report appears to provide a reasonable estimate of comorbidity. For certain QOL assessments, self-reported disease burden may provide a more accurate estimate of comorbidity than existing measures that use different methodologies, and that were originally validated against other outcomes. Investigators adjusting for comorbidity in studies using QOL outcomes may wish to consider using subjective comorbidity measures that incorporate disease severity.
==== Body
Background
The goal of caring for persons with chronic medical conditions is frequently to maximize quality of life (QOL) rather than to 'cure' illness. Therefore interventions to improve processes of care for this population often assess QOL outcomes such as physical functioning, overall health status, and emotional well being. These outcomes are, by definition, subjective. The values assigned to these outcomes are most meaningful to the patients themselves. However, these subjective outcomes have been shown to correlate with mortality, health care utilization, job loss, and many other more 'quantifiable' outcomes [1-3].
The outcomes of a chronic condition may be affected by coexisting (comorbid) chronic conditions as well as the index condition of interest and analyses must adjust for this effect of comorbidity. Multiple instruments have been developed and validated to quantify comorbidity for purposes of statistical adjustment and clinical decision making. The majority of these use medical record review or administrative data as sources of information; observation during clinical encounters and self report have also been used for this purpose. These instruments have primarily been validated against 'objective' health outcomes such as mortality, length of stay, and cost of care [4-13]. We are aware of two such instruments that have been validated against QOL outcomes [5,14]. In addition, many of these instruments were designed for use in hospitalized patients or populations characterized by specific illnesses.
Self-reported information about comorbidity and the burden it imposes can provide information about the concurrent impact of multiple disease states on QOL outcomes. Self-reported comorbidity information is also efficient in studies in which other information, such as QOL outcomes, is collected by survey. Instruments designed to assess comorbidity by self-report have reported significant correlations between comorbidity score and utilization, QOL, mortality and hospitalization [15-20].
It is important to incorporate assessment of disease severity into comorbidity measurement [6]. Some self-report instruments incorporate various weighting systems for this purpose and two of these have been validated in hospitalized populations [15,18]. We have developed a self-report instrument that incorporates disease severity by quantifying the respondent's subjective 'disease burden' which we define as the number of self-identified comorbid conditions weighted by the degree to which each condition limits daily activity. We hypothesized that a subjective measure of comorbidity such as this may be more strongly correlated with QOL outcomes than measures of comorbidity previously validated against other, more objective, health outcomes.
Our goals in this investigation were to validate this newly-developed instrument against a presumed 'gold standard' of chart review, and to conduct an initial comparison of this instrument with other well known measures of comorbidity (chart review of number of conditions, the Charlson Comorbidity Index and the RxRisk score) by correlating these measures with selected QOL outcomes.
Methods
Study setting and sample selection
The study setting was a Health Maintenance Organization (HMO) in the United States that provides primary, specialty and hospital care for persons of all ages. Due to the use of an electronic medical record, both primary and specialty providers can enter diagnoses and assessments into a single patient record. Participants were selected from a stratified random sample of HMO members age 65 or older with 0 (8%), 1 (10%), 2 (12%), or 3 or more (69%) chronic medical conditions. We sampled this age group based on the high prevalence of comorbid conditions in older adults [21]. The stratification was performed with a modified version of the RxRisk comorbidity assessment instrument that uses administrative pharmacy data to determine an estimated disease count [4]. As one of the goals of our investigation was to assess issues of importance to persons with multiple comorbidities, we oversampled members with a greater number of chronic conditions. Due to the pilot nature of the study, we used consecutive random sampling in increments of single mailings until we had sufficient sample size to evaluate the instrument. We calculated that we would need a sample size of 139 for an expected proportion (sensitivity and specificity) of 0.90 to have a 95% confidence interval with a total width of 0.10.
Instrument development
We searched the literature to determine the health conditions most frequently assessed in measuring comorbidity [4,5,7,17,22-26]. From this we assembled a list of 25 common chronic conditions and coupled it with a scale that asked respondents to report for each condition a) whether they had the condition, and b) if so whether it interfered with their daily activities "not at all' (a weight of 1) to "a lot" (a weight of 5). These responses then provided a measure of 'disease burden' (comorbidity) that resulted from weighting each reported condition by the degree of limitation. These conditions are listed in Table 2. Depression is absent from the list of morbidities as it was assessed as a separate outcome measure. As there is a known correlation between comorbidity and physical dimensions of QOL, we chose overall health status and physical functioning as our primary outcomes of interest [14]. We also investigated depression and self-efficacy as secondary outcomes important in caring for persons with multiple morbidities.
Table 2 Sensitivity and Specificity of Self-Report Relative to Chart Review (N = 1511)
Prevalence
Medical condition2 Mean Self-Report Disease Burden Self-Report n (%) Chart Review n (%) Sensitivity (%) Specificity (%)
Angina/coronary artery disease 2.2 19 (12) 39 (25) 41 97
Asthma 2.7 20 (13) 12 (8) 100 94
Back pain 3.0 61 (39) 37 (24) 92 77
Bronchitis, chronic/COPD 3.1 20 (13) 20 (13) 70 96
Cancer (within the past 5 yrs) 1.7 11 (7) 20 (13) 55 100
Cholesterol, elevated 1.6 81 (52) 78 (50) 89 85
Colon problem (e.g., diverticulitis, irritable bowel) 2.8 21 (14) 12 (8) 75 92
Congestive heart failure 2.5 23 (15) 23 (15) 74 96
Diabetes 2.2 31 (20) 32 (21) 88 98
Hard of hearing 2.6 75 (48) 32 (21) 81 61
Hypertension 1.8 95 (61) 103 (66) 84 83
Kidney disease 2.0 6 (4) 17 (11) 35 100
Nerve condition 3.4 9 (6) 15 (10) 47 99
Osteoarthritis 2.8 72 (46) 69 (44) 73 75
Osteoporosis 2.1 30 (19) 22 (14) 86 92
Overweight 2.4 70 (45) 28 (18) 96 66
Poor circulation (e.g., peripheral vascular disease) 3.0 44 (28) 14 (9) 93 78
Rheumatic disease, other 3.2 5 (3) 4 (3) 75 99
Rheumatoid arthritis 3.2 25 (16) 4 (3) 75 86
Stomach problem (e.g., gastritis, peptic disease) 2.3 46 (30) 40 (26) 75 86
Stroke 2.4 16 (10) 20 (13) 60 97
Thyroid disorder 1.5 39 (25) 41 (26) 90 98
Vision problem 2.3 98 (63) 109 (70) 78 72
1 Total N = 156, 151 participants reported 1 or more conditions.
2 For most conditions, an example or two were provided to illustrate the diagnostic category. For example, 'other rheumatic disease" was presented as "rheumatic disease such as fibromyalgia or lupus"; and "nerve condition" was presented as "nerve condition such as Parkinson's disease or multiple sclerosis."
Survey administration
We pre-tested the instrument for clarity and ease of completion with volunteers who were age 65 or older and had more than one chronic medical condition. Pre-testing was conducted in one-on-one interviews in which the volunteer completed the survey and then provided detailed feedback to the interviewer on the content and comprehension of the measure. Any recommended changes were incorporated into the subsequent version of the instrument. It was then mailed to respondents as a component of another pilot survey that assessed potential barriers to the medical self-care process. The complete questionnaire included validated questions that assessed physical functioning and general health status, a depression screen, and an adapted and concurrently validated assessment of general self-efficacy. We used the physical functioning measure and the general health status single question from the Short-Form 36®, the depression screen from the Behavioral Risk Factor Surveillance System, and a concurrently validated adaptation of the general self-efficacy scale (our coefficient alpha = 0.76) [1,27-29]. We used these assessments as our primary and secondary QOL outcomes of interest for the current investigation. The investigation was approved by the Institutional Review Board of the participating HMO and informed consent was obtained from all participants.
Comparison with chart review
We compared each participant's responses with diagnoses listed in their electronic medical record. We reviewed assessments from all outpatient encounters over the two years preceding the survey and accepted at least two chart-documented assessments of a chronic condition as an active diagnosis. Requiring two rather than one chart diagnosis may reduce the sensitivity of self-report [30]. However, we based our decision on the assumption that a recurrence of a chronic diagnosis would reasonably have been communicated to the patient, and therefore he or she might be expected to list that diagnosis in their response to our survey. Either two recorded outpatient diagnoses or one inpatient diagnosis have been suggested as a reasonable standard for a confirmed diagnosis [31]. In our chart review, we also counted previously documented chronic conditions that were likely to persist (e.g. hearing loss). We did not count diagnoses of chronic problems that had been surgically corrected and required no further management (e.g. cataract surgery).
Comparison with other measures of comorbidity
In addition to calculating comorbidity with our instrument, for each respondent we quantified level of comorbidity using two other validated comorbidity measurement tools. These two methods were the RxRisk score and the Charlson comorbidity index [4,7]. We chose these based on both their common use and the contrast they provided in methodologies since they use different methods of data collection and have been validated against different outcomes. The RxRisk score is a measure of comorbidity that incorporates age, gender, health insurance benefit status and an RxRisk category based on diagnoses derived from administrative pharmacy data. It was originally developed and validated to identify chronic conditions and to predict cost of health care, and subsequently revised to assess disease burden in certain populations [4,32]. We used administrative pharmacy data to apply the RxRisk tool to our study population. The Charlson comorbidity index is a widely used comorbidity measure that was originally developed to predict one-year mortality following hospitalization. The score is based on chart review for specified diagnostic criteria. It has been subsequently adapted and revalidated to assess longer term mortality, disability, hospital readmission and length of stay and has been revised into formats that utilize either ICD-9 diagnosis codes or questionnaire [6-8,25]. We calculated the Charlson comorbidity score using chart review.
Statistical methods
We calculated sensitivity and specificity for each condition using the chart report as the 'gold standard.' We also calculated sensitivity and specificity for each participant to indicate the percent of positive and negative conditions on which the respondent and chart agree relative to the total positive or negative conditions in the chart. Thus specificity and sensitivity by condition reflect respondents' overall tendency to accurately report a given condition relative to chart report, and sensitivity and specificity by participant reflect respondents' overall tendency to accurately report on all of their conditions in comparison to the gold standard of chart review. (Note that sensitivity and specificity analyses used self-reported presence or absence of conditions for comparison rather than the weighted disease burden score.) In order to further compare self-reported disease burden with our 'gold standard' of chart review, for each condition we entered self-reported disease burden followed by chart report of that condition into limited logistic regression models to assess the relative contributions of each of these independent variables to the predictive accuracy of the model for each of our outcome measures [33].
We calculated Spearman correlations between disease burden from the new instrument, disease count by chart review, the Charlson index and the RxRisk score, with our QOL outcomes of interest: measures of overall health status, physical functioning, positive depression screen, and level of self-efficacy.
Results
After two consecutive single mailings, 157 individuals completed the survey. The response rate of 28% was obtained without the use of strategies typically employed to increased response, such as multiple mailings or more active follow-up. Characteristics of respondents are noted in Table 1. Mean age was 75, health status ranged from excellent to poor and respondents reported an average of 5.9 chronic conditions. Respondents did not differ from non-respondents with regard to age, gender, number of chronic conditions (as estimated by the initial screen with the RxRisk instrument), or duration of HMO membership.
Table 1 Characteristics of study population (N = 156)
Characteristic Number (%)
Age (mean, range) 75.0, 67–94
Gender
Male 75 (48.1)
Female 77 (49.4)
Missing, chose not to answer 4 (2.6)
Marital status
Married 101 (64.7)
Widowed 33 (21.2)
Divorced/separated 18 (11.5)
Missing, chose not to answer 4 (2.6)
Education level
Did not graduate high school 16 (10.3)
High school graduate 42 (26.9)
Some college 42 (26.9)
College graduate 20 (12.8)
Post-college 31 (19.9)
Missing, chose not to answer 5 (3.2)
Household income (mean category)
Less than $15,000 22 (14.1)
$15,000–30,000 38 (24.4)
$30,000–45,000 29 (18.6)
$45,000–60,000 20 (12.8)
$60,000–75,000 12 (7.7)
$75,000–90,000 1 (0.6)
More than $90,000 5 (3.2)
Missing, chose not to answer, don't know 29 (18.6)
Race
Caucasian 142 (91.0)
African-American 1 (0.6)
Other 7 (4.5)
Missing, chose not to answer 6 (3.8)
Hispanic ethnicity
Yes 5 (3.2)
No 131 (84.0)
Missing, chose not to answer 20 (12.8)
Health status
Excellent 6 (3.9)
Very Good 43 (27.6)
Good 67 (43.0)
Fair 30 (19.2)
Poor 4 (2.6)
Missing 6 (3.9)
Level of Comorbidity (mean, range of each)
Number of Self-Reported Conditions 5.9, 0–16
Self-Reported Disease Burden* 13.9, 0–51
*Total score of limitations due to conditions (Sum of weights from 1 to 5 for each condition present).
One hundred fifty-one respondents reported at least one of the conditions and 6 reported none. In analyses by condition, median sensitivity of patient report of a condition relative to a 'gold standard' of chart review was 75% (range 35% to 100%) and median specificity was 92% (range 61% to 100%). In analyses by respondent, sensitivities (agreement on number of conditions positive relative to chart review) ranged from 14% (n = 1) to 100% (n = 53); the median was 83%. Sensitivities were not calculated for the ten respondents who did not agree with the chart on any conditions, including those who agreed with their medical record that they had none of the conditions (n = 2). Specificities by respondent ranged from 59% (n = 1) to 100% (n = 34); the median was 91%. Sensitivity and specificity of self-report of each condition relative to chart review are reported in Table 2. (Not included on the table are results for 2 of the original 25 conditions: liver disease and alcoholism. Two respondents and one separate chart reported alcohol abuse, and no respondents or charts reported liver disease.)
In order to assess the relative contributions of self reported diseases and disease count by chart review to the outcomes of general health status and physical functioning, we entered these two variables into limited logistic regression models. In these models containing only these two variables, the predictive accuracy of the model (as measured by the c-statistic) was not significantly different using each of the two variables, implying comparable contributions of either measure. C-statistics for overall health status ranged from 0.521 ("other rheumatic disease) to 0.669 ("osteoarthritis"); and for physical functioning ranged from 0.515 ("other rheumatic disease") to 0.679 ("overweight").
QOL outcomes of interest correlated most strongly with self-reported disease burden, followed by number of conditions by chart review, self-reported number of conditions, the Charlson index score and the RxRisk score. Although all measures of comorbidity except the RxRisk score showed comparable p values (p <= 0.001) for the outcomes of health status and physical functioning, the correlations for disease burden were significantly stronger than those for self-reported number of conditions or Charson comorbidity score for these outcomes. Table 3 lists these correlations for our primary outcomes of interest – overall health status and physical functioning – and our secondary outcomes of positive screen for depression and self-efficacy.
Table 3 Correlations Between Measures of Comorbidity and QOL Outcomes (N = 1562)
Self reported disease burden1 Chart review number of conditions Self reported number of conditions3 Charlson comorbidity score [7] Rx-risk score [4]
Overall health status* (n = 150) 0.60 p < 0.001 0.56 p < 0.001 0.477 p < 0.001 0.48 p < 0.001 0.17 P = 0.037
Physical functioning* (n = 137) -0.63 p < 0.001 -0.52 p < 0.001 -0.482 p < 0.001 -0.41 p < 0.001 -0.18 p = 0.035
Depression screen* (n = 153) -0.29 p < 0.001 -0.25 p = 0.002 -0.240 p = 0.003 -0.12 p = 0.140 -0.05 p = 0.559
Self-efficacy* (n = 145) -0.32 p < 0.001 -0.22 p = 0.008 -0.305 p < 0.001 -0.14 p = 0.096 0.10 p = 0.234
* For health status, a higher score implies worse perceived health; for other outcomes, a higher number implies a better functioning, less depression or greater self-efficacy.
1 Total score of degree of limitation due to each positive condition (1 = not at all to 5 = a lot).
2 Due to missing scale scores, total n ranged from 137 (physical activity) to 154 (social activity).
3 Number of conditions from the list that were positively reported by the respondent.
Discussion
It is important to incorporate assessment of comorbidity into studies involving QOL outcomes for persons with chronic medical conditions, as coexisting conditions may substantially affect outcomes of interest such as physical functioning, overall health status, depression and self-efficacy. In our study population, patients with multiple chronic medical conditions accurately reported a majority of common comorbid conditions relative to chart review. In addition, they were aware of most of their own diagnoses. Furthermore, self-reported disease burden correlated well with QOL outcomes, and correlated more strongly than did the two other measures of comorbidity that we used for comparison. This is consistent with our hypothesis that, for investigations using QOL outcomes, it is most appropriate to adjust for comorbidity using a subjective measure of comorbidity.
Previous investigations that have compared self-report with administrative data reported 59–79%, 72–73%, and 78–83% agreement on diagnoses of hypercholesterolemia, diabetes, and hypertension respectively; and 56% and 69% agreement on stroke and myocardial infarction [30,34]. In our investigation we expanded the number of conditions for comparison to 23 and additionally assessed respondents' tendencies to accurately report all of their own conditions. Certain diagnoses were reported with high levels of sensitivity and specificity, while others were not.
A sensitivity greater than specificity may be due to either 'over-reporting' by participants or 'under-reporting' in the chart. Examples from our list included asthma, back pain, overweight and hard-of-hearing. We suspect that, for the first case, some participants reported COPD as asthma. For the remaining cases, we suspect that the conditions were under-reported in the chart – either because they had not been brought to medical attention or because they had not been assessed as isolated problems in the context of medical visits during the period covered by the chart review.
Sensitivity was substantially less than specificity for angina, nerve conditions, cancer and kidney disease. Although there may be a tendency to under-report chronic conditions, and respondents are more likely to report conditions with more severe symptoms [17,35]; we re-reviewed charts of persons with these diagnoses to see if we could determine the cause of the discrepancies. From these repeat chart reviews, we concluded that these discrepancies were due to wording based more on symptoms than diagnosis (angina), under-reporting of conditions with stable or few symptoms (renal and neurological), and possible perceptions of cure or remission after acute treatment (cancer). In addition we analyzed the demographic and health characteristics (from Table 1) of respondents for each of these four conditions to see if any demographic or disease characteristics were likely to predict a low agreement with chart review and found no patterns.
In our assessments of sensitivity and specificity, we assumed that the presence of a diagnosis in the chart was a 'gold standard' – an assumption that may not be entirely accurate. We suspect that diagnoses for which there are obvious medical treatments – especially medications – are more likely to be recorded in the chart. Chart diagnoses may be less accurate for conditions for which a person is less likely to seek (or for which a provider is less likely to offer) specifically biomedical solutions.
We found a high correlation between our measure of disease burden and our QOL outcomes of interest, as compared to lower correlations between two other comorbidity indices and these same outcomes. However, the correlations between the other comorbidity indices and health status and physical functioning were also significant and have been noted previously [36]. The correlations between the Charlson and RxRisk scores and our secondary outcomes of interest (depression screen and self-efficacy) were not significant. Based on the pattern of these associations, we suggest that assessment of comorbidity is a function of the outcome of interest, the population studied, and the different (subjective versus objective) aspects of comorbidity measured by each instrument. The effect of comorbidities on QOL outcomes may be most accurately assessed when subjective measures are used to adjust for comorbidity. In contrast, for situations in which mortality, for example, is the outcome of interest, comorbidity should be assessed using instruments that have been developed for that purpose. These suggestions are consistent with the notion that 'complete' measurement of all health states requires both self-reported and objectively reported measures [37].
It is certainly possible that one comorbidity measure may work for many situations. Other self-report instruments have been shown to predict mortality and hospitalization in addition to QOL [15,16,18]. We are also aware of at least two investigations in which comorbidity measured by chart review correlated with QOL outcomes [5,14]. The two instruments with which we compared our own instrument use different methodologies and were originally developed to assess comorbidity in studies investigating the objective outcomes of mortality and cost of care respectively [4,7]. The Charlson index has been subsequently validated against length of stay, post operative complications, discharge to nursing home, disability, hospital readmission and hospital charges [6,8,38-40]. The RxRisk score has subsequently been adapted and validated against administrative data on diagnoses and disease burden in certain populations [4,32]. Our investigation adds to the growing body of knowledge on measuring comorbidity by highlighting the different results that may be obtained when using different methodologies to adjust for comorbidity in studies assessing QOL outcomes.
We did not incorporate additional measures of comorbidity, such as those that use administrative data into our analysis [8,12,13]. Previous comparative studies suggest that chart-review-based measures may be slightly more accurate than administrative data-based comorbidity measures in predicting objective outcomes such as mortality and length of hospital stay [6,38,41]. Further investigation is necessary to assess association of comorbidity measured by administrative data with QOL outcomes.
As with any initial validation effort, the generalizability of our conclusions is limited by the characteristics of the population studied – a relatively small HMO population aged 65 years or older. It is possible that this population is relatively 'well-educated' regarding the number and type of their medical conditions. If so, some of the sensitivities we report may be at the upper end of the spectrum that may be anticipated from self-report. In addition, we terminated the sampling process when we attained a sample size sufficient to test our primary hypothesis, without maximizing response rate. Thus, the findings in this sample may not represent the associations of a broader population. Although respondents did not differ significantly from non-respondents on RxRisk comorbidity score, more motivated or knowledgeable participants may have been more likely to respond promptly to our survey. Correlations and sensitivities could be lower when examined in a less motivated population or those with a lower knowledge base. Specifically, self-report may be less reliable in the geriatric sub-population that may suffer from cognitive impairment. Additional validation studies will be required in order to assess the usefulness of this instrument in other populations and for different QOL and other outcomes. We anticipate that these changes will strengthen our results for sensitivity in comparison to chart review and that they will not change the overall correlations with our outcomes of interest.
Disease burden (as we defined it) may in itself constitute a substantial portion of any patient's assessment of health status and physical functioning. Our incorporation of perceived limitation into a disease count may be similar to other investigations that have coupled a simple disease count with a health status measure such as the SF-36® and found that doing so strengthened the relationship between comorbidity and utilization and mortality [16,19]. However, models that attempt to explain the relationship between symptom burden, overall quality of life and physical functioning note that these outcomes are also affected by environmental characteristics, individual personality, expectations, values, and social and psychological supports [42,43]. What we refer to as disease burden explains part, but not all, of our QOL outcomes as is illustrated by the values of our c-statistics. To the extent that investigations that use QOL outcomes concentrate on participants with one index condition and need to adjust for comorbidities, a subjective measure of disease burden using self-report may be an accurate way to account for the effect of other coexisting conditions with regard to that outcome.
Finally, depression is both an important potential comorbidity for anyone with chronic illness as well as an equally important component of the QOL outcome of emotional well being. We chose to treat it as the latter. As depression severity independently contributes to general QOL over and above other coexisting chronic illness, we suspect that including depression on our list of conditions would have increased the strength of correlations between self-reported disease burden and general health status [44,45].
Conclusion
Assessing comorbidity is relevant to investigations of populations with multiple medical conditions and should be incorporated into the associated analyses. Not only is self-report likely to give a reasonable estimate of comorbidity, for investigations using QOL outcomes, self-reported disease burden (or other subjective assessments of comorbidity) may provide a more accurate comorbidity adjustment than measures that have been validated against other outcomes. If this finding is confirmed by additional investigation, subjective measures of comorbidity that incorporate disease severity should be added to QOL assessments for populations with high rates of comorbidity.
Authors' contributions
EB conceived the study, designed the comorbidity instrument, supervised survey administration and drafted the manuscript. JE participated in the design of the study, performed the statistical analysis, and participated in the data review and manuscript preparation. JS consulted on all phases of the study design, data review and analysis, and participated in the manuscript preparation.
Acknowledgements
This project was funded by an internal research grant from Kaiser Permanente, Colorado.
Portions of this material were previously presented in poster format at the annual HMO Research Network Conference, Santa Fe, NM. April 2004.
==== Refs
Ware JE SF-36(R) Health Survey Update 2005
Lawrence WF Fryback DG Margin PA Klein R Klein BEK Health status and hypertension: A population-based study J Clin Epidemiol 1996 49 1239 1245 8892491 10.1016/S0895-4356(96)00220-X
Wagner EH LaCroix AZ Grothaus LC Hecht JA Responsiveness of health status measures to change in older adults J Am Geriatr Soc 1993 41 241 248 8440846
Fishman PA Goodman MJ Hornbrook MC Meenan RT Bachman DJ O'Keefe Rosetti MC Risk adjustment using automated ambulatory pharmacy data. TheRxRisk model. Med Care 2003 41 84 99 12544546 10.1097/00005650-200301000-00011
Greenfield S Apolone G McNeil BJ Cleary PD The importance of co-existent disease in the occurrence of postoperative complications and one-year recovery in patients undergoing total hip replacement. Comorbidity and outcomes after hip replacement. Med Care 1993 31 141 154 8433577
de Groot V Beckerman H Lankhorst GJ Bouter LM How to measure comorbidity: a critical review of available methods J Clin Epidemiol 2003 56 221 229 12725876 10.1016/S0895-4356(02)00585-1
Charlson ME Pompei P Ales KL MacKenzie CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chron Dis 1987 40 373 383 3558716 10.1016/0021-9681(87)90171-8
Deyo RA Cherkin DC Ciol MA Adapting a clinical comorbidity index for use with ICD-9 administrative databases J Clin Epidemiol 1992 45 613 619 1607900 10.1016/0895-4356(92)90133-8
Katz JN Chang LC Sangha O Fossel AH Bates DW Can comorbidity be measured by questionnaire rather than medical record review? Med Care 1996 34 73 84 8551813 10.1097/00005650-199601000-00006
Linn BS Linn MW Lee G Cumulative illness rating scale J Am Geriatr Soc 1968 16 622 626 5646906
Kaplan MH Feinstein AR The importance of classifying initial co-morbidity in evaluating the outcome of diabetes mellitus J Chron Dis 1973 27 387 404 4436428 10.1016/0021-9681(74)90017-4
Starfield B Weiner J Mumford L Steinwachs D Ambulatory care groups: a categorization of diagnoses for research and management Health Serv Res 1991 26 53 74 1901841
Elixhauser AE Steiner C Harris R Coffey RM Comorbidity measures for use with administrative data Med Care 1998 36 8 27 9431328 10.1097/00005650-199801000-00004
Fortin M Lapointe L Hudon C Vanasse A Ntetu AL Maltais D Multimorbidity and quality of life in primary care: a systematic review. Health and Quality of Life Outcomes 2004 2 51 15380021 10.1186/1477-7525-2-51
Sangha O Stucki G Liang MH Fossel AH Katz JN The self-administered comorbidity questionnaire: a new method to assess comorbidity for clinical and health services research Arthritis Rheum 2003 49 156 163 12687505 10.1002/art.10993
Fan VS Au D Heagerty P Deyo RA McDonell MB Fihn SD Validation of case-mix measures derived from self-reports of diagnoses and health J Clin Epidemiol 2002 55 371 380 11927205 10.1016/S0895-4356(01)00493-0
Klabunde CN Reeve BB Harlan LC Davis WW Potosky AL Do patients consistently report comorbid conditions over time? Results from the prostate cancer outcomes study Med Care 2005 43 391 400 15778642 10.1097/01.mlr.0000156851.80900.d1
Chaudhry S Jin L Meltzer D Use of a self-report-generated Charlson Comorbidity Index for predicting mortality Med Care 2005 43 607 615 15908856 10.1097/01.mlr.0000163658.65008.ec
Hornbrook MC Goodman MJ Chronic disease, functional health status, and demographics: a multi-dimensional approach to risk adjustment Health Services Research 1996 31 283 307 8698586
Selim AJ Fincke BG Ren XS Lee A Rogers WH Miller DR Skinner KM Linzer R Kazis LE Comorbidity assessments based on patient report. Results from the Veterans Health Study Jounral of Ambulatory Care Management 2004 27 281 295
Anderson GF Partnership for solutions: Better Lives for People with Chronic Conditions. 2000 2000 Robert Wood Johnson Foundation and Johns Hopkins University.
Verbrugge LM Lepkowski JM Imanaka Y Comorbidity and its impact on disability The Milbank Quarterly 1989 67 450 484 2534562
Fried LP Bandeen-Roche K Kasper JD Guralnik JM Association of comorbidity with disability in older women: The Women's Health and Aging Study Journal of Clinical Epidemiology 1999 52 27 37 9973071 10.1016/S0895-4356(98)00124-3
Parkerson GRJ Broadhead WE Tse C The Duke Severity of Illness Checklist (DUSOI) for measurement of severity and comorbidity J Clin Epidemiol 1993 46 379 393 8483003 10.1016/0895-4356(93)90153-R
Charlson M Szatrowski TP Peterson J Gold J Validation of a combined comorbidity index J Clin Epidemiol 1994 47 1245 1251 7722560 10.1016/0895-4356(94)90129-5
Ware JE Bayliss MS Rogers WH Kosinski M Tarlov AR Differences in 4-year health outcomes for elderly and poor, chronically ill patients treated in HMO and fee-for-service systems. Results from the Medical Outcomes Study JAMA 1996 276 1039 1047 8847764 10.1001/jama.276.13.1039b
National Center for Chronic Disease Prevention and Health Promotion Behavior Risk Factor Surveillance System 2005
Arroll B Khin N Kerse N Screening for depression in primary care withtwo verbally asked questions: cross sectional study BMJ 2003 327 1144 1146 14615341 10.1136/bmj.327.7424.1144
Bosscher RJ Smit JH Confirmatory factor analysis of the General Self-Efficacy Scale Behaviour Research and Therapy 1998 36 339 343 9642852 10.1016/S0005-7967(98)00025-4
Robinson JR Young TK Roos LL Gelskey DE Estimating the burden of disease. Comparing administrative data and self-reports Med Care 1997 35 932 947 9298082 10.1097/00005650-199709000-00006
Soumerai SB McLaughlin TJ Ross-Degnan D Casteris CS Bollini P Effects of a limit on Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental health services by patients with schizophrenia The New England Journal of Medicine 1994 331 650 655 8052275 10.1056/NEJM199409083311006
Sloan KL Sales AE Liu CF Fishman P Nichol P Suzuki NT Sharp ND Construction and characteristics of the RxRisk-V: a VA-adapted pharmacy-based case-mix instrument. Med Care 2003 41 761 764 12773842 10.1097/00005650-200306000-00009
Braitman LE Davidoff F Predicting clinical states in individual patients. Annals of Internal Medicine 1996 125 406 412 8702092
Martin LM Leff M Calonge N Garrett C Nelson DE Validation of self-reported chronic conditions and health services in a managed care population American Journal of Preventive Medicine 2000 18 215 218 10722987 10.1016/S0749-3797(99)00158-0
Rector TS Wickstrom SL Shah M Greenlee NT Rheault P Rogowski J Freedman V Adams J Escarce JJ Specificity and sensitivity of claims-based algorithms for identifying members of Medicare+Choice health plans that have chronic medical conditions Health Serv Res 2004 39 1839 1857 15533190 10.1111/j.1475-6773.2004.00321.x
Mandelblatt JS Bierman AS Gold K Zhang Y Ng JH Maserjan N Hwang Y Meropol NJ Hadley J Silliman RA Constructs of burden of illness in older patients with breast cancer: A comparison of measurement methods Health Serv Res 2001 36 1085 1107 11775669
Murray CJL Chen LC Understanding morbidity change Population and Development Review 1992 18 481 595
Kieszak SM Flanders WD Kosinski AS Shipp CC Karp H A comparison of the Charlson Comorbidity Index derived from medical record dat and administrative billing data J Clin Epidemiol 1999 52 137 142 10201654 10.1016/S0895-4356(98)00154-1
Melfi C Holleman E Arthur D Datz B Selecting a patient characteristics index for the prediction of medical outcomes using administrative claims data J Clin Epidemiol 1994 48 917 926 7782800 10.1016/0895-4356(94)00202-2
Schneeweiss S Maclure M Use of comorbidity socres for control of confounding in studies using administrative databases Int J Epidemiol 2000 29 891 898 11034974 10.1093/ije/29.5.891
Malenka DJ McLerran D Roos N Fisher ES Wennberg JE Using administrative data to describe casemix:a cmoparison with the medical record J Clin Epidemiol 1994 47 1027 1032 7730905 10.1016/0895-4356(94)90118-X
Wilson IB Cleary PD Linking clinical variables with health-related quality of life: A conceptual model of patient outcomes. JAMA 1995 273 59 65 7996652 10.1001/jama.273.1.59
Kaplan SH Kravitz RL Greenfield S A critique of current uses of health status for the assessment of treatment effectiveness and quality of care. Med Care 2000 38 II-184 II-191 10982106
Noel PH Williams JW Unutzer J Worchel J Lee S Cornell J Katon W Harpole LH Hunkeler E Depression and comorbid illness in elderly primary care patients: Impact on multiple domains of health status and well-being Annals of Family Medicine 2004 2 555 562 15576541 10.1370/afm.143
Ruo B Rumsfeld JS Hlatky MA Liu H Browner WS Whooley MA Depressive symptoms and health-related quality of life JAMA 2003 290 215 221 12851276 10.1001/jama.290.2.215
|
16137329
|
PMC1208932
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Sep 1; 3:51
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-51
|
oa_comm
|
==== Front
Hum Resour HealthHuman Resources for Health1478-4491BioMed Central London 1478-4491-3-41604277810.1186/1478-4491-3-4ResearchUtility of a thematic network in primary health care: a controlled interventional study in a rural area Coma del Corral Maria Jesús [email protected] Luquín Pedro [email protected] Guevara José [email protected] Movilla Angel [email protected] Torres Gerardo [email protected] Garcia Javier [email protected] Research Unit, Hospital General Yagüe, Burgos, Spain2 Department of Nephrology, Hospital General Yagüe, Burgos, Spain3 Health Center of Primary Care, Hospital General Yagüe, Burgos, Spain2005 22 7 2005 3 4 4 16 4 2005 22 7 2005 Copyright © 2005 del Corral et al; licensee BioMed Central Ltd.2005del Corral et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
UniNet is an Internet-based thematic network for a virtual community of users (VCU). It supports a virtual multidisciplinary community for physicians, focused on the improvement of clinical practice. This is a study of the effects of a thematic network such as UniNet on primary care medicine in a rural area, specifically as a platform of communication between specialists at the hospital and doctors in the rural area.
Methods
In order to study the effects of a thematic network such as UniNet on primary care medicine in a rural area, we designed an interventional study that included a control group. The measurements included the number of patient displacements due to disease, number of patient hospital stays and the number of prescriptions of drugs of low therapeutic utility and generic drug prescriptions by doctors. These data were analysed and compared with those of the control center.
Results
Our study showed positive changes in medical practice, reflected in the improvement of the evaluated parameters in the rural health area where the interventional study was carried out, compared with the control area. We discuss the strengths and weaknesses of UniNet as a potential medium to improve the quality of medical care in rural areas.
Conclusion
The rural doctors had an effective, useful, user-friendly and cheap source of medical information that may have contributed to the improvement observed in the medical quality indices.
==== Body
Background
The Internet has tremendous potential for the delivery of instructional materials, allowing collaboration among a virtual community of users (VCU). The future success of the Internet in medicine, among other areas, will depend on their ability to enable physicians to participate in continuous medical education at a time and location convenient for them, along with useful communication. Functions such as electronic mail, chat rooms, mailing lists and computer conferencing enable physicians to establish and maintain communication and collaboration with colleagues investigating the same subjects. This encourages creation of virtual networks, such as UniNet, that allow physicians to share experiences and learn from one another [1,2].
The UniNet project (University Network of Thematic Resources for Virtual User Communities: ) is the first real pilot experience that proposes integrated virtual thematic services for a VCU [3]. A VCU is a group of Internet users who share a set of common aspects of the "knowledge society". UniNet began operating by the end of 1996. The UniNet project aims to be universal, free of language or geographical boundaries and open to all the interests represented in the "knowledge society", but it covers mainly scientific, academic and cultural topics [4]. Currently the major emphasis is on medical and health science VCUs.
The UniNet project is based on the voluntary and altruistic cooperative work of scientists and professionals of many countries across five continents. UniNet also intends to supply information and communication channels on the Internet to every member of the "knowledge society". Another important mission of UniNet is to provide the best resources for the inexperienced user, even one who has no previous knowledge of computers [5,6]. UniNet supports a virtual network on Internet, for a community of doctors oriented to supporting the best practices in patient care.
Goal
Our objective was to evaluate the usefulness of a thematic network such as UniNet, specifically in medical family practice. The goal of the study is to analyse the effects of the thematic network UniNet in the following four outcomes of a primary care medical practice in a rural area: patients' referrals, number of hospital stays, use of prescriptions of drugs of low therapeutic utility and generic drug prescriptions. We tried to quantify the benefits of doctors having access to medical information of high quality, more continuing medical education and easy coordination with specialist physicians.
Methods
At the end of 1998, we presented a research proposal based on the study – before and after – to connect the doctors from a basic zone of health (BZH), located in a rural area, to UniNet through the public Internet. A grant helped us to provide computer resources and Internet connection to the rural medical center, allowing rural health care professionals to join a VCU at the UniNet network for collaborative work with medical specialists from the General Yagüe Hospital by web, text conference and e-mail.
Design
We designed a controlled interventional study on family doctors in a rural health area. A before-and-after study was performed with a control group in order to evaluate the educational intervention.
We pursued this study in a BZH in the province of Burgos, with 5700 inhabitants affiliated with our public health system, and eight doctors – seven of them primary care physicians and one paediatric specialist – who care for patients in three different doctors' offices, located in the larger towns of this rural area: a central office and two peripheral doctors' offices. This BZH is located more than 100 km from their reference center, the General Yagüe Hospital, and the road has three mountain passes. A control group was used: an adjacent BZH, with a population of 4378 affiliated to the public health system, five doctors and a similar environment, nutritional habits, etc. The centers of these areas are 7.5 km apart, and they have the same kind of population.
The intervention consisted of the establishment of a local area network (LAN) with free access to the Internet, and access to the UniNet network. Computer equipment was installed at the offices of every rural doctor. These doctors continuously used different services available on the Internet to retrieve information, perform searches in bibliographic databases, etc. The different services used were:
1. The World Wide Web, which was used as a complex entity with information from a variety of source types, including databases accessible on the Web and from the hospital library's collection of Web-based e-journals and databases. Also, at the UniNet site the doctors could find relevant information including courses, a doctoral program, and congresses organized on the network, such as the 3rd Congress of Nephrology on the Internet (CIN2003), organized by us during the the study. In addition, search engines were used to retrieve and store specific information related to the required health field.
2. Electronic mail (e-mail) was used as a tool of private communication by the rural and specialized doctors. We included the use of a mailing list to transmit information to the participants in this study, and tried to encourage their participation in other meetings, such as the online congress.
3. Chat sessions, or text conferencing, was the most relevant tool. Sessions of continuing medical education were planned as updates on diverse medical subjects as requested by the rural doctors. These sessions were provided by means of the simultaneous connection of the specialized doctors involved and the rural doctors in a private channel of text-conferencing.
All these interventions, including the initial training in Internet tools, occurred during a 12-month period. The global coordination of the project was made by the Research Unit of the General Yagüe Hospital. The educational activities were conducted by specialists from diverse medical departments: nephrology, haematology, rheumatology, neurology, gynaecology, etc. The specialists from the Department of Nephrology led the communication with the BZH and took the lead role in coordinating the educational activities.
For these purposes, each doctor had a personal computer installed in his or her office. In addition, these doctors had access to the databases at the hospital's library, as well as other existing ones freely available via the Internet (Pubmed, Embase, etc).
It took nine months to make the site operational. The first three months of the project were spent on administrative efforts, including equipment orders and installation. During the next three months the software was installed and the network and computers were configured. Afterwards, another six months was used for scheduling and training.
The work plan began with the installation of the local area network (LAN) and the computing equipment. Next, an Introduction to Internet course was delivered, and in the following months, the doctors were trained in computer operation and how to use the diverse services available via the Internet: Web, electronic mail, online databases and text conferencing.
Between January and June 2001, when it was agreed that the doctors had had enough training, we launched the first steps of the continuing medical education program, including clinical sessions, bibliographic reviews, conferences to maintain currency, debate, etc., using live communication by text conferencing, hosted on the UniNet network.
Generally we planned and ran a weekly conference session with the hospital's doctors along with those in the BZH location. A program of subjects of interest presented by the rural doctors was developed. The procedure was as follows: a written document related to the item was sent to the participants before the conference via e-mail, so that it could be read by the rural doctors in advance. Later, at a text conference, all the doctors could discuss the subject more effectively. Every session included bibliographic directions and practical questions. In addition, there was continuous interaction between the specialists and the primary care doctors during the week via e-mail, etc.
In order to measure the effectiveness of these procedures with regard to the effect on the patients, the variation of four parameters, corresponding to the years 2000 (before the intervention) and 2001 (the year of the intervention), was compared.
1. Referral to a hospital specialist. This was defined as every consultation of the patients from the Basics Zone of Health (the BZH of the study and BZH control), requested by their family doctor, to the departments of the hospital that corresponded to any of the CIE-10 codes.
2. Number of hospital stays. This means the number of nights that the patient of the BZH stayed in the hospital, occupying a bed.
3. The proportion of prescriptions of drugs of low therapeutic utility (UTB) with regard to the total of prescribed drugs. One considers the crude rate and the standardized rate, adjusted for the active population versus retired persons; the Spanish population census and the direct method of standardization were used to adjust the rate. The UTB drugs are found in a list of UTB, described annually by the Directorate of Public Health (INSALUD)
4. Generic drug prescription: the proportion of prescribed generic medicines relative to the total.
All these data were gathered through a blind and independent method with respect to this study by the Directorate of Public Health in Burgos. The doctors of the BZH of the study did not know the evaluation parameters, and the BZH control remained isolated from this investigation.
Results
The number of consultation visits to specialists for patients of both BZH shows a reduction of the number and rates of derivation in both, but this reduction is more pronounced in the BZH under study, where the intervention took place. The total reduction, measured by difference between the relative risk reduction in the studied BZH and the BZH control was 4.6% in nephrology, 19.36% in all medical specialties and 13.89% in global consultations, as we can see in Table 1.
Table 1 Number (N) and rate of consultations per 1000 inhabitants (°/°°) of nephrology specialists to all medical specialties, and total of consultations (medical and surgical specialties) of patients from both centers, relative to the reference hospital. RRR: relative risk reduction.
Consultations Center Year 2000
N (°/°°) Year 2001
N (°/°°) Variation 2001–2000
N (°/°°) RRR Difference in RRR (Total effect)
Nephrology Studied BZH 15 (2.63) 10 (1.75) -5 (-0.88) -33.46% -4.6%
Control 10 (2.28) 7 (1.6) -3 (-0.68) -29.86%
Medical (including Nephrology) Studied BZH 701 (122.98) 422 (74.05) -279 (-48.93) -39.78% -19.36%
Control 475 (108.5) 378 (86.34) -95 (-22.16) -20.42%
Totales (Medical and Surgical Specialities) Studied BZH 1685 (295.61) 1189 (208.59) -496 (-87.01) -29.43% -13.89%
Control 1164 (265.87) 983 (224.53) -181 (-41.34) -15.54%
Hospital admissions (Table 2), measured in absolute numbers, rate of derivation by 1000 inhabitants and the average stay all decreased in both BZH with respect to the nephrology department. This rate is lower in the studied BZH than in the BZH control and it does not vary in the studied BZH, while it decreases in the BZH control. The total average hospital stay decreased in the patients from the studied BZH, whereas it increased for the patients in the BZH control.
Table 2 Hospitalization of patients in the Department of Nephrology and the total hospitalization of patients for any specialty at the study center and control center: number (N), rate per 1000 inhabitants (°/°°) and average stay (AS). RRR: relative risk reduction in the rate.
Hospitalization Center Year 2000
N (°/°°)
AS Year 2001
N (°/°°)
AS Variation 2001–2000
N (°/°°)
AS RRR Difference in RRR
(Total effect)
Nephrology Studied BZH 5 (0.87)
8.4 days 4 (0.7)
8.75 days -1 (-0.17)
+0.35 days -19.54% 0.19%
Control 10 (2.28)
5.9 days 8 (1.83)
14.25 days -2 (-0.45)
+ 8.35 days -19.73%
Total Studied BZH 447 (78.42)
8.44 days 449 (78.77)
7.12 days -2 (+ 0.35)
-1.32 days 0.44% 4.93%
Control 410 (93.65)
7.38 days 388 (88.62)
8.65 days -22 (-5.03)
+ 1.27 days -5.37%
With respect to the prescription of drugs of low therapeutic utility (UTB), there was a reduction of 0.47% in the study BZH, whereas in the control BZH the reduction was of 0.04%, as shown in Table 3. The percentage of UTB, adjusted for the active population versus retired persons, shows a reduction of 0.87% in the study BZH and an increase of 0.32% in the control BZH. The difference between RRR was a reduction of 15.31%, as shown in Table 4.
Table 3 Percentage of prescription of drugs of low therapeutic utility (UTB) in the rural study center and the control center. RRR = relative risk reduction.
% UTB Year 2000 Year 2001 Variation 2001–2000 RRR Difference in RRR (Total effect)
Studied BZH 6.75% 6.28% -0.47% -6.96% -6,26%
Control 5.50% 5.46% -0.04% -0.72%
Table 4 Percentage of UTB drugs prescribed relative to the active population versus retired persons in the study center and the control center. RRR= relative risk reduction.
Adjusted UTB Year 2000 Year 2001 Variation 2001 – 2000 RRR Difference in RRR (Total effect)
Studied BZH 8.84% 7.97% -0.87% -9.84% -15.31%
Control 5.85% 6.17% 0.32% 5.47%
With regard to the proportion of generic drug prescriptions, the results are shown in Table 5. The period of study coincides with an institutional directive to preferentially prescribe generic drugs in order to cut costs. As we can see, both centers increased their generic medicine prescription rate, but the study BZH did so at a greater rate.
Table 5 Percentage of generic drug prescriptions compared to the total number of prescribed drugs. RRR = relative risk reduction.
Generics Year 2000 Year 2001 Variation 2001–2000 RRR Difference in RRR (Total effect)
Studied BZH 5.3% 8.53% 3.23% 60.94% 36.74%
Control 5.0% 6.21% 1.21% 24.20%
Discussion
The results of our study showed positive changes in medical practice, reflected in the improvement of the evaluated parameters in the study BZH compared with the control BZH during the study period. We will now examine the real effect of the intervention on those positive changes; that is to say, the influence that it is possible to attribute to the UniNet network on the improvement of the medical practice indicators. The limitations of our study are related to its design, in that it is quasi-experimental. However, the existence of a control group of similar characteristics permits us to avoid many of these difficulties and lends weight to the results.
The analysis of Dorsch [7] indicates that rural doctors appear to have the same basic information needs as their urban counterparts, and that both groups rely on colleagues and personal libraries as their main sources of information. Rural practitioners, however, tend to make less use of journals and online databases and ask fewer clinical questions, a difference that correlates with geographical and demographic factors. Rural practitioners find many barriers to information access, including lack of time, isolation, inadequate library access, lack of equipment and skills, higher costs and an inadequate Internet infrastructure.
The Internet removes some of the isolation barriers and it has the potential to facilitate communication among rural health professionals and urban specialists. This was the goal of UniNet, and it was specifically the objective of this study. The integration of professionals into a VCU such as UniNet will not totally compensate for the lack of local library services, but electronic communications, such as synchronous text conferencing, provide a way to incorporate learning activities through expert collaboration, with a useful and practical model of continuing medical education. The formation of these virtual learning networks can allow physicians to reflect upon, generalize and discuss the applications of new information with their peers [8]. A common pitfall of many information technology programs in rural health centers is to give priority to building the technical infrastructure rather than focusing on meeting local needs and developing local expertise and ownership [9]. Text conferencing is a long-distance communication tool that is simple, powerful and cheap [10], all reasons why it is a useful and valid tool for the objectives.
Curran et al. [1] consider that the Internet is useful for continuing medical learning because it enables personal one-to-one communications: that is, the doctors, both students and instructors, may communicate with their peers through electronic mail. The Internet enables access to library resources, interactive multimedia tutorials and other related clinical and academic resources. Asynchronous or synchronous group communications allow participation in collaborative discussions with colleagues, instructors and experts via asynchronous computer conferencing or online, real-time chat groups and the like.
The limited Internet usage in our professional environment constituted the main hurdle for the study [11]. When we initiated this study, only one of the doctors used Internet services. It required a relatively long training period to achieve the mandatory prerequisite that most professionals in the BZH develop basic Internet abilities. The motivational level among the clinical staff to use the workstations varied, depending on their age, work style and the presence of a leader who would lead by example, as we found in the rural health center. Regardless of the importance of the information needed, busy clinicians will use the service only if the computer connections and interfaces are convenient and easy to use [12].
The project achieved its goals of ensuring that family doctors had access to good-quality, user-friendly, cost-effective medical information. In addition, the family doctors were assisted by a group of specialists at the hospital who served as mentors. The location of the computer was a key factor in ensuring that it was well used for clinical decision making. The doctors at the BZH under study learned how to search Medline, how to locate and gain access to the best information resources on the Internet and how to request materials through the General Yagüe Hospital library. Recently, the National Library of Medicine has developed a similar project, introducing aids that allow access to bibliographic databases (Medline), through the Internet in rural health centers, with initial success [13].
More than a technical tool to facilitate access, information technology can also serve to build and strengthen communities of like-minded individuals and institutional networks. Therefore it is necessary to form a VCU that allows for information selection. A deluge of information, much of which concerns technologically sophisticated treatments or expensive medications, is not always applicable to a rural health center. Information is relevant and useful if it emphasizes prevention and promotes evidence based, cost-effective treatment strategies [14-16].
The Internet differs from other media in one major way: the communication process is bidirectional, with an information source in the active role and the receiver in a more passive role [17]. Westberg and Miller [18] propose a model in which the academic health center integrates and distributes a wide range of electronic and human resources. This model will require substantial funding for online, full-text journal collections and networked bibliographic databases that may be even more expensive than print collections. The advent of electronic full-text journals may improve the information access of rural health professionals.
According to Marshall [19]: "the timely use of publicly accessible, electronic databases containing bibliographic and full-text information has the potential to assist in the maintenance of health professional competence, decrease the isolation and lack of up-to-date knowledge experienced by health professionals practicing outside of major population centers, and improve the quality of patient care by narrowing the gap between the publication of scientific findings and their application by researchers and clinicians". Searching for medical information via computers is certainly still not the norm [20]. Although computers provide access to an overabundance of up-to-date information, many physicians still find it more convenient to read a textbook or a journal or to ask a colleague. The utility of a VCU such as UniNet lies in its access to biomedical knowledge, colleagues and to faculty [21-24].
Conclusion
Our study found that providing computer resources and Internet connection to a rural medical center enabled rural doctors to join a VCU hosted by the UniNet network for collaborative work with medical specialists and allowed for access to high-quality medical information. Through this network, rural doctors had an effective, useful, user-friendly and cheap source of medical information, which may be related to the improvements observed in the medical quality indices.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All the authors contributed equally to the design, data collection, data analysis, drafting and completion of this article.
Acknowledgements
This work was financed by grant FIS 99/0324 (Fondo de Investigación Sanitaria – Ministerio de Sanidad) and cofinanced by the European Union (FEDER funds). We also wish to thank Jose Nazario, PhD, for assistance in revising the English translation of this paper.
==== Refs
Curran VR Hoekman T Gulliver W Landells I Hatcher L Web-based continuing medical education (I): field test of a hybrid computer-mediated instructional delivery system J Contin Educ Health Prof 2000 20 97 105 11232226 10.1002/chp.1340200206
Sánchez-Ramos M Coma del Corral MJ Moro-Rodríguez E Conde Martín AF Vieira e Silva CR Hernández Martín A Pat-UniNet: Creación y desarrollo foro internacional, multilingüe, de diagnóstico en Patología Patología 2003 36 149 158
Hawa Attourah M Coma del Corral MJ The virtual communities of users and the integrated virtual thematic services First International Congress of Neuropsychology in the Internet, Nov 1-Dec 15 1999, Invited Conference
Coma del Corral MJ Martin-Alganza A Hawa-Attourah M La comunicacion en Internet. UniNet: la red universitaria de servicios telematicos integrados Rev Neurol 1998 26 992 995 9658476
Coma del Corral MJ Martin-Alganza A Hawa Attourah M Telematic services and communication through scientific irc: virtual user communities of biomedicine in UniNet INABIS '98 – 5th Internet World Congress on Biomedical Sciences at McMaster University, Canada, Dec 7–16th 1998, Invited Symposium (II).
Fernandez Carro A Douglas M Da Costa A Lopez M Developments in educational biomedical courses, meetings and symposia INABIS '98 – 5th Internet World Congress on Biomedical Sciences at McMaster University, Canada, Dec 7–16th Invited Symposium 1998
Dorsch JL Information needs of rural health professionals: a review of the literature Bull Med Libr Assoc 2000 88 346 354 11055302
Curran VR Hoekman T Gulliver W Landells I Hatcher L Web-based continuing medical education. (II): Evaluation study of computer-mediated continuing medical education J Contin Educ Health Prof 2000 20 106 119 11232219 10.1002/chp.1340200207
Black J Health information for the developing world. Obsession with high technology Br Med J 1994 309 1437 7695732
Coma del Corral MJ Alfaro L, García Rojo M, Puras A Integración de recursos telemáticos Manual de Telepatología Club de Informática Aplicada de la SEAP Ciudad Real 2001 199 203
Rauch S Holt MC Horner M Rambo N Community hospitals and the Internet: lessons from pilot connections Bull Med Libr Assoc 1994 82 401 406 7841910
Byrnes JA Kulick TA Schwartz DG Information-seeking behavior changes in community-based teaching practices J Med Libr Assoc 2004 92 334 340 15243639
McDuffee DC AHEC library services: from circuit rider to virtual librarian Bull Med Libr Assoc 2000 88 362 366 11055304
Jabbour S Nishtar S Prabhakaran D Chockalingam A Achutti A Agrawal A Luna MA Garofalo F Wielgosz A Information and communication technology in cardiovascular disease prevention in developing countries: hype and hope. Report of the International Collaboration on Information Use in Cardiovascular Health Promotion in Developing Countries Int J Cardiol 2003 92 105 111 14659841 10.1016/S0167-5273(03)00093-7
Bennett NL Casebeer LL Kristofco RE Strasser SM Physicians' Internet information-seeking behaviors J Contin Educ Health Prof 2004 24 31 38 15069910 10.1002/chp.1340240106
Beniger M Pluye P Shortcomings of health information on the Internet Health Promot Int 2003 18 381 386 14695369 10.1093/heapro/dag409
van Woerkum CM The Internet and primary care physicians: coping with different expectations Am J Clin Nutr 2003 77 1016S 1018S 12663310
Westberg EE Miller RA The basis for using the Internet to support the information needs of primary care J Am Med Inform Assoc 1999 6 6 25 9925225
Marshall JG Characteristics of early adopters of end-user online searching in the health professions Bull Med Libr Assoc 1989 77 48 55 2720210
Steiner V Hartmann J Ronau T MedReach: building an Area Health Education Center medical information outreach system for northwest Ohio J Med Libr Assoc 2002 90 317 322 12113517
Coma MJ Díaz-Alersi R Almeida E Textoconferência em MedicinaIntensiva: A experiencia Uninet Rev Port Med Intensiv 2001 10 79 82
Díaz-Alersi R Coma MJ Almeida E Nuñez A Comunidades virtuales de usuarios en Uninet Rev Port Med Intensiv 2001 10 74 75
Gantenbein RD Designing an Internet-based collaboratory for biomedical research Biomed Sci Instrum 2002 38 399 404 12085639
Bers MU Gonzalez-Heydrich J Demaso DR Use of a computer-based application in a pediatric hemodialysis unit: a pilot study J Am Acad Child Adolesc Psychiatry 2003 42 493 496 12649637
|
16042778
|
PMC1208933
|
CC BY
|
2021-01-04 16:37:43
|
no
|
Hum Resour Health. 2005 Jul 22; 3:4
|
utf-8
|
Hum Resour Health
| 2,005 |
10.1186/1478-4491-3-4
|
oa_comm
|
==== Front
Int J Behav Nutr Phys ActThe International Journal of Behavioral Nutrition and Physical Activity1479-5868BioMed Central London 1479-5868-2-91607638610.1186/1479-5868-2-9ResearchPsychosocial predictors of eating habits among adults in their mid-30s: The Oslo Youth Study follow-up 1991–1999 Kvaavik Elisabeth [email protected] Nanna [email protected] Grethe S [email protected] Knut-Inge [email protected] Department of Nutrition, University of Oslo, Norway2 Department of Public Health and Primary Health Care, University of Bergen, Norway3 Department of Research and Development, Haukeland University Hospital, Bergen, Norway2005 2 8 2005 2 9 9 14 12 2004 2 8 2005 Copyright © 2005 Kvaavik et al; licensee BioMed Central Ltd.2005Kvaavik et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The predictive value of the psychosocial constructs of Theory of Planned Behaviour (TPB) on subsequent dietary habits has not been previously investigated in a multivariate approach that includes demographic factors and past dietary behaviour among adults. The aim of this study was to investigate to what extent TPB constructs, including intention, attitudes, subjective norms, perceived behavioural control, and perceived social norms, measured at age 25 predicted four eating behaviours (intake of fruits and vegetables, whole grains, total fat and added sugar) eight years later.
Methods
Two hundred and forty men and 279 women that participated in the Oslo Youth Study were followed from 1991 to 1999 (mean age 25 and 33 years, respectively). Questionnaires at baseline (1991) included the constructs of the TPB and dietary habits, and at follow-up (1999) questionnaires included demographic factors and diet. For the assessment of diet, a food frequency questionnaire (FFQ) with a few food items was used at baseline while an extensive semi-quantitative FFQ was used at follow-up.
Results
Among men, attitudes, subjective norms and previous eating behaviour were significant predictors of fruit and vegetable intake, while education and past eating behaviour were predictive of whole grain intake in multivariate analyses predicting dietary intake at follow-up. For women, perceived behavioural control, perceived social norms and past behaviour were predictive of fruit and vegetable intake, while subjective norms, education and past eating behaviour were predictive of whole grain intake. For total fat intake, intention was predictive for men and perceived behavioural control for women. Household income and past consumption of sugar-rich foods were significant predictors of added sugar intake among men, while past intake of sugar-rich foods was a significant predictor of added sugar intake among women.
Conclusion
After adjusting for potential confounding factors, all psychosocial factors assessed among young adults appeared predictive of one or more eating behaviours reported eight years later. Results point to the influence of psychosocial factors on future eating behaviours and the potential for interventions targeting such factors.
==== Body
Background
The relationship between psychosocial factors and dietary behaviours over time is not clear [1,2]. With the exception of intervention studies that have used psychosocial models to predict dietary changes among adults over periods ranging from a few weeks to a few years [3-6] only a few studies have used longitudinal designs to investigate the relationship between psychosocial factors and dietary behaviours among adults [7-10]. A study by Armitage and Conner [7] employed the Theory of Planned Behaviour (TPB, including the constructs behavioural intention, attitudes, subjective norms and perceived behavioural control) and reported intentions assessed at baseline to be predictive of dietary behaviour (fat intake) three months later. Conner and colleagues [8] also employed the TPB and found that healthy eating behaviour was predicted from intentions and perceived past behaviour measured six years earlier. Kristal and colleagues [9] applied the Stages of Change Model in their study and implemented a follow-up survey after two years. They found that reduction in fat intake and increase in fruit and vegetable intake were differently predicted by psychosocial factors. Respondents who were in the later stages of change for eating a low fat diet and who read food labels made the largest reductions in fat intake, while changes in fruit and vegetable intakes were small and did not reach significance across psychosocial factors. Patterson and colleagues [10] used the psychosocial factors beliefs, knowledge and perceived norms about diet and cancer to investigate the prediction of dietary change over three years. They found that respondents with a strong belief in the diet-cancer connection and those with knowledge about recommendations regarding diet and cancer reduced their fat intake and increased their fibre intake (knowledge only) more than those with no beliefs and no knowledge about the recommendations. However, food composition knowledge and perceived pressure to eat a healthful diet did not predict changes in fat or fibre intake. Given that the TPB constructs – attitudes toward behaviour, subjective norms, perceived behavioural control and behavioural intention – differ from other models, dietary predictions found with the TPB model can not be directly compared to predictions found via other models. Thus, further investigation is needed to determine the ability of TPB's psychosocial constructs to predict dietary intakes among adults. In addition, dietary habits are known to vary with socioeconomic status, with those in higher social classes having more healthful diets than those in lower social classes [11-15]. The association between social class and diet is also seen in Norway [16,17]. Few studies have investigated long term associations between psychosocial and socioeconomic factors and diet among adult men and women [9,18-21].
A previous cross-sectional analysis of Oslo Youth Study participants (at mean age 25 years) found that components of the TPB accounted for 32% of the variance in behavioural intention to eat healthier food during the four weeks following the survey [22]. The aim of the current study was to utilize a longitudinal design to examine to what extent these TPB constructs measured at age 25 years predicted key elements of eating behaviour eight years later. In particular, we were interested in examining the explanatory power of the psychosocial constructs after adjustment for potential confounders, such as socio-demographic factors and previous eating habits.
Subjects and methods
Design and subjects
The Oslo Youth Study was initiated in 1979 with participants in 5th–7th grade (mean age 13 years, range 11–16 years) from six schools in Oslo, Norway. The purposes of the study were to obtain epidemiological data on risk factors for cardiovascular disease and cancer and to evaluate the effects of a controlled intervention program to prevent the onset of smoking, increase physical activity and improve eating habits [23]. Participants were invited to take part in a follow-up survey in 1981. In 1991 and 1999, the same subjects, with average ages of 25 and 33 years, respectively, were invited to participate in follow-up studies involving self-administered questionnaires. In 1991, psychosocial constructs taken from the TPB and related to healthy eating were included for the first time. Data from subjects who participated in both the 1991 and 1999 surveys are reported here. In 1991, 706 of 947 eligible subjects participated (74.6%); of these, 526 also participated in 1999. However, as seven participants did not complete the dietary questionnaire in 1999, the final cohort consists of 519 subjects or 73.5% of those eligible in 1991.
As part of the Oslo Youth Study, an intervention program was implemented between the 1979 and 1981 surveys in half of the participating schools. Both students who received the intervention and the controls are included in this study. The intervention is described in detail elsewhere [24].
The Oslo Youth study was approved by the Norwegian Data Inspectorate, as well as the City of Oslo's health authorities.
Outcome measures – dietary habits
In 1999, participants completed a validated quantitative food frequency questionnaire designed to assess usual diet during the past year [25]. The questionnaire included 180 food items grouped together according to the typical Norwegian diet and meal pattern. Questions were phrased as to tap the usual intake, and both frequencies (10 response alternatives) and amounts (ranging between four and 14 response alternatives dependent on food item) were reported by the participants. Four outcome variables were constructed: 1) Fruit and vegetable intake (grams per day, including boiled potatoes in accordance with the Norwegian dietary recommendations [26] and a maximum of 150 grams of orange juice); 2) Whole grain intake (grams per day, including whole wheat bread, unsweetened breakfast cereals and oat porridge); 3) Total fat intake (percent of total energy intake); and 4) Added sugar intake (percent of total energy intake).
Predictor variables
The constructs from TPB (intention, attitudes, subjective norms and perceived behavioural control with respect to healthy eating) were assessed in 1991 [22]. In the 1991 questionnaire, "healthy food" was loosely defined as "foods low in fat, sugar and salt." Behavioural intention was measured by one question: "How likely is it that you will eat healthier food during the next four weeks?" with response options ranging from (1) "Very unlikely" to (5) "Very likely."
Attitudes
Seven beliefs were assessed by a probability scale ranging from (1) "Very unlikely" to (5) "Very likely" (e.g., "If I am eating healthier food the next four weeks my cholesterol level will be reduced"). In addition to the belief regarding reducing cholesterol levels, beliefs about reducing body weight, be more fit, reducing risk of coronary heart disease, looking younger, reducing cancer risk and enjoying food more were assessed. The corresponding outcomes were measured by means of the question "How important is it for you to [e.g., reduce your cholesterol level]?" using a scale ranging from (1) "Not important at all" to (4) "Very important." Each belief item was multiplied by the corresponding outcome item and the products were used to construct an indirect measure of attitude (range: 1–20, Cronbach's α: 0.80).
Subjective norms
Normative beliefs were assessed by six items on a scale ranging from (1) "Very unlikely" to (5) "Very likely": "Do you believe that your parents/siblings/friends/partners/physician/co-workers think that you should eat healthier food the next four weeks?" Response alternatives included "I do not know," coded as neutral and "I do not have parents/siblings etc.," coded as missing. Motivation to comply with significant others was measured by the following question: "How important is it for you to comply with...?" on a scale ranging from (1) "Not important at all" to (4) "Very important." The responses to each normative belief were multiplied by the corresponding item for motivation to comply. The products were used to construct a subjective norm scale (range: 1 – 20, Cronbach's α: 0.83).
Perceived behavioural control
Perceived behavioural control was assessed by six questions measuring participants' beliefs in eating healthier foods under specific circumstances and two questions measuring participants' beliefs in preparing healthier dishes when busy or when tired. The question "To what extent do you believe you are able to eat healthier food if you are...?" addressed these eight situations, e.g., "with people who eat unhealthy food." In addition, a global question was asked about the extent to which participants felt able to prepare healthier dishes. The response scale ranged from (1) "Very little" to (4) "Very much." The responses to each of the nine beliefs were used to construct a perceived behavioural control scale (range: 1 – 4, Cronbach α: 0.82).
Perceived social norms
In addition to the constructs from TPB noted above, perceived social norms were measured by asking participants five questions on how important healthy eating was to their parents, partner, best friend(s), siblings, and co-workers using a scale ranging from (1) "Not important at all" to (4) "Very important." Those who checked "I do not know" were coded neutral and those who checked "I do not have....." were coded as missing. The responses to each of the five questions were used to construct a perceived social norms scale (range: 1 – 4, Cronbach's α: 0.69). For a listing of all items, see Øygard and Rise, 1996 [22].
Dietary habit score
In 1991, a food frequency questionnaire was used to assess dietary intake. The questionnaire included 30 food items and the five response alternatives regarding frequencies ranged from seldom/never to more than once daily. Information about amounts was not collected, and therefore an assessment of total energy intake was not possible. A Fruit and vegetable score was composed of reported intake of fruits, vegetables and orange juice. A Whole grain score was composed of reported intake of whole wheat bread and breakfast cereals. A Fat score was composed of reported consumption of plant margarine, butter, whole fat milk, French fries/potato chips, meat balls, hamburgers and sausages and, similarly, a Sugar score was composed of reported intake of chocolate/sweets, cakes/buns and sugar sweetened, carbonated soft drinks. All scores for single food items ranged from (1) "Eat never or seldom" to (5) "Eat several times daily." A composite score was composed by summing values of single food items included in the specific score and dividing by the number of food items included, resulting in a range from one to five for all four dietary habit scores.
Demographic variables measured in 1999
Educational attainment reported in 1999 was classified in five categories, ranging from "9 years of elementary/secondary school (or less)" to "More than 4 years at college/university." The household's annual income was reported in six categories ranging from "<NOK 200000" to "≥NOK 600000." If household income was missing, personal income was used. This was the case for four women (two married, one single and one divorced) and for five men (all married). Marital status was classified as "Married or co-habiting" versus "Single," which also included those divorced (n = 18) or widowed (n = 1).
Statistics
Unpaired t-tests and chi-square tests were used to compare baseline characteristics between follow-up participants and non-participants, and to compare men and women.
Pearson's correlation coefficient was used to investigate bivariate associations between all independent and dependent variables. Linear regression analysis was used to predict eating habits at follow-up. The models were tested for interaction by gender and by intervention/control status in 1979/81. As there was significant interaction with gender, men and women were analyzed separately. There was no interaction with intervention status, thus intervention and control groups were combined in the analyses. In the regression analysis, we entered baseline psychosocial factors in model one, demographic factors in model two and past eating behaviour in model three.
The statistical software package SPSS 11.0 for Windows was used in all analyses.
Attrition analysis
Of the 706 subjects who participated in 1991, 180 did not participate in 1999. No significant baseline differences were found between those who participated and those who did not in 1999.
Results
Significant gender differences were observed for baseline dietary and psychosocial factors (Table 1). Compared to men, women reported higher scores on fruit and vegetable intake, attitudes and perceived behavioural control, while men reported higher scores on fat and sugar intake and perceived social norms. At follow-up in 1999, no significant differences existed between men and women regarding the dependent dietary variables (Table 2).
Table 1 Psychosocial and dietary factors at baseline (mean age 25 years). The Oslo Youth Study 1991
Men (n = 238*) Women (n = 279*) p-value**
Fruit and vegetable score Range
Mean (SD) 1.0 – 5.0
2.6 (0.8) 1.0 – 5.0
3.0 (0.9) <0.001
Whole grain score Range
Mean (SD) 1.0 – 5.0
2.5 (1.0) 1.0 – 5.0
2.5 (1.0) 0.925
Fat score Range
Mean (SD) 1.0 – 5.0
2.0 (0.8) 1.0 – 4.3
1.8 (0.7) <0.001
Sugar score Range
Mean (SD) 1.0 – 5.0
2.1 (0.7) 1.0 – 5.0 2.0
(0.7) 0.004
Attitude Range
Mean (SD) 2.4 – 16.7
8.7 (3.1) 2.4 – 18.9
9.9 (3.1) <0.001
Subjective norm Range
Mean (SD) 1.2 – 15.0
5.6 (3.1) 1.2 – 18.3
5.7 (3.1) 0.612
Perceived behavioural control Range
Mean (SD) 1.1 – 4.0
2.4 (0.5) 1.2 – 4.0
2.6 (0.5) <0.001
Perceived social norms Range
Mean (SD) 1.0 – 4.0
2.8 (0.5) 1.0 – 4.0
2.6 (0.5) 0.029
* n differ slightly between different variables due to missing values. ** p-value for difference between men and women.
Table 2 Dietary and demographic factors at follow up (mean age 33 years). The Oslo Youth Study 1999.
Variable Categories Men (n = 240*) Women (n = 279*) p-value**
Fruit and vegetable intake Grams per day, mean (SD) 335 (192) 367 (213) 0.074
Whole grain intake Grams per day, mean (SD) 143 (128) 125 (99) 0.070
Total fat intake Per cent from total energy intake, mean (SD) 31.5 (5.9) 32.1 (5.9) 0.250
Added sugar intake Per cent from total energy intake, mean (SD) 10.9 (7.7) 10.3 (6.3) 0.403
Education ≤9 years 6.7 5.8 0.004
10 – 11 years 23.3 11.6
12 years 25.4 29.1
13 – 17 years 25.4 35.3
≥17 years 19.2 18.2
Household income previous year <200 000 NOK 8.4 12.7 0.048
200 000 – 299 000 NOK*** 15.8 23.6
300 000 – 399 000 NOK 19.2 13.1
400 000 – 499 000 NOK 24.8 20.0
500 000 – 599 000 NOK 17.9 15.6
≥600 000 NOK 13.7 14.9
Marital status Married or co-habitant 71.5 73.2 0.696
Single 28.5 26.8
Children, number No children 46.3 30.4 0.002
1 child 20.8 24.6
2 children 25.0 37.0
3 children or more 7.9 8.0
* n differ slightly between different variables due to missing values. ** p-value for difference between men and women. ***NOK = Norwegian krone.
Table 3 presents bivariate correlations between independent variables in 1991 (except dietary scores) and 1999. The correlation coefficients ranged from zero to moderately high. The strongest internal correlations were found among the psychosocial factors, between attitude and subjective norms (0.38 and 0.48 for women and men, respectively), and between attitude and intentions (0.47 and 0.50 for women and men, respectively). For men, the only psychosocial and demographic factors associated with dietary intakes were perceived social norms (with added sugar intake) and education in 1999 (with whole grain intake in 1999). For women, subjective norms, perceived behavioural control and perceived social norms measured in 1991 were associated with two or three of the dietary habits measured in 1999. Furthermore, all demographic factors were associated with fruit and vegetable intake among women, while education in addition were associated with whole grain, fat and sugar intake among women. Bivariate correlation coefficients (Pearson's r) between dietary intake in 1991 and 1999 among men were 0.31 for fruits and vegetables, 0.28 for whole grains, 0.13 for fat and 0.34 for sugar. Corresponding values for women were 0.41, 0.34, 0.17 and 0.30, respectively, and all p-values were <0.001 except for fat intake (p = 0.05 for men and p < 0.01 for women).
Table 3 Inter-correlations (Pearson's r) between independent and dependent variables by gender (N≈502). The Oslo Youth Study 1991–99.
Men/Women 1 2 3 4 5 6 7 8 9 10 11 12 13
1) Attitude 1991 0.48*** 0.20** 0.23*** 0.50*** -0.05 -0.07 0.12 -0.14* 0.08 0.01 0.06 0.04
2) Subjective norms 1991 0.38*** -0.07 0.11 0.31*** -0.11 -0.05 0.04 0.02 -0.09 -0.05 0.08 0.09
3) Perceived behavioural control 1991 0.13* -0.14* 0.17** 0.28*** 0.21** -0.09 0.08 -0.17** 0.02 0.10 -0.02 -0.12
4) Perceived social norms 1991 0.10 0.05 0.22*** 0.11 0.03 -0.13* 0.09 -0.02 0.09 0.03 -0.01 -0.13*
5) Intention 1991 0.47*** 0.20* 0.31*** 0.11 0.06 0.02 0.06 0.02 0.00 0.08 -0.11 -0.06
6) Education 1999 -0.02 -0.18** 0.14* 0.14* -0.04 0.21** -0.09 -0.14* 0.08 0.20** -0.10 -0.05
7) Household income 1999 -0.11 -0.05 0.08 -0.01 -0.09 0.28*** -0.44*** 0.16* 0.05 0.05 0.00 -0.10
8) Marital status 1999 0.04 -0.02 0.03 -0.07 0.02 -0.04 -0.53*** -0.30*** -0.08 0.00 -0.09 0.01
9) Number of children 1999 -0.05 -0.04 0.04 0.08 -0.03 -0.18** 0.10 -0.35*** -0.02 -0.04 0.04 0.10
10) Fruit and vegetable intake 1999 0.05 -0.06 0.27*** 0.22*** 0.09 0.15* 0.12* -0.18** 0.13* 0.20** -0.02 -0.14*
11) Whole grain intake 1999 0.08 -0.19** 0.16** 0.07 0.03 0.25*** 0.04 0.03 -0.11 0.21*** -0.22*** -0.22**
12) Total fat intake 1999 0.01 0.08 -0.29*** -0.08 -0.03 -0.15* -0.11 0.00 0.05 -0.20** -0.26*** -0.10
13) Added sugar intake 1999 0.05 0.12* -0.03 -0.12* 0.04 -0.13* -0.08 -0.02 0.02 -0.12* -0.12* -0.07
Upper half represents men, and lower half represents women. All variables are coded in ascending order except marital status that is coded 1 = married, 2 = single/divorced/widow.
*p < 0.05, **p < 0.01, ***p < 0.001.
Table 4 and 5 present results from multivariate analyses of predictors of dietary habits at follow-up. For men, all significant predictors in model one and two remained significant in the consecutive models. This was, however, not so for women. In model two, education was predictive of women's fruit and vegetable intake, while it lost its significance in model three when adjusting for previous behaviour (Table 4). Subjective norms and perceived social norms appeared predictive of women's intake of added sugar in model one; however, this significance disappeared when adjusting for previous behaviour in model three (Table 5). In the final models for men, attitudes, baseline subjective norms and corresponding eating behaviour at baseline remained significant predictors of intake of fruits and vegetables, while education and corresponding eating behaviour at baseline remained predictive of whole grain intake (Table 4). For women, perceived behavioural control, perceived social norms and corresponding eating behaviour at baseline were predictive of fruit and vegetable intake, while subjective norms and education, in addition to corresponding eating behaviour at baseline, were predictive of whole grain intake in the multivariate models (Table 4). For total fat intake at follow-up, intention and marital status were became significant predictors for men and perceived behavioural control was the only significant predictor for women (Table 5). For added sugar intake, household income and past intake of sugar-rich foods were significant predictors for men (Table 5). For women, only past intake remained a significant predictor of added sugar intake at follow-up (Table 5).
Table 4 Baseline (age 25) predictors of fruits and vegetable and whole grain intake at follow up (age 33 years). Multiple linear regression analyses; unstandardized (B) and standardized regression coefficients (β). The Oslo Youth Study 1991 – 1999.
Dietary habits at follow-up Predictors Daily fruits and vegetable intake Daily whole grain intake
Model 1 Men Women Men Women
Psychosocial factors at baseline B β B β B β B β
Attitude 10.43 0.17* 1.07 0.02 -1.51 -0.04 3.84 0.13
Subjective norm -12.92 -0.18* -4.41 -0.05 -2.72 -0.06 -7.91 -0.22**
Perceived behaviour control -11.48 -0.03 88.07 0.21** 23.68 0.09 22.18 0.12
Perceived social norms 33.52 0.08 79.74 0.18** 5.98 0.02 4.18 0.02
Intention -5.33 -0.03 4.99 0.02 12.08 0.09 1.57 0.02
R2/R2 adjusted, % 3.7/1.6 10.1/8.4 2.1/0.0 7.1/5.4
Model 2
Model 1 + Demographic factors at follow-up
Attitude 11.17 0.18* 1.66 0.02 -1.31 -0.03 3.67 0.12
Subjective norm -12.51 -0.17* -2.46 -0.03 -1.56 -0.03 -7.41 -0.20**
Perceived behaviour control -17.02 -0.04 85.70 0.20** 16.62 0.06 18.88 0.10
Perceived social norms 36.35 0.09 66.43 0.15* 2.64 0.01 2.19 0.01
Intention -6.54 -0.03 7.36 0.02 10.89 0.08 1.55 0.02
Education 9.70 0.06 25.40 0.13* 18.03 0.17* 15.24 0.17*
Household income 2.03 0.02 -4.21 -0.03 1.92 0.02 1.21 0.02
Marital status -27.58 -0.06 -73.40 -0.15 9.61 0.03 -1.86 -0.01
Children -4.16 -0.02 17.05 0.08 -0.37 -0.00 -8.30 -0.08
R2/R2 adjusted, % 4.7/0.8 14.2/11.2 5.1/1.3 11.8/8.7
Model 3
Model 2 + baseline eating behaviour
Attitude 12.40 0.20* 0.91 0.01 -2.42 -0.06 3.85 0.12
Subjective norm -11.06 -0.15* 0.01 0.00 1.02 0.02 -5.76 -0.16*
Perceived behaviour control -28.44 -0.07 59.38 0.14* 11.60 0.04 10.72 0.06
Perceived social norms 19.52 0.05 52.97 0.12* -10.87 -0.04 -6.04 -0.03
Intention -3.04 -0.02 1.16 0.01 7.21 0.06 -0.79 -0.01
Education 0.08 0.00 13.05 0.07 14.63 0.14* 11.02 0.13*
Household income -0.98 -0.01 -2.97 -0.02 -1.96 -0.02 0.54 0.01
Marital status -37.43 -0.09 -56.32 -0.12 17.87 0.06 -1.79 -0.01
Children -4.73 -0.03 11.99 0.05 0.63 0.01 -8.33 -0.09
Baseline eating behaviour† 75.27 0.31** 79.65 0.32** 31.79 0.25*** 30.68 0.31***
R2/R2 adjusted, % 13.2/9.3 23.1/20.1 10.7/6.6 20.2/17.1
Marital status is coded 1 = married, 2 = single/divorced/widow, while all other variables are coded in ascending order. † The corresponding eating behaviour in 1991: fruit and vegetable score in 1991 for fruit and vegetable intake in 1999 (grams per day), whole grain score in 1991 for whole grain intake in 1999 (grams per day), *p < 0.05, **p < 0.01, ***p < 0.001.
Table 5 Baseline (age 25) predictors of total fat and added sugar intake at follow up (age 33 years). Multiple linear regression analyses; unstandardized (B) and standardized regression coefficients (β). The Oslo Youth Study 1991 – 1999.
Dietary habits at follow-up Predictors Total fat intake Added sugar intake
Model 1 Men Women Men Women
Psychosocial factors at baseline B β B β B β B β
Attitude 0.25 0.13 0.06 0.03 0.25 0.10 0.03 0.02
Subjective norm 0.20 0.09 0.05 0.02 0.23 0.08 0.31 0.13*
Perceived behaviour control 0.16 0.01 -3.31 -0.29*** -1.29 -0.08 0.72 0.06
Perceived social norms -0.34 -0.03 -0.15 -0.01 -2.20 -0.13* -1.83 -0.14*
Intention -1.24 -0.21** 0.14 0.02 -0.78 -0.10 -0.16 -0.03
R2/R2 adjusted, % 4.1/2.0 8.4/6.6 4.3/2.2 3.3/1.5
Model 2
Model 1 + Demographic factors at follow-up
Attitude 0.27 0.14 0.04 0.02 0.30 0.12 0.02 0.01
Subjective norm 0.18 0.08 0.03 0.02 0.20 0.07 0.28 0.12
Perceived behaviour control 0.36 0.03 -3.06 -0.27*** -1.31 -0.08 0.98 0.08
Perceived social norms -0.20 -0.02 -0.37 -0.03 -2.53 -0.15* -1.97 -0.15*
Intention -1.23 -0.21** 0.05 0.01 -0.83 -0.11 -0.18 -0.03
Education -0.43 -0.09 -0.27 -0.05 0.28 0.04 -0.41 -0.07
Household income -0.16 -0.04 -0.42 -0.12 -0.77 -0.15* -0.27 -0.07
Marital status -1.97 -0.15 -0.76 -0.06 -0.07 -0.00 -0.63 -0.04
Children -0.01 0.00 0.41 0.04 0.86 0.11 0.08 0.01
R2/R2 adjusted, % 6.7/2.9 10.2/7.1 7.0/3.2 4.6/1.2
Model 3
Model 2 + baseline eating behaviour
Attitude 0.23 0.12 0.05 0.03 0.13 0.05 -0.02 -0.01
Subjective norm 0.17 0.08 0.03 0.01 0.07 0.02 0.16 0.07
Perceived behaviour control 0.41 0.03 -2.91 -0.25*** -1.45 -0.09 1.24 0.10
Perceived social norms -0.31 -0.03 -0.29 -0.02 -2.30 -0.14* -1.58 -0.12
Intention -1.16 -0.20* 0.09 0.02 -0.23 -0.03 0.15 0.02
Education -0.43 -0.09 -0.27 -0.05 0.74 0.12 -0.30 -0.05
Household income -0.19 0.05 -0.37 -0.10 -0.89 -0.17* -0.26 -0.07
Marital status -2.16 -0.17* -0.64 -0.05 -0.65 -0.04 -0.90 -0.06
Children -0.06 -0.01 0.22 0.04 0.42 0.06 0.0 0.00
Baseline eating behaviour† 0.92 0.12 0.56 0.07 3.65 0.35*** 2.34 0.26***
R2/R2 adjusted, % 8.0/3.9 10.7/7.1 17.6/13.9 10.6/7.1
Marital status is coded 1 = married, 2 = single/divorced/widow, while all other variables are coded in ascending order. † The corresponding eating behaviour in 1991: fat score in 1991 for total fat intake in 1999 (per cent of energy from total fat), sugar score in 1991 for added sugar intake in 1999 (per cent of energy from added sugar), *p < 0.05, **p < 0.01, ***p < 0.001.
Discussion
Results of this study are an important addition to the literature on psychosocial predictors of eating behaviour. By employing a longitudinal design and adjusting for socio-demographic confounders, as well as previous eating behaviour, we found that attitudes, subjective norms, perceived behavioural control, perceived social norms and intention to eat healthier food the next four weeks emerged as significant predictors of one or several eating behaviours eight years later. Of the socio-demographic factors, only education was positively associated with healthy eating for both sexes, while a higher income was associated with a low sugar intake and to be single or divorced was associated with a lower fat intake among men. Overall, the factors examined accounted for 4% to 20 % of the variation in follow-up eating behaviour.
Fruit and vegetable intake
For men, neither subjective norms nor attitude was significantly correlated with fruit and vegetable intake in bivariate analyses. However, in multivariate analyses, these factors appeared to be significantly associated with the intake of fruits and vegetables at follow-up. Because no significant associations existed between psychosocial factors and fruit and vegetable intake among men in bivariate analyses, multivariate associations might be artefacts, but they could also be a suppression phenomenon, see under Internal correlations below. Perceived behavioural control and perceived social norms measured at baseline remained predictive of women's intake of fruits and vegetables. In the 1991 survey, perceived social norms were found to be predictive of healthy eating [27]. In 1999, subjective norms were negatively associated with intake of fruits and vegetables because of the way the question was asked: "Do you believe that your parents, etc. think that you should eat healthier food the next four weeks?" Thus, those already having a healthy diet most likely did not believe their significant others expected them to eat healthier food. Conner and colleagues [8] investigated the longitudinal relationship between the constructs of TPB assessed at two time points six months apart (time 1 (T1) and time 2 (T2)) and eating behaviour assessed six years after T2. They found that intentions and the interaction between intentions and intention stability (defined as stability in intention between T1 and T2 six years prior to the last follow-up in their study) and between perceived past behaviour and intention stability were predictive of fruit and vegetable intake six years later. In our study, the constructs of TPB were measured only at age 25; hence, we can not address the stability of the psychosocial constructs. In an earlier publication by Conner and colleagues [28], they concluded that both stability in intentions and perceived behavioural control were important for future dietary behaviour. Our study adds to previous research in that perceived behavioural control, as well as perceived social norms among women measured once several years prior to dietary behaviour assessment, appear to be predictive of dietary behaviour.
Social class and family situation are factors shown to be predictive of dietary habits [15,29-31]. However, none of the socio-demographic factors in our study emerged as predictors of fruit and vegetable intake when we adjusted for past behaviour. This lack of an association may indicate that the association is mediated through other variables, such as dietary habits and psychosocial factors.
Whole grain intake
Education and previous behaviour remained significant predictors of whole grain intake among men and women in multivariate analyses, while the subjective norms construct was significant only for women. This is consistent with previous findings showing that a higher social class is associated with higher rates of consumption of whole grain foods [14,32]. As for fruit and vegetable intake, the subjective norms construct was negatively associated with whole grain intake. Patterson and colleagues [10] did not find that perceived norms, which resembled subjective norms in our study, explained fibre intake three years later.
Fat intake
Intention to eat healthier food measured at baseline was significantly and negatively associated with men's fat intake at follow up, even after adjusting for past behaviour. Perceived behavioural control remained a significant predictor of fat intake among women in multiple regression analyses. This is in contrast to the findings of Conner and colleagues, who found that intentions and the interaction between intentions and intention stability were the only significant predictors of fat intake six years later [8]. While Conner and colleagues' study included both genders, women constituted 83% of their sample. Among men in our study, marital status appeared to be a significant predictor of fat intake; to be single/divorced was associated with a lower fat intake when adjusted for psychosocial factors and past behaviour. This is in contrast to a Finnish study finding that married men and women had diets more in line with the dietary guidelines than not married men and women [15]. We have no explanation for our finding, and it may also be spurious.
Sugar intake
In the final model, perceived social norms and household income were predictive of men's sugar intake. In a previous report from the Oslo Youth Study, perceived social norms, represented by partners, were also predictive of healthy eating when adjusted for education [27]. For women, perceived social norms were predictive of sugar intake; however, when we included past behaviour in the model, past behaviour was the only factor that was predictive of women's sugar intake. A cross-sectional study by Grogan and colleagues examined gender differences in attitudes and behaviour using the Theory of Reasoned Action regarding eating sweet snacks [33]. The authors found that perceived social pressure and attitudes toward sweet snacks were associated with women's intentions to eat sweet snacks, while only attitudes were associated with men's intentions. Both men's and women's intentions were associated with reported intake of sweet snacks. However, the perceived social pressure construct in that study resembled the subjective norms construct in our study, and in that respect, our findings were similar. However, in our study, subjective norms lost its significance in predicting sugar intake when we included demographic factors and past intake of food high in sugar in the models.
Past eating behaviour
Scores representing past behaviour were predictive of intake of all dependent dietary measures for both sexes, with the exception of fat. Few studies have investigated the stability of nutrient intakes [34] and eating habits [35] during adult years. A study by Mulder and colleagues [35] on the stability of lifestyle behaviour over four years among adult men 30 – 39 years of age at baseline found the correlation coefficient between the dietary scores (including meal pattern, sweet and salty snacks, fruit and attitudes toward eating fat and fibre) at time one and time two to be 0.57. Our findings indicate that past habits are important predictors of current habits even when adjusting for socio-demographic and psychosocial factors and when taking into account that a different measure to assess food intake was used at baseline of the study. Conner and Armitage [36] have proposed that past behaviour may predict future behaviour as a moderator of the relationship between the TPB variables, as a source of information and as a mediator of TPB variables. They argue that there are good reasons to incorporate frequency of past behaviour as predictors of current behaviour in the TPB alongside intentions and perceived behavioural control. The results of our study support the view that past behaviour has an independent predictive value of fruit and vegetable intake, whole grain intake and sugar intake when taking the TPB constructs into account. Previous studies have shown that past behaviours were predictive of future behaviours independent of intentions, attitudes, norms and perceived behavioural control (PBC) [37]. Conner and Armitage [36] reported that after accounting for PBC and intentions, past behaviour, on average, could explain 13% (3% to 28%) of the observed variance in behaviour. In our study, an additional 6.2% on average (ranging from 0.5% to 10.6%) of variance in behaviour was explained by past behaviour beyond what was explained by the TPB variables and demographic factors. Given that the TPB is not often used to investigate dietary behaviours, comparing the explained variance between studies is difficult. The total explained variance in dietary habits in our study is, however, comparable to findings reported by Conner and colleagues [8]. The low explained variances found in our study and in other studies investigating psychosocial and demographic factors' prediction of dietary habits point to other variables having impact on dietary habits. Such factors have not been examined in this study, but physical environment [38,39], as well as taste, cost and convenience have been proposed as important to dietary behaviours [40,41].
Gender differences
Dietary differences between men and women have previously been demonstrated, with women generally reporting healthier eating habits than those reported by men [42]. Our results at baseline agree with this finding, as women had higher scores on fruit and vegetable intake and lower scores on fat and sugar intake compared to scores among men. However, at follow-up we observed no statistically significant differences between men and women's dietary intakes. In a representative and random sample of Norwegians 16 to 79 years of age in 1993 that applied the same method as the 1999 follow up of the Oslo Youth Study, researchers reported that women had higher intakes of vegetables and of fruit and berries compared to men, while men had higher intakes of cereals compared to women [43]. Also, fat and sugar intake differed between men and women in the previous study, while in our study the differences were not statistically significant. Even though most findings from our study showed the same patterns as in previous studies, differences between men and women did not reach statistical significance. This may be due to differing age groups and places of residence, as these factors also influence dietary intakes assessed with this method [44].
Gender differences in the TPB constructs regarding healthier eating have not been reported previously, but a study applying the Theory of Reasoned Action found women were under more social pressure not to eat sweet snacks than were men [33]. Barker and colleagues demonstrated that fat-phobic and fibre-philic attitudes were more prevalent among women, and that fat-phobic attitudes were inversely related to fat intake among women, but not among men [45]. A Norwegian study showed that, compared to men, women were more prone to consider foods that were in accordance with dietary recommendations as healthy, and less prone to consider fat- and protein-rich foods as healthy [46]. However, the gender difference disappeared when including "trust in experts" in the model, indicating that women's higher trust in experts might be one reason they ranked fish, fruits, vegetables and potatoes as healthy foods. This is similar to Grogan and colleagues' findings that women felt more pressure from health experts than did men to avoid eating sweet snacks [33]. Our results support this finding by indicating that men and women might use different psychosocial bases to carry out certain behaviours, but as we can see no certain patterns regarding gender, we can not conclude about how TPB constructs predict men's and women's eating behaviour differently.
Limitations
The relatively high attrition rate in this study is comparable to attrition rates found in previous longitudinal studies with similar follow-up periods [47,48]. There were, however, no baseline differences between participants and drop-outs, and we do not think that the attrition seen in this study is a threat to the validity of the observed prospective relationships.
The psychosocial factors in this study were constructed around healthier eating defined as "food low in fat, sugar and salt." Being aware that fruit, vegetables and whole meal bread was not included in the definition of a healthy diet in the 1991 survey is important. In 1991, the dietary focus in Norwegian society, as well as in the Oslo Youth Study, was on reducing the intake of fat and salt. The dietary focus in 1991 influenced the way the TPB questions were phrased and this might have influenced the observed associations between psychosocial factors and dietary habits. Baranowski and colleagues [1] claim that the influence on dietary habits varies by foods, and that the predictive value of TPB appears to be higher in predicting intake of a single food item or narrow categories of foods. Also the variability in measurement may play a role in the prediction of, for instance, fat eating patterns [1], which can be measured by means of total fat intake in grams per day, as per cent of total energy from fat or as foods high in fat. All this will contribute to differing prediction of different behaviours by psychosocial factors.
The time period for which intentions and attitudes toward healthier eating was applied at baseline in this study was "the next four weeks." It is therefore remarkable that the constructs of TPB contributed to the prediction of eating habits assessed eight years later. Results might point to the stability of the underlying psychosocial constructs. These findings are in agreement with previous studies investigating the stability of TPB constructs over time and as predictors of food choice over time [7,8].
The methodology of assessing diet in 1991 and 1999 differed. The dietary method used in 1999 made it possible to compute total intake of energy, macro- and micronutrients, while the questionnaire used in 1991 only enabled assessment of intake frequencies. Prediction of single nutrients in 1999 by means of previous intake was difficult as we did not have measures of the same nutrient eight years earlier. However, the single food items measured in 1991 are good sources for the particular 1999 nutrients [49]. Despite the differing methods used to assess dietary intakes, past behaviour was predictive of current behaviour for all items except fat intake, indicating a high degree of stability in dietary habits. Dietary scores composed of intake of specific food items and nutrients are shown to be valid for evaluating diet quality among adults [50,51].
Internal Correlations
In multivariate analyses of prediction of fruit and vegetable for men, the high correlations between subjective norms and attitudes might explain why variables not significant in bivariate analyses became significantly associated with the diet under investigation in multivariate analyses. Subjective norms might act as suppressors on attitudes, and vice versa, in bivariate analyses between each of these constructs and the intake of fruit and vegetables (negative confounding). However, in the multivariate analyses, these constructs will be mutually adjusted and the association between each of them and fruit and vegetable intake will become significant. However, the observed associations between attitude, subjective norms and fruit and vegetable intake among men in the multivariate analyses might also be an artefact. The other modest correlations between independent variables in this study are not regarded as a threat to the validity of the results.
Conclusion
Despite the psychosocial predictor variables investigated in this study being operationalized in terms of predicting healthy eating four weeks later, attitudes, subjective norms, perceived behavioural control, intentions and perceived social norms all appeared predictive of one or more specific eating behaviours reported eight years later. This was the case even after adjusting for demographic factors and past corresponding eating behaviour. Results point to the influence of psychosocial factors on future eating behaviours among adults and the potential for interventions targeting such factors on future behaviours.
Future research should focus on further development of appropriate assessment tools for psychosocial constructs, whether such constructs are stable over time and applying parallel measures of dietary intakes over time among representative samples of adults.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
EK was responsible for the formulation of the research question, the data analyses and writing the paper. NL contributed to formulate the research question and assisted with data analyses and writing the paper. GST and KIK were responsible for the design of the overall study, data collection and writing the paper. KIK contributed to formulating the specific research question and supervised data analyses. All authors have read and approved the final manuscript.
==== Refs
Baranowski T Cullen KW Baranowski J Psychosocial correlates of dietary intake: advancing dietary intervention Annu Rev Nutr 1999 19 17 40 10448515 10.1146/annurev.nutr.19.1.17
Satia JA Kristal AR Patterson RE Neuhouser ML Trudeau E Psychosocial factors and dietary habits associated with vegetable consumption Nutrition 2002 18 247 254 11882398 10.1016/S0899-9007(01)00758-4
Anderson AS Cox DN McKellar S Reynolds J Lean ME Mela DJ Take Five, a nutrition education intervention to increase fruit and vegetable intakes: impact on attitudes towards dietary change Br J Nutr 1998 80 133 140 9828754
Armitage CJ Evidence that implementation intentions reduce dietary fat intake: a randomized trial Health Psychol 2004 23 319 323 15099174 10.1037/0278-6133.23.3.319
Kristal AR Glanz K Tilley BC Li S Mediating factors in dietary change: understanding the impact of a worksite nutrition intervention Health Educ Behav 2000 27 112 125 10709796
Satia JA Kristal AR Curry S Trudeau E Motivations for healthful dietary change Public Health Nutr 2001 4 953 959 11784408
Armitage CJ Conner M The theory of planned behaviour: Assessment of predictive validity and 'perceived control' Br J Soc Psychol 1999 38 35 54 10.1348/014466699164022
Conner M Norman P Bell R The theory of planned behavior and healthy eating Health Psychol 2002 21 194 201 11950110 10.1037//0278-6133.21.2.194
Kristal AR Hedderson MM Patterson RE Neuhouser M Neuhauser ML Predictors of self-initiated, healthful dietary change J Am Diet Assoc 2001 101 762 766 11478472 10.1016/S0002-8223(01)00191-2
Patterson RE Kristal AR White E Do beliefs, knowledge, and perceived norms about diet and cancer predict dietary change? Am J Public Health 1996 86 1394 1400 8876507
Giskes K Turrell G Patterson C Newman B Socio-economic differences in fruit and vegetable consumption among Australian adolescents and adults Public Health Nutr 2002 5 663 669 12372160 10.1079/PHN2002339
Kant AK Dietary patterns and health outcomes J Am Diet Assoc 2004 104 615 635 15054348 10.1016/j.jada.2004.01.010
Laaksonen M Prättälä R Helasoja V Uutela A Lahelma E Income and health behaviours. Evidence from monitoring surveys among Finnish adults J Epidemiol Community Health 2003 57 711 717 12933778 10.1136/jech.57.9.711
Lang R Thane CW Bolton-Smith C Jebb SA Consumption of whole-grain foods by British adults: findings from further analysis of two national dietary surveys Public Health Nutr 2003 6 479 484 12943564 10.1079/PHN2002453
Roos E Lahelma E Virtanen M Prättälä R Pietinen P Gender, socioeconomic status and family status as determinants of food behaviour Soc Sci Med 1998 46 1519 1529 9672392 10.1016/S0277-9536(98)00032-X
Holmboe-Ottesen G Wandel M Mosdøl A Social inequality and diet. [Nor] Tidsskr Nor Laegeforen 2004 124 1526 1528 15195158
Johansson L Thelle DS Solvoll K Bjørneboe GE Drevon CA Healthy dietary habits in relation to social determinants and lifestyle factors Br J Nutr 1999 81 211 220 10434847
Krebs-Smith SM Heimendinger J Patterson BH Subar AF Kessler R Pivonka E Psychosocial factors associated with fruit and vegetable consumption Am J Health Promot 1995 10 98 104 10160052
Kvaavik E Meyer HE Selmer RM Egeland G Tverdal A [Food habits in Hedmark related to gender, education and marital status] [Norwegian] Tidsskr Nor Laegeforen 1999 119 3406 3409 10553335
Trudeau E Kristal AR Li S Patterson RE Demographic and psychosocial predictors of fruit and vegetable intakes differ: implications for dietary interventions J Am Diet Assoc 1998 98 1412 1417 9850109 10.1016/S0002-8223(98)00319-8
Van Duyn MA Kristal AR Dodd K Campbell MK Subar AF Stables G Association of awareness, intrapersonal and interpersonal factors, and stage of dietary change with fruit and vegetable consumption: a national survey Am J Health Promot 2001 16 69 78 11727591
Øygard L Rise J Predicting the intention to eat healthier food among young adults Health Education Research 1996 11 453 461
Tell GS Epidemiology and prevention of chronic disease risk factors in adolescents: The Oslo Youth Study PhD-thesis 1987 University of Oslo, Department of preventive medicine
Klepp K-I Øygard L Tell GS Vellar OD Twelve year follow-up of a school-based health education programme. The Oslo Youth Study Eur J Public Health 1994 4 195 200
Andersen LF Solvoll K Johansson LR Salminen I Aro A Drevon CA Evaluation of a food frequency questionnaire with weighed records, fatty acids, and alpha-tocopherol in adipose tissue and serum Am J Epidemiol 1999 150 75 87 10400557
National Nutrition Council Recommendations increased intake of fruits and vegetables Report 1998 [Nor] Oslo
Øygard L Klepp KI Influences of social groups on eating patterns: a study among young adults J Behav Med 1996 19 1 15 8932658 10.1007/BF01858171
Conner M Sheeran P Norman P Armitage CJ Temporal stability as a moderator of relationships in the Theory of Planned Behaviour Br J Soc Psychol 2000 39 469 493 11190681 10.1348/014466600164598
Billson H Pryer JA Nichols R Variation in fruit and vegetable consumption among adults in Britain. An analysis from the dietary and nutritional survey of British adults Eur J Clin Nutr 1999 53 946 952 10602352 10.1038/sj.ejcn.1600877
Dowler E Inequalities in diet and physical activity in Europe Public Health Nutr 2001 4 701 709 11683565
Roos G Johansson L Kasmel A Klumbiene J Prättälä R Disparities in vegetable and fruit consumption: European cases from the north to the south Public Health Nutr 2001 4 35 43 11255494
Mishra G Ball K Arbuckle J Crawford D Dietary patterns of Australian adults and their association with socioeconomic status: results from the 1995 National Nutrition Survey Eur J Clin Nutr 2002 56 687 693 12080411 10.1038/sj.ejcn.1601391
Grogan SC Bell R Conner M Eating sweet snacks: gender differences in attitudes and behaviour Appetite 1997 28 19 31 9134092 10.1006/appe.1996.0067
Dunn JE Liu K Greenland P Hilner JE Jacobs DR Jr Seven-year tracking of dietary factors in young adults: the CARDIA study Am J Prev Med 2000 18 38 45 10808981 10.1016/S0749-3797(99)00114-2
Mulder M Ranchor AV Sanderman R Bouma J van den Heuvel WJ The stability of lifestyle behaviour Int J Epidemiol 1998 27 199 207 9602399 10.1093/ije/27.2.199
Conner M Armitage CJ Extending the Theory of Planned Behavior: A Review and Avenues for Further Research Journal of Applied Social Psychology 1998 28 1429 1464
Ouellette JA Wood W Habit and Intention in Everyday Life: The Multiple Processes by Which Past Behaviour Predicts Future Behavior Psychological Bulletin 1998 124 54 74 10.1037//0033-2909.124.1.54
French SA Story M Jeffery RW Environmental influences on eating and physical activity Annu Rev Public Health 2001 22 309 335 11274524 10.1146/annurev.publhealth.22.1.309
Swinburn B Egger G The runaway weight gain train: too many accelerators, not enough brakes BMJ 2004 329 736 739 15388619 10.1136/bmj.329.7468.736
Drewnowski A The role of energy density Lipids 2003 38 109 115 12733741
Drewnowski A Obesity and the food environment: dietary energy density and diet costs Am J Prev Med 2004 27 154 162 15450626 10.1016/j.amepre.2004.06.011
Johansson L Andersen LF Who eats 5 a day?: intake of fruits and vegetables among Norwegians in relation to gender and lifestyle J Am Diet Assoc 1998 98 689 691 9627628 10.1016/S0002-8223(98)00156-4
Johansson L Solvoll K Bjørneboe G-EA Drevon CA Dietary habits among Norwegian men and women Scandinavian Journal of Nutrition 1997 41 63 70
Johansson L Solvoll K Norkost 1997 Report 1997 [Nor] Directorate for Health and Social Affairs, Oslo
Barker ME Thompson KA McClean SI Attitudinal dimensions of food choice and nutrient intake Br J Nutr 1995 74 649 659 8541271
Wandel M Fagerli RA Norwegians' opinions of a healthy diet Journal of Nutrition Education 1999 31 339 346
Barnekow-Bergkvist M Hedberg G Janlert U Jansson E Adolescent determinants of cardiovascular risk factors in adult men and women Scand J Public Health 2001 29 208 217 11680773 10.1080/140349401316983635
Myers L Coughlin SS Webber LS Srinivasan SR Berenson GS Prediction of adult cardiovascular multifactorial risk status from childhood risk factor levels. The Bogalusa Heart Study Am J Epidemiol 1995 142 918 924 7572972
Johansson L Development in dietary intake in Norway Report 2002 [Nor] National council for nutrition and physical activity. Oslo
Rafferty AP Anderson JV McGee HB Miller CE A healthy diet indicator: quantifying compliance with the dietary guidelines using the BRFSS Prev Med 2002 35 9 15 12079436 10.1006/pmed.2002.1056
Weinstein SJ Vogt TM Gerrior SA Healthy Eating Index scores are associated with blood nutrient concentrations in the third National Health And Nutrition Examination Survey J Am Diet Assoc 2004 104 576 584 15054343 10.1016/j.jada.2004.01.005
|
16076386
|
PMC1208934
|
CC BY
|
2021-01-04 16:38:04
|
no
|
Int J Behav Nutr Phys Act. 2005 Aug 2; 2:9
|
utf-8
|
Int J Behav Nutr Phys Act
| 2,005 |
10.1186/1479-5868-2-9
|
oa_comm
|
==== Front
Immun AgeingImmunity & ageing : I & A1742-4933BioMed Central London 1742-4933-2-111602273410.1186/1742-4933-2-11ResearchHeparan sulfate proteoglycan induces the production of NO and TNF-α by murine microglia Bussini Simona [email protected] Lucia [email protected] Elio [email protected] Emilio [email protected] Giancarlo [email protected] Marco [email protected] Nereo [email protected] Pierluigi [email protected] Department of Neurological Sciences, Centre for Excellence on Neurodegenerative Diseases and "Dino Ferrari" Center, University of Milan, Fondazione IRCCS "Ospedale Maggiore Policlinico, Mangiagalli e Regina Elena", Via F. Sforza 35, 20122 Milan, Italy2 Dept. Preclinical Sciences, University of Milano, 20157 - Milano and E.Medea Scientific Institute 23842 - Bosisio Pasini, Italy2005 16 7 2005 2 11 11 22 4 2005 16 7 2005 Copyright © 2005 Bussini et al; licensee BioMed Central Ltd.2005Bussini et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A common feature of Alzheimer's disease (AD) pathology is the abundance of activated microglia in neuritic plaques containing amyloid-beta protein (Aβ) and associated molecules including heparan sulfate proteoglycan (HSPG). Besides the role as pathological chaperone favouring amyloidogenesis, little is known about whether or not HSPG can induce microglial activation. Cultures of primary murine microglia were used to assess the effect of HSPG on production of proinflammatory molecules that are known to be present in neuritic plaques of AD.
Results
HSPG stimulated up-regulation of tumor necrosis factor-alpha (TNF-α), production of inducible nitric oxide synthase (iNOS) mRNA and accumulation of TNF-α protein and nitrite (NO2-) in a time- and concentration-dependent manner. The effects of HSPG were primarily due to the property of the protein core as indicated by the lack of microglial accumulation of TNF-α and NO2- in response to denaturated HSPG or heparan sulfate GAG chains (HS).
Conclusion
These data demonstrate that HSPG may contribute to chronic microglial activation and neurodegeneration seen in neuritic plaques of AD.
==== Body
Introduction
Senile plaques in the Alzheimer disease (AD) brain are characterized by the presence of an amyloid core consisting of fibrillar Aβ, surrounded by a wreath of dystrophic neurites and activated microglial cells. Reactive microglia release a variety of potentially neurotoxic compounds, including cytokines and free radicals. Many research groups have provided evidence that deposition of aggregated Aβ is centrally involved in the chronic inflammatory process occurring in senile plaques of AD brain [1,2].
Aβ is a 39- to 42-amino-acid peptide that arises from proteolytic processing of the amyloid precursor protein (APP) [3,4]. The Aβ peptide that is found in the senile plaques and cerebrovascular deposits exists as a multimeric aggregate with a fibrillar appearance [5]. Several other molecules have also been shown to be associated with Aβ deposits, and in vitro studies of fibrillogenesis suggest they may be important in the aggregation and persistence of the Aβ fibrils in vivo. These include apolipoprotein E, laminin, acetylcholinesterase, α1-antichymotrypsin and heparan sulfate proteoglycan (HSPG) [6-11].
HSPG is a multifunctional macromolecule characterized by a core polypeptide to which glycosaminoglycans (GAGs) are covalently attached. There are at least four different classes of HSPG present in AD, which are either associated with the cell membrane or with the extracellular matrix [12]. HSPG has been consistently associated with both diffuse and neuritic plaques [13,14]. Its early presence in AD pathological alterations as well as its immunohistochemical colocalization with all varieties of Aβ plaques, irrespective of their stage of maturation, have suggested that HSPG could play an active role in plaque formation. In this regard it has been proposed that HSPG facilitates Aβ deposition and/or promotes Aβ persistence by inhibiting clearance mechanisms, thus augmenting the formation of Aβ deposits in AD [15]. Consistent with this hypothesis, in vitro studies have shown that HSPG can bind with high affinity to Aβ as well as to APP and it protects Aβ from protease degradation [16-19].
Besides its function in amyloidogenic pathways, HSPG might contribute to AD pathogenesis also through activation of microglial cells. This possibility has never been investigated. To study this we have assessed in "in vitro" cultures of mouse microglial cells, two known markers of their activation, i.e. production of the proinflammatory cytokine tumor necrosis factor-α (TNF-α) and expression of the mRNA for the inducible nitric oxide (NO) synthase (iNOS). This enzyme is generated in microglia in response to a variety of pro-inflammatory cytokines and bacterial products, such as lipopolysaccharide (LPS). In addition, we have measured in the culture media supernatants the accumulation of NO2-, a good proxy for generation of NO by iNOS. Our results show that HSPG might contribute to neurodegeneration in neuritic plaques of AD also through activation of microglial cells and the ensuing increased inflammatory response.
Materials and methods
Reagents
Heparan sulfate proteoglycan (HSPG), heparan sulfate (HS), lipopolysaccharide (LPS, from Escherichia coli 026.B6) were purchased from Sigma (St Louis, MO, USA) and dissolved in clinical pyrogen-free H2O. Levels of endotoxin in HSPG and HS stocks were measured by E-TOXATE (Limulus Amebocyte Lysate, LAL) kit purchased from Sigma (St Louis, MO, USA).
Preparation of Microglial Cultures
Mice were obtained from Charles River Laboratories, Inc. (Wilmington, MA, USA) and were used according to institutional guidelines that are in compliance with national (D.I. no. 116, G.U. suppl. 40, Feb. 18, 1992, Circolare No.8, G.U., 14 Luglio 1994) and international law and policies (EEC Council Directive 86/609, OJ L358, 1 Dec. 12, 1987; Guide for the Care and Use of Laboratory Animals, U.S. National Research Council, 1996). Primary murine microglial cultures were prepared as previously described [20]. Briefly, cerebral cortical cells from day 1-old mice were dissociated with 0.25% trypsin and 0.1% DNAse (Sigma) and plated in 75 cm2 culture flasks (Corning, Acton, MA) in Dulbecco's Modified Eagle's Medium (DMEM) (Invitrogen Corporation Grand Island, NY, USA) containing 10% heat-inactivated FBS (Invitrogen Corporation Grand Island, NY, USA) and 100 μg/ml gentamicin (Invitrogen Corporation Grand Island, NY, USA). Dissociated glial cultures were maintained at 37°C with 5% CO2 and medium was replenished 4 days after plating. On day twelve of culture, flasks were shaken for 2 hours. The culture media supernatants (containing predominantly microglia and oligodendrocytes) were then collected and the cells were plated in either 96-well tissue plates (Nunc, Roskild, DK) at a concentration of 4 × 104 cells per 100 μl per well, or in 48-well tissue plates at a concentration of 25 × 104 per 500 μl per well, and maintained at 37°C with 5% CO2 for 1 hour. Loosely adherent oligodendrocytes were then removed from the cultures by gentle shaking of the culture plates by hand. After pouring off the culture media supernatants (containing oligodendrocytes), the adherent microglia were maintained at 37°C with 5% CO2 for subsequent treatment. The purity of microglial cultures was routinely assessed by staining with the F4/80-antibody (Serotec, Oxford, UK), which recognizes a glycoprotein expressed predominantly in microglia/macrophages cells [21], and found to be in the range of 96–98% in all cell preparations.
Exposure of microglia to HSPG
Microglial cells were incubated with HSPG at a concentration of 5, 15, 30, 40 μg/ml for 2, 4, 12, 24 and 48 h. After the indicated times, culture supernatants were harvested, and frozen at -70°C until assayed for levels of secreted TNF-α and nitrite (NO2-). After 4 h total RNA for each conditions was isolated from cells plated in 48-well tissue culture plates, using the TRIzol reagent protocol (Invitrogen; Carlsbad, CA, USA). In some experiments cells were exposed to 15 μg/ml of HS or HSPG denaturated at 90°C for 10 min. After treatment at indicated times, culture media supernatants were assayed for NO2- accumulation and TNF-α release as described below.
TNF-α assay
Antigenic mouse TNF-α was detected by ELISA system from Biosurce (Camarillo, CA, USA), and based on the quantitative "sandwich" enzyme immunoassay technique. The sensitivity of the assay was 10 pg/ml.
Nitrite assay
Nitrite (NO2-) is a stable end-product used extensively as an indicator of NO production by cultured cells. In our experimental conditions, NO2- accumulation was assayed by the Griess reaction, according to the method previously described [22]. Briefly, culture media supernatants were mixed with equal amounts of Griess reagent (p-aminobenzene sulfonamide 1%, naphtylethylenediamide 0.1% in phosphoric acid 2.5%) in 96-well plates: samples were incubated at room temperature for 10 min, and subsequently absorbance was read at 540 nm using a microplate reader. NO2- concentrations were calculated in accordance with a sodium nitrite standard curve.
RT-PCR
One μg total RNA quantitated spectrophotometrically and isolated from each condition was reverse transcribed using oligo(dT)20-primers and Superscript II-Reverse Transcriptase according to the manifacturer's protocol (Invitrogen, Carlsbad, CA, USA). c-DNA equivalent to 20 ng of total RNA was subjected to subsequent PCR analysis in a total volume of 30 μl containing 25 pmol of primers specific for TNF-α, iNOS and glyceraldehyde phosphate dehydrogenase (GAPDH; used as an internal control) (Table I), 10 mM Tris-HCl pH 8.3 (at 25°C), 50 mM KCl, 10% DMSO, 1.25 mM MgCl2, 250 μM each of dATP, dCTP, dGTP, dTTP, and 1.5 units AmpliTaq DNA-polymerase (Roche, Branchburg, NJ, USA). PCR was performed at the following conditions: (1) 2 min at 93°C; (2) 30 sec at 93°C, 30 sec at 60°C (TNF-α) or at 68°C (iNOS and GAPDH), 45 sec at 72°C for 26 cycles (iNOS), 24 cycles (TNF-α) or 20 cycles (GAPDH); (3) 10 min. at 72°C. PCR products were analysed on 1.5% agarose gel containing 10 μg/ml ethidium bromide. Controls included RNA subjected to the RT-PCR procedure without addition of reverse transcriptase and PCR performed in the absence of c-DNA which always yelded negative results.
Table I Oligonucleotide Primers Used for cDNA amplification
Probe Cycles Orientation Tann Sequence Length of PCR Fragments (bp)
GAPDH 20 sense
antisense 68 5'TGAAGGTCGGTGTGAACGGATTTGGC3'
5'CATGTAGGCCATGAGGTCCACCAC3' 983
iNOS 26 sense
antisense 68 5'CCCTTCCGAAGTTTCTGGCAGCAGC3'
5'GGCTGTCAGAGCCTCGTGGCTTTGG3' 493
TNFα 24 sense
antisense 60 5'TTCTGTCTACTGAACTTCGGGTGATCGGTCC3'
5'GTATGAGATAGCAAATCGGCTGACGGTGTGGG3' 378
Statistical analysis
Data are expressed as means ± standard deviations (SD). Statistical evaluation was performed by repeated measures ANOVA (analysis of variance) followed by Dunnet's test for specific comparisons. Statistical significance was set at P < 0.05.
Results
1. HSPG triggers production of NO2- and TNF-α in cultured microglia
To test whether the interaction of HSPG with microglia could induce the production of proinflammatory and potentially cytotoxic mediators, we assayed the accumulation of NO2- as an indirect measure of NO production from mouse primary microglia stimulated with HSPG and, for comparison, with LPS. As shown in Fig. 1, microglia in resting conditions did not release detectable NO2- even after a 48-h incubation, whereas HSPG induced significant accumulation of NO2- in culture media supernatants in the range of that observed with LPS. The effect of HSPG on NO2- production was time and concentration-dependent, with maximal accumulation observed at 48 h (7.5. ± 0.5-fold increase over control at 30 μg/ml HSPG; P < 0.05, n = 9) (Fig. 1A) and production already significantly increased after a 24 h exposure to 15 μg/ml (5.2 ± 0.4-fold increase over control; P < 0.05, n = 9) (Fig. 1B). We also investigated whether or not HSPG was able to induce the production of TNF-α by microglia. Untreated cells constitutively produced very small amounts of TNF-α, whereas their stimulation with HSPG or LPS resulted in the release of significant levels of TNF-α, with maximal accumulation observed at 4 h (25 ± 1.8-fold increase over control; P < 0.05, n = 9), followed by a decrease at later times (Fig. 2A). Concentration-response studies demonstrated that the amount of TNF-α released into the culture media supernatants increased with increasing concentrations of HSPG (Fig. 2B). TNF-α release was already significantly increased after a 24 h exposure to 15 μg/ml HSPG (11.6 ± 0.85.-fold increase over control; P < 0.05, n = 9). Specificity of the effects of HSPG on microglial activation was confirmed by LAL test that excluded trace levels of endotoxin in our HSPG stocks.
Figure 1 Effect of HSPG on the accumulation of NO2- from murine microglia. In panel A, time course of NO2- production by murine microglia in response to HSPG. Microglial cells were cultured in 96-well plates and stimulated with 30 μg/ml HSPG for up to 48 h. In panel B, dose-dependent effect of HSPG on the accumulation of NO2- by murine microglia. Microglial cells were cultured in 96-well plates and stimulated for 24 h with increasing concentrations of HSPG, or 100 ng/ml LPS. Mean values ± SD of assays performed with culture media supernatants collected and pooled from triplicate wells for each condition are shown (n = 9). Both panels depict a representative experiment out of three performed with similar results. **p < 0.01.
Figure 2 Effect of HSPG on TNF-α release by murine microglia. In panel A, time course of TNF-α production by murine microglia in response to HSPG. Microglial cells were cultured in 96-well plates and stimulated with 30 μg/ml HSPG for up to 24 h. In panel B, dose-dependent effect of HSPG on the accumulation of TNF-α by murine microglia. Microglial cells were cultured in 96-well plates and stimulated for 24 h with increasing concentrations of HSPG or 100 ng/ml LPS. Mean values ± SD of assays performed with culture media supernatants collected and pooled from triplicate wells for each condition are shown (n = 9). Both panels depict a representative experiment out of three performed with similar results. **p < 0.01.
2. HSPG induces expression of iNOS and TNF-α mRNA in cultured microglia
To determine whether the production of NO2- and TNF-α triggered by HSPG reflected induction of iNOS and TNF-α mRNA, RT-PCR analysis was performed on microglia total RNA, using probes complementary to the mouse macrophage iNOS and TNF-α coding sequences. In resting conditions the mRNA expression for iNOS and TNF-α in microglia was absent. On the contrary, mRNA levels for iNOS and TNF-α were clearly induced after stimulation of the cells for 4 h with HSPG at the concentration of 15 μg/ml or 100 ng/ml LPS (Fig. 3).
Figure 3 Effect of HSPG on iNOS and TNF-α mRNA expression in murine microglia. One-day-old microglial cells were cultured in 48-well plates and then stimulated for 4 hours with 15 μg/ml HSPG or 100 ng/ml LPS. Total RNA was extracted and analysed by RT-PCR. The densitometric evaluation of the bands obtained from three independent experiments in duplicates are reported (n = 6); values are expressed as the relative level of iNOS/GAPDH and TNFα/GAPDH. *p < 0.05.
3. Microglial activation induced by HSPG is mediated primarily by the protein core
To examine whether the protein core or the GAGs of HSPG were involved in mediating microglial activation, the effects of HSPG were compared with those obtained using HSPG denaturated at 90°C for 10 min (HSPG-hd) or HS-GAG chains (HS). As shown in Fig. 4, exposure of microglia to 30 μg/ml HSPG-hd almost completely abolished the production of NO2- (1 ± 0.2 μM; P < 0.05, n = 9) and particularly of TNF-α (98 ± 10 pg/ml; n = 9). Stimulation with 30 μg/ml HS only slightly affected release of TNF-α (200 ± 18 pg/ml; P < 0.05, n = 9) but not accumulation of NO2- (0.8 ± 0.1 μM; n = 9). The lack of ability of HSPG-hd and HS to maintain NO2- accumulation and TNF-α release appeared to be mediated at the transcriptional level since HSPG-hd and HS failed to increase both iNOS and TNF-α mRNA transcription (Fig. 5).
Figure 4 Effect of heat denaturated HSPG and HS-GAG chains on NO2- and TNF-α release by murine microglia. Microglial cells were cultured in 96-well plates and stimulated with 15 μg/ml HSPG, 15 μg/ml heat denaturated HSPG (HSPG-hd) or 15 μg/ml HS-GAG chains (HS). After 24 h, culture media supernatants were assayed for NO2- accumulation (A) and TNF-α release (B). Mean values ± SD of assays performed with culture media supernatants collected and pooled from triplicate wells for each condition are shown (n = 9). Both panels depict a representative experiment out of three performed with similar results. *p < 0.05, **p < 0.01.
Figure 5 Effect of heat denaturated HSPG and HS-GAG chains on iNOS and TNF-α mRNA expression in murine microglia. Microglial cells were cultured in 48-well plates and then stimulated for 4 hours with 15 μg/ml HSPG, 15 μg/ml heat denaturated HSPG (HSPG-hd), 15 μg/ml HS-GAG chains (HS) or 100 ng/ml LPS. Total RNA was extracted and analysed by RT-PCR. The densitometric evaluation of the bands obtained from three independent experiments in duplicates are reported (n = 6); values are expressed as the relative level of iNOS/GAPDH and TNF-α/GAPDH. *p < 0.05.
Discussion
The identification in senile plaques of pathologic stimuli that can lead to microglial activation represents one of the important issues of immunologic research in AD. Previous studies have shown that microglia upon stimulation with Aβ can produce proinflammatory and cytotoxic mediators, and that these mediators play a role in the pathogenesis of AD [23-26]. The demonstration of the indirect neuronal injury prompted us to investigate if, in addition to Aβ, others molecules present in senile plaques could be involved in similar mechanisms of microglial activation. We have identified HSPG as a potential candidate in this process on the basis that this molecule has been found to be associated with Aβ peptide-containing deposits and suggested to act as pathological chaperone, increasing β-pleated structure within Aβ [27,28].
We demonstrate that HSPG is able to induce the release of NO2- and TNF-α by cultured primary murine microglia. By assessing iNOS and TNF-α mRNA expression with RT-PCR we have also shown that the release of NO and TNF-α in HSPG-stimulated microglial cells is due to the induction of iNOS and TNF-α gene expression. These findings suggest that, in addition to Aβ, also HSPG is able to activate microglia with production of proinflammatory molecules known to be present in the brain of AD patients.
The precise mechanism by which microglia mediate neuronal cell injury in AD is incompletely understood and several mediators have been proposed, among them NO and TNF-α [29-31]. Even though physiological levels of NO may influence synaptic efficacy by regulating neurotrasmitter release [32], excess NO may cause neuronal degeneration by combining with oxygen radicals such as superoxide anion to form the highly toxic peroxynitrite ion [33]. Similarly, TNF-α has been reported to be trophic to rat hippocampal neurons [34]. However, transgenic mice that overexpress TNF-α exibit severe inflammation and neurodegeneration [35]. Moreover, in vitro studies have shown that some Aβ-induced microglial activities, including neurotoxicity and chemokine production, are mediated through release of endogenous TNF-α [23,36,37]. That TNF-α is presumably involved in AD pathology is also supported by its elevated levels in the serum, CSF and cerebral cortex of AD patients [38,39]. In view of the information summarized above, it is conceivable that generation of NO and TNF-α from microglial cells is one means by which HSPG may enhance the inflammatory reaction in neuritic plaques, and thus pathogenesis of AD. However, it must be pointed out that astrocytes represent the main source of NO in the plaques whereas release of neurotoxic levels of NO by microglia has been shown in vitro [40,23,24,41]. Future studies will be needed in order to determine whether HSPG also activates astrocytes to produce NO and TNF-α.
The lack of inflammation and microglial activation described in diffuse plaques opens the question about the actual proinflammatory role played by HSPG at early stages of plaque evolution. Our results show that activation of microglial cells by HSPG is concentration-dependent, both in terms of NO2- accumulation and TNF-α release in the culture medium, and that 15 μg/ml of the compound is sufficient to trigger the activation process. Although HSPG has been immunohistologically localized to senile plaques its biochemical isolation from these structures, and thus its quantification, have yet to be performed [42]. Thus, we cannot establish at present whether the threshold for microglial activation we found can explain the discrepancies in immunohistochemical localization of HSPG in diffuse plaques lacking inflammation.
The cellular source of HSPG in mammalian brain is represented by microglia and astrocytes which have been shown by immunofluorescence and/or western blotting to express HSPG both in vitro and in vivo [43,44]. Nevertheless, the factors implicated in HSPG biosynthesis and deposition have been poorly investigated. Recently regulation of HSPG by injury and IL-1α has been demonstrated in astrocytes and microglia [45]. Therefore, the stimulation of microglial cells by HSPG, followed by increased production of proinflammatory cytokines could stimulate further HSPG formation in an autocrine, feed-forward manner.
The importance of HSPG deposition in AD pathogenesis is supported by the fact that HSPG binds Aβ, accelerates Aβ fibril formation and maintains Aβ fibril stability [46]. Most of these effects of HSPG have been shown to be due to its associated HSGAG side-chains, as also suggested by the lack of extracellular Aβ deposits in transgenic mice overexpressing HSPG protein core [47]. This contrasts with our results showing that the proinflammatory role of HSPG is primarily mediated by its protein core. However, in these transgenic mice HSPG was detected only inside the cells, i.e. without accumulation of the compound in the extracellular environment, which instead is a hallmark of AD neurodegeneration. Studies in these animals, therefore, do not allow drawing conclusions about the role of extracellular HSPG in AD.
In conclusion, the potential participation of HSPG in NO and TNF-α-mediated neuronal injury induced by microglia adds a novel biological role of this molecule in the pathogenesis of AD. These data indicate that HSPG plays an immunomodulatory role in the activation of microglia, in addition to that proposed in amyloidogenic pathways and demonstrate another mechanism by which immune responses may be triggered in AD brain. Therefore, a further understanding of the role of HSPG in the pathogenesis of AD would assist in the development of rational, targeted therapeutic strategies to combat this neurodegenerative disorder.
Acknowledgements
We are grateful to financial support from the MANAD project of the European Community and from "Ricerca Finalizzata Alzheimer 2000" of the Italian Ministery of Health.
==== Refs
Neuroinflammation Working Group Inflammation and Alzheimer's disease Neurobiol Aging 2000 21 383 421 10858586 10.1016/S0197-4580(00)00124-X
Meda L Baron P Scarlato G Glial activation in Alzheimers disease: the role of Aβ and its associated proteins Neurobiol Aging 2001 22 885 893 11754995 10.1016/S0197-4580(01)00307-4
Goldgaber D Lerman MI McBride OW Saffioti U Gajdusek DC Characterization and chromosomal localization of a cDNA encoding brain amyloid of Alzheimer's disease Science 1987 235 877 880 3810169
Kang J Lemaire HG Unterbeck A Salbaum JM Masters CL Grzeschik KH Multhaup G Beyreuther K Muller-Hill B The precursor of Alzheimer's disease amyloid A4 protein resembles a cell-surface receptor Nature 1987 325 733 736 2881207 10.1038/325733a0
Roher A Wolfe D Palutke M Kukuruga D Purification, ultrastructure, and chemical analysis of Alzheimer disease amyloid plaque core protein Proc Natl Acad Sci USA 1986 83 2662 2666 3458224
Wisniewski T Castano EM Golabek A Vogel T Frangione B Acceleration of Alzheimer's fibril formation by apolipoprotein E in vitro Am J Pathol 1994 145 1030 1035 7977635
Wood SJ Chan W Wetzel R Seeding of Aβ fibril formation is inhibited by all three isotypes of apolipoprotein E Biochemistry 1996 35 12623 12628 8823200 10.1021/bi961074j
Monji A Tashiro K Yoshida I Tashiro N Laminin inhibits Abeta42 fibril formation in vitro Brain Res 1998 788 187 190 9555007 10.1016/S0006-8993(97)01542-4
Alvarez A Bronfman F Perez CA Vicente M Garrido J Inestrosa NC Acetylcholinesterase, a senile plaque component, affects the fibrillogenesis of amyloid-β-peptide Neurosci Lett 1995 201 49 52 8830310 10.1016/0304-3940(94)12127-C
Eriksson S Janciauskiene S Lannfelt L α1-Antichymotrypsin regulates Alzheimer β-amyloid peptide fibril formation Proc Natl Acad Sci USA 1995 92 2313 2317 7892264
Castillo GM Ngo C Cummings J Wight TN Snow AD Perlecan binds to the β-amyloid proteins (Aβ) of Alzheimer's disease, accelerates Aβ fibril formation, and maintains Aβ fibril stability J Neurochem 1997 69 2452 2465 9375678
van Horssen J Wesseling P van der Heuvel L de Waal R Verbeek M Heparan sulphate proteoglycans in Alzheimer's disease and amyloid-related disorders Lancet Neurol 2003 2 482 492 12878436 10.1016/S1474-4422(03)00484-8
Snow AD Wight TN Proteoglycans in the pathogenesis of Alzheimer's disease Neurobiol Aging 1989 10 481 497 2682326 10.1016/0197-4580(89)90108-5
Snow AD Mar H Nochlin D Sekiguchi RT Kimata K Koike Y Wight TN Early accumulation of heparan sulfate in neurons and in the beta-amyloid protein containing lesions of Alzheimer's disease and Down's syndrome Am J Pathol 1990 137 1253 1270 2146882
Snow AD Sekiguchi RT Nochlin D Kalaria RN Kimata K Heparan sulfate proteoglycan in diffuse plaques of hippocampus but not in cerebellum of Alzheimer's disease brain Am J Pathol 1994 144 337 347 8311117
Buée L Ding W Delacourte A Fillit H Binding of secreted human neuroblastoma proteoglycans to the Alzheimer's amyloid A4 peptide Brain Res 1993 601 154 163 8431762 10.1016/0006-8993(93)91706-X
Snow AD Kinsella MG Parks E Sekiguchi RT Miller JD Kimata K Wight TN Differential binding of vascular cell-derived proteoglycans (perlecan, biglycan, decorin, and versican) to the beta-amyloid protein of Alzheimer's disease Arch Biochem Biophys 1995 320 84 95 7793988 10.1006/abbi.1995.1345
Narindrasorasak S Lowry DE Gonzalez-DeWhitt PA Poorman RA Greenberg BD Kisilevsky R High affinity interactions between the Alzheimer's beta-amyloid precursor proteins and the basement membrane form of heparan sulfate proteoglycan J Biol Chem 1991 266 12878 12883 1906461
Gumpta-Bansal R Frederickson CA Brunden KR Proteoglycan-mediated inhibition of Aβ proteolysis J Biol Chem 1995 270 18666 18671 7629198 10.1074/jbc.270.31.18666
Agresti C Aloisi F Levi G Heterotypic and homotypic cellular interactions influencing the growth and differentiation of biopotential oligodendrocytes-type-2 astrocytes progenitors in cultures Dev Biol 1991 114 16 29 1995394 10.1016/0012-1606(91)90474-H
Lawson L Perry V Dri P Gordon S Heterogeneity in the distribution and morphology of microglia in the normal adult mouse brain Neuroscience 1990 39 151 170 2089275 10.1016/0306-4522(90)90229-W
Rockett KA Awburn MM Aggarwal BB Cowden WB Clark IA In vivo induction of nitrite and nitrate by tumor necrosis factor, lymphotoxin, and interleukin-1: possible roles in malaria Infect Immun 1991 60 3725 3730 1500182
Meda L Cassatella MA Szendrei G Otvos L Baron P Villalba M Ferrari D Rossi F Activation of microglial cells by β-amyloid protein and IFN-γ Nature 1995 374 647 650 7715705 10.1038/374647a0
Goodwin JL Uemura E Cunnick JE Microglial release of nitric oxide by synergistic action of β-amyloid and IFNγ Brain Res 1995 692 207 214 8548305 10.1016/0006-8993(95)00646-8
London JA Biegel D Pachter JS Neurocytopathic effects of β-amyloid-stimulated monocytes: a potential mechanism for central nervous system damage in Alzheimer's disease Proc Natl Acad Sci USA 1996 93 4147 4152 8633031 10.1073/pnas.93.9.4147
Meda L Baron P Prat E Scarpini E Scarlato G Cassatella MA Rossi F Proinflammatory profile of cytokine production by human monocytes and murine microglia stimulated with β-amyloid(25–35) J Neuroimmunol 1999 93 45 52 10378868 10.1016/S0165-5728(98)00188-X
Kisilewsky R Snow A The potential significance of sulphated glycosaminoglycans as a common constituents of all amyloids: or, perhaps amyloid is not a misnomer Med Hypothesis 1988 26 231 236 10.1016/0306-9877(88)90125-9
Wisniewski T Frangione B Molecular biology of Alzheimer's amyloid-Dutch variant Mol Neurobiol 1992 6 75 86 1463589
Eikelenboom P Zhan SS vanGool WA Allsop D Inflammatory mechanisms in Alzheimer's disease Trends Pharmacol Sci 1994 15 447 450 7886816 10.1016/0165-6147(94)90057-4
Rogers J Infammation as a pathogenic mechanism in Alzheimer's disease Arzneimittelforschung 1995 45 439 442 7763341
McGeer PL Schulzer M McGeer EG Arthritis and anti-inflammatory agents as possible protective factors for Alzheimer's disease Neurology 1996 47 425 432 8757015
Brenman JE Bredt DS Synaptic signalling by nitric oxide Curr Opin Neurobiol 1997 7 374 378 9232800 10.1016/S0959-4388(97)80065-7
Smith MA Rickey HP Sayre LM Beckman JS Perry G Widespread peroxynitrite-mediated damage in Alzheimer's disease J Neurosci 1997 17 2653 2657 9092586
Barger SW Horster D Furukawa K Goodman Y Krieglstein J Mattson MP TNFs alpha, and beta protect neurons against amyloid beta-peptide toxicity: evidence for involvement of a kappa B-binding factor, and attenuation of peroxide, and Ca+ accumulation Proc Natl Acad Sci USA 1995 92 328 332 7568127
Akassoglou K Probert L Kontogeorgos G Kollias G Astrocyte-specific but not neuron-specific transmembrane TNF triggers inflammation and degeneration in the central nervous system of transgenic mice J Immunol 1997 158 438 445 8977220
Meda L Bernasconi S Bonaiuto C Sozzani S Zhou D Otvos L Mantovani A Rossi F Cassatella MA β-amyloid (25–35) peptide and IFNγ synergistically induce the production of the chemotactic cytokine MCP-1/JE in monocytes, and microglial cells J Immunol 1996 157 1213 1218 8757628
Shalit F Sredni B Rosenblatt-Bin H Kazimirsky G Brodie C Huberman M Beta-amyloid peptide induces TNFα and nitric oxide production in murine macrophages cultures Neuroreport 1997 8 3577 3580 9427329
Fillit H Ding WH Buee L Kalman J Altstiel L Lawlor B Wolf-Klein G Elevated circulating TNF levels in Alzheimer's disease Neurosci Lett 1991 129 318 320 1745413 10.1016/0304-3940(91)90490-K
Tarkowski E Blennow K Wallin A Tarkowski A Intracerebral production of TNF-α, a local neuroprotective agent, in Alzheimer's disease and vascular dementia J Clin Immunol 1999 19 223 230 10471976 10.1023/A:1020568013953
Wallace MN Geddes JG Farquhar DA Masson MR Nitric oxide synthase in reactive astrocytes adjacent to β-amyloid plaques Exp Neurol 1997 144 266 272 9168828 10.1006/exnr.1996.6373
Masayuki I Sunamoto M Ohnishi K Ichimori Y β-amyloid protein-dependent nitric oxide production from microglial cells and neurotoxicity Brain Res 1996 720 93 100 8782901 10.1016/0006-8993(96)00156-4
Snow AD Mar H Nochlin D Kimata K Kato M Suzuki S Hassell J Wight TN The presence of heparan sulfate proteoglycans in the neuritic plaques and congophilic angiopathy of Alzheimer's disease Am J Pathol 1988 133 456 463 2974240
Su JH Cummings BJ Cotman CW Localization of heparan sulfate glycosaminoglycan and proteoglycan core protein in aged brain and Alzheimer's disease Neuroscience 1992 51 801 813 1488123 10.1016/0306-4522(92)90521-3
Miller JD Cummings J Maresh GA Walker DG Castillo GM Ngo C Kimata K Kinsella MG Wight TN Snow AD Localization of perlecan (or perlecan-related macromolecule) to isolated microglia in vitro and to microglia/macrophages following infusion of beta-amyloid protein into rodent hippocampus Glia 1997 21 228 243 9336237 10.1002/(SICI)1098-1136(199710)21:2<228::AID-GLIA6>3.0.CO;2-2
Garcia de Yebenez EG Ho A Damani T Fillit H Blum M Regulation of the heparan sulfate proteoglycan, perlecan, by injury and interleukin-1α J Neurochem 1999 73 812 820 10428080 10.1046/j.1471-4159.1999.0730812.x
Castillo GM Cummings JA Ngo C Yang W Snow AD Novel purification and detailed characterization of perlecan isolated from the Engelbreth-Holm-Swarm tumor for use in an animal model of fibrillar Aβ amyloid persistence in brain J Biochem 1996 120 433 444 8889831
Hart M Li L Tokunaga T Lindsey R Hassell JR Snow AD Fukuchi K Overproduction of perlecan core protein in cultured cells and transgenic mice J Pathol 2001 194 262 269 11400157 10.1002/1096-9896(200106)194:2<262::AID-PATH882>3.0.CO;2-W
|
16022734
|
PMC1208935
|
CC BY
|
2021-01-04 16:36:31
|
no
|
Immun Ageing. 2005 Jul 16; 2:11
|
utf-8
|
Immun Ageing
| 2,005 |
10.1186/1742-4933-2-11
|
oa_comm
|
==== Front
J Circadian RhythmsJournal of Circadian Rhythms1740-3391BioMed Central London 1740-3391-3-101614454710.1186/1740-3391-3-10Short PaperLight-dark cycle synchronization of circadian rhythm in blind primates Silva Mayara MA [email protected] Alex M [email protected] John F [email protected] Laboratório de Cronobiologia, Departamento de Fisiologia, CB/UFRN, Natal, Brazil2005 6 9 2005 3 10 10 17 8 2005 6 9 2005 Copyright © 2005 Silva et al; licensee BioMed Central Ltd.2005Silva et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recently, several papers have shown that a small subset of retinal ganglion cells (RGCs), which project to the suprachiasmatic nucleus (SCN) and contain a new photopigment called melanopsin, are the photoreceptors involved in light-dark entrainment in rodents. In our primate colony, we found a couple of common marmosets (Callithrix jacchus) that had developed progressive and spontaneous visual deficiency, most likely because of retinal degeneration of cones and/or rods. In this study, we evaluated the photoresponsiveness of the circadian system of these blind marmosets.
Methods
Two blind and two normal marmosets were kept in cages with a controlled light-dark cycle (LD) to study photoentrainment, masking, and phase response to a dark pulse.
Results
Blind marmosets were entrained with the new LD cycle when light onsets were delayed and advanced by 6 hours. In constant light conditions, blind marmosets free-ran with a period of 23.2 hours, while normal animals free-ran with a period of 23.6 hours. All marmosets responded to dark pulses in the early subjective day with phase delays and with phase advances in the late subjective day.
Conclusion
Our results demonstrate that light can synchronize circadian rhythms of blind marmosets and consequently, that this species could be a good primate model for circadian photoreception studies.
==== Body
Introduction
In mammals, circadian rhythms of physiological and behavioral variables are driven by a master circadian pacemaker, the suprachiasmatic nucleus (SCN). To be useful, the circadian clock must be synchronized to the light and dark alternation of the real world's day-night cycles. In mammals, the eyes are required for photoentrainment. Because only rods and cones were known to be ocular photoreceptors, it was generally assumed that circadian photoreception relied on these cells. However, studies on rd/rd mice, which lack rod photoreceptors, and more recent studies on rd/rd cl mice, which lack all functional rods and cones, have provided overwhelming evidence that these classical photoreceptors are not required for photoentrainment [1,2]. Recently, several papers have shown that a small subset of retinal ganglion cells (RGCs) that project to the SCN and contain a new photopigment called melanopsin serve as photoreceptors involved in light-dark entrainment in rodents [3-7]. Melanopsin was also found in humans, other primates, rats, and mice [8]. Several studies have shown that the melanopsin photoreceptors not only regulate the circadian system, but also contribute to both papillary light reflex and acute alterations in motor activity, and may be involved in a broad range of physiological and behavioral responses to light [8].
The common marmoset (Callithrix jacchus) is a small neo-tropical primate found in the northeast of Brazil and is easily adapted to laboratory use. It is a diurnal animal and has a bimodal circadian pattern. The motor activity of these animals displays a stable circadian rhythm in constant light. Light pulses cause delays when given in the early subjective night and phase advances when given in the late subjective night [9,10]. Studies of morphology of retinal ganglion cells in marmosets have shown a sex-linked polymorphism of cone pigment expression, such that all males are dichromats and the majority of females are trichromats [11].
In our primate colony, we found a couple of common marmosets that had developed progressive and spontaneous visual deficiency. They were living in semi-natural conditions and were active during the day and inactive at night. Ophthalmoscopic examination did not show opacities of the cornea, lens, and vitreous, nor lesions of the optic nerve, nor vascular retinopathies. The most important feature in fundoscopy was a bilateral pigmentation in the macular region, which is similar to retinal rod degeneration. As it is known that ganglion cells are generally preserved in retinal degeneration disease [12], we proposed that these marmosets have a retinal degeneration of cones and/or rods. Based on previous studies, we believe that this retinal degeneration of rods and cones does not impair expression of circadian rhythmicity, photoentrainment, masking, and phase response to a dark pulse. In order to test this hypothesis, these marmosets were transferred to the laboratory so that the light-dependent features of their circadian system could be studied.
Methods
Four marmosets (one normal male, one blind male, one normal female, and one blind female), with an average age of eight years and average weight of 354 g, were housed in individual cages in a room with attenuated noise, controlled temperature (average temperature of 25.5°C), water at libitum and food daily available for 8.5 hours. The animals were first exposed to an LD cycle of 24 hours (LD 12:12). Illuminance was 150 lux during the light phase and 1 lux during the dark phase. The marmosets were adapted to the laboratory for 10 days. After two weeks in LD 12:12, the time of lights-on was delayed by 6 hours; three weeks later, it was advanced by 6 hours. Four weeks later, the marmosets were placed in constant light conditions for 4 weeks. The marmosets were then returned to LD 12:12 for 3 more weeks and then again to constant light for 4 weeks.
General circadian locomotor activity of the marmosets was measured using an infrared motion sensor above the cage. Output from the sensors was integrated with an IBM-compatible computer running data acquisition software. Analyses of rhythm characteristics and graphical output, actograms, were undertaken using the El Temps computer program (Diez-Noguera, Barcelona, Spain). The free-running period of the locomotor activity rhythm under constant light was computed by the chi square periodogram procedure [13] with a global risk level (α) of p < 0.05. Under LL, the onset of activity, designated as circadian time (CT) 0, was used as the phase reference point for the onset of the subjective day. Phase-shifts were determined as the difference between projected times of activity onset on the day after dark stimulation. The dark stimulation consisted of 2-hour pulses of darkness. Experiments were in compliance with the institutional guidelines of the Universidade Federal do Rio Grande do Norte and Sociedade Brasileira de Neurociência e Comportamento.
Results and discussion
The marmosets were submitted to two behavioral tests. In the first one, a non-smelling object (such as a pen or a key) was placed two centimeters away from each animal's face. The normal marmosets directed their sight to the object and tried to grab and bite it; the blind animals did not react to the objects at all. The same test was repeated with the objects in movement, and again only the normal marmosets reacted by directing their sight to the moving object. The second test took place in a room with dim light (10 lux). A spotlight was placed on one side of the animals and directed to their faces. The normal marmosets turned their faces towards the light, while the blind ones did not.
As shown in Figure 1, blind marmosets were clearly synchronized to the external cycles. Like normal animals, blind animals showed a normal biphasic activity circadian rhythm, with a more intense bout of activity at the beginning of the light phase and a second bout near the end. However, this bimodal pattern was less prominent in blind animals (Figure 2). Additionally, blind marmosets showed a shorter active phase compared to normal animals. After we shifted the light phase by 6 hours (first a delay and then an advance), blind marmosets were entrained to the new light-dark cycle, but their entrainment was much slower compared to the normal marmosets (Figures 1 and 2). The blind animals synchronized only after 12–14 days, while the normal animals did so after 3–4 days. During entrainment, the phase angle of activity onset in relation to the LD cycle was different in blind and normal marmosets (see Figure 2).
Figure 1 Light-dark cycle synchronization of circadian rhythms in blind primates. Shown are representative double-plotted motor activity records of a blind marmoset during photoentrainment, advance and delay of light phase, and free-run in constant light condition (LL). Time is indicated at the bottom, day at left side, and the light-dark cycle (LD) or LL on the right side. The boxes represent the light phase of the LD cycle.
Figure 2 Circadian pattern of motor activity in blind and normal marmoset. (A) Representative double-plotted motor activity of a blind (left) and a normal (right) animal. (B) Wave form plot of activity of blind (left) and normal (right) marmoset. Both animals showed a bimodal pattern of activity, but this bimodal pattern was less prominent in the blind animal, which showed a shorter active phase compared to the normal animal. (C) Periodogram of activity rhythms for blind (left) and normal (right) marmoset in the LD cycle. (D) Entrainment of the blind marmoset after the time of lights-on was delayed by 6 hours. (E) Entrainment of the sighted marmoset after the time of lights-on was delayed by 6 hours.
The animals were then placed in constant light conditions in order to determine if they had functional circadian clocks. As show in Figure 1, the marmosets showed free-running circadian rhythms. Free-running periods were significantly different between the two groups: blind marmosets showed a 23.2-hour period while normal marmosets free-ran showed a period of 23.6 hours. This shorter period in blind marmosets could be explained by their lower activity and, consequently, decreased motor activity feedback to the circadian system, or by the participation of classical photoreceptors (rods and cones) in the generation of free-running circadian rhythms. However, it could also be explained by the suggestion that the loss of rods and cones has an impact on the nature of light information reaching the SCN [8].
For photic entrainment to occur, the circadian oscillator must respond differently to light at different phases of its cycle. Phase response curves (PRC) are useful descriptions of these phase-dependent responses. A number of non-photic stimuli, both pharmacological and non-pharmacological, have been identified as able to induce phase shifts in mammalian circadian clocks as a function of the circadian phase that the stimulus is presented [14]. The PRC of non-photic stimuli (including dark pulses presented to animals kept in constant light) is 180° out of phase with photic stimuli. We tested the phase-shift response of blind marmosets using dark pulses of 2-hour duration. When the dark pulse was given in the early subjective day, it caused a phase delay; when given in the late subjective day, it caused a phase advance. This result is an important contribution to the discussion about the non-photic phase shift in this species. Our results agree with Glass et al [14], in whose studies the qualitative similarities between the phase responses to entraining photic and non-photic stimuli in marmosets and nocturnal mammals were demonstrated.
Many of the non-photic stimuli that induce phase shifts in the circadian clocks also induce an acute increase in locomotor activity in nocturnal mammals, and it appears that at least some of the phase shifting effects of these agents is due to the induction of activity and/or arousal [15]. In the present study in marmosets, the phase shifts produced by dark pulses were not due to the inhibition of activity. The dark pulses produced an inhibition of activity (negative masking) in the sighted marmosets but not in the blind ones, despite the fact that both groups showed phase shifts with dark pulses.
The response of the circadian system to different stimuli, photic and non-photic, is of great importance because implies that circadian systems are in fact able to use many sources of information. As the marmoset is a social animal, we also investigated social synchronization between these animals. Blind marmosets showed different activity onset during the free-running phase, but they showed a stable phase angle. The two normal marmosets showed the same behavior but with different free-running periods from the blind marmosets. Therefore, despite the fact that the four animals were in the same room, the blind marmosets were not synchronized with the normal ones.
One limitation of this study is the small number of animals, but the results of the two normal marmosets are similar to other studies that were conducted in our laboratory [16]. Considering studies previously carried out in rodents along with our present results, it is possible to infer that the blind marmosets had normal retinal ganglion cells, which are required to synchronize their circadian clocks to the LD cycle. In the absence of classical photoreceptors, photosensitive ganglion cells are sufficient for photic entrainment [17].
Conclusion
Ours results constitute the first experimental evidence that non-classical retinal photoreceptors can provide photic information to the circadian system of primates and diurnal animals. The blind marmosets may provide an excellent model for the study of photoreception and entrainment in primates. Possible benefits are obvious, such as the development of strategies to solve the problem of synchronization in blind humans or to study retinal degeneration.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MMAS: Participated in all experiments, in the analysis and discussion of the results, and in the writing of the manuscript.
AMA: Participated in all experiments, in the analysis and discussion of the results, and in the writing of the manuscript.
JFA: Participated in all experiments, in the analysis and discussion of the results, and in the writing of the manuscript.
All authors read and approved the final manuscript.
Acknowledgements
We thank Dr. Alexandre Bezerra Gomes for performing the ophthalmoscopic exam in our animals. Supported by grants from CNPq (to JFA and MMAS) and PPq-UFRN (to AMA).
==== Refs
Foster RG Keeping an eye on the time: the Cogan Lecture Invest Ophth Vis Sci 2002 43 1286 1298
Lucas RJ Douglas RH Foster RG Characterization of an ocular photopigment capable of driving pupillary constriction in mice Nature Neurosci 2001 4 621 625 11369943 10.1038/88443
Berson DM Dunn FA Takao M Phototransduction by retinal ganglionar cells that set the circadian clock Science 2002 295 1070 1073 11834835 10.1126/science.1067262
Provencio I Rollag MD Castrucci AM Photoreceptive net in the mammalian retina. This mesh of cells may explain how some blind mice can still tell day from night Nature 2002 415 493 11823848 10.1038/415493a
Ruby NF Brennan TJ Xie X Cao V Franken P Heller HC O'Hara BF Role of melanopsin in circadian responses to light Science 2002 298 2211 2213 12481140 10.1126/science.1076701
Hattar S Liao HW Takao M Berson MD Yau KW Melanopsin-containing retinal ganglion cells: architecture, projections, and intrinsic photosensitivity Science 2002 295 1065 1070 11834834 10.1126/science.1069609
Lucas RJ Hattar S Takao M Berson MD Foster RG Yau KW Diminished pupillary reflex at high irradiances in melanopsin-knockout mice Science 2003 299 245 247 12522249 10.1126/science.1077293
Foster RG Hankins M Lucas RJ Jenkins A Muñoz M Thompson S Appleford JM Bellingham J Non-rod, non-cone photoreception in rodents and teleost fish. Molecular clocks and light signal Novartis Found Symp 2003 253 3 30 14712912
Erket HG Characteristics of the circadian activity rhythm in common marmoset (Callithrix j. jacchus) Am J Primatol 1989 17 271 286 10.1002/ajp.1350170403
Wechselberger E Erkert HG Characteristics of the light-induced phase response of circadian activity rhythms in common marmoset, Callithrix j. jacchus (Primates Cebidae) Chronobiol Int 1994 11 275 284 7828210
Ghosh KK Goodchild AK Sefton AE Martin PR Morphology of retinal cells in new world monkey, the marmoset Callithrix jacchus J Comp Neurol 1996 366 76 92 8866847 10.1002/(SICI)1096-9861(19960226)366:1<76::AID-CNE6>3.0.CO;2-H
Chierzi S Cenni MC Maffei L Pizzorusso T Porciatti V Ratto GM Strettoi P Protection of retinal ganglionar cells and preservation of function after optic nerve lesion in bcl-2 transgenic mice Vision Res 1998 38 1537 1543 9667018 10.1016/S0042-6989(97)00332-5
Sokolove PG Bushell WN The chi square periodogram: its utility for analysis of circadian rhythms J Theor Biol 1978 72 131 160 566361 10.1016/0022-5193(78)90022-X
Glass JD Tardif SD Clements R Mrosovsky N Photic and nophotic circadian phase resenting in a diurnal primate, the common marmoset Am J Physiol 2001 280 R191 R197
Mrosovsky N Locomotor activity and nonphotic influences on circadian clocks Biol Ver Camb Philos Soc 1996 71 343 372
Mendes ALB Influência de pistas sociais sobre a sincronização do ritmo circadiano de atividade motora no sagui, Callithrix jacchus, ao ciclo claro-escuro Master Thesis 2003 Psychobiology Graduate Program, UFRN
Mrosovsky N Contribution of classic photoreceptors to entrainment J Comp Physiol [A] 2003 189 69 73
|
16144547
|
PMC1208936
|
CC BY
|
2021-01-04 16:39:13
|
no
|
J Circadian Rhythms. 2005 Sep 6; 3:10
|
utf-8
|
J Circadian Rhythms
| 2,005 |
10.1186/1740-3391-3-10
|
oa_comm
|
==== Front
J Exp Clin Assist ReprodJournal of Experimental & Clinical Assisted Reproduction1743-1050BioMed Central London 1743-1050-2-101609553010.1186/1743-1050-2-10CommentaryClarifying the role of three-dimensional transvaginal sonography in reproductive medicine: an evidenced-based appraisal Raine-Fenning Nick [email protected] Arthur C [email protected] Obstetrics & Gynaecology, Queen's Medical Centre, University Hospital NHS Trust, NURTURE, B Floor, East Block, Nottingham, NG 7 2UH UK2 Department of Radiology and Radiological Sciences, Vanderbilt University Medical Center, 1116 21st Avenue, South, Nashville, TN 37232-2675 USA3 Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, 1116 21st Avenue, South, Nashville, TN 37232-2675 USA2005 11 8 2005 2 10 10 26 5 2005 11 8 2005 Copyright © 2005 Raine-Fenning and Fleischer; licensee BioMed Central Ltd.2005Raine-Fenning and Fleischer; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This overview describes and illustrates the clinical applications of three-dimensional transvaginal sonography in reproductive medicine. Its main applications include assessment of uterine anomalies, intrauterine pathology, tubal patency, polycystic ovaries, ovarian follicular monitoring and endometrial receptivity. It is also useful for detailed evaluation of failed and/or ectopic pregnancy. Three-dimensional color Doppler sonography provides enhanced depiction of uterine, endometrial, and ovarian vascularity.
==== Body
Background
Conventional sonography provides two-dimensional views of three-dimensional structures that an experienced ultrasonographer has to dynamically examine in order to create their own three-dimensional impression of the object of interest [1]. In contrast, three-dimensional sonography allows the simultaneous assessment of individual sectional planes, which dependent upon the particular field of interest may be examined in one of several different viewing modalities to maximise the information available and improve spatial awareness (Fig. 1) [2,3]. Uniquely, three-dimensional sonography allows demonstration of the coronal plane perpendicular to the transducer face facilitating the identification of surface irregularities which can then be accounted for during volume measurement [4]. The digital technology central to its development also means that three-dimensional imaging lends itself to telemedicine, as it allows the storage of large datasets without loss of information that may be subsequently analysed off-line and reappraised by experts in a 'virtual real-time consultation' [5]. Two-dimensional color Doppler sonography provides a subjective estimation of uterine and ovarian vascularity. It is limited, however, by providing flow depiction in a single plane as opposed to the sample volume as obtained by three-dimensional imaging (Fig. 1). Advocates of three-dimensional sonography [6] suggest that these features offer the user the following advantages in comparison to two-dimensional sonography:
Figure 1 Two-dimensional vs three-dimensional transvaginal color Doppler sonography. (a) Diagram (left) and two-dimensional TV-CDS (right) showing arcuate, radial, and spiral vessels in follicular phase. (b) Diagram (left) are two-dimensional TV-CDS (right) showing changes in endometrial vascularity throughout the menstral cycle. During the luteal phase, several spiral vessels are detected. (c) Diagram (top) and multiplanar images (left) and magnified 3D TV-CDS (right) of corpus luteum. The multiplanar image shows the vascular "wreath" surrounding the functioning corpus luteum in the longitudinal (top left), short (axial) (top right) and coronal (bottom right) planes. The combined gray scale and 3D TV-CDS image (both left) which is also magnified and shown as the right depicts the numerous vessels surrounding this corpus luteum.
• accurate measurement of organ dimensions and volumes
• improved anatomic and blood flow information
• improved assessment of complex anatomic anomalies
• a better specificity in regard to the confirmation of normality
• standardisation of the sonographic examination procedure
• reduced scanning times with cost-effective use of equipment and sonographer time
• telemedicine and tertiary consultation
This review will critically appraise the evidence to see if the potential advantages of three-dimension over two-dimension may be substantiated in the context of reproductive medicine and specifically ask if they actually lead to an improved diagnostic capability. To begin it is first necessary to outline how three-dimensional sonography may be used to quantify volume and blood flow as these measurements form the basis of many studies.
Three-dimensional Data Analysis
When one thinks of three-dimensional imaging in terms of its measurement capability the most obvious parameter considered is that of volume. Whilst volume may be estimated from measurements made with conventional two-dimensional sonography, such measurements use various formulae based upon certain geometric assumptions [7]. Volume estimation based on three-dimensional sonography still involves a degree of geometric assumption, as data are reconstructed based upon their most probable position within a Cartesian grid system, but utilises much more information. There are two basic methods employed to calculate volume from a three-dimensional dataset: the conventional 'full planar' or 'contour' method (Fig. 2a) and the more recently introduced 'rotational' method possible through the VOCAL-imaging program (Virtual Organ Computer-aided AnaLysis™) (Fig. 2b) which also generates a three-dimensional model of the object of interest (Fig. 3). Volume calculation by either of these techniques has proven highly reliable and valid both in vitro and in vivo [8-15]. Both techniques involve the manual delineation of the object of interest in the multiplanar display that shows the three perpendicular planes characteristic of three-dimensional sonography but there are other advantages to the 'rotational' technique in that it facilitates assessment of blood flow in a novel manner through the quantification of the power Doppler signal both within the defined volume of interest and also within the surrounding tissue through the application of a shell parallel to the originally defined surface contour (Fig. 4). Three indices of vascularity are calculated: the Vascularisation Index (VI) reflects the ratio of power Doppler information within the total dataset relative to both colour and grey information, the Flow Index (FI) represents the mean power Doppler signal intensity and the Vascularisation Flow Index represents a combination of the two (Fig. 5) [16]. The exact relationship of these indices to true flow and vascularity in vivo remains to be established but they have been shown to vary both within an individual and between different subjects suggesting they could have a valuable role in identifying and categorising differences between patient groups (Fig. 6). Importantly, the indices may be calculated in a reproducible manner between observers [17] following the three-dimensional acquisition of power Doppler data which itself has also been shown to be reliable [18].
Figure 2 The techniques of volume calculation. Both figures show the typical multiplanar display of a three-dimensional sonographic dataset of the uterus with the three mutually related orthogonal planes at 90-degrees to one another. The upper left image in both displays represents the longitudinal plane (the A plane), the upper right image the transverse plane (the B plane) and the lower left image the coronal plane (the C plane). Volume calculation can be conducted in either the B or C plane. (a) shows the conventional technique of volume calculation in which a series of 'slices' are taken through the volume of interest whilst the contour is outlined in another plane (the transverse plane in this case). The distance between consecutive slices can be varied according to the degree of change in the surface contour and increased for more complex structures. (b) shows the rotational technique of volume calculation in which the dataset is rotated through 180° about a central axis defined by the application of two callipers. The number of planes available for volume calculation are determined by the rotation step shown in the lower left of the image. Here the 30-degree rotation step has been used and the contour outlined in the coronal or C plane using the manual mode. The resultant three-dimensional model is shown in the lower right of the image.
Figure 3 Three-dimensional 'wire' models of the endometrium. When VOCAL is used to calculate volume a three-dimensional model is generated that may be rotated and examined in its natural orientation. In this model the coronal view is seen on the left clearly demonstrates the fundal defect of an arcuate uterus whilst the actual anteflexion of the uterus may be appreciated in the lateral view on the right.
Figure 4 Three-dimensional power Doppler angiography of the uterine blood supply. A three-dimensional dataset containing power Doppler information has been acquired from the uterus. VOCAL has then been used to define the myometrial-endometrial border and to apply a shell 5 mm outside of that contour to define the sub-endometrium, which can be clearly seen to be more vascular than the endometrium itself.
Figure 5 The 'Histogram'. The three indices of vascularity have been calculated for the endometrial model together with the mean grey value. The Vascularisation Index (VI) reflects the degree of power Doppler information within the model and is considered as a percentage therefore whilst the Flow Index (FI) and Vascularisation Flow Index (VFI) include information on the mean power Doppler signal intensity referenced against a scale from zero to one hundred to indicate the minimal and maximum range accordingly.
Figure 6 Intra-subject variation in endometrial blood flow. The degree of power Doppler information seen within Figure 6a is clearly superior to that seen in Figure 6b and this difference is quantifiable through the 'histogram' facility. These images also serve to demonstrate how vascularity is independent of morphometry and varies throughout different phases of the menstrual cycle as the less vascular homogenous endometrium characteristic of the luteal phase in Figure 6b is of greater volume than that in Figure 6a, which is the same endometrium at the end of follicular phase.
Clinical applications of three-dimensional sonography
Having established the key features of three-dimensional sonography in terms of its measurement ability and improved spatial awareness let us now examine how these have been applied in the diagnosis of subfertility and subsequent monitoring of treatment.
Investigation of subfertility
Three-dimensional ultrasound has been used to diagnose uterine anomalies, assess tubal patency and to exclude intrauterine and ovarian pathology.
• Uterine anomalies
Congenital uterine anomalies are associated with an increased risk of repeated first and second trimester miscarriage and preterm delivery [19]. A meta-analysis of published retrospective data comparing pregnancy outcome before and after hysteroscopic septoplasty indicated a marked improvement after surgery, which itself has minimal postoperative sequelae [20]. Accurate and reliable diagnosis is important therefore as it allows the identification of patients at risk of these complications and timely surgical intervention. This is undoubtedly the area where three-dimensional sonography has contributed the most and has become the investigation of choice in units where available. This reflects its ability to demonstrate both the endometrial cavity and the myometrium simultaneously in the coronal plane as shown by Jurkovic et al. in their pioneering study of 61 patients with a history of recurrent miscarriage or infertility (Figs. 7) [21]. Hysterosalpingography had shown the presence of a normal uterus in 44 (72.1%) patients, an arcuate uterus in nine (14.8%) and a major fusion defect in three cases (4.9%). Whilst two-dimensional sonography was associated with five false-positive diagnoses of arcuate uteri and three of major uterine anomalies, three-dimensional sonography agreed with hysterosalpingography in all of these cases. Shortly afterwards Raga et al. and Wu et al. both reported similarly favourable results suggesting three-dimensional sonography offered a 100% specificity for the exclusion of uterine anomalies and was able to differentiate between the different anomalies [22,23].
Figure 7 Uterine anomalies. Three-dimensional sonography has become the 'gold standard' investigation for the diagnosis and exclusion of congenital uterine anomalies. Its extremely high sensitivity and specificity relate to its ability to demonstrate the plane coronal perpendicular to the transducer face and in doing so allow visualisation of the fundal contour and comparison of the myometrium with the endometrium throughout the uterine length. Figure 7a shows normal cornna with straight contours at the upper aspect of the cavity in contrast to the characteristic concave contour seen in arcuate uteri (Fig. 7b) and the deeper contour of various length seen in sub-septate uteri (Figs. 7c, 7d & 7e). Any indentation of the fundal contour may also be appreciated in the coronal plane as seen in the multiplanar display in Figure 7f. This uterus had been considered normal with conventional ultrasound, which only provides the longitudinal and transverse views seen in the upper two images, but laparoscopy had demonstrated a bulky uterus with a possible fundal defect and a follow-up three-dimensional ultrasound confirmed the presence of a significant septum. The size of the septal defect can be measured as shown in Figure 7g but may be less important than the remaining length of the cavity shown as a bold dashed line.
Three-dimensional sonography has since been used to determine the prevalence of uterine anomalies in various patient groups and to characterise outcome on the basis of the anomaly. As many as 24% of women with recurrent pregnancy loss may have uterine anomalies [14] which is roughly four times that seen in low-risk women where the prevalence is in the order of 5 to 6% [25,26]. In terms of the type of anomaly a similar distribution is seen between different groups with arcuate uteri being the most common, followed by subseptate then bicornuate uteri with the more complex anomalies such as uterus didelphys and single uterine horns the least prevalent. Women with a subseptate uterus have a significantly higher proportion of first-trimester loss and women with an arcuate uterus a significantly greater proportion of second-trimester loss (p < 0.01) and preterm labor (p < 0.01) compared to women with a normal uterus [24]. Another important finding derived from these three-dimensional studies has been that outcome is related not only to the degree of defect but also to the remaining cavity length which is significantly shorter in both arcuate and subseptate uteri in women with recurrent miscarriage [26]. These measurement techniques and classifications used allow comparison between these studies as a degree of standardisation has been used, based on measurement of fundal distortion from the mid-point of an imaginary horizontal line joining the upper aspects of the cornuae to the upper aspect of the uterine cavity, that has been shown to be reliable between observers examining stored three-dimensional datasets (Fig. 7f) [27].
If we revisit the advantages proposed at the start of the review we can see that the majority are already satisfied in respect to uterine anomalies. Three-dimensional sonography offers a reliable and standardised tool to diagnose, differentiate and quantify uterine anomalies. Three-dimensional sonography has significantly added to our understanding of uterine anomalies qualifying their effect on reproductive outcome and thereby helping the clinician counsel patients accordingly and confidently.
• Intrauterine pathology
La Torre et al. compared three-dimensional sonography with conventional imaging with and without saline contrast in their study of twenty-three patients in whom subsequent hysteroscopy revealed the presence of 16 endometrial polyps [28]. Standard two-dimensional sonography demonstrated a relatively poor specificity of only 69.5% suggesting the presence of polyps in 23 patients. This was improved to 94.1% when two-dimensional sonography was used in conjunction with saline infusion as only 17 patients were then thought to have polyps. Three-dimensional sonography performed almost as well diagnosing the presence of polyps in 18 patients with a specificity of 88.8% and subsequently correctly identified all 16 polyps when used in conjunction with saline infusion. A similar improvement in specificity with three-dimensional sonography has been shown by Sylvestre et al. in their study of 209 subfertile patients thought to have an intrauterine lesion on transvaginal two-dimensional sonography or hysterosalpingography [29]. Using saline infusion sonography with two-dimensional and then three-dimensional sonography, 92 patients were subsequently identified as having a variety of intrauterine lesions suggesting a sensitivity and specificity of 97% and 11% for two-dimensional sonography, 87% and 45% for three-dimensional sonography and 98% and 100% for two-dimensional saline infusion sonography. Of 59 patients that had undergone hysteroscopy, the sensitivity and positive predictive value of saline infusion sonography were 98% and 95% when performed in combination with two-dimensional sonography and 100% and 92% with three-dimensional sonography respectively. The study clearly demonstrates how simple contrast media potentially increase the specificity of two-dimensional sonography as 55% (116 of 209) of patients were found to have normal cavities following the infusion of saline. This was largely due to the correct localization of leiomyomas as intramural rather than submucosal (54 of 101). This is extremely important in the clinical setting, as whilst three-dimensional sonography improves further on two-dimensional imaging it is not currently available in all units and its use therefore has significant implications in terms of cost and training. Saline infusion on the other hand may be undertaken easily using various dedicated systems or modifications of readily available cheaper catheters such as a paediatric feeding tube. Figure 8 shows the typical three-dimensional sonography appearances of endometrial polyps at different stages of the menstrual cycle whilst Figure 9 demonstrates a sub-mucous fibroid at three-dimensional saline infusion sonography.
Figure 8 Endometrial polyps. Endometrial polyps are best appreciated during the late follicular phase when they can be seen against the plump heterogenous endometrium with its characteristic trilaminar pattern may be seen (Fig. 8a). Polyps may also be seen towards the end of menses however or at the 'down-regulation' scan when many women are still bleeding (see below) with the menstrual fluid acting as an in vivo contrast agent (Fig. 8b). The coronal view may be used to locate the position of intrauterine pathology and confirm the presence of individual smaller polyps suitable often confused as large single fibroids with conventional imaging that are suitable for routine polypectomy rather then hysteroscopic resection (Fig. 8c)
Figure 9 Three-dimensional saline infusion contrast sonography. Two-dimensional ultrasound had shown a persistently thick endometrium despite several weeks of pituitary suppression prior IVF treatment and saline infusion sonography has proven unhelpful simply confirming a thick endometrium despite very low serum oestradiol levels. A three-dimensional acquisition was undertaken therefore both before (Fig. 9a) and during the instillation of saline (Fig. 9b) and subsequently analysed off-line to reduce the amount of saline used and overall time needed for the examination. A large sub-mucosal leiomyoma with a broad base was seen originating from the left posterior aspect of the uterus that almost entirely occupied the cavity (Fig. 9b). This allowed the cancellation of the planned IVF treatment with the patient continuing her 'down-regulation' medication for a further two-months at which stage the fibroid was removed transcervically using a modified operative hysteroscope.
• Tubal patency
The combination of contrast media and three-dimensional sonography has also been used to assess tubal patency. Kiyokawa et al. found three-dimensional saline sonohysterosalpingography was able to demonstrate the entire contour of the uterine cavity in 96% of cases compared to only 64% cases with conventional X-ray hysterosalpingography (p < 0.005) and was associated with a positive predictive value and specificity of predicting tubal patency of 100% in 25 unselected infertile patients [30]. Sladkevicius et al. found three-dimensional power Doppler imaging demonstrated free spill almost twice as often as conventional imaging (114 versus 58 tubes respectively) when used during hysterosalpingo-contrast sonography [31]. Sankpal et al. reported less promising results with the same technique when they reassessed tubal patency in 15 women who had normal X-ray HSG examinations within the previous year [32]. An important distinction was their use of saline rather than a positive contrast agent but it maybe that the technique has a distinct learning curve and numbers were small.
• Polycystic Ovaries
Several groups have used three-dimensional sonography to demonstrate ovarian volume and vascularity are increased in polycystic ovarian syndrome [33,36]. Three-dimensional sonography also allows for the measurement of stromal volume through the calculation and subtraction of total follicular volume from total ovarian volume (Fig. 10). Using this technique Kyei-Mensah et al showed stromal volume was positively correlated with serum androstenedione concentrations (p < 0.01) in 26 women with clinical evidence of polycystic ovaries [37]. However, using a similar approach, Nardo et al. were unable to demonstrate any relationship between serum FSH, LH or testosterone and ovarian stromal volume in 23 infertile women with clomiphene citrate-resistant PCOS at the same stage of the menstrual cycle [38].
Figure 10 Polycystic ovaries. Three-dimensional sonography facilitates objective assessment of the ovarian stroma, through measurement of its mean grey signal intensity, its vascularity and its volume, which may be calculated by subtracting total follicular volume from total ovarian volume (Fig. 9A). Ovarian blood flow is increased and associated with significantly higher three-dimensional indices of vascularity than ovaries with a normal appearance (Fig. 10b).
Assisted reproduction treatment
Transvaginal sonography is used on a daily basis to monitor the response to treatment and to guide the transvaginal collection of oocytes and subsequent transcervical transfer of embryos to the uterus. Three-dimensional sonography may be used in any of these areas but has largely been applied as a predictor of ovarian response and as a determinant of endometrial receptivity. To clarify the current evidence in relation to the role of three-dimensional sonography in these areas it is necessary to first outline the general principles involved in assisted reproduction treatment.
• Principles of in vitro fertilisation therapy
The first stage is to obtain control over the hypothalamic-pituitary-ovarian axis through the induction of pituitary 'down-regulation' achieved through a continuous and highly supraphysiological dose of a gonadotrophin releasing hormone (GnRH) analogue. Follicular development is then induced through the administration of natural or synthetic Follicle Stimulating Hormone (FSH) until at least three large follicles measuring 18 mm or more are present when the resumption of meiosis is initiated though a single dose of human Chorionic Gonadotrophin (hCG), which has a similar structure and pharmacodynamic effect to Luteinising Hormone (LH) but is more readily available. Exposure to hCG is essential prior to the collection of the oocytes to ensure that they are mature and therefore capable of being fertilised when incubated with sperm during in vitro fertilisation (IVF) or when injected with a single sperm in intracytoplasmic sperm injection (ICSI) treatment. This would eventually lead to ovulation in vivo if the oocytes were not physically retrieved and this procedure is critically timed therefore to occur within 34 to 36 hours of the administration of hCG.
The dose and duration of FSH treatment is dependent upon each individual patients response to stimulation and is adjusted dynamically during treatment dependent on the number of follicles developing and on the amount of oestrogen they produce collectively as measured in the serum. The response and eventual number of mature oocytes retrieved is referred to as the patients 'ovarian reserve'. Prediction of ovarian reserve has obvious importance as it allows treatment to be tailored to the individual potentially increasing the number of oocytes retrieved.
Fertilisation of an oocyte is confirmed by the presence of two pronuclei 18 to 24 hours after IVF or ICSI. The embryos are allowed to develop in vitro until the second or third day after oocyte retrieval when most units would transfer a maximum of two to the recipient uterus selected on the basis of their quality as determined by a subjective grading system. The chance of the implantation occurring is not only dependent upon the embryo stage and grade but also on the endometrium. Just how important 'endometrial receptivity' is in the overall process remains uncertain but the endometrium is likely to play more than a passive role.
With a better understanding of processes involved in controlled ovarian stimulation and assisted reproduction let us now turn to the application of three-dimensional sonography and critically review its current role.
• Three-dimensional markers of 'ovarian reserve'
Of the sonographic markers suggested as predictive of ovarian response the three that have been specifically addressed by three-dimensional sonographic studies are antral follicle counts, ovarian volume and ovarian blood flow.
Pellicer et al. were amongst the first to use three-dimensional sonography as an adjunct to conventional markers of ovarian reserve when they examined ovarian volume and the number of 'selectable follicles' measuring 2–5 mm in a small group of low responders on day three of the menstrual cycle [39]. Both the number of selectable follicles and the total number of antral follicles were significantly decreased in the 'low responder' group who also demonstrated significantly higher serum FSH levels despite having values within the normal range. Ovarian volume measurements, however, were similar between the two groups. Pohl et al. also used three-dimensional sonography to quantify the number of follicles of varying diameter in 113 patients following 'down-regulation' but prior to ovarian stimulation ([40]. Patients with a higher number of follicles measuring between 5 and 10 mm were younger (p < 0.01), had a significantly higher number of oocytes retrieved (p < 0.0001) and were more likely to conceive (p < 0.05) (Fig. 12). Kupesic et al also suggest that the antral follicle count is a better predictor than three-dimensional measures of ovarian volume and blood flow [41]. A minimal ovarian volume may be important however (Fig. 11). Schild et al. noted a pregnancy rate of only 6.7% (1 of 15) in patients with a minimum unilateral ovarian volume of ≤ 3 cm3, which represented a single standard deviation below the mean, versus 21.9% (30 of 137) in patients with an initial minimum ovarian volume above 3 cm3[42]. This difference was not significant however and cancellation rates due to poor ovarian response or failed fertilisation were similar in both groups.
Figure 11 Ovarian volume calculation. Ovarian volume may be measured in the same way as endometrial volume through the manual delineation of the ovarian cortex.
Figure 12 Controlled ovarian stimulation. Three-dimensional sonography may be used to facilitate standard measurement of follicular diameter (a). During the latter stages of ovarian stimulation the ovaries may contain many follicles, which become progressively harder to measure reliably between observers, as there are no landmarks to aid orientation. Off-line analysis of stored three-dimensional datasets reduces the time spent with the patient and may potentially improve the number of mature oocytes retrieved by identifying the most appropriate time for oocyte collection. Evidence also suggests that three-dimensional rendering allows demonstration of the cumulus oophorus complex in follicles containing a mature oocyte (b).
There is no doubt that antral follicle counts, when used in categorical classifications, are an important predictor of 'ovarian reserve' and may be measured with a high level of agreement both between and within observers [43]. Tree-dimensional sonography, however, does not appear to offer any significant advantage over two-dimensional imaging even at higher follicle counts when interobserver reliability is reduced. Ovarian volume has a limited predictive ability that does not appear to supersede that of antral follicle counts. Do measurements of ovarian vascularity add anything to validate the use of three-dimensional sonography as a marker of ovarian reserve?
Jarvela et al. used three-dimensional power Doppler angiography after pituitary 'down-regulation' and during gonadotrophin stimulation to compare ovarian vascularity in 33 women with normal ovarian reserve, as judged by antral follicle counts, to 12 women who had demonstrated a previous poor response [44]. The number of oocytes retrieved correlated with the antral follicle count (R = 0.458, p < 0.01) and ovarian volume (R = 0.388, p < 0.05) but not with ovarian vascularity. All three indices of vascularity were shown to increase significantly during gonadotrophin stimulation in the group with normal ovarian reserve only but this was related to the antral follicle count reiterating the importance of this marker as an independent variable. Kupesic et al. similarly showed the number of oocytes retrieved and subsequent conception rate to be greater in patients with a greater ovarian volume and a greater ovarian stromal vascularity but not independently of a higher number of antral follicles [45].
• Three-dimensional markers of 'endometrial receptivity'
Schild et al. measured endometrial thickness and volume in a total of 47 IVF cycles on the day of oocyte retrieval [46]. There were no significant differences between the group of fifteen patients that conceived (31.9%) and the remaining 32 non-pregnant women in terms of the mean endometrial thickness (10.8 ± 2.3 mm versus 11.8 ± 3.4 mm) or volume (4.9 ± 2.2 cm3 versus 5.8 ± 3.4 cm3) respectively. Yaman et al. reported similar findings with no differences in endometrium volume (4.16 ± 1.97 cm3 versus 4.53 ± 1.79 cm3), endometrium thickness (11 ± 2 mm versus 11 ± 2 mm) between 21 pregnant and 44 non-pregnant women on the day of hCG administration [47]. However, whilst there was no absolute endometrial thickness required for pregnancy a minimal endometrial volume above 2.5 cm3 favoured pregnancy. Raga et al. also described a cut-off point in endometrial volume in their study of 72 patients on the day of embryo transfer [48]. Implantation rates were significantly lower (p < 0.05) in those women with endometria measuring less than 2 cm3 and no pregnancies were observed at volumes below 1 cm3. Even if there is a lower limit in endometrial volume below which pregnancy will not occur conventional measurement of endometrial thickness is already known to have a similar negative predictive value with conception less likely in patients with an endometrium measuring less than 5 mm in diameter.
Kupesic et al. reported more predictive information could be derived at the time of embryo transfer when three-dimensional power Doppler was used to quantify endometrial vascularity [49]. Of 89 patients studied successful conception cycles were associated with a significantly higher endometrial flow index (13.2 ± 2.2 versus 11.9 ± 2.4, p < 0.05). Wu et al. also found three-dimensional power Doppler angiography to be an important determinant of 'endometrial receptivity' but on the day of hCG administration in 54 patients undergoing their first IVF cycle [50]. The subendometrial vascularisation flow index (VFI) proved the best predictor of conception being superior to the vascularisation index (VI), flow index (FI) and endometrial volume in the receiver operating characteristics curve analysis. Interestingly, three-dimensional sonography may also be used to examine endometrial vascularity and determine 'endometrial receptivity' prior to ovarian stimulation. Schild et al. reported significantly lower indices of vascularity at down-regulation in 15 patients that subsequently conceived (20%) than in 60 non-conception cycles (p < 0.05) with the flow index the strongest predictive factor of IVF success (p < 0.05) [51]. Endometrial measurements were once again not correlated with outcome. This may reflect a more profound pituitary suppression but is more likely to reflect patients responsive to exogenous hormonal therapy.
Applications in early pregnancy
• Miscarriage
Gestational sac volume has been proposed as representative of the competence of the early uteroplacental unit and therefore a potential predictor of pregnancy outcome [52]. In one of the first reported applications of three-dimensional sonography, Steiner et al. studied 38 pregnancies between 5 and 11 weeks gestation and found gestational sac volume measurements to significantly correlated with gestational age (r = 0.74, p < 0.001) and more than two standard deviations below the mean in three of five embryonic or anembryonic pregnancy failures [52]. However, Acharya and Morgan have recently shown a clear relationship between gestational sac volume and sac diameter also exists in patients with embryonic pregnancy failure and that measurements are unable to predict eventual outcome during conservative management [53]. Babinszki et al. did find a relationship between gestational sac and yolk sac volume and adverse outcome in a group of 49 patients who had conceived following treatment for subfertility but crown rump length performed equally as well [54].
Currently, therefore, one may conclude that whilst there is a distinct relationship between gestational age and three-dimensional measurements of gestational and yolk sac volume, these parameters do not appear to improve upon the predictive value of current tests in determining the eventual outcome in either viable or non-viable intrauterine pregnancies.
• Ectopic pregnancy
The improved spatial orientation afforded by three-dimensional sonography has been used to evaluate endometrial shape in cases of pregnancy of unknown location in an attempt to improve the sensitivity and specificity of sonography. Rempen et al. described the persistence of a distinct symmetry with regard to the median longitudinal axis of the uterus in the coronal plane in 90% of extrauterine pregnancies which was lost in intrauterine pregnancies [55]. Su et al. reported a case of an anembryonic cervical pregnancy diagnosed at 10 weeks associated with a large arteriovenous malformation managed conservatively with selective uterine artery embolisation [56]. Three-dimensional power Doppler ultrasonography proved useful in the initial diagnosis subsequent monitoring of the response to treatment. These studies suggest a potential role for three-dimensional sonography but in the absence of properly blinded data it is impossible to predict whether it will improve upon conventional sonography. The spatial orientation and additional information afforded by the coronal plane do lend themselves to assessment of the endometrial cavity as illustrated so elegantly by the three-dimensional studies on uterine anomalies so one should be optimistic in this regard. Certainly in our own unit we have found three-dimensional imaging to be beneficial in several cases of pregnancy of unknown location and particularly so in twin pregnancies following assisted reproduction treatment and embryo transfer where heterotropic pregnancy is so much more likely (Fig. 13).
Figure 13 Ectopic pregnancy. The spatial orientation of three-dimensional ultrasound and its ability to demonstrate the coronal plane of the uterus in its natural position may help in the diagnosis and exclusion of ectopic pregnancy. This is particularly true in cases of heterotropic pregnancy where an intrauterine pregnancy is associated with an ectopic twin gestation. Spontaneous heterotropic pregnancy is rare but may complicate 0.5% of pregnancies arising from IVF treatment with the ectopic gestation located in the cornua (Fig. 13a) or cervix (Fig. 13b).
Conclusion
There is sufficient evidence to support the notion that the theoretical advantages of three-dimensional sonography are indeed translated into clinical practice in the field of reproductive medicine. The spatial orientation and additional information derivable from individual sectional planes has greatly enhanced our knowledge of uterine anomalies and contributed to our understanding of how these affect pregnancy outcome and may offer insight into the location of pregnancies of unknown location. Quantitative three-dimensional analysis of volume and vascularity has proven less powerful and whilst individual studies suggest a potential role for such measurements these do not appear to out perform current assessments.
For now three-dimensional sonography largely remains an exciting research tool with the converted applying it in different forms and areas of interest and in doing so unearthing significant new information about normal physiology and pathophysiological change that direct further work. Three-dimensional sonography offers too much to be ignored and, as history has shown with previous developments, will gradually become commonplace in most units. Our role is to continue to test it prospectively but to remain realistic and to examine how it may be most appropriately applied in the clinical setting. Be a pioneer, embrace it and you will be rewarded handsomely!
==== Refs
Linney AD Deng J Three-dimensional morphometry in ultrasound Proc Inst Mech Eng 1999 213 235 45 10.1243/0954411991534942
King DL King DL JrShao MY Three-dimensional spatial registration and interactive display of position and orientation of real-time ultrasound images J Ultrasound Med 1990 9 525 32 2214012
Brunner M Obruca A Bauer P Feichtinger W Clinical application of volume estimation based on three-dimensional ultrasonography Ultrasound Obstet Gynecol 1995 6 358 61 8590209 10.1046/j.1469-0705.1995.06050358.x
Maymon R Herman A Ariely S Dreazen E Buckovsky I Weinraub Z Three-dimensional vaginal sonography in obstetrics and gynaecology Hum Reprod Update 2000 6 475 84 11045878 10.1093/humupd/6.5.475
Heer IM Strauss A Muller-Egloff S Hasbargen U Telemedicine in ultrasound: new solutions Ultrasound Med Biol 2001 27 1239 43 11597365 10.1016/S0301-5629(01)00420-3
Pretorius DH Nelson TR Three-dimensional ultrasound Ultrasound Obstet Gynecol 1995 5 219 21 7600200 10.1046/j.1469-0705.1995.05040219.x
Gilja OH Hausken T Berstad A Odegaard S Measurements of organ volume by ultrasonography Proc Inst Mech Eng [H] 1999 213 247 59 10420778 10.1243/0954411991534951
Kyei-Mensah A Maconochie N Zaidi J Pittrof R Campbell S Tan SL Transvaginal three-dimensional ultrasound: reproducibility of ovarian and endometrial volume measurements Fertil Steril 1996 66 718 22 8893673
Raine-Fenning NJ Campbell BK Collier J Brincat MB Johnson IR The reproducibility of endometrial volume acquisition and measurement with the VOCAL-imaging program Ultrasound Obstet Gynecol 2002 19 69 75 11851972 10.1046/j.0960-7692.2001.00608.x
Farrell T Leslie JR Chien PF Agustsson P The reliability and validity of three dimensional ultrasound volumetric measurements using an in vitro balloon and in vivo uterine model Bjog 2001 108 573 82 11426890 10.1016/S0306-5456(00)00148-0
Riccabona M Nelson TR Pretorius DH Three-dimensional ultrasound: accuracy of distance and volume measurements Ultrasound Obstet Gynecol 1996 7 429 34 8807760 10.1046/j.1469-0705.1996.07060429.x
Raine-Fenning NJ Clewes JS Kendall NR Bunkheila AK Campbell BK Johnson IR The interobserver reliability and validity of volume calculation from three-dimensional ultrasound datasets in the in vitro setting Ultrasound Obstet Gynecol 2003 21 283 91 12666225 10.1002/uog.61
Riccabona M Nelson TR Pretorius DH Davidson TE In vivo three-dimensional sonographic measurement of organ volume: validation in the urinary bladder J Ultrasound Med 1996 15 627 32 8866444
Kyei-Mensah A Zaidi J Pittrof R Shaker A Campbell S Tan SL Transvaginal three-dimensional ultrasound: accuracy of follicular volume measurements Fertil Steril 1996 65 371 6 8566265
Amer A Hammadeh ME Kolkailah M Ghandour AA Three-dimensional versus two-dimensional ultrasound measurement of follicular volume: are they comparable? Arch Gynecol Obstet 2003 268 155 7 12774236 10.1007/s00404-002-0315-6
Pairleitner H Steiner H Hasenoehrl G Staudach A Three-dimensional power Doppler sonography: imaging and quantifying blood flow and vascularization Ultrasound Obstet Gynecol 1999 14 139 43 10492874 10.1046/j.1469-0705.1999.14020139.x
Raine-Fenning NJ Campbell BK Clewes JS Kendall NR Johnson IR The reliability of the semi-quantification of ovarian, endometrial and sub-endometrial perfusion using three-dimensional power Doppler angiography and shell-imaging Ultrasound Obstet Gynecol 2003 22 633 639 14689538 10.1002/uog.923
Jarvela IY Sladkevicius P Tekay AH Campbell S Nargund G Intraobserver and interobserver variability of ovarian volume, gray-scale and color flow indices obtained using transvaginal three-dimensional power Doppler ultrasonography Ultrasound Obstet Gynecol 2003 21 277 82 12666224 10.1002/uog.62
Kupesic S Clinical implications of sonographic detection of uterine anomalies for reproductive outcome Ultrasound Obstet Gynecol 2001 18 387 400 11779002 10.1046/j.0960-7692.2001.00539.x
Homer HA Li TC Cooke ID The septate uterus: a review of management and reproductive outcome Fertil Steril 2000 73 1 14 10632403 10.1016/S0015-0282(99)00480-X
Jurkovic D Geipel A Gruboeck K Jauniaux E Natucci M Campbell S Three-dimensional ultrasound for the assessment of uterine anatomy and detection of congenital anomalies: a comparison with hysterosalpingography and two-dimensional sonography Ultrasound Obstet Gynecol 1995 5 233 7 7600203 10.1046/j.1469-0705.1995.05040233.x
Raga F Bonilla-Musoles F Blanes J Osborne NG Congenital Mullerian anomalies: diagnostic accuracy of three-dimensional ultrasound Fertil Steril 1996 65 523 8 8774280
Wu MH Hsu CC Huang KE Detection of congenital mullerian duct anomalies using three-dimensional ultrasound J Clin Ultrasound 1997 25 487 92 9350567 10.1002/(SICI)1097-0096(199711/12)25:9<487::AID-JCU4>3.0.CO;2-J
Woelfer B Salim R Banerjee S Elson J Regan L Jurkovic D Reproductive outcomes in women with congenital uterine anomalies detected by three-dimensional ultrasound screening Obstet Gynecol 2001 98 1099 103 11755560 10.1016/S0029-7844(01)01599-X
Jurkovic D Gruboeck K Tailor A Nicolaides KH Ultrasound screening for congenital uterine anomalies Br J Obstet Gynaecol 1997 104 1320 1 9386036
Salim R Regan L Woelfer B Backos M Jurkovic D A comparative study of the morphology of congenital uterine anomalies in women with and without a history of recurrent first trimester miscarriage Hum Reprod 2003 18 162 6 12525460 10.1093/humrep/deg030
Salim R Woelfer B Backos M Regan L Jurkovic D Reproducibility of three-dimensional ultrasound diagnosis of congenital uterine anomalies Ultrasound Obstet Gynecol 2003 21 578 82 12808675 10.1002/uog.127
La Torre R De Felice C De Angelis C Coacci F Mastrone M Cosmi EV Transvaginal sonographic evaluation of endometrial polyps: a comparison with two dimensional and three dimensional contrast sonography Clin Exp Obstet Gynecol 1999 26 171 3 10668146
Sylvestre C Child TJ Tulandi T Tan SL A prospective study to evaluate the efficacy of two- and three-dimensional sonohysterography in women with intrauterine lesions Fertil Steril 2003 79 1222 5 12738522 10.1016/S0015-0282(03)00154-7
Kiyokawa K Masuda H Fuyuki T Koseki M Uchida N Fukuda T Three-dimensional hysterosalpingo-contrast sonography (3D-HyCoSy) as an outpatient procedure to assess infertile women: a pilot study Ultrasound Obstet Gynecol 2000 16 648 54 11169373 10.1046/j.1469-0705.2000.00327.x
Sladkevicius P Ojha K Campbell S Nargund G Three-dimensional power Doppler imaging in the assessment of Fallopian tube patency Ultrasound Obstet Gynecol 2000 16 644 7 11169372 10.1046/j.1469-0705.2000.00302.x
Sankpal RS Confino E Matzel A Cohen LS Investigation of the uterine cavity and fallopian tubes using three-dimensional saline sonohysterosalpingography Int J Gynecol Obstet 2001 73 125 9 10.1016/S0020-7292(01)00363-0
Jarvela IY Mason HD Sladkevicius P Kelly S Ojha K Campbell S Characterization of normal and polycystic ovaries using three-dimensional power Doppler ultrasonography J Assist Reprod Genet 2002 19 582 90 12503891 10.1023/A:1021267200316
Wu MH Tang HH Hsu CC Wang ST Huang KE The role of three-dimensional ultrasonographic images in ovarian measurement Fertil Steril 1998 69 1152 5 9627309 10.1016/S0015-0282(98)00092-2
Dolz M Osborne NG Blanes J Raga F Abad-Velasco L Villalobos A Polycystic ovarian syndrome: assessment with color Doppler angiography and three-dimensional ultrasonography J Ultrasound Med 1999 18 303 13 10206219
Pan HA Wu MH Cheng YC Li CH Chang FM Quantification of Doppler signal in polycystic ovary syndrome using three-dimensional power Doppler ultrasonography: a possible new marker for diagnosis Hum Reprod 2002 17 201 6 11756388 10.1093/humrep/17.1.201
Kyei-Mensah AA LinTan S Zaidi J Jacobs HS Relationship of ovarian stromal volume to serum androgen concentrations in patients with polycystic ovary syndrome Hum Reprod 1998 13 1437 41 9688367 10.1093/humrep/13.6.1437
Nardo LG Buckett WM White D Digesu AG Franks S Khullar V Three-dimensional assessment of ultrasound features in women with clomiphene citrate-resistant polycystic ovarian syndrome (PCOS): ovarian stromal volume does not correlate with biochemical indices Hum Reprod 2002 17 1052 5 11925404 10.1093/humrep/17.4.1052
Pellicer A Ardiles G Neuspiller F Remohi J Simon C Bonilla-Musoles F Evaluation of the ovarian reserve in young low responders with normal basal levels of follicle-stimulating hormone using three-dimensional ultrasonography Fertil Steril 1998 70 671 5 9797096 10.1016/S0015-0282(98)00268-4
Pohl M Hohlagschwandtner M Obruca A Poschalko G Weigert M Feichtinger W Number and size of antral follicles as predictive factors in vitro fertilization and embryo transfer J Assist Reprod Genet 2000 17 315 8 11042827 10.1023/A:1009448810413
Kupesic S Kurjak A Predictors of IVF outcome by three-dimensional ultrasound Hum Reprod 2002 17 950 5 11925388 10.1093/humrep/17.4.950
Schild RL Knobloch C Dorn C Fimmers R van der Ven H Hansmann M The role of ovarian volume in an in vitro fertilization programme as assessed by 3D ultrasound Arch Gynecol Obstet 2001 265 67 72 11409477 10.1007/s004040000138
Scheffer GJ Broekmans FJ Bancsi LF Habbema JD Looman CW Te Velde ER Quantitative transvaginal two- and three-dimensional sonography of the ovaries: reproducibility of antral follicle counts Ultrasound Obstet Gynecol 2002 20 270 5 12230451 10.1046/j.1469-0705.2002.00787.x
Jarvela IY Sladkevicius P Kelly S Ojha K Campbell S Nargund G Quantification of ovarian power Doppler signal with three-dimensional ultrasonography to predict response during in vitro fertilization Obstet Gynecol 2003 102 816 22 14551013 10.1016/S0029-7844(03)00693-8
Kupesic S Kurjak A Bjelos D Vujisic S Three-dimensional ultrasonographic ovarian measurements and in vitro fertilization outcome are related to age Fertil Steril 2003 79 190 7 12524087 10.1016/S0015-0282(02)04567-3
Schild RL Indefrei D Eschweiler S Van der Ven H Fimmers R Hansmann M Three-dimensional endometrial volume calculation and pregnancy rate in an in-vitro fertilization programme Hum Reprod 1999 14 1255 8 10325273 10.1093/humrep/14.5.1255
Yaman C Ebner T Sommergruber M Polz W Tews G Role of three-dimensional ultrasonographic measurement of endometrium volume as a predictor of pregnancy outcome in an IVF-ET program: a preliminary study Fertil Steril 2000 74 797 801 11020526 10.1016/S0015-0282(00)01493-X
Raga F Bonilla-Musoles F Casan EM Klein O Bonilla F Assessment of endometrial volume by three-dimensional ultrasound prior to embryo transfer: clues to endometrial receptivity Hum Reprod 1999 14 2851 4 10548635 10.1093/humrep/14.11.2851
Kupesic S Bekavac I Bjelos D Kurjak A Assessment of endometrial receptivity by transvaginal color Doppler and three-dimensional power Doppler ultrasonography in patients undergoing in vitro fertilization procedures J Ultrasound Med 2001 20 125 34 11211132
Wu HM Chiang CH Huang HY Chao AS Wang HS Soong YK Detection of the subendometrial vascularization flow index by three-dimensional ultrasound may be useful for predicting the pregnancy rate for patients undergoing in vitro fertilization-embryo transfer Fertil Steril 2003 79 507 11 12620431 10.1016/S0015-0282(02)04698-8
Schild RL Holthaus S d'Alquen J Fimmers R Dorn C van Der Ven H Quantitative assessment of subendometrial blood flow by three-dimensional-ultrasound is an important predictive factor of implantation in an in-vitro fertilization programme Hum Reprod 2000 15 89 94 10611195 10.1093/humrep/15.1.89
Steiner H Gregg AR Bogner G Graf AH Weiner CP Staudach A First trimester three-dimensional ultrasound volumetry of the gestational sac Arch Gynecol Obstet 1994 255 165 70 7695361 10.1007/s004040050047
Acharya G Morgan H Does gestational sac volume predict the outcome of missed miscarriage managed expectantly? J Clin Ultrasound 2002 30 526 31 12404517 10.1002/jcu.10107
Babinszki A Nyari T Jordan S Nasseri A Mukherjee T Copperman AB Three-dimensional measurement of gestational and yolk sac volumes as predictors of pregnancy outcome in the first trimester Am J Perinatol 2001 18 203 11 11444364 10.1055/s-2001-15499
Rempen A The shape of the endometrium evaluated with three-dimensional ultrasound: an additional predictor of extrauterine pregnancy Hum Reprod 1998 13 450 4 9557855 10.1093/humrep/13.2.450
Su YN Shih JC Chiu WH Lee CN Cheng WF Hsieh FJ Cervical pregnancy: assessment with three-dimensional power Doppler imaging and successful management with selective uterine artery embolization Ultrasound Obstet Gynecol 1999 14 284 7 10586481 10.1046/j.1469-0705.1999.14040284.x
|
16095530
|
PMC1208937
|
CC BY
|
2021-01-04 16:40:17
|
no
|
J Exp Clin Assist Reprod. 2005 Aug 11; 2:10
|
utf-8
|
J Exp Clin Assist Reprod
| 2,005 |
10.1186/1743-1050-2-10
|
oa_comm
|
==== Front
J Immune Based Ther VaccinesJournal of Immune Based Therapies and Vaccines1476-8518BioMed Central London 1476-8518-3-61613892910.1186/1476-8518-3-6ReviewPsoriatic arthritis: Pathogenesis and novel immunomodulatory approaches to treatment Cassell Sarah [email protected] Arthur [email protected] Center for Innovative Therapy, Division of Rheumatology, Allergy, and Immunology, The University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0943, USA2005 2 9 2005 3 6 6 5 7 2005 2 9 2005 Copyright © 2005 Cassell and Kavanaugh; licensee BioMed Central Ltd.2005Cassell and Kavanaugh; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Psoriatic arthritis (PsA) is a chronic inflammatory arthropathy characterized by the association of arthritis and psoriasis. PsA runs a variable course, from mild synovitis to severe, progressive, erosive arthropathy. The pathogenesis of PsA involves alteration in the components of the immune response, although the exact cause of PsA is unknown. A number of patients with severe peripheral arthritis fail to respond to standard conventional therapy. Advances in biotechnology and in our understanding of the immunopathogenesis of PsA have led to great interest and progress in regards to biologic treatments for PsA. Notable success achieved with recently introduced biologic therapies has paved the way for further research and develpoment of additional therapies that should improve outcomes for affected patients.
==== Body
Introduction
Psoriatic arthritis (PsA) is a chronic inflammatory arthropathy characterized by the association of arthritis and psoriasis. Joint involvement is heterogeneous, and may consist of spondyloarthropathy, as well as oligoarticular and polyarticular peripheral arthritis. PsA runs a variable course, from mild synovitis to severe, progressive, erosive arthropathy. PsA is classified as one of the subtypes of spondyloarthropathy, sharing clinical features such as asymmetric joint involvement, an oligoarticular arthritis pattern, a similar frequency in men and women, the common occurrence of enthesitis and dactylitis, infrequent rheumatoid factor and anti-cyclic-citrullinated-peptide seropositivity, and extra-articular manifestations such as iritis.
Epidemiology
Psoriasis occurs in about 2% of the population [1]. PsA has been reported in 7% to 42% of patients with psoriasis [2]. The prevalence of PsA in the US has been estimated as 0.67% [3]. However, estimates of prevalence are variable, due in part to the heterogeneity of the disease as well as a lack of validated diagnostic criteria [4].
In general, skin involvement precedes joint disease, often by years. However, PsA precedes skin psoriasis in about 15% of patients, and the two occur simultaneously in about 20%. Some reports suggest that PsA is more common in patients with severe psoriasis [5,6]. A recent study suggested a correlation between the extent of skin and joint severity only among patients with simultaneous onset of skin and joint manifestations [7].
Pathogenesis
The exact cause of PsA is unknown, although genetic, environmental, and immunologic factors clearly play important roles. The pathogenic connection between psoriasis and arthritis is not clear, although both are immunologically mediated.
Genetic factors
Most studies document a familial predisposition to both psoriasis and PsA. More than 40% of patients with PsA have first degree family members with either skin or joint disease [8,9]. Several genetic susceptibility loci have been proposed, with the strongest effect residing within the major histocompatibility complex (MHC). Population studies in PsA have shown increased frequency of HLA-B13, B17, B27, B38, B39, DR4 and DR7 [8,10,11]. In a comparison of 158 patients with PsA to 101 patients with uncomplicated psoriasis, HLA-B7 and B27 were more common among patients with PsA, whereas B17, Cw6 and DR7 were more common among patients with uncomplicated psoriasis [8]. Some of these associations may be confounded by linkage disequilibrium. HLA-B27 has been associated with spinal disease in which radiological sacroiliitis is present. A symmetric pattern of peripheral PsA appears related to HLA-DR4 [8,12]. The strongest susceptibility locus for psoriasis is on chromosome 6p, termed PSORS1 [1,13-16]. Other psoriasis susceptibility loci are located on chromosomes 17q25 (PSORS2), 4q34 (PSORS3), 1q (PSORS4), 3q21 (PSORS5), 19p13 (PSORS6), 1p (PSORS7), and 17q25 (RUNX1) [1].
Other genes within the MHC region and non-HLA associations have been explored. A TNF-α promoter polymorphism or a gene in linkage disequilibrium with TNF-α may predispose or increase susceptibility to psoriasis and PsA [17]. One study looking at the TNFβ+252 and TNFα-308 polymorphisms did not find the alleles more frequently in PsA patients than matched controls, but did find both alleles were significantly associated with the presence of joint erosions and the progression of joint erosions in early PsA [18]. A meta-analysis showed the TNFα-238 variant in Caucasian PsA patients was a significant risk factor for PsA [19]. A recent study pointed to Cw6 and MHC class I chain-related A (MICA)-A9 as being the strongest genetic susceptibility factors for PsA [20].
Environmental factors – infection, trauma
Both viral and bacterial infections have been implicated as causative agents in PsA. Support for the role of bacterial antigens in the pathogenesis of psoriasis and PsA comes from indirect observation of enhanced humoral and cellular immunity to gram-positive bacteria typically found in the psoriatic plaques [21]. However, psoriatic plaques often get secondarily infected, thus the cause-effect relationship of bacteria and psoriasis is difficult to prove. One study of sera from patients with PsA showed higher levels of antibody to streptococcal exotoxin, which provides some evidence of a link between streptococcal infection and articular inflammation [22]. The possibility that PsA might be virally induced has been proposed, although never confirmed [23,24]. Physical trauma may result in the onset of psoriasis (Koebner Phenomenon) and, theoretically, PsA at the sites of injury. This association would highlight potential association between innate and specific immunity. There are a number of case reports suggesting a possible role for trauma in PsA, but this has not been evaluated in a prospective manner.
Immunologic factors
Both psoriasis and PsA are immunologically mediated. Characteristic pathologic features of PsA are synovial lining layer cell hyperplasia, inflammatory cell accumulation and prominent vascularity. T lymphocytes, particularly CD8+ cells, may play important pathogenic roles. Activated T cells have been noted in affected tissue, both skin and joint [25,26]. A predominance of CD8+ T lymphocytes with clonal expansion have been found in PsA synovial fluid leading to the proposal that CD8+ T cells drive the immune response [27]. This is further supported by the fact that CD8+ T cells also dominate the infiltrate at marrow sites adjacent to entheseal inflammation, an early area of involvement [28]. An analysis of T cell receptor beta chain variable (TCRβV) gene repertoires revealed common expansions in both skin and synovial inflammatory sites, suggesting an important role for cognate T cell responses in the pathogenesis of PsA and that the inciting antigen may be identical or homologous between afflicted skin and synovium [29].
The cytokine network in the psoriatic skin and synovium is dominated by monocyte and T-cell derived cytokines: IL-1β, IL-2, IL-10, IFN-γ and TNF-α [30]. In PsA synovium, higher levels of IFN-γ, IL-2 and IL-10 have been detected than in psoriatic skin. One study of cytokine staining in PsA synovium showed IL-1α, IL-1β, IL-8, IL-15, IFN-γ and TNF-α staining localized to the lining layer and perivascular macrophages [31]. These cytokines can induce proliferation and activation of synovial and epidermal fibroblasts, leading to fibrosis in patients with longstanding PsA. TNF-α, a key proinflammatory cytokine, induces the production of other inflammatory cytokines such as IL-1, IL-6, and granulocyte-macrophage colony-stimulating factor, chemokines such as IL-6, degradative enzymes such as several matrix metalloproteinases (MMPs) and other factors. TNF-α mediates a number of biological processes that can result in joint damage including stimulation of bone resorption, inhibition of bone formation, and inhibition of synthesis of proteoglycans [32,33]. Angiogenic factors such as TNF-α and vascular endothelial growth factor (VEGF) may contribute to vascular proliferation [34,35].
While the mechanisms governing psoriatic skin and joint involvement are similar, there are distinctions. For example, cutaneous lymphocyte associated (CLA) antigen, an adhesion molecule that identifies lymphocytes that preferentially traffic to the skin, is upregulated on lymphocytes in psoriatic skin but is minimally expressed on cells in the PsA synovium [36].
Clinical Features
Wright and Moll recognized several patterns of PsA: isolated distal interphalangeal disease, peripheral oligoarthritis, peripheral polyarthritis, and spondyloarthropathy. These clinical phenotypes are not fixed but are interchangeable, and individual patients can switch phenotypes [37]. The most important distinction as regards outcome appears to be oligo- versus poly-articular joint involvement.
Extra-articular manifestations of PsA are important aspects of the disease, the most common is the psoriatic skin lesion, which may affect all areas of the skin. Dactylitis is typical in PsA and presents as inflammation of the whole digit, joints and tendon sheaths. Enthesitis, inflammation at the site of tendon, ligament or synovial membrane insertion into bone, is characteristic of PsA and may represent the earliest site of involvement. Other extra-articular manifestations include the presence of iritis, mouth ulcers, and urethritis.
PsA has several characteristic radiographic features which include lack of periarticular osteopenia, destruction of interphalangeal joints with widening of the joint spaces, pencil-in-cup changes in the hands and feet, ankylosis, periosteal reaction, and spur formation [38].
The course of PsA is variable. Patients who have five or more involved joints at presentation are more likely to have progressive disease. Some patients have few episodes and completely recover, but recent studies demonstrated that many patients have persistent and severe courses [39-41]. Damage in PsA occurs early and progresses over time, with increasing deformities and limitation of daily activity [42]. Patients with PsA have increased mortality compared to the general population. More severe disease, as manifested by higher ESR and radiologic scores at presentation, is a predictive factor of mortality [43].
Treatment of PsA
Conventional treatment
Mild joint symptoms may respond to non-steroidal anti-inflammatory drugs (NSAIDs) [42]. Systemic steroids can be used, but may cause side effects and rebound worsening of psoriasis [44]. Patients who are unresponsive to NSAID therapy or who have progressive disease may require disease modifying anti-rheumatic drugs (DMARDs) (eg methotrexate [MTX], leflunomide, sulfasalazine [SSZ], cyclosporine [CsA]).
MTX is considered by many rheumatologists the DMARD of choice because of its remarkable efficacy in ameliorating both skin and joint disease, its rapid onset, and its acceptable safety profile [45,46]. However, 16–30% of patients treated with MTX discontinue it because of toxicity [47,48]. Leflunomide, an antipyrimidine drug that interferes with T-cell activation, has been shown to be effective in improving both joint and skin symptoms [49]. The most common side effects seen with leflunomide are diarrhea and increased transaminases. SSZ has been shown to be helpful for peripheral arthritis but not for axial disease [50]. CSA improves both joint inflammation and skin lesions in PsA, but is not frequently used because of its toxicities, the most worrisome being hypertension and nephrotoxicity [48,51]. Likewise, gold compounds and other drugs have been reported to ameliorate arthritis in some PsA patients, but are rarely used secondary to side effects and toxicities.
Biologic Agents
In recent years, greater understanding of immunopathology and advances in biotechnology facilitating the ability to design and produce novel biologic agents have led to exciting breakthroughs in the treatment of autoimmune disease, including psoriasis and PsA [52]. The development of novel biologic agents has been further encouraged by the unmet need for better treatments and the positive results with their use in other autoimmune diseases, particularly rheumatoid arthritis (RA). The most significant experience of the use of biologics in treatment of PsA is with TNF-α inhibitors.
Tumor necrosis factor-alpha (TNF-α) inhibitors
Given its pro-inflammatory potential and its elevated levels in RA and PsA, TNF-α was identified as an attractive target for biologic therapies. TNF-α inhibitors have been used with great success to suppress joint inflammation in RA, inducing not only marked improvement in the signs and symptoms of disease, but also substantially improved functional status and quality of life [53-55]. Additionally, they have been shown to attenuate the progression of radiographic joint damage. Adverse effects have been reported, but in general these agents are well-tolerated. These encouraging results spurred interest in using TNF-α inhibitors in PsA. Currently there are three TNF-α inhibitors available: 1) etanercept, a fusion protein consisting of a dimer of the extracellular portion of the type II TNF receptor (p75) linked to the Fc portion of IgG1, 2) infliximab, a chimeric monoclonal antibody specific for TNF-α, and 3) adalumimab, a human monoclonal antibody specific for TNF-α.
1. Etanercept
Etanercept has been proven effective for the treatment of PsA [56,57]. The first double-blind, placebo controlled clinical trial of etanercept in PsA was in 60 patients with long-standing disease. The etanercept group showed significant improvement in all measures of disease activity compared with the placebo group at 12 weeks. The primary endpoint for arthritis activity, the Psoriatic Arthritis Response Criteria (PsARC), a composite index, was achieved by 87% versus 23% of the etanercept and placebo groups respectively [58]. A secondary endpoint was the American College of Rheumatology composite response criteria (ACR), a score based on 20%, 50%, or 70% improvement [59]. ACR20 responses were 73% and 13% in the etanercept and placebo groups respectively [56]. For psoriasis, the primary endpoint was 75% improvement in the Psoriasis Area and Severity (PASI) score (PASI75). Of patients with >3% body surface involvement, 26% of etanercept treated patients achieved PASI75 versus none in the placebo treated group [56]. Disability, as assessed by responses on the health assessment questionnaire (HAQ), significantly improved in the etanercept group. An open-label extension of this study revealed sustained efficacy in joints, further improvement of skin disease, ability to decrease or discontinue concomitant methotrexate and prednisone, and continued tolerability [60].
Another phase III clinical trial of etanercept in 205 patients with PsA confirmed and extended earlier findings. ACR20 response rates were achieved by 59% of the etanercept group and 15% of the placebo group at 12 weeks (P < 0.001). This clinical response was sustained for 24 weeks. Of those meeting criteria for PASI evaluation, the etanercept group showed on average 47% improvement compared to no improvement in the placebo group (P < 0.001) [57].
Etanercept has been observed to slow and halt radiographic structural damage in PsA. A one year study of 205 patients revealed that at twelve months the radiographic disease progression in the etanercept group was inhibited (Sharp score: -.03 units) compared with worsening in the placebo group (Sharp score: +1.00 units) (p = .0001) [61].
2. Infliximab
Open-label studies of infliximab in PsA showed significant decreases in the signs and symptoms of joint inflammation and skin disease [62-64]. This led to double blind, placebo controlled trials, which also revealed positive results [65,66]. The infliximab multinational psoriatic arthritis controlled trial (IMPACT) enrolled 104 patients in a double blind, randomized, placebo-controlled trial for 16 weeks, followed by blinded single-crossover design through 50 weeks [65]. ACR20/50/70 responses at week 16 were 69%/49%/29% in the active treatment group compared to 8%/0%/0% in the placebo group. These results were sustained at 50 weeks with ACR 20/50/70 responses in the infliximab group of 72%/54%/35%. Of the placebo-treated patients who crossed over to active treatment at week 16, ACR20/50/70 responses increased to 77%/49%/30%. This study also assessed dactylitis and enthesitis, two important characteristics of PsA that had not previously been included in clinical trials. Significant improvements were seen in dactylitis and enthesitis with infliximab therapy. Of particular note in this study was the dramatic improvement in skin psoriasis seen with infliximab treatment. Thus, PASI75 was achieved by 12 of 14 infliximab patients whereas there was overall worsening of skin scores in the placebo treated group. This effect was sustained at week 50. Also, 8 of 16 placebo patients who switched to infliximab treatment after week 16 achieved PASI75 at week 50.
These results were confirmed with the subsequent larger phase 3 IMPACT 2 study [67]. In this trial, 200 patients with active PsA were randomized to receive infliximab or placebo for 24 weeks. ACR 20/50/70 scores at week 24 in the infliximab group were 54%/41%/27% and 11%/4%/2% in the placebo group. Again, skin improvement was very impressive, with 60% of the infliximab group achieving PASI75 at week 24, whereas only 1% of the placebo group did. Statistically significant improvements in measures of functional status and quality of life (measured by HAQ and SF-36, respectively) were seen, as were improvements in dactylitis and enthesopathy.
Two studies have shown that infliximab can inhibit radiographic disease progression. In a double-blind, placebo controlled trial of 200 PsA patients (IMPACT2), patients treated with infliximab had significantly less radiographic disease progression at week 24, as measured by the van der Heijde-Sharp method modified for PsA (-0.7 +/- 2.53 versus .82 +/- 2.62, for infliximab versus placebo treated patients respectively; p < 0.001) [68]. An analysis of patients from the IMPACT1 study showed that at 50 weeks, radiographic progression of disease was inhibited in both the group treated with infliximab throughout the trial as well as in the group receiving infliximab from week 16 through week 50 [69].
3. Adalumimab
Adalumimab was assessed in PsA in a phase III, placebo-controlled, double blind study, the Adalimumab Effectiveness in PsA Trial (ADEPT) [70]. 151 patient received adalumimab and 162 received placebo. Adalumimab treated patients showed rapid improvements. At week 24 ACR20, 50, and 70 scores for adalumimab were 57%, 39%, and 23% respectively versus 15%, 6%, and 1% for placebo. PASI50, 75 and 90 scores for adalumimab and placebo respectively were 75%, 59%, and 42% versus 12%, 1%, and 0% [70].
Adalimumab was also shown to inhibit radiographic disease progression. In the ADEPT trial, at week 24 mean change in modified total Sharp Scores (mTSS) was -0.2 in infliximab treated patients compared with +1.0 in placebo treated patients (p <= .001). All patients were allowed to go into an open label extension after week 24. Patients who started in the placebo arm and crossed to the adalimumab open label arm at week 24 had mTSS scores of +1.0 and +1.0 at weeks 24 and 48 respectively, showing no further radiographic progression after they started adalimumab. The patients originally in the adalimumab arm who extended into open label treatment had mTSS scores of -0.2 and 0.1 at weeks 24 and 48 respectively. Assessments at week 48 showed that adalimumab maintained the lack of radiographic change [71].
With all TNF-α inhibitors there have been concerns about safety issues, particularly infections, serious infections and opportunistic infections such as reactivation of latent tuberculosis. Appropriate monitoring for signs and symptoms of infection is required before and during treatment. While other adverse events have been reported at relatively low rates, careful monitoring of patients on these new biologic agents is quite important.
Alefacept
Another biologic agent in development for PsA is alefacept, which was approved in the US for the treatment of psoriasis in 2003. Alefacept is a human LFA-3/IgG1 fusion protein and is under clinical investigation for the treatment of PsA and RA. The LFA-3 portion of alefacept binds to CD2 receptors on T cells to block the natural interaction between LFA-3 on antigen-presenting cells and CD2 on T cells. Blockade of the LFA-3/CD2 interaction, a key co-stimulatory pathway, can inhibit T-cell activation. The IgG1 portion of alefacept can bind to FcγRIII (CD16) IgG receptors on accessory cells (e.g. natural killer cells) and may induce granzyme-mediated apoptosis [52,72].
Alefacept was evaluated as a treatment for psoriasis in multicenter, randomized, placebo-controlled, double blind study. Two hundred twenty-nine patients with chronic psoriasis received intravenous injection of alefacept at different dosages. The mean reduction in the PASI score 12 weeks after treatment was greater in the alefacept groups than the placebo group [73].
A small study suggested that alefacept may also improve both skin and joint symptoms in PsA [74]. In a single center open-label study, 11 patients with PsA received intravenous 7.5 mg alefacept once weekly for 12 weeks. Synovial tissue biopsies of an index joint were obtained by arthroscopy at baseline and at weeks 4 and 12. Clinically, some degree of improvement in arthritis was observed in six patients (55%) at the completion of the treatment. A similar proportion of patients achieved 50% amelioration of skin disease. This study supports the notion that T cell activation plays an important role in chronic inflammatory diseases and effective blockade of the LFA-3/CD2 interaction may be useful for treating PsA.
Additionally, a double blind, placebo controlled trial assessed the combination of alefacept and methotrexate in 185 PsA patients. An ACR20 response was achieved by 54% of the alefacept group versus 24% of the placebo group. 53% of the alefacept group achieved PASI50 compared with 17% of the placebo group [75]. Adverse events in this trial occurring at >5% included: back pain, nasopharyngitis, nausea, URI and increased ALT. There were no serious infections and the serious adverse event rate was 2% [76].
Efalizumab
Leukocyte function associate antigen-1 (LFA-1) is an adhesion molecule expressed on T lymphocytes. It interacts with its ligand, intercellular adhesion molecule (ICAM-1), in ways that may be relevant to the pathogenesis of psoriasis including: stabilizing the binding of antigen-presenting cells to T lymphocytes, facilitating migration of T lymphocytes from circulation into skin, and activation of T lymphocytes [77]. Efalizumab is a humanized monoclonal IgG antibody that binds to the alpha-subunit (CD11) of LFA-1 and prevents LFA-1 binding to ICAM-1. In two recent phase 3, randomized, double-bind, placebo-controlled trials, efaluzimab showed efficacy in treating moderate to severe plaque psoriasis, and was recently approved for this use by the US FDA.
Leonardi et al, reported a study of 498 psoriasis patients that showed PASI75 scores at 12 weeks in the treatment groups were achieved in 32.6% of patients versus 2.4% of placebo-treated patients [77]. The most common adverse events (headache, fever, chills, nausea, and myalgias) were more frequent in the efalizumab-treated group only during the first two injections, and then decreased to rates similar to placebo. A second study randomized 556 psoriasis patients for twelve weeks with continuation in an open label study [78]. At 12 weeks, PASI50/75 were 58.5%/26.6% respectively in efaluzimab-treated patients compared with 13.9%/4.3% in placebo treated patients. These numbers increased at week 24. Patient reported outcomes (dermatology life quality index and itching scale) also improved. Interestingly, during the second twelve weeks there was an increased incidence of arthritis (5.6%); 12 of these 19 cases had had a prior history of arthritis.
Preliminary results from a phase II study of efalizumab in 117 PsA patients showed that treatment did not reach statistical significance as far as achieving an ACR20 reponse at twelve weeks [79].
Other types of biologic agents and future directions
The introduction of TNF-α inhibitors and their tremendous clinical impact has generated considerable interest in exploring other avenues for the treatment of PsA. In addition, it is worth noting that despite the tremendous success achieved in PsA patients treated with TNF-α inhibitors, approximately one-third of patients with moderate to severe PsA have negligible or insufficient responses to such treatment. This has provided the impetus for the development of biologic agents targeting other aspects of the dysregulated immune system. Several promising biologic agents, directed at targets other than TNF-α, are currently under study (Table 1).
Table 1 Biologic agents under consideration for the treatment of Psoriatic arthritis
Suppression of inflammatory mediators
target agent comment
IL-1 Anakinra IL-1 receptor antagonist
IL-8 ABXIL-8 human anti-IL-8 mAb
Modulation of the function of Anti-inflammatory mediators
target agent comment
IL-10 rIL-10 recombinant human Th2 cytokine
IL-11 rIL-11 recombinant human Th2 cytokine
Alteration of T cell number and function interaction
target agent comment
IL-12 anti-IL-12 mAb several in development
CD25 (IL-2 receptor) Daclizumab humanized anti-CD25 mAb
CD2 Alefacept human LFA-3/IgG fusion protein
CD11a (LFA-1) Efalizumab humanized anti-CD11a mAb
TCR/CD3 huOKT3γ1(ala-ala) humanized anti-CD3 mAb
CD80/CD86 IDEC-114 humanized anti-CD80 mAb
CTLA4Ig fusion protein of CTLA-4/Ig
CD40/CD40L IDEC-131 humanized anti-CD154 mAb
mAb, monoclonal antibody; rIL, recombinant interleukin; IL-2R, LFA, leukocyte function associated antigen; TCR, T-cell receptor; CTLA4Ig, cytotoxic T-lymphocyte-associated antigen 4/immunoglobulin
One approach is the targeting of other inflammatory mediators. ABXIL-8 (Abgenix Inc, Fremont, CA), a human anti-IL-8 monoclonal antibody, binds free IL-8 and may deactivate it in the skin. Effects of IL-8 include T cell and neutrophil activation and chemotaxis, as well as keratinocyte proliferation. IL-8 may also play a role in the vascular responses found in psoriasis [80]. A phase II trial in psoriasis showed some improvement in patients' PASI as well as in histological responses [81]. IL-1, many of the activities of which overlap with TNF, has been suggested to be of potential importance in the pathogenesis of joint and other inflammation [82]. Anakinra (IL-1ra) a homologue of the naturally occurring IL-1 receptor antagonist, has been approved for use in moderate to severely active RA. Other IL-1 inhibiting agents are in development. To date there have not been controlled clinical trials of IL-1 inhibitors in PsA.
Another approach that would suppress inflammation involves the therapeutic use of anti-inflammatory cytokines. For example, among its various activities, Il-10 inhibits INF-γ and promotes TH2 biased cytokine secretion. IL-10 is relatively deficient in psoriatic skin, although it is found in high levels in synovium and serum of PsA patients [83]. Recombinant IL-10 (rIL-10) was used in a phase II trial in 14 patients with chronic plaque psoriasis; 71% had more than a 50% reduction of PASI scores [84]. It has also been studied in PsA which showed modest improvements in skin but not articular disease [85]. Recombinant human IL-11 (rhIL-11) has been shown to have anti-inflammatory activity in vitro and in vivo and has been tested in 12 patients with psoriasis. They showed some improvement in PASI scores [86]. However, there are no published reports of it being used in PsA.
Another therapeutic strategy is to target the number or function of immunocompetent cells central to the propagation of the disease. Several therapies have targeted T cells, which have been suggested to play a central role in orchestrating the immune driven inflammation in PsA. Daclizumab, a humanized antibody to the α-subunit of the IL-2 receptor, blocks the binding of IL-2, a vital growth factor for T cells. One trial in 19 psoriatic patients showed IL-2 blockade during the first 4 weeks and variable desaturation after that, which correlated with reversal in disease improvement that had been achieved. Patients with pretreatment PASI score of <36 showed mean reduction in severity by 30% at eight weeks [87]. HuOKT3γ1 (ala-ala), a non-FcR-binding monoclonal antibody to CD3 (a component of the T cell receptor complex), modulates the function of T cells without decreasing their numbers. A phase I/II study in seven patients with PsA showed 6/7 achieving ACR70 responses at 30 days and all seven had transient, dose dependent depletion of T cells [88]. CD28 is a cell-surface protein on mature T cells and binds to two ligands, CD80 and CD86 on antigen-presenting cells. Blocking this interaction results in incomplete T cell activation. CTLA-4, a natural inhibitor of CD28, binds to CD80/86 molecules with high affinity and competes with CD28. CTLA-4-immunoglobulin (CTLA4Ig) was developed to block the CD 28 and CD80/86 interactions. A phase 1 trial in psoriasis patients showed dose dependent improvement in skin involvement [89]. A CTLA-4-Ig construct, abatacept, is in late phase development for the treatment of RA. It will likely be studied in PsA in the near future. IDEC-114, a humanized anti-CD80 monoclonal antibody, has also been developed to block this interaction. A phase I/II trial of IDEC-114 in 35 psoriatic patients showed 40% of patients achieved PASI50 [90]. Finally, therapies directed at inhibiting IL-12, a cytokine central in driving Th1 biased immune responses, are in the early phases of investigation in psoriasis and PsA.
Conclusion
Appreciation of the unmet clinical need for affected patients, greater understanding of the underlying immunopathophysiology of this common autoimmune disease, and progress in biopharmaceutical development have paved the way for the development of novel biologic agents for PsA. Following closely upon the successes achieved in RA, there have been dramatic clinical efficacy achieved with TNF inhibitors. Substantial improvements have been noted not only as far as the signs and symptoms of arthritis, but also in dactylitis and enthesitis and in skin involvement. Moreover, improvements in functional status and quality of life, and attenuation in the progression of radiographic damage have been achieved. Driven by this success, biologic agents targeting other components of the dysregulated immune response that play major roles in pathogenesis of PsA are actively under study. In the foreseeable future, we can expect further exciting development in immunomodulatory therapies for psoriatic arthritis.
==== Refs
Schon M Boeknck WH Psoriasis NEJM 2005 352 1899 1912 15872205 10.1056/NEJMra041320
Gladman DD Psoriatic arthritis Rheum Dis Clin North Am 1998 24 829 44 9891713 10.1016/S0889-857X(05)70044-2
Lawrence RC Hochberg MC Kelsey JL Estimate of the prevalence of selected arthritis and musculoskeletal diseases in the United States J Rheumatol 1989 16 427 41 2746583
Gladman DD Farewell VT Nadeau C Clinical indicators of progression in psoriatic arthritis: multivariate relative risk mode J Rheumatol 1995 22 675 9 7791162
Little H Harvie JN Lester RS Psoriatic arthritis in severe psoriasis Can Med Assoc J 1975 112 317 9 1109748
Leonard DG O'Duffy JD Rogers RS Prospective analysis of psoriatic arthritis in patients hospitalized for psoriasis Mayo Clin Proc 1978 53 511 8 682678
Elkayam O Ophir J Yaron M Caspi D Psoriatic arthritis: interrelationships between skin and joint manifestations related to onset, course and distribution Clin Rheumatol 2000 19 301 5 10941813 10.1007/s100670070005
Gladman DD Anhorn KAB Schachter RK Mervart H HLA antigens in psoriatic arthritis J Rheumatol 1986 13 586 92 3735281
Gladman DD Shuckett R Russell ML Thorne JC Schachter RK Psoriatic arthritis – an analysis of 220 patients Quart J Med 1987 62 127 41 3659255
Espinoza LR Gerber LH, Espinoza Psoriatic arthritis: further epidemiologic and genetic consideration Psoriatic arthritis 1985 Grune & Stratton, Orlando Florida 9 32
Sakkas LI Loqueman N Bird H Vaughan RW Welsh KI Panayi GS HLA class II and T cell receptor gene polymorphisms in psoriatic arthritis and psoriasis J Rheumatol 1990 17 1487 90 2273489
Eastmon CJ Psoriatic arthritis. Genetics and HLA antigens Baillieres Clin Rheumatol 1994 8 263 76 8076387
Tomfohrde J Silverman A Barnes R Gene for familial psoriasis susceptibility mapped to the distal end of human chromosome 17q Science 1994 264 1141 5 8178173
Matthews D Fry L Powels A Evidence that a locus for familial psoriasis maps to chromosome 4q Nat Genet 1996 13 231 3 8841203 10.1038/ng1096-231
Burden A Javed S Bailey M Genetics of psoriasis: paternal inheritance and a locus on chromosome 6p J Invest Dermatol 1998 110 958 60 9620305 10.1046/j.1523-1747.1998.00213.x
Veal CD Clough RL Barber RC Identification of a novel psoriasis susceptibility locus at 1p and evidence of epistasis between PSORS1 and candidate loci J Med Genet 2001 38 7 13 11134234 10.1136/jmg.38.1.7
Hohler T Kruger A Schneider PM A TNF-α promoter polymorphism is associated with juvenile onset psoriasis and psoriatic arthritis J Invest Dermatol 1997 109 562 5 9326391 10.1111/1523-1747.ep12337469
Balding J Kane D Lingstone W Mynett-Johnson L Bresnihan B Smith O FitzGerald O Cytokine Gene Polymorphisms: Associations with Psoriatic Arthritis Susceptibility and Severity Arth & Rheum 2003 48 1408 1413 12746914 10.1002/art.10935
Rahman P Siannis F Butt C Farewell V Peddle L Pellet F Gladman D Meta-Analysis of TNF-Alpha Polymorphisms in Caucasian Psoriatic Arthritis Populations Ann Rheum Dis 2005 64 325 15155371
Gonzalez S Martinez-Borra J Lopez-Vazquez A MICA rather than MICB, TNFA, or HLA-DRB1 is associated with susceptibility to psoriatic arthritis J Rheumatol 2002 29 973 8 12022360
Vasey FB Gerber LH, Espinoza Etiology and pathogenesis of psoriatic arthritis Psoriatic arthritis 1985 Grune & Stratton, Orlando Florida 45 57
Vasey FB Deitz C Fenske NA Possible involvement of group A streptococci in the pathogenesis of psoriatic arthritis J Rheumatol 1982 9 719 22 6757423
Njobvu P McGill P Psoriatic arthritis and human immunodeficiency virus infection in Zambia J Rheumatol 2000 27 1699 702 10914854
Taglione EV Martini ML Galluzzo P Hepatitis C virus infection: prevalence in psoriasis and psoriatic arthritis J Rheumatol 1999 26 370 2 9972971
Panayi G Immunology of psoriasis and psoriatic arthritis Baillieres Clin Rheumatol 1994 8 419 27 8076395
Veale DJ Barnes L Rogers S Fitzgerald O Immunohistochemical markers for arthritis in psoriasis Ann Rheum Dis 1994 53 450 4 7944617
Costello PJ Winchester RJ Curran SA Psoriatic arthritis joint fluids are characterized by CD8 and CD4 T cell clonal expansions that appear antigen driven J Immunol 2001 166 2878 86 11160357
Laloux L Voisin MC Allain J Immunohistological study of enthesis in spondyloarthropathies: comparison in rheumatoid arthritis and osteoarthritis Ann Rheum Dis 2001 60 316 21 11247858 10.1136/ard.60.4.316
Tassiulas I Duncan SR Centola M Clonal characteristics of T cell infiltrates in skin and synovium of patients with psoriatic arthritis Hum Immunol 1999 60 479 91 10408797 10.1016/S0198-8859(99)00034-8
Ritchlin C Haas-Smith SA Hicks D Patterns of cytokine production in psoriatic synovium J Rheumatol 1998 25 1544 52 9712099
Danning CL Illei GG Hitchon C Macrophage-derived cytokine and nuclear factor kappa B p65 expression in synovial membrane and skin of patients with psoriatic arthritis Arthritis Rheum 2000 43 1244 56 10857783 10.1002/1529-0131(200006)43:6<1244::AID-ANR7>3.0.CO;2-2
Saklatvala J Tumor necrosis factor alpha stimulates resorption and inhibits synthesis of proteoglycan in cartilage Nature 1986 322 547 9 3736671 10.1038/322547a0
Bertolini DR Nedwin GE Bringman TS Stimulation of bone resorption and inhibition of bone formation in vitro by human tumor necrosis factors Nature 1986 319 516 8 3511389 10.1038/319516a0
Ettehadi P Greaves MW Wallach D Elevated tumor necrosis factor alpha biological activity in psoriatic skin lesions Clin Exp Immunol 1994 96 146 51 8149659
Fearon U Reece R Smith J Synovial cytokine and growth factor regulation of MMPs/TIMPs: implications for erosions and angiogenesis in early rheumatoid and psoriatic arthritis patients Ann N Y Acad Sci 1999 878 619 21 10415789
Pitzalis C Cauli A Pipitone N Cutaneous lymphocyte antigen-positive T lymphocytes preferentially migrate to the skin but not to the joint in psoriatic arthritis Arthritis Rheum 1996 39 137 45 8546722
McHugh NJ Balachrishnan C Jones SM Progression of peripheral joint disease in psoriatic arthritis: a 5-yr prospective study Rheumatology 2003 42 778 83 12730539 10.1093/rheumatology/keg217
Resnick D Niwayama G Psoriatic arthritis Diagnosis of bone joint disorders 1981 WB Saunder Co, Philadelphia 1103 29
Gladman DD Shuckett R Russell ML Psoriatic arthritis – an analysis of 220 patients Quart J Med 1987 62 127 41 3659255
Jones SM Armas JB Cohen MG Psoriatic arthritis: Outcome of disease subsets and relationship of joint disease to nail and skin disease Br J Rheumatol 1994 33 834 9 8081668
Kane D Stafford L Bresnihan B Fitzgerald O A prospective, clinical and radiological study of early psoriatic arthritis: an early synovitis clinic experience Rheumatology 2003 Oct 1
Pringle F A multidisciplinary approach to psoriatic arthropathy Community Nurse 1999 5 21 2 10513533
Gladman DD Farewell VT Wong K Husted J Mortality studies in psoriatic arthritis: results from a single outpatient center. II. Prognostic indicators for death Arthritis Rheum 1998 41 1103 10 9627021 10.1002/1529-0131(199806)41:6<1103::AID-ART18>3.0.CO;2-N
Christophers E Mrowietz U Freedberg IM, Eisen AZ, Wolff K Psoriasis Fitzpatrick's Dermatology in General Medicine 1999 5 McGraw Hill, New York 495 521
Black RL O'Brian WM Van Scott EJ Methotrexate therapy in psoriatic arthritis: double-blind study on 21 patients JAMA 1964 189 141 5
Willkens RF Williams HJ Ward JR Randomized, double-blind, placebo controlled trial of low-dose pulse methotrexate in psoriatic arthritis Arthritis Rheum 1984 27 376 81 6712754
Marquerie L Flipo RM Grardel B Use of disease-modifying antirheumatic drugs in patients with psoriatic arthritis Joint bone spine 2002 69 275 81 12102274 10.1016/S1297-319X(02)00396-2
Spadaro A Taccari E Mohtadi B Life-table analysis of cyclosporin A treatment in psoriatic arthritis: comparison with other disease-modifying antirheumatic drugs Clin Exp Rheumatol 1997 15 609 14 9444416
Kaltwasser JP Nash P Gladman D Rosen CF Behrens F Jones P Wollenhaupt J Falk F Mease P for the Treatment of Psoriatic Arthritis Study Group Efficacy and Safety of Leflunomide in the Treatment of Psoriatic Arthritis and Psoriasis: A Multinational, Double Blind, Randomized, Placebo Controlled Clinical Trial Arthritis Rheum 2004 50 1939 50 15188371 10.1002/art.20253
Clegg DO Reda DJ Mejias E Comparison of sulfasalazine and placebo in the treatment of psoriatic arthritis. A Department of Veterans Affairs Cooperative Study Arthritis Rheum 1996 39 2013 20 8961906
Ellis CN Fradin MS Messana JM Cyclosporine for plaque-type psoriasis: results of a multidose, double-blind trial N Engl J Med 1991 324 277 84 1986287
Braun J Sieper J Role of novel biological therapies in psoriatic arthritis Biodrugs 2003 17 187 99 12749755
Kavanaugh A Cohen S Cush J The evolving use of TNF inhibitors in rheumatoid arthritis J Rheumatol 2004 31 1881 4 15468347
Moreland LW Cohen SB Baumgartner SW Long-term safety and efficacy of etanercept in patients with rheumatoid arthritis J Rheumatol 2001 28 1238 44 11409115
Bondeson J Maini RN Tumor necrosis factor as a therapeutic target in rheumatoid arthritis and other chronic inflammatory diseases: the clinical experience with infliximab (REMICADE) Int J Clin Pract 2001 55 211 16 11351775
Mease P Goffe BS Metz J Etanercept in the treatment of psoriatic arthritis and psoriasis: a randomized trial Lancet 2000 356 385 90 10972371 10.1016/S0140-6736(00)02530-7
Mease P Kivitz A Burch F Improvement in disease activity in patients with psoriatic arthritis receiving etanercept (Enbrel): results of a phase 3 multicenter clinical trial (abstract) Arthritis Rheum 2001 44 S90
Clegg DO Reda DJ Mejias E Comparison of sulfasalazine and placebo in the treatment of psoriatic arthritis Arthritis Rheum 1996 39 2013 20 8961906
Felson DT Andersen JJ Boers M American College of Rheumatology preliminary definition of improvement in rheumatoid arthritis Arthritis Rheum 1995 38 727 35 7779114
Mease PJ Goffe BS Metz J Enbrel (etanercept) in patients with psoriatic arthritis and psoriasis (poster) Ann Rheum Dis 2001 60 146 11156548
Mease PJ Kivitz AJ Burch FX Siegel EL Cohen SB Ory P Salonen D Rubenstein J Sharp JT Tsuji W Etanercept Treatment of Psoriatic Arthritis: Safety, Efficacy, and Effect on Disease Progression Arthritis Rheum 2004 50 2264 72 15248226 10.1002/art.20335
Salvarani C Cantini F Olivieri I Efficacy of infliximab in resistant psoriatic arthritis Arthritis Rheum 2003 49 541 5 12910561 10.1002/art.11201
Ogilvie AL Antoni C Dechant C Treatment of psoriatic arthritis with antitumour necrosis factor-alpha antibody clears skin lesions of psoriasis resistant to treatment with methotrexate Br J Dermatol 2001 144 587 9 11260020 10.1046/j.1365-2133.2001.04089.x
Van den Bosch F Kruithof E Baeten D Effects of a loading dose regimen of three infusions of chimeric monoclonal antibody to tumor necrosis factor α (infliximab) in spondyloarthropathy: an open pilot study Ann Rheum Dis 2000 59 428 33 10834859 10.1136/ard.59.6.428
Antoni C Kavanaugh A Kirkham B The one-year results of the infliximab multinational psoriatic arthritis controlled trial Arthritis Rheum 2003 ACR abstract S604
Van den Bosch F Kruithof E Baeten D Randomized double-blind comparison of chimeric monoclonal antibody to tumor necrosis factor alpha (infliximab) versus placebo in active spondyloarthropathy Arthritis Rheum 2002 46 755 65 11920412 10.1002/art.511
Antoni Krueger De Vlam Birbara Beutler Guzzo Zhou Dooley Kavanaugh Infliximab Improves Signs and Symptoms of Psoriatic Arthritis: Results of the IMPACT 2 Trial Ann Rheum Dis 2005 000 1 8
van der Heijde D Kavanaugh A Beutler A Guzzo C Zhou B Dooley L Antoni CE Krueger GG Gladman D Infliximab Inhibits Progression of Radiographic Damage in Patients With Active Psoriatic Arthritis: Results for Impact 2 Trial Ann Rheum Dis 2005 64 109
Antoni CE Kavanaugh A Gladman D Wassenberg S Zhou B Beutler A Bermester G Furst DE Weisman M Ebner W Kalden JR Smolen J van der Heijde D Ann Rheum Dis 2005 64 107
Mease P Gladman D Ritchlin C Ruderman E Steinfeld S Choy E Perdok R Weinberg M Adalimumab Therapy in Patients with Psoriatic Arthritis: 24-week Results of a Phase III Study Arthritis Rheum 2004 4097
Mease PJ Sharp JI Ory P Gladman DD Ritchlin CT Choy EH Weinberg EH Adalimumab Treatment Effects on Radiographic Progression of Joint Disease in Patients with Psoriatic Arthritis: Result from ADEPT Ann Rheum Dis 2005 64 320
Weinberg JM An overview of infliximab, etanercept, efalizumab, and alefacept as biologic therapy for psoriasis Clin Ther 2003 25 2487 505 14667953 10.1016/S0149-2918(03)80313-2
Ellis CN Krueger GG Treatment of chronic plaque psoriasis by selective targeting of memory effector T lymphocytes N Engl J Med 2001 345 248 55 11474662 10.1056/NEJM200107263450403
Kraan MC van Kuijk AWR Dinant HJ Alefacept treatment in psoriatic arthritis. Reduction of the effector T cell population in peripheral blood and synovial tissue is associated with improvement of clinical signs of arthritis Arthritis Rheum 2002 46 2776 84 12384938 10.1002/art.10543
Mease P Gladman D Keystone E Efficacy of Alefacept in Combination with Methotrexate in the Treatment of Psoriatic Arthritis Ann Rheum Dis 2005 64 324
Gladman D Mease P Keystone E Safety of Alefacept in Combination with Methotrexate in the Treatment of Psoriatic Arthritis Ann Rheum Dis 2005 64 324 325
Leonardi Papp Gordon Menter Feldman Cato Walicke Compton Gottlieb Extended efalizumab therapy improves chronic plaque psoriasis: Results from a randomized phase III trial J Am Acad Dermatol 2005 52 425 433 15761420 10.1016/j.jaad.2004.09.029
Menter Gordon Carey Hamilton Glazer Caro Li Gulliver Efficacy and Safety Observed During 24 Weeks of Efalizumab Therapy in Patients with Moderate to Severe Plaque Psoriasis Arch Dermatol 2005 141 31 38 15655139 10.1001/archderm.141.1.31
Barker JN Jones ML Mitra RS Modulation of keratinocyte-derived interleukin-8 which is chemotactic for neutrophils and T lymphocytes Am J Pathol 1991 139 869 876 1681733
Singri P West DP Gordon KB Patterns of cytokine production in psoriatic synovium Arch Dermatol 2002 138 657 663 12020229 10.1001/archderm.138.5.657
Bresnihan B Cunnane G Interleukin-1 receptor antagonist Rheum Dis Clin North Am 1998 24 187 199 10.1016/S0889-857X(05)70029-6
Ritchlin C Haas-Smith SA Hicks D Patterns of cytokine production in psoriatic synovium J Rheum 1998 25 1544 1552 9712099
Reich K Garbe C Blaschke V Response of psoriasis to interleukin-10 is associated with suppression of cutaneous type 1 inflammation, downregulation of epidermal interleukin-8/CXCR2 pathway and normalization of keratinocyte maturation J Invest Dermatol 2001 116 319 329 11180010 10.1046/j.1523-1747.2001.01248.x
McInnes IB Illei GG Danning CL IL-10 improves skin disease and modulates endothelial activation and leukocyte effector function in patients with psoriatic arthritis J Immunol 2001 167 2161 2168
Trepicchio WL Ozawa M Walter IB Interleukin-11 therapy selectively downregulates type I cytokine proinflammatory pathways in psoriasis lesions J Clin Invest 1999 104 1527 1537 10587516
Krueger JG Walters IB Miyazawa M Successful in vivo blockade of CD25 (high-affinity interleukin 2 receptor) on T cells by administration of humanized anti-Tac antibody to patients with psoriasis J Am Acad Dermatol 2000 43 448 458 10954656 10.1067/mjd.2000.106515
Utset TO Auger JA Peace D Modified anti-CD3 therapy in psoriatic arthritis: a phase I/II clinical trial J Rheum 2002 29 1907 1913 12233885
Abrams JR Kelley SL Hayes E Blockade of T lymphocyte costimulation with cytotoxic T lymphocyte-associated antigen 4-Immunoglobulin (CTLA4Ig) reverses the cellular pathology of psoriatic plaques, including the activation of keratinocytes, dendritic cells, and endothelial cells J Exp Med 2000 192 681 693 10974034 10.1084/jem.192.5.681
Gottlieb A Kang S Linden K Lebwohl M Menter A Abdulghani A Goldfrab M Cheiffo N Tortoritis Evaluation of safety and clinical activity of multiple doses of the anti-CD80 monoclonal antibody, galiximab, in patients with moderate to severe plaque psoriasis Clin Imm 2004 111 28 37 10.1016/j.clim.2004.01.006
|
16138929
|
PMC1208938
|
CC BY
|
2021-01-04 16:37:45
|
no
|
J Immune Based Ther Vaccines. 2005 Sep 2; 3:6
|
utf-8
|
J Immune Based Ther Vaccines
| 2,005 |
10.1186/1476-8518-3-6
|
oa_comm
|
==== Front
J Neuroengineering RehabilJournal of NeuroEngineering and Rehabilitation1743-0003BioMed Central London 1743-0003-2-221605352310.1186/1743-0003-2-22MethodologyManaging variability in the summary and comparison of gait data Chau Tom [email protected] Scott [email protected] Sue [email protected] Bloorview MacMillan Children's Centre, Toronto, Canada2 Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Canada2005 29 7 2005 2 22 22 30 4 2005 29 7 2005 Copyright © 2005 Chau et al; licensee BioMed Central Ltd.2005Chau et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Variability in quantitative gait data arises from many potential sources, including natural temporal dynamics of neuromotor control, pathologies of the neurological or musculoskeletal systems, the effects of aging, as well as variations in the external environment, assistive devices, instrumentation or data collection methodologies. In light of this variability, unidimensional, cycle-based gait variables such as stride period should be viewed as random variables and prototypical single-cycle kinematic or kinetic curves ought to be considered as random functions of time. Within this framework, we exemplify some practical solutions to a number of commonly encountered analytical challenges in dealing with gait variability. On the topic of univariate gait variables, robust estimation is proposed as a means of coping with contaminated gait data, and the summary of non-normally distributed gait data is demonstrated by way of empirical examples. On the summary of gait curves, we discuss methods to manage undesirable phase variation and non-robust spread estimates. To overcome the limitations of conventional comparisons among curve landmarks or parameters, we propose as a viable alternative, the combination of curve registration, robust estimation, and formal statistical testing of curves as coherent units. On the basis of these discussions, we provide heuristic guidelines for the summary of gait variables and the comparison of gait curves.
==== Body
Introduction
Definition of variability
In quantitative gait analysis, variability is commonly understood to be the fluctuation in the value of a kinematic (e.g. joint angle), kinetic (e.g. ground reaction force), spatio-temporal (e.g. stride interval) or electromyographic measurement. This fluctuation may be observed in repeated measurements over time, across or within individuals or raters, or between different measurement, intervention or health conditions. In this paper, we will focus on the variability in two types of data: unidimensional gait variables and single-cycle, prototypical gait curves, as these are the most common abstractions of spatio-temporal, kinematic and kinetic data, typically collected within a gait laboratory.
Measurement
Many different analytical methods have been proposed for estimating the variability in gait variables. The most widely used measures are those relating to the second moment of the underlying probability distribution of the gait variable of interest. Examples include, standard deviation (e.g., [1-4]), coefficient of variation (e.g., [5-8]) and coefficient of multiple correlation (e.g., [9,10]). Other less conventional variability measures have also been suggested. For example, Kurz et al. demonstrated an information-theoretic measure of variability, where increased uncertainty in joint range-of-motion (ROM), and hence entropy, reflected augmented variability in joint ROM [11].
For gauging variability among gait curves, some distance-based measures have been put forth, including the mean distance from all curves to the mean curve in raw 3-dimensional spatial data [12], the point-by-point intercurve ranges averaged across the gait cycle [13] and the norm of the difference between coordinate vectors representing upper and lower standard deviation curves in a vector space spanned by a polynomial basis [14]. Instead of reporting a single number, an alternative and popular approach to ascertain curve variability has been to peg prediction bands around a group of curves. Recent research on this topic has demonstrated that bootstrap-derived prediction bands provide higher coverage than conventional standard deviation bands [15-17].
Additionally, various summary statistics, such as the intra-class correlation coefficient [8] and Pearson correlation coefficient [18], for estimating gait measurement reliability, repeatability or reproducibility have been deployed in the assessment of methodological, environmental and instrumentation or device-induced variability. Principal components and multiple correspondence analyses have also been applied in the quantification of variability in both gait variables and curves, as retained variance and inertia, respectively, in low dimensional projections of the original data [19].
Sources of variability
As depicted in Figure 1, the numerous sources of variability in gait measurements can be loosely categorized as either internal or external to the individual being observed [20].
Figure 1 Sources of variability in empirical gait measurements.
Internal
Internal variability is inherent to a person's neurological, metabolic and musculoskeletal health, and can be further subdivided into natural fluctuations, aging effects and pathological deviations. It is now well known that neurologically healthy gait exhibits natural temporal fluctuations that are governed by strong fractal dynamics [21-23]. The source of these temporal fluctuations may be supraspinal [24] and potentially the result of correlated central pattern generators [25]. One hierarchical synthesis hypothesis purports that these nonlinear dynamics are due to the neurological integration of visual and auditory stimuli, mechanoreception in the soles of the feet, along with vestibular, proprioceptive and kinesthetic (e.g., muscle spindle, Golgi tendon organ and joint afferent) inputs arriving at the brain on different time scales [24,26]. Internal variability in gait measurements may be altered in the presence of pathological conditions which affect natural bipedal ambulation. For example, muscle spasticity tends to augment within-subject variability of kinematic and time-distance parameters [10] while Parkinson's disease, particularly with freezing gait, leads to inflated stride-to-stride variability [27] and electromyographic (EMG) shape variability and reduced timing variability in the EMG of the gastrocnemius muscle [28]. Similarly, recent studies have reported increased stride-to-stride variability due to Huntington's disease [29], amplified swing time variability due to major depressive and bipolar disorders [30], and heightened step width [31] and stride period [32] variability due to natural aging of the locomotor system.
External
Aside from mechanisms internal to the individual, variability in gait measurements may also arise from various external factors, as shown in Figure 1. For example, influences of the physical environment, such as the type of walking surface [33], the level of ambient lighting in conjunction with type of surface [34] and the presence and inclination of stairs [35] have been shown to affect cadence, step-width, and ground reaction force variability, respectively, in certain groups of individuals. Assistive devices, such as canes or semirigid ankle orthoses may reduce step-time and step-width variability [36] while different footwear (soft or hard) can affect the variability of knee and ankle joint angles, possibly by altering peripheral sensory inputs [14].
Variability may also originate from the nature of the instrumentation employed. This variability is often appraised by way of test-retest reliability studies. Some recent examples include the reproducibility of measurements made with the GAITRite mat [8], 3-dimensional optical motion capture systems [9,18], triaxial accelerometers [37], insole pressure measurement systems [4], and a global positioning system for step length and frequency recordings [7].
Experimenter error or inconsistencies may also contribute, as an external source, to the observed variability in gait data. Besier et al. contend that the repeatability of kinematic and kinetic models depends on accurate location of anatomical landmarks [38]. Indeed, various studies have confirmed the exaggerated variability in kinematic data due to differences in marker placement between trials [9,39] and between raters [40]. Finally, analytical manipulations, such as the computation of Euler angles [9] or the estimation of cross-sectional averages [41] may also amplify the apparent variability in gait data.
Clinical significance of variability
The magnitude of variability and its alteration bears significant clinical value, having been linked to the health of many biological systems. Particularly in human locomotion, the loss of natural fractal variability in stride dynamics has been demonstrated in advanced aging [32] and in the presence of neurological pathologies such as Parkinson's disease [42], and amyotrophic lateral sclerosis [42]. In some cases, this fractal variability is correlated to disease severity [32]. Variability may also serve as a useful indicator of the risk of falls [43] and the ability to adapt to changing conditions while walking [44]. Stride-to-stride temporal variability may be useful in studying the developmental stride dynamics in children [45]. Natural variability has been implicated as a protective mechanism against repetitive impact forces during running [14] and possibly a key ingredient for energy efficient and stable gait [46]. Variability is not always informative and useful and in fact may lead to discrepancies in treatment recommendations. For example, due to variability in static range-of-motion and kinematic measurements, Noonan et al. found that different treatments were recommended for 9 out of 11 patients with cerebral palsy, examined at four different medical centres [13].
Dealing with variability
Given the ubiquity and health relevance of variability in gait measurements, it is critical that we summarize and compare gait data in a way that reflects the true nature of their variability. Despite the apparent simplicity of these tasks, if not conducted prudently, the derived results may be misleading, as we will exemplify. In fact, there are to date many open questions relating to the analysis of quantitative gait data, such as the elusive problem of systematically comparing two families of curves.
The objectives of this paper are twofold. First, we aim to review some of the analytical issues commonly encountered in the summary and comparison of gait data variables and curves, as a result of variability. Our second goal is to demonstrate some practical solutions to the selected challenges, using real empirical data. These solutions largely draw upon successful methods reported in the statistics literature. The remainder of the paper addresses these objectives under two major headings, one on gait variables and the other on gait curves. The paper closes with some suggestions for the summary and comparison of gait data and directions for future research on this topic.
Gait random variables
Unidimensional variables which are measured or computed once per gait cycle will be referred to as gait random variables. This category includes spatio-temporal parameters such as stride length, period and frequency, velocity, single and double support times, and step width and length, as well as parameters such as range-of-motion of a particular joint, peak values, and time of occurrence of a peak, which are extracted from kinematic or kinetic curves on a per cycle basis.
Due to variability, univariate gait measures and parameters derived thereof should be regarded as stochastic rather than deterministic variables [47,48]. In this random variable framework, a one-dimensional gait variable is represented as X and governed by an underlying, unknown probability distribution function FX, or density function . A realization of this random variable is written in lower case as x.
Inflated variability and non-robust estimation
It has been recently demonstrated that typical location and spread estimators used in quantitative gait data analysis, i.e. mean and variance, are highly susceptible to small quantities of contaminant data [48]. Indeed, a few spurious or atypical measurements can unduly inflate non-robust estimates of gait variability. The challenge in the summary of highly variable univariate gait data lies in reporting location and spread, faithful to the underlying data distribution and minimally influenced by extraordinary observations.
Here, we focus on the issue of inflated variability and non-robust estimation by examining four different spread estimators, applied to stride period data from a child with spastic diplegic cerebral palsy. As stated above, the coefficient of variation and standard deviation are routinely employed in the summary of gait variables. Given a sample of N observations of a gait variable X, i.e., {x1,..., xN}, the coefficient of variation is defined as,
where the numerator is simply the sample standard deviation and the denominator, , is the sample mean. We also include two other estimators, although seldom used in gait analysis, to illustrate the qualitative differences in estimator robustness. The interquartile range of the sample is defined as
IQR(X) = x0.75 - x0.25 (2)
where x0.75 and x0.25 are the 75% and 25% quantiles. The q-quantile is defined as where as usual, FX is the probability distribution of X. Equivalently, the q-quantile is the value, xq, of the random variable where . That is, q × 100 percent of the random variable values lie below xq. We also introduce the median absolute deviation [49],
MAD(X) = med (|X - med(X)|) (3)
where med(X) is the median of the sample, or the 50% quantile as defined above. This last estimator is, as the name implies, the median of the absolute difference between the sample values and their median value. We are interested in studying how these different estimators perform when estimating the spread in a gait variable, the observations of which may contain outlying values or contaminants. In the left pane of Figure 2, we show a set of stride period data recorded from a child with spastic diplegia. The top graph shows the raw data with a number of obvious outliers with atypically long stride times. We adopted a common outlier definition, labeling points more than 1.5 interquartile ranges away from the sample median as extreme values. According to this definition there were 21 outlying observations. In the bottom graph, the outliers have been removed. The bar graph on the right-hand side of Figure 2 portrays the spread estimates of the stride period data, computed with each estimator introduced above, with and without the outliers.
Figure 2 Robust vs. non-robust estimators of parameter spread. The left pane shows a sequence of stride periods with outliers (top) and after removal of outliers (bottom). The right pane is a bar graph showing the values of four different spread estimators before and after outlier removal.
We note immediately that the spread estimates in the presence of outliers are higher. The standard deviation and coefficient of variation change the most, dropping 42 and 36 percent in value, respectively, upon outlier removal. This observation is particularly important in the comparison of gait variables, as inflated variability estimates will diminish the probability of detecting significant differences when they do in fact exist. In contrast, the interquartile range and median absolute deviation, only change by 21 and 11%, respectively. We see that these latter estimates are more statistically stable, in that they are not as greatly influenced by the presence of extreme observations.
To more fully comprehend estimator robustness or lack thereof, the field of robust statistics offers a valuable tool called influence functions, which as the name implies, summarizes the influence of local contaminations on estimated values. Their use in gait analysis was first introduced in the context of stride frequency estimation [48].
We first introduce the concept of a functional, which can be understood as a real-valued function on a vector space of probability distributions [50]. In the present context, functionals allow us to think of an estimator as a function of a probability distribution. For example, for the interquartile range, the functional is simply, .
Let the mixture distribution Fz, ε describe data governed by distribution F but contaminated by a sample z, with probability ε. The influence function at the contamination z is defined as
where T(·) is the functional for the estimator of interest. The influence function for a particular estimator measures the incremental change in the estimator, in the presence of large samples, due to a contamination at z. Clearly, if the impact of this contaminant on the estimated value is minimal, then the estimator is locally robust at z. Influence functions can be analytically derived for a variety of common gait estimators (see for example, [48]), including those mentioned above. For the sake of analytical simplicity and practical convenience, we will instead use finite sample sensitivity curves, SC(z), which can be defined as,
SC(z) = (N + 1){T(x1,..., xN, z) - T(x1,..., xN)} (5)
where as above, T(·) is the functional for the estimator in question, and z is the contaminant observation. When N → ∞ the sensitivity curve converges to the influence function for many estimators. Like the asymptotic influence functions, sensitivity curves describe the local impact of a contamination z on the estimator value. For the purposes of computer simulation, the functional T(x1,..., xN, z) and T(x1,..., xN) are simply the evaluations of the estimator of interest at the augmented and original samples, respectively. Figure 3 depicts the sensitivity curves for the estimators introduced in the stride period example. To generate these curves, we used the cleansed stride period data (without outliers) and incrementally added a deviant stride period from 0.5 below the lowest sample value to 0.5 above the highest sample value. The sample mean for this data was 1.41 seconds.
Figure 3 Sensitivity curves for various estimators of gait parameter variability based on the stride period example.
We observe that both standard deviation and coefficient of variation have quadratic sensitivity curves with vertices close to the sample mean. In other words, as contaminants take on extreme low or high values, the estimated values are unbounded. Clearly, these two estimators are not robust, explaining their high sensitivity to the outliers in the stride period data. In contrast, both the interquartile range and median absolute deviation have bounded sensitivity curves, in the form of step functions. The median absolute deviation is actually not sensitive to contaminant values above 1.1 seconds whereas the interquartile range has a constant sensitivity to contaminant values over 1.6. Since most of the outliers in the stride period data were well above the mean, this difference explains the considerably lower sensitivity of the median absolute deviation to outlier influence.
From this example, we appreciate that estimators of gait variable spread (i.e. variability) should be selected with prudence. The popular but non-robust variability measures of standard deviation and coefficient of variation both have 0 breakdown points [51], meaning that only a single extreme value is required to drive the estimators to infinity. Indeed, as seen in Figure 2, the presence of a small fraction of outliers can unduly inflate our estimates of gait variability. Outlier management [52], with methods such as outlier factors [53] or frequent itemsets [54], represents one possible strategy to reduce unwanted variability when using these non-robust estimators. Apart from the addition of a computational step, this strategy introduces the undesirable effects of outlier smearing and masking [55], which need to be carefully addressed.
In contrast, outliers need not be explicitly identified with robust estimation, hence circumventing the above complications and abbreviating computation. The interquartile range and median absolute deviation, have breakdown points of 0.25 and 0.5, respectively [51]. Practically, this means that these estimators will remain stable (bounded) until the proportion of outliers reaches 25% and 50% of the sample size, respectively. To circumvent explicit outlier detection and its associated issues altogether, and in the presence of noisy data, which often result from spatio-temporal recordings and parameterizations of kinematic and kinetic curves, robust estimators may thus be preferable in the summary of gait variables.
Non-gaussian distributions
Even in the absence of outliers, univariate gait data may not adhere to a simple, unimodal gaussian distribution. In fact, distributions of gait measurements and derived parameters may be naturally skewed, leptokurtic or multimodal [56]. Neglecting these possibilities, we may summarize gait data with location and spread values which do not reflect the underlying data distribution.
Semi-parametric estimation
As an example, consider the hip range-of-motion extracted from 45 strides of 9 able-bodied children. A histogram of the data is plotted in Figure 4. Assuming that the data are gaussian distributed, we arrive at maximum likelihood estimates for the mean and standard deviation, i.e. 40.4 ± 5.1. However, the histogram clearly appears to be bimodal. A Lilliefors test [57] confirms significant departure from normality (p = 0.02). A number of approaches could be undertaken to find the underlying modes. One could perform simple clustering analysis [58], such as k-means clustering. Doing so reveals two well-defined clusters, the means and standard deviations of which are reported in Table 1. Alternatively, one could attempt to fit to the data, a convex mixture density of the form,
Figure 4 Multimodal parameter distribution. Shown here is a histogram of hip range-of-motion (45 strides from 9 able-bodied children) with two possible distribution functions overlaid: unimodal normal probability distribution (solid line) and bimodal gaussian mixture distribution (dashed line).
Table 1 Summary of bimodal ROM data
Mixture distribution k-means clustering Normal distribution
Mode # 1 37.7 ± 2.4 37.7 ± 2.6 40.4 ± 5.1
Mode # 2 49.1 ± 3.5 47.7 ± 3.0 -
Mixing proportion (mode I/mode 2) 0.71/0.29 0.73/0.27 -
Critical value (lower) 33.35 32.96 30.40
Critical value (upper) 53.89 51.70 50.40
where Wi is a scalar such that ∑i Wi = 1 to preserve probability axioms, NC is the number of clusters or modes and is a gaussian density with mean μi and variance . The fitting of (6) is known as semi-parametric estimation as we do not assume a particular parametric form for the data distribution per se, but do assume that it can modeled by a mixture of gaussians. In the present case, NC = 2 and we can use a simple optimization approach to determine the parameters of the mixture. In particular, we determined the parameter vector [W1, W2, μ1, σ1, μ2, σ2] to minimize the objective function , where nj is the number of points within an interval of length Δ around xj and N is the number of points in the sample. The latter term in the objective function is a crude probability density estimate [59]. As seen in Table 1, the results of fitting this bimodal mixture yields similar results to those obtained from clustering.
What are the implications of naively summarizing these data with a unimodal normal distribution? First of all, the probabilities of observing range-of-motion values between 35 and 39 degrees, where most of the observations occur, would be underestimated. Likewise, ROM values between 39 and 48 degrees, where the data exhibit a dip in observed frequencies, would be grossly overestimated. These discrepancies are labeled as regions B and C in Figure 4. More importantly, the discrepancies in the tails of the distributions, regions A and D, suggest that statistical comparisons with other data, say pathological ROM, would likely yield inconsistent conclusions, depending on whether the mixture or simple distribution was assumed. Indeed, as seen in Table 1 the lower critical value of the simple normal distribution for a 5% significance level is too low. This could lead to exagerrated Type II errors. Similarly, the upper critical value is not high enough, potentially leading to many false positive (Type I) errors.
The above example depicts bimodal data. However, the mixture distribution method can be applied to arbitrary non-normal data distributions, regardless of the underlying modality. Fitting such distributions can be accomplished by the well-established expectation-maximization algorithm [60]. For a comprehensive review of other semi-parametric and non-parametric estimation methods, see for example [59].
Parametric estimation
When we have some a priori knowledge about the underlying data distribution, we can adopt a simpler approach to summarize the gait data. In particular, we could fit the data to a specific parametric form. As an example, consider the task of comparing two sets of stride period data from two children with spastic diplegia, with identical gross motor function classification scores [61]. The histograms of strides for both children are shown in Figure 5. It is known that stride period data tend to be right-skewed [56]. A careful examination of the bottom graph indicates that the histogram is indeed right-skewed. In fact, the skewness value is 1.7 and Lilliefors test for normality [57] confirms significant departure from normality (p < 10-5). We thus determine the maximum likelihood gamma distribution for these data. The gamma distribution has the following parametric form [62],
Figure 5 Comparison of stride period distributions between 2 children with spastic diplegia. In each graph, the dashed line is the normal probability distribution estimated for the data. The solid line is the gamma distribution fit to the data.
where a is the shape parameter, b is the scale parameter and Γ(·) is the gamma function. The gamma distribution fits are plotted as solid lines in Figure 5.
As in the previous example, we consider the consequence of assuming that the data are normally distributed. Do these two children have similar stride periods? To answer this question, one may hastily apply a t-test, assuming that the stride period distributions are gaussian. The results of this test reveal no significant differences (p = 0.31), as reported in Table 2. To visualize the departure from normality, the maximum likelihood normal probability distribution fits to the stride data are superimposed on each histogram as a dashed curve. Note that the tails of the distribution are overly broad, particularly in the bottom graph. This diminishes the likelihood of detecting genuine significant differences between the data sets. Table 2 summarizes the maximum likelihood estimates of the distribution parameters under the two different distributional assumptions. Under the gamma distribution assumption, the stride periods between the two children are statistically different (p = 0.036) according to a Monte Carlo simulation of differences between 104 similarly distributed gamma random variables, which contradicts the previous conclusion. We have arbitrarily chosen the gamma distribution in this example as it appears to describe well the positively skewed data. However, there are many other parametric forms that could be fit to gait data in general. See for example [62,63].
Table 2 Statistical comparison of stride periods under different distributional assumptions
Child No. strides Gaussian distribution Gamma distribution
uZ σZ a b
1 24 1.36 0.158 79.19 0.0171
2 23 1.74 0.734 7.513 0.232
p = 0.31 p = 0.036
In brief, the issue of non-normal distributions of measured gait variables or derived parameters, may lead to inaccurate reports of population means and variability and error-prone statistical testing. In fact, as the last example has shown, different distributional assumptions may lead to different statistical conclusions. Without a priori knowledge about the form of the distribution, one possible solution is to use a general mixture distribution to summarize the gait data. When we have some a priori knowledge about the underlying distribution, we can simply summarize the data using a known non-gaussian distribution, such as the gamma distribution exemplified above for the right-skewed stride period data. In either case, it is generally advisable to routinely check for significant departure from normality using such tests for normality as Pearson's Chi-square [64] or Lilliefors [57].
We remark that mixture models typically have a larger number of parameters than simple unimodal models. As a general rule-of-thumb, one should thus consider that mixture models generally require more data points for their estimation [59]. In particular, note that in any hypothesis test, the requisite sample size is dependent on the anticipated effect size, the desired level of significance and the specified level of statistical power [65]. For specific guidelines and methodology relating to sample size determination, the reader is referred to literature on sample size considerations in general hypothesis testing [66], normality testing [67], and other distributional testing [68].
Single-cycle gait curves
Kinematic, kinetic and metabolic data are often presented in the form of single-cycle curves, representing a time-varying value over one complete gait cycle. Time is often normalized such that the data vary over percentages of the gait cycle rather than absolute time. Examples include curves for joint angles, moments and powers, ground reaction forces, and potential and kinetic energy. Due to variability from stride-to-stride, these measurements do not generate a single curve, but a family of curves, each one slightly different from the other. We will consider a family of gait curves as realizations of a random function [69-71]. Let Xj (t) denote a discrete time function, i.e. a gait curve, where for convenience and without loss of generality, t is a positive integer and t = 1,..., 100. We further assume that the differences among curves at each point in time are independently normally distributed. Each sample curve, Xj (t), can thus be represented as [70],
Xj (t) = f(t) + εj (t) j = 1,..., N t = 1,..., 100 (8)
where f(t) is the true underlying mean function, εj (t) ~ (0, σj (t)2) are independent, normally distributed, gaussian random variables with variance σj (t)2 and N is the number of curves observed. With this formulation in mind, we now address four prevalent challenges in analyzing gait curves, namely, undesired phase variation, robust estimation of spread, the difficulty with landmark analysis and lastly, the comparison of curves as whole objects rather than as disconnected points.
Phase variation
It has been recognized that within a sample of single-cycle gait curves, there is both amplitude and phase variation [71-73]. Typically, when we describe variability in gait curves, we refer to amplitude variability. However, unchecked phase variation, that is the temporal misalignment of curves, can often lead to inflated amplitude variability estimates [72,73]. Computing cross-sectional averages over a family of malaligned gait curves can lead to the cancellation of critical shape characteristics and landmarks [74]. This issue presents a significant challenge when summarizing a series of curves for clinical interpretation and treatment planning. On the one hand, the presentation of a large number of different curves can be overwhelmingly difficult to assimilate. On the other hand, a prototypical average curve which does not reflect the features of the individual curves is equally uninformative.
Curve registration [71] is loosely the process of temporally aligning a set of curves. More precisely, it is the alignment of curves by minimizing discrepancies from an iteratively estimated sample mean or by allineating specific curve landmarks. Sadeghi et al. demonstrated the use of curve registration, particularly to reduce intersubject variability in angular displacement, moment and power curves [72,73]. Additionally, they reported that curve characteristics, namely, first and second derivatives and harmonic content were preserved while peak hip angular displacement and power increased upon registration [72]. This latter finding confirms that averaging unregistered curves may eliminate useful information.
Judging by the few gait papers employing curve registration, the method appears largely unknown among the quantitative gait analysis community. Here, we briefly outline the the global registration criterion method [71,75].
Since each gait curve is a discrete set of points, it is useful to estimate a smooth sample function for each observed sample curve. Given the periodic nature of gait curves, the Fourier transform provides an adequate functional representation of each curve. The basic principle is then to repeatedly align a set of sample functions to an iteratively estimated mean function. The agreement between a sample function and the mean function can be measured by a sum-of-squared error criterion. The goal of registration is to find a set of temporal shift functions such that the evaluation of each sample function at the transformed temporal values minimizes the sum-of-squared error criterion. The sample mean is re-estimated at each iteration with the current set of time-warped curves. As an optimization problem, the curve registration procedure is the iterative minimization of the sum-of-squared criterion J,
where N is the number of sample curves, T is the time interval of relevance, wi(·) is the time-warping function and is the iteratively estimated mean based on the current time-warped curves Xi (wi (s)). For greater methodological details, the reader is referred to [71,72,75]. This global registration criterion method is only one of several possibilities for curve alignment. Related methods which are applicable to gait data include dynamic time warping based on identified curve landmarks [41] and latency corrected ensemble averaging [28].
We exemplify the impact of accounting for undesirable phase variation using ankle angular displacement data from a child with spastic diplegla. The top left graph of Figure 6 depicts the unregistered curves, exhibiting excessive dorsiflexion throughout the gait cycle and the absence of the initial valley during loading response. Below this graph are the aligned curves. Note particularly the alignment of the large valley at pre-swing and the peak in swing phase.
Figure 6 Accounting for phase variation. On the left, we portray unregistered (top graph) and registered (bottom graph) ankle angle curves from a child with spastic diplegia. On the right are the mean (top) and standard deviation (bottom) curves before (dashed line) and after (solid line) curve registration.
The right column of Figure 6 indicates that the differences in the mean and standard deviation curves before and after registration are non-trivial, with maximum changes of +15% and -51%, respectively. The post-registration mean curve not only exhibits heightened but shifted peaks (3 – 5% of the gait cycle). This observation suggests that simple cross-sectional averaging without alignment may not only diminish useful curve features but can also inadvertently misrepresent the temporal position of key landmarks. Inaccurate identification of these landmarks, such as the minimum dorsiflexion at the onset of swing phase in this example, could be problematic when attempting to coordinate spatio-temporal and EMG recordings with kinematic curves. The bottom right graph shows a dramatic decrease in variability after registration, particularly in terminal stance. This finding is in line with the tendency towards variability reduction reported by Sadeghi et al. [72].
While curve registration is useful for mitigating unwanted phase variation in gait curves, there may be instances where phase variability is itself of interest [3]. In such instances, curve registration can still be useful in providing information about the relative temporal phase shifts among curves. Because curve registration actually changes the temporal location of data, it should not be applied in studies concerned with temporal stride dynamic characterizations, such as scaling exponents [21] or Lyapunov exponents [44]. At present, only a few gait studies have applied curve registration to manage undesired phase variability. However, the evidence in those studies, along with the example above, supports further research and exploratory application of curve registration to fully grasp its merits and limitations in quantitative gait data analyses. For now, curve registration appears to be the most viable solution to the challenge of summarizing a family of temporally misaligned gait curves. In the ensuing sections, we will demonstrate how curve registration can be used advantageously, in conjunction with other methods to address other curve summary and comparison challenges.
Robustness of spread estimation
We have already seen that curve registration can mitigate amplitude variability in a family of gait curves. The robust measurement of variability in gait curves is itself a non-trivial challenge. One may need to estimate the variability in a group of curves for the purposes of classifying a new observation as belonging to the same population, or not [15]. Alternatively, knowledge of the variability among curves can help in the statistical comparison of two populations of curves [16], say arising from two different subject groups or pre- and post-intervention.
As in gait variables, the challenge lies in robustly estimating the spread of a sample of gait curves and to avoid fallacious under or overestimation. The intuitive and perhaps most popular way of estimating curve variability is the calculation of the standard deviation across the sample of curves, for each point in the gait cycle. This yields upper, UX, and lower bands, LX, around the sample of curves, i.e.
UX (t) = μX (t) + σX (t) t = 1,..., 100
LX (t) = μX (t) - σX (t) (10)
where , for t = 1,..., 100, is the sample mean curve. Lenhoff et al. argued, by way of empirical examples and systematic cross-validation, that standard deviation bands provide inadequate coverage of the sample curves [15]. They instead supported the use of bootstrap prediction bands [76] which, in their study, provided close to the targeted 90% coverage of the sample curves. Two subsequent studies [16,17] have adopted the 90% bootstrap bands for the classification of new curves and for the comparison between groups of curves. The usefulness of bootstrap prediction bands for clinical identification of pathological deviations in kinematic curves has also been demonstrated [77]. In this section, we provide further evidence to support the use of bootstrap prediction bands and argue that they are more stable than standard deviation bands.
The basic idea of the bootstrap method is to create a large number of bootstrap subsets by resampling the curves Xj, j = 1,..., N with replacement. For each subset, the bootstrap mean and standard deviation are calculated. One then checks how many of the sample curves are "covered" by the bootstrap standard deviation bands. A curve is considered covered, if its maximum absolute standardized difference from the bootstrap mean is less than the bootstrap constant C. The number of covered curves averaged over all the bootstrap subsets then yields the coverage probability for the given bootstrap constant, C. The upper and lower bootstrap prediction bands can then be written as,
The reader is referred to [15] for details for practical computer implementation of the above procedure.
To exemplify issues of robust spread estimation, we consider knee angle curves from a child with spastic diplegia. Initially standard deviation and bootstrap bands are computed for the data prior to curve registration. The maximum absolute deviation from the sample mean curve is reported in Table 3. For both methods, the maximum spread decreases significantly upon registration, suggesting that there is significant inflated variability in the unaligned curve sample. Once the curves are aligned, one suspicious curve, plotted as a thin dashed line in Figure 7, becomes evident. The standard deviation bands around the sample with and without this outlying curve are shown on the left side of Figure 7. The maximum spread, that is maxt and C, for standard deviation and bootstrap bands, respectively, are labeled on each graph. We see that by removing the outlying curve, both the standard deviation and bootstrap bands become narrower. In fact, as seen in Table 3, the maximum standard deviation decreases by a dramatic 27%. Thus it appears that the variability among a group of curves, as estimated by both standard deviation and bootstrapping, can be minimized by curve registration and further reduced by the subsequent removal of outlying curves.
Table 3 Maximum spread estimates: registered and unregistered data
Bootstrap bands Standard deviation bands
Data set max C Change max Change
unregistered data 12.5 - 4.7 -
registered data 9.5 -24% 3.96 -16%
registered data without outlier 8.0 -16% 2.91 -27%
Figure 7 Estimation of spread in a group of registered knee angle curves from a 13-year old child with spastic diplegia. The left column depicts the standard deviation bands with (top graph) and without (bottom graph) an apparent outlying curve (thin dashed line). The 90% bootstrap prediction bands are plotted on the right, again with (top graph) and without (bottom graph) the outlying curve.
To further understand the robustness properties of the two spread estimators, we generate sensitivity curves using the 45 knee angle curves introduced in Figure 4. These curves are first registered to minimize unwanted phase variability. In the case of gait curves, the contaminant is not a single point, but an entire curve. For convenience, we choose the following contaminant,
where δ ∈ ℝ and δmin ≤ δ ≤ δmax. In other words, the contaminant is just a shifted version of the sample mean curve, . For simulating the sensitivity curve, we choose δmin = -50 and δmax = 50, recognizing that in practice, we would never observe deviations of this magnitude. This large range does however, gives us a more complete picture of the sensitivity curves. We proceed to define the sensitivity curves for the standard deviation and bootstrap estimates as follows,
where is the variance of the uncontaminated sample and
is the variance of the contaminated sample. In the above, is the mean curve of the contaminated sample. The notations CX and CX, z represent the bootstrap constants determined using the original and contaminated data, respectively. In other words, these sensitivity curves will reflect the influence of a contaminant curve, z(t), on the maximum estimated spread across a group of curves, over the gait cycle. Figure 8 summarizes the results of evaluating (15) over the simulated contaminants defined in (13).
Figure 8 Sensitivity curves for the standard deviation bands and 90% bootstrap estimated prediction bands. Here, each point on a sensitivity curve represents the difference between the maxima of the bands estimated with clean and contaminated data.
We note that, as in the univariate case, the standard deviation exhibits quadratic sensitivity with vertex at the zero deviation curve. This parabolic sensitivity curve indicates that the standard deviation bands are not locally robust to contaminant curves. In contrast, the sensitivity curve for the bootstrap bands is not smooth and quartic in nature. The lack of smoothness is due to the random resampling inherent in the bootstrap method, such that with each contaminant curve, slightly different bootstrap samples are used in estimating the 90% prediction bands. Initially, as the contaminant curve deviates from the mean curve, the sensitivity is negative, meaning that the width of the estimated bands are smaller than those for the uncontaminated data. Indeed, the actual value of the bootstrap constant initially decreases, likely to counter the accompanying sharp increase in the standard deviation bands. In other words, as the standard deviation bands widen, a smaller bootstrap constant is required to cover 90% of the sample curves. However, as the contaminant curve deviates farther from the mean, the slope of standard deviation sensitivity increases in magnitude more slowly. With a smaller change in standard deviation band per unit of deviation of the contaminant curve, the bootstrap constant necessarily increases to maintain 90% coverage. This reasoning accounts for the subsequent increase in the tails of the bootstrap sensitivity curve. Finally, we note that overall, the bootstrap sensitivity curve, although apparently unbounded, traverses a much smaller range than the standard deviation curve. This would suggest that with the kinematic data employed in this example, the bootstrap coverage bands enjoy greater stability than their highly sensitive standard deviation cousins.
In brief, the foregoing discussion further supports the use of bootstrap coverage bands in robustly summarizing the variability within a family of gait curves. Moreover, curve registration and outlier removal can further tighten the location of the prediction bands.
Problems with simple parameterizations
It is common to compare specific landmarks or features of gait curves to gauge the impact of an intervention or to determine differences among different subject populations. However, the identification of curve features is inherently problematic. Indeed, the multiplicity of peaks and valleys across two different groups of curves may be inconsistent. As an example, Figure 9 portrays the vertical ground reaction force of an able-bodied child on the left with the typical loading response peak, mid-stance valley and terminal stance peak [78]. On the right is the vertical ground reaction force from the intact side of a child with an above-knee amputation, walking with a prosthetic lower limb. Note that there are at least three distinct peaks in the mean curve (thin dotted line). Attempts to compare the able-bodied and amputee force profiles are encumbered by the unclear choice of corresponding peaks. This is a common pitfall of comparing gait curves on the basis of specific landmarks.
Figure 9 Inconsistency in multiplicity and location of local extrema. Graphs portray registered vertical ground reaction force curves from an able-bodied child (left) and from a child with above-knee prosthesis (right). The dotted line on the right is the mean curve while dashed line is the wavelet reconstructed mean curve.
The wavelet transform has been touted as a useful method for uncovering intrinsic trends in data [79,80]. Hence, it may be possible to extract an underlying low frequency trend from the amputee force curve for the sake of striking a comparison with the able-bodied curve. To this end, we decomposed the mean force curve for the child with amputation using a 4 – level coiflet wavelet transform [81]. We reconstructed the force curve using only the approximation coefficients. The resulting trend line is plotted on the right graph of Figure 9 as the thick dashed line and more closely resembles the expected force profile. Extraction of the extrema yields plausible peak locations at 17% and 44% of the gait cycle and a valley at 30%. These locations are comparable to those for the able-bodied child (peaks at 12% and 44% and valley at 26% of the gait cycle), but suggest a slightly extended loading response phase.
The extraction of the trend line in this example illustrates that in some curves, the desired landmarks may be concealed by the fluctuations of higher frequency signal components and hence may be salvageable. However, even when landmarks are clearly identifiable among curves, they reflect only a very microscopic view of the entire curve. For example, two curves could have identical landmarks, but pronounced differences in shape characteristics. We therefore do not advocate the isolated use of simple parameterizations or landmarks for routine comparison of curves. Rather, the comparison of two sets of curves should be based on the entire curve and not isolated parameterizations. We suggest however, that landmark analysis and simple parameterizations can be meaningful as a post-hoc procedure, i.e. investigating how curves are similar or different, only after statistically significant differences among curves or lack thereof have been established. We therefore suggest to first statistically compare entire gait curves as unified objects, and reserve landmarks for post-hoc analysis. In the following section, we describe how such a statistical test may be carried out.
Comparison of gait curves as coherent entities
If gait curves were strictly deterministic, one could simply define a distance measure between two curves and be done. However, due to stride-to-stride variability, an extension of the univariate statistical test is needed, to determine if one set of curves could have arisen from the same statistical distribution as another. Alternatively, one could test whether the average difference between two sets of curves is approximately zero, within the critical values of an expected distribution of differences. The fundamental challenge is to compare families of gait curves as coherent entities rather than as unconnected, independent points. One way to consider curves as a whole rather than as disjoint points is to give them an appropriate functional representation. One can then compare the functional representations of the curves. Exploiting this principle, Fan and Lin [70] proposed a general method for comparing two sets of discrete time-sampled curves. In their method, the discrete Fourier transform of the standardized difference between the mean curves of two families of curves is computed. Only selected low frequency components of the transform, which encompass the majority of signal energy, are retained. These coefficients are then subjected to the adaptive Neyman test which yields the probability that the two families of curves have similar means. To the best of our knowledge, the adaptive Neyman statistic [69] has not yet been applied in the gait literature for the comparison of empirical gait curves. We therefore outline below, in some detail, the proposed procedure that we have adapted from Fan and Lin [70]. Suppose that we would like to compare two families of gait curves, {Xj (t), j = 1,..., NX} and {Yj (t), j = 1,..., NY}, with t = 1,..., 100. The null hypothesis is that the difference between the means of the two families of curves is zero. In the random function formulation given by Equation (8), we can write, H0 : fX (t) - fY (t) = 0, where fX (t) and fY (t) are the true underlying mean curves. For notational convenience, we will let t = 0,..., T - 1, where we have chosen T = 100 in the previous examples. The main steps of the test are as follows.
1. Compute the sample mean curves, μX (t) and μY (t), where and likewise for μY (t)
2. Compute the sample variance curves, and , where and likewise for .
3. Align the two mean curves using the global registration criterion method. Denote the registered curves as and . This step does not appear in the original formulation of Fan and Lin [70].
4. Compute the standardized difference Z(t) between the registered means,
5. Compute the discrete Fourier decomposition, , of the standardized difference,
where k = 0,..., T/2, Real(·) and Imag(·) denote the real and imaginary components of the complex Fourier coefficient , respectively, and k denotes the Fourier frequency.
6. Form a new vector of coefficients E, of length T + 1, by pairing real and imaginary coefficients of the complex Fourier coefficients, , as follows,
7. Estimate the adaptive Neyman statistic, TAN(E) for the vector defined above. This proceeds in two steps.
(a) Determine the optimal the number of coefficients to retain to maximize , where Ei are the elements of the vector defined above and 1 <m <T + 1. This optimal value of m, denoted , maximizes the power of the adaptive Neyman statistic [70]. The maximum statistic value is written as,
where Var(E2), is the variance of the square of the elements of E obtained in step 6.
(b) Let K = ln(T ln T). Compute the following final transformed test statistic value [70],
Here, we have explicited indicated that the statistic has been computed for the vector E of Fourier coefficients. Asymptotically, this statistic has an exponential of an exponential distribution [69], that is, P(TAN ≤ x) → exp(-exp(-x)), as T becomes arbitrarily large.
8. Estimate the p-value of the computed test statistic value, TAN(E), by Monte Carlo simulation of a large number, say 106, of vectors, Yi, i = 1,..., 106, each of length T and whose elements are drawn from a standard normal distribution, i.e. Yi ~ (0, 1), ∀i. The rationale is that when two sets of curves arise from the same random function, the standardized differences of their Fourier coefficients are normally distributed around 0. For each normal vector, Yi, evaluate TAN(Yi) as in step 7 above. When the null hypothesis of no differences is true, the probability of observing an adpative neyman statistic as extreme as TAN(E) is estimated as,
where H(·) is the heaviside function, where H(x) = 1 only if x > 0 and is 0 otherwise. In the examples below, we simulated 106 such vectors to estimate the probability of observing TAN.
We exemplify the above procedure with two kinematic data sets taken from a child with an above-knee amputation, wearing two different types of prosthetic devices. Of interest is whether the use of a swing phase control mechanism within the prosthetic knee affects gait. Figure 10 depicts the ankle angular displacements with (dashed line) and without (solid line) the swing phase control mechanisms. By visual inspection, the curves look similar in magnitude and exhibit a slight phase difference. We would expect statistical testing to conclude that the curves are not different. The top right graph is the standardized difference between the registered mean curves. Note that most values fluctuate around 0. The bottom right graph is a stem plot of selected Fourier coefficients of the Fourier transform of the standardized difference. Note the concentration of energy in the low frequencies and the distribution of coefficients above and below 0. It is thus not surprising that the result of the adaptive Neyman test statistic, yielded TAN = -9.01 which corresponds to p = 1 for = 2. In other words, it is very likely that the two groups of curves came from the same distribution of random functions. In the present context, the swing phase mechanism had no effect on the ankle angular displacement.
Figure 10 Comparison of ankle angle curves (left pane) from an above-knee amputee using a prosthetic knee without (solid line) and with (dashed line) a swing phase control mechanism. The right pane depicts the standardized difference between the registered mean curves for the two groups (top graph) and the corresponding first 30 coefficients of its discrete Fourier transform.
To illustrate the statistical detection of differences, we draw upon a second example involving kinematic curves from an adult subject pre- and post-surgical replacement of the ankle. In Figure 11, the pre-surgery curves do not have well resolved peaks or valleys whereas post-surgically, distinct peaks and valleys emerge with substantial magnitude. On the basis of this visual inspection, one would anticipate that statistical testing should indicate that the pre- and post-surgery curves are indeed different. The standardized difference between the registered mean curves exhibits relatively large fluctuations around 0 and the retained Fourier coefficients are nearly all positive, resulting in a positively skewed coefficient distribution. The adaptive Neyman statistic value for these coefficients is TAN = 5.99 corresponding to p = 2.6 × 10-5 with = 6. Hence, the statistical test indicates that there is strong evidence for rejecting the null hypothesis. It appears that surgery has significantly altered the gait curves. Once significant statistical difference has been established, one can then seek to identify specific characteristics which differentiate the two sets of curves. For example, the post-surgical curves exhibit a well-defined valley, towards plantar flexion at toe-off and a strong first dorsiflexion peak in terminal stance. Both of these extrema are absent in the presurgery curves.
Figure 11 Comparison of ankle angle curves (left pane) from one individual before and after total ankle replacement surgery. The right pane portrays the standardized difference between the registered mean curves of each group (top graph) and the first 30 coefficients of its discrete Fourier transform.
Note that we have not said anything about the requisite sample sizes for the statistical comparison of gait curves. Clearly, as in unidimensional power analysis [65], the required sample size depends on the effect size, significance level and specified power. To the best of our knowledge, no power-sample size tables have been derived for the adaptive Neyman statistic at the time of writing. For insights on the topic, the interested reader can refer to authoritative works [65,82] on power analysis in the univariate case. The statistical testing demonstrated here can be extended to compare more than two groups of curves, using high-dimensional analysis of variance [70]. Further, when the standardized difference curve is not smooth, wavelet denoising can be used to identify the frequency bands where the majority of signal energy is concentrated [69]. The adaptive Neyman statistic introduced here is only one of several possibilities for objectively and rigorously testing differences among curves. Other alternatives include an ANOVA test for functional data [83] and functional canonical correlation analysis [84]. The procedure outlined in this section formalizes the comparison of gait curves as coherent entities. The method provides a means of statistically confirming overall similarities and differences that we may detect by visual inspection, but may have difficulty quantifying with conventional time and frequency domain parameterizations.
Recommendations
We summarize the foregoing discussions by proposing some heuristic guidelines for dealing with the aforementioned variability issues in gait variables and curves. For gait variables or parameters, the suggested solution pathways are shown in Figure 12.
Figure 12 Heuristic guidelines for summary of gait variables or parameters.
For gait curves, the suggested procedures for summary and comparison are summarized in Figure 13. A few comments beyond the above discussions are in order. Note that robust estimation is suggested in the summary of gait curves, as after registration, there may still be curves which appear atypical, in amplitude or overall shape. Location estimation of gait curves was only discussed in the context of the adaptive Neyman test, but is included in Figure 13 for completeness. In the comparison of curves, post-hoc analysis would encompass the comparisons of conventional curve parameterizations or landmarks (e.g. peaks and valleys), as investigative procedures to explain the formally established statistical differences or lack thereof.
Figure 13 Heuristic guidelines for the summary (on the left) and comparison (on the right) of gait curves.
Future directions
This paper has only skimmed the tip of the iceberg in the discussion and demonstration of several promising analytical approaches for practically addressing variability issues in gait data summary and comparison. The topics of curve registration and bootstrap estimates of curve variability, although not necessarily new to gait data analyses, have been seldom studied and applied in the gait research community. The handful of studies to date on these subjects, have provided strong initial evidence for potentially improving the rigor and objectivity of gait data interpretation. Examples in the present paper lend further credence to these methods. Systematic comparisons of these techniques with conventional parameterizations, summary statistics, and even expert interpretation of gait data, would lead to a greater appreciation of their relative merits and limitations in gait data analyses. For example, would the use of registration and bootstrapping to consolidate gait data improve the consistency of clinical decision-making? Given the propensity for variability inflation in gait data, the topic of robust estimation needs to be studied in greater depth, in terms of contaminant influences and possibly adaptive estimators [49]. Likewise, the rigorous statistical comparison of gait curves as coherent entities rather than uncorrelated sets of points, is a promising area of research in gait variability analyses. This stream of study is only in the embryonic stages but promises to strengthen the comparison of quantitative gait data and to complement its subjective interpretation, a pratice which has been debated in literature [85-87].
Authors' contributions
T. Chau wrote the entire manuscript and carried out the majority variability analyses reported herein.
S. Redekop collected most of the empirical data reported herein, carried out all the kinematic and kinetic data analyses, wrote programs for data extraction, identified the datasets and helped in their interpretation.
S. Young provided the literature review for the manuscript and contributed significantly to revising various drafts.
Acknowledgements
The primary author would like to acknowledge the support of the Canada Research Chairs program, the Natural Sciences and Engineering Research Council, the REMAD Foundation, the Canada Foundation for Innovation, the Ontario Innovation Trust and the Bloorview Research Institute. The authors also acknowledge Jan Andrysek who contributed the amputee data.
==== Refs
Owings T Grabiner M Step width variability, but not step length variability or step time variability, discriminates young and older adults during treadmill locomotion Journal of Biomechanics 2004 37 935 938 15111081 10.1016/j.jbiomech.2003.11.012
Danion F Varraine E Bonnard M Pailhous J Stride variability in human gait: the effect of stride frequency and stride length Gait and Posture 2003 18 69 77 12855302 10.1016/S0966-6362(03)00030-4
Kao J Ringenbach S Martin P Gait transitions are not dependent on changes in intralimbcoordination variability Journal of Motor Behavior 2003 35 211 214 12873836
Randolph A Nelson M Akkapeddi S Levin A Alexandrescu R Reliability of measurements of pressures applied on teh foot during walking by a computerized insole sensor system Archives of Physical Medicine and Rehabilitation 2000 81 573 578 10807094 10.1016/S0003-9993(00)90037-6
del Olmo M Cudeiro J Temporal variability of gait in Parkinson disease: effeccts of a rehabilitation program based on rhythmic sound cues Parkinsonism and Related Disorders 2005 11 25 33 15619459 10.1016/j.parkreldis.2004.09.002
Cavanagh P Perry J Ulbrecht J Derr J Pammer S Neuropathic diabetic patients do not have reduced variability of plantar loading durig gait Gait & Posture 1998 7 191 199 10200384 10.1016/S0966-6362(98)00011-3
Terrier P Schutz Y Variability of gait patterns during unconstrained walking assesed by satellite positioning (GPS) European Journal of Applied Physiology 2003 90 554 561 12905048 10.1007/s00421-003-0906-3
Menz H Latt M Tiedemann A Kwan M Lord S Reliability of the GAITRite(R) walkway system for the quantification of temporo-spatial parameters of gait in young and older people Gait & Posture 2004 20 20 25 15196515 10.1016/S0966-6362(03)00068-7
Growney E Meglan D Johnson M Cahalan T An K Repeated measures of adult normal walking using a video tracking system Gait & Posture 1997 6 147 162 10.1016/S0966-6362(97)01114-4
Steinwender G Saraph V Scheiber S Zwick E Uitz C Hackl K Intrasubject repeatability of gait analysis data in normal and spastic children Clinical Biomechanics 2000 15 134 139 10627329 10.1016/S0268-0033(99)00057-1
Kurz M Stergiou N The aging human neuromuscular system expresses less certainty for selecting joint kinematics during gait Neuroscience Letters 2003 348 155 158 12932817 10.1016/S0304-3940(03)00736-5
Abel R Rupp M Sutherland D Quantifying the variability of a complex motor task specifically studying the gait of dyskinetic CP children Gait & Posture 2003 17 50 58 12535726 10.1016/S0966-6362(02)00054-1
Noonan K Halliday S Browne R O'Brien S Kayes K Feinberg J Interobserver variability of gait analysis in patients with cerebral palsy Journal of Pediatric Orthopaedics 2003 23 279 287 12724586 10.1097/00004694-200305000-00001
Kurz M Stergiou N The spanning set indicates that variability during the stance period of running is affected by footwear Gait & Posture 2003 17 132 135 12633773 10.1016/S0966-6362(02)00064-4
Lenhoff M Santer T Otis J Peterson M Williams B Backus S Bootstrap prediction and confidence bands: a superior statistical method for analysis of gait data Gait and Posture 1999 9 10 17 10575065 10.1016/S0966-6362(98)00043-5
Duhamel A Bourriez J Devos P Krystkowiak P Destee A Derambure P Defebvre L Statistical tools for clinical gait analysis Gait and Posture 2004 20 204 212 15336292 10.1016/j.gaitpost.2003.09.010
Murray-Weir M Root L Peterson M Lenhoff M Wagner C Marcus P Proximal femoral varus rotation osteotomy in cerebral palsy: a prospective gait study Journal of Pediatric Orthopaedics 2003 23 321 329 12724594 10.1097/00004694-200305000-00009
Westhoff B Hirsch M Hefter H an dR Krauspe AW Test-retest reliability of 3-dimensional computerized gait analysis Sportverletzung-sportschaden 2004 18 76 79 15164292 10.1055/s-2004-813229
Chau T A review of analytical techniques for gait data: Part I: fuzzy, statistical and fractal methods Gait & Posture 2001 13 49 66 11166554 10.1016/S0966-6362(00)00094-1
Schwartz M Trost J Wervey R Measurement and management of errors in quantitative gait data Gait and Posture 2004 20 196 203 15336291 10.1016/j.gaitpost.2003.09.011
Hausdorff J Peng C Ladin Z Wei J Goldberger A Is walking a random walk? Evidence of long-range correlations in stride interval of human gait Journal of Applied Physiology 1995 78 349 358 7713836
West B Griffin L Allometric Control, Inverse Power Laws and Human Gait Chaos, Solitons and Fractals 1999 10 1519 1527 10.1016/S0960-0779(98)00149-0
Griffin L West D West B Random Stride Intervals with Memory Journal of Biological Physics 2000 26 185 202 10.1023/A:1010322406831
Hausdorff J Purdon P Peng C Ladin Z Wei J Goldberger A Fractal dynamics of human gait: stability of long-range correlations in stride interval fluctuations Journal of Applied Physiology 1996 80 1448 1457 8727526
West B Scafetta N Nonlinear dynamical model of human gait Physical Review E 2003 67 1063 1065 10.1103/PhysRevE.67.051917
Hausdorff J Peng C Multiscaled randomness: A possible source of 1/f noise in biology Physical Review E 1996 54 2154 2157 10.1103/PhysRevE.54.2154
Hausdorff J Schaafsma J Balash Y Bartels A Gurevich T Giladi N Impaired regulation of stride variability in Parkinson's disease subjects with freezing gait Experimental Brain Research 2003 149 187 194
Miller R Thaut M Mclntosh G Rice R Components of EMG symmetry and variability in parkinsonian and healthy elderly gait Electromyography and motor control – electroencelphalography and clinical neurophysiology 1996 101 1 7 10.1016/0013-4694(95)00209-X
Hausdorff J Cudkowicz M Firtion R Wei J Goldberger A Gait variability and basal ganglia disorders: stride-to-stride variations in gait cycle timing in Parkinson's disease and Huntington's disease Movement Disorders 1998 13 428 437 9613733 10.1002/mds.870130310
Hausdorff J Peng C Goldberger A Stoll A Gait unsteadiness and fall risk in two affective disorders: a preliminary study BMC Psychiatry 2004 4 39 15563372 10.1186/1471-244X-4-39
Owings T Grabiner M Variability of step kinematics in young and older adults Gait & Posture 2004 20 26 29 15196516 10.1016/S0966-6362(03)00088-2
Hausdorff J Mitchell S Firtion R Peng C Cudkowicz M Wei J Goldberger A Altered fractal dynamics of gait: reduced stride-interval correlations with aging and Huntington's disease Journal of Applied Physiology 1997 82 262 269 9029225
Menz H Lord S Fitzpatrick R Acceleration patterns of the head and pelvis when walking on level and irregular surfaces Gait and Posture 2003 18 35 46 12855299 10.1016/S0966-6362(02)00159-5
Richardson J Thies S Demott T Ashton-Miller J Interventions improve gait regularity in patients with peripheral neuropathy while walking on an irregular surface under low light Journal of the American Geriatrics Society 2004 52 510 515 15066064 10.1111/j.1532-5415.2004.52155.x
Stacoff A Diezi C Luder G Stussi E Quervain IKD Ground reaction forces on stairs: effects of stair inclination and age Gait & Posture 2005 21 24 38 15536031 10.1016/j.gaitpost.2003.11.003
Richardson J Thies S DeMott T Ashton-Miller J A comparison of gait characteristics between older women with and without peripheral neuropathy in standard and challenging environments Journal of the American Geriatrics Society 2004 52 1532 1537 15341557 10.1111/j.1532-5415.2004.52418.x
Henriksen M Lund H Moe-Nilssen R Bliddal H Danneskiod-Samsoe B Test-retest reliability of trunk accelerometric gait analysis Gait & Posture 2004 19 288 297 15125918 10.1016/S0966-6362(03)00069-9
Besier T Sturnieks D Alderson J Lloyd D Repeatability of gait data using a functional hip joint centre and a mean helical knee axis Journal of Biomechanics 2003 36 1159 1168 12831742 10.1016/S0021-9290(03)00087-3
Carson M Harrington M Thompson N O'Connor J Theologis T Kinematic analysis of a multi-segment foot model for research and clinical applications: a repeatability analysis Journal of Biomechanics 2001 34 1299 1307 11522309 10.1016/S0021-9290(01)00101-4
Maynard V Bakheit A Oldham J Freeman J Intra-rater and inter-rater reliability of gait measurements with CODA mpx30 motion analysis system Gait & Posture 2003 17 59 67 12535727 10.1016/S0966-6362(02)00051-6
Wang K Gasser T Alignment of curves by dynamic time warping The Annals of Statistics 1997 25 1251 1276 10.1214/aos/1069362747
Hausdorff J Lertratanakul A Cudkowicz M Peterson A Kaliton D Goldberger A Dynamic markers of altered gait rhythm in amyotrophic lateral sclerosis Journal of Applied Physiology 2000 88 2045 2053 10846017
Hausdorff J Rios D Edelberg H Gait variability and fall risk in community-living older adults: a 1-year prospective study Archives of Physical Medicine and Rehabilitation 2001 82 1050 1056 11494184 10.1053/apmr.2001.24893
Buzzi U Stergiou N Kurz M Hageman P Heidel J Nonlinear dynamics indicates aging affects variability during gait Clinical Biomechanics 2003 18 435 443 12763440 10.1016/S0268-0033(03)00029-9
Hausdorff J Zemany L Peng C Goldberger A Maturation of gait dynamics: stride-to-stride variability and its temporal organization in children Journal of Applied Physiology 1999 86 1040 1047 10066721
Goldberger A Amaral L Hausdorff J Ivanov P Peng C Stanley H Fractal dynamics in physiology: alterations with disease and aging PNAS 2002 99 2466 2472 11875196 10.1073/pnas.012579499
Stokes V Thorstensson A Lanshammar H From stride period to stride frequency Gait & Posture 1998 7 35 38 10200373 10.1016/S0966-6362(97)00024-6
Chau T Parker K On the robustness of stride frequency estimation IEEE Transactions on Biomedical Engineering 2004 51 294 303 14765702 10.1109/TBME.2003.820396
Shevlyakov G Vilchevski N Robustness in data analysis 2002 Utrecht: VSP
Kreyszig E Introductory functional analysis 1989 New York: Wiley
Wilcox R Introduction to robust estimation and hypothesis testing 1997 San Diego: Academic Press
Hodge V Austin J A survey of outlier detection methodologies Artificial Intelligence Review 2004 22 85 126
Saltenis V Outlier detection based on the distribution of distances between data points Informatica 2004 15 399 410
He Z amd JZX Huang XX Deng S A frequent pattern discovery method for outlier detection Lecture Notes in Computer Science 2004 3129 726 732
Tolvi J Genetic algorithms for outlier detection and variable selection in linear regression models Soft Computing 2004 8 527 533 10.1007/s00500-003-0310-2
Chau T Rizvi S Automatic stride interval extraction from long, highly variable noisy gait timing signals Human Movement Science 2002 21 495 514 12450681 10.1016/S0167-9457(02)00125-2
Conover W Practical nonparametric statistics 1998 third New York: John Wiley & Sons
Duda R Hart P Stork D Pattern Classification 2000 New York: Wiley Interscience
Scott D Multivariate density estimation 1992 Wiley series in probability and statistics, New York: Wiley Interscience
McLachlan G Krishnan T The EM algorithm and extensions 1997 Wiley series in probability and statistics, New York: John Wiley & Sons
Palisano R Rosenbaum P Walter P Russel D Wood E Galuppi B Development and reliability of a system to classify gross motor function in children with cerebral palsy Developmental Medicine and Child Neurology 1997 39 214 223 9183258
Papoulis A Probability, random variables and stochastic processes 1991 New York: McGraw Hill
Kazakos D Papantoni-Kazakos P Detection and estimation 1990 New York: Computer Science Press
Bendat J Piersol A Random data 2000 New York: Wiley
Cohen J Statistical power analysis for the behavioral sciences 1988 Lawrence Erlbaum Associates
Asraf R Brewer J Conducting tests of hypotheses: the need for an adequate sample size Australian Educational Researcher 2004 31 79 94
Baxter M Beardah C Westwood S Sample size and related issues in the analysis of lead isotope data Journal of Archaeological Science 2000 27 973 980 10.1006/jasc.1999.0546
Kundu D Manglick A Discriminating between the Weibull and log-normal distributions Naval Research Logistics 2004 51 893 905 10.1002/nav.20029
Fan J Test of significance based on wavelet thresholding and Neyman's truncation Journal of the American Statistical Association 1996 91 674 688
Fan J Lin S Test of significance when data are curves Journal of the American Statistical Association 1998 93 1007 1021
Ramsay J Silverman B Functional data analysis 1997 New York: Springer Verlag
Sadeghi H Mathieu P Sadeghi S Labelle H Continuous curve registration as an intertrial gait variability reduction technique IEEE Transactions on Neural Systems and Rehabilitation Engineering 2003 11 24 30 12797722 10.1109/TNSRE.2003.810428
Sadeghi H Allard P Shafie K Mathieu P Sadeghi S Prince F Ramsay J Reduction of gait variability using curve registration Gait & Posture 2000 12 257 264 11154937 10.1016/S0966-6362(00)00085-0
Kneip A Gasser T Statistical tools to analyze data representing a sample of curves Annals of Statistics 1992 20 1266 1305
Ramsay J Li X Curve registration Journal of the Royal Statistical Society – Series B 1998 60 351 363 10.1111/1467-9868.00129
Olshen R Biden E Wyatt M Sutherland D Gait analysis and the bootstrap The Annals of Statistics 1989 17 1419 1440
Sutherland D Kaufman K Campbell K Ambrosini D Wyatt M Clinical use of prediction regions for motion analysis Developmental Medicine and Child Neurology 1996 38 773 781 8810708
Perry J Gait analysis: normal and pathological function 1992 New Jersey: SLACK Inc
Andreas E Trevino G Using wavelets to detect trends Journal of Atmospheric and Oceanic Technology 1997 14 554 564 10.1175/1520-0426(1997)014<0554:UWTDT>2.0.CO;2
Meyer F Wavelet-based estimation of a semiparametric generalized linear model of fMRI time-series IEEE Transactions on Medical Imaging 2003 22 315 322 12760549 10.1109/TMI.2003.809587
Cooklev T Nishihara A Biorthogonal Coiflets IEEE Transactions on Signal Processing 1999 47 2582 2588 10.1109/78.782213
Bausell R Li Y Power analysis for experimental research: a practical guide for the biological, medical and social sciences 2002 Cambridge: Cambridge University Press
Cuevas A Febrero M Fraiman R An anova test for functional data Computational Statistics and Data Analysis 2004 47 111 122 10.1016/j.csda.2003.10.021
Leurgans S Moyeed R Silverman B Canonical correlation analysis when the data are curves Journal of the Royal Statistical Society B 1993 55 725 740
Watts HG Gait laboratory analysis for preoperative decision making in spastic cerebral palsy: Is it all it's cracked up to be? Journal of Pediatric Orthopaedics 1994 14 703 4 7814579
Davis RB Reflections on clinical gait analysis Journal of Electromyography and Kinesiology 1997 7 251 7 11369268 10.1016/S1050-6411(97)00008-4
Sisto SA Lisa JAD The Value of Information Resuling from Instrumented Gait Analysis: The Physical Therapist Gait Analysis in the Science of Rehabilitation, Monograph 002 1998 Department of Veteran Affairs 76 84
|
16053523
|
PMC1208939
|
CC BY
|
2021-01-04 16:37:40
|
no
|
J Neuroengineering Rehabil. 2005 Jul 29; 2:22
|
utf-8
|
J Neuroeng Rehabil
| 2,005 |
10.1186/1743-0003-2-22
|
oa_comm
|
==== Front
J Negat Results BiomedJournal of Negative Results in Biomedicine1477-5751BioMed Central London 1477-5751-4-51612488210.1186/1477-5751-4-5ResearchUnsatisfactory gene transfer into bone-resorbing osteoclasts with liposomal transfection systems Laitala-Leinonen Tiina [email protected] Institute of Biomedicine, Department of Anatomy, University of Turku, Turku, Finland2005 29 8 2005 4 5 5 18 1 2005 29 8 2005 Copyright © 2005 Laitala-Leinonen; licensee BioMed Central Ltd.2005Laitala-Leinonen; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Bone-resorbing osteoclasts are multinucleated cells that are formed via fusion of their hematopoietic stem cells. Many of the details of osteoclast formation, activation and motility remain unsolved. Therefore, there is an interest among bone biologists to transfect the terminally differentiated osteoclasts and follow their responses to the transgenes in vitro. Severe difficulties in transfecting the large, adherent osteoclasts have been encountered, however, making the use of modern cell biology tools in osteoclast research challenging. Transfection of mature osteoclasts by non-viral gene transfer systems has not been reported.
Results
We have systematically screened the usefulness of several commercial DNA transfection systems in human osteoclasts and their mononuclear precursor cell cultures, and compared transfection efficacy to adenoviral DNA transfection. None of the liposome-based or endosome disruption-inducing systems could induce EGFP-actin expression in terminally differentiated osteoclasts. Instead, a massive cell death by apoptosis was found with all concentrations and liposome/DNA-ratios tested. Best transfection efficiencies were obtained by adenoviral gene delivery. Marginal DNA transfection was obtained by just adding the DNA to the cell culture medium. When bone marrow-derived CD34-positive precursor cells were transfected, some GFP-expression was found at the latest 24 h after transfection. Large numbers of apoptotic cells were found and those cells that remained alive, failed to form osteoclasts when cultured in the presence of RANKL and M-CSF, key regulators of osteoclast formation. In comparison, adenoviral gene delivery resulted in the transfection of CD34-positive cells that remained GFP-positive for up to 5 days and allowed osteoclast formation.
Conclusion
Osteoclasts and their precursors are sensitive to liposomal transfection systems, which induce osteoclast apoptosis. Gene transfer to mononuclear osteoclast precursors or differentiated osteoclasts was not possible with any of the commercial transfection systems tested. Osteoclasts are non-dividing, adherent cells that are difficult to grow as confluent cultures, which may explain problems with transfection reagents. Large numbers of αvβ3 integrin on the osteoclast surface allows adenovirus endocytosis and infection proceeds in dividing and non-dividing cells efficiently. Viral gene delivery is therefore currently the method of choice for osteoclast transfection.
==== Body
Background
Osteoclasts are bone-resorbing cells that are highly polarized when physiologically active [1]. Their mononuclear precursors are hematopoietic in origin, and remain non-adherent in culture until they differentiate further from the multipotent cell lineage [2,3]. Monocytes, macrophages and osteoclasts derive from the same precursor cells [4]. Multinuclear osteoclasts are formed by fusion of their committed mononuclear precursor cells and RANKL is the major growth factor inducing osteoclast formation [5]. Osteoclast morphology and activity is highly dependent on the matrix that they are cultured on, bone being their natural substrate. Mature osteoclasts undergo several cycles of activation and inactivation, where bone is resorbed in the active state and cells migrate in the resting state. Eventually, the cells die apoptotically and, in vivo, new bone formation by osteoblastic cells takes place to fill the resorption lacuna.
Cell transfection is used in biomedical research to study the role of individual gene products in vitro or in vivo. Viral and non-viral gene transfer systems are available from several suppliers, and several cell lines and primary cells can efficiently be transfected [6,7]. Physiological barriers, including the plasma membrane, still cause transfection difficulties with distinct cell types. Cell-surface glycosaminoglycans inhibit transfection in vitro [8], suggesting that efficient gene transfer is as a sum of many positively affecting parameters. Inside cells, DNA needs to escape from the endosomes before their maturation into lysosomes [9]. Cell-specific targeting of gene transfer particles would also be beneficial, and manipulating the gene transfer complexes by adding targeting proteins or peptides is currently under research [10].
When plasmid DNA is transfected to cells, it needs to be transported to the nucleus to reach the transcription machinery [11,12]. Nuclear transport may be achieved either during mitosis when the nuclear membrane becomes disrupted or by transport through the nuclear pores. Transfection of non-dividing cells may be obtained by activating nuclear uptake by inserting nuclear localization signals into the transgene [13,14].
Adenoviral gene transfer into osteoclasts has been shown to work well [15]. This is probably due to the numerous αvβ3 integrin receptors that are located on the osteoclast plasma membrane [16]. Reports describing non-viral transfection on mature, adherent osteoclasts have not been found. There are also reports describing transfection of macrophages, like RAW264.7, that have after non-viral gene transfer been induced to form multinuclear giant cells [17]. It still remains controversial, however, whether these cells are polykaryons or truly osteoclasts capable of bone resorption. Due to a wish to study osteoclast migration and bone matrix removal in a more physiological context, we cultured osteoclasts and their early mononuclear precursors on bone and used these cultures for transfection. Earlier work in our laboratory suggested that other conventional transfection methods like calcium phosphate, DEAE-Dextran, electroporation, scrape-loading and hypotonic shock cannot be used. In the current paper we present data on the unsuccessful use of liposomal systems in the transfection of mature human osteoclasts and their mononuclear precursors in vitro.
Results
Transfection reagent-DNA ratio
Transfection reagents have specific reagent-to-DNA ratios that affect transfection efficiency and toxicity. In order to determine which ratios to use in the following experiments, we decided to test three ratios. On the basis of the morphological analysis of the cells, one test ratio was chosen for further analysis. Although disappointing at this stage, a more detailed study was continued to determine whether decreasing incubation time after transfection would allow transgene expression.
Apoptosis index
Cell death is the major problem encountered when using liposomal transfection systems. Therefore we counted the number of apoptotic cells from Hoechst staining using a conventional fluorescence microscope. Cultured osteoclasts were incubated with the transfection reagents for 2 h, followed by a 4 h, 8 h or 24 h culture period. In the baseline control, where no transfection reagents or adenoviruses were added, only some apoptotic nuclei were found and multinuclear osteoclasts remained polarized and active, as determined by actin ring morphology (Figure 1, [18]) and resorption activity measurements (Figures 2 and 3). When samples treated with the transfection reagents were evaluated, large numbers of apoptotic nuclei were seen and only some nuclei remained unfractionated (Figure 4). Intact osteoclasts could not be found in these samples, and resorption activity was totally lost. The lack of a dose-response suggests that even smaller amounts of liposomes or PEI were not tolerable to the osteoclasts. Some apoptotic nuclei were also seen in the adenovirus-treated samples, but the majority of the nuclei remained intact and many osteoclasts remained actively resorbing bone.
Figure 1 Visualization of actin rings and TRACP-positive cells in osteoclast cultures. Osteoclasts were differentiated in the presence of RANKL, M-CSF and TGF-β1 for 7 days, followed by fixation and staining of actin rings (a-c) and TRACP (d). Baseline control is shown in a and d, and adenovirus-infected cells 4 h post infection are shown in b. A typical view of the cells incubated 2 h with transfection reagents and 4 h in fresh medium is shown in c.
Figure 2 Number of actin rings in osteoclast cultures. Cells were treated with transfection reagents for 2 h, followed by culture for 4 h, 8, or 24 h. Cells were stained with phalloidin and number of actin rings was counted to quantitate actively resorbing osteoclasts. BL, baseline with no additions; Ad, adenoviral infection of GFP; T1-T8, transfection reagents as shown in Tables 1 and 2. ANOVA: p < 0,001
Figure 3 Number of new resorption pits in osteoclast cultures. Bone slices were biotinylated before cells were treated with transfection reagents for 2 h, followed by culture for 4 h, 8, or 24 h. Biotinylated resorption pits were visualized with FITC-labelled streptavidin and all resorption pits were stained with TRITC-WGA lectin. Resorption occurring after transfection was determined as pits emitting only red fluorescence. BL, baseline with no additions; Ad, adenoviral infection of GFP; T1-T8, transfection reagents as shown in Tables 1 and 2. The baseline control shown in the insert shows the staining pattern of the resorption pits before transfection (green) and overall resorption activity during the whole culture period (red). Yellow colour determines areas where both fluorochromes overlap. ANOVA: p < 0,001
Figure 4 Apoptosis index in osteoclast cultures. Cells were treated with transfection reagents for 2 h, followed by culture for 4 h, 8, or 24 h. Nuclei were stained with Hoechst and apoptotic osteoclasts were counted with a fluorescence microscope. BL, baseline with no additions; Ad, adenoviral infection of GFP; T1-T8, transfection reagents as shown in Tables 1 and 2. ANOVA: p < 0,001
Viability assay
In order to determine whether any combination of transfection reagent concentration and incubation time would allow cell survival, we cultured osteoclasts on 96 well plates and measured dead and live cell fluorescence with a microplate reader. As can be seen from Figure 5, we could not avoid killing cells with the transfection reagents. When the samples were monitored in more detail after cytochemical staining for the osteoclast marker enzyme TRACP [19], it became evident that a total loss of osteoclasts occurred already after a 1 h treatment with transfection reagents. Adenoviral gene delivery also resulted in osteoclast death and decreased viability, but the majority of the cells remained alive and many cells expressed the transgene.
Figure 5 Viability index in osteoclast cultures. Cells were treated with transfection reagents for 2 h, followed by culture for 4 h, 8, or 24 h. Osteoclast differentiation cultures were performed on collagen-coated plates to allow the use of the microplate reader. After transfection, cells were stained with Calcein AM and EthD and fluorescence of the dyes was measured using appropriate band pass filters. BL, baseline with no additions; Ad, adenoviral infection of GFP; T1-T8, transfection reagents as shown in Tables 1 and 2. ANOVA: p < 0,001
We also wanted to check if it would be possible to transfect the non-adherent CD34-positive mononuclear cells and then induce osteoclast differentiation. The Live-Dead assay was thus performed also with the mononuclear precursor cells. As can be seen from Figure 6, the viability indexes remained somewhat higher but far too low as compared to the baseline control or to the adenovirus-treated samples.
Figure 6 Viability index in CD34-positive mononuclear cell cultures. Cells were treated with transfection reagents for 2 h, followed by culture for 4 h, 8, or 24 h. CD34-positive cells were grown on collagen-coated plates and after transfection, cells were stained with Calcein AM and EthD. Fluorescence of the dyes was measured using the microplate reader and appropriate filter sets. BL, baseline with no additions; Ad, adenoviral infection of GFP; T1-T8, transfection reagents as shown in Tables 1 and 2. ANOVA: p < 0,001
Transfection efficiency
GFP expression was followed in adherent osteoclasts and in non-adherent mononuclear precursors transfected for 4 hours and cultured in fresh medium for 1 h, 24 h, 48 h or 5 days. No GFP expression was noticed in osteoclasts after transfection with any of the transfection systems tested (Table 1). In comparison, adenoviral delivery of the transgene resulted in a 15% transfection efficiency of multinuclear osteoclasts. When CD34-positive non-adherent precursor cells were transfected, some cells were positive 24 h and 48 h after transfection, but no positive cells were seen on day 5 with any of the transfection reagents tested (Table 2). In the adenovirus-infected cultures, multiple GFP expressing cells was seen 24 h and 48 h after infection and some cells also 5 days after infection. These data suggest that transfection of the osteoclasts or the mononuclear precursor cells was not feasible with the conventional transfection methods.
Table 1 Transfection efficiency (% of live cells) in mature osteoclast cultures
1 h 24 h 48 h 5 d
Baseline 0 0 0,12 ± 0,042 0
Adenovirus 1,2 ± 0,095 9,7 ± 1,4 15,2 ± 3,4 3,8 ± 0,96
T1: Metafectene 0 0 0 0
T2: Lipofectamine plus 0 0 0 0
T3: Tfx-50 0 0 0 0
T4: FuGene6 0 0 0 0
T5: DOTAP 0 0 0 0
T6: DOSPER 0 0 0 0
T7: JetPei 0 0 0 0
T8: DuoFect 0 0 0 0
GFP-expressing and negative osteoclasts were counted using fluorescence microscopy and phase optics, and transfection efficiencies were counted. ANOVA: p = 1,4 × 10-8, n = 5.
Table 2 Transfection efficiency (% of live cells) in CD34-positive mononuclear cell cultures
1 h 24 h 48 h 5 d
Baseline 0 0,13 ± 0,031 0,47 ± 0,08 0
Adenovirus 2,7 ± 0,12 13,1 ± 1,4 22,6 ± 3,4 4,2 ± 0,96
T1: Metafectene 0 0,12 ± 0,042 0,23 ± 0,053 0
T2: Lipofectamine plus 0 0,24 ± 0,060 0,30 ± 0,11 0
T3: Tfx-50 0 0,10 ± 0,037 0,41 ± 0,091 0
T4: FuGene6 0 0,23 ± 0,071 0,22 ± 0,13 0
T5: DOTAP 0 0 0,56 ± 0,064 0
T6: DOSPER 0 0,16 ± 0,046 0,28 ± 0,13 0
T7: JetPei 0 0,20 ± 0,050 0,50 ± 0,15 0
T8: DuoFect 0 0 0,12 ± 0,056 0
GFP-expressing and negative osteoclasts were counted using fluorescence microscopy and phase optics, and transfection efficiencies were counted. ANOVA: p = 2,3 × 10-11, n = 5.
Discussion
Osteoclasts are cells that need to be cultured as primary cells or as a differentiation culture from bone marrow-derived mononuclear precursor cells. The natural substrate of osteoclasts is bone, and seeding the cells on a non-natural substrate, like plastic or glass, has a major effect on the regulation of gene expression and cell morphology [18,20]. Therefore we aimed at transfecting multinuclear osteoclasts adhered to bovine cortical bone, a widely used system in osteoclast research. Adenoviral transfection of osteoclasts was used in this study as the reference gene transfer system, while it has been shown to work also with osteoclasts [15]. CAR-receptor bound adenoviruses are internalized via endocytosis after attachment to αv integrins, which are widely distributed on the osteoclast surface. Although viral gene delivery is at it's best very efficient and rapid, a strong promoter may drive excessive transgene production and interfere with normal cell physiology. The use of human pathogens, like adeno- and lentiviruses, also requires special attention and authorization, while conventional transfection methods can be used in any laboratory.
Commercial modifications of liposomal gene delivery systems and PEI-dependent endosomal disruption systems were systematically evaluated to determine whether any of the concentration-incubation time combinations would result in osteoclast transfection. To our disappointment, however, none of the 8 transfection systems could provide satisfactory osteoclast transfection efficiency. GFP-tagged actin was used as the transgene for easy monitoring of gene transfer, but no transfected osteoclasts were noticed. Adenoviral gene delivery was the only method capable of providing sufficient transfection efficiency. Among the non-adherent mononuclear precursor cells, an equally poor transfection rate was obtained. The most striking effect was the vast induction of apoptosis with both cationic liposomes and with PEI-dependent endosomal proton sponges. When uptake of the transfection reagent-packed DNA into the cells was monitored in more detail, it could be noted that most of the molecules never penetrated the plasma membrane. It was recently shown that cell-surface glycosaminoglycans are capable of inhibiting transfection [8]. The osteoclast plasma membrane is coated with large amounts of hyaluronic acid and other glycoproteins (for review see [21]), and this may explain why the transfection reagents are unable to deliver their cargo to the plasma membrane. Another explanation for the lack of transfection may be the low cell density. While commercial transfection reagents are suggested to be used in sub-confluent to confluent cell cultures, our osteoclast cultures were appr. 50% confluent (Figure 1d). Our cells were non-dividing, and this may also contribute to the transfection difficulties.
Mature osteoclasts cannot be grown as suspension cultures and confluency is difficult to control. However, osteoclasts take up plasma membrane-impermeable DNA- and RNA molecules from culture medium [22-24]. For antisense and siRNA-research, it would be optimal to increase the uptake and intracellular availability of gene knockdown-molecules in osteoclast cultures. While viral gene transfer is difficult to control, the primary choice for gene knockdown experiments would be a non-viral system that allows transgene packaging, protection and sufficient bioavailability.
Conclusion
Although many cell lines and some primary cells are easy to transfect using calcium phosphate, DEAE-dextran, electroporation, scrape loading or liposomal transfection systems, these systems cannot be used on multinuclear osteoclasts. These large, adherent, non-dividing cells are fragile and undergo apoptosis rapidly when challenged chemically or mechanically. Optimal cells for commercial transfection systems should be in sub-confluent, rapidly dividing growth phase, which cannot be provided in osteoclast cultures. Microinjection may be used for osteoclast transfection, if only a few transfected osteoclasts are enough and the expertise is available. For proper transfection of higher numbers of osteoclasts, however, the only rational tools are the viral delivery systems.
Materials and methods
Cell culture
Human bone marrow-derived CD34-positive mononuclear cells were cultured on bovine cortical bone slices in the presence of M-CSF (33 ng/ml, R&D Systems, UK) and RANKL (66 ng/ml, Peprotech, UK) as suggested by the supplier (Cambrex, USA). TGF-β1 (1 ng/ml, R&D Systems, UK) was added on day 3, and adherent, terminally differentiated osteoclasts were transfected on day 7. When non-adherent osteoclast precursors were used, the transfections were performed on day 1. Cells were cultured in high-glucose DMEM supplemented with 10% heat-inactivated fetal calf serum, 20 mM HEPES, 100 U/ml penicillin and 100 mg/ml streptomycin (all from Gibco Invitrogen, UK). Cells were grown in 96 well plates with 200 μl of medium for fluorescence measurements with a plate reader. Bovine cortical bone slices were 150-180 μm thick transversal sections that were sonicated and sterilized by dipping in 70% ethanol before use. A control group of cells attached to glass coverslips coated with type I collagen (BD Biosciences, Belgium) was also included. Non-attached cells were transfected in wells containing type I collagen-coated glass coverslips or bone slices.
Transfection systems
The plasmid containing EGFP-actin (Clontech, USA) was transfected to the cells to allow fluorescent visualization of transfected actin filaments. For liposome-mediated transfection, Metafectene (Biontex, USA), Lipofectamine Plus (Gibco Invitrogen, UK), Tfx-50 (Promega Corp, USA) and FuGene6, DOTAP and DOSPER (all from Roche, Germany) were used according to the supplier's instructions. Reagent/DNA ratios were as follows: 1 μg plasmid DNA was complexed with 1.5, 3.0 or 6.0 μl of FuGene6 or Lipofectamine Plus transfection reagent; or with 2.0, 3.0 or 4.0 μl of Tfx-50 or Metafectene transfection reagent; or with 5, 7.5 or 10 μg of DOTAP; or with 3, 7.5 or 12 μg of DOSPER. Also the endosomal disruption-based transfection systems JetPei (PolyTransfection, USA) and DuoFect (Quantum Appligene, USA) were used according to the manufacturer's instructions. For DuoFect transfection, 50 μM deferrioxamine was added to the culture medium 24 h before transfection. With these systems, 1 μg plasmid DNA was complexed with 0.5, 0.75 or 1.0 μl of DuoFect transfection reagent or with 1.5, 3 or 4.5 μl of JetPei transfection reagent.
To test the optimal transfection reagent-to-DNA ratio, cells were incubated with transfection reagents for 2 h the presence of serum, dipped in warm PBS and transferred onto fresh culture plates containing medium and osteoclast growth factors for an additional culture period of 48 h. Cell morphology and transgene expression were monitored microscopically and the following reagent-to-DNA ratios were chosen to be used in the future experiments: 1 μg plasmid DNA was complexed with 3.0 μl of FuGene6, Lipofectamine Plus, Tfx-50 or Metafectene transfection reagent; or with 7.5 μg of DOTAP or DOSPER; or with 1.0 μl of DuoFect; or with 4.5 μl of JetPei transfection reagent. In the following experiments, cells were incubated with transfection reagents for 2 h in the presence of serum, dipped in warm PBS and transferred onto fresh culture plates containing medium and osteoclast growth factors for an additional culture period of 4 h, 8 h or 24 h. Transgene expression and cell viability were evaluated with help of a fluorescence microscope (Leica) and a microplate reader (Victor2, Wallac).
A commercial adenovirus resulting in the expression of GFP under the CMV promoter was used as the transfection control (QBiogene, USA). Cells were infected with 5000 virus particles of Ad5.CMV-GFP in 100 μl medium for 1 h, after which 100 μl of fresh medium and osteoclast growth factors were added. GFP expression and cell viability was evaluated as already described.
Transfection efficiency and viability
Transgene expression in the cells was monitored under fluorescence microscope 1 h, 24 h, 48 h and 5 days after transfection, and all GFP-positive mononuclear cells and osteoclasts (cells with at least 3 nuclei) were counted. For counting apoptotic cells, 3% paraformaldehyde-2% sucrose was used for fixing the cells prior to staining nuclei with Hoechst as suggested by the supplier (Molecular Probes, USA). Apoptotic nuclei were counted under fluorescence microscope. To monitor cell viability in detail, we stained dead and live cells with the Live/Dead-system (Molecular Probes, USA). Cells grown on 96 well plates were stained after transfection by adding 7 μM Calcein AM (stained live cells) and 5 μM ethidium homodimer-1 (EthD, detected dead cells) to the cell cultures that were washed with warm PBS. Cells were incubated with the dyes for 45 min in 100 μl PBS, followed by fluorescence intensity measurements using exitation/emission filter sets of 495/520 nm (Calcein AM) and 530/642 nm (EthD). Viability indexes were counted by dividing the live cell fluorescence by the dead cell fluorescence.
Morphological analysis
The effects of transfection reagents on the morphology of cultured cells were monitored during culture with phase optics, and more detailed morphological analysis was performed on fixed samples. Cells were fixed in 3%PFA-2% sucrose for 15 min. To monitor confluency and osteoclast formation capacity in the cultures, cells were fixed and stained for TRACP with the Leukocyte Acid Phosphatase kit (Sigma, USA). Bone resorbing osteoclasts were determined by actin ring staining with AlexaFluor488 Phalloidin (Molecular Probes, USA). Resorption activity was monitored in the samples by biotinylating the existing resorption pits immediately before transfection with sulfo-NHS-biotin (Pierce, USA) as described before [25]. After transfection and further culture, samples were fixed and biotin was detected with FITC-streptavidin (DAKO, Denmark) and all resorption pits were stained with TRITC-WGA lectin (Sigma Aldrich, USA).
Statistical analysis
Data are expressed as mean ± SD of four replicas and all experiments were independently performed twice (n = 8). Differences from the control were examined for statistical significance by analysis of variance and student's T-test. A p-value less than 0,05 was considered significant.
Authors' contributions
TLL is responsible for the content of this article.
Acknowledgements
Jukka Rissanen and Salla Ylönen are acknowledged for technical assistance and prof. H. Kalervo Väänänen for scientific advice. This work was financially supported by the National Technology Agency of Finland.
==== Refs
Väänänen HK Zhao H Mulari M Halleen JM The cell biology of osteoclast function J Cell Sci 2000 113 377 381 10639325
Suda T Takahashi N Martin TJ Modulation of osteoclast differentiation Endocr Rev 1992 13 66 80 1555533 10.1210/er.13.1.66
Blair HC Athanasou NA Recent advances in osteoclast biology and pathological bone resorption Histol Histopathol 2004 19 189 199 14702187
Athanasou NA Quinn J Immunophenotypic differences between osteoclasts and macrophage polykaryons: immunohistological distinction and implications for osteoclast ontogeny and function J Clin Pathol 1990 43 997 1003 2266187
Lacey DL Timms E Tan HL Kelley MJ Dunstan CR Burgess T Elliott R Colombero A Elliott G Scully S Osteoprotegerin ligand is a cytokine that regulates osteoclast differentiation and activation Cell 1998 93 165 176 9568710 10.1016/S0092-8674(00)81569-X
Zhang G Budker VG Ludtke JJ Wolff JA Naked DNA gene transfer in mammalian cells Methods Mol Biol 2004 245 251 264 14707386
Banerjee R Mahidhar YV Chaudhuri A Gopal V Rao NM Design, synthesis and transfection biology of novel cationic glycolipids for use in liposomal gene delivery J Med Chem 2001 44 4176 4185 11708919 10.1021/jm000466s
Ruoponen M Honkakoski P Tammi M Urtti A Cell-surface glycosaminoglycans inhibit cation-mediated gene transfer J Gene Med 2004 6 405 414 15079815 10.1002/jgm.522
Xu Y Szoka FC Jr Mechanism of DNA release from cationic liposome/DNA complexes used in cell transfection Biochem 1996 35 5616 5623 8639519 10.1021/bi9602019
Sokoloff AV Wong SC Ludtke JJ Sebestyen MG Subbotin VM Zhang G Budker T Bachhuber M Sumita Y Wolff JA A new peptide ligand that targets particles and heterologous proteins to hepatocytes in vivo Mol Ther 2003 8 867 782 14664787 10.1016/j.ymthe.2003.09.009
Lechardeur D Lukacs GL Intracellular barriers to non-viral gene transfer Curr Gene Ther 2002 2 183 194 12109215
Lukacs GL Haggie P Seksek O Lechardeur D Freedman N Verkman AS Size-dependent DNA mobility in cytoplasm and nucleus J Biol Chem 2000 275 1625 1629 10636854 10.1074/jbc.275.3.1625
Ludtke JJ Zhang G Sebestyen MG Wolff JA A nuclear localization signal can enhance both the nuclear transport and expression of 1 kb DNA J Cell Sci 1999 112 2033 2041 10341220
Aronsohn AI Hughes JA Nuclear localization signal peptides enhance cationic liposome-mediated gene therapy J Drug Target 1998 5 163 169 9606006
Tanaka S Takahashi T Takayanagi H Miyazaki T Oda H Nakamura K Hirai H Kurokawa T Modulation of osteclast function by adenovirus vector-induced epidermal growth factor receptor J Bone Miner Res 1998 13 1714 1720 9797480
Nesbitt SA Helfrich M Horton M Biochemical characterization of human osteoclast integrins. Osteoclasts express alpha v beta 3, alpha 2 beta 1 and alpha v beta 1 integrins J Biol Chem 1993 268 16737 16745 8344953
Lean JM Jagger CJ Kirstein B Fuller K Chambers TJ Hydrogen Peroxide Is Essential for Estrogen-Deficiency Bone Loss and Osteoclast Formation Endocrinology 2005 146 728 735 15528306 10.1210/en.2004-1021
Lakkakorpi P Tuukkanen J Hentunen T Järvelin K Väänänen K Organization of osteoclast microfilaments during the attachment to bone surface in vitro J Bone Miner Res 1989 4 817 825 2692403
Minkin C Bone acid phosphatase: Tartrate-resistant acid phosphatase as a marker of osteoclast function Calcif Tissue Int 1982 34 285 290 6809291
Vääräniemi J Halleen JM Kaarlonen K Ylipahkala H Alatalo SA Andersson G Kaija H Vihko P Väänänen HK Intracellular machinery for matrix degradation in bone-resorbing osteoclasts J Bone Miner Res 2004 19 1432 1440 15312243
Prince CW Roles of hyaluronan in bone resorption BMC Muskuloskeletal Disorders 2004 5 12 10.1186/1471-2474-5-12
Laitala T Väänänen HK Inhibition of bone resorption in vitro by antisense RNA and DNA molecules targeted against carbonic anhydrase II or two subunits of vacuolar H+-ATPase J Clin Invest 1994 93 2311 2318 8200964
Laitala-Leinonen T Löwik C Papapoulos S Väänänen HK Inhibition of intravacuolar acidification by antisense RNA decreases osteoclast differentiation and bone resorption in vitro J Cell Sci 1999 112 3657 3666 10523502
Zhao H Parikka V Laitala-Leinonen T. Väänänen HK Impaired osteoclast polarization and inhibition of bone resorption induced by oligodeoxynucleotide targeted to small GTPase Rab 7 J Biol Chem 2001 276 39295 302 11514537 10.1074/jbc.M010999200
Salo J Bone resorbing osteoclasts reveal two basal plasma membrane domains and transcytosis of degraded matrix material PhD thesis 2002 University of Oulu, Department of Anatomy and Cell Biology
|
16124882
|
PMC1208940
|
CC BY
|
2021-01-04 16:37:37
|
no
|
J Negat Results Biomed. 2005 Aug 29; 4:5
|
utf-8
|
J Negat Results Biomed
| 2,005 |
10.1186/1477-5751-4-5
|
oa_comm
|
==== Front
Lipids Health DisLipids in Health and Disease1476-511XBioMed Central London 1476-511X-4-171614454910.1186/1476-511X-4-17ResearchDiet, lifestyle factors and hypercholesterolemia in elderly men and women from Cyprus Polychronopoulos Evangelos [email protected] Demosthenes B [email protected] Anna [email protected] Department of Dietetics – Nutrition, Harokopio University, Athens, Greece2005 6 9 2005 4 17 17 26 8 2005 6 9 2005 Copyright © 2005 Polychronopoulos et al; licensee BioMed Central Ltd.2005Polychronopoulos et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
We sought to investigate the single and combined effect of Mediterranean diet, being physically active, moderate alcohol use, and non-smoking on clinical status of 150 elderly people from Cyprus.
Methods
The study comprises individuals enrolled in surveys from Greece and Cyprus. This work includes 53 apparently men and 97 women, aged 65 to 100 years, from various areas of Cyprus. The cohort study was conducted between 2004 and 2005. A diet score that assesses the inherent characteristics of the Mediterranean diet was developed (range 0–55) and then a healthy index was calculated that evaluated four lifestyle habits (range 0 – 4), i.e. non-smoking, alcohol intake, physical activity and adherence to the Mediterranean diet (i.e. above the median of the score).
Results
65% participants had hypercholesterolemia (total serum cholesterol > 200 mg/dl or use of lipid lowering agents). Moreover, 32% of the participants reported physically active, 5% reported smoking habits and 4% that they have stopped smoking during the past decade, while 8% reported alcohol drinking. A positive association was observed between prevalence of hypercholesterolemia and smoking habits (odds ratio = 4.3, p = 0.03), while an inverse association was observed between hypercholesterolemia, alcohol drinking (odds ratio = 0.3, p = 0.04) and adherence to a Mediterranean diet (odds ratio = 0.77, p = 0.02), controlled for age, sex, and other factors.
Conclusion
Adherence to a Mediterranean diet and healthful lifestyle is associated with reduced odds of having hypercholesterolemia among elderly people.
==== Body
Introduction
During the past years several observational studies and clinical trials have revealed the adverse effect of abnormal blood lipid levels on the progression of atherosclerosis, and consequently the development of cardiovascular disease [1-5]. However, it should be noted that the strength of the relationships between blood lipids levels and atherosclerosis might be influenced by several lifestyle-related factors, like smoking, physical activity and psychosocial conditions. In particular, the Framingham Heart Study first reported that glucose intolerance, blood pressure levels and smoking habits modify the effect of total cholesterol on cardiovascular risk [1]. It is also well known that dietary patterns, like the Mediterranean diet, are strongly related with blood lipids levels, as well as with the prevalence and the management of dyslipidemia [6-10]. Recently, Yusuf et al. [11] reported that, among others, smoking, dietary habits and alcohol intake, as well as regular physical activity account for most of the risk of myocardial infarction worldwide in both sexes and at all ages in all regions. Moreover, in elderly people cardiovascular disease is the leading cause of death around the world, while hypercholesterolemia is among the major risk factors for the development of the disease. However, very few studies have evaluated the role of diet and other lifestyle-related factors in the prevalence of hypercholesterolemia in the elderly.
Given the lack of current data regarding the levels of blood lipids in Cyprus population, we investigated the distribution, awareness and management of high blood lipids levels, in a random sample of 65 years and older adults without any clinical evidence of cardiovascular disease. Moreover, we evaluated the association of dietary habits, smoking and physical activity with their blood lipid levels.
Methods
Population of the study
This study is a health and nutrition survey, which is being carried out in various areas of Cyprus (Arsos, Lemessos, Pachna, Pafos, Empa, Kallepia, Yeroskipou). From November 2004 to May 2005, 188 men and women, 65 years and older, and without any clinical history of cardiovascular disease were sere selected to participate in the study. Of them, 53 men and 97 women were finally agreed to participate (80% participation rate). People living in institutions were excluded from the study. The sampling was random and multistage according to the population of each city. All participants interviewed by specialists who used a standard questionnaire. The number of enrolled participants is adequate to evaluate standardised differences between the investigated parameters greater than 0.5, achieving statistical power >0.80 at <0.05 probability level (P-value).
In this work a special attention was given to people with hypercholesterolemia. In particular, hypercholesterolemia was defined as total serum cholesterol levels greater than 200 mg/dl or the use of lipid-lowering agents. Moreover, we recorded from participants' medical records high and low density lipoprotein (HDL, LDL) cholesterols, and triglycerides levels, as well as any special treatment.
Dietary assessment
Consumption of 15 food groups was measured as an average per week during the past year through a semi-quantitative food-frequency questionnaire [12]. The frequency of consumption was then quantified approximately in terms of the number of times a month a food was consumed. Alcohol consumption was measured by daily ethanol intake, in wineglasses (100 ml and 12% ethanol concentration). Then we developed a dietary score according to the Mediterranean dietary pattern which consists of: (a) daily consumption: of non refined cereals and products (whole grain bread, pasta, brown rice, etc), vegetables (2 – 3 servings/day), fruits (6 servings/day), olive oil (as the main added lipid) and dairy products (1 – 2 servings/day), (b) weekly consumption: of fish (4 – 5 servings/week), poultry (3 – 4 servings/week), olives, pulses, and nuts (3 servings/week), potatoes, eggs and sweets (3 – 4 servings/week) and monthly consumption: of red meat and meat products (4 – 5 servings/month). In particular, for the consumption of food items that are close to this dietary pattern we assigned score 0 for rare or no consumption, 1 for 1 to 4 times/month, 2 for 5 to 8 times, 3 for 9 to 12 times/month, 4 for 13 to 18 times/month and 5 for almost daily consumption. On the other hand, for the consumption of foods that are away from this traditional diet, like meat and meat products, we assigned the opposite scores (i.e. 0 for almost daily consumption to 5 for rare or no consumption). Higher values of the suggested dietary score indicates adherence to the traditional Mediterranean diet (i.e. which is also characterized by moderate consumption of fat and high monounsaturated: saturated fat ratio).
Lifestyle habits
To evaluate physical activity status of the patients during the past year we used a modified short version of a self-reported questionnaire, the International physical activity questionnaire (IPAQ) for the elderly [13]. Based on this questionnaire we assessed the frequency (times per week), duration (in minutes per time) and intensity of sports or occupation related physical activity. Participants who did not report any physical activities were defined as sedentary. For the rest of the participants we calculated a combined score by multiplying the weekly frequency, duration and intensity of physical activity. The upper tertile of the score classified participants as "highly" physical active, the medium tertile as "moderately" active and the lowest tertile as "low" physical active.
Finally, current smokers were defined as those who smoked at least one cigarette per day or have stopped cigarette smoking during the past 12 months. Former smokers were defined as those who had stopped smoking more than one year previously. The rest of them were defined as never smokers or rare smokers.
The healthy index
To evaluate status of the participants we have developed a healthy index using dietary habits, smoking and physical activity status. In particular, participants who had diet score greater than the median (i.e. 36 for men and women) had a contribution to the healthy index score equal to 1, while the rest had a contribution equal to 0. Especially for alcoholic beverages intake we assigned in the healthy index the value of 1 for consumption of 3 or less wineglasses per day and the value of 0 for none or consumption of more than 3 wineglasses per day. Regarding the contribution of physical activity status in the healthy index we assigned the value of 1 for those who were in the middle or upper tertile of the physical activity score and the value of 0 in the rest of them. We have also assigned the value of 1 in the healthy index score into those who reported non-smokers or have stopped smoking for at least 15 years and the value of 0 into those who reported current or former smokers. Then we calculated the healthy lifestyle index by summing the individuals' scores for diet, alcohol intake, smoking and exercise. Thus, a 0 to 4 index was developed.
Other measurements
The study's questionnaire also included demographic characteristics like age, gender, financial status (average annual income during the past three years) and education level. The educational level of the participants (as a proxy of social status) was measured by the years of school. Body mass index was calculated as weight (in kilograms) divided by standing height (in meters squared). Obesity was defined as body mass index > 29.9 Kg/m2. Participants who reported blood pressure levels were greater or equal to 140/90 mm Hg or were under antihypertensive medication were classified as hypertensives. Diabetes mellitus as a blood sugar > 125 mg/dl or the use of antidiabetic medication.
Statistical analysis
Continuous variables are presented as mean values ± standard deviation. The categorical variables are presented as absolute and relative (%) frequencies. Associations between continuous variables and group of patients were evaluated through the analysis of variance (ANOVA), after controlling for equality of variances (homoscedacity). Due to multiple comparisons we applied the Bonferroni correction to correct for the inflation of Type-I error. Associations between categorical variables were tested by the use of the chi-squared test, without the correction of continuity. Correlations between continuous variables were tested by the use of Pearson's correlation coefficient. Multiple logistic regression analysis evaluated the association of the healthy index as well as individual scores on the likelihood of having hypercolesterolemia. All statistical calculations were performed on the SPSS version 12.0 software (SPSS Inc, Chicago, IL, U.S.A.).
Results
Demographic, clinical and behavioural characteristics of the participants are presented in Table 1.
Table 1 Socio-demographic and lifestyle characteristics of the participants (% by gender)
Men (n = 53) Women (n = 97) P
Age (years) 79 ± 8 75 ± 7 0.003
Education status (years of school) 5.7 ± 2 4.0 ± 2 0.001
Physical inactivity (%) 59 73 <0.001
Smoking habits (%) <0.001
Current smoker 11 1
Former smoker 47 1
Never smoker 42 98
Marital status (%) <0.001
Never married 2
Married 90 56
Divorced 4
Widowed 10 38
Diet score (range 0 – 55) 36 ± 3 35 ± 3 0.15
Alcohol consumption (%) 15 4 0.02
Sixty five percent of the participants had hypercholesterolemia (60% of men and 68% of women, p = 0.34). Total serum cholesterol levels were 217 ± 38 mg/dl in men and 234 ± 42 mg/dl in women (p = 0.046). In addition to this information, 16% of men and 42% of women had cholesterol levels above 240 mg/dl (p < 0.001). Treatment of hypercholesterolemia was as follows: 37% of men and 53% of women were on special diet (p = 0.19), 81% of men and 63% of women on pharmaceutical treatment, i.e. statin and/or diet (p = 0.12) and 8% of men and 7% of women were untreated (p = 0.89). It is of interest that 90 out of 150 (60%) participants were unaware about the normal limits for total cholesterol levels. In addition, HDL cholesterol levels were 47 ± 9 mg/dl in men and 57 ± 14 mg/dl in women (p = 0.016), LDL cholesterol levels were 134 ± 19 mg/dl in men and 146 ± 32 mg/dl in women (p = 0.20) and triglycerides levels were 167 ± 75 mg/dl in men and 141 ± 62 mg/dl in women (p = 0.09). We also observed that 46% of men and 18% of women had low HDL-cholesterol levels (<40 mg/dl), 23% of men and 27% of women had high LDL-cholesterol levels (>100 mg/dl) and 52% of men and 40% of women had high triglycerides levels (>150 mg/dl). A group of people with particular interest is those who have normal total cholesterol, but low HDL cholesterol levels. In our population, 27% of men and women who had normal total cholesterol levels (i.e. <200 mg/dl) had HDL cholesterol levels lower than 35 and 45 mg/dl, respectively.
A positive association was observed between prevalence of hypercholesterolemia and smoking habits (p = 0.03), hypertension (p = 0.07), obesity (p = 0.06), while an inverse association was observed between hypercholesterolemia and alcohol drinking (p = 0.04). No associations were observed between hypercholesterolemia and financial status (p = 0.46), physical activity (p = 0.38) and presence of diabetes (p = 0.45).
It is known that age is a factor that correlates well with blood lipid levels. In our study, age was positively and significantly associated with all blood lipids measurements in both men and women. Thus further analysis confirmed previous findings between genders, after adjusting for age and controlling for other potential confounders that correlate with age and sex, like smoking habits, physical activity status and dietary habits (results not shown in Tables or Figure).
Since several investigators claimed for the effect of social status on blood lipid levels we evaluated the distribution of lipids in relation to years of school as a proxy of socio-economic level of the participants. After controlling for several potential confounders, like age, sex, body mass index, physical activity status, smoking and dietary habits, we found that none of the cholesterol levels had a consistent positive association with education status (data not presented in text or Tables). Moreover, no statistically significant associations were observed between blood lipids levels and income of the participants.
Blood lipids and "healthy lifestyle status"
A secondary goal of this work was to assess blood lipid levels with dietary habits and other lifestyle characteristics. At first we evaluated the effect of Mediterranean diet on the investigated lipids. We revealed that greater adherence to Mediterranean diet (i.e. diet score > median value of 36) was associated with 23% lower likelihood of having hypercholesterolemia (odds ratio = 0.77, p = 0.02), after controlling for age, sex, body mass index, smoking habits and physical activity status. Afterwards we focused our interest on participants who were under lipid-lowering treatment and also adopted the Mediterranean diet. We found that the aforementioned combination was associated with 26% reduction in total serum cholesterol levels (p =< 0.001) and a 29% reduction in LDL cholesterol levels (p = 0.001), compared to those who were untreated and away to the Mediterranean diet. Of particular interest is that the observed reductions were higher than the reductions achieved by diet or statin treatment alone (p for interaction < 0.05). The effect of this combination on the other rest lipids was not statistically significant.
The association of the "healthy index" on the likelihood of having hypercholesterolemia is presented in Table 3. As we can see presence of 2 or more protective factors (i.e. healthy index > 1) seems to be associated with about 53% lower risk of having hypercholesterolemia. The calculation of the population attributable risk was showed that from 20% to 31% of hypercholesterolemic people could be prevented through this healthy lifestyle pattern. However, when we evaluated the components of the "healthy index" (Mediterranean diet, physical activity, moderate alcohol drinking and abstinence from smoking) on the prevalence of hypercholesterolemia no significant associations were observed. By the exception of Mediterranean diet mentioned above.
Table 3 Results from logistic regression analysis that evaluated the diet, physical activity, alcohol intake and smoking (healthy index) on the likelihood of having hypercholesterolemia
Healthy index OR (95% CI) PAR (%) P
# protective factors 0 (reference category) 1.00
1 0.33; 0.05 to 3.06 25 0.33
2 0.47; 0.22 to 0.97 20 0.01
3 0.20; 0.08 to 0.41 30 0.001
4 0.15; 0.07 to 0.29 31 0.001
OR = odds ratio; PAR = population attributable risk.
Variables also entered in the model were age, sex, body mass index, history of hypertension, and diabetes mellitus.
Discussion
In this work we evaluated the distribution of blood lipid in a random, population-based, sample of elderly people from Cyprus. We observed that roughly six out of ten participants had high blood lipid levels. Although the latter finding may be influenced by the compliance to medication and other confounders, it is of great importance for the public health strategies in the studied population. Regarding the various dietary characteristics, we observed that adherence to the Mediterranean diet resulted a significant reduction on the likelihood of having hypercholesterolemia. In addition, we revealed the synergistic effect of Mediterranean diet with statin treatment in the management of blood lipids. Finally, a healthy lifestyle, including adherence to the Mediterranean diet, abstinence from smoking, physical activity, even in elderly people seems to be associated with a considerable reduction of the burden of hypercholesterolemia.
Epidemiology of blood lipids among older adults
There are very few epidemiological studies that have assessed blood lipids levels in the elderly. Based on the ATTICA study [14] that evaluated, among others, blood lipids among Greek adults the investigators reported that 48% of men and 55% of women, aged > 50 years, had high total cholesterol levels (i.e. >200 mg/dl). In another Mediterranean population, the Portuguese, Costa et al. [15] reported that the prevalence of total cholesterol levels > 200 mg/dl among elderly people was 57%. In the EPICARDIAN study [16], the investigators reported that the prevalence of hypercholesterolemia among elderly Spanish people was 68%. Moreover, roughly 50% of white adult men and women in USA had total blood cholesterol levels over 200 mg/dl, as reported in the National Health and Nutrition Examination Survey (NHANES) III study [17]. These reports are in accord with our findings since we observed that 6 out of ten men and 7 out of ten women had high total serum cholesterol levels. Studies show that a higher percentage of women than men have total blood cholesterol of 200 mg/dl or higher, beginning at age 50 [1-3]. The later was confirmed by our study, too.
Regarding HDL cholesterol there is a wide range of scientific evidence which suggests that it plays a role in the development of coronary heart disease, especially in the elderly [1,3]. The NHANES III study [16] reported that 18% of middle aged men and 6% of middle aged women had HDL cholesterol levels below 35 mg/dl. In our study, 46% of men and 18% of women had low HDL-cholesterol levels. The increased rates may attribute to the increased age of our sample. Studies suggest that even for those with normal levels of total cholesterol, risk for myocardial infarction is high when HDL cholesterol is low. We observed that 27% of men and women, who had desirable total cholesterol levels, had low HDL cholesterol levels. The later may underline the importance of total-to-HDL cholesterol ratio for the evaluation of blood lipids and the prevention of atherosclerotic disease, at population level. According to several observational and clinical studies LDL cholesterol levels of 100 mg/dl or greater have a significant contribution for the development and the progression of coronary heart disease [1]. We observed that 23% of men and 27% of women had high LDL-cholesterol levels (>100 mg/dl). In the NHANES III these rates were 22% and 17%, respectively. Recently the Adult Treatment Plan (ATP) III [4] based initiation and treatment goals for dietary and pharmacological therapy on LDL cholesterol levels, number of pre-existing risk factors and previous experience of coronary heart disease. According to these guidelines individuals without coronary heart disease and less than two risk factors should initiate dietary therapy when LDL cholesterol levels exceed 160 mg/dl. Taking into account the high prevalence of the cardiovascular risk factors in our elderly population, it seems that a considerable proportion of men and women should be under lipid lowering agents. Finally, concerning triglycerides levels the ATP III suggests a cut off point of 150 mg/dl for defining elevated levels. In the present study, approximately one half of men and women had triglycerides levels of 150 mg/dl or higher. Unfortunately, population based data from other elderly cohorts regarding triglycerides levels are lacking; this makes the comparisons with our findings difficult. Nevertheless, it is noteworthy that a considerable proportion of men and women are at risk because of on their triglyceride levels.
Furthermore, we observed a positive association between prevalence of hypercholesterolemia and smoking habits, hypertension, and obesity, while no associations were observed between hypercholesterolemia and financial status, physical activity and presence of diabetes. The association between hypercholesterolemia with other co-morbidities has already been reported in other studies before [1-5]. This fact makes the presence of high blood levels of great importance for public health, and emerge measures for the management and control of hypercholesterolemia.
Dietary and lifestyle management of blood lipids
It has already been reported that dietary habits usually influence blood lipids [7]. Thus, we evaluated lipid levels under the prism of the adoption or not of the Mediterranean diet. This diet has already been related with the reduction of all cause and cardiovascular disease mortality, due to its effect on blood pressure levels, body mass index, platelet aggregation, plasma fibrinogen and other haemostaseological factors [8-10]. Moreover, Petridou et al. [18] studying blood samples from Greek adolescents reported that the traditional Mediterranean pattern of living and eating was associated with a favorable lipid profile. However, benefits from this dietary pattern on blood lipids have rarely been reported in the literature, especially in the elderly. In our study we revealed that adherence to the Mediterranean diet is associated with a significant reduction in the likelihood of having hypercholesterolemia, after controlling for various potential confounders. Thus, based on this observation we could state another hypothesis of a pathophysiological mechanism by which Mediterranean diet may reduce cardiovascular risk, through the moderation of the oxidation process. Moreover, we revealed the additive effect of Mediterranean diet with statin treatment on blood lipids. Since the levels of uncontrolled or untreated dyslipidemia seem high in our population the previous finding could of high public health interest. However, the latter could be confounded by the better compliance to the treatment by people who were "closer" to the Mediterranean dietary pattern; which could not be assessed by the present study.
In addition, we observed an association between a "healthy index" and the likelihood of having hypercholesterolemia. In particular, presence of 2 or more protective factors (i.e. healthy index > 1) is associated with 53% lower risk of having hypercholesterolemia, while from 20% to 31% of hypercholesterolemic people could be prevented through this healthy lifestyle pattern. It is of interest that, by the exception of Mediterranean diet, none of the other components of the "healthy index" (physical activity, moderate alcohol drinking and abstinence from smoking) were associated with the prevalence of hypercholesterolemia in our elderly people.
Limitations
At this point it should be noted that the extrapolation of our findings into the general population might be under scrutiny. One of the main reasons is the moderate participation rate (68%), which is acceptable for population-based studies, like the present one, but may state hypotheses that the lifestyles of those who agreed to participate and of those who did not could be different. One other limitation is the cross-sectional design of the study. Thus, the observed benefits from the Mediterranean diet on oxidised LDL cholesterol levels, or the effect of physical activity on HDL cholesterol levels should be further investigated by randomised clinical trials.
Conclusion
The present study revealed that a large proportion of elderly men and women have blood lipid abnormalities. Moreover, despite the aforementioned limitations, Mediterranean diet seems to be an effective, non-pharmacological, intervention for the management of high blood lipids levels. In addition, a healthy lifestyle that includes abstinence from smoking, and physical activity together with moderate alcohol drinking seems to be attractive in the prevention of hypercholesterolemia. Based on the likelihood that modification of lipid levels will be beneficial, especially, to the elderly that are at higher risk for coronary heart disease, we suggest that screening for these abnormalities is considered essential and must be followed by active and effective interventions.
Table 2 Clinical and anthropometric characteristics of the participants (% by gender)
Men (n = 53) Women (n = 97) P
Hypertension (%) 60 58 0.85
Diabetes mellitus (%) 28 18 0.17
History of coronary heart disease (%) 11 5 0.20
Body mass index (kg/m2) 29 ± 4 30 ± 6 0.17
Obesity (%) 34 52 0.03
Waist circumference (cm) 107 ± 8 104 ± 8 0.11
Hip circumference (cm) 107 ± 7 113 ± 11 0.001
Waist to hip ratio 1.00 ± 0.06 0.92 ± 0.05 0.001
==== Refs
Gotto AM Jr Pearson TA, Criqui MH, Luepker RV, Oberman A, Wilson M Lipid and lipoprotein disorders Primer in Preventive Cardiology 1994 Dallas, Tex: American Heart Association 107 129
Ginsberg HN Lipoprotein metabolism and its relationship to atherosclerosis Med Clin North Am 1994 78 1 20 8283926
Wilson PWF Abbott RD Castelli WP High density lipoprotein cholesterol and mortality Arteriosclerosis 1988 8 737 741 3196218
Executive summary of the third report of the National Cholesterol Educational Program Expert panel on detection, evaluation, and treatment of high blood cholesterol in adults JAMA 2001 285 2486 2497 11368702 10.1001/jama.285.19.2486
Gran B Major differences in cardiovascular risk indicators by educational status. Results from a population based screening program Scand J Soc Med 1995 23 9 16 7784858
Denke M Cholesterol lowering diets. A review of evidence Arch Int Med 1995 155 17 26 7802517 10.1001/archinte.155.1.17
Trichopoulou A Kouris-Blazos A Wahlqvist M Gnardellis C Lagiou P Polychronopoulos E Vassilakou T Lipworth L Trichopoulos D Diet and overall survival in elderly people Brit Med J 1995 311 1457 1460 8520331
de Lorgeril M Salen P Martin J-L Monjaud I Delaye J Mamelle N Mediterranean diet, traditional risk factors and the rate of cardiovascular complications after myocardial infarction. Final report of the Lyon Diet Heart Study Circulation 1999 99 779 785 9989963
Pitsavos C Panagiotakos DB Chrysohoou C Skoumas J Papaioannou I Stefanadis C Toutouzas P The effect of Mediterranean diet on the risk of the development of acute coronary syndromes in hypercholesterolemic people: a case-control study (CARDIO2000) Coron Artery Dis 2002 13 295 300 12394655 10.1097/00019501-200208000-00008
Trichopoulou A Costacou T Bamia C Trichopoulos D Adherence to a Mediterranean diet and survival in a Greek population N Engl J Med 2003 348 2599 2608 12826634 10.1056/NEJMoa025039
Yusuf S Hawken S Ounpuu S Dans T Avezum A Lanas F McQueen M Budaj A Pais P Varigos J Lisheng L INTERHEART Study Investigators Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study Lancet 2004 364 937 52 15364185 10.1016/S0140-6736(04)17018-9
Katsouyanni K Rimm EB Gnardellis C Trichopoulos D Polychronopoulos E Trichopoulou A Reproducibility and relative validity of an extensive semi-quantitative food frequency questionnaire using dietary records and biochemical markers among Greek schoolteachers Int J Epidemiol 1997 26 S118 S127 9126540 10.1093/ije/26.suppl_1.S118
Craig CL Marshall AL Sjostrom M Bauman AE Booth ML Ainsworth BE Pratt M Ekelund U Yngve A Sallis JF Oja P International physical activity questionnaire: 12-country reliability and validity Med Sci Sports Exerc 2003 35 1381 95 12900694
ATTICA study
Costa J Borges M Oliveira E Gouveia M Carneiro AV Incidence and prevalence of hypercholesterolemia in Portugal: a systematic review. Part I Rev Port Cardiol 2003 22 569 77 12879647
Gabriel R Saiz C Susi R Alonso M Vega S Lopez I Cruz Cardenal MM Gomez-Gerique JA Porres A Muniz J EPICARDIAN Study Epidemiology of lipid profile of the Spanish elderly population: the EPICARDIAN study Med Clin (Barc) 2004 122 605 9 15142507 10.1157/13061065
Third report on Nutrition monitoring in the United States 1995 I Washington DC. Government Printing Office
Petridou E Malamou H Doxiadis S Pantelakis S Kanellopoulou G Toupadaki N Trichopoulou A Flytzani V Trichopoulos D Blood lipids in Greek adolescents and their relation to diet, obesity, and socioeconomic factors Ann Epidemiol 1995 5 286 291 8520710 10.1016/1047-2797(94)00094-A
|
16144549
|
PMC1208941
|
CC BY
|
2021-01-04 16:39:18
|
no
|
Lipids Health Dis. 2005 Sep 6; 4:17
|
utf-8
|
Lipids Health Dis
| 2,005 |
10.1186/1476-511X-4-17
|
oa_comm
|
==== Front
Malar JMalaria Journal1475-2875BioMed Central London 1475-2875-4-351604279310.1186/1475-2875-4-35ResearchUse of insecticide-treated nets (ITNs) following a malaria education intervention in Piron, Mali: a control trial with systematic allocation of households Rhee Michelle [email protected] Mahamadou [email protected] Sharon [email protected] Willi [email protected] Julie [email protected] Ogobara [email protected] Health Research and Policy Department, Stanford Medical School, Stanford University, Stanford, CA 94305, USA2 Malaria Research and Training Centre, Department of Epidemiology of Parasitic Diseases, Faculty of Medicine, Pharmacy and Odonto-Stomatology, University of Bamako Mali, P.O. Box 1805, Bamako, Mali3 San Francisco Department of Public Health, Suite 500, 25 Van Ness Avenue, San Francisco, CA 94102, USA2005 25 7 2005 4 35 35 10 5 2005 25 7 2005 Copyright © 2005 Rhee et al; licensee BioMed Central Ltd.2005Rhee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Insecticide-treated nets (ITNs) reduce malaria morbidity and mortality, but use is limited. A barrier to ITN use may be lack of knowledge regarding malaria transmission and prevention. This study is a controlled trial comparing ITN use and malaria knowledge levels between households in Piron, Mali, undertaken in 2003.
Methods
Households received net impregnation services either with or without antecedent education. The main outcome measure was ITN use, defined as impregnation of at least one of the household's existing bednets with insecticide during the study. Knowledge about malaria and prevention practices was assessed pre- and post- educational intervention. Results were analysed by household and by individual.
Results
Forty-nine percent (34/70) of households who received the educational component impregnated their nets in comparison to 35% (22/62) of households who did not (OR = 1.6 CI = 0.8–3.3, P = 0.19). In individual analysis, ITN use was significantly greater in participants who had received the educational intervention (48%) vs. individuals who did not (33%, OR = 1.9, P = 0.012). Knowledge levels about malaria significantly increased for each individual pre- versus post- educational intervention (average change score = 2.13, standard deviation = 1.97, t = -17.78, P < 0.001), although there was no difference found between educational (change score = 2.14) and control groups (change score = 2.12).
Conclusion
It is possible to educate individuals about malaria and to implement net impregnation services with limited resources. Greater accessibility to net-impregnation services is necessary but not sufficient to increase ITN use.
==== Body
Introduction
A number of studies have demonstrated that the use of insecticide-treated nets (ITNs) is effective in reducing malaria-related morbidity and mortality [1-6]. A 25% reduction in all-cause mortality for children one to nine years of age was detected during the first year of the Gambian National Bednet Program [2]. In Kilifi District, Kenya, a 33% reduction in mortality and a 44% reduction in hospital admissions for severe malaria were also found [5].
Although these trials have demonstrated that ITNs are an effective malaria control strategy, there have been many challenges to ITN distribution, acceptance and utilization when trying to implement large-scale ITN programs [2]. Knowledge about the cause of malaria and about the existence of ITNs was low in many malaria-endemic communities [7-9]. For those areas which have been reached by publicity campaigns, high cost and lack of access were some of the reasons stated as to why ITNs were not used [10-12].
Since 1993, one of the Malian National Malaria Control Program's (NMCP) main objectives was to have 90% of net users in Mopti region treating their bednets with insecticide, but it has only achieved 10–30% usage rates to date (NMCP, unpublished data). In 2000, a household survey was conducted in four villages of Mopti region in order to identify the barriers to ITN use. Although a government media campaign about ITNs reached all villages, knowledge about malaria and about the benefits of ITN use was highly variable among the four villages. Households treating their nets with insecticide had significantly higher levels of knowledge about malaria and its prevention [13]. Reasons why people did not impregnate their bednets included: not knowing anything about ITNs, cost, and not having net impregnation services readily available in the village. In the village of Piron, ten of 73 households stated that they had previously treated their bednets and had seen the benefits of ITNs but were not retreating their nets, because there were no net treatment services available in close proximity to their households.
Based on these findings, a net-impregnation service was installed within Piron run by the community itself. An antecedent household-level educational program that promoted ITNs by relating use with malaria prevention was also implemented for half the village households during the study period. The objectives of this study were to measure the impact of education plus service availability on the level of knowledge about malaria and on ITN use.
Materials and methods
Study Design
This was a controlled intervention study performed in Piron, Mali of the effects of education on ITN use, as defined by impregnation of at least one of the households' existing bednets with insecticide, and malaria knowledge. Briefly, pre-test questionnaires were administered to 133 households during the intervention period. Households were then systematically assigned to receive either an educational component or not. All households were then offered the chance to have their nets impregnated by village agent trainees during the net-treatment campaign period. Lastly, individuals were given a post-test questionnaire during the evaluation period to measure ITN use and change in knowledge about malaria. The study design is shown Figure 1.
Figure 1 Study design and timeline.
Setting
This study was conducted during July-August 2003 in Piron village located in Bandiagara District, Mopti region, a sahelian area where malaria is endemic. Malaria prevalence rates for Piron, measured during the dry season for children ages one to nine years, was 61% (Keita et al., personal communication), but transmission rates are much higher at the end of the rainy season in September-October. Piron has no electricity or running water and no health facilities are available within the village. Although Piron is only 17 kilometres away from Goundaka – the closest village with health and net impregnation facilities – the roads are impassable during the rainy season and very few individuals have a method of transportation other than walking.
Study design
Sampling design
A complete village sample of adults (estimated total population 1000, with 300 adults) was carried out with the household as the unit of allocation. To be eligible for the household survey, subjects had to have children. The Mopti Region Health Information System estimated 150 households in Piron, although only 133 were present at the time of study. Every second household was systematically assigned in equal proportions to receive the educational intervention with net-treatment services (study group) or net-treatment services alone (control group).
The protocol was reviewed and approved by the Administrative Panel for Human Subjects in Medical Research at Stanford University in the United States and the Institutional Ethical Committee at the Faculty of Medicine, Pharmacy and Odonto-stomatology, University of Bamako, in Mali. Informed consent was obtained from the village leaders and all individuals participating in the study. All individuals were given standard treatment for simple illnesses regardless of their participation in the study.
Intervention period (Days 1–8)
Two trained field guides, along with two local village guides who introduced them to households as is customary, went door-to-door to orally administer the pre-intervention questionnaires in the local languages to consenting individuals. Questionnaires were first developed in English, translated into French and then into Peul or Dogon, and back-translated to French and then English for verification.
Ten pre-test questions assessed individuals' knowledge of malaria, including ability to define malaria, to recognise signs and symptoms of the disease and knowledge of transmission, susceptibility, prevention and treatment. Additional information solicited in the one-on-one, in-depth interviews included individuals' demographic information, including socio-economic status (SES), and specific malaria prevention practices used such as ITNs, bednets, sprays and mosquito coils.
The intervention consisted of education about signs and symptoms of malaria, susceptibility, transmission and prevention of malaria and information about the benefits of ITN use, including, how, when and where to impregnate nets. Husband and wife/wives from each household assigned to the education group were educated together as a unit immediately after pre-test questionnaires were administered. Households in the control group received no antecedent education. All households were given the opportunity to have their nets treated at the end of the intervention study period, during the establishment of permanent net-impregnation services within the village. To the extent possible, interviewers were blinded to group assignment by reassigning interviewers during the post-education evaluation period.
Net-impregnation period (Days 9–16)
A community-based skills training format [14] was used to train four members of the community to run the net-impregnation program within their village. They were educated on the following parameters: malaria transmission and prevention, how to treat bednets with insecticide and how to use a self-monitoring system to collect data on how many nets are treated, how often and who uses them. They were also trained in recognising signs and symptoms of simple malaria and correct treatment dosages for different age groups.
During the training period, nets were impregnated by the village agent trainees for an at-cost price of 200 CFAF (1 US$ = 560 francs de la Communauté Financière d'Afrique [CFAF] at the time of study) per net or curtain regardless of size. Because cost was previously identified as a major inhibiting factor to using ITNs, the minimal price was charged that would permit purchase of more insecticide from the Mopti Regional Health Centre, thus enhancing long-term sustainability. At the end of the study, a village meeting was held to discuss the parameters of the program, namely the toxicity of the permethrin and the price of net impregnation they must charge in order to keep the program operating.
Evaluation period (Days 17–22)
The primary outcome measure was ITN use, as defined by impregnation of at least one of the households' existing bednets with insecticide, at the end of the study period. Change in knowledge about malaria transmission and prevention was assessed by comparing responses to questions pre- and post-intervention. After the post-test evaluation survey, control households received the educational component.
Statistical analysis
Data were analysed with SPSS (Version 11.5). 'Intent-to-treat' analysis assessed the success of the educational intervention by measuring the ITN use for households in the intervention versus control arms using logistic regression analysis with household as the unit of allocation, weighting responses according to the number of individuals within each household who participated in the study. The weighting scheme took into account that there can be one or more household decision-makers, that everyone in the household had equal access to ITNs and that some household members may sleep under a single bednet. In a separate analysis, ITN use was analysed by individual, that is, by 'treatment-per-protocol' logistic regression analysis, under the rationale that households typically had more than one bednet and because education uptake was highly individualised. Nonetheless, men and women were educated as a household unit.
In determining change in knowledge, each of ten interview questions pertaining to malaria disease, symptoms, transmission, and prevention were assigned a point value of "1" for each correct answer and "0" for an incorrect answer to questions with only one response or "-1" for an incorrect answer to questions with greater than one response and summed for each respondent. This method of scoring responses produced a normal distribution when assessing the change in knowledge for individuals from baseline to post-intervention surveys. The paired t-test was used to measure change in pre- and post-test scores within subject. Mean change in household knowledge levels between control subjects and subjects receiving the educational component were also weighted according to household numbers in linear regression analysis.
All regression analyses were adjusted for non-randomized characteristics; that is, the date when individuals received the educational component, which field worker conducted the education and baseline characteristics which were significantly different between intervention and control groups. Beta values (B), odds ratios (OR), adjusted odds ratios (AOR) and 95% confidence intervals (CI) are presented.
Results
Demographic characteristics
In July 2003, 133 households (276 individuals) underwent systematic allocation: 70 households (147 individuals, 61 men and 86 women) to receive an educational component with the offer to treat their bednets and 63 households (129 individuals, 57 men and 72 women) to receive the offer of net-impregnation services alone (Figure 1). Of those approached, there was a 100% participation rate. No one crossed over between groups. Five individuals from the control group were lost to follow-up: a husband and wife from one household, and two men and one woman from separate households (62 households = 124 individuals, 54 men and 70 women). The 'intention-to-treat' and 'treatment-per-protocol' analysis was based on the remaining 132 households with one to four responders per household and 271 individuals, respectively.
Most baseline characteristics were similar in the two groups (Table 1). The mean age was 40 years and ages ranged from 14–80 in both groups. The individuals were Muslim (100%) and were primarily Dogon ethnicity (82–84%) with the other 15–18% being a mixture of Peul, Sonnike, Bambara and Maraka (primarily Peul). All residents were subsistence-level agriculturists, with less than 40% of both groups having another form of work for monetary income. Less than 15% of the population was literate, with slightly more than half of that percentage being literate in French. There was a significant difference in previous ITN use (education = 13% vs. control = 2%, P = 0.02) and median number of deceased children (education = 1 vs. control = 2, P = 0.014) between the two groups. SES, whether measured as earning income or possessions, was balanced across education and control groups.
Table 1 Baseline demographic characteristics of individuals.*
Household characteristics Education group
(N = 70) Control group
(N = 62) P Value
Total individuals- no. 147 124
Women- no. (%) 86 (58.5) 70 (56.5) 0.73
Men- no. (%) 61 (41.5) 54 (43.5)
Age-years 40.4 ± 13.5 40.3 ± 13.7 0.99
Ethnicity- no. (%)
Dogon 124 (84) 102 (82) 0.22
Peul 18 (12) 12 (10)
Other 5 (3) 10 (8)
Median number of responders per household 2 2
Household size- median (25 and 75 quartiles) 5 (4, 8) 5 (4, 8) 0.84
Number of children total- median (25 and 75 quartiles) 5 (2, 7) 5 (3, 8) 0.16
Living- median (25 and 75 quartiles) 3 (2, 5) 3 (2, 5) 0.78
Deceased- median (25 and 75 quartiles) 1 (0, 3) 2 (1, 4) 0.014
At least one responder is literate- no. (%)† 15 (21) 11 (18) 0.60
At least one responder is a trader- no. (%)‡ 13 (19) 17 (27) 0.23
At least one responder earns an income- no. (%) 24 (34) 24 (39) 0.60
Socio-economic status (SES) § 0.9 ± 1.6 0.8 ± 1.1 0.53
Malaria prevention methods used currently-
no. (%)¶
ITNs used previously 9 (13) 1 (2) 0.02
Untreated nets 69 (99) 57 (92) 0.10
Insecticide sprays 6 (9) 5 (8) 0.92
Mosquito coils 12 (17) 9 (15) 0.68
*Plus-minus values are means ± standard deviations (SD).
† Literate in any language, having studied for at least one year.
‡ All individuals were subsistence-level farmers.
§SES was calculated based on the number of material possessions. Each item was weighted according to their market value. SES value of 1 is approximately equal to 135,000 CFA ($241).
¶Percentages represent households who had at least one individual using stated method. Most stated that they used these methods against the nuisance of mosquitoes, not for malaria. None of the households were using ITNs at the time of study.
ITN use
At baseline interview, none of the 132 households were using ITNs. The most common reasons for not treating their nets were cost (59%), availability (23%) and lack of knowledge regarding the effectiveness of ITNs in preventing malaria (11%). However, 93% of those who did not treat their nets during the study stated that cost was the main reason. Other malaria prevention methods are summarised in Table 1, with untreated bednets being the most common malaria prevention method (96%).
Forty-two percent (56/132) of households impregnated at least one of their nets with insecticide during the study. Forty-nine percent (34/70) of households who received the educational component impregnated their nets in comparison to 35% (22/62) of households who did not, although this difference was not statistically significant (OR = 1.6 CI = 0.8–3.3, P = 0.19). When stratified by day of pre-test interview, however, households in the intervention group who were interviewed during the first two days were significantly more likely to impregnate their nets than control group households (OR = 12.67, CI = 1.18–135.96, P = 0.04). When the household analysis was adjusted for significantly different baseline variables, none of them influenced the effect of the education intervention on ITN use.
Households in the education group tended to impregnate more nets per household (N = 1.01) than those in the control group (N = 0.66, t = 1.90, P = 0.059). Thus when analysed by individual, 48% (71/147) versus 33% (41/124) of individuals in education and control groups, respectively, impregnated their nets (unadjusted OR = 1.9, CI = 1.15–3.10, P = 0.012). Individuals assigned to the education group were still significantly more likely to impregnate their bednets than control group individuals when analysis was adjusted for day of baseline interview and pre-test interviewer. Interview day was significantly associated with ITN use independently of the educational intervention (AOR = 0.52, CI = 0.40–0.67, P < 0.001) as was the interviewer (AOR = 2.32, CI = 1.06–5.08, P = 0.015). Moreover, there was a significant interaction between education and interviewer (AOR = 3.69, CI = 1.33–10.25, P = 0.01), indicating that one field guide was much more successful in convincing individuals to use ITNs. In the multivariate model that includes the interaction between field guide and education group, the effect of the education intervention was greater for the male interviewer (AOR = 3.89) than for the female interviewer (AOR = 2.44). Of note, the adjusted effect of the educational intervention on individual ITN use with either interviewer is greater than the unadjusted effect (OR = 1.9).
Malaria knowledge assessment
Pre-intervention malarial disease recognition was high (Table 2). Most individuals could identify malaria as the most common disease in their village (93%), recognise malaria based on clinical symptoms (98%), were familiar with the term malaria in their local language (99%) and state at least one symptom of the disease (93%). Knowledge about who was most susceptible to malaria (72%) and malaria treatment methods (87%) was also relatively high. However, knowledge of prevention was more limited. Only 35% of individuals knew that malaria was transmitted by mosquitoes and less than 40% of people knew that one could prevent malaria. Only 17% of those individuals stated that using ITNs was an important method of prevention.
Table 2 Pre- and post-intervention levels of knowledge about malaria for all individuals in Piron, Mali.*
Education group Control group
Pre-intervention Post-intervention Pre-intervention Post-intervention
Question N (%) N (%) N (%) N (%)
1. What is the most common disease in your village? 137 (93) 147 (100) 115 (93) 124 (100)
2. If you or your child has the following symptoms: fever, headache, vomiting and chills, what is the disease? 144 (98) 147 (100) 121 (98) 124 (100)
3. Do you know what malaria is? 145 (99) 146 (99) 122 (98) 124 (100)
4. Name at least one symptom of malaria. 139 (95) 147 (100) 112 (90) 124 (100)
5. Who is most susceptible to malaria? 53 (36) 130 (88) 41 (33) 107 (86)
6. How is malaria transmitted? 111 (76) 131 (89) 84 (68) 118 (95)
7. Can you prevent malaria? 53 (36) 131 (89) 46 (37) 102 (82)
8. How can you prevent malaria? 49 (33) 131 (89) 47 (38) 102 (82)
9. Can you treat malaria? 132 (90) 147 (100) 111 (90) 122 (98)
10. How can you treat malaria? 129 (88) 147 (100) 107 (86) 123 (99)
Total knowledge score
(Maximum 10 points) † 7.43 ± 1.85 9.55 ± -.96 7.31 ± 1.88 9.44 ± 1.06
*One point was given for any correct answer, 0 points was given for an incorrect answer, and one point was subtracted for any incorrect answer for questions with multiple answers. This method of scoring responses produced a normal distribution when assessing the change in knowledge for individuals from baseline to post-intervention surveys. Correct answers to the questions are: 1. Malaria, 2. Malaria, 3. Yes, 4. Fever, headache, chills, yellow eyes, vomiting, diarrhoea, 5. Pregnant women, children, 6. Mosquito bite, 7. Yes, 8. Insecticide-treated net (ITN), untreated net, modern medicine, traditional medicine, 9. Yes, 10. Modern medicine, traditional medicine. All pre- and post-intervention differences were significant (P < 0.001) except for questions two and three. †Plus-minus values are means ± standard deviations (SD).
Within subject, change in knowledge score pre- versus post-educational intervention was also significant (average change score = 2.13, standard deviation = 1.97, t = -17.78, P < 0.001). However, there was no significant difference in responses to individual questions or in total knowledge score between the group which received the educational component and the control group when analysed by household (education group change score = 2.14 vs. control group change score = 2.12 P = 0.96) or by individual (education group change score = 2.12 vs. control group change score = 2.13 P = 0.98). When household analysis was adjusted for baseline characteristics, none affected the impact of the intervention on knowledge scores. When the analysis was adjusted for time of interview, it was found that individuals (and households) who were interviewed later during the intervention period had significantly lower changes in knowledge score (B = -0.34, CI = -0.56 to -0.11, P = 0.004).
Discussion
The study results suggest that education increases net-treatment rates. Antecedent education was significantly associated with ITN use at the individual level, but not at the household level. Additionally, knowledge about malaria increased overall for the entire village, but no significant difference was found between education and control groups.
Other studies have found that ITN use increased when individuals received health promotional activities about ITNs [15,16], although few studies have examined the effects of ITN educational interventions in a control trial. There are several reasons that could explain why a difference was not found at the household level. First, the power of our study to detect differences between intervention and control households may have diminished over time as information diffused from one household to another. In fact, households that were interviewed during the first two days of the study were more likely to impregnate their nets if they received the educational intervention compared to control. Second, simply being in the village may have had an immeasurable influence on net treatment rates. Almost everyone had heard about ITNs previously, but was not using them. Taking time with each individual and/or household to offer net impregnation services from an 'expert' source may have been sufficient to change behaviour. Finally, any ITN use in a household use is a conservative estimate of total ITN use because most households owned more than one net and many people often share a net. Of note, the analysis of ITN use had more power to detect a difference between intervention and control groups, and the effect remained significant after adjusting for within household correlation.
Another important objective of this study was to install net-impregnation services within the village run by community members themselves so that accessibility did not prevent ITN use. A previous study done in Mali demonstrated that individuals who used ITNs were predominantly from communities that had net-treatment services in their village [13]. Other studies have also shown that community involvement is an important factor to the success of net-impregnation programs [12,17,18].
Approximately 40% of the village impregnated their nets, which illustrates that availability of services and education can increase ITN use, but there remain substantial barriers to achieving the NMCP goal of 90% use. Cost is clearly a factor as individuals who made a monetary income were more likely to impregnate their nets. Of those who did not impregnate their nets during this study, 93% of individuals said that cost was the main factor. Similar results have been found in other studies [10,19,20]. However, the socio-economic data collected in this study suggest that many individuals were able to afford the promotional price of treating one net. Guigemde et al. have shown that treating nets with insecticide is affordable to many individuals in malaria-endemic areas, but they are not aware of how much they spend on other, often less effective, prevention methods [21]. Individuals may also need time to see that impregnation is more effective than untreated nets alone. It was found that more individuals impregnated their nets at later time points during the promotional service once they had seen others benefit from ITN use.
There are limitations to the study. The biggest shortcoming is that the study was done within one village, thus placing control and intervention groups in close proximity. There is a strong possibility that knowledge diffused between individuals who received the educational component and those individuals who did not, as suggested by the fact that almost everyone knew more about malaria after the education intervention. A more rigorous study design would use several widely separated villages rather than households within the same village. However, the limited time and resources precluded identifying and implementing the project in multiple villages with comparable demographic factors.
Second, the field guide who administered the questionnaire and provided the education affected the impact of the education intervention on ITN use, as was demonstrated by the interaction between education and field guide. There is no reason to believe that the educational intervention was delivered differently by each field guide, as both were given the same training and tested on the reproducibility of their presentations. One field guide may simply be a better salesman in promoting net treatment. Regardless of field guide influence, individuals receiving education were still significantly more likely to impregnate their bednets than control group individuals, as shown by the fully adjusted analysis.
Third, there was limited power in this study to detect small effect sizes. The study was planned to have sufficient power to detect a 50% increase in ITN use given the size of the village. The number of households required to detect the effect size observed in this study (400) is larger than the village itself. As mentioned above, resources were not sufficient to include other villages.
Finally, sustainability of net-impregnation program in the long-term was not determined during this study. Although there was not complete ITN use in Piron at the conclusion of the study, increased ITN use may be seen in Piron after the rainy season has finished when malaria transmission is heaviest and the advantages of ITNs are more pronounced. A follow-up study is needed in order to see if this is correct.
Conclusion
Despite potential limitations, the data suggest that it is possible to educate individuals about malaria and to implement net impregnation services in villages with limited resources. ITNs are currently one of the most viable options for reducing malaria-related morbidity and mortality. Although ITN use is the primary method recommended by the World Health Organization for malaria reduction and control, implementation worldwide has been slow. Some of the major barriers to ITN use are lack of access determined by locality of services, misunderstanding the costs in relation to the benefits of ITNs and low levels of knowledge about malaria prevention. Greater accessibility to net-impregnation services is necessary but not sufficient to increase ITN use. The present study evaluated a project to simultaneously increase access to ITN and malaria prevention knowledge using a controlled study design. This study also illustrates the difficulties associated with behaviour change intervention research. More studies such as this one, which combines program delivery using minimal resources with evidence-based evaluation to address major local health concerns in developing countries, need to be done.
Authors' contributions
MR participated in the design and coordination of the study, carried out the field work and data collection, performed statistical analysis of the data collected and drafted the manuscript. MS participated in the design and coordination of the study and contributed to the writing of the manuscript. SP participated in the design of the study, performed statistical analysis of the data collected and contributed to the writing of the manuscript. WM participated in the design of the study and contributed to the writing of the manuscript. JP participated in the design of the study and contributed to the writing of the manuscript. OD participated in the design and coordination of the study and contributed to the writing of the manuscript.
Acknowledgements
This project received financial support from the Stanford Medical School Community Partners Medical Scholars Program, the West African Research Association, and the American Medical Women's Association and was sponsored by the Malaria Research and Training Centre at the University of Bamako, Mali. We thank Dr. Mamadou Namory Traore, Dr. Bourema Plea and Dr. Youssouf Coulibaly for helping to facilitate the project and to organise visits to Piron. Special thanks to Dr. Massambou Sacko and all members of the Mali National Malaria Control Program for donating the permethrin used in this project. We would also like to thank Mr. Amadou Thiande Traore for his invaluable help with implementing the net-impregnation service. Many thanks to all the field guides Mr. Oumar Niare, Mrs. Mariam Maiga, Mr. Nouhoum Degoga and Mr. Bani Kansaye and to all of the residents of Piron who participated in this study.
==== Refs
Alonso PL Lindsay SW Armstrong Shellenberg JRM Keita K Gomez P Shenton FC Hill AG David PH Fegan G Cham K Greenwood BM A malaria control trial using insecticide-treated bed nets and targeted chemoprophylaxis in a rural area of The Gambia, West Africa: 6. The impact of the interventions on mortality and morbidity from malaria Trans R Soc Trop Med Hyg 1993 87 37 44 8212109 10.1016/0035-9203(93)90174-O
D'Alessandro U Olaleye B McGuire W Langerock P Bennett S Aikins M Thomson M Cham M Greenwood BM Mortality and morbidity from malaria in Gambian children after introduction of an impregnated bednet programme Lancet 1995 346 462 467 7637479 10.1016/S0140-6736(95)91321-1
Huailu C Wen Y Wuanmin K Chongyi L Large-scale spraying of bednets to control mosquito vectors and malaria in Sichuan, China Bull World Health Organ 1995 73 321 328 7614664
Binka F Kubaje A Adjuik M Williams L Lengeler C Maude G Armah G Kajihara B Adiamah J Smith P Impact of permethrin impregnated bednets on child mortality in Kassena-Nankana district, Ghana: a randomised control trial Trop Med Int Health 1996 1 147 154 8665378
Nevill C Some E Mung'ala V Mutemi W New L Marsh K Lengeler C Snow R Insecticide-treated bednets reduce mortality and severe morbidity from severe malaria among children on the Kenyan coast Trop Med Int Health 1996 1 139 146 8665377
Hawley WA ter Kuile FO Steketee RS Nahlen BL Terlouw DJ Gimnig JE Shi YP Vulule JM Alaii JA Hightower AW Kolczak MS Kariuki SK Phillips-Howard PA Implications of the western Kenya permethrin-treated bed net study for policy, program implementation, and future research Am J Trop Med Hyg 2003 168 173 12749501
Aikins MK Pickering H Greenwood BM Attitudes to malaria, traditional practices and bednets (mosquito nets) as vector control measures: a comparative study in five West African countries J Trop Med Hyg 1994 97 81 86 8170007
Ziba C Slutsker L Chitsulo L Steketee R Use of malaria prevention measures in Malawian households Trop Med Parasitol 1994 45 70 73 7915045
Onwujekwe O Akpala C Ghasi S Shu E Okonkwo P How do rural households perceive and prioritise malaria and mosquito nets? A study in five communities of Nigeria Public Health 2000 114 407 410 11035466 10.1016/S0033-3506(00)00373-5
Kachur S Phillips-Howard P Odhacha A Ruebush T Oloo A Nahlen B Maintenance and sustained use of insecticide-treated bednets and curtains three years after a controlled trial in western Kenya Trop Med Int Health 1999 4 728 735 10588766 10.1046/j.1365-3156.1999.00481.x
Nuwaha F People's perception of malaria in Mbarara, Uganda Trop Med Int Health 2002 7 462 470 12000657 10.1046/j.1365-3156.2002.00877.x
Okrah J Traore C Pale A Sommerfeld J Muller O Community factors associated with malaria prevention by mosquito nets: an exploratory study in rural Burkina Faso Trop Med Int Health 2002 7 240 248 11903986 10.1046/j.1365-3156.2002.00856.x
Rhee M Sissoko M Perry S Dicko A McFarland W Donmbo O Malaria Prevention practices in Mopti region Mali Eamj 2005 82
Elder J Geller E Hovell M Mayer J Motivating Health Behavior 1996 Albany: Delmar Publishers Inc
Lin K Aung S Lwin S Min H Aye NN Webber R Promotion of insecticide-treated mosquito nets in Myanmar Southeast Asian J Trop Med Public Health 2000 31 444 447 11288998
Minja H Schellenberg JA Mukasa O Nathan R Abdulla S Mponda H Tanner M Lengeler C Obrist B Introducing insecticide-treated nets in the Kilombero Valley, Tanzania: the relevance of local knowledge and practice for an information, education and communication (IEC) program Trop Med Int Health 2001 6 614 623 11555427 10.1046/j.1365-3156.2001.00755.x
Makemba AM Winch PJ Kamazima SR Makame V Sengo F Lubega PB Minjas J Shiff C Community-based sale, distribution and insecticide impregnation of mosquito nets in Bagamoyo district, Tanzania Health Policy Plan 1995 10 50 59
Kroeger A Avinna A Ordonnez-Gonzalez J Escandon C Community cooperatives and insecticide-treated materials for malaria control: a new experience in Latin America Malaria J 2002 1 15 10.1186/1475-2875-1-15
Cham MK Olaleye B D'Alessandro U Aikins M Cham B Maine N Williams LA Mills A Greenwood BM The impact of charging for insecticide on the Gambian National Impregnated Programme Health Policy Plan 1997 12 240 247 10173405 10.1093/heapol/12.3.240
Winch PJ Makemba AM Makame VR Mfaume MS Lynch MC Premji C Minjas JN Shiff CJ Social and cultural factors affecting rates of regular retreatment of mosquito nets with insecticide in Bagamoyo District, Tanzania Trop Med Int Health 1997 2 760 770 9294546 10.1046/j.1365-3156.1997.d01-376.x
Guigemde TR Dao F Curtis V Traore A Sondo B Testa J Ouedraogo JB Household expenditures in the town of Bobo-Dioulasso, Burkina Faso Trans R Soc Trop Med Hyg 1994 88 285 287 7974661 10.1016/0035-9203(94)90079-5
|
16042793
|
PMC1208942
|
CC BY
|
2021-01-04 16:26:57
|
no
|
Malar J. 2005 Jul 25; 4:35
|
utf-8
|
Malar J
| 2,005 |
10.1186/1475-2875-4-35
|
oa_comm
|
==== Front
Microb Cell FactMicrobial Cell Factories1475-2859BioMed Central London 1475-2859-4-231612021310.1186/1475-2859-4-23ReviewMonitoring protein stability in vivo Ignatova Zoya [email protected] Biotechnologie II, Technische Universität Hamburg-Harburg, Denickestr. 15, 21073 Hamburg, Germany2005 24 8 2005 4 23 23 26 7 2005 24 8 2005 Copyright © 2005 Ignatova; licensee BioMed Central Ltd.2005Ignatova; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Reduced protein stability in vivo is a prerequisite to aggregation. While this is merely a nuisance factor in recombinant protein production, it holds a serious impact for man. This review focuses on specific approaches to selectively determine the solubility and/or stability of a target protein within the complex cellular environment using different detection techniques. Noninvasive techniques mapping folding/misfolding events on a fast time scale can be used to unravel the complexity and dynamics of the protein aggregation process and factors altering protein solubility in vivo. The development of approaches to screen for folding and solubility in vivo should facilitate the identification of potential components that improve protein solubility and/or modulate misfolding and aggregation and may provide a therapeutic benefit.
==== Body
Introduction
The protein folding problem has challenged scientists for more than three decades. The process of folding has been studied extensively in the test tube, commonly carried out at low protein concentrations in order to minimize any parallel off-pathway aggregation processes. The environment, however, that a newly synthesized protein encounters in the cell is far from these idealized in vitro conditions. The intracellular space presents a highly crowded limited volume predicted to favor association processes [26]. Given the potential for intermolecular interactions with other cellular components, the kinetic and thermodynamic parameters of folding/unfolding processes in the cell may be quite different from those ideal in vitro solutions. The complex intracellular environment significantly affects thermodynamic stability through changes in the local osmolarity and oxidation potential in response to stress conditions [3,25]. Reduced protein stability is detrimental to function and can lead to aggregation or degradation [2].
There are multiple instances where low in vivo stability of a protein can lead to negative outcomes. Recombinant protein production faces difficulties producing some proteins in an active form in expression hosts. Unfortunately, many recombinant proteins are not efficiently expressed in a folded or soluble form, which is an obstacle for the pharmaceutical industry and biotechnology. Protein-folding diseases are yet another area where protein solubility plays an important role. Aberrant folding and stability in the cell is associated with the pathology of many neurodegenerative diseases (e.g., Alzheimer's, Huntington's, Parkinson's disease) [2,16]. Solubility and folding efficiency of a protein can be maintained by sequence alterations, which necessitate a significant structural knowledge about the protein of interest to choose appropriate amino acids for mutation. As a result, there is an urgent need to develop methods to accurately monitor folding and unfolding events on fast timescales, directly in living cells. Such efficient selective methods to monitor folding and/or solubility in vivo could be also used to assay for compounds inhibiting aggregation and screening for variants with an improved solubility by natural or induced mutations.
In-Cell Spectroscopy
The potential of the nuclear magnetic resonance (NMR) technique, widely used in vivo for identification of small molecules, has been extended towards determination of the conformation of proteins inside living cells [10,17]. The chemical shifts as a specific fingerprint of the local environment of the atomic nuclei within a protein provide residue specific information on the conformational changes due to unfolding or binding events, or other post-translational modifications. Studies using [15N,1H]-HSQC [18,19] or magic angle spinning binding experiments [10] have been used to characterize cytoplasmic and periplasmic biomolecules in bacterial strains. The in vivo NMR-spectra detect additional conformations not observable in in vitro purified samples: The in-cell spectrum of calmodulin reveals different occupation of the four calcium binding sites that is connected to the tight regulation of the intracellular calcium concentration [19]. A crucial factor, however, that limits the application of the in-cell NMR experiments is the rotational correlation time or tumbling rate of the protein in the cellular environment. The intracellular space is highly viscous and decreases the tumbling rate, causing a broadening of the peaks or even their disappearance. The introduction of the TROSY approach [13] has increased the molecular weight range of proteins that are amenable to NMR spectroscopy beyond the 100 kDa mark. This can be applied to virtually any protein, but there is still the viscosity limit of the intracellular space.
An interesting twist using conventional mass spectrometry technique, widely applied to the determination of conformational states of proteins, is a method to measure the Stability of Unpurified proteins from rates of H/D exchange (SUPREX) developed by Oas and coworkers [5]. This approach enables the determination of protein stability directly in the cell. This technique is a variant of the Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry [4] and determines the protein stability in vivo by measuring the hydrogens that exchange with solvent deuterons at a rate depending on the thermodynamic stability of the protein [5]. The SUPREX approach has been applied to estimate the folding thermodynamics by direct in-cell urea titration. The thermodynamic stability of the small monomeric N-terminal fragment of λ repressor (λ6–85) estimated with this technique shows no change in stability versus measurements made in vitro [5]. Additionally, the effects of introduced mutations can be quantified as changes of the free energy (ΔΔGoN-M) between the native and unfolded states of mutated protein (ΔGoM) compared to the wild-type counterpart (ΔGoN). The bacterial cells are viable up to 3 M urea, which is sufficient to map the entire unfolding transition curves in vivo. Although some successful examples to determine conformational changes and stabilities have been reported so far, both techniques, NMR and mass spectrometry, are far from being established as routine tests for estimating the in vivo stability. A serious limitation arises if the target protein interacts with any cellular components (e.g., chaperones, DNA, membranes).
Protein-Reporter-Based Approach
The fusion of a protein of interest to a reporter protein whose function depends on the solubility of the chimera has been successfully used to reveal in vivo protein stability and folding efficiency, provided the reporter does not change the solubility of the target protein [Fig. 1 is a cartoon summarizing different tagging approaches]. Here, the term solubility is used interchangeably with folding, and it denotes both the chemical solubility of the folded native state and the absence of aggregation due to misfolding events. This technique is based on the attachment of a specific protein tag that equips the resulting chimera with a unique property that can be exploited to monitor certain functions of the target protein in vivo. Davidson and coworkers used the resistance against chloramphenicol generated by fusing chloramphenicol acetyltransferase (CAT, 25 kDa) to the C-terminus of the protein of interest to select soluble variants from a large pool of mutants created by a random mutagenesis [11].
Figure 1 Strategies for site-specific labeling of proteins to detect folding and stability in vivo. (A) The autofluorescent green protein is appended to the protein of interest (here maltose-binding protein) and the latter dictates the spectroscopic behavior of the protein-reporter tag. The same fusion scheme applies for enzymes (e.g., CAT) whose activity serves as readout of the analysis. (B) Complementation assay. The target protein (here maltose-binding protein) is sandwiched between two reporter proteins or between the two halves of one reporter. Efficient folding brings both parts of the reporter unit into sufficient proximity to generate a readout signal. (C) Small-molecule labeling using peptide tags. The peptide tag can be introduced in the middle of the protein sequence (here cellular retinoic acid binding protein with a tetracysteine motif highlighted in yellow and FlAsH ligated to it) and the intensity of the emitted fluorescent signal can be used to determine conformationally distinct populations, provided the specific signatures of the folded and unfolded states are pre-determined for each experiment.
A specific implementation of the reporter-based approach is the structural complementation that involves the splitting of the reporter protein into two components that need to combine to execute their function. Wigley et al. [23] have designed an elegant assay for monitoring of the solubility/folding based on complementarity of two fragments originating from β-galactosidase. The α-fragment of β-galactosidase, typically 50–90 residues in length, when attached C-terminally to the protein of interest, can complement its ω-fragment, coexpressed in the soluble cytoplasmic fraction, to reconstitute the functional protein. The assembly occurs only if the target protein is expressed in the soluble form and the complementation event restores the enzymatic functions. The relatively small size of the α-peptide does not perturb the structure of the target proteins and does not have any dramatic effect on the solubility of the chimera [23]. As a bimolecular reaction the assembly is significantly affected by the concentration of the complementing parts and is determined by the equilibrium association/dissociation constant Kd, which is a significant hurdle to study kinetics of folding/aggregation. An advantage this system shares with the CAT-based system is the potential for using an easily detectable phenotypic selection for screening of mutants with improved folding efficiency.
An alternative to the complementarity principle is the split-technique by which target protein is tagged at both termini with two halves of the reporter protein [9]. The protein of interest is sandwiched between the two halves of ubiquitin, which assemble into a functional ubiquitin moiety only after an accomplished folding of the target protein [15]. The reconstituted ubiquitin serves as a substrate of ubiquitin-specific proteases, normally present in the cytosol and the nucleus of all eukaryotic cells, and the release of the reporter from the chimera is the readout of the assay. Detection is accomplished ex vivo after the intermolecular association event has taken place. It may perturb the equilibrium established within the cell due to this physical separation of the cellular components. The method is qualitative, allowing the study of alterations of the structure caused by mutations or intracellular environment but it cannot be used for measurements and for determination of the destabilization energy (ΔΔGoN-M) caused by the corresponding mutation.
The use of autofluorescent proteins as reporter protein tags, such as green fluorescent protein (GFP) and its relatives with improved spectral properties, has contributed greatly to our understanding of intracellular trafficking, localization and protein-protein interactions that could not have been achieved with other technologies [20]. The autocatalytically-generated broad fluorescent signal of GFP is easily detectable in a bulk cell suspension or under the microscope, and this key feature makes it an attractive fusion counterpart for studying intracellular folding and unfolding processes. Waldo and coworkers [21] showed that the folding trajectory of a protein of interest fused upstream to the GFP dictates its fluorescent behavior and the emitted fluorescent signal is directly proportional to the amount of the correctly folded protein. Later on, Glockshuber and coworkers [14] propose an approach for monitoring of protein solubility based on in vivo Förster-resonance-energy-transfer (FRET) measurements of two proteins, reminiscent of the idea of the complementation assay. The target protein is labeled at both termini with GFP and its blue-shifted variant, BFP, and efficient folding brings both fluorescent proteins into sufficient proximity to generate a FRET-signal. The readouts of the fluorescent reporter protein-based methods for protein solubility in vivo are simply detectable signals and allow a fast selection of the soluble variants of any protein of interest. Although enhanced variants of GFP with significantly accelerated folding and fluorescence acquisition have been developed, unfortunately, their bulky size (238 amino acids) may perturb the structure of the target protein. The relatively long folding trajectory of GFP itself ranging from minutes to hours [6], even of the improved Venus or T-Sapphire variants, is out of the time window of early intramolecular (folding) and intermolecular (aggregation) events. Furthermore, large surface contacts between the target protein molecules (e.g. protein aggregation) can undesirably alter the fluorescence of the reporter [24].
All reporter-based approaches monitor the protein folding and the solubility as a result of efficient folding at its equilibrium. For understanding the mechanisms of protein misfolding in the cell and its impact for disease and biotechnology, it is necessary to follow the process from its early initial steps. Insertion of the reporter molecule within the sequence of the target protein can probably capture early folding/unfolding events. Unfortunately, the bulky reporter proteins can be appended only to either the N- or C-terminus. Short peptide sequences that mediate the labeling of the protein of interest with small synthetic molecules, conferring unique spectroscopic properties, present an alternative solution to overcome these restrictions. Such approaches are discussed in the next section.
Lighting-up Folding Processes with Small Molecules
The prototype for the in vivo labeling of fusion proteins with small synthetic molecules is based on the formation of stable complexes between biarsenical derivatives of fluorescein (FlAsH) or resorufin (ReAsH) and peptides containing tetracysteine motif developed in the lab of Roger Tsien [1,7]. The two arsenoid groups of the dye selectively coordinate the four cysteine residues from the naturally uncommon tetracysteine motif (Cys-Cys-Xaa-Xaa-Cys-Cys, where Xaa denotes any amino acid, preferably Pro-Gly, [1]), which can be incorporated at virtually any place in the target protein sequence due to the small size. The dye ligand is membrane permeable and non-fluorescent; however it emits an intense fluorescent signal upon high affinity binding to the cysteine residues (2–4 pM Kd [1]). The background fluorescence can be minimized by treatment of the cells with ethane dithiol that pairs the unpaired cysteines from endogenous proteins displaying a weak affinity for FlAsH.
Dye ligation to the tetracysteine receptor is not connected to a characteristic emission shift that can be used as a specific fingerprint for detection of a distinct conformational species. Incorporation of the tetracysteine sequence in flexible structural regions (e.g., loops) within the protein sequence of the cellular retinoic acid binding protein, however, leads to a significant sensitivity of the dye quantum yield on the conformational state of the host protein, with a denatured state hyperfluorescent compared to the native one [8]. The FlAsH-mediated fluorescence lacks specific quantum yields repeatedly observed for distinct conformational populations and the frame conditions (e.g., intensity of folded and unfolded state) need to be determined for each protein. Upon a set-up of these frame conditions the intensity of the FlAsH fluorescence can be used to capture early misfolding events and to monitor the time course of aggregate formation, provided the FlAsH has been internalized in the cells before induction of the protein synthesis [8]. The key advantage of this system is that it allows to follow the whole protein folding history, starting from the early misfolding events assumed to generate the most toxic species in the aggregation process. Moreover, the FlAsH-mediated fluorescence can be used to monitor the equilibrium stability of a protein of interest by direct in-cell urea titration [8], allowing thus to determine the direct effect of a mutation on the protein stability and the aggregation propensity of mutants.
Recent improvements of the original FlAsH-methodology have expanded the utility of this approach, providing a powerful tool for studying conformational dynamics in vivo. Nakanishi and coworkers [12] modified the bensoic acid moiety and developed another biarsenical analog of the Nile red (BarNile) with an environmentally sensitive fluorophore used to monitor conformational changes caused by binding of metal ions. The FlAsH-based techniques hold a great promise as a tool for study the protein misfolding and aggregation kinetics in the cell with a time window tracking the early aggregation phases.
Conclusions and Future Perspectives
Off-pathway misfolding and aggregation are highly undesirable processes both reflecting the cell's fitness, resulting in physiological dysfunctions (e.g., neurodegeneration), and the production of recombinant proteins. In vivo investigations, in the same background the newly synthesized protein encounters, are tremendously helpful to elucidate the mechanisms of protein folding/misfolding. The general need, however, goes towards detecting soluble and insoluble species after the biosynthesis process has reached equilibrium. Besides the development of methods with easy readouts, the approaches have to be conceptually focused on spanning the time window towards the early folding/unfolding events. Real-time kinetic studies are necessary to track the entire process and they will accelerate the development of strategies to suppress the undesired off-pathway unfolding reactions. In addition to understanding the mechanisms of folding and aggregation within the cell, the direct monitoring of the protein solubility in vivo has wide practical applicability. For instance, in the recombinant production of proteins, culture conditions can be modulated and optimized by monitoring their direct effect on the protein solubility. Of particular interest would be to engineer expression hosts, e.g., use of bacterial strains deficient in certain intracellular proteases, or strains with up-regulated expression of cellular chaperones and to follow in real time their influence on the recombinant protein production. The direct monitoring of the folding ability in vivo opens up the possibility of high-throughput screening for the production of proteins with improved solubility by random mutagenesis and facilitating the design of novel protein structures.
Acknowledgements
The author is a recipient of the Heisenberg fellowship of the DFG (IG 73/1-1). I appreciate critical reading and fruitful discussion of the manuscript by Lila Gierasch, Kenneth S. Rotondi, Joanna Swain, and Beena Krishnan.
==== Refs
Adams SR Campbell RE Gross LA Martin BR Walkup GK Yao Y Llopis J Tsien RY New biarsenical ligands and tetracysteine motifs for protein labeling in vitro and in vivo: synthesis and biological applications J Am Chem Soc 2002 124 6063 6076 12022841 10.1021/ja017687n
Dobson CM Protein misfolding, evolution and disease Trends Biochem Sci 1999 24 329 32 10470028 10.1016/S0968-0004(99)01445-0
Frand AR Cuozzo JW Kaiser CA Pathways for protein disulphide bond formation Trends Cell Biol 2000 10 203 210 10754564 10.1016/S0962-8924(00)01745-1
Ghaemmaghami S Fitzgerald MC Oas TG A quantitative high-throughput screen for protein stability Proc Natl Acad Sci U S A 2000 97 8296 8301 10890887 10.1073/pnas.140111397
Ghaemmaghami S Oas TG Quantitative protein stability measurement in vivo Nat Struct Biol 2001 8 879 882 11573094 10.1038/nsb1001-879
Heim R Cubitt AB Tsien RY Improved green fluorescence Nature 1995 373 663 664 7854443 10.1038/373663b0
Griffin BA Adams SR Jones J Tsien RY Fluorescent labeling of recombinant proteins in living cells with FlAsH Methods Enzymol 2000 327 565 578 11045009
Ignatova Z Gierasch LM Monitoring protein stability and aggregation in vivo by real-time fluorescent labeling Proc Natl Acad Sci U S A 2004 101 523 528 14701904 10.1073/pnas.0304533101
Johnsson N Varshavsky A Split ubiquitin as a sensor of protein interactions in vivo Proc Natl Acad Sci U S A 1994 91 10340 10344 7937952
Wieruszeski JM Bohin A Bohin JP Lippens G In vivo detection of the cyclic osmoregulated periplasmic glucan of Ralstonia solanacearum by high-resolution magic angle spinning NMR J Magn Reson 2001 151 118 123 11444945 10.1006/jmre.2001.2348
Maxwell KL Mittermaier AK Forman-Kay JD Davidson AR A simple in vivo assay for increased protein solubility Prot Sci 1999 8 1908 1911
Nakanishi J Maeda M Umezawa Y A new protein conformation indicator based on biarsenical fluorescein with an extended benzoic acid moiety Anal Sci 2004 20 273 278 15055950 10.2116/analsci.20.273
Pervushin K Riek R Wider G Wuthrich K Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution Proc Natl Acad Sci U S A 1997 94 2366 12371 10.1073/pnas.94.23.12366
Philipps B Hennecke J Glockshuber R FRET-based in vivo screening for protein folding and increased protein stability J Mol Biol 2003 327 239 49 12614622 10.1016/S0022-2836(03)00077-9
Raquet X Eckert JH Muller S Johnsson N Detection of altered protein conformations in living cells J Mol Biol 2001 305 927 938 11162103 10.1006/jmbi.2000.4239
Ross CA Poirier MA Protein aggregation and neurodegenerative disease Nat Med 2004 10 S10 S17 15272267 10.1038/nm1066
Serber Z Dötsch V In-cell NMR spectroscopy Biochemistry 2001 40 14317 14323 11724542 10.1021/bi011751w
Serber Z Keatinge-Clay AT Ledwidge R Kelly AE Miller SM Dötsch V High-resolution macromolecular NMR spectroscopy inside living cells J Am Chem Soc 2001 123 2446 2447 11456903 10.1021/ja0057528
Serber Z Ledwidge R Miller SM Dötsch V Evaluation of parameters critical to observing proteins inside living Escherichia coli cells by in-cell NMR spectroscopy J Am Chem Soc 2001 123 8895 8901 11552796 10.1021/ja0112846
Tsien RY The green fluorescent protein Annu Rev Biochem 1998 67 509 544 9759496 10.1146/annurev.biochem.67.1.509
Waldo GS Standish BM Berendzen J Terwilliger TC Rapid protein-folding assay using green fluorescent protein Nat Biotechnol 1999 17 691 695 10404163 10.1038/10904
Wieruszeski JM Bohin A Bohin JP Lippens G In vivo detection of the cyclic osmoregulated periplasmic glucan of Ralstonia solanacearum by high-resolution magic angle spinning NMR J Magn Reson 2001 151 118 23 11444945 10.1006/jmre.2001.2348
Wigley WC Stidham RD Smith NM Hunt JF Thomas PJ Protein solubility and folding monitored in vivo by structural complementation of a genetic marker protein Nat Biotechnol 2001 19 131 6 11175726 10.1038/84389
Wurth C Guimard NK Hecht MH Mutations that reduce aggregation of the Alzheimer's Aβ42 peptide: an unbiased search for the sequence determinants of Aβ amyloidogenesis J Mol Biol 2002 319 1279 1290 12079364 10.1016/S0022-2836(02)00399-6
Yancey PH Clark ME Hand SC Bowlus RD Somero GN Living with water stress: evolution of osmolyte systems Science 1982 217 1214 1222 7112124
Zimmerman SB Minton AP Macromolecular crowding: biochemical, biophysical, and physiological consequences Annu Rev Biophys Biomol Struct 1993 22 27 65 7688609 10.1146/annurev.bb.22.060193.000331
|
16120213
|
PMC1208943
|
CC BY
|
2021-01-04 16:24:35
|
no
|
Microb Cell Fact. 2005 Aug 24; 4:23
|
utf-8
|
Microb Cell Fact
| 2,005 |
10.1186/1475-2859-4-23
|
oa_comm
|
==== Front
Microb Cell FactMicrobial Cell Factories1475-2859BioMed Central London 1475-2859-4-251612239210.1186/1475-2859-4-25ReviewOrganic acid toxicity, tolerance, and production in Escherichia coli biorefining applications Warnecke Tanya [email protected] Ryan T [email protected] Department of Chemical and Biological Engineering, UCB424/ECCH120, University of Colorado, Boulder, CO 80309, USA2005 25 8 2005 4 25 25 8 7 2005 25 8 2005 Copyright © 2005 Warnecke and Gill; licensee BioMed Central Ltd.2005Warnecke and Gill; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Organic acids are valuable platform chemicals for future biorefining applications. Such applications involve the conversion of low-cost renewable resources to platform sugars, which are then converted to platform chemicals by fermentation and further derivatized to large-volume chemicals through conventional catalytic routes. Organic acids are toxic to many of the microorganisms, such as Escherichia coli, proposed to serve as biorefining platform hosts at concentrations well below what is required for economical production. The toxicity is two-fold including not only pH based growth inhibition but also anion-specific effects on metabolism that also affect growth. E. coli maintain viability at very low pH through several different tolerance mechanisms including but not limited to the use of decarboxylation reactions that consume protons, ion transporters that remove protons, increased expression of known stress genes, and changing membrane composition. The focus of this mini-review is on organic acid toxicity and associated tolerance mechanisms as well as several examples of successful organic acid production processes for E. coli.
==== Body
Review
Biorefining Platforms
Biorefining promises the development of efficient processes for the conversion of renewable sources of carbon and energy into large volume commodity chemicals. It has been estimated that such bioprocesses already account for 5% of the 1.2 trillion dollar US chemical market [1], with some projecting future values of up to 50% of the total US chemical market generated through biological means. While the attractiveness of such bioprocesses has been recognized for some time [2,3], recent advances in biological engineering and associated sciences [4-15], several biorefining success stories [16-18], and instability in the price and future availability of oil [19], have collectively reinvigorated interest in the large scale production of chemicals through biological routes. Nevertheless, many challenges still remain for the economical bio-production of commodity chemicals. Such challenges encompass the need to not only inexpensively convert biomass into usable sources of carbon and energy but also to engineer microbes to produce relevant chemicals at high titers and productivities while minimizing the generation of byproducts that might foul downstream processes [1,20,21]. One model for addressing the latter of such challenges involves the generation of platform organisms that can be easily engineered and re-engineered to produce a variety of building block chemicals that are amenable to conversions to higher value products via traditional catalytic routes (see Figure 1). Although chemical pretreatment of raw materials impairs viability of platform organisms, this review will focus on product toxicity issues associated with the production of organic acids in E. coli (for further information on sugar extraction from raw materials see Zaldavar, et al. [22] and Knauf, et al. [23]).
Figure 1 Conceptual model of toxicity in biorefining applications. Sugars are extracted from waste biomass for use as feedstock for platform organisms in a biorefinery. Metabolically engineered microorganisms convert sugars into valuable platform chemicals that are then further derivatized to large-volume chemicals. Product and feedstock toxicity are observed, thus limiting productivity of biorefining applications.
The US Department of Energy (USDOE) recently released a prioritized list of building block chemicals for future biorefining endeavors. Priority was assigned based on the projected value of the platform chemical and potential derivatives as well as what technological developments were required for the production of the chemical and associated derivatives [21]. The report emphasized the importance of organic acids to the future of biorefining efforts (eight of the top twelve chemicals were organic acids, see Table 1 in additional file 1). The USDOE is not the first to recognize the importance of organic acids. In fact, there is a rich literature describing microbial production of organic acids [17,20,24,25], including several successful commercial bioprocesses [26-28]. Product toxicity is one of the primary challenges in the development of organic acid bioprocesses based on the use of platform host organisms, such as E. coli. In particular, while E. coli is known to survive very high concentrations of acids (pH = 2) when passing through the mammalian stomach, E. coli are surprisingly acid sensitive in exponential phase when cultured planktonically [29,30]. Moreover, undissociated organic acids, which pass freely through the outer and plasma membranes of E. coli [31,32], dissociate upon entry into the slightly alkaline cytoplasm releasing protons that lower internal pH (pHi) and anions that specifically inhibit different aspects of metabolism resulting in impaired growth [33-35]. Titers and productivities of 50–100 g/L and 2–3 g/L·hr are expected for the economical manufacturing of most building block acids by fermentation. The pKa values range from 3–5 for these organic acids, which would result in a pH reduction to around 2.0 for titers of 50 g/L. This highlights a key challenge in the metabolic engineering of organic acid production hosts. That is, high titers result in the addition of protons to the culture, which either result in a decreased pH or the addition of large volumes of base titrant. At low pH, organic acids are undissociated, thus they pass freely through the membrane and inhibit growth. At high pH, the process is less efficient due to base requirements and because export of the organic acid cannot proceed by free diffusion alone (for a more detailed discussion of organic acid export issues see Van Maris et al. [36]). What is desired, therefore, is a platform organism that not only produces high levels of organic acid chemicals but also is tolerant to any associated toxicity.
Many microbes are capable of producing platform chemicals by aerobic and anaerobic fermentation processes [22]. L-lactic acid has traditionally been produced by lactic acid bacteria. Although many lactic acid bacteria strains have been studied extensively [37], the ability to produce optically pure L-lactic acid is hampered by the presence of both L and D lactate dehydrogenase genes [38]. Pure L-lactic acid must therefore be produced via another pathway, as the racemic acid product is not useful for downstream conversion into polylactic acid. A number of other microorganisms have been used for industrial fermentation of several of the building block organic acids identified in Table 1. Large scale production of amino acids has been accomplished in Corynebacteriumglutamicum [39], succinic acid has been produced by Actinobacillus succinogenes [40], and itaconic acid production has been carried out with Aspergillus terrus [41]. While successful, the future application of these organisms as platform hosts is limited when compared with E. coli. E. coli is advantageous as a platform host because it is the most well characterized model organism, it has been used in recombinant processes for over 20 years, there are a wide variety of good genetic tools, and it is sensitive to many antibiotics used in genetic engineering efforts [42]. Moreover, the completion of the E. coli genome sequence has already enabled many functional genomics studies and proven useful in metabolic engineering efforts [43]. Finally, E. coli grows quickly in minimal media and maintains the ability to metabolize both 5 and 6 carbon sugars, which is a specific advantage over the use of industrially relevant yeast strains [22]. This mini-review will describe the basic mechanisms underlying organic acid toxicity and associated tolerance pathways in E. coli followed by a short discussion of several metabolic engineering strategies employed for the production of organic acids in E. coli.
Organic Acid Toxicity in E. coli
One of the primary factors contributing to the toxicity of organic acids is their ability to diffuse across E. coli cellular membranes when undissociated as opposed to the restricted passage of dissociated protons and anions (see Figure 2) [31,32]. Diffusion of dissociated acids is limited to secondary transport, which is known to involve H+/monocarboxylic acid symporters. However, the detailed mechanism and specificities of the transporters remain unknown [31]. E. coli maintain a cytoplasmic pH (pHi = 7.5) that is most often higher than that of the external media and typically well above the pKa of organic acids [44,45]. As a result, organic acids exist in the dissociated form within the cytoplasm. Thus, diffusing organic acids entering into the cytoplasm will dissociate and disrupt the pHi and anion pool of the cytoplasm. The resulting increase in internal acidity can affect the integrity of purine bases [46] and result in denaturing of essential enzymes inside the cell [35], both of which negatively affect cell viability.
Figure 2 An overview of organic acid toxicity and tolerance mechanisms in E. coli. Diffusion of undissociated acid molecules can occur freely in acidic medium but is limited to transport systems at neutral or basic pH. The toxic effects associated with organic acids are the result of both anion specific affects on metabolism as well as increased internal proton concentrations. Affects on internal pH are mitigated by transport of protons out of the membrane, consumption of protons by decarboxylation reactions, and, more generally, induction of stress regulons. Anion specific tolerance mechanisms are not well characterized.
Organic acid anions affect cell growth in a variety of manners. Increased anion concentration has been shown to lead to an increased transport of potassium ions into the cell, which increases turgor pressure [47,48]. To maintain a constant turgor pressure and cell volume, glutamate is transported out of the cell [48]. This transport activity concomitantly disrupts the osmolarity of the cytoplasm, which in turn lowers the cell's growth potential and viability. In addition to this general anion effect, there are also effects specific to each organic acid. It has been proposed that enzymes involved in protein synthesis are sensitive to a combination of two unrelated mechanisms, including the acidification of pHi and the formation of an anionic pool [35]. Although this finding implies that the organic inhibition due to the anion pool could be acid specific, the details describing this dual inhibition mechanism remain unclear. Kirkpatrick et al. reported proteins exhibiting increased expression in response to extracellular acetate [33]. Among these are the OppA transporter, RpoS regulon, several amino acid uptake proteins, DNA binding proteins, and extreme-acid preiplasmic chaperones. Interestingly, when formate was introduced in place of acetate the expression of the previously mentioned proteins was repressed, indicating that the response was anion specific. This finding introduces new challenges in addressing organic acid tolerance. Specifically, it highlights the need to engineer both pH and as well as specific anion tolerance into host organisms.
Finally, production of organic acids might include intermediates that are themselves toxic. For example, 3-hydroxypropionic acid (3HP) is closely related to the antimicrobial compound Reuterin. Reuterin describes the hydroxypropionaldehyde (HPA) system including HPA, HPA dimer, and HPA hydrate. Reuterin is inhibitory to several bacteria, including E. coli, at concentrations as low as 0.03–0.05 g/L [49-51]. It is thought that the toxicity could be the result of inhibition of DNA synthesis [52]. It has been postulated that the reactivity of the aldehyde group of HPA causes DNA damage similarly to formaldehyde, which is the aldehyde analog of formic acid [49]. Intermediate toxicity can be managed either by optimization of the production pathway in the host or by engineering tolerance to the intermediate itself.
Organic Acid Tolerance in E. coli
E. coli has a remarkable ability to remain viable under a broad range of pH conditions. This ability is essential for its survival in the mammalian digestive system where pH can vary between pH = 2–8. Several different acid tolerance mechanisms have been identified in E. coli. While each mechanism is capable of providing some degree of tolerance, they are regulated differently and confer varying levels of tolerance.
Although most acid tolerance systems are activated in stationary phase, acid tolerance as low as pH = 3 has been observed in exponential phase E. coli grown under aerobic conditions, which is advantageous from a productivity standpoint [30]. Although the underlying tolerance mechanism is not known, such tolerance can be reliably activated by adapting cells at sublethal pH values between 4.3 and 5.8 [53]. E. coli that exhibit growth phase tolerance remain viable at pH values on the same order as stationary phase tolerance, however the percent survival is significantly lower. Lin et al. reported 1% survival of the original culture following acid adaption at pH 4.3 followed by acid challenge at pH 3.3 compared to 0.0001% survival for unadapted cultures. This is compared to stationary-phase cultures, which exhibited up to 50% survival.
Three stationary phase acid resistance systems have been studied in the most detail [29,30]. These systems confer the highest levels of tolerance and are believed to be responsible for stationary phase E. coli survival when passing through the mammalian stomach. Acid resistance system 1 (AR1) is activated in slightly acidic media (pH 5.5) in the absence of extracellular glucose or amino acids. E. coli grown aerobically under these conditions retain viability under acid challenges as low as pH = 2.5 [54]. This system is also referred to as the oxidative or glucose-repressed system, since the expression of this system is thought to be regulated either directly or indirectly by RpoS and cyclicAMP receptor protein (CRP) [55,56]. Acid resistance system 2 (AR2) is activated in E. coli grown in aerobic conditions in acidic complex media. This system requires the presence of extracellular glucose and glutamate and is dependent upon genes encoding glutamate decarboxylase (gadAB) and a glutamate:GABA antiporter (gabC) [30]. Under such conditions, E. coli have been demonstrated to exhibit acidic resistance up to a pH of 2. The mechanism involves the expenditure of excess cytoplasmic protons during amino acid decarboxylation reactions (see Figure 2), thus raising the internal pH [54,55]. Acid resistance system 3 (AR3) parallels the mechanisms of AR2 with several slight deviations [30,54,55]. AR3 is activated under anaerobic conditions, in complex media with added glucose. It also involves amino acid decarboxylation reactions to lower the internal pH, but requires extracellular arginine in place of glutamate. AR3 also requires increased expression of arginine decarboxylase and an arginine: agmatine antiporter for increased acid tolerance.
Finally, several general acid tolerance mechanisms that regulate the physical properties of the membrane or the effectiveness of ion transport have been identified. These active responses, or those that occur as a result of the cell's ability to sense pH changes, are independent of growth and are induced by pH shifts as small as 0.2 pH units [57]. The first response is the ability of the microorganism to adjust membrane properties, such as lipid content, thus effectively changing the proton permeability [57]. Another cellular response to acid shock is the induction of genes responsible for repairing and preventing lethal cellular damage. Specifically, increased expression of the oxyR and soxR regulatory genes has been observed by transcriptional profiling of acid tolerant phenotypes [45,58]. These systems regulate the removal of damaging oxidizing agents, thus preventing further DNA damage under acidic stress [46]. Finally, acid tolerance can be achieved by adjusting the ionic transporter efficiency, effectively regulating the anion and cation balance as a means of maintaining a constant internal pH [47].
Organic Acid Production in E. coli
Metabolic and genetic engineering, directed evolution, and classic strain selection have all been employed in the development of E. coli strains that produce building block organic acids, including lactic-acid, succinic acid, and 3HP [17,25,59,60]. Improved titers have been achieved due to optimization of fermentation conditions and relavant pathways utilized. However, titer limitations exist when fermentation is carried out in unbuffered media, which allows the pH to acidify due to increased acid concentration. Alternatively large amounts of base titrant are required to raise the pH of the media during the organic acid production leaving the final acid molecule in the undissociated form. Following production under these conditions, large volumes of acid must be added to recover the acid in the protonated form. Metabolic and genetic engineering of acid tolerance into production strains, making fermentation at a pH less than the pKa of the acid produced possible, would circumvent the need for the additional consumption of acid and base titrants, and thus lower the overall production cost. Similarly, engineering strain fitness to increase productivity at a decreased pH would improve productivity and reduce base consumption.
Lactic acid production is one of the most successful examples to date of the engineering of large volume chemical production in E. coli. E. coli was selected as a favorable host strain due to its ability to consume both pentose and hexose sugars and to generate optically pure L-lactic acid, which is the desired product for downstream polylactic acid (PLA) production [61,62]. An effective lactic acid producing strain of E. coli was created by induced expression of the L-specific lactic acid dehydrogenase (LDH) gene from Streptococcus bovis. High titers (50–75 g/L) were observed under controlled pH (pH = 7) and anaerobic conditions. Titers were drastically decreased (10–20 g/L) as the pH was allowed to drop with increasing acid production [59]. However, allowing the pH to fall below the pKa of lactic acid also resulted in decreased concentration of the acid in the undissociated form, which facilitated the subsequent isolation of the protonated acid. Interestingly, the choice of host strain made a significant difference in lactic acid production [59]. Those constructed from an E. coli B strain showed a titer of almost twice that produced from K12 derivatives. The increased production was attributed primarily to differences in the native growth characteristics rather than increased acid tolerance.
Economically competitive titers of succinic acid have also been achieved in E. coli. Strains were engineered to limit flux to other anaerobic byproducts normally formed during fermentation [60]. Specifically, succinic acid production was optimized by redirecting the metabolic flux at the pyruvate node away from lactate and formate through inactivation of the pyruvate-formatelyase and lactate dehydrogenase [60,63]. The maximum yield in succinic acid production was approximately 50 g/L in pH controlled cultures. However, similar to lactic acid studies, succinic acid production was significantly repressed when pH was not kept at neutral levels.
A final example of metabolic engineering organic acid production in E. coli was reported by Cargill in 2001 [17]. Suthers and Cameron engineered a 2-step glycerol to 3HP pathway in E. coli. Glycerol was first converted to 3HPA via a glycerol dehydratase enzyme (dhaB – isolated from Klebsiella pneumoniae). 3HPA was then converted to 3HP via an aldehyde dehydrogenase (ald). This first pathway was not ideal for several reasons including a very low reported titer (0.2 g/L), the use of the more expensive glycerol as opposed to glucose, and the generation of the highly toxic 3-HPA (reuterin) compound. Selifinova et al. later proposed five additional pathways for the production of 3-HP directly from glucose in E. coli [36]. Results for each of such pathways have yet to be reported. One issue that has yet to be addressed is how to fulfill the desire to produce 3-HP at a pH below the pKa = 4.51 of 3-HP, which would lessen the dependency on large volumes of base titrant to retain neutral pH at high titers.
Metabolic engineering of E. coli organic acid tolerance represents an important future opportunity. As discussed above, E. coli possess several systems for surviving pH as low as 2.0, which is much lower than what is required for an economical biorefining process. Since induction of these systems is well characterized and the relevant genes are known in many cases, future efforts might be better focused on the development of multi-stage fermentations that allow for generation of biomass prior to induction of acid tolerance and, ultimately, acid production. Future genetic engineering efforts might focus on engineering tolerance against the less well characterized metabolic effects associated with increased organic acid anion concentrations. For example, the addition of acetate, benzoate, and propionate to culture media at a concentration of 8 mM has been observed to inhibit growth of E. coli up to 50% [35]. The acetate inhibition is thought to be caused by limited methionine pools combined with increasing concentrations of homocysteine, a toxic intermediate, due to inactivation of a key enzyme in the methionine synthesis pathway, which can be countered by the addition of methionine to the media. This finding established that growth inhibition is the result of both of lowered pH and specific anionic effects, which decreases the activity of key enzymes. Thus, engineering tolerance to specific organic acid anion effects by increased expression of inhibited enzymes could aid in increasing overall process productivity.
Conclusion
Organic acids are a valuable sector of the industrial chemical market, which have already been successfully produced through microbial fermentation. However, product titers have been variable, ranging from less than 1 g/L to concentrations cost competitive with current petrochemical production processes. These fermentation processes have been limited in E. coli due to product and intermediate toxicity. Toxicity is directly measured by growth inhibition, which specifically decreases productivity. This review highlighted what is known about organic-acid toxicity and tolerance mechanisms in E. coli. Specifically, E. coli are growth inhibited by the increase in both proton and associated anion concentrations that are characteristic of organic-acid production processes. While several acid-tolerance mechanisms have been characterized in E. coli, anion specific mechanisms require additional study. Thus, future metabolic engineering efforts that seek to improve understanding of these issues within the context of organic-acid biorefining applications should prove useful.
Supplementary Material
Additional File 1
Table 1: Organic acids for platform biorefining applications. (* see references [64,65])
Click here for file
==== Refs
Bachmann R Riese J From promise to profit Industrial Biotechnology 2005 1 9 15
Leeper SA Ward TE Andrews GF Production of Organic Chemicals via Bioconversion: A Review for Potential 1991 US DOE's Idaho Operations Office
Herrera S Industrial biotechnology - a chance at redemption Nature Biotechnology 2004 22 671 675 15175684
Crameri A Raillard S Bermudez E Stemmer WPC DNA shuffling of a family of genes from diverse species accelerates directed evolution Nature 1994 370 389 390 8047147
Patnaik R Louie S Gavrilovic V Perry K Stemmer WPC Ryan CM del Cardayré S Genome shuffling of Lactobacillus for improved acid tolerance Nature Biotechnology 2002 20 707 712 12089556
Canada KA Iwashita S Shim H Wood TK Directed evolution of toluene ortho-monooxygenase for enhanced 1-naphthol synthesis and chlorinated ethene degradation Journal Of Bacteriology 2002 184 344 349 11751810
Farmer WR Liao JC Improving lycopene production in Escherichia coli by engineering metabolic control Nature Biotechnology 2000 18 533 537 10802621
Fodor SPA Read JL Pirrung MC Stryer L Lu AT Solas D Light-Directed, Spatially Addressable Parallel Chemical Synthesis Science 1991 251 767 773 1990438
Gill RT Wildt S Yang YT Ziesman S Stephanopoulos G Genome-wide screening for trait conferring genes using DNA microarrays Proceedings Of The National Academy Of Sciences Of The United States Of America 2002 99 7033 7038 11997466
Martin VJJ Pitera DJ Withers ST Newman JD Keasling JD Engineering a mevalonate pathway in Escherichia coli for production of terpenoids Nature Biotechnology 2003 21 796 802 12778056
Ohnishi J Mitsuhashi S Hayashi M Ando S Yokoi H Ochiai K Ikeda M A novel methodology employing Corynebacterium glutamicum genome information to generate a new L-lysine-producing mutant Applied Microbiology And Biotechnology 2002 58 217 223 11876415
Schmidt-Dannert C Umeno D Arnold FH Molecular breeding of carotenoid biosynthetic pathways Nature Biotechnology 2000 18 750 753 10888843
Bailey JE Towards a science of metabolic engineering Science 1991 252 1668 1675 2047876
Bailey JE Sburlati A Hatzimanikatis V Lee K Renner WA Tsai PS Inverse metabolic engineering: A strategy for directed genetic engineering of useful phenotypes Biotechnology And Bioengineering 2002 79 568 579 12209828
Stephanopoulos G Vallino JJ Network Rigidity And Metabolic Engineering In Metabolite Overproduction Science 1991 252 1675 1681 1904627
Higley DP Sun Y Acid-dyeable polymer compositions. US patent 938760 2004
Suthers PF Cameron DC Production of 3-Hydroxypropionic acid in recombinant organisms. PCT WO 01-16346 2001
Gatenby AA Haynie SL Nagarajan Method for the production of 1,3-propanediol by recombinant organisms. WO 9821339 1998
Deffeyes KS Hubbert's Peak: The Impending World Oil Shortage 2001 Princeton, NJ , Princeton University Press
Chotani G Dodge T Hsu A Kumar M LaDuca R Trimbur D Weyler W Sanford K The commercial production of chemicals using pathway engineering Biochimica Et Biophysica Acta-Protein Structure And Molecular Enzymology 2000 1543 434 455
Werpy T Petersen G Top Value Added Chemicals from Biomass Volume 1: Results of Screening for Potential Candidates from Sugars and Synthetic Gas 2004 Oak Ridge, TN , U.S. Department of Energy
Zaldivar J Nielsen J Olsson L Fuel ethanol production from lignocellulose: a challenge for metabolic engineering and process integration Applied Microbiology And Biotechnology 2001 56 17 34 11499926
Knauf M Moniruzzaman M Lignocellulosic biomass processing: A perspective International Sugar Journal 2004 106 147 150
Trends in Life Sciences: US Market 2002 Washington DC , International Access Corporation (IAC)
Selifonova OV Jessen H Gort SJ Selmer T Buckel W 3-Hydroxypropionic acid and other organic compounds. PCT WO 02/42418 2002
Ngheim NP Donnelly MI Millard CS Stols L Method for the production of dicarboxylic acids 1999
Reichstein T Process for the manufacture of levoascorbic acid (vitamin C). US patent 2,265,121 1941
Skory CD Fungal lactate dehydrogenase gene and constructs for the expression thereof. US Patent 535381 2000
Richard H Foster JW Escherichia coli glutamate- and arginine-dependent acid resistance systems increase internal pH and reverse transmembrane potential Journal Of Bacteriology 2004 186 6032 6041 15342572
Richard HT Foster JW Acid resistance in Escherichia coli Advances In Applied Microbiology, Vol 52 Advances In Applied Microbiology 2003 52 167 186
Poole RC Halestrap AP Transport Of Lactate And Other Monocarboxylates Across Mammalian Plasma-Membranes American Journal Of Physiology 1993 264 C761 C782 8476015
Walter A Gutknecht J Monocarboxylic Acid Permeation Through Lipid Bilayer-Membranes Journal Of Membrane Biology 1984 77 255 264 6699907
Kirkpatrick C Maurer LM Oyelakin NE Yoncheva YN Maurer R Slonczewski JL Acetate and formate stress: Opposite responses in the proteome of Escherichia coli Journal Of Bacteriology 2001 183 6466 6477 11591692
Roe AJ McLaggan D Davidson I O'Byrne C Booth IR Perturbation of anion balance during inhibition of growth of Escherichia coli by weak acids Journal Of Bacteriology 1998 180 767 772 9473028
Roe AJ O'Byrne C McLaggan D Booth IR Inhibition of Escherichia coli growth by acetic acid: a problem with methionine biosynthesis and homocysteine toxicity Microbiology-Sgm 2002 148 2215 2222
van Maris AJA Konings WN van Dijken JP Pronk JT Microbial export of lactic and 3-hydroxypropanoic acid: implications for industrial fermentation processes Metabolic Engineering 2004 6 245 255 15491854
Narayanan N Roychoudhury PK Srivastava A L (+)lactic acid fermentation and its product polymerization Electronic Journal Of Biotechnology 2004 7 167 U2
Saitoh S Ishida N Onishi T Tokuhiro K Nagamori E Kitamoto K Takahashi H Genetically engineered wine yeast produces a high concentration of L-lactic acid of extremely high optical purity Applied And Environmental Microbiology 2005 71 2789 2792 15870375
Ikeda M Nakagawa S The Corynebacterium glutamicum genome: features and impacts on biotechnological processes Applied Microbiology And Biotechnology 2003 62 99 109 12743753
Zeikus JG Jain MK Elankovan P Biotechnology of succinic acid production and markets for derived industrial products Applied Microbiology And Biotechnology 1999 51 545 552
Willke T Vorlop KD Biotechnological production of itaconic acid Applied Microbiology And Biotechnology 2001 56 289 295 11548996
Nelson DL Cox MM Lehninger Principles of Biochemistry 2000 3rd New York, NY , Worth Publisher
Harrington CA Rosenow C Retief J Monitoring gene expression using DNA microarrays Current Opinion In Microbiology 2000 3 285 291 10851158
Goulbourne E Matin M Zychlinsky E Matin A Mechanism Of Delta-pH Maintenance In Active And Inactive Cells Of An Obligately Acidophilic Bacterium Journal Of Bacteriology 1986 166 59 65 3007441
Maurer LM Yohannes E Bondurant SS Radmacher M Slonczewski JL pH regulates genes for flagellar motility, catabolism, and oxidative stress in Escherichia coli K-12 Journal Of Bacteriology 2005 187 304 319 15601715
Choi SH Baumler DJ Kaspar CW Contribution of dps to acid stress tolerance and oxidative stress tolerance in Escherichia coli O157 : H7 Applied And Environmental Microbiology 2000 66 3911 3916 10966408
Kroll RG Booth IR The Relationship Between Intracellular Ph, The Ph Gradient And Potassium-Transport In Escherichia-Coli Biochemical Journal 1983 216 709 716 6365081
McLaggan D Naprstek J Buurman ET Epstein W Interdependence Of K+ And Glutamate Accumulation During Osmotic Adaptation Of Escherichia-Coli Journal Of Biological Chemistry 1994 269 1911 1917 7904996
Sung HW Chen CN Chang Y H.F. L Biocompatibility Study of Biological Tissues Fixed by a Natural Compound (Reuterin) Produced by Lactobacillus Reuteri Biomaterials 2002 23 3203 3214 12102192
Ganzle MG Reutericyclin: biological activity, mode of action, and potential applications Applied Microbiology And Biotechnology 2004 64 326 332 14735324
Luthi-Peng Q Dileme FB Puhan Z Effect of glucose on glycerol bioconversion by Lactobacillus reuteri Applied Microbiology And Biotechnology 2002 59 289 296 12111160
Rasch M The influence of temperature, salt and pH on the inhibitory effect of reuterin on Escherichia coli International Journal of Food Microbiology 2002 72 225 231 11845821
Goodson M Rowbury RJ Habituation to normally lethal acidity by prior growth of Escherichia coli at a sub-lethal acid pH value Lett Appl Microbiol 1989 8 77 79
Lin JS Smith MP Chapin KC Baik HS Bennett GN Foster JW Mechanisms of acid resistance in enterohemorrhagic Escherichia coli Applied And Environmental Microbiology 1996 62 3094 3100 8795195
Castanie-Cornet MP Penfound TA Smith D Elliott JF Foster JW Control of acid resistance in Escherichia coli Journal Of Bacteriology 1999 181 3525 3535 10348866
Lin JS Lee IS Frey J Slonczewski JL Foster JW Comparative-Analysis Of Extreme Acid Survival In Salmonella-Typhimurium, Shigella-Flexneri, And Escherichia-Coli Journal Of Bacteriology 1995 177 4097 4104 7608084
Booth IR The regulation of intracellular pH in bacteria Novartis Found Symposium 1999 221 19 37
Storz G Imlay JA Oxidative Stress Current Opinions in Microbiology 1999 2 188 194
Dien BS Nichols NN Bothast RJ Recombinant Escherichia coli engineered for production of L-lactic acid from hexose and pentose sugars Journal Of Industrial Microbiology & Biotechnology 2001 27 259 264 11687940
Vemuri GN Eiteman MA Altman E Effects of growth mode and pyruvate carboxylase on succinic acid production by metabolically engineered strains of Escherichia coli Applied And Environmental Microbiology 2002 68 1715 1727 11916689
Chang DE Jung HC Rhee JS Pan JG Homofermentative production of D- or L-lactate in metabolically engineered Escherichia coli RR1 Applied And Environmental Microbiology 1999 65 1384 1389 10103226
Gruber PR Hall ES Kolstad JJ Iwen ML Benson RD Borchardt RL Continuous process for manufacture of lactide polymers with controlled optical purity. US Patent 5142023 1992 Cargill, Inc.
Donnelly MI Millard CS Clark DP Chen MJ Rathke JW A novel fermentation pathway in an Escherichia coli mutant producing succinic acid, acetic acid, and ethanol Applied Biochemistry And Biotechnology 1998 70-2 187 198 9627382
Marz U RGA-103R Worldwide Markets for Fermentation Ingredients
Paster M Pellegrino JL Carole TM Energetics Incorporated Industrial Bioproducts: Today and Tomorrow 2003 Washington DC , US Department of Energy, Office of Energy Efficiency and Renewable Energy & Office of the Biomass Program
|
16122392
|
PMC1208944
|
CC BY
|
2021-01-04 16:24:35
|
no
|
Microb Cell Fact. 2005 Aug 25; 4:25
|
utf-8
|
Microb Cell Fact
| 2,005 |
10.1186/1475-2859-4-25
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-281608350110.1186/1476-4598-4-28ResearchTumor suppressor in lung cancer 1 (TSLC1) alters tumorigenic growth properties and gene expression Sussan Thomas E [email protected] Mathew T [email protected] Yoshinori [email protected] Roger H [email protected] Department of Physiology, Johns Hopkins University School of Medicine, Baltimore, MD 21205-2185, USA2 Tumor Suppression & Functional Genomics Project, National Cancer Center Research Institute, Tokyo 104-0045, Japan2005 5 8 2005 4 28 28 25 4 2005 5 8 2005 Copyright © 2005 Sussan et al; licensee BioMed Central Ltd.2005Sussan et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Introduction of cDNA or genomic clones of the tumor suppressor in lung cancer 1 (TSLC1) gene into the non-small cell lung cancer line, A549, reverses tumorigenic growth properties of these cells. These results and the observation that TSLC1 is down-regulated in a number of tumors suggest that TSLC1 functions as a critical switch mediating repression of tumorigenesis.
Results
To investigate this mechanism, we compared growth properties of A549 with the TSLC1-containing derivative. We found a G1/S phase transition delay in 12.2. Subtractive hybridization, quantitative PCR, and TranSignal Protein/DNA arrays were used to identify genes whose expression changed when TSLC1 was up-regulated. Members of common G1/S phase regulatory pathways such as TP53, MYC, RB1 and HRAS were not differentially expressed, indicating that TSLC1 may function through an alternative pathway(s). A number of genes involved in cell proliferation and tumorigenesis were differentially expressed, notably genes in the Ras-induced senescence pathway. We examined expression of several of these key genes in human tumors and normal lung tissue, and found similar changes in expression, validating the physiological relevance of the A549 and 12.2 cell lines.
Conclusion
Gene expression and cell cycle differences provide insights into potential downstream pathways of TSLC1 that mediate the suppression of tumor properties in A549 cells.
RIS1Ras-induced senescenceNSCLClung cancerA549TSLC1
==== Body
Background
Non-small cell lung cancer (NSCLC) includes squamous and large cell carcinomas and adenocarcinoma. NSCLC accounts for approximately 75% of all lung cancers diagnosed in the United States [1]. Genetic mutations that activate oncogenes such as KRAS2 and NRAS [2], and loss of function in tumor suppressors such as RB1, TP53, PPP2R1B, CDKN2A, and TSLC1 have been demonstrated in NSCLC tumors [3-7].
A549 is derived from an NSCLC adenocarcinoma and displays several properties that are characteristic of transformed cells, including a short cell cycle, loss of contact inhibition, and rapid development of tumors following injection into athymic mice [8]. Introduction of a 1.1 Mb YAC derivative containing the TSLC1 gene into A549 restored TSLC1 expression to normal levels, creating the stable cell line, 12.2 [8]. 12.2 cells do not develop tumors following injection into athymic mice. TSLC1 protein is down-regulated or lost in NSCLC and a number of other neoplastic diseases, including pancreatic [7], hepatocellular [7], breast [9], prostate [10], nasopharyngeal [11], gastric [12], and cervical cancers [13]. Reduction or loss of TSLC1 expression is also observed in cell lines derived from esophageal, ovarian, endometrial, small-cell lung and colorectal tumors [14].
The product of TSLC1 is a transmembrane glycoprotein that forms dimers both within a cell and between adjacent cells to promote cell-cell adhesion [15]. This protein contains structural domains homologous to members of the immunoglobulin superfamily, NCAM adhesion proteins, and the nectin family of Ca2+-independent cell-cell adhesion proteins [7,16]. It contains two protein-protein interaction domains that are required for tumor suppressor activity [17]. TSLC1 interacts with the actin cytoskeleton through DAL-1, which implies that it plays a role in cell motility [18]. The TSLC1 gene has been isolated in a number of different experimental paradigms and has received multiple names as a consequence, including IGSF4, BL2, ST17, SynCAM1, SgIGSF, RA175, and NECL2 [16,19-22].
Because TSLC1 by itself can reverse tumorigenic and metastatic properties of the highly aggressive A549 cell line, it is of interest to identify downstream effectors of this potent tumor suppressor. Identification of genes or pathways activated by TSLC1 would help to characterize the molecular switch from tumorigenic to non-tumorigenic growth. We characterized the growth differences that result from restoration of TSLC1 expression to normal levels and used a number of approaches to identify the underlying changes in gene expression. Several genes involved in Ras-induced senescence, endometrial stromal cell decidualization and trophoblast implantation in the uterus were differentially regulated. Additional genes contributing to cell growth, adhesion, and energy production showed altered expression, as well. We did not find evidence that TSLC1 works through any of several previously-characterized cell cycle regulatory pathways. Several expression changes were confirmed in the small amounts of tumor and normal tissue obtained from histological specimens. Thus analysis of this tumor suppressor in the readily accessible A549/12.2 cell system may provide insights into a new gene expression cascade involved in suppression of transformation.
Results
TSLC1 Alters Growth Properties of A549 Cells
Introduction of the TSLC1 gene or cDNA into adenocarcinoma-derived A549 cells restores its expression to normal levels and suppresses many tumorigenic properties of this line [7,8]. We extended observations about the inhibitory effect of TSLC1 expression on A549 cell growth [8] by showing that 12.2 cells expanded to only 28% of A549 levels after five days (Table 1 and Fig. 1). This same result was seen with WST-1 reagent, which showed that 48 hours after plating there was a significantly reduced number of viable 12.2 cells relative to A549 (data not shown).
Table 1 Expansion rate of A549 and 12.2 cell lines. Expansion rates of A549 and 12.2 cells were determined by counting cells 24 h and 120 h after plating 5 × 104 cells. Results of two independent experiments are shown.
Cell line 24 hours 120 hours Fold Increase (120 h/24 h) Growth Upregulation (A549/12.2)
A549 5.73 × 104 2.27 × 106 39.7 3.40
12.2 4.53 × 104 5.28 × 105 11.7
A549 2.81 × 104 8.12 × 105 28.9 3.79
12.2 3.59 × 104 2.74 × 105 7.6
Figure 1 Cell doubling assay of A549 and 12.2 cells. Cell number was counted at Day 2 and Day 5, and normalized to Day 1. The results of two independent experiments are shown.
We used flow cytometry to examine how TSLC1 affects apoptosis and cell cycle. Rates of apoptosis in A549 and the TSLC1-expressing 12.2 cell lines were compared after staining with annexin V. No difference was detected in the number of apoptotic cells (Fig. 2A and 2B). Next, we stained cells with propidium iodide and examined cell cycle profiles of A549 and 12.2 (Fig. 2C and Table 2). The 12.2 cell line showed a significant accumulation of cells in G1 phase (74.4%) compared to A549 (60.4%). Fewer 12.2 cells were seen in S and G2/M phase (17.2 and 8.9%, respectively) when compared to A549 (28.2 and 11.9%, respectively). Thus, the decreased growth rate of 12.2 is due to reduced cell division, which occurs at least in part to a delay at the G1/S phase checkpoint, resulting in delayed progression into S phase.
Figure 2 Flow cytometry analysis of apoptosis and cell cycle in A549 and 12.2. Histograms of (A) A549 or (B) 12.2 cells stained with annexin V-FITC. Percentages are averages of four experiments; histograms are from one representative experiment. (C) Cell cycle histograms of A549 and 12.2. Cells were fixed, stained with propidium iodide, and analyzed for DNA content by flow cytometry analysis. The left peak represents 2N cells in G1 phase and the right peak represents 4N cells in G2/M phase.
Table 2 Cell cycle profiles of cell lines determined by flow cytometry. Flow cytometry analysis of cell cycle profiles was determined for A549 and 12.2. Percentages were quantified using CellQuest software. A549 and 12.2 profiles are the means of three experiments. The difference between A549 and 12.2 is statistically significant (p < 0.005).
Cell Cycle Stage Percent Total Cells
A549 12.2
G1 60.4+/-0.3 74.4+/-1.8
S 28.2+/-0.5 17.2+/-2.3
G2/M 11.9+/-0.7 8.9+/-1.0
Expression of Signaling Pathway Genes
Differences in growth rates and cell cycle profiles between A549 and 12.2 led us to examine expression of several known checkpoint and signaling pathway genes using quantitative RT-PCR (qPCR). Alterations in the Ras/p53 pathway result in abnormal G1/S transition in some NSCLC [23]. However, mRNA levels of HRAS, p19, RB1, and TP53 were not substantially different between A549 and 12.2 (Table 3). The minor differences in expression levels were not coordinately regulated in a way that would explain lengthening of the G1/S transition in 12.2 cells. We also examined MYC and cyclin D1 (CCND1), which promote G1 to S phase transition. Neither CCND1 nor MYC, its upstream regulator, were differentially expressed. Thus, neither of these established pathways appears to be responsible for the G1 delay.
Table 3 Comparison of gene expression in A549 and 12.2 cell lines. Genes expressed differentially between A549 and 12.2 cells were identified by qPCR, subtractive hybridization (SH), and/or TransSignal DNA-protein array (TS) as indicated. Relative quantification differences were determined for those genes analyzed by either qPCR or TS. ND, not determined.
Gene Up-Regulated Assay Fold Expression Difference Functional Class
Fibrinogen beta chain (FGB) A549 SH/qPCR 2737.5+/-1058.5 Adhesion
Fibrinogen gamma chain (FGG) A549 SH ND Adhesion
Adenomatosis polyposis coli (APC)* A549 qPCR 1.7+/-0.2 Cellular Growth
Cyclin D1 (CCND1)† A549 qPCR 1.6+/-0.0 Cellular Growth
β-catenin 1 (CTNNB1)* 12.2 qPCR 1.1+/-0.0 Cellular Growth
Dishevelled 1 (DVL1)* A549 qPCR 1.1+/-0.3 Cellular Growth
v-Ha-ras Harvey rat sarcoma viral oncogene homolog (HRAS)† A549 qPCR 1.4+/-0.3 Cellular Growth
Lymphoid enhancer-binding factor 1 (LEF1)* 12.2 qPCR 3.4+/-0.4 Cellular Growth
v-myc myelocytomatosis viral oncogene homolog (MYC)† A549 qPCR 1.3+/-0.2 Cellular Growth
Interleukin 23-alpha (p19)† A549 qPCR 1.0+/-0.0 Cellular Growth
Tumor protein p53 (TP53)† A549 qPCR 1.7+/-0.3 Cellular Growth
Retinoblastoma 1 (RB1)† 12.2 qPCR 1.3+/-0.4 Cellular Growth
S100 calcium-binding protein P (S100P) A549 SH/qPCR 1680.0+/-955.4 Cellular Growth
Transcription factor 4 (TCF4)* 12.2 qPCR 6.8+/-2.0 Cellular Growth
Transcription factor 7-like 2 (TCF7L2)* 12.2 qPCR 2.2+/-0.7 Cellular Growth
Transmembrane 4 superfamily member (TM4SF4) A549 SH/qPCR 475.2+/-328.0 Cellular Growth
Centromere protein E (CENPE) A549 SH Cellular Growth
Heat shock protein 70B (HSPA6) A549 SH/qPCR 2.4+/-0.8 Chaperone
Insulin-like growth factor binding protein 1 (IGFBP1) A549 SH/qPCR 15.9+/-7.4 Decidualization [41]
Retinoid X receptor RXR (DR-1) A549 TS 2.7 Decidualization [42]
Aldehyde dehydrogenase 1 (ALDH1) A549 SH/qPCR 4.2+/-0.1 Decidualization-Implantation [43]
Annexin A2 (Lipocortin II) (ANXA2) 12.2 SH/qPCR 3.1+/-1.0 Decidualization-Implantation [43]
Metallothionein-IF (MTIF) 12.2 SH ND Decidualization-Implantation [43]
Metallothionein-IG (MT1G) 12.2 SH/qPCR 37.1+/-10.4 Decidualization-Implantation [43]
Cadherin 11 (OB-cadherin, osteoblast) (CDH11) A549 SH/qPCR 5.6+/-2.7 Decidualization-Luteal/Adhesion [34,44]
Fibroblast growth factor 9 (FGF9) A549 SH ND Decidualization-Proliferation [45]
Promyelocytic leukemia gene (PML) A549 SH ND Decidualization-Proliferation
Similar to aldehyde dehydrogenase 6 (ALDH1A3) A549 SH ND Dehydrogenase
3-alpha hydroxysteroid dehydrogenase type II (AKR1C3) A549 SH ND Dehydrogenase
Dihydrodiol dehydrogenase 2 (AKR1C2) A549 SH/qPCR 22.9+/-2.4 Dehydrogenase
Kinesin 2, light chain (KNS2) 12.2 SH/qPCR 3.9+/-0.0 Intracellular Trafficking
Matrix metalloproteinase 1 (MMP1) 12.2 qPCR 1.2+/-0.4 Invasion
Vascular endothelial growth factor (VEGF) 12.2 qPCR 4.2+/-0.3 Invasion
Ferritin, heavy chain (FTH1) 12.2 SH/qPCR 2.7+/-1.7 Iron Binding
Metallothionein-IE (MT1E) 12.2 SH/qPCR 14.8+/-8.5 Metal Homeostasis
Metallothionein-IH (MT1H) 12.2 SH ND Metal Homeostasis
Metallothionein-IL (MT1L) 12.2 SH/qPCR 5.8+/-0.9 Metal Homeostasis
Metallothionein-IR (MT1R) 12.2 SH ND Metal Homeostasis
Manganese-containing superoxide dismutase (SOD2) A549 SH/qPCR 2.0+/-0.6 Mitochondrial
Microsomal glutathione transferase (MGST1) A549 SH/qPCR 2.3+/-0.1 Mitochondrial
NAD(P)H menadione oxidoreductase 1, dioxin-inducible (NQO1) A549 SH/qPCR 3.1+/-0.1 Mitochondrial
Solute carrier family 25 member 5 (SLC25A5) A549 SH/qPCR 3.4+/-0.3 Mitochondrial
Myelin proteolipid protein (PLP) A549 SH ND Myelin Constituent
Ribosomal protein S6 (RPS6) 12.2 SH/qPCR 2.5+/-1.2 Protein Kinase
v-ets erythroblastosis virus E26 oncogene homolog 2 (ETS2) A549 qPCR 1.2+/-0.0 Ras-Induced Senescence [29]
Metallothionein-II (MTII) 12.2 SH ND Ras-Induced Senescence [29]
Ras induced senescence 1 (RIS1) 12.2 SH/qPCR 80.2+/-62.3 Ras-Induced Senescence [29]
Tissue inhibitor of metalloproteinase 1 (TIMP1) 12.2 qPCR 2.2+/-0.1 Ras-Induced Senescence [29]
CAAT box general (CBF) 12.2 TS 2.5 Transcription Factor
CCAAT displacement protein (CDP) 12.2 TS 2 Transcription Factor
E2F transcription factor 1 (E2F-1) 12.2 TS 2.4 Transcription Factor
Early growth response (EGR) 12.2 TS 3.4 Transcription Factor
Estrogen receptor (ERE) 12.2 TS 2.5 Transcription Factor
GATA binding protein (GATA) 12.2 TS 2.1 Transcription Factor
Glucocorticoid receptor (GRE) 12.2 TS 3 Transcription Factor
Nuclear factor of activated T-cells, cytoplasmic (NF-ATc) 12.2 TS 3.1 Transcription Factor
Signal transducer and activator or transcription 3 (STAT3) A549 TS 2 Transcription Factor
Signal transducer and activator or transcription 4 (STAT4) A549 TS 2.3 Transcription Factor
Upstream transcription factor (USF-1) A549 TS 2.5 Transcription Factor
Transcription co-activator Sp110 A549 SH ND Transcription Factor
Similar to eukaryotic translation initiation factor 2B, subunit 1 (EIF2B1) A549 SH ND Translation
Insulin-like 4 (INSL4) A549 SH/qPCR 27.0+/-15.8 Trophoblast [46]
Keratin 8 (KRT8) A549 SH ND Trophoblast [47]
cDNA clone DKFZp761C1(AL157447) A549 SH ND Unknown
cDNA clone FLJ20643 (AK000650) A549 SH ND Unknown
FLJ14639 (NM_032815) A549 SH ND Unknown
Hit clone 451B21 (AL033522) A549 SH ND Unknown
HSA1p34 genomic sequence (AL009181) A549 SH ND Unknown
Rhomboid family 1 (Z69719) (RHBDF1) A549 SH ND Unknown
RP11-2H8 (AC089984) A549 SH ND Unknown
RP11-389E6 (AL359836) A549 SH ND Unknown
RP11-478J18 (AC011700) A549 SH ND Unknown
RP11-7F24 (AC018841) A549 SH ND Unknown
RP1-20C7 (AL136304) A549 SH ND Unknown
SR+89 (Z69706) A549 SH ND Unknown
HSA14 genomic sequence (AL135745) 12.2 SH ND Unknown
* Wnt/β-catenin pathway
†G1/S transition
Alterations in the Wnt/β-catenin pathway are commonly observed in colorectal cancers and other solid tumors, including NSCLC. We detected no differences in mRNA levels of three upstream genes in the Wnt/β-catenin pathway, disheveled (DVL1), adenomatosis polyposis coli (APC), and β-catenin (CTNNB1). However, transcriptional regulators downstream of β-catenin, TCF4, TCF7L2, and LEF1, were up-regulated in the suppressed 12.2 cell line, by 6.8-, 2.2-, and 3.4-fold, respectively (Table 3).
A549 cells are metastatic in experimental systems [24], and elevated TSLC1 expression is correlated with lower levels of metastasis and invasion in esophageal squamous cell carcinoma [25]. Accordingly, we examined levels of several genes involved in angiogenesis and metastasis. Increased expression of the angiogenic factor, VEGF, is correlated with metastasis and poor prognosis in solid tumors [26]. Surprisingly, this gene was up-regulated 4.2+/-0.3 fold in the suppressed 12.2 cells. Metalloproteinases have roles in various stages of primary tumor progression, invasion and metastasis. In A549 and 12.2, matrix metalloproteinase 1 (MMP1) showed no expression difference, while tissue inhibitor of metalloproteinase 1 (TIMP1) showed a small (2.2+/-0.2-fold) up-regulation in 12.2. These results show that some genes associated with metastasis are altered by TSLC1, but not necessarily as predicted by previously described pathways in tumorigenic cells.
Proteomic Analysis of Transcription Factors
We characterized changes in transcription factor (TF) expression more broadly using the TranSignal Protein/DNA Array I (Panomics Inc., Redwood City, CA). Of the 54 TFs analyzed, 4 were down-regulated 2-fold or more in the 12.2 line and 8 were up-regulated in 12.2 (Table 3). In general, those up-regulated in 12.2 have been previously associated with repression of tumorigenesis. E2F1, which regulates the G1/S phase transition and has tumor suppressor properties, is increased in 12.2, consistent with our results from cell cycle analysis. The signal transducer and activator of transcription (Stat) proteins, which promote cellular proliferation, were down-regulated in 12.2. Interestingly, several TFs commonly disregulated in cancer, including AP-1, c-Myb, Ets, Sp1, Myc-associated factor X, NFκB and p53, were not altered between A549 and 12.2. (For complete list of TFs analyzed, see . These results are consistent with qPCR expression analysis, which did not show differences in cDNA levels for TP53, MYC, and ETS2 (Table 3).
Differential gene expression analysis
Subtractive hybridization was performed between A549 cells and the suppressed 12.2 cells to identify genes differentially expressed when TSLC1 is restored to normal levels. The procedure was performed in both directions to produce two populations of cDNA enriched for messages up-regulated in A549 or in 12.2, respectively. Each of the differentially expressed cDNA populations was hybridized to Genome Systems cDNA arrays, identifying 41 genes greatly over-expressed in the tumorigenic A549 line and 18 genes over-expressed in the suppressed 12.2 cell line (Table 3). The differentially expressed genes represented a variety of functional classes, including those with roles in cell proliferation, cell survival, protein phosphorylation, immune response, cell adhesion, or detoxification. Several genes whose products are localized to the mitochondria were also differentially expressed.
Relative transcript levels for 31 of the differentially-expressed genes were quantified using qPCR (Table 3). Expression was normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and alpha-tubulin (TUBA1). qPCR showed that TSLC1 expression was 2.8+/-0.8 times higher in 12.2 than in A549, as expected. Twenty-one of the 31 genes analyzed by qPCR showed expression differences of 2-fold or more in the direction predicted by subtractive hybridization. One gene, complement component C5 (CCC5), showed a 13.8+/-4.7-fold down-regulation in 12.2 cells, contrary to results expected after subtractive hybridization. The nine remaining genes changed less than 2-fold.
Several genes involved in cellular proliferation were among those expressed at high levels in A549 but low levels in 12.2 (Table 3). Cadherin 11 (CDH11), which plays a role in cell-cell interactions was also down-regulated in 12.2. Furthermore, several mitochondrial genes were down-regulated in 12.2 (Table 3). The genes that were most highly up-regulated with the restoration of TSLC1 expression in 12.2 cells were the candidate tumor suppressor Ras-induced senescence 1 (RIS1), metallothionein 1G (MT1G) and metallothionein 1E (MT1E).
Gene expression in tumor vs. normal tissue
We compared gene expression in normal vs. tumor tissue to determine whether differences in 12.2 and A549 cells reflect changes that occur in vivo (Table 4). Tissue was recovered from pathological specimens and transcript levels were measured by qPCR, normalized to GAPDH as described [14]. TSLC1 and RIS1 levels were lower than normal in 5/5 tumor specimens, while S100 calcium-binding protein P (S100P) and insulin-like growth factor binding protein 1 (IGFBP1) levels were elevated in 5/5 and 4/5 tumors, respectively. Thus, key differences observed between the transformed (A549) and suppressed (12.2) cell lines reflect physiological differences seen in tumor vs. normal tissue.
Table 4 Gene expression profiles in NSCLC tumor vs. normal lung parallel those in A549 and 12.2 Relative fold expression differences, determined by qPCR, were determined in tumors and compared to normal lung tissue from same patient. Positive values represent higher expression in tumor; negative values represent higher expression in normal tissue.
Relative Gene Expression in Tumors
TSLC1 RIS1 S100P IGFBP1
Patient 1 -25.7 -2.3 33.3 600
Patient 2 -23.8 -6.4 2.1 2.3
Patient 3 -16.9 -7.1 28.1 -8.4
Patient 4 -344.3 -43.9 3.2 6.1
Patient 5 -32.4 -44.9 16.1 1.9
Cell Lines Down in A549 Down in A549 Up in A549 Up in A549
Discussion
Restoration of TSLC1 expression to normal levels in A549 cells reverses several transformed properties of this line, slowing the growth rate, restoring contact inhibition, eliminating its ability to form tumors in nude mice and blocking metastasis. In this study, we have shown that restoration of TSLC1 expression in 12.2 cells reduced cell growth by 3.6-fold. This change in growth rate dynamics was not due to an increase in the number of apoptotic cells. Rather, flow cytometry revealed that 12.2 cells experience a delay relative to A549 cells in progression from G1 to S phase of the cell cycle. These differences in growth related properties between A549 and 12.2 were similar to those observed previously when tumor cells were transfected with a vector containing TSLC1 cDNA or genomic clone [7,8,25]. These phenotypic differences led us to examine expression of genes associated with adhesion, invasion, basal metabolism, cell growth, senescence and apoptosis in A549 and 12.2 to identify classes of genes altered by restoration of TSLC1 expression in 12.2.
The most highly up-regulated gene in 12.2 cells was the putative tumor suppressor, Ras-induced senescence 1 (RIS1). RIS1 is located at 3p21.3, which is a region that is deleted in many human tumors [27]. Also, RIS1 is located within a 1 Mb human chromosomal region that is commonly deleted during tumor formation in human/mouse microcell hybrids that are passaged through severe combined immunodefficient (SCID) mice [28]. This region, called chromosome 3 common eliminated region 1 (CER1), has been posited to contain one or more currently unidentified tumor suppressors. Our results provide support for RIS1 as a candidate.
Coordinate up-regulation of RIS1, metallothionein II, and TIMP1 was observed in Ras-induced senescent cells [29]. Consistent with this study, we found increased levels of RIS1, several metallothioneins, and TIMP1 in 12.2 cells. Activation of these genes suggests that TSCL1-mediated inhibition of tumorigenesis may be related to the Ras-induced senescence pathway. However, previous studies showed that RIS1 expression is dependent on ETS2, an inducer of Ras-induced senescence, in human fibroblast IMR90 cells [29]. ETS2 was not differentially expressed between A549 and 12.2 cells in this study (Table 3) suggesting that RIS1 is activated in 12.2 cells through a different pathway.
Several genes were identified that have known roles in cancer. S100P, which was down-regulated in the non-tumorigenic 12.2 cell line, is involved in cell growth and has been previously implicated in prostate [30] and breast cancer progression [31]. It is over-expressed in lung adenocarcinomas [32], which frequently exhibit reduced levels of TSLC1 expression[7]. However, down-regulation of S100P in A549 using antisense RNA was not sufficient to alter growth related phenotypes by itself (data not shown). Insulin-like growth factor 4 (INSL4), also down-regulated in 12.2, has been reported to be over-expressed in highly invasive breast cancer cells [33]. Reduced expression of S100P and INSL4 in 12.2 may contribute to the slower growth rate and loss of tumorigenic properties in the 12.2 cell line when TSLC1 expression is restored to normal levels.
Several differentially expressed genes identified in this study have been previously shown, in non-overlapping experiments, to be differentially expressed during various stages of endometrial stromal cell decidualization and trophoblast implantation (Table 3). The relationship between these processes and neoplastic transformation in NSCLC is not clear. However, it is interesting that these seemingly unrelated events show similar patterns of gene expression changes. Decidualization and implantation are characterized by high levels of proliferation and tissue invasion; properties shared with transformed cells. Together, these observations suggest that TSLC1 may repress transformed growth via some of the same pathways that regulate proliferation in endometrial cells during various stages of decidualization.
Subtractive hybridization and qPCR showed that restoration of TSLC1 lowered expression of CDH11, an adhesion protein that may enhance cellular invasion [34]. CDH11 is expressed in highly invasive but not in noninvasive breast cancer cell lines [34,35]. It has been shown to associate with β-catenin (CNNTB1)[36]. While we did not see a significant difference in expression of upstream genes in the Wnt-1/β-catenin pathway (DVL1, CDKN2A, CNNTB1) between A549 and 12.2, we did see expression differences in CTNNB1-dependent transcription factors LEF1, TCF4, and TCF7L2, which were up-regulated in 12.2. This may be a consequence of down-regulation of CDH11 leading to lower levels of CNNTB1 sequestered at the plasma membrane. This unbound CNNTB1 could then translocate to the nucleus to activate downstream genes. Retera et. al. [37] demonstrated that CNNTB1 expression is reduced in NSCLC primary tumors and metastases. Our results suggest that downstream effectors of CNNTB1, such as LEF1, TCF4, and TCF7L2, may be involved in suppressing tumorigenic properties in 12.2.
Although the growth difference in A549 and 12.2 is characterized by a significant delay at G1/S in the latter, we did not find significant changes in gene expression for common G1/S phase regulators HRAS, p19, RB1, TP53, MYC, and CCND1. Also, the increased expression of VEGF in 12.2 cells contrasts with observations from many tissues which show that this gene is up-regulated in tumor cells. This suggests that TSLC1 does not suppress tumorigenesis through any of these common pathways. However, several of these genes are regulated at the protein level or through localization to the nucleus. In order to address this concern, we examined c-Myc and cyclin D1 protein levels by western blot, and found no difference in expression between A549 and 12.2 (data not shown). Furthermore, the Panomics TranSignal Protein/DNA Array found no difference in expression of Myc-associated factor X, NFκB, c-Myb, and AP-1.
Cell lines are artifactual by definition, and they do not perfectly replicate in vivo conditions. However, comparison of key gene expression patterns in matched tumor-normal tissue pairs showed that our results with A549 and 12.2 are representative of in vivo expression levels. These results validate the physiological relevance of our in vitro expression analysis in a model system that is far more amenable to experiment than is the minute amount of material recovered from histological specimens.
It is notable that acute expression of high levels of TSLC1 in A549 cells has a somewhat different effect on cell cycle profiles than does the long-term restoration of this gene in 12.2 cells. Infection for 3 or 5 days with adenovirus vectors expressing a TSLC1 cDNA (Ad-TSLC1) induced apoptosis and increased annexin V staining in infected cultures [38]. This contrasts with stable restoration of TSLC1 expression in the 12.2 line, which does not demonstrate elevation in annexin V staining. Since the 12.2 cell line was selected after transfection of TSLC1, it adds valuable insights into the normal function of TSLC1 in non-transformed cells. TSLC1 has the ability to suppress the transformed growth properties of A549, and it alters the gene expression profile of A549 to resemble that of normal relative to transformed lung tissue. A part of its normal function as a potent tumor suppressor may be to regulate cell growth by initiating apoptosis in those rare cells that initiate neoplastic transformation.
Conclusion
Restoration of TSLC1 levels in the tumorigenic A549 cell line resulted in a loss of transformed growth properties, including a reduced cell doubling rate and a delayed progression from G1 to S phase during the cell cycle. This corresponded with a change in the gene expression profile, including changes in genes with roles in Ras-induced senescence and endometrial decidualization. Other genes with roles in cell proliferation were also altered when TSLC1 levels were restored, including IGFBP1, S100P, and INSL4. TSLC1 does not appear to act through any of several well-characterized cell growth regulatory pathways.
Elucidating the mechanisms by which TSLC1 represses tumorigenesis would have an important impact on the understanding of cancer biology in the lung, as well as in the numerous other tissues where TSLC1 has been associated with cancer progression. This study reveals several cellular phenotypes associated with TSLC1 expression and provides insights into the genes and molecular pathways induced by TSLC1.
Methods
Cell Culture and Tumor Samples
The A549 cell line (American Type Culture Collection [ATCC], Manassas, VA) was cultured in Dulbecco's modified Eagle's medium (Invitrogen Corp., Carlsbad, CA) supplemented with 10% fetal bovine serum (Hyclone Laboratories, Inc., Logan UT), 1X non-essential amino acids and 1% penicillin-streptomycin (Invitrogen Corp., Carlsbad, CA) in 5% CO2. The suppressed 12.2 cell line was created by transfecting YAC derivative y939-95 into A549 cells [8]. 12.2 cells were cultured under the same conditions as A549, with the addition of 500 μg/ml G418 (Mediatech, Inc., Herndon, VA), except for cell growth assays, in which G418 was omitted.
For the growth assays, duplicate aliquots of 5 × 104 cells were plated in six-well dishes. After 24, 48, and 120 h, cells were trypsinized and three aliquots from each well were counted using a hemacytometer. The average of six counts (three each, for two wells) is reported here.
For the WST-1 cellular proliferation assay (Roche Applied Science, Indianapolis, IN), 1 × 104 cells were cultured for 48 h. Samples were incubated with WST-1 reagent for 1 h, and absorbance was measured at 450 nm and 620 nm.
Five primary NSCLC tumors and corresponding non-cancerous lung tissues from the same patients were surgically resected and histologically diagnosed at National Cancer Center, Japan. All samples were immediately frozen after surgical resection and stored at -135°C. The analyses of human samples were carried out in accordance with the institutional guidelines.
Proteomic Analysis
Proteomic analysis was performed using the TranSignal Protein/DNA Array I (Panomics Inc., Redwood City, CA). Nuclear protein extracts from A549 or 12.2 were incubated with an excess of biotinylated cis-binding elements (CBE) of 54 common transcription factors (TFs). Unbound CBE were removed and the protein/DNA complexes were separated, leaving labeled CBE which represent the relative protein levels of the 54 TFs. These were hybridized to a DNA array and visualized using streptavidin-HRP. Hybridized probe was quantified using the AlphaImager v5.5 software (Alpha Innotech Corp. San Leandro, CA).
Subtractive Hybridization
Total RNA was isolated from A549 and 12.2 using Trizol reagent (Invitrogen Corp.), and poly(A) mRNA was purified using the PolyATract mRNA Isolation System II (Promega, Madison, WI). Subtractive hybridization was performed between A549 and 12.2 in both directions using the PCR-Select cDNA Subtraction Kit (Clontech, Palo Alto, CA).
The enriched pools of cDNA were hybridized to human Gene Discovery Array cDNA nylon filters (Genome Systems, Inc., St. Louis, MO). Samples (15 μl) of the final PCR reaction from each subtractive hybridization reaction were radioactively labeled by random prime-labeling with [32P]dCTP [39,40] and purified using ProbeQuant G-50 Sephadex columns (Amersham Pharmacia Biotech, Inc., Piscataway, NJ). The filters were prehybridized for 2 h at 42°C in buffer consisting of 0.75 M NaCl, 0.1 M Na2HPO4, 0.1% Na4P2O7-10H2O, 0.15 M Tris (pH 7.5), 5X Denhardt's solution, 2% SDS, and 100 μg/ml sheared salmon testis DNA (Sigma-Aldrich, St. Louis, MO). Probes were hybridized overnight at 42°C in the same buffer.
The membranes were washed in 2X SSC for 5 min at room temperature, twice in 2X SSC with 1% SDS for 30 min at 68°C, and twice in 0.6X SSC with 1% SDS for 30 min at 68°C. The filters were then rinsed in room-temperature 2X SSC and placed on film for 3 days for the A549 over-expressed population and 2 weeks for the 12.2 over-expressed isolates. Identities of associated EST sequences for positive clones were obtained from the Genome Systems website . EST sequences were analyzed by BLAST, using the non-redundant database to obtain gene annotation for positive clones.
Quantitative PCR
RNA was isolated from A549, 12.2, or the antisense clones with Trizol reagent (Invitrogen Corp.) and used to generate cDNA using Superscript II reverse transcriptase (Invitrogen Corp.). Quantitative PCR (qPCR) was carried out using the LightCycler rapid thermal cycler system and the SYBR Green FastStart PCR kit (Roche Diagnostics Ltd., Lewes UK). Primers were used at 0.5 μM and MgCl2 at 4 mM. Samples were heat-denatured for 10 min., then cycled 55 times at 95°C for 10 sec., 58°C for 5 sec., and 72°C for 20 sec. At the completion of the cycling, a melting curve analysis was performed to detect the presence of multiple products. A standard curve was generated based on serial dilutions of PAC 66B10 and primers for marker 66B10.SP6 (CCTGGTAGTGGATTTCCCAA and ATGCCATTCAGTTTGTTCCC). Samples were normalized to glyceraldehyde 3-phosphate (ACCACAGTCCATGCCATCAC and TCCACCACCCTGTTGCTGTA). Primers for each gene were designed using the Primer3 program .
Flow Cytometry
For the apoptosis assay, A549 and 12.2 cells were grown without antibiotic selection. Cells were trypsinized and stained with annexin V and propidium iodide as recommended (BD Biosciences Pharmingen, San Diego, CA). Cells were analyzed on a Becton Dickinson FACScan.
For cell cycle studies, 2 × 106 cells were collected and resuspended in 1 ml cold PBS, and 4 ml of -20°C 100% ethanol was slowly added. Cells were stored at -20°C overnight, recovered by centrifugation and resuspended in 1 ml PBS. RNase A (20 μg/ml) (Sigma-Aldrich Chemical Co., St. Louis, MO) was added, and the samples were incubated at 37°C for 30 min. Samples were incubated in 100 μg/ml propidium iodide (Sigma-Aldrich Chemical Co.) at room temperature for at least 1 h prior to analysis on a Becton Dickinson FACScan.
Authors' contributions
TES carried out qPCR, protein analysis, flow cytometry, and drafted the manuscript. MTP performed the subtractive hybridization. YM acquired and analyzed tumor and normal lung tissue. RHR was responsible for the study design and coordinated data analysis.
Acknowledgements
We would like to thank David Graham (Johns Hopkins School of Medicine, Baltimore, MD) for his comments on the manuscript. Insightful comments on the design and interpretation of these experiments were provided by Chi Van Dang (Johns Hopkins School of Medicine, Baltimore, MD). This work was supported by PHS awards HD24605 and HD38384 (RHR).
==== Refs
Ihde DC Minna JD Non-small cell lung cancer. Part I: Biology, diagnosis, and staging Curr Probl Cancer 1991 15 61 104 1649734
Suzuki Y Orita M Shiraishi M Hayashi K Sekiya T Detection of ras gene mutations in human lung cancers by single-strand conformation polymorphism analysis of polymerase chain reaction products Oncogene 1990 5 1037 1043 2197591
Birrer MJ Minna JD Genetic changes in the pathogenesis of lung cancer Annu Rev Med 1989 40 305 317 2658754 10.1146/annurev.me.40.020189.001513
Kishimoto Y Murakami Y Shiraishi M Hayashi K Sekiya T Aberrations of the p53 tumor suppressor gene in human non-small cell carcinomas of the lung Cancer Res 1992 52 4799 4804 1324794
Sachse R Murakami Y Shiraishi M Hayashi K Sekiya T DNA aberrations at the retinoblastoma gene locus in human squamous cell carcinomas of the lung Oncogene 1994 9 39 47 8302602
Wang SS Esplin ED Li JL Huang L Gazdar A Minna J Evans GA Alterations of the PPP2R1B gene in human lung and colon cancer Science 1998 282 284 287 9765152 10.1126/science.282.5387.284
Kuramochi M Fukuhara H Nobukuni T Kanbe T Maruyama T Ghosh HP Pletcher M Isomura M Onizuka M Kitamura T Sekiya T Reeves RH Murakami Y TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer Nat Genet 2001 27 427 430 11279526 10.1038/86934
Murakami Y Nobukuni T Tamura K Maruyama T Sekiya T Arai Y Gomyou H Tanigami A Ohki M Cabin D Frischmeyer P Hunt P Reeves RH Localization of tumor suppressor activity important in nonsmall cell lung carcinoma on chromosome 11q Proc Natl Acad Sci U S A 1998 95 8153 8158 9653156 10.1073/pnas.95.14.8153
Allinen M Peri L Kujala S Lahti-Domenici J Outila K Karppinen SM Launonen V Winqvist R Analysis of 11q21-24 loss of heterozygosity candidate target genes in breast cancer: indications of TSLC1 promoter hypermethylation Genes Chromosomes Cancer 2002 34 384 389 12112527 10.1002/gcc.10079
Fukuhara H Kuramochi M Fukami T Kasahara K Furuhata M Nobukuni T Maruyama T Isogai K Sekiya T Shuin T Kitamura T Reeves RH Murakami Y Promoter Methylation of TSLC1 and Tumor Suppression by Its Gene Product in Human Prostate Cancer Jpn J Cancer Res 2002 93 605 609 12079507
Hui AB Lo KW Kwong J Lam EC Chan SY Chow LS Chan AS Teo PM Huang DP Epigenetic inactivation of TSLC1 gene in nasopharyngeal carcinoma Mol Carcinog 2003 38 170 178 14639656 10.1002/mc.10156
Honda T Tamura G Waki T Jin Z Sato K Motoyama T Kawata S Kimura W Nishizuka S Murakami Y Hypermethylation of the TSLC1 gene promoter in primary gastric cancers and gastric cancer cell lines Jpn J Cancer Res 2002 93 857 860 12716461
Steenbergen RD Kramer D Braakhuis BJ Stern PL Verheijen RH Meijer CJ Snijders PJ TSLC1 gene silencing in cervical cancer cell lines and cervical neoplasia J Natl Cancer Inst 2004 96 294 305 14970278
Fukami T Fukuhara H Kuramochi M Maruyama T Isogai K Sakamoto M Takamoto S Murakami Y Promoter methylation of the TSLC1 gene in advanced lung tumors and various cancer cell lines Int J Cancer 2003 107 53 59 12925956 10.1002/ijc.11348
Masuda M Yageta M Fukuhara H Kuramochi M Maruyama T Nomoto A Murakami Y The tumor suppressor protein TSLC1 is involved in cell-cell adhesion J Biol Chem 2002 277 31014 31019 12050160 10.1074/jbc.M203620200
Shingai T Ikeda W Kakunaga S Morimoto K Takekuni K Itoh S Satoh K Takeuchi M Imai T Monden M Takai Y Implications of nectin-like molecule 2/IGSF4/RA175/SgIGSF/TSLC1/SynCAM1 in cell-cell adhesion and transmembrane protein localization in epithelial cells J Biol Chem 2003
Mao X Seidlitz E Ghosh K Murakami Y Ghosh HP The cytoplasmic domain is critical to the tumor suppressor activity of TSLC1 in non-small cell lung cancer Cancer Res 2003 63 7979 7985 14633730
Yageta M Kuramochi M Masuda M Fukami T Fukuhara H Maruyama T Shibuya M Murakami Y Direct association of TSLC1 and DAL-1, two distinct tumor suppressor proteins in lung cancer Cancer Res 2002 62 5129 5133 12234973
Gomyo H Arai Y Tanigami A Murakami Y Hattori M Hosoda F Arai K Aikawa Y Tsuda H Hirohashi S Asakawa S Shimizu N Soeda E Sakaki Y Ohki M A 2-Mb sequence-ready contig map and a novel immunoglobulin superfamily gene IGSF4 in the LOH region of chromosome 11q23.2 Genomics 1999 62 139 146 10610705 10.1006/geno.1999.6001
Biederer T Sara Y Mozhayeva M Atasoy D Liu X Kavalali ET Sudhof TC SynCAM, a synaptic adhesion molecule that drives synapse assembly Science 2002 297 1525 1531 12202822 10.1126/science.1072356
Wakayama T Ohashi K Mizuno K Iseki S Cloning and characterization of a novel mouse immunoglobulin superfamily gene expressed in early spermatogenic cells Mol Reprod Dev 2001 60 158 164 11553913 10.1002/mrd.1072
Urase K Soyama A Fujita E Momoi T Expression of RA175 mRNA, a new member of the immunoglobulin superfamily, in developing mouse brain Neuroreport 2001 12 3217 3221 11711859 10.1097/00001756-200110290-00015
Niklinski J Niklinska W Laudanski J Chyczewska E Chyczewski L Prognostic molecular markers in non-small cell lung cancer Lung Cancer 2001 34 Suppl 2 S53 8 11720742 10.1016/S0169-5002(01)00345-2
Hirai K Shimada H Ogawa T Taji S The spread of human lung cancer cells on collagens and its inhibition by type III collagen Clin Exp Metastasis 1991 9 517 527 1752082 10.1007/BF01768580
Ito T Shimada Y Hashimoto Y Kaganoi J Kan T Watanabe G Murakami Y Imamura M Involvement of TSLC1 in progression of esophageal squamous cell carcinoma Cancer Res 2003 63 6320 6326 14559819
Dvorak HF Vascular permeability factor/vascular endothelial growth factor: a critical cytokine in tumor angiogenesis and a potential target for diagnosis and therapy J Clin Oncol 2002 20 4368 4380 12409337 10.1200/JCO.2002.10.088
Kok K Naylor SL Buys CH Deletions of the short arm of chromosome 3 in solid tumors and the search for suppressor genes Adv Cancer Res 1997 71 27 92 9111863
Yang Y Kiss H Kost-Alimova M Kedra D Fransson I Seroussi E Li J Szeles A Kholodnyuk I Imreh MP Fodor K Hadlaczky G Klein G Dumanski JP Imreh S A 1-Mb PAC contig spanning the common eliminated region 1 (CER1) in microcell hybrid-derived SCID tumors Genomics 1999 62 147 155 10610706 10.1006/geno.1999.5952
Barradas M Gonos ES Zebedee Z Kolettas E Petropoulou C Delgado MD Leon J Hara E Serrano M Identification of a candidate tumor-suppressor gene specifically activated during Ras-induced senescence Exp Cell Res 2002 273 127 137 11822868 10.1006/excr.2001.5434
Mousses S Bubendorf L Wagner U Hostetter G Kononen J Cornelison R Goldberger N Elkahloun AG Willi N Koivisto P Ferhle W Raffeld M Sauter G Kallioniemi OP Clinical validation of candidate genes associated with prostate cancer progression in the CWR22 model system using tissue microarrays Cancer Res 2002 62 1256 1260 11888886
Guerreiro Da Silva ID Hu YF Russo IH Ao X Salicioni AM Yang X Russo J S100P calcium-binding protein overexpression is associated with immortalization of human breast epithelial cells in vitro and early stages of breast cancer development in vivo Int J Oncol 2000 16 231 240 10639564
Beer DG Kardia SL Huang CC Giordano TJ Levin AM Misek DE Lin L Chen G Gharib TG Thomas DG Lizyness ML Kuick R Hayasaka S Taylor JM Iannettoni MD Orringer MB Hanash S Gene-expression profiles predict survival of patients with lung adenocarcinoma Nat Med 2002 8 816 824 12118244
Brandt B Roetger A Bidart JM Packeisen J Schier K Mikesch JH Kemming D Boecker W Yu D Buerger H Early placenta insulin-like growth factor (pro-EPIL) is overexpressed and secreted by c-erbB-2-positive cells with high invasion potential Cancer Res 2002 62 1020 1024 11861376
Pishvaian MJ Feltes CM Thompson P Bussemakers MJ Schalken JA Byers SW Cadherin-11 is expressed in invasive breast cancer cell lines Cancer Res 1999 59 947 952 10029089
Feltes CM Kudo A Blaschuk O Byers SW An alternatively spliced cadherin-11 enhances human breast cancer cell invasion Cancer Res 2002 62 6688 6697 12438268
Shibata T Ochiai A Kanai Y Akimoto S Gotoh M Yasui N Machinami R Hirohashi S Dominant negative inhibition of the association between beta-catenin and c-erbB-2 by N-terminally deleted beta-catenin suppresses the invasion and metastasis of cancer cells Oncogene 1996 13 883 889 8806677
Retera JM Leers MP Sulzer MA Theunissen PH The expression of beta-catenin in non-small-cell lung cancer: a clinicopathological study J Clin Pathol 1998 51 891 894 10070329
Mao X Seidlitz E Truant R Hitt M Ghosh HP Re-expression of TSLC1 in a non-small-cell lung cancer cell line induces apoptosis and inhibits tumor growth Oncogene 2004 23 5632 5642 15184878 10.1038/sj.onc.1207756
Feinberg AP Vogelstein B A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity Anal Biochem 1983 132 6 13 6312838 10.1016/0003-2697(83)90418-9
Schwartz DC Cantor CR Separation of yeast chromosome-sized DNAs by pulsed field gradient gel electrophoresis Cell 1984 37 67 75 6373014 10.1016/0092-8674(84)90301-5
Liu HC Mele C Catz D Noyes N Rosenwaks Z Production of insulin-like growth factor binding proteins (IGFBPs) by human endometrial stromal cell is stimulated by the presence of embryos J Assist Reprod Genet 1995 12 78 87 7545484
Tarrade A Rochette-Egly C Guibourdenche J Evain-Brion D The expression of nuclear retinoid receptors in human implantation Placenta 2000 21 703 710 10985974 10.1053/plac.2000.0568
Kao LC Tulac S Lobo S Imani B Yang JP Germeyer A Osteen K Taylor RN Lessey BA Giudice LC Global gene profiling in human endometrium during the window of implantation Endocrinology 2002 143 2119 2138 12021176 10.1210/en.143.6.2119
Chen GT Getsios S MacCalman CD Cadherin-11 is a hormonally regulated cellular marker of decidualization in human endometrial stromal cells Mol Reprod Dev 1999 52 158 165 9890746 10.1002/(SICI)1098-2795(199902)52:2<158::AID-MRD6>3.0.CO;2-3
Tsai SJ Wu MH Chen HM Chuang PC Wing LY Fibroblast growth factor-9 is an endometrial stromal growth factor Endocrinology 2002 143 2715 2721 12072406 10.1210/en.143.7.2715
Mock P Frydman R Bellet D Diawara DA Lavaissiere L Troalen F Bidart JM Pro-EPIL forms are present in amniotic fluid and maternal serum during normal pregnancy J Clin Endocrinol Metab 1999 84 2253 2256 10372742 10.1210/jc.84.6.2253
Morrish DW Linetsky E Bhardwaj D Li H Dakour J Marsh RG Paterson MC Godbout R Identification by subtractive hybridization of a spectrum of novel and unexpected genes associated with in vitro differentiation of human cytotrophoblast cells Placenta 1996 17 431 441 8899872 10.1016/S0143-4004(96)90025-9
|
16083501
|
PMC1208945
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Aug 5; 4:28
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-28
|
oa_comm
|
==== Front
Mol CancerMolecular Cancer1476-4598BioMed Central London 1476-4598-4-291608350710.1186/1476-4598-4-29ReviewArtificial neural networks for diagnosis and survival prediction in colon cancer Ahmed Farid E [email protected] Department of Radiation Oncology, Leo W Jenkins Cancer Center, The Brody School of Medicine at East Carolina University, Greenville, NC 27858, USA2005 6 8 2005 4 29 29 19 2 2005 6 8 2005 Copyright © 2005 Ahmed; licensee BioMed Central Ltd.2005Ahmed; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ANNs are nonlinear regression computational devices that have been used for over 45 years in classification and survival prediction in several biomedical systems, including colon cancer. Described in this article is the theory behind the three-layer free forward artificial neural networks with backpropagation error, which is widely used in biomedical fields, and a methodological approach to its application for cancer research, as exemplified by colon cancer. Review of the literature shows that applications of these networks have improved the accuracy of colon cancer classification and survival prediction when compared to other statistical or clinicopathological methods. Accuracy, however, must be exercised when designing, using and publishing biomedical results employing machine-learning devices such as ANNs in worldwide literature in order to enhance confidence in the quality and reliability of reported data.
ANNbackpropagationnodesperceptronperformancetrainingweights
==== Body
1 Introduction and Development of Artificial Neural Networks
Artificial neural networks (ANNs) are regression devices containing layers of computing nodes (crudely analogous to the mammalian biological neurons) with remarkable information processing characteristics. They are able to detect nonlinearities that are not explicitly formulated as inputs, making them capable of learning and adaptability. They possess high parallelism, robustness, generalization and noise tolerance, which make them capable of clustering, function approximation, forecasting and association, and performing massively parallel multifactorial analyses for modeling complex patterns, where there is little a priori knowledge [1]. Artificial neural models possessing such characteristics are desirable because: (a) nonlinearity allows better fit to the data, (b) noise-insensitivity leads to accurate prediction in the presence of uncertain data and measurement errors, (c) high parallelism implies fast processing and hardware failure-tolerance, (d) learning and adaptability permits the system to update and/or modify its internal structure in response to changing environment, and (e) generalization enables application of the model to unlearned data [2].
In the early 1940s, McCulloch and Pitts [3] explored the competitive abilities of networks made up of theoretical mathematical models when applied to the operation of simple artificial neurons. When these early neurons were combined, it was possible to construct networks capable of computing any of the finite basic Boolean logical functions, including symbolic logic. The system comprised of an artificial neuron and input (stimuli) was referred to as "the Perceptron", which established a mapping between input activity and output signal. The next important milestone was the development of the first trainable network perceptron by Rosenblatt, 1959 [4] and Widrow & Hoff, 1960 [5], initially as a linear model having two layers of neurons or nodes (an input and an output layer) and a single layer of interconnections with variables (weights) that were adjustable during training. Some models increased their computational capabilities by adding additional optical filters and layers with fixed random weights, or other layers with unchanging weights. However, these single layers of trainable weights were limited to only solving linear problems. By 1974, Werbos [6] expanded the network to have nonlinear capabilities, modeling with two layers of weights that were trainable in a general fashion, and that accomplished nonlinear discrimination and functional approximation. These original algorithms were named "back-error propagation, BP" and the networks called multilayer perceptrons (MLPs). In BP, the network error (i.e., difference between the predicted and true outcome) constitutes two steps: forward activation to produce a solution, and a backward propagation of the computed error to modify the weights (usually carried out through fitting the weights of the model by a certain function, such as squared error or maximum likelihood, using a gradient optimization method) [Figure 1]. Rumelhart and McClelland popularized ANNs in 1986 [7], and a variety of ANN paradigms have been developed over the last 46 years [2]. In fact, over 50 different ANN types exist. Some applications may be solved using different ANN types, whereas others may only be solved by a specific ANN type. Some networks are capable of solving perceptual problems, while others are more tailored for data modeling and functional approximation [8]. Within cancer research alone, ANNs have been applied to image processing, outcome prediction, treatment-response forecasting, diagnosis and staging [1] [Figure 2].
Figure 1 A three fully interconnected feedforward BP neural network (FFNN), with a single hidden layer. From reference 2, with permission.
Figure 2 Application of ANNs to problem solving: (a) pattern classification (i.e., assigning an unknown input pattern to any of prespecified classes based on properties that are characteristic to a given class); (b) clustering (i.e., clusters or classes are formed by exploring the similarities or dissimilarities between the input patterns based on their inter-correlations); (c) functional approximation or modeling (i.e., training an ANN on input-output data to approximate the underlying rule relating the inputs to outputs); (d) forecasting or predicting (i.e., training an ANN on samples for a time series [t(1) to t(n)] representing a certain phenomenon at a given scenario and then using it for other scenarios to predict the behavior at a subsequent time [t(n + 1)], and (e) association (i.e., developing a pattern by training an ANN to construct the corrupted or missing data).
After demonstrating the utility of ANNs to various applied problems, mathematicians established a theoretical basis for the conceptual capabilities of the MLPs. They showed by a general function approximation theorem that, with appropriate internal parameters (or weights), a neural network could approximate an arbitrary nonlinear function [2]. Thus, ANNs should not be viewed as "black boxes", but as tools that are capable of learning and outcome prediction. Due to the fact that classification tasks, prediction issues and decision support problems are considered functional approximation problems, then ANNs could be applied to problem solving in various domains, and a major research effort has been dedicated to ways of adjusting weights to the best-fitting functional approximations and training parameters [8].
When an ANN is trained on a set of data, it builds a predictive model that reflects a minimization in error when the network's prediction (its output) is compared with a known or expected outcome. Training, which is analogous to biological learning, is carried by a "teacher" program that loads in training cases from a database and adjusts the weights and thresholds value of the network to minimize the error between the real-world outputs and the network generated outputs for the training case inputs. The network would then be validated with available data, and performance measurements [e.g., the mean squared error (MSE), the full range of sensitivity and specificity values (i.e., receiver operating characteristic, ROC, plot associated with the continuous variable output, 0 to 1), and confidence and prediction intervals] can ascertain the network's level of success in arriving at a meaningful prediction unique to each input. Traditionally in medicine, expert opinions have been developed from clinicians' experience and search of the literature. Today, however, ANNs and multivariate analysis can be used to analyze the multitude of data simultaneously and to learn tends in population, thus expanding the "localized" knowledge to a more "global" knowledge, which can be accessed by other practitioners [8].
2 Theory and Performance Measures Behind the Feedforward Artificial Neural Netwoks (FFANN)
The feedforward BP MLP can be viewed basically as a set of equations that are linked together through shared variables in a formation diagramed as a set of interconnected nodes in a network capable of general functional approximation that provides learning capabilities [9]. Variables for inclusion in the final network architecture are usually chosen by a sensitivity analysis method, which tests each input variable by dropping it from the input list and determining the resulting loss of predictive accuracy. Only variables that result in a significant loss of accuracy when dropped are retained in the final network's architecture. Classification tasks like tumor staging, diagnosis, or predicting survival can be performed by FFANNs [10].
FFANN is typically organized as a set of interconnected layers of artificial intermediate (hidden) nodes depicted as a row or collection of nodes, each receiving input from other nodes, connected together to form the network (Figure 1). The MLP has an associated output activation level known as a "squashing" or "activation" function; the most popular is the sigmoid function [f(x)] expressed as:
f(x) = 1/[1 + exp(-x)] ..................................... (1), where x is the input to the squashing function (Figure 3). This sigmoid function, which may have no biological significance, can then be expressed as Sj (the sum of the products of the incoming activation levels with their associated weights):
Figure 3 The activation, squashing, or sigmoid function f(x).
, where Wji, is the incoming weight for unit i; ai, is activation value of unit i; and n, is the number of units that send connections to unit j.
The majority of biomedical studies utilize three-layer networks (input, intermediate and output), in which layers are fully connected (Figure 1). Each connection has an associated weight (w) that corresponds to synaptic junctions in biological systems. Equation (1) becomes:
, where ai, k represents the activation value of node 1 in layer k, and Wji, k represents the weight associated with the connection from the ith node of the kth layer to the jth node of layer k + 1. In a three layer node ANN, there exist two types of weights, and k = 1 or 2. Whereas a network with too few hidden nodes would be incapable of differentiating complex patterns, a network with too many hidden nodes lead to poor generalization for untrained data (Figure 4) [2].
Figure 4 Effect of hidden layer size on network generalization. From reference 2; with permission.
The most popular approach to finding the optimal number of hidden nodes (HN) is by trial and error. Other statistical methods such as cross validation, bootstrapping or pruning have been used. Livingstone & Manallack's [11] suggested that hidden node can be empirically expressed as: HN = M·O/W ..................................... (4), where M is the number of training examples; O, is the number of outputs; and W, is the number of weights, and HN is usually >3 to ensure good generalization and avoid memorizing the training set. Thus, if there are 240 training cases and a single output, the network should not have more than 80 weights. In a network with 10 inputs, this corresponds to having a single hidden layer with 6 units [12].
In theory, there are some problems for which it may be better to use a network with two hidden layers because the overall number of nodes will be less than it would be in a single-hidden layer net. However, for most biomedical applications there is no substantial practical evidence that more than one hidden layer will add meaningfully to the predictive capabilities of a network. Therefore, for practicality, most medical applications use a single hidden-layer networks [12].
In an untrained ANN, the weights of all interconnections are set to be small random numbers. The ANN is then trained (i.e., presented with a training data set that provides inputs and desired outputs of the network). The weights are continuously adjusted by algorithms such as gradient descent computation et sequa, that seek to find a minimum in the error surface so that the network computes the desired output [13]. The amount of network error (or mean square error, MSE) is expressed as:
, where di, p is the desired output of output unit i for input pattern p; P, is the total number of patterns in the data set; n, is the number of output units, and the sum is taken over all data patterns and all output units. The root mean square (RMS) is the square root of the MSE.
Gradient descent weigh training starts with inputting a data pattern to the network in order to determine the activation values of the input nodes. This is followed by forward propagation, in which the hidden layer updates its activation value followed by updates to the output layer (as depicted in equation 3). Next, the desired (known) outputs are submitted to the network. A calculation is then carried out to assign a value to the amount of error associated with each output node. The formula for this error value (δ) is expressed as:
δj, 3 = (dj - aj, 3) f(x) (Sj, 3) ..................................... (6), where dj, is the desired output for output unit j; aj, 3, is the actual output for output unit j (layer 3); f(x), is the squashing function; and Sj, 3 is the incoming sum for output unit j in equation (2).
After these error values become known, weights (from unit i to j) on the incoming connections to each output neuron can be updated according to the following equation:
ΔWji, k = ηδj, k+1 ai, k ..................................... (7), in which k = 2 during updating the layer of weights on connections that terminate at the output layer (see Figure 5).
As the BP ensues, an error value (δ) is then calculated for each hidden node as follows:
δi, 2 = (∑δj, 3 Wji, 2) f(x) (Si, 2) ..................................... (8).
After the error values are known, weights on the incoming connections to each hidden neuron can then be updated. The updated equation (# 7) is used again, substituting k = 1 for weights on connections that start at the first layer. The derivation of the above equations is based on the gradient descent approach and uses the chain rule and interconnected structure of the network [6,8,13]. A general function approximation theorem has been proven for a three layer MLP, showing that they are capable of approximating any nonlinear function in such a way that creating the functional form and fitting the function are performed at the same time (Figure 5), unlike nonlinear regression in which a fit is forced to a prechosen function, giving the ANN an advantage over traditional statistical multivariate regression techniques [14].
Figure 5 Updating or adjusting the value of the weight along a single connection. From reference 8; with permission.
Figure 6 is a graphical representation of an arbitrary nonlinear function approximation performed by an ANN. The function computes a value y = f(x) for every value of x. The ANN is trained to input the value of x, and to output an approximation of f(x). ANN weights are available, which are capable of approximating any arbitrary nonlinear function [8].
Figure 6 Example of a nonlinear functional approximation configured by ANN weights. (a) Illustration showing a function f(x). (b) A neural network configuration to determine an approximation to f(x), given the input x. Modiefied from reference 8.
It should be noted that there has to be patterns (or predictive factors) present in the training data for the ANN to learn successfully; otherwise the network's performance will be low. To measure how well a single output ANN matches data with known outcome, performance metrics include the MSE and RMS. The area under the ROC, or AUROCC, [in which sensitivity can be blotted as a function of (1 - specificity)] (Figure 7) is an acceptable performance measure to use with a single output classification neural network [15]. AUROCC gives a definitive measure of the classifier's discrimination ability that is not dependent upon the choice of the decision threshold. It is identical to the probability that given a positive case and a negative case, the network output will be higher for the positive case. Algorithms are available for calculating the AUROCC [15]. Although the AUROCC provides a useful measure of discrimination (i.e., how well a prediction model can rank patients), it does not, however, provide much insight into calibration (which refers to the correspondence between predicted and actual probabilities). Calibration curves, which are plots of actual against predicted probabilities, are very useful for visually determining accuracy, and can generally help a physician make better inferences before he provides a predicted probability to a patient under evaluation [16].
Figure 7 ROC curve for validating cases from an ANN. Modified from reference 10.
Other measures of the network's performance include the kappa value and the information given. Unlike the AUROCC, these measures require an output threshold to be chosen. Kappa is the actual improvement in classification rate over the chance rate divided by the maximum possible improvement over the chance rate. A value of 1 indicates perfect classification, and a value of 0 indicates classification at the chance rate [12].
The information gain refers to the decrease in classification uncertainty after having observed the network output. Algorithms are available for finding the output threshold that maximizes the information gain [17]. If the relative costs of different types of misclassification (i.e., the cost of false negative or false positive) are known, then an overall cost measure can be calculated. Alternatively, the network can be tuned to output these values directly [12].
When training an ANN, three non-overlapping sets of data are used: (a) the training set, (b) the validation (or testing) set, and (c) the verification set. The training set is used for adjustment of weights during training, whereas the testing set is used to decide when to stop training; otherwise, the ANN will learn features in the training set that are not present in the wider population of cases, a phenomenon known as "fitting to noise" or "overfitting". The performance measures should be made on both the training and test sets. However, only if the testing set has been used to set the network's weights or evaluate its structure, will it reflect the network's performance on future data; this practice of splitting the data into a training set and a test set is referred to as "cross validation" [18]. Another method for estimating the error rate of a prediction rule is "data splitting" [19]. Both cross validation and data splitting methods are suitable if there is plenty of data available. For small data sets, the "bootstrap" method is used. A drawback of the bootstrap is that a large number of samples (20 ≤ B samples ≤ networks) must be trained [20]. As seen in a classical training curve (Figure 8) where RMS of the training and testing sets are plotted as a function of the number of HNs or training cycles, the RMS on the training set decreases as more training is carried, but the testing set has a minimum when reached the RMS begins to increase. Training beyond the inflection point results in overfitting. It is imperative to not overfit the network during training, which can be achieved by methods such as restricting the topology of the network (i.e., decreasing the number of nodes), or early stopping, or by using weight decay. If computationally possible, one should consider the use of a Bayesian approach that averages over several plausible networks [21].
Figure 8 Criteria for termination of training and selection of an optimum ANN architecture. Modified from references 2 and 8.
The time it takes to train a neural net increases exponentially with the number of network inputs and the number of network nodes, and polynomially with the number of training examples. A network with 200 inputs trained on a few thousand examples takes about four hours to train on a computer. Therefore, it is important to include only those inputs and examples that seem relevant to the task at hand. This is not a significant problem for medical decision-making, as they are generally small. For example, a typical medical net has 20 inputs and is generally trained on ~350 samples [12]. An ANN-based system is considered to have learned if it can: (a) handle imprecise, fuzzy, noisy and probabilistic information without noticeable adverse effect on response quality, and (b) generalize from the tasks it has learned to an unknown ones [2].
It should be remembered that because the ANN has undergone generalized learning, it becomes capable of interpolating or extrapolating results from new incoming data. However, it is important to ascertain that the incoming data do not extend beyond the range of values used in training. Data outside acceptable limits should either not be processed or results flagged and viewed with caution. For data within the range of network training, a measure of confidence can be made by calculation of a confidence or a prediction interval. It should be kept in mind that a trained, tested and verified ANN does not provide a unique solution because its final trained resting state depends on several factors such as number of original randomized starting weights, number of and order of presentation of cases, and the number of testing cycles. Other mathematical alternatives employed during training such as the use of momentum, adjusting the learning constant, "jogging" the weights, etc., may have implications. Therefore, for a particular application such as cancer prediction, a frequency distribution of the network versus the outcome probability can be produced and a central tendency such as mean, mode, measure of variance and nonparametric predictive intervals (in case of skewed nonparametric distributions) are plotted, producing what is called "prevalence-value accuracy" plots. Then it could be stated, for example, that with 90% confidence, the probability of the expected outcome will occur with such a range, having a median value of such a number [22].
Rather than putting faith in "black box" systems, the workings of a neural net set used in survival analysis on censored data could be explained by exploring the interactions between predictive values and survival rates, which leads to useful insights into the roles played by different prognostic variables in determining patient outcomes [23]. In a performance measure approach dubbed a "sensitivity analysis", each input is varied and the corresponding change in output is measured. The ratio of change in output over input (δy/δxi) is then averaged over all samples to produce a sensitivity parameter for each input. The inputs can then be ranked according to sensitivity. Sensitivity analysis, however, could be misleading because y may not be a linear function of x. The sensitivity measures are dependent on the particular example used to train the network, and also on the initial weight setting of the network [12].
Another performance measure approach known as "key factor justification" has been used to explain individual decisions. In that approach, each input to the ANN is reversed. If the output decision is consequently reversed, then a key factor could accordingly be identified. If no single key factor could be ascertained, then pairs of variables, or triples, could be reversed together and the output is observed. However, going beyond triples may require excessive computational capabilities [24].
3 General Application and Improving Performance of ANNs
ANNs have been applied to problem solving in various fields including: (a) pattern classification, (b) clustering (class separation), (c) functional approximation (modeling), (d) forecasting, (e) association (e.g., image completion), and (f) optimization (e.g., finding a solution that minimizes an objective function (see Figure 2) [2].
In the military and electronic arenas, ANN applications include automatic target recognition, control of flying aircrafts, engine combustion optimization, adaptive switching, circuits and fault detection in complex systems [8]. In the financial field, a decision support role for ANNS to predict stock market fluctuations and commodity trading has been envisioned [8]. In the biological domain, ANN application to samples' characterization, identification and interaction include: interpreting pyrolysis mass spectrometery, GC and HPLC data; pattern recognition of DNA, RNA, protein structure and microscopic images; prediction of microbial growth, biomass and shelf-life of food products; and identification of microorganisms and molecules [2]. In the medical and behavioral sciences, image analysis has resulted in systems capable of diagnosis and prognosis of various diseases, (including cancer), classification of cancer subtypes, predicting tumor sensitivity to drugs, identification of potential biomarkers, analysis of gene expression data, medical imaging and radiological diagnosis, analysis of wave forms, outcome prediction, identification of pathological specimens, interpretation of laboratory data, evaluation of epidemiologic data, waveform analysis (including electroencephalography, electromyogram, electrocardiogram and Doppler ultrasound), length of stay in intensive care units following various diseases/surgery, and predicting admission decisions in psychiatric wards [25-30].
The performance of an ANN depends on network parameters, the network weights and the type of transfer functions used. A disadvantage of using FFANNs is that they require the initialization and adjustment of many individual parameters to optimize their classification performance. When optimized manually, these adjustments can take days or even weeks to complete one set of experiments for estimating one outcome on single database [31]. The lengthy process of manually optimizing a feedforward BP ANN provided the incentive to develop an automated system that could fine-tune the network parameters without user supervision. A new stopping criterion (i.e., the logarithmic sensitivity index) was introduced that provided a balance between sensitivity and specificity of the output classification. The network automatically monitored the classification performance to determine when was the best time to stop training after no noticeable improvement in the performance measure (either highest correct classification rate, lowest mean squared error, or highest log-sensitivity index value) occurred in the subsequent 500 epochs. Using these automated ANNs, experiments performed on three medical databases showed that the optimal network parameter settings found by the automated system were similar to those found manually, and that automated networks performed equally well or better than the manually optimized ANNs, and the best classification performance was achieved using the log-sensitivity index as a stopping criterion [31].
When using an ANN, three important issues needs to be addressed that the solution to which will significantly influence the overall performance of the ANN with regard to two considerations: (a) recognition rate to new patterns, and (b) generalization performance to new data sets that have not been presented during network training [32]. These issues are: (i) the selection of data patterns for network training [33], (ii) the selection of an appropriate and efficient training algorithm from a large number of possible training algorithms found in the literature such as BP and its many variants [34] and the second-order algorithms [35], just to name a few. New training algorithms with faster convergence properties and less computational requirements are being developed, and (iii) determination of network size. This is a more difficult problem to solve. It is necessary to find a network structure small enough to meet certain performance specifications. In practice, this is carried by training a number of networks with different sizes, and the smallest network that can fulfill all or most of the required performance requirements is selected. In an attempt to develop a systematic procedure for an automatic determination and/or adaptation of the network architecture to an incremental constructive training scheme, input-side and output-side training could be separated in order to improve the input-side training effectiveness and efficiency, and to obtain better generalization performance capabilities. Two pruning methods for improving the input-side redundant connections were also developed that resulted in smaller networks without degrading or compromising their performance. Moreover, numerical simulations demonstrated the potential and advantages of the proposed data pattern selection/training and size determination strategies when compared to other existing techniques in the literature [32].
4 Applications of ANNs to Colon Cancer Diagnosis
Microarray data are becoming powerful tools in clinical diagnosis, particularly for tumor classification because they simultaneously record gene expression levels of thousands of genes. These data are characterized by high dimensionality because a large number of gene expression input vastly exceeds the number of sampling, which may lead to overfitting. This situation necessitates dimensionality reduction through either using a reduction algorithm, or selecting a small set of genes as input to the classifier in a supervised way [36], or by employing cross validation to avoid overfitting [37].
Both unsupervised clustering methods and supervised methods have been used for classification [38]. I have employed colon cancer as an example to show how supervised ANNs have an advantage over clustering methods (which were shown to be incapable of detecting subtle differences between biological classes) in classification if some prior knowledge of the classes is available.
There is an important subtle distinction between sporadic colon adenomas and cancers (SACs) and inflammatory bowl disease-related dysplasia or cancer (IBDNs) because SACs can be managed by polypectomy alone, whereas IBDNs require a life-threatening subtotal colectomy. A microarray study was conducted to evaluate the ability of ANN and hierarchical cluster analysis to discriminate between these types of cancer based on hybridizing 8064 cDNA clones to mRNAs derived from 39 colon neoplastic specimens [1]. GeneFinder software was used to select 1192 clones that showed significantly different mean square expression levels between IBDNs and SACs (P = 0.001). A BP FFNN, with two hidden layers and 1192 inputs (representing the selected genes) was constructed, and the output was set at 0 for IBDNs and 1 for SACs using the software program MatLab (Math Works, Inc., Nattick, MA). The ANN was learned using a training set of 5 IBDNs and 22 SACs. The test set comprised the remaining data samples consisting of 3 IBDNs and 9 SACs. ANN approximations were evaluated using regression analysis that compared expected output (Target) with ANN output following training, and unpaired 2-sided Student t-test was also used to evaluate the statistical differences between the net defined IBDNs versus SACs (i.e, 0 vs. 1). Hierarchial clustering was performed using the program Cluster (Stanford University, Palo Alto, CA). Whereas the network correctly diagnosed 12 of 12-blinded samples, hierarchial analysis failed, probably because of noise in the database. Only by using an iterative process to reduce the number of clones used for diagnosis to 97, could cluster analysis separate the two types of lesions. Even with this reduced clone set, ANN still retained its capacity for correct diagnosis of the two types of colon cancer [1].
Another microarray study employed a combination selection method in conjunction with ensemble neural network to analyze cancer data, including that of the colon. The principle of the method was based on the assumption that combining various feature selection mechanisms to chose top-ranked genes will avail more information, and by using an ensemble combining the output of several ANNs into an aggregate output, features can be analyzed more effectively due to the stability of the networks and robustness of the answers [39]. The authors employed the public database of Alon et al [40] containing 62 samples (40 colon tumors and 22 normal tissue samples). They chose 2,000, out of ~6,500 expressed genes, based on their confidence in the measured expression level to assemble networks consisting of 100 members. No fresh samples were available for testing the network ensemble. Nevertheless, using this ensemble, the predictive accuracy of adopting leave-one-out cross validation (LOOCV) and 10-fold cross validation was 91.94% and 90.32%, respectively, as compared to 85.48% obtained by using various boosting algorithms in combination with LOOCV. However, a drawback of the ANNs ensemble approach is the increased computational complexity and the additional time needed to perform the analysis [39].
5 Application of FFNN to Predicting Survival in Colon Cancer
It is currently difficult to predict when and if a particular patient will die after surgical and adjuvant chemotherapeutic treatment of colon cancer, especially at the intermediate Dukes; B and C stages, using available techniques based on histopathological TNM staging and employing univariate and multivariate regression analysis [41].
A 5-year follow-up data from 334 patients treated for colorectal cancer (CRC) were used to train 284 patients and validate 50 patients using 6 FFNN with BP, containing from 2 to 15 hidden units designed to predict death within 9, 12, 15, 18, 21 and 24 months using the logistic activation function with continuous output on the interval 0, 1. Furthermore, the trained 12-months ANN was then applied to 2-years follow up on patients from a second institution. The network predictions of which individual patients would die within 12 months were also compared with those of two consulting surgeons [42]. Results showed that all 6 ANNs were able to achieve an overall predictive accuracy of death at 95% CI ≥ 80% at the first institution, with a mean sensitivity and specificity of 60% and 88%, respectively. Furthermore, the trained 12-months ANN achieved an overall predictive accuracy for death of 90% (95% CI 84–96) when applied to death from the second institution, compared with an overall accuracy of 79% (71 – 87) and 75% (66 – 84) for CRC surgeons. Thus, ANNs predicted outcome for CRC death more accurately than clinicopathological methods. Moreover, once trained in one institution, ANNs were able to accurately predict outcome for patients from an unrelated institution [42].
In another study to predict a 5-year survival after primary treatment of colon carcinoma in the National Cancer Data Base (NCDB), UK, 37,500 cases treated between the years 1985 and 1993, and not used in model development, were analyzed by an ANN model and compared with a standard Cox parametric logistic regression [10]. A FFNN with two hidden layers that contained 4 and 3 hidden neurons, respectively, and one output layer was selected. Eleven input variables were chosen by a sensitivity analysis method (including race; sex; age; tumor location, size, behavior; histopathology; surgery, chemo or radiation therapy, hormonal or other cancer-directed therapy) and only the variables that resulted in significant loss of accuracy when dropped were retained in the final network architecture, Training of the network was accomplished by using a standard second order conjugate gradient descent method. A validation set representing 25% of randomly chosen data was employed for validation. The area under the ROC curve was used to measure the overall predictive accuracy of the network. The ANN yielded a ROC area of 87.6%. At sensitivity to mortality of 95%, the specificity was 41%. The logistic regression yielded a ROC area of 82%, and sensitivity to mortality of 95% gave a specificity of only 27%. Thus, the ANN found a strong pattern in the database predictive of 5-year survival status, whereas the logistic regression produced somewhat less accurate, but good results [10]. In another study by the same group of investigators [43] aiming at predicting 5-year survival associated with CRC using the same ANN and Cox regression and ROC to compare data, the logistic regression model gave a result of 66% and the ANN gave 78%, indicating that the neural network approach was more superior compared to regression analysis in predicting colon cancer survival.
A fourth study compared ANNs to TNM staging to predict 5-year survival of patients with CRC, using the area under the ROC as a measure of accuracy. Variables for patient care evaluation (PCE) database used for analysis included: age, race, gender, signs and symptoms (e.g., changes in bowel habits, obstruction, jaundice, occult blood, and others), diagnostic and extent-of-disease tests (e.g., endoscopy, radiography, barium enema, colonoscopy, CT, biopsy, CEA antigen, X-ray, liver function tests and others), and histoipathological parameters. A test set of 5,007 training cases, and a validating set of 3,005 cases was used. A FFNN BP composed of an input, a hidden and an output layer was used. The ANNs prediction of 5-year survival was significantly more accurate than the TNM staging (ANN 0.815 versus TNM 0.737, p < 0.001). Adding commonly collected demographic and anatomic variable to the TNM variables further increased the accuracy of the ANN (0.869). Thus, the ANNs were significantly more accurate than the TNM staging system when both used the TNM prognostic factors alone, and prognostic factors added to ANN further increased the predictive prognostic accuracy [44].
6 Conclusion
There are advantages and disadvantages to FFANNs when applied to biomedical decision-making. Advantages include: (a) requirement for less formal statistical training to develop, (b) having a better discriminating power than other regression models, (c) can be developed using multiple different training algorithms, (d) their parallel nature enable them to accept a certain amount of inaccurate data without a serious effect on predictive accuracy (i.e., graceful degradation), (e) having the ability to accurately detect complex nonlinear relationships between independent and dependent variables, and all possible interactions between variables, as they make no assumptions about those variables, (f) reduce the number of false positives without significantly increasing the number of false negatives, and (g) they may allow for individual case prediction. On the other hand, disadvantages include: (a) considered as "black box" methods, one cannot exactly understand what interactions are being modeled in their hidden layers as compared to "white box" statistical models, (b) have limited abilities to identify possible causal relationships, (c) model development is empirical; thus, providing low decision insight, and many methodological issues remain to be solved, (d) models prone to overfitting, (e) require lengthy development and time to optimize, (f) they are more difficult to use in the field because of computational requirements, and (g) there is conflicting evidence as to whether or not they are better than traditional regression statistical models for either data classification, or for predicting outcome [21,45,46].
Despite their theoretical advantages, ANNs do not universally outperform standard regression techniques for several reasons: (a) because from a practical point of view, only a limited amount of data that may be related to the outcome of interest can be collected, and these data are mostly based on studies in which a standard regression model was used, and therefore only factors that were significant in a regression models are collected in subsequent studies. Therefore, nonlinear functions, or those that involve interaction with other variables may not have emerged as "significant" in the regression analysis and therefore are not reflected in the literature as important prognostic factors, (b) all variables and outcomes are measured with error(s). A nonlinear relation when measured with an error may well be adequately represented by a linear model, (c) there exist data barriers beyond which mathematical models are unable to make predictions in biological systems, and (d) regression models are superior to ANNs when drawing inferences and interpretations based on outputs [47]. In addition to insight into the disease process, regression models provide explicit information regarding the relative importance of each independent variable. This information can be valuable in planning subsequent interventions, in eliminating unnecessary tests or procedures that are unrelated to the outcome of interest, and in determining which are the most critical data to store in the database [47].
Although the representation of a complex risk structure by nonlinear machine-learning methods such as ANN or classification and regression tree (CART) could provide as insight into the underlying nature of a disease, ease of interpretation is not a typical feature of the network representation of a complex relationship. However, a suitable network approach could outperform other approaches, provided that the underlying disease has sufficient complex interactions because the ability to represent arbitrary relationships is a well known property of neural networks. However, one of the main problems in using ANNs to provide support for therapy decisions is the need for a high level of trust in the predictions of such a model on the part of both the physician and the patient under examination. This need requires that a good generalization capability must be convincingly demonstrated. In a clinical context with a small data set, the key to good generalization lies in optimized complexity reduction techniques. Thus, improvement in these techniques will play an important role in increasing confidence in the application of ANNs to the clinical setting [48].
From earlier analysis on colon cancer, it is evident that FFANN enhanced diagnosis and prognosis when compared to other statistical methods, and increased survival prediction when compared to logistic regression or clinicopathological staging. However, the uncritical use of ANNs for prognostic and diagnostic classification of cancer, including colon cancer, can lead to the following mistakes: (1) the reported error rates for some ANNs may underestimate the true misclassification probabilities; for example, by not showing the cross validation error rates in the learning, validation and/or test sets, and in some cases by having a too small size of the test set; (2) fitting of biologically implausible functions to describe the probability of class membership when overfitting occurs, as overfitting generally occurs if the ratio between the number of observations and the number of parameters is smaller than two; (3) incorrectly and/or failure to report or describe the complexity of the network (i.e., number of parameters, the number of hidden layers and hidden units to calculate the number of fitted weights, etc.) will not allow the reader to judge the magnitude of overfitting, (4) use of inadequate statistical competitors or statistical methods to compare the performance of the networks. A fair comparison of the performance of FFNNs and statistical methods must be based on tools of similar flexibility like nearest-neighbor methods, generalized additive models, CART or logistic regression models with quadratic terms and multiplicative interaction terms, which is not usually carried out; (5) inefficient comparison with statistical methods without proving the significance of the differences between the observed misclassification rates, and (6) naive application of ANNs to survival data such as omitting censored cases (which lead to bias), and using the number of the time interval as an additional input unit, which causes the estimated survival probabilities not to depend on the length of the time intervals [45]. Avoiding the above mistakes when reporting the results of ANNs is a good science, as this will improve the confidence in the reliability of data reported in the scientific literature by using an unsupervised method such as ANN for data analysis, whose use has nevertheless been steadily on the rise?
Acknowledgements
I wish to express my gratitude to my colleague Dr. Paul Vos of The Department of Biostatistics, School of Allied Health, East Carolina University for his comments and insightful discussions.
==== Refs
Selaru FM Xu Y Yin J Zou T Liu TC Mori Y Abraham JM Sato F Wang S Twigg C Olaru A Shustova V Leytin A Hytiroglou P Shibata D Harpaz N Meltzer SJ Artificial neural networks distinguish among subtypes of neoplastic colorectal lesions Gastroenterol 2002 122 606 613 10.1053/gast.2002.31904
Basheer IA Hajmeer M Artificial neural networks: fundamentals, computing, design and application J Microbiol Meth 2000 43 3 31 10.1016/S0167-7012(00)00201-3
McCulloch WS Pitts W A logical calculus of the ideas imminent in nervous activity Bull Math Biphys 1943 5 115 133
Rosenblatt F The perceptron: a probalistic method for information storage in the brain Psych Rev 1959 65 386 407
Widrow B Hoff M August IRE WESCON Convention Record 1960 Part 4 96 104
Werbos PJ Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences PhD Thesis 1974 Harvard University, Cambridge, MA
Rumelhart DE McClelland JL Parallel Distributed Processing 1986 1 and 2 MIT Press, Cambridge, MA
Dayhoff JE DeLeo JM Artificial neural networks: opening the black box Cancer 2001 91 1615 1635 11309760 10.1002/1097-0142(20010415)91:8+<1615::AID-CNCR1175>3.0.CO;2-L
Hornik K Stinchcomb X White X Multilayer feed forward networks are universal approximations Neural Net 1993 2 359 366 10.1016/0893-6080(89)90020-8
Snow PB Kerr DJ Brandt JM Rodvold DM Neural network and regression predictors of 5-year survival after colon carcinoma treatment Cancer 2001 91 1673 1678 11309767 10.1002/1097-0142(20010415)91:8+<1673::AID-CNCR1182>3.0.CO;2-T
Livingstone DJ Manallack DT Statistics using neural networks: chance effects J Med Chem 1993 36 1295 1297 8487267 10.1021/jm00061a023
Penny W Frost D Neural networks in clinical medicine Med Decis Making 1996 16 386 398 8912300
Mehrotra K Mohan CK Ranka S Elements of Artificial Neural Networks 1997 MIT Press, Cambridge, MA
Hornik K Some new results on networks approximation Neural Net 1989 6 1969 1972
Hanley JA McNeil BJ The meaning and use of the area under a receiver operating characteristic (ROC) curve Radiology 1982 143 29 36 7063747
Kattan M Statistical prediction models, artificial neural networks and the sophism "I am a patient, not a statistics" J Clin Oncol 2002 20 885 887 11844805 10.1200/JCO.20.3.791
Somoza E Mossman D Comparing and optimizing diagnostic tests: an information-theoretical approach Med Decis Making 1992 12 179 188 1513208
Efron B Estimating the error rate of a prediction rule: improvement on cross-validation J Am Stat Assoc 1983 78 316 331
Picard RR Berk KN Data splitting Am Stat 1990 44 140 147
Efron B Tibshirani R An Introduction to Bootstrap 1993 Chapman and Hall, New York
Dreiseitl S Ohno-Machado L Logistic regression and artificial neural network classification models: a methodology review J Biomed Inform 2002 35 352 359 12968784 10.1016/S1532-0464(03)00034-0
Remaley AT Sampson ML DeLeo JM Remaley NA Farsi BD Zweig MH Prevelence-value-accuracy plots: a new method for comparing diagnostic tests based on misclassification costs Clin Chem 1999 45 934 941 10388467
De Laurentiis M Ravdin PM A technique for using neural network analysis to perform survival analysis of censored data Cancer Lett 1994 77 127 138 8168059 10.1016/0304-3835(94)90095-7
Gallant S Neural Network Learning and Expert Systems 1994 MIT Press, Cambridge, MA
Walker CR Frize M Artificial neural networks "ready to use" for decision making in the neonatal intensive care unit Pediat Res 2004 56 6 8 15128926 10.1203/01.PDR.0000129654.02381.B9
Astion ML Wilding P Application of neural networks to the interpretation of laboratory data in cancer diagnosis Clin Chem 1992 38 34 38 1733603
Baset WG Application of artificial neural networks to clinical medicine Lancet 1995 346 1135 1138 7475607 10.1016/S0140-6736(95)91804-3
Fu LM Fu-Liu CS Multi-class cancer subtype classification based on gene expression signatures with reliability analysis FEBS Lett 2004 561 186 190 15013775 10.1016/S0014-5793(04)00175-9
Le Blanc M Kooperberg C Grogan TM Miller TP Directed indices for exploring gene expression data Bioinformatics 2003 19 686 693 12691980 10.1093/bioinformatics/btg079
Ball G Mian S Holding F Allibone RO Lowe J Al S Li G McCardle S Ellis ID Creaser C Rees RC An integrated approach utilizing artificial neural network and SELDI mass spectrometry for the classification of human tumors and rapid identification of potential biomarkers Bioinformatics 2002 18 395 404 11934738 10.1093/bioinformatics/18.3.395
Ennett CM Frize M Charette E Improvement and automation of artificial neural networks to estimate medical outcome Med Engineer Phys 2004 26 321 328 10.1016/j.medengphy.2003.09.005
Ma L Khorasani K New training strategy for constructive neural networks with application to regression problems Neural Networks 2004 17 589 609 15109686 10.1016/j.neunet.2004.02.002
Tetko IV Efficient partition of learning data sets for neural network training Neural Networks 1997 10 1361 1374 12662480 10.1016/S0893-6080(97)00005-1
Stager F Agrawal M Three methods to speed up the training of feedforward and feedback perceptron Neural Networks 1997 10 1435 1443 12662484 10.1016/S0893-6080(97)00053-1
Shepherd AJ Second-order Methods for Neural Networks 1997 Springer, London
Golub TR Slonin DK Tamayo P Huard C Gassenbeck M Mesirov JP Coller H Loh ML Downing JR Caligiuri MA Bloomfield CD Lander ES Molecular classification of cancer class discovery and class prediction by gene comparison monitoring Science 1999 286 531 537 10521349 10.1126/science.286.5439.531
Tibshirani R Hastie T Narasimhan B Chu G Diagnosis of multiple cancer types by shrunken centroids of gene expression Proc Natl Acad Sci USA 2002 99 6567 6577 12011421 10.1073/pnas.082099299
Ringnér M Peterson C Microarray-based cancer diagnosis with artificial networks BioTechniques 2003 39 530 535
Liu B Cui Q Jiang T Ma S A combinational feature selection and ensemble neural network method for classification of gene expression data BMC Bioinf 2004 51 131
Alon U Barkai N Notterman DA Gissh K Ybarra S Mack D Levine AJ Broad pattern of gene expression revealed by clustering analysis of tumor and normal colon tissue provided by oligonucleotide arrays Proc Natl Acad Sci USA 1999 96 6745 6750 10359783 10.1073/pnas.96.12.6745
Newland RC Dent OF Lyttle MNB Chapuis DS Bokey EL Pathological determination of survival associated with colorectal cancer with lymph node metastasis Cancer 1994 73 2076 2082 8156513
Bottaci L Drew PJ Huntley JE Hadfield MB Farouk R Lee PWR Macintyre IMC Duthie GS Monson JRT Artificial neural networks applied to outcome prediction for colorectal cancer patients in separate institutions Lancet 1997 350 469 472 9274582 10.1016/S0140-6736(96)11196-X
Grumett S Snow P Kerr D Neural networks in the prediction of survival in patients with colorectal cancer Clin Colorect Cancer 2003 2 239 244
Burke HB Goodman PH Rosen DB Henson DE Weinstein JN Artificial neural networks improve the accuracy of cancer survival prediction Cancer 2001 91 857 862.29
Schwarzer G Vach W Schumacher M On the misuses of artificial neural networks for prognostic and diagnostic classification in oncology Stat Med 2000 19 541 561 10694735 10.1002/(SICI)1097-0258(20000229)19:4<541::AID-SIM355>3.0.CO;2-V
Kattan MW Comparison of Cox regression with other methods for determining predictive models and normograms J Urol 2003 170 S6 S10 14610404 10.1097/01.ju.0000094764.56269.2d
Sargent DJ Comparison of artificial networks with other statistical approaches Cancer 2001 91 1636 1942 11309761 10.1002/1097-0142(20010415)91:8+<1636::AID-CNCR1176>3.0.CO;2-D
Kates RE Berger U Ulm K Harbeck N Graeff H Schmitt M Jain LL Performance of neural nets, CART, and Cox models for censored survival data Proceeding 3rd International Conference on Knowledge-Based Intelligent Information Engineering Systems Adelaide, Australia, 31st August to 1st September 1999
|
16083507
|
PMC1208946
|
CC BY
|
2021-01-04 16:36:35
|
no
|
Mol Cancer. 2005 Aug 6; 4:29
|
utf-8
|
Mol Cancer
| 2,005 |
10.1186/1476-4598-4-29
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-221610216910.1186/1744-8069-1-22ResearchGenetic alteration of anxiety and stress-like behavior in mice lacking CaMKIV Shum Fanny WF [email protected] Shanelle W [email protected] Yong-Seok [email protected] Bong-Kiun [email protected] Min [email protected] Department of Physiology, Faculty of Medicine, University of Toronto, Toronto, Ontario, M5S 1A8, Canada2 Department of Biological Sciences, College of Natural Sciences, Seoul National University, Seoul 151–742, South Korea2005 15 8 2005 1 22 22 1 6 2005 15 8 2005 Copyright © 2005 Shum et al; licensee BioMed Central Ltd.2005Shum et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Calcium-calmodulin-dependent protein kinase IV (CaMKIV) phosphorylates the major transcription factor cyclic AMP-response element binding protein (CREB), which plays a role in emotional behavior. Here, CaMKIV knockout mice (CaMKIV-/-) were tested in a battery of stress and anxiety-related behavioral tests, to determine if CaMKIV plays a role in emotional behavior. CaMKIV-/-exhibited a decrease in anxiety-like behavior in both the elevated plus maze and dark-light emergence tests when compared to wild-type mice. Both the acoustic startle response and prepulse inhibition of startle were decreased with the deletion of CaMKIV. In addition, CaMKIV-/- mice displayed a lack of stress-induced analgesia following restraint or cold swim stress. Our results demonstrate a key role for CaMKIV in anxiety and stress-related behavior.
anxietyCaMKIVknockout micemicroarraystress-induced analgesia
==== Body
Introduction
Calcium-calmodulin-dependent protein kinase IV (CaMKIV) plays a role in the activity-dependent phosphorylation of cyclic AMP-responsive element binding protein (CREB) and CRE modulator (CREM), which regulate the expression of genes involved in neuroplasticity [1], learning and memory [2-4], emotional behavior [5-7] and molecular changes induced by antidepressants [8]. Several protein kinase cascades regulate CREB function in the CNS [1,9], these include the cAMP signaling pathway and the Ca2+-calmodulin dependent protein kinase pathway. Among different Ca2+-dependent protein kinases, CaMKIV is detected predominantly in the nuclei of neurons [10,11], therefore CaMKIV may play a unique role in the phosphorylation of CREB and in the regulation of neuronal gene expression [12].
CaMKIV is normally expressed in the amygdala and hippocampus, two brain structures involved in the regulation of anxiety and CaMKIV deficient mice exhibit defects in contextual and auditory fear memory [13]. A recent study reported that the CaMKIV signaling pathway may play a role in the excitation-mediated regulation of neuropeptides involved in the pathophysiology of anxiety in vitro [14]. However, molecular and physiological roles of CaMKIV in emotional behavior have yet to be investigated.
Previous studies have implicated that both CREB [5] and CREM [6] are activated by CaMKIV and both have been shown to play a role in emotional behavior. In the present study we wanted to determine if the deletion of CaMKIV would result in changes in anxiety and stress-related behaviors. Here we report that CaMKIV-/- mice exhibit decreased anxiety-like behaviors in several anxiety paradigms and develop less stress-induced analgesia. Our results demonstrate a key role for CaMKIV in mediating changes in anxiety and stress-related behaviors.
Materials and methods
Subjects
All subjects were 8–12 weeks old male mice. The CaMKIV-/- mice were derived as described [15] and bred for several generations on the C57BL/6 background (F12–F16). While wild-type littermates were used in some experiments, C57BL/6 mice were purchased from Charles River to use as controls in others. We do not feel that this represents a problem with the genetics background since the CaMKIV transgenic line can be considered congenic with C57BL/6 and preliminary results showed that there was no difference in the behavior of wild-type littermates from C57BL/6 mice. Mice were housed on a 12-h light-dark schedule with food and water available ad libitum. All experiments were carried out in accordance with the rules and regulations of the Animal Care and Use Committee at the University of Toronto. All efforts were made to minimize the animal's suffering and to reduce the number of animals used. No visual difference between C57BL/6 and CaMKIV-/- mice was noticeable and experiments were performed blind when possible.
Elevated plus-maze test
The elevated plus maze (Med Associates, St Albans, Vt) consists of two open arms and two closed arms situated opposite each other and separated by a 6 cm square center platform. Each runway is 6 cm wide and 35 cm long. The open arms have lips that are 0.5 cm high and the closed arms are surrounded on three sides by 20 cm walls. The floors and walls are black polypropylene. For each test, the animal was placed in the center square and allowed to move freely for five minutes. Open arm entries were defined as the mouse having all four paws onto the open arm. The number of entries and time spent in each arm was recorded.
Light/Dark box
The testing apparatus consisted of a rectangular Plexiglas box (44 × 8.5 × 25 cm) equally divided into a light, open topped, compartment connected by a door (17 cm in height) to a dark, closed topped, opaque compartment. Each mouse was placed in the light box and was allowed 10 sec to explore before the door to the dark box was opened. Each animal was tested for 10 min. The time spent in the light compartment and the number of light/dark transitions were recorded.
Dark-light emergence Task
The apparatus consisted of a black opaque plastic box (13 × 11 × 8.5 cm) with a small opening (3 × 6 cm) placed along one side of the open field evenly illuminated by white overhead lighting. The exit was faced out into the open field. Mice were individually placed into the box with the exit blocked for a 5 min habituation period. At the end of the habituation period, the exit was opened and the mice were allowed to freely explore the open field for 5 min. The latency to emerge from the box, time out of the box, time spent assessing the open field (scanning the open field with the head but less than four paws out of the box) and box/open field transitions were recorded.
Acoustic startle and prepulse inhibition (PPI)
Testing was conducted in a startle chamber from Med Associates (Med Associates Inc., St. Albans, VT). The startle chamber consists of a nonrestrictive Plexiglas cylinder (4.4 cm) in diameter, resting on a platform in a ventilated and sound attenuating chamber. A high-frequency speaker mounted behind the cylinder produced all the acoustic and prepulse stimuli. Mouse movements were detected and transduced by a stablimeter mounted under the cylinder and digitized and stored by a computer and interface assembly. Startling stimulus began at 80 dB and 1 ms readings were recorded to obtain the amplitude of the animal's startle or prepulse response to each stimulus. For acoustic startle, mice were acclimated in the chamber for 20 min and 80 trials of 80, 85, 90, 95, 100, 110 and 120 dB noise bursts were presented over a 45-min test session. The inter-trial interval varied randomly from 10 to 20 s, with an average of 15 s. For PPI, each test session consisted of 55 trials and was initiated with a 10-min acclimation period. Testing began after the initial exploratory behavior had diminished. Three different trial types were presented at random: 20 ms prepulse of 80, 90 or 100 dB, 100 ms before a startle noise burst (112 dB), startle noise burst (112 dB) alone and background noise alone. For both acoustic startle and PPI the background noise was at 70 dB.
Stress induced nociceptive tests
To induce stress, mice were forced to swim in water (10°C) for 3 min or were restrained for 30 min. Animals were individually restrained in small cylindrical tubes with a diameter slightly larger than a mouse's body. Responses to the hot-plate and tail-flick were measured at different time points up to 60 min after stress. Data are presented as the mean response latency (s) or maximum possible inhibition (MPI = (response latency – baseline response latency)/ (cut-off time – baseline response latency) × 100). The total effect of stress over time, the area under the curve (MPI versus time) was used.
GeneChip analysis
Total RNA was isolated from the forebrains of two C57BL/6 or CaMKIV-/- mice using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and purified with RNeasy mini kit (Qiagen). RNA quality was checked visually on an agarose gel and also by Agilent BioAnalyzer. Combined RNA was subjected to gene chip analysis (Affymetrix Mouse 430 2.0). First-strand synthesis, fragmentation, hybridization and washing were done according to GeneChip (Affymetrix Inc.) protocol. The array was scanned, and information was extracted using the GeneChip Expression Analysis program (Affymetrix) and analyzed using the GeneChip Operating System (GCOS, Affymetrix).
Reverse transcription-PCR analysis
First strand-cDNAs were synthesized from total RNA isolated from the forebrains of C57BL/6 and CaMKIV-/- mice previously extracted for the GeneChip experiment (n = 2 for C57BL/6 and CaMKIV-/- mice) using SuperScript III RT (Invitrogen) with oligo(dT) as a primer. The primers for RT-PCR were as follows: Oxytocin: sense, 5'-TTGCTGCCTGCTTGGCTTAC-3', antisense, 5'-TATTCCCAGAAAGTGGGCTC-3', arginine vasopressin: sense, 5'- ACACTACGCTCTCCGCTTGT-3', antisense, 5'- GGGCAGGTAGTTCTCCTCCT-3', Transthyretin: sense, 5'- ATGGTCAAAGTCCTGGATGC-3', antisense, 5'- CAGAGTCGTTGGCTGTGAAA-3', connexin 43: sense, 5'- GGACTGCTTCCTCTCACGTC-3', antisense, 5'- CAGCTTGTACCCAGGAGGAG-3' and synaptotagmin 1: sense 5'-CAATACTGCCATTCCCTCGT-3', antisense, 5'- GTAGCAGGCTCACCTTCCTG-3'. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was ampli?ed as an internal control by using the primer sets: sense, 5-AACGACCCCTTCATTGAC-3' antisense 5'-TCCACGACATACTCAGCAC-3'. PCR conditions were adjusted to be in a linear range of amplification.
Data analysis and Statistics
Results were analyzed by t-test, paired t-test, ONE-WAY ANOVA, TWO-WAY ANOVA followed by post-hoc Student-Newman-Keuls test to identify significant differences. All data are expressed as mean ± S.E.M. In all cases, P < 0.05 was considered statistically significant.
Results
Decreased anxiety-related behaviors in CaMKIV-/- mice
To determine if CaMKIV plays a role in the expression of anxiety-like behaviors, CaMKIV-/- mice were tested in the elevated plus-maze, dark-light emergence test, light/dark box, acoustic startle and for the prepulse inhibition of startle. Previous studies showed that general locomotor activity was similar between CaMKIV-/- mice in an open field when compared to wild-type mice, however, CaMKIV-/- mice spent significantly more time exploring the center of the open field, which is indicative of a reduction in anxiety-like behavior [16]. To further assess anxious behaviors in CaMKIV-/- mice, we used the elevated plus maze, which is a well-validated, conventional test for anxiety-related behavior in mice [17]. In this test, an increase in anxiety correlates with a decrease in exploration of the open arms of the maze. The number of entries into the closed arms or the total number of arm entries can be taken as a measure of locomotor activity. CaMKIV-/- mice spent significantly more time in the open arms of the plus-maze (Fig. 1b; P < 0.001) and displayed an increase in the number of open arm visits (Fig. 1a; P < 0.01) when compared to wild-type mice. The number of entries into the closed arms (Fig. 1c; P < 0.05) and the total number of arm entries was also significantly increased (Fig. 1d; P < 0.001), suggesting an increase in locomotor activity.
Figure 1 Decreased anxiety-like behavior in CaMKIV-/- mice. a, CaMKIV-/- mice (n = 14 mice) displayed a significantly increased number of open arm entries (t(27) = -3.89, P < 0.001) compared with wild-type mice (n = 19 mice) in the elevated plus-maze. b, The percentage of time spent by CaMKIV-/- mice in the open arms was significantly greater than wild-type mice (n = 19 mice, t(31) = -3.96, P < 0.01). c, d, There was a significant increase in the number of closed arm entries in CaMKIV-/- mice (n = 14 mice, t(27) = -2.21, P < 0.05) as well as total arm entries (n = 14 mice, t(27) = -3.90, P < 0.001) when compared to wild-type mice (n = 19 mice).
Increased exploration in the dark-light emergence test
To strengthen our hypothesis that CaMKIV may play a role in the regulation of anxiety-like behaviors, we tested the performance of CaMKIV-/- mice in the dark-light emergence test [18]. CaMKIV-/- mice again displayed a significant decrease in anxiety-like behavior. We found that the overall time spent in the open field was significantly greater in CaMKIV-/- mice (Fig. 2a; P < 0.05). We further examined changes in anxiety-related behavior using the light/dark box [19]. Anxiogenic agents decrease, while anxiolytic drugs increase the amount of time spent in the light half of the chamber [20]. The total time spent in the light compartment did not significantly differ between wild-type and CaMKIV-/- mice (Fig. 2b). However, CaMKIV-/- mice displayed a significant decrease in the number of light/dark transitions (Fig 2c; P < 0.05). Since this test is based on a mouse's natural aversion to brightly lit spaces, an overall decrease in exploration by CaMKIV-/- mice may represent a reduction in conflict between the two chambers. Although CaMKIV-/- mice did not display a consistent anxiety-like phenotype in the light/dark box paradigm, evidence suggests that similar rodent behavioral tests may measure different forms of anxiety-like behavior [21,22]. Previous studies have found targeted gene mutations can produce anxiety-like phenotypes in some tests but not others [22,23]. Therefore, the decrease in anxiety-like behavior seen in the elevated plus-maze and dark-light emergence test is likely to reliably represent the anxiety phenotype in CaMKIV-/- mice.
Figure 2 Dark-light emergence test and Light/Dark box paradigm. a, In the dark-light emergence test, CaMKIV-/- mice (n = 7 mice) spent significantly more time outside the chamber compared to wild-type mice (n = 8 mice, t(13) = -2.65, P < 0.05,). b, In the light/dark box, no difference was seen between CaMKIV-/- and wild-type mice in the time spent in the light compartment. c, However, wild-type mice (n = 10 mice) displayed an increase in the number of light/dark transitions compared with CaMKIV-/- mice (n = 6, t(14) = 2.44, P < 0.05).
Impaired startle and prepulse inhibition of startle
The acoustic startle response is elicited by a sudden loud acoustic stimulus and is characterized by a coordinated contraction of the muscles of the neck and extremities. The percent prepulse inhibition (PPI) is an index of sensorimotor gating. We examined the acoustic startle response and PPI in mice lacking CaMKIV. Consistent with a reduction in anxiety-like behaviors, CaMKIV-/- mice displayed a decrease in baseline acoustic startle amplitudes (Fig. 3a; P < 0.05, P < 0.001, and P < 0.001 for 100 dB, 110 dB and 120 dB respectively). Similarly, CaMKIV-/- mice showed a significant reduction in prepulse inhibition of the startle reflex at prepulse intensities of 90 and 100 dB (Fig 3b; P < 0.001 for both intensities). These results further support our hypothesis that CaMKIV may play a role in emotional behaviors.
Figure 3 Decreased acoustic startle and PPI. a, Baseline startle amplitudes were significantly decreased at 100, 110 and 120 dB in CaMKIV-/- mice (wild-type mice, n = 7 mice, CaMKIV-/- n = 13 mice, at intensities of 100, 110, 120 dB, q = 3.44, P < 0.05, q = 5.02, P < 0.001, q = 6.55, P < 0.001 respectively). b, Prepulse inhibition of the startle reflex was significantly decreased at prepulse intensities of 90 and 100 dB (wild-type, n = 10 mice, CaMKIV-/-, n = 18 mice, at intensities of 90 dB, q = 3.98, P < 0.01, 100 dB, q = 4.05, P < 0.01).
CaMKIV plays a role in stress-induced analgesia
To determine if CaMKIV plays a role in stress-induced behavioral changes, CaMKIV-/- and wild-type mice were tested for stress-induced analgesia after either restraint or swim stress. Consistent with previous studies [24], restraint stress induced analgesic effects in the tail-flick and hotplate tests in wild-type mice (Fig. 4a–d). In contrast, CaMKIV-/- mice failed to show any analgesia in both tests (P < 0.01 for both tests and area under the curve P < 0.001 and P < 0.05 for hotplate and tail-flick tests respectively). We can exclude the possibility that this is due to altered spinal nociceptive transmission since CaMKIV-/- mice have unaltered baseline tail-flick and hotplate responses [13]. Similar results were obtained using the cold-swim stress test (Fig. 4e–f; P < 0.05 comparisons of CaMKIV-/- and wild-type mice using area under the curve). Here, we measured behavioral nociceptive responses on the hot-plate after animals were forced to swim in cold water. CaMKIV-/- mice exhibited significantly reduced stress-induced analgesia when compared to wild-type controls. Taken together, these results suggest a role for CaMKIV in the regulation of stress-induced analgesia.
Figure 4 Restraint and cold swim stress. a, c, Antinociceptive responses in the hot-plate (n = 8 wild-type mice, n = 9 CaMKIV-/- mice, F(1,7) = 68.21, P < 0.01) and tail-flick test (F(1,7) = 13.59, P < 0.01) following restraint stress (30 min) was significantly decreased in CaMKIV-/- mice. b, d, antinociceptive effect presented as area under the curve between 0 to 60 min after restraint stress in the hot-plate test (t(15) = 4.13, P < 0.001) and tail-flick test (t(15) = 2.86, P < 0.05) show a lack of stress-induced analgesia in CaMKIV-/- mice. e, f, Similarly, hotplate responses after cold-swim test (n = 11 wild-type mice, n = 12 CaMKIV-/- mice) was significantly reduced in CaMKIV-/- mice (F(1,6) = 7.39, P < 0.05, area under the curve, t(21) = 2.26, P < 0.05)
Changes in Stress/anxiety-related gene expression
Several studies show that altering CREB function can change anxiety-like behaviors [7,25] and since CaMKIV can directly modulate the activity of this major transcription factor, GeneChip analysis was performed on the forebrain of CaMKIV-/- mice to uncover any changes in the expression levels of genes related to emotional behavior. We compared gene expression profiles between CaMKIV-/- and wild-type mice and found 233 genes to be differentially expressed with lower expression levels in CaMKIV-/- mice than in wild-type animals. We filtered out the top five anxiety/stress-related genes among down-regulated genes based on previous literature (Table 1). Of particular interest, we found that the expression level of the neuropeptide oxytocin, which is known to mediate emotional behaviors such as social recognition [26], aggression [27] and stress-induced analgesia [24], was decreased in CaMKIV-/- mice by more than twofold (2.83 fold, P < 0.05). To confirm this GeneChip data, reverse transcription-polymerase chain reaction (RT-PCR) was performed to measure the mRNA expression level of oxytocin in CaMKIV-/- mice (Fig. 5). The results of the RT-PCR were consistent with those from GeneChip analysis, showing a decrease in oxytocin mRNA in CaMKIV-/-. This suggests that, as an upstream regulator of CREB, CaMKIV may play a role in the regulation of anxiety-related genes such as oxytocin. Since oxytocin has been shown to mediate stress-induced analgesia in mice [24] and our gene expression analysis showed that the basal expression of oxytocin is reduced after the deletion of CaMKIV, this result is consistent with the lack of stress induced analgesia in CaMKIV-/- mice.
Table 1 Top 5 down-regulated anxiety-related gene expression in CaMKIV deficient mice
Probe Gene name Accession Symbol LocusLink Gene Description
1450794_at Arginine vasopressin NM_009732.1 Avp 11998 neurohypophysial peptide[35]
1420556_at Oxytocin NM_011025.1 Oxt 18429 neurohypophysial peptide[37]
1451580_a_at Transthyretin BC024702.1 Ttr 22139 transporter of thyroxine and vitamin A[41]
1415801_at Connexin 43 M63801.1 14609 subunit of gap junction channel[42]
1421990_a Synaptotagmin 1 NM_009306.1 Syt1 20979 Ca2+ sensor for transmitter release[43]
Figure 5 Decreased oxytocin mRNA expression in CaMKIV-/- mice. RT-PCR analysis revealed that the mRNA level of oxytocin was decreased in CaMKIV-/- compared with wild-type mice. RT-PCR analysis confirmed the altered expression of oxytocin identified from GeneChip analysis.
Discussion
In the present study, we show that the genetic disruption of CaMKIV in mice results in a decrease in anxiety-like behavior and the abolishment of stress-induced analgesia. CaMKIV-/- mice display reduced anxiety-like behaviors in the elevated plus-maze, dark-light emergence test, and in the acoustic startle reflex and PPI. Furthermore, these mice lack stress-induced analgesia induced by forced swim or restraint stress.
Several kinase cascades regulate CREB function in the CNS [1,9], these include the cAMP signaling pathway and Ca2+-calmodulin-dependent protein kinase pathway. The nuclear location of CaMKIV, its ability to phosphorylate CREB and its broad expression throughout forebrains areas suggests that CaMKIV may play an important role in higher brain function. CaMKIV may also play a role in CREB phosphorylation by modulating other kinase pathways. Studies in cell culture systems demonstrated the regulatory role of CaMKIV in MAP kinase and cAMP pathways [28,29]. Disturbances in any of these pathways could potentially disrupt the control of CREB-mediated anxiety-related gene expression. Recent data showed the involvement of the CaMKIV cascade in antidepressant mechanisms [30]. In-vitro experiments have suggested a role for the CaMKIV signaling pathway in the excitation-mediated regulation of corticotrophin-releasing hormone (CRH) synthesis [14].
Previous studies of CREB mutant mice reported an increase in anxiety-like behavior in several behavioral paradigms including the elevated plus-maze, black and white box and open field [5]. Although mice lacking different isoforms of CREB responded differently to certain stressful situations, all CREB mutants displayed anxiety-like responses in all behavior models [7,25]. In another study, the CREB-related transcription factor CREM was shown to be involved in the control of anxiety-like behavior. CREM mutant mice were hyperactive in the open field but displayed reduced anxiety-like behaviors in the elevated plus-maze and zero maze [6]. From our data, CaMKIV-/- mice displayed reduced anxiety-like behavior by spending more time in the open arms of the elevated plus-maze, which correlates with the phenotype of the CREM mutant mice. However, mice lacking CaMKIV did not show hyperlocomotor activities in the open field [15]. Our results show that CaMKIV-/- mice made significantly more entries into the open and closed arms of the elevated plus-maze, suggesting an increase in locomotor activity. One explanation for this discrepancy may be because CaMKIV-/- mice have less anxiety so they may spend more time exploring the anxious environment of the elevated plus maze when compared to wild-type mice. These results suggest that CaMKIV-/- mice have a decrease in anxiety in the plus-maze compared to wild-type mice, and hint that CaMKIV may play a role in regulating levels of anxiety. Therefore, the contrasting phenotypes of between CaMKIV, CREB and CREM knockout mice emphasize the complexity of the transcription factor and genetics underlying such emotional behavior. These studies further suggest that CaMKIV may play an important role in anxiety-like behavior through its regulation of CREB since CREB mutant mice displayed alterations in emotional behavior [5,7,25].
Microarray analysis provides us the opportunity to screen for changes in thousands of genes at the same time [31] and this technology was applied to several studies of gene expression ranging from complex clinical diseases such as schizophrenia [32,33] to human cancer [34]. We used microarray analysis to assay the relative gene expression levels in the forebrain of CaMKIV-/- and wild-type mice. We found at least 200 genes down-regulated, five of which were stress/anxiety-related genes (Table 1). Of these genes, two were neuroendocrine hormones; vasopressin and oxytocin. Mice lacking the vasopressin VIa receptor exhibit markedly reduced anxiety-like behavior [35] while mice lacking oxytocin display elevated anxiety-like behavior [36,37] and lack stress-induced analgesia [24]. Interestingly, other genes found to be reduced in CaMKIV-/- mice include the intracellular transportor gene, transthyretin; secretory vesicle protein, synaptotagmin 1, and a gap junction protein, connexin 43. The down-regulation of these genes suggests that the deletion of CaMKIV may affect the expression of genes involved in the regulation of presynaptic terminals and its structure. These genes are shown to be anxiety/stress related and may have other roles in addition to anxiety. The GeneChip data included in this study may help us to find a mechanism for the role of CaMKIV in anxiety and such a mechanism may include oxytocin.
Of interest, accumulating evidence suggests that oxytocin possesses anxiolytic properties and is important for stress-induced analgesia [37-39]. Oxytocin in the amygdala is essential for social recognition and the control of anxiety [26,36,40] and central administration of oxytocin is anxiolytic and attenuates the stress response. CaMKIV-/- mice, with reduced oxytocin levels, display a lack of stress-induced analgesia. Contrary to observations in oxytocin deficient mice [37-39], which displayed heightened anxiety behaviors, the present data showed that CaMKIV-/- mice display lower anxiety-like behaviors. Our findings imply that changes in oxytocin expression may contribute to the changes in anxiety level of CaMKIV-/- mice. Thus, it appears that CaMKIV may regulate the expression of many anxiety-related genes, including oxytocin, but that this regulation cannot solely account for the behavioral phenotype. While future studies are needed to thoroughly characterize changes in gene expression that result from the deletion of CaMKIV and how these changes affect behavioral phenotypes, our initial GeneChip data provides a good starting point for the dissection of molecular changes responsible for the anxiety phenotype of CaMKIV-/- mice.
Our results suggest that a nuclear protein kinase, with broad forebrain distribution and the ability to affect gene expression through the activation of CREB, plays a role in anxiety-like behavior and that modulation of CaMKIV activity may prove useful in modifying anxious behavior. Future studies are needed to elucidate the role of CaMKIV in the molecular changes involved in the modulation of anxiety-like behaviors.
Acknowledgements
The authors would like to thank Dr. Talal Chatila for the generous gift of CaMKIV-/- mice. This work was supported by grants from the EJLB-CIHR Michael Smith Chair in Neurosciences and Mental Health in Canada and Canada Research Chair.
==== Refs
West AE Griffith EC Greenberg ME Regulation of transcription factors by neuronal activity Nat Rev Neurosci 2002 3 921 931 12461549 10.1038/nrn987
Bourtchuladze R Frenguelli B Blendy J Cioffi D Schutz G Silva AJ Deficient long-term memory in mice with a targeted mutation of the cAMP-responsive element-binding protein Cell 1994 79 59 68 7923378 10.1016/0092-8674(94)90400-6
Josselyn SA Shi C Carlezon WAJ Neve RL Nestler EJ Davis M Long-term memory is facilitated by cAMP response element-binding protein overexpression in the amygdala J Neurosci 2001 21 2404 2412 11264314
Impey S Smith DM Obrietan K Donahue R Wade C Storm DR Stimulation of cAMP response element (CRE)-mediated transcription during contextual learning Nat Neurosci 1998 1 595 601 10196567 10.1038/2830
Valverde O Mantamadiotis T Torrecilla M Ugedo L Pineda J Bleckmann S Gass P Kretz O Mitchell JM Schutz G Maldonado R Modulation of anxiety-like behavior and morphine dependence in CREB-deficient mice Neuropsychopharmacology 2004 29 1122 1133 15029152 10.1038/sj.npp.1300416
Maldonado R Smadja C Mazzucchelli C Sassone-Corsi P Altered emotional and locomotor responses in mice deficient in the transcription factor CREM Proc Natl Acad Sci U S A 1999 96 14094 14099 10570204 10.1073/pnas.96.24.14094
Barrot M Olivier JD Perrotti LI DiLeone RJ Berton O Eisch AJ Impey S Storm DR Neve RL Yin JC Zachariou V Nestler EJ CREB activity in the nucleus accumbens shell controls gating of behavioral responses to emotional stimuli Proc Natl Acad Sci U S A 2002 99 11435 11440 12165570 10.1073/pnas.172091899
Nibuya M Nestler EJ Duman RS Chronic antidepressant administration increases the expression of cAMP response element binding protein (CREB) in rat hippocampus J Neurosci 1996 16 2365 2372 8601816
Shaywitz AJ Greenberg ME CREB: a stimulus-induced transcription factor activated by a diverse array of extracellular signals Annu Rev Biochem 1999 68 821 861 10872467 10.1146/annurev.biochem.68.1.821
Nakamura Y Okuno S Sato F Fujisawa H An immunohistochemical study of Ca2+/calmodulin-dependent protein kinase IV in the rat central nervous system: light and electron microscopic observations Neuroscience 1995 68 181 194 7477923 10.1016/0306-4522(95)00092-W
Matthews RP Guthrie CR Wailes LM Zhao X Means AR McKnight GS Calcium/calmodulin-dependent protein kinase types II and IV differentially regulate CREB-dependent gene expression Mol Cell Biol 1994 14 6107 6116 8065343
Kasahara J Fukunaga K Miyamoto E Differential effects of a calcineurin inhibitor on glutamate-induced phosphorylation of Ca2+/calmodulin-dependent protein kinases in cultured rat hippocampal neurons J Biol Chem 1999 274 9061 9067 10085155 10.1074/jbc.274.13.9061
Wei F Qiu CS Liauw J Robinson DA Ho N Chatila T Zhuo M Calcium calmodulin-dependent protein kinase IV is required for fear memory Nat Neurosci 2002 5 573 579 12006982 10.1038/nn855
Yamamori E Asai M Yoshida M Takano K Itoi K Oiso Y Iwasaki Y Calcium/calmodulin kinase IV pathway is involved in the transcriptional regulation of the corticotropin-releasing hormone gene promoter in neuronal cells J Mol Endocrinol 2004 33 639 649 15591024
Ho N Liauw JA Blaeser F Wei F Hanissian S Muglia LM Wozniak DF Nardi A Arvin KL Holtzman DM Linden DJ Zhuo M Muglia LJ Chatila TA Impaired synaptic plasticity and cAMP response element-binding protein activation in Ca2+/calmodulin-dependent protein kinase type IV/Gr-deficient mice J Neurosci 2000 20 6459 6472 10964952
Ko SW Chatila T Zhuo M Contribution of CaMKIV to injury and fear- induced ultrasonic vocalizations in adult mice Mol Pain 2005 1 10 15813959 10.1186/1744-8069-1-10
Lister RG The use of a plus-maze to measure anxiety in the mouse Psychopharmacology (Berl) 1987 92 180 185 3110839
Smith GW Aubry JM Dellu F Contarino A Bilezikjian LM Gold LH Chen R Marchuk Y Hauser C Bentley CA Sawchenko PE Koob GF Vale W Lee KF Corticotropin releasing factor receptor 1-deficient mice display decreased anxiety, impaired stress response, and aberrant neuroendocrine development Neuron 1998 20 1093 1102 9655498 10.1016/S0896-6273(00)80491-2
Crawley J Goodwin FK Preliminary report of a simple animal behavior model for the anxiolytic effects of benzodiazepines Pharmacol Biochem Behav 1980 13 167 170 6106204 10.1016/0091-3057(80)90067-2
Kliethermes CL Anxiety-like behaviors following chronic ethanol exposure Neurosci Biobehav Rev 2005 28 837 850 15642625 10.1016/j.neubiorev.2004.11.001
Belzung C Le Pape G Comparison of different behavioral test situations used in psychopharmacology for measurement of anxiety Physiol Behav 1994 56 623 628 7972418 10.1016/0031-9384(94)90311-5
van Gaalen MM Steckler T Behavioural analysis of four mouse strains in an anxiety test battery Behav Brain Res 2000 115 95 106 10996412 10.1016/S0166-4328(00)00240-0
Bale TL Contarino A Smith GW Chan R Gold LH Sawchenko PE Koob GF Vale WW Lee KF Mice deficient for corticotropin-releasing hormone receptor-2 display anxiety-like behaviour and are hypersensitive to stress Nat Genet 2000 24 410 414 10742108 10.1038/74263
Robinson DA Wei F Wang GD Li P Kim SJ Vogt SK Muglia LJ Zhuo M Oxytocin mediates stress-induced analgesia in adult mice J Physiol 2002 540 593 606 11956346 10.1113/jphysiol.2001.013492
Barrot M Wallace DL Bolanos CA Graham DL Perrotti LI Neve RL Chambliss H Yin JC Nestler EJ Regulation of anxiety and initiation of sexual behavior by CREB in the nucleus accumbens Proc Natl Acad Sci U S A 2005 102 8357 8362 15923261 10.1073/pnas.0500587102
Ferguson JN Young LJ Hearn EF Matzuk MM Insel TR Winslow JT Social amnesia in mice lacking the oxytocin gene Nat Genet 2000 25 284 288 10888874 10.1038/77040
Winslow JT Hearn EF Ferguson J Young LJ Matzuk MM Insel TR Infant vocalization, adult aggression, and fear behavior of an oxytocin null mutant mouse Horm Behav 2000 37 145 155 10753584 10.1006/hbeh.1999.1566
Enslen H Tokumitsu H Stork PJ Davis RJ Soderling TR Regulation of mitogen-activated protein kinases by a calcium/calmodulin-dependent protein kinase cascade Proc Natl Acad Sci U S A 1996 93 10803 10808 8855261 10.1073/pnas.93.20.10803
Wu GY Deisseroth K Tsien RW Activity-dependent CREB phosphorylation: convergence of a fast, sensitive calmodulin kinase pathway and a slow, less sensitive mitogen-activated protein kinase pathway Proc Natl Acad Sci U S A 2001 98 2808 2813 11226322 10.1073/pnas.051634198
Tiraboschi E Tardito D Kasahara J Moraschi S Pruneri P Gennarelli M Racagni G Popoli M Selective phosphorylation of nuclear CREB by fluoxetine is linked to activation of CaM kinase IV and MAP kinase cascades Neuropsychopharmacology 2004 29 1831 1840 15138445 10.1038/sj.npp.1300488
Brown PO Botstein D Exploring the new world of the genome with DNA microarrays Nat Genet 1999 21 33 37 9915498 10.1038/4462
Miyakawa T Leiter LM Gerber DJ Gainetdinov RR Sotnikova TD Zeng H Caron MG Tonegawa S Conditional calcineurin knockout mice exhibit multiple abnormal behaviors related to schizophrenia Proc Natl Acad Sci U S A 2003 100 8987 8992 12851457 10.1073/pnas.1432926100
Mirnics K Middleton FA Marquez A Lewis DA Levitt P Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex Neuron 2000 28 53 67 11086983 10.1016/S0896-6273(00)00085-4
Alizadeh AA Eisen MB Davis RE Ma C Lossos IS Rosenwald A Boldrick JC Sabet H Tran T Yu X Powell JI Yang L Marti GE Moore T Hudson JJ Lu L Lewis DB Tibshirani R Sherlock G Chan WC Greiner TC Weisenburger DD Armitage JO Warnke R Levy R Wilson W Grever MR Byrd JC Botstein D Brown PO Staudt LM Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling Nature 2000 403 503 511 10676951 10.1038/35000501
Bielsky IF Hu SB Szegda KL Westphal H Young LJ Profound impairment in social recognition and reduction in anxiety-like behavior in vasopressin V1a receptor knockout mice Neuropsychopharmacology 2004 29 483 493 14647484 10.1038/sj.npp.1300360
Bale TL Davis AM Auger AP Dorsa DM McCarthy MM CNS region-specific oxytocin receptor expression: importance in regulation of anxiety and sex behavior J Neurosci 2001 21 2546 2552 11264328
Amico JA Mantella RC Vollmer RR Li X Anxiety and stress responses in female oxytocin deficient mice J Neuroendocrinol 2004 16 319 324 15089969 10.1111/j.0953-8194.2004.01161.x
Windle RJ Shanks N Lightman SL Ingram CD Central oxytocin administration reduces stress-induced corticosterone release and anxiety behavior in rats Endocrinology 1997 138 2829 2834 9202224 10.1210/en.138.7.2829
Windle RJ Kershaw YM Shanks N Wood SA Lightman SL Ingram CD Oxytocin attenuates stress-induced c-fos mRNA expression in specific forebrain regions associated with modulation of hypothalamo-pituitary-adrenal activity J Neurosci 2004 24 2974 2982 15044536 10.1523/JNEUROSCI.3432-03.2004
Ferguson JN Aldag JM Insel TR Young LJ Oxytocin in the medial amygdala is essential for social recognition in the mouse J Neurosci 2001 21 8278 8285 11588199
Sousa JC Grandela C Fernandez-Ruiz J de Miguel R de Sousa L Magalhaes AI Saraiva MJ Sousa N Palha JA Transthyretin is involved in depression-like behaviour and exploratory activity J Neurochem 2004 88 1052 1058 15009661 10.1046/j.1471-4159.2003.02309.x
Theis M Jauch R Zhuo L Speidel D Wallraff A Doring B Frisch C Sohl G Teubner B Euwens C Huston J Steinhauser C Messing A Heinemann U Willecke K Accelerated hippocampal spreading depression and enhanced locomotory activity in mice with astrocyte-directed inactivation of connexin43 J Neurosci 2003 23 766 776 12574405
Geppert M Goda Y Hammer RE Li C Rosahl TW Stevens CF Sudhof TC Synaptotagmin I: a major Ca2+ sensor for transmitter release at a central synapse Cell 1994 79 717 727 7954835 10.1016/0092-8674(94)90556-8
|
16102169
|
PMC1208947
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Aug 15; 1:22
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-22
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-251613733710.1186/1744-8069-1-25ResearchSpinal cord NR1 serine phosphorylation and NR2B subunit suppression following peripheral inflammation Caudle Robert M [email protected] Federico M [email protected] Valle-Pinero Arseima Y [email protected] Michael J [email protected] Department of Oral and Maxillofacial Surgery and Diagnostic Sciences, University of Florida College of Dentistry, Gainesville, FL 32610, USA2 Department of Neuroscience, University of Florida College of Medicine, McKnight Brain Institute, Gainesville, FL 32610, USA3 Pain and Neurosensory Mechanisms Branch, National Institutes of Dental and Craniofacial Research, National Institutes of Health, Bethesda, MD 20892, USA2005 2 9 2005 1 25 25 22 7 2005 2 9 2005 Copyright © 2005 Caudle et al; licensee BioMed Central Ltd.2005Caudle et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Spinal cord N-methyl-D-aspartate (NMDA) receptors are intimately involved in the development and maintenance of central sensitization. However, the mechanisms mediating the altered function of the NMDA receptors are not well understood. In this study the role of phosphorylation of NR1 splice variants and NR2 subunits was examined following hind paw inflammation in rats. We further examined the level of expression of these proteins following the injury.
Results
Lumbar spinal cord NR1 subunits were found to be phosphorylated on serine residues within two hours of the induction of hind paw inflammation with carrageenan. The enhanced NR1 serine phosphorylation reversed within six hours. No phosphorylation on NR1 threonine or tyrosine residues was observed. Likewise, no NR2 subunit phosphorylation was observed on serine, threonine or tyrosine residues. An analysis of NR1 and NR2 protein expression demonstrated no change in the levels of NR1 splice variants or NR2A following the inflammation. However, spinal cord NR2B expression was depressed by the hind paw inflammation. The expression of NR2B remained depressed for more than one week following initiation of the inflammation.
Conclusion
These data suggest that NR1 serine phosphorylation leads to an initial increase in NMDA receptor activity in the spinal cord following peripheral injury. The suppression of NR2B expression suggests compensation for the enhanced nociceptive activity. These data indicate that spinal cord NMDA receptors are highly dynamic in the development, maintenance and recovery from central sensitization following an injury. Thus, chronic pain therapies targeted to NMDA receptors should be designed for the exact configuration of NMDA receptor subunits and post-translational modifications present during specific stages of the disease.
==== Body
Background
Central sensitization is a form of plasticity in the spinal cord that alters the input/output relationship of the neuronal pain processing circuitry. Central sensitization is symptomatically expressed as allodynia, pain to normally non-painful stimuli, and hyperalgesia, an enhanced sensation of pain to typically painful stimuli. When an individual is injured central sensitization encourages the protection of the injured area by enhancing the pain experience. The individual is then motivated to guard the damaged tissue until it is healed. As a rule, central sensitization will be reversed as the injury heals. However, on occasion it fails to resolve and becomes the patient's primary disease. This disease is referred to as chronic pain. Thus, the molecular processes that induce and reverse central sensitization are important to understanding, preventing and treating chronic pain.
Recent work on pain processing has highlighted the central role of N-methyl-D-aspartate (NMDA) receptors in central sensitization. NMDA receptors were found to play a major role in hyperalgesia, allodynia, and expanded receptive fields when central sensitization had been induced by peripheral injury [1-5]. These findings using NMDA receptor antagonists indicated that NMDA receptors initiated events that lead to neuronal plasticity in the spinal cord and that the NMDA receptors themselves participated in the maintenance of central sensitization. Central sensitization is the result of an increase in intracellular calcium, which enhances synaptic inputs from primary nociceptors. NMDA receptors conduct much of this calcium from the extracellular space through their ionophore. The net effect of the increased calcium is an increased number of effective synapses on dorsal horn neurons and enhanced neuronal excitability [1,6,7].
Central sensitization, it must be noted, is distinct from the frequently studied phenomenon of windup, which is rapidly reversed when the peripheral stimulus ceases. Windup is produced by the well documented voltage dependent magnesium block of the NMDA receptor's ion channel. The magnesium block enables the receptor to integrate nociceptive signals that arrive in the spinal cord via C-fibers. The net result of the integration is that the later stimuli in a series produces greater responses in dorsal horn neurons even when the stimuli are identical to the first event [8-10]. Windup does not lead to a prolonged enhancement of dorsal horn neuronal excitability like central sensitization, but may induce central sensitization by increasing intracellular calcium levels. Thus, although NMDA receptors are involved in both central sensitization and windup their role in the two processes is distinct [10].
Recently, Zou and colleagues examined the role of NMDA receptor subunit phosphorylation in the development of central sensitization [11]. These investigators found that capsaicin injection into the hind paw of rats resulted in an ipsilateral accumulation of phosphorylated NR1 subunits in spinothalamic tract neurons. Zou detected the phosphorylation using an antibody that was selective for phosphorylated serine 897 on the NR1 subunit. Phosphorylation of serine 897 on NR1 results in the accumulation of NMDA receptors in synapses [12]. Zou et al. further demonstrated that PKA mediated the phosphorylation of serine 897 and that the enhanced activity of spinothalamic tract neurons was sensitive to PKA inhibitors[13]. Similarly, Brenner and colleagues demonstrated that noxious heat applied to the hind paw of rats produces an increase in serine 896 phosphorylation of NR1 [14]. This phosphorylation was demonstrated to be mediated by activation of protein kinase C (PKC). Previously, it was demonstrated that PKC phosphorylation of serine 896 acts in concert with PKA phosphorylation of serine 897 to induce membrane insertion of NMDA receptors [15]. The findings from these two groups indicate that a relatively brief nociceptive stimulus in the periphery might induce the migration of new NMDA receptors to synapses between the primary and secondary nociceptive afferents. Presumably, the newly inserted NMDA receptors participate in enhancing synaptic activity and in inducing central sensitization. In contrast to the NR1 serine phosphorylation it was notable that these short duration stimuli did not induce any detectable change in NR1 protein expression.
In an excitotoxity induced spinal cord injury model, we examined NR1 phosphorylation and found that an increase in NR1 serine phosphorylation was associated with spontaneous behaviors that indicate the expression of hyperalgesia and allodynia [16]. We further found that, in contrast to the short term peripheral nociception models, NR1 protein was up regulated in the spinal cord following the excitotoxic injury. These data suggest that sustained injuries produce not only changes in NMDA receptor phosphorylation, but also changes in protein expression. The changes in NMDA receptor protein could lead to a more extended period of central sensitization than the readily reversed NMDA receptor phosphorylation.
In addition to NR1 subunit phosphorylation, it was recently demonstrated that NR2B subunits in the spinal cord were phosphorylated on tyrosine residues following either hind paw inflammation with Complete Freund's Adjuvant (CFA) or mustard oil [17]. The NR2B phosphorylation was blocked by inhibitors of Src type kinases. Interestingly, NR2A was reported not to be phosphorylated following the CFA injury. In addition, these investigators reported no change in the expression of NR2 protein in the spinal cord following the CFA treatment.
Overall, the NMDA receptor phosphorylation studies suggest that phosphorylation of spinal cord NR1 and NR2B subunits leads to enhanced NMDA receptor function and central sensitization. However, it is unclear if extended periods of central sensitization are maintained by NMDA receptor phosphorylation or if changes in NMDA receptor subunit protein expression contribute to the continuance of the sensitized state as was observed in the spinal cord injury model [16]. In the present study we examined phosphorylation on serine, threonine and tyrosine residues, as well as protein expression, in both NR1 and NR2 subunits following carrageenan induced hind paw inflammation. The goal was to determine if there is a transition from NMDA receptor subunit phosphorylation to changes in NMDA receptor subunit protein expression during an extended period of nociceptive input and the relationship of these changes to central sensitization.
Results
Characterization of NR1 Splice Variant Selective Antibodies
Four antibodies were raised in rabbits against four regions of the NR1 subunit for identifying specific splice variants containing the N1, C1, C2 and C2' cassettes [18]. NR1 splice variant clones were transfected into COS-7 cells and the protein was harvested and run on western blots. The western blots were probed with the antibodies. As demonstrated in figure 1 the western blots demonstrated that the antibodies were highly selective for the appropriate NR1 splice variants.
Figure 1 Selectivity of rabbit polyclonal antibodies raised against splice variants of the NR1 subunit of the NMDA receptor. Antibodies were raised against synthetic peptides and used for western blots. The peptide sequences used to raise the antibodies were SKKRNYENLDQLSYDNKRGPC, DRKSGRAEPDPKKKATFRAC, PRRAIEREEGQLQLC, and QYHPTDITGPLNLSDPS for exons 5 (N1), 21 (C1), 22 (C2) and 22 minus (C2') respectively. The antibodies were purified with affinity columns and tested against cloned splice variants of the NR1 subunit transfected into COS-7 cells. The three numbers at the top of the columns identify the splice variant clones. The first number indicates the presence (1) or absence (0) of N1, the second C1 and the third C2. The protein was compared to protein taken from non-transfected cells and rat brain homogenates.
Time Course of Carrageenan Induced Enhancement of Nociception
Sixteen rats received injections of 6 mg of carrageenan into the left hind paw and another sixteen rats received vehicle injections. Eight of the carrageenan rats and eight of the vehicle treated rats were then used for thermal nociceptive testing and the remaining animals were used for the mechanical nociception testing. The time course of the thermal testing demonstrated that the limb withdrawal latency was reduced by carrageenan injection within half an hour (Figure 2A). This reduced withdrawal latency reached a peak by 2 to 6 hours and recovery to baseline levels was achieved within 24 hours (ANOVA, Dunnett's test p < 0.05). Although there was a trend for greater thermal hyperalgesia at 6 hours when compared to the 2 hour time point this difference did not reach significance (paired ttest, p = 0.12). In contrast the non-injected paws demonstrated no significant changes in withdrawal latency. No change in paw withdrawal latency was observed in the vehicle injected animals (data not shown).
Figure 2 Time course of carrageenan induced inflammation on heat hyperalgesia and mechanical allodynia. A. Paw withdrawal latencies from a thermal stimulus before and following injection of 6 mg carrageenan into the plantar surface of the left hind paw. Data are means ± SEM (N = 8 rats). Asterisks indicate p < 0.05 ANOVA followed by Dunnett's test. B. Thresholds for paw withdrawal from a mechanical stimulus before and following the injection of 6 mg carrageenan into the plantar surface of the left hind paw. Data are means ± SEM (N = 8 rats). Asterisks indicate p < 0.05 ANOVA followed by Dunnett's test.
In the mechanical allodynia assay the carrageenan injected paws demonstrated significantly enhanced sensitivity to mechanical stimuli within half an hour. The enhanced sensitivity reached a peak within 6 hours. However, in contrast to the thermal assay the rats did not completely return to baseline levels for at least 2 days, but were fully recovered by 5 days (Figure 2B). The non-injected paws and vehicle treated animals (data not shown) demonstrated no significant change in mechanical sensitivity during the entire time course.
NR1 Phosphorylation
To evaluate NR1 phosphorylation rats were injected with either vehicle or 6 mg of carrageenan into the left hind paw. The ipsilateral lumbar spinal cord tissue was harvested and NR1 subunits were precipitated with selective antibodies for the N1 or C1 cassettes [18] or an antibody that recognizes all 8 of the NR1 splice variants. The splice variant selective antibodies were raised in house (Figure 1). The protein was then run on western blots and probed with anti-phosphoserine, anti-phosphothreonine or anti-phosphotyrosine antibodies. A band at ~118 kd was identified by the anti-phospho antibodies, which was consistent with NR1 [14].
Carrageenan injected into the left hind paw induced a significant increase in serine phosphorylation of NR1 protein precipitated with the globally reactive NR1 antibody (N = 6 rats/time point) and with protein precipitated with the C1 cassette selective antibody (N = 8 rats/time point) within 2 hours of the carrageenan injection (ANOVA followed by Dunnett's test on raw data, p < 0.05) (Figure 3A). The phosphorylation of serine residues begins to resolve within 6 hours of the treatment. Interestingly, NR1 protein precipitated with N1 cassette selective antibodies did not demonstrate an increase in serine phosphorylation in response to carrageenan treatment (N = 8 rats per time point, ANOVA followed by Dunnett's test on raw data, p > 0.05).
Figure 3 Serine phosphorylation of spinal cord NR1 subunits following carrageenan induced hind paw inflammation. A. Time course of serine phosphorylation on NR1 subunits immunoprecipitated by antibodies that recognized all NR1 subunits, N1 containing splice variants or C1 containing splice variants. The western blots at the top are representative experiments where the immunoprecipitated protein was probed with anti-phosphoserine antibodies. The bands migrated at approximately 118 kd. The graph summarizes the band density data from the western blots. Data are means ± SEM of the normalized band density data (NR1: N = 6 rats per time point, N1: N = 5 rats per time point, C1: N = 6 rats per time point). Asterisks indicate p < 0.05 ANOVA followed by Dunnett's test on the raw data. B. Comparison of serine phosphorylation on NR1 subunits from the dorsal and ventral spinal cord four hours following the injection of either saline or carrageenan (6 mg) into the left hind paws. At the top is a representative western blot demonstrating the immunoprecipitated NR1 after it was probed with anti-phosphoserine antibodies. The graph is a summary of the normalized band density data for control animals (N = 5), saline injected animals (N = 5) and carrageenan injected animals (N = 5). Data are means ± SEM. Asterisks indicate p < 0.05 ANOVA followed by Dunnett's test.
To determine if the changes in phosphorylation were in the dorsal horn the isolated spinal cord segments of 8 carrageenan injected rats, 8 saline injected rats and 8 non-injected rats were further divided into dorsal and ventral halves and analyzed for serine phosphorylation. Two hours following the carrageenan injections the animals demonstrated an increase in serine phosphorylation in the dorsal horn, but not in the ventral horn (Figure 3B) (ANOVA followed by Dunnett's test on raw data, p < 0.05).
Phosphothreonine and phosphotyrosine antibodies demonstrated no changes in phosphorylation of C1 precipitated NR1 (N = 9 rats/time point) or total NR1 protein (N = 8 rats/time point) (Figure 4A and 4B) (ANOVA followed by Dunnett's test on raw data, p < 0.05). The western blots probed with the anti-phosphotyrosine antibodies were extremely faint (see Figure 4B) indicating that NR1 is not phosphorylated on tyrosine residues to any significant degree.
Figure 4 Phosphorylation of threonine and tyrosine residues on spinal cord NR1 subunits. No phosphorylation of NR1 threonine or tyrosine residues was found in response to peripheral inflammation. A. Lumbar spinal cord NR1 subunits were immunoprecipitated with an antibody that recognizes all NR1 splice variants or an antibody that recognizes splice variants with the C1 cassette before and after left hind paw inflammation with carrageenan. The protein was run on western blots and probed with an anti-phosphothreonine antibody. The western blots at the top are representative experiments and the graph at the bottom presents the mean ± SEM band density of all experiments (Total NR1; N = 8 rats per time point, C1 cassette; N = 9 rats per time point). ANOVA p > 0.05. The molecular weight of the bands on the western blots was ~118 kd. B. NR1 subunit protein was immunoprecipitated with an antibody that recognizes all NR1 splice variants or an antibody that recognizes splice variants with the C1 cassette, run on western blots and probed with an anti-phosphotyrosine antibody. The western blots at the top are representative experiments and the graph at the bottom presents the mean ± SEM band density of all experiments (Total NR1; N = 4 rats per time point, C1 cassette; N = 4 rats per time point). ANOVA p > 0.05.
NR2 Phosphorylation
The NR2 subunits NR2A and NR2B were immunoprecipitated from rat spinal cord extracts as described in the methods section, run on western blots and probed with anti-phosphoserine, anti-phosphothreonine or anti-phosphotyrosine antibodies. The anti-phospho antibodies recognized proteins on the western blots with a molecular weight of approximately 180 kd, which is consistent with NR2A and NR2B [17].
An examination of phosphorylation of NR2A (N = 5 rats/time point) and NR2B (N = 9 rats/time point) demonstrated that there was no significant change in phosphorylation on serine residues (Figure 5A) (ANOVA on raw data, p > 0.05) after carrageenan treatment. Similarly, no change in threonine or tyrosine phosphorylation was detected on either NR2A (N = 6 rats/time point) or NR2B (N = 5 rats/time point) (Figure 5B and 5C) (ANOVA on raw data, p > 0.05).
Figure 5 Phosphorylation of NR2A and NR2B subunits. NR2A and NR2B were immunoprecipitated from rat lumbar spinal before and after carrageenan induced inflammation of the left hind paw. The protein was run on western blots and probed with anti-phosphoserine, anti-phosphothreonine or anti-phosphotyrosine antibodies. Single bands were identified on the western blots that migrated at ~180 kd. A. Phosphoserine on NR2A and NR2B subunits. The western blots at the top demonstrate representative experiments and the graph below presents the mean ± SEM band density for all experiments (NR2A N = 4 rats per time point, NR2B N = 4 rats per time point). ANOVA p > 0.05. B. Phosphothreonine on NR2A and NR2B subunits. The western blots at the top demonstrate representative experiments and the graph below presents the mean ± SEM band density for all experiments (NR2A N = 4 rats per time point, NR2B N = 4 rats per time point). ANOVA p > 0.05. C. Phosphotyrosine on NR2A and NR2B subunits. The western blots at the top demonstrate representative experiments and the graph below presents the mean ± SEM band density for all experiments (NR2A N = 4 rats per time point, NR2B N = 4 rats per time point). ANOVA p > 0.05.
NR1 Protein
In western blots of spinal cord extracts the NR1 proteins were probed with antibodies selective for the N1 (N = 8 rats/time point), C1 (N = 8 rats/time point), C2 (N = 4 rats/time point) or C2' (N = 4 rats/time point) cassettes and all NR1 splice variants (N = 6 rats/time point). Extracts were prepared from animals without carrageenan injections, and 2, 6 and 24 hours following carrageenan injection into the hind paw. All of the antibodies recognized single bands on the westerns with a molecular weight of approximately 118 kd. Lane loading was monitored by probing for actin (data not shown). No change in spinal cord NR1 protein or splice variant expression was found following carrageenan inflammation (Figure 6) (ANOVA on the raw data, p > 0.05).
Figure 6 Peripheral inflammation does not influence NR1 protein expression in the lumbar spinal cord. Spinal cords were harvested before and after carrageenan injection into the left hind paw of rats, homogenized and run on western blots. The western blots were probed with antibodies that recognized all NR1 splice variants (N = 6 rats per time point), splice variants containing the N1 cassette (N = 8 rats per time point), splice variants with the C1 cassette (N = 8 rats per time point), splice variants with the C2 cassette (N = 4 rats per time point) and splice variants with the C2' cassette (N = 4 rats per time point). The western blots at the top are representative experiments and the graph a summary of all experiments. Bars are the mean ± SEM. ANOVA p > 0.05.
NR2 Protein
To examine changes in NR2 protein spinal cord extracts were collected at 2, 6, 24, 48 and 120 hours following carrageenan injections and probed on western blots with antibodies to NR2A (N = 4 rats/time point) or NR2B (N = 4 rats/time point). Lane loading was monitored with antibodies to actin. The western blots for NR2A indicated no significant change in the expression of protein (ANOVA on raw data, p > 0.05) (Figure 7). However, there was a significant decrease in the expression of NR2B that was observed 24 hours following the carrageenan injection (ANOVA, Dunnett's test on raw data, p < 0.05). This decrease in NR2B lasted throughout the remainder of the week.
Figure 7 Expression of spinal cord NR2A and NR2B protein following peripheral inflammation. Rat lumbar spinal cords were harvested before and after carrageenan induced hind paw inflammation. The tissue was homogenized and run on western blots. The membranes were probed with antibodies that were selective for either NR2A (N = 4 rats per time point) or NR2B (N = 4 rats per time point). Single bands were identified at ~180 kd. The western blots are representative for NR2B and actin. Bars on the graph are means ± SEM of band density data normalized to the control bands. Asterisks indicate p < 0.05 ANOVA followed by Dunnett's test on the raw band densities.
Discussion
The involvement of spinal cord N-methyl-D-aspartate (NMDA) receptors in hyperalgesia and allodynia is well documented in the literature [2-5,19-21]. However, the mechanisms by which the receptors mediate the development and maintenance of allodynia and hyperalgesia in a chronic pain state are not well understood. Long lasting changes in the NMDA receptors either through post-translational modifications or expression of different subunits may lead to a chronic enhancement of nociception. In this study we used a relatively long lasting nociceptive stimulus to evaluate NR1 and NR2 subunit phosphorylation and protein expression. We found that spinal cord NR1 subunits possessing the C1 cassette are phosphorylated for a few hours following carrageenan injection (Figures 2 and 3). The NR1 phosphorylation was found only in the dorsal horn indicating that the phosphorylation was associated with nociception. Furthermore, the NR1 serine phosphorylation paralleled thermal hyperalgesia with recovery within 24 hours. The serine phosphorylation peaked at 2 hours following the carrageenan injections while the thermal hyperalgesia peaked in the range of 2 to 6 hours. This finding suggests that the NR1 serine phosphorylation may be a contributing factor to the thermal hyperalgesia.
The inflammation did not produce phosphorylation of serines on NR2 subunits or phosphorylation of threonine and tyrosine residues on either NR1 or NR2 subunits. The NR1 phosphorylation data are consistent with the shorter duration stimuli used by Zou et al. [11,13] and Brenner et al. [14], and with our previous spinal cord injury study [16]. However, we did not find that NR2B subunits were phosphorylated on tyrosine residues as was reported by Guo et al. [17]. The reason for this discrepancy is not known, but it could be due to the differences in inflammation protocols. We used carrageenan to inflame the hind paws while Guo et al. used Complete Freund's Adjuvant (CFA) to induce inflammation.
In a previous study, Prybylowski and colleagues demonstrated that NR1 subunits containing the C1 cassette in the spinal cord of rats represented only about 5 percent of the total NR1 protein [22], which suggests that C1 containing NR1 subunits may not play a significant role in the spinal cord. However, the group did not determine whether these NR1 subunits were localized to any specific area of the cord such as the dorsal horn or whether they were distributed throughout the lamina. It is possible that the C1 containing NR1 subunits are localized primarily in the superficial lamina, thus enhancing their abundance within neurons in the nociceptive pathways. From the work of Brenner et al. [14] and Zou et al. [13] it is clear that NR1 subunits are readily phosphorylated on serines 890, 896 and 897 by serine/threonine kinases during nociception, which is consistent with our findings. These serine residues reside in the C1 cassette indicating that NR1 subunits containing C1 are critical to nociceptive processing in the spinal cord. Interestingly, Prybylowski et al. also demonstrated that the majority of the NR1 subunits that contain N1 do not contain the C1 cassette [22]. This finding is consistent with our data demonstrating that when N1 is used as the precipitating epitope we do not find any injury induced phosphorylation of NR1 (Figure 3A) suggesting an absence of C1 in the precipitated protein. The sum of the data therefore suggests that NR1 subunits containing the C1 cassette and lacking the N1 cassette represent a dynamic pool of NMDA receptors that are rapidly recruited to the neuronal membrane following a nociceptive stimulus.
We further investigated the expression of NR1 splice variants, NR2A and NR2B protein. These data indicated no significant changes in NR1 splice variants or NR2A protein (Figures 6 and 7). However, NR2B expression was depressed by 24 hours following the carrageenan inflammation and did not recover during the 5 day period examined. The lack of change in NR1 subunits is consistent with the findings of Zou et al. [11,13], Brenner et al. [14] and Gaunitz et al. [23], but not with our spinal cord injury study [16]. These findings suggest that NR1 protein expression is not influenced by peripheral injuries, but is altered by central nervous system injuries. NR2B protein expression, in contrast, was suppressed by the peripheral inflammation. Guo et al. [17] did not find a change in NR2B protein following CFA induced peripheral inflammation and Gaunitz et al. [23] found no change in NR2B mRNA 6 hours following peripheral formalin injection. However, Gaunitz et al. did find a modest increase in NR2A and NR2C mRNA. From our findings we may conclude that the NR2A mRNA increases may not translate into increased levels of NR2A protein. We, however, did not examine NR2C protein.
The suppression of NR2B following peripheral inflammation could have significant effects on nociception and the choice of treatments for chronic pain. NR2B subunits are located primarily in laminas I and II of the dorsal horn [24]. These subunits are involved in windup [25] and central sensitization [26] suggesting a major role for NR2B subunits in the function of NMDA receptors mediating nociception. Furthermore, recent work by Tan and colleagues [27] demonstrated that selective knockdown of NR2B in the dorsal horn using siRNA can suppress formalin induced nocifensive behaviors. Interestingly, our data demonstrate a strong negative correlation between the recovery period of mechanical allodynia and the level of NR2B protein expression (Figure 8, Deming linear regression, r2 = -0.93, p = 0.0363). This might indicate that the suppression of NR2B is used to compensate for the enhanced nociceptive barrage. It is also possible; however, that NR2B is replaced by either NR2C or NR2D during a persistent nociceptive barrage. Thus, the NMDA receptors may still remain functional as the NR2B protein decreases. We have not yet examined NR2C or NR2D to confirm this hypothesis.
Figure 8 Correlation between NR2B protein and recovery of mechanical allodynia following carrageenan induced hind paw inflammation. Data from 6 hours to 120 hours from the NR2B western blots (figure 7) and the mechanical allodynia data (figure 2) were plotted against each other. Deming linear regression indicates r2 = -0.93, p = 0.0363.
Several investigators have demonstrated that pharmacological agents that target NR2B subunits, such as Ifenprodil, can be used to control pain [28-31]. The loss of NR2B subunits in the spinal cord as the nociceptive stimulus progresses suggests that pharmacological agents targeting NR2B may be less effective in chronic pain than in acute pain. This idea is supported by the work of Nakazato and colleagues and De Vry et al. who recently demonstrated that intrathecally administered NR2B selective agents are ineffective at blocking mechanical allodynia in nerve constriction injury models [32,33]. These findings contrast with data that demonstrate NR2B selective antagonists suppress allodynia in acute nociception models [34] and that intrathecal siRNA knockdown of NR2B suppresses formalin induced nociceptive behaviors [27]. The efficacy of intrathecal NR2B antagonists in acute models and the lack of effect in chronic injury models support the concept that NR2B is down regulated in the spinal cord as the injury progresses.
The temporal association of thermal hyperalgesia with NR1 phosphorylation while recovery of mechanical allodynia is correlated with the suppression of NR2B suggests that the two behavioral phenomena have distinct mechanisms. Zou et al. and Brenner et al. have demonstrated that NR1 phosphorylation is mediated by the serine/threonine kinases protein kinase C (PKC) and protein kinase A (PKA) [11,13,14]. Their findings suggest that phosphorylation of NR1 induces the migration of NMDA receptors from the endoplasmic reticulum to synapses leading to enhanced transmission of nociceptive information [14,35,36]. It is possible that the phosphorylated receptors remain in the synapses and remain functional following dephosphorylation. The dephosphorylation of the receptors may reduce stimulation of dorsal horn neurons produced by C-fibers transmitting thermal information. In contrast, the enhanced number of NMDA receptors remaining in synapses may still conduct a significant amount of information from A-delta fibers that transmit mechanical information. Only as the number of NR2B subunits is reduced do the synapses return toward normal function. Presumably, the loss of NR2B subunits would prevent the remaining NR1 subunits from forming functional receptors in much the same way as siRNA to NR2B was found to suppress nociceptive behaviors by selectively reducing the amount of NR2B protein [27]. Thus, thermal hyperalgesia may be mediated by the enhanced function of phosphorylated NMDA receptors, while mechanical allodynia may simply require more NMDA receptors in the synapse.
Alternatively, the phosphorylated NR1 subunits may be inserted into C-fibers synapses while the NR2B subunits are located primarily in synapses associated with A-delta fibers. Thus, in the sensitized neurons suppression of the NR1 phosphorylation results in reduction of thermal hyperalgesia and a decrease in NR2B expression produces an inhibition of mechanical allodynia. These hypotheses are highly speculative at this time as the differences observed between the two forms of stimuli may actually be mediated by changes in the periphery rather than centrally. However, it is intriguing to consider that thermal hyperalgesia might be controlled via kinase inhibitors, whereas established mechanical allodynia may require selective NMDA receptor antagonists or suppression of NMDA receptor gene expression to be controlled.
In summary, our data indicate that peripheral inflammation with carrageenan results in a transient increase in serine phosphorylation of spinal cord dorsal horn NR1 subunits and a long lasting decrease in the expression of NR2B. These data indicate that NMDA receptors in the spinal cord are highly dynamic and may represent a moving target for pharmacological control of chronic pain.
Methods
Male Sprague Dawley rats (200–300 g, Harlan Sprague Dawley, Indianapolis, IN) were maintained on a 12 hour light/dark cycle and fed standard rodent chow and water ad libitum. All experiments were approved by the University of Florida Institutional Animal Care and Use Committee.
Hind Paw Inflammation
Inflammation was induced in the left hind paw of the rats by subcutaneous injection of carrageenan (6 mg in 150 μl of saline) (Sigma, St. Louis, MO) into the plantar surface. Control animals received only saline (150 μl) injections. Animals used for behavioral tests were tested prior to the inflammation and at 0.5, 2, 6, 24, 48 and 120 hours following the injections. Rats used for western blots were euthanized (CO2 inhalation) and the spinal cord tissue was harvested at 2, 6 and 24 hours following the hind paw injections. For the analysis of NR2 proteins tissue was harvested at 2, 6, 24, 48 and 120 hours following the hind paw inflammation.
Thermal Nociception Assay
Thermal nociception was measured using the method of Hargreaves et al. [37]. Briefly, the rats were placed on a clear plastic surface and allowed 15 minutes to accommodate to the enclosure. An infrared light was directed onto a hind paw's plantar surface approximately in the middle of the foot. The latency for the animal to remove its foot from the path of the light was used as the dependent measure for thermal sensitivity. The light intensity was initially adjusted to produce latencies of approximately 8 s and a cutoff time of 20 s was used to prevent injury to the animals. Both hind paws were tested 3 times and the average of the 3 tests was used as the paw withdrawal latency for that time point. A rest period of at least 2 minutes was observed before the animals were retested to prevent sensitization of the paws.
Mechanical Allodynia Assay
An electronic Von Frey device (Ugo Basile, Italy) was used for analyzing mechanical sensitivity in the rats. The animals were placed on a wire screen and allowed 15 minutes to become accustomed to the device. A small steel rod (~0.5 mm in diameter) with a blunt end was pressed against the plantar surface of the paw and a pressure gradient was applied (0 – 50 g) over the course of 20 s. The force at which the rat moved its paw was used as the dependent measure of mechanical sensitivity. Both hind paws were tested on each rat 3 times and the average of the 3 tests was used as the data point for that time period. A rest period of at least 2 minutes was observed before the animals were retested to prevent sensitization.
Immunoprecipitation and Western Blots
The rats were euthanized at the indicated times with CO2 and the spinal cords were removed by pressure ejection with 5 mls ice cold phosphate buffered saline (pH 7.4). The L2 to L5 region of the cord was isolated and divided in half lengthwise. The side ipsilateral to the paw injection was retained for the experiments. The tissue was immediately sonicated and boiled for 10 minutes in denaturing buffer (10 mls/g tissue) containing 1% SDS, 10 mM Tris pH 7.4 and 0.4 mM sodium ortho-vanadate. The solution was then centrifuged at 25,000 g for 30 minutes and the supernatant collected. A sample of the supernatant containing 500 μg of protein was diluted to 500 μl with deionized water and mixed with 500 μl 2 × immunoprecipitation buffer containing 2% triton X-100, 300 mM NaCl, 20 mM Tris pH 7.4, 2 mM EDTA, 2 mM EGTA, 0.4 mM sodium ortho-vanadate, 0.4 mM PMSF and 1% NP-40. The anti-NR1 or anti-NR2 antibodies (2–5 μg) were added to the solution and incubated over night at 4°C. The proteins were then precipitated with agarose beads coupled to the appropriate anti-IgG antibody (Sigma, St. Louis, MO). The beads were isolated and washed in immunoprecipitation buffer. The beads were then suspended in 30 μl immunoprecipitation buffer and diluted with 30 μl 2 × Laemmli buffer (BioRad, Hercules, CA) and boiled for 10 minutes. The samples were run on 3 – 20% SDS-PAGE gels and semi-dry transferred to PVDF membranes. Phospho-serine, -threonine and -tyrosine were then examined using the appropriate phospho-selective anti-bodies (Sigma, St. Louis, MO). The bands were viewed using HRP chemiluminescence and film. Band density was determined using Scion Image (Scion corporation, Gaithersburg, MD). To verify that the immunoprecipitations were quantitative samples of the supernatant following the removal of the beads were run on western blots and probed with the appropriate anti-NR1 or anti-NR2 antibodies. If the precipitated protein was found in the supernatant the precipitation procedure was repeated.
To analyze NR1 and NR2 protein a sample of the denaturing buffer containing 20 μg of protein was mixed with an equal volume of 2 × Laemmli buffer, boiled (10 minutes) and run as described above. The membranes were then probed with the appropriate antibodies. Lane loading was verified with anti-actin antibodies. The density of the actin bands was determined. If the density of any band in the gel diverged more than one standard deviation from the mean of all the bands the gel was discarded and the experiment repeated. All statistics were performed on the raw band density, but were presented graphically as data normalized to the non-inflamed control data to enhance clarity.
Rabbit antibodies to the exon 5 region (N1), exon 21 region (C1), exon 22 C-terminus (C2) and exon 22 minus C-terminus (C2') cassettes [18] of the NR1 subunit were raised at the National Institute for Dental and Craniofacial Research, National Institutes of Health using the following peptide sequences (N1) SKKRNYENLDQLSYDNKRGPC, (C1) DRKSGRAEPDPKKKATFRAC, (C2) PRRAIEREEGQLQLC and (C2') QYHPTDITGPLNLSDPS. The antibodies were affinity purified and tested for specificity using COS-7 cells transfected with specific NR1 splice variants and western blots (Figure 1). The secondary HRP coupled antibodies were purchased from Sigma (ST. Louis, MO). The remainder of the antibodies were purchased from Chemicon International (Temecula, CA), BD Biosciences (San Jose, CA) or Santa Cruz Biotechnology (Santa Cruz, CA).
Data Analysis
Western blot band density was measured using Scion Image (Scion Corp., Frederick, MD). All statistical analyses were conducted on the raw data. Statistical analysis consisted of ANOVA followed by Dunnett's tests or Deming linear regression using Prism statistical software (Graphpad Software inc., San Diego, CA). Statistical significance was assigned to p ≤ 0.05.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RMC designed the study, ran the behavioral experiments and prepared the manuscript. FMP performed immunoprecipitations and western blots. AYD performed immunoprecipitations and western blots. MJI raised the NR1 splice variant selective antibodies.
Acknowledgements
This work was supported by grants from the National Institute on Drug Abuse DA13166 (RMC), the McKnight Brain Institute Brain and Spinal Cord Injury Research Trust Fund (RMC), and the National Institute of Dental and Craniofacial Research intramural research program (MJI).
==== Refs
Vierck CJ Cannon RL Fry G Maixner W Whitsel BL Characteristics of temporal summation of second pain sensations elicited by brief contact of glabrous skin by a preheated thermode Journal of Neurophysiology 1997 78 992 1002 9307129
Urch CE Rahman W Dickenson AH Electrophysiological studies on the role of the NMDA receptor in nociception in the developing rat spinal cord Developmental Brain Research 2001 126 81 89 11172889 10.1016/S0165-3806(00)00141-3
Woolf CJ Thompson SWN The Induction and Maintenance of Central Sensitization Is Dependent on N-Methyl-D-Aspartic Acid Receptor Activation - Implications for the Treatment of Postinjury Pain Hypersensitivity States Pain 1991 44 293 299 1828878 10.1016/0304-3959(91)90100-C
Ren K Dubner R Central nervous system plasticity and persistent pain Journal of Orofacial Pain 1999 13 155 163 10823030
Ren K Dubner R Nmda Receptor Antagonists Attenuate Mechanical Hyperalgesia in Rats with Unilateral Inflammation of the Hindpaw Neuroscience Letters 1993 163 22 26 7905196 10.1016/0304-3940(93)90220-F
Ren K Hylden JLK Williams GM Ruda MA Dubner R The Effects of A Noncompetitive Nmda Receptor Antagonist, Mk-801, on Behavioral Hyperalgesia and Dorsal Horn Neuronal-Activity in Rats with Unilateral Inflammation Pain 1992 50 331 344 1454389 10.1016/0304-3959(92)90039-E
Ren K Williams GM Hylden JLK Ruda MA Dubner R The Intrathecal Administration of Excitatory Amino-Acid Receptor Antagonists Selectively Attenuated Carrageenan-Induced Behavioral Hyperalgesia in Rats European Journal of Pharmacology 1992 219 235 243 1358641 10.1016/0014-2999(92)90301-J
Ji RR Woolf CJ Neuronal plasticity and signal transduction in nociceptive neurons: implications for the initiation and maintenance of pathological pain Neurobiol Dis 2001 8 1 10 11162235 10.1006/nbdi.2000.0360
Woolf CJ Thompson SW King AE Prolonged primary afferent induced alterations in dorsal horn neurones, an intracellular analysis in vivo and in vitro J Physiol (Paris) 1988 83 255 266 3272296
Woolf CJ Windup and central sensitization are not equivalent Pain 1996 66 105 108 8880830 10.1016/0304-3959(96)03114-4
Zou XJ Lin Q Willis WD Enhanced phosphorylation of NMDA receptor 1 subunits in spinal cord dorsal horn and spinothalamic tract neurons after intradermal injection of capsaicin in rats Journal of Neuroscience 2000 20 6989 6997 10995844
Crump FT Dillman KS Craig AM cAMP-dependent protein kinase mediates activity-regulated synaptic targeting of NMDA receptors Journal of Neuroscience 2001 21 5079 5088 11438583
Zou X Lin Q Willis WD Role of protein kinase a in phosphorylation of NMDA receptor 1 subunits in dorsal horn and spinothalamic tract neurons after intradermal injection of capsaicin in rats Neuroscience 2002 115 775 786 12435416 10.1016/S0306-4522(02)00490-6
Brenner GJ Ji RR Shaffer S Woolf CJ Peripheral noxious stimulation induces phosphorylation of the NMDA receptor NR1 subunit at the PKC-dependent site, serine-896, in spinal cord dorsal horn neurons Eur J Neurosci 2004 20 375 384 15233747 10.1111/j.1460-9568.2004.03506.x
Scott DB Blanpied TA Ehlers MD Coordinated PKA and PKC phosphorylation suppresses RXR-mediated ER retention and regulates the surface delivery of NMDA receptors Neuropharmacology 2003 45 755 767 14529714 10.1016/S0028-3908(03)00250-8
Caudle RM Perez FM King C Yu CG Yezierski RP N-methyl-D-aspartate receptor subunit expression and phosphorylation following excitotoxic spinal cord injury in rats Neurosci Lett 2003 349 37 40 12946581 10.1016/S0304-3940(03)00700-6
Guo W Zou SP Guan Y Ikeda T Tal M Dubner R Ren K Tyrosine phosphorylation of the NR2B subunit of the NMDA receptor in the spinal cord during the development and maintenance of inflammatory hyperalgesia Journal of Neuroscience 2002 22 6208 6217 12122079
Zukin RS Bennett MV Alternatively spliced isoforms of the NMDARI receptor subunit Trends Neurosci 1995 18 306 313 7571011 10.1016/0166-2236(95)93920-S
Petrenko AB Yamakura T Baba H Shimoji K The role of N-methyl-D-aspartate (NMDA) receptors in pain: a review Anesth Analg 2003 97 1108 1116 14500166 10.1213/01.ANE.0000081061.12235.55
Riedel W Neeck G Nociception, pain, and antinociception: current concepts Zeitschrift fur Rheumatologie 2001 60 404 + 11826734 10.1007/s003930170003
Willis WD Role of neurotransmitters in sensitization of pain responses Role of Neural Plasticity in Chemical Intolerance 2001 933 142 156
Prybylowski KL Grossman SD Wrathall JR Wolfe BB Expression of splice variants of the NR1 subunit of the N-methyl-D-aspartate receptor in the normal and injured rat spinal cord Journal of Neurochemistry 2001 76 797 805 11158251 10.1046/j.1471-4159.2001.00069.x
Gaunitz C Schuttler A Gillen C Allgaier C Formalin-induced changes of NMDA receptor subunit expression in the spinal cord of the rat Amino Acids 2002 23 177 182 12373534 10.1007/s00726-001-0125-3
Nagy GG Watanabe M Fukaya M Todd AJ Synaptic distribution of the NR1, NR2A and NR2B subunits of the N-methyl-d-aspartate receptor in the rat lumbar spinal cord revealed with an antigen-unmasking technique Eur J Neurosci 2004 20 3301 3312 15610162 10.1111/j.1460-9568.2004.03798.x
Kovacs G Kocsis P Tarnawa I Horvath C Szombathelyi Z Farkas S NR2B containing NMDA receptor dependent windup of single spinal neurons Neuropharmacology 2004 46 23 30 14654094 10.1016/S0028-3908(03)00339-3
Harris J Joules C Stanley C Thomas P Clarke RW Glutamate and tachykinin receptors in central sensitization of withdrawal reflexes in the decerebrated rabbit Exp Physiol 2004 89 187 198 15123548 10.1113/expphysiol.2003.002646
Tan PH Yang LC Shih HC Lan KC Cheng JT Gene knockdown with intrathecal siRNA of NMDA receptor NR2B subunit reduces formalin-induced nociception in the rat Gene Ther 2005 12 59 66 15470478 10.1038/sj.gt.3302376
Parsons CG Danysz W Quack G Glutamate in CNS disorders as a target for drug development: an update Drug News Perspect 1998 11 523 569 15616669 10.1358/dnp.1998.11.9.863689
Chazot PL The NMDA receptor NR2B subunit: a valid therapeutic target for multiple CNS pathologies Curr Med Chem 2004 11 389 396 14965239 10.2174/0929867043456061
Smith PF Therapeutic N-methyl-D-aspartate receptor antagonists: will reality meet expectation? Curr Opin Investig Drugs 2003 4 826 832 14619404
LoGrasso P McKelvy J Advances in pain therapeutics Curr Opin Chem Biol 2003 7 452 456 12941418 10.1016/S1367-5931(03)00078-4
Nakazato E Kato A Watanabe S Brain but not spinal NR2B receptor is responsible for the anti-allodynic effect of an NR2B subunit-selective antagonist CP-101,606 in a rat chronic constriction injury model Pharmacology 2005 73 8 14 15452358 10.1159/000081069
De Vry J Kuhl E Franken-Kunkel P Eckel G Pharmacological characterization of the chronic constriction injury model of neuropathic pain Eur J Pharmacol 2004 491 137 148 15140630 10.1016/j.ejphar.2004.03.051
Minami T Matsumura S Okuda-Ashitaka E Shimamoto K Sakimura K Mishina M Mori H Ito S Characterization of the glutamatergic system for induction and maintenance of allodynia Brain Res 2001 895 178 185 11259776 10.1016/S0006-8993(01)02069-8
Hawkins LM Prybylowski K Chang K Moussan C Stephenson FA Wenthold RJ Export from the endoplasmic reticulum of assembled N-methyl-d-aspartic acid receptors is controlled by a motif in the c terminus of the NR2 subunit J Biol Chem 2004 279 28903 28910 15102836 10.1074/jbc.M402599200
Prybylowski K Fu Z Losi G Hawkins LM Luo J Chang K Wenthold RJ Vicini S Relationship between availability of NMDA receptor subunits and their expression at the synapse J Neurosci 2002 22 8902 8910 12388597
Hargreaves K Dubner R Brown F Flores C Joris J A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia Pain 1988 32 77 88 3340425 10.1016/0304-3959(88)90026-7
|
16137337
|
PMC1208948
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Sep 2; 1:25
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-25
|
oa_comm
|
==== Front
Nutr JNutrition Journal1475-2891BioMed Central London 1475-2891-4-241613893010.1186/1475-2891-4-24ReviewChildhood obesity, prevalence and prevention Dehghan Mahshid [email protected] Noori [email protected] Anwar T [email protected] Population Health Research Institute, McMaster University, Hamilton, Canada2 School of Nursing and Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Canada3 Department of Clinical Epidemiology and Biostatistics, and Population Health Research Institute, McMaster University, Hamilton, Canada2005 2 9 2005 4 24 24 6 6 2005 2 9 2005 Copyright © 2005 Dehghan et al; licensee BioMed Central Ltd.2005Dehghan et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Childhood obesity has reached epidemic levels in developed countries. Twenty five percent of children in the US are overweight and 11% are obese. Overweight and obesity in childhood are known to have significant impact on both physical and psychological health. The mechanism of obesity development is not fully understood and it is believed to be a disorder with multiple causes. Environmental factors, lifestyle preferences, and cultural environment play pivotal roles in the rising prevalence of obesity worldwide. In general, overweight and obesity are assumed to be the results of an increase in caloric and fat intake. On the other hand, there are supporting evidence that excessive sugar intake by soft drink, increased portion size, and steady decline in physical activity have been playing major roles in the rising rates of obesity all around the world. Consequently, both over-consumption of calories and reduced physical activity are involved in childhood obesity.
Almost all researchers agree that prevention could be the key strategy for controlling the current epidemic of obesity. Prevention may include primary prevention of overweight or obesity, secondary prevention or prevention of weight regains following weight loss, and avoidance of more weight increase in obese persons unable to lose weight. Until now, most approaches have focused on changing the behaviour of individuals in diet and exercise. It seems, however, that these strategies have had little impact on the growing increase of the obesity epidemic. While about 50% of the adults are overweight and obese in many countries, it is difficult to reduce excessive weight once it becomes established. Children should therefore be considered the priority population for intervention strategies. Prevention may be achieved through a variety of interventions targeting built environment, physical activity, and diet. Some of these potential strategies for intervention in children can be implemented by targeting preschool institutions, schools or after-school care services as natural setting for influencing the diet and physical activity. All in all, there is an urgent need to initiate prevention and treatment of obesity in children.
==== Body
Introduction
Childhood obesity has reached epidemic levels in developed countries. Twenty five percent of children in the US are overweight and 11% are obese. About 70% of obese adolescents grow up to become obese adults [1-3]. The prevalence of childhood obesity is in increasing since 1971 in developed countries (Table 1). In some European countries such as the Scandinavian countries the prevalence of childhood obesity is lower as compared with Mediterranean countries, nonetheless, the proportion of obese children is rising in both cases [4]. The highest prevalence rates of childhood obesity have been observed in developed countries, however, its prevalence is increasing in developing countries as well. The prevalence of childhood obesity is high in the Middle East, Central and Eastern Europe [5]. For instance, in 1998, The World Health Organization project monitoring of cardiovascular diseases (MONICA) reported Iran as one of the seven countries with the highest prevalence of childhood obesity. The prevalence of BMI (in percentage) between 85th and 95th percentile in girls was significantly higher than that in boys (10.7, SD = 1.1 vs. 7.4, SD = 0.9). The same pattern was seen for the prevalence of BMI > 95th percentile (2.9, SD = 0.1 vs. 1.9, SD = 0.1) [6]. In Saudi Arabia, one in every six children aged 6 to 18 years old is obese [7]. Furthermore, in both developed and developing countries there are proportionately more girls overweight than boys, particularly among adolescent [6,8,9].
Table 1 Changes in the prevalence of overweight and obesity in some developed countries
Country/Year Age/yr Study (author) Change in obesity
USA
1973–1994 5–24 Bogalusa [67] Mean level increased 0.2 kg/yr, twofold increase in prevalence of obesity
1971–1974 6–19 NHANES I [68] Relatively stable
1976–1980 6–19 NHANES II [68] Relatively stable
1988–1994 6–19 NHANES III [68] Doubled to 11%
1999–2000 6–19 NHANES IV [68] Increased by 4%
Japan
1974–1993 6–14 Kotani [69] Doubled (5% to 10%)
UK
1984–98 7–11 Lobstein [70] Changed from 8% to 20%
Spain
1985/6 to 1995/6 6–7 Moreno [71] Changed from 23% to 35%
France
1992–1996 5–12 Rolland-Cachera [72] Changed from 10% to 14%
Greece
1984–2000 6–12 Krassas [73] Increased by 7%
Overweight and obesity in childhood have significant impact on both physical and psychological health; for example, overweight and obesity are associated with Hyperlipidaemia, hypertension, abnormal glucose tolerance, and infertility. In addition, psychological disorders such as depression occur with increased frequency in obese children [10]. Overweight children followed up for 40 [11] and 55 years [12] were more likely to have cardiovascular and digestive diseases, and die from any cause as compared with those who were lean.
Definition of childhood obesity
Although definition of obesity and overweight has changed over time [13,14], it can be defined as an excess of Body Fat (BF). There is no consensus on a cutoff point for excess fatness of overweight or obesity in children and adolescents. Williams et al. [15] measured skin fold thickness of 3320 children aged 5–18 years and classified children as fat if their percentage of body fat was at least 25% and 30%, respectively, for males and females. The Center for Disease Control and Prevention defined overweight as at or above the 95th percentile of BMI for age and "at risk for overweight" as between 85th to 95th percentile of BMI for age [16,17]. European researchers classified overweight as at or above 85th percentile and obesity as at or above 95th percentile of BMI [18].
There are also several methods to measure the percentage of body fat. In research, techniques include underwater weighing (densitometry), multi-frequency bioelectrical impedance analysis (BIA) and magnetic resonance imaging (MRI). In the clinical environment, techniques such as body mass index (BMI), waist circumference, and skin fold thickness have been used extensively. Although, these methods are less accurate than research methods, they are satisfactory to identify risk. While BMI seems appropriate for differentiating adults, it may not be as useful in children because of their changing body shape as they progress through normal growth. In addition, BMI fails to distinguish between fat and fat-free mass (muscle and bone) and may exaggerate obesity in large muscular children. Furthermore, maturation pattern differs between genders and different ethnic groups. Studies that used BMI to identify overweight and obese children based on percentage of body fat have found high specificity (95–100%), but low sensitivity (36–66%) for this system of classification [19]. While health consequences of obesity are related to excess fatness, the ideal method of classification should be based on direct measurement of fatness. Although methods such as densitometry can be used in research practice, they are not feasible for clinical settings. For large population-based studies and clinical situations, bioelectrical impedance analysis (BIA) is widely used. Cross-sectional studies have shown that BIA predicts total body water (TBW), fat-free mass (FFM), and fat mass or percentage of body fat (%BF) among children [20-23]. Also, it has been shown that BIA provides accurate estimation of changes on %BF and FFM over time [24]. Waist circumference, as a surrogate marker of visceral obesity, has been added to refine the measure of obesity related risks [25]. Waist circumference seems to be more accurate for children because it targets central obesity, which is a risk factor for type II diabetes and coronary heart disease. To the best of our knowledge there is no publication on specific cut off points for waist circumference, but there are some ongoing studies.
Causes of obesity
Although the mechanism of obesity development is not fully understood, it is confirmed that obesity occurs when energy intake exceeds energy expenditure. There are multiple etiologies for this imbalance, hence, and the rising prevalence of obesity cannot be addressed by a single etiology. Genetic factors influence the susceptibility of a given child to an obesity-conducive environment. However, environmental factors, lifestyle preferences, and cultural environment seem to play major roles in the rising prevalence of obesity worldwide [26-29]. In a small number of cases, childhood obesity is due to genes such as leptin deficiency or medical causes such as hypothyroidism and growth hormone deficiency or side effects due to drugs (e.g. – steroids) [30]. Most of the time, however, personal lifestyle choices and cultural environment significantly influence obesity.
Behavioral and social factors
I. Diet
Over the last decades, food has become more affordable to larger numbers of people as the price of food has decreased substantially relative to income and the concept of 'food' has changed from a means of nourishment to a marker of lifestyle and a source of pleasure. Clearly, increases in physical activity are not likely to offset an energy rich, poor nutritive diet. It takes between 1–2 hours of extremely vigorous activity to counteract a single large-sized (i.e., >=785 kcal) children's meal at a fast food restaurant. Frequent consumption of such a diet can hardly be counteracted by the average child or adult [31].
Calorie intake
although overweight and obesity are mostly assumed to be results of increase in caloric intake, there is not enough supporting evidence for such phenomenon. Food frequency methods measure usual diet, but estimate caloric intake poorly [32]. Other methods such as 24-hour recall or food diaries evaluate caloric intakes more accurately, however, estimate short-term not long-term intake [32]. Total energy intake is difficult to measure accurately at a population level. However, a small caloric imbalance (within the margin of error of estimation methods) is sufficient over a long period of time to lead to obesity. With concurrent rise in childhood obesity prevalence in the USA, the National Health and Nutrition Examination Survey (NHANES) noted only subtle change in calorie intake among US children from the 1970s to 1988–1994. For this period, NHANES III found an increase calorie intake only among white and black adolescent females. The same pattern was observed by the latest NHANES (1999–2000). The Bogalusa study which has been following the health and nutrition of children since 1973 in Bogalusa (Louisiana), reported that total calorie intake of 10-year old children remained unchanged during 1973–1988 and a slight but significant decrease was observed when energy intake was expressed per kilogram body weight [33]. The result of a survey carried out during the past few decades in the UK suggested that average energy intakes, for all age groups, are lower than they used to be [34]. Some small studies also found similar energy intake among obese children and their lean counterparts [6,35-37].
Fat intake
while for many years it has been claimed that the increase in pediatric obesity has happened because of an increase in high fat intake, contradictory results have been obtained by cross-sectional and longitudinal studies. Result of NHANES has shown that fat consumption of American children has fallen over the last three decades. For instance; mean dietary fat consumption in males aged 12–19 years fell from 37.0% (SD = 0.29%) of total caloric intake in 1971–1974 to 32.0% (SD = 0.42%) in 1999–2000. The pattern was the same for females, whose fat consumption fell from 36.7% (SD = 0.27%) to 32.1% (SD = 0.61%) [38,39]. Gregory et al. [40] reported that the average fat intake of children aged 4–18 years in the UK is close to the government recommendation of 35% energy. On the other hand, some cross-sectional studies have found a positive relationship between fat intake and adiposity in children even after controlling for confounding factors [41,42]. The main objection to the notion that dietary fat is responsible for the accelerated pediatric obesity epidemic is the fact that at the same time the prevalence of childhood obesity was increasing, the consumption of dietary fat in different populations was decreasing. Although fat eaten in excess leads to obesity, there is not strong enough evidence that fat intake is the chief reason for the ascending trend of childhood obesity.
Other dietary factors
there is a growing body of evidence suggesting that increasing dairy intake by about two servings per day could reduce the risk of overweight by up to 70% [43]. In addition, calcium intake was associated with 21% reduced risk of development of insulin resistance among overweight younger adults and may reduce diabetes risk [44]. Higher calcium intake and more dairy servings per day were associated with reduced adiposity in children studied longitudinally [45,46]. There are few data reporting the relation between calcium or dairy intake and obesity among children.
Between 1970 and 1997, the United State Department of Agriculture (USDA) surveys indicated an increase of 118% of per capita consumption of carbonated drinks, and a decline of 23% for beverage milk [47]. Soft drink intake has been associated with the epidemic of obesity [48] and type II diabetes [49] among children. While it is possible that drinking soda instead of milk would result in higher intake of total energy, it cannot be concluded definitively that sugar containing soft drinks promote weight gain because they displace dairy products.
II. Physical Activity
It has been hypothesized that a steady decline in physical activity among all age groups has heavily contributed to rising rates of obesity all around the world. Physical activity strongly influenced weight gain in a study of monozygotic twins [50]. Numerous studies have shown that sedentary behaviors like watching television and playing computer games are associated with increased prevalence of obesity [51,52]. Furthermore, parents report that they prefer having their children watch television at home rather than play outside unattended because parents are then able to complete their chores while keeping an eye on their children [53]. In addition, increased proportions of children who are being driven to school and low participation rates in sports and physical education, particularly among adolescent girls [51], are also associated with increased obesity prevalence. Since both parental and children's choices fashion these behaviors, it is not surprising that overweight children tend to have overweight parents and are themselves more likely to grow into overweight adults than normal weight children [54]. In response to the significant impact that the cultural environment of a child has on his/her daily choices, promoting a more active lifestyle has wide ranging health benefits and minimal risk, making it a promising public health recommendation.
Prevention
Almost all public health researchers and clinicians agree that prevention could be the key strategy for controlling the current epidemic of obesity [55]. Prevention may include primary prevention of overweight or obesity itself, secondary prevention or avoidance of weight regains following weight loss, and prevention of further weight increases in obese individuals unable to lose weight. Until now, most approaches have focused on changing the behavior of individuals on diet and exercise and it seems that these strategies have had little impact on the growing increase of the obesity epidemic.
What age group is the priority for starting prevention?
Children are often considered the priority population for intervention strategies because, firstly, weight loss in adulthood is difficult and there are a greater number of potential interventions for children than for adults. Schools are a natural setting for influencing the food and physical activity environments of children. Other settings such as preschool institutions and after-school care services will have similar opportunities for action. Secondly, it is difficult to reduce excessive weight in adults once it becomes established. Therefore it would be more sensible to initiate prevention and treatment of obesity during childhood. Prevention may be achieved through a variety of interventions targeting built environment, physical activity and diet.
Built Environment
The challenge ahead is to identify obesogenic environments and influence them so that healthier choices are more available, easier to access, and widely promoted to a large proportion of the community (Table 2). The neighborhood is a key setting that can be used for intervention. It encompasses the walking network (footpaths and trails, etc.), the cycling network (roads and cycle paths), public open spaces (parks) and recreation facilities (recreation centers, etc.). While increasing the amount of public open space might be difficult within an existing built environment, protecting the loss of such spaces requires strong support within the community. Although the local environment, both school and the wider community, plays an important role in shaping children's physical activity, the smaller scale of the home environment is also very important in relation to shaping children's eating behaviors and physical activity patterns. Surprisingly, we know very little about specific home influences and as a setting, it is difficult to influence because of the total numbers and heterogeneity of homes and the limited options for access [56]. Of all aspects of behavior in the home environment, however, television viewing has been researched in greatest detail [57-59].
Table 2 Some interventions strategies that could be considered for prevention of childhood obesity
I. Built environment
1. Walking network
a. Footpaths (designated safe walking path)
b. Trails (increasing safety in trails)
2. The cycling network
a. Roads (designated cycling routes)
b. Cycle paths
3. Public open spaces (parks)
4. Recreation facilities (providing safe and inexpensive recreation centers)
II. Physical activity
1. Increasing sports participation
2. Improving and increasing physical education time
3. Use school report cards to make the parents aware of their children's weight problem
4. Enhancing active modes of transport to and from school
a. Walking e.g. walking bus
b. Cycling
c. Public transport
III. TV watching
1. Restricting television viewing
2. Reducing eating in front of the television
3. Ban or restriction on television advertising to children
IV. Food sector
1. Applying a small tax on high-volume foods of low nutritional value (e.g. soft drinks, confectionery, and snack foods)
2. Food labeling and nutrition 'signposts' (e.g. logos for nutritious foods)
3. Implementing standards for product formulation
Physical activity
Stone et al. [60] reviewed the impact of 14 school-based interventions on physical activity knowledge and behavior. Most of the outcome variables showed significant improvements for the intervention. One interdisciplinary intervention program in the USA featured a curriculum-based approach to influence eating patterns, reduce sedentary behaviors (with a strong emphasis on television viewing), and promote higher activity levels among children of school grades 6 to 8. Evaluation at two years showed a reduction in obesity prevalence in girls (OR = 0.47; 95%CI: 0.24 – 0.93), but not in boys (OR = 0.85; 95%CI: 0.52 – 1.39) compared to controls. The reduction in television viewing (by approximately 30 min/day) was highly significant for both boys and girls. Increases in sports participation and/or physical education time would need policy-based changes at both school and education sector levels [61]. Similarly, increases in active modes of transport to and from school (walking, cycling, and public transport) would require policy changes at the school and local government levels, as well as support from parents and the community. In some communities a variety of such programs have been implemented e.g. road crossings, 'walking bus', and designated safe walking and cycling routes [51].
Effects of dietary pattern and TV watching
It appears that gains can be made in obesity prevention through restricting television viewing. Although, it seems that reduced eating in front of the television is at least as important as increasing activity [58]. Fast foods are one of the most advertised products on television and children are often the targeted market. Reducing the huge volume of marketing of energy-dense foods and drinks and fast-food restaurants to young children, particularly through the powerful media of television, is a potential strategy that has been advocated. Television advertising to children under 12 years of age has not been permitted in Sweden since commercial television began over a decade ago, although children's television programs from other countries, and through satellite television, probably dilute the impact of the ban in Sweden. Norway, Denmark, Austria, Ireland, Australia, and Greece also have some restrictions on television advertising to young children [51]. The fact that children would still be seeing some television advertisements during adult programs or other types of marketing, such as billboards, does not contradict the rationale for the control on the television watching of young children.
Food Sector
Food prices have a marked influence on food-buying behaviour and, consequently, on nutrient intake [62]. A small tax (but large enough to affect sales) on high-volume foods of low nutritional value, such as soft drinks, confectionery, and snack foods, may discourage their use. Such taxes currently applied in some parts of the USA and Canada [63]. In addition, food labeling and nutrition 'signposts' such as logos that indicate that a food meets certain nutrition standards might help consumers make choices of healthy foods. An example is the 'Pick the Tick' symbol program run by the National Heart Foundations in Australia and New Zealand [64]. The 'Pick the Tick' symbols made it easier for consumers to identify healthier food choices and are frequently used by shoppers. In addition, the nutrition criteria for the products serve as 'de facto' standards for product formulation, and many manufacturers will formulate or reformulate products to meet those standards.
Effectiveness of the prevention methods
It has been shown that focusing on reducing sedentary behaviour and encouraging free play has been more effective than focusing on forced exercise or reducing food intake in preventing already obese children from gaining more weight [65]. Recent efforts in preventing obesity include the initiative of using school report cards to make the parents aware of their children's weight problem. Health report cards are believed to aid prevention of obesity. In a study in the Boston area, parents who received health and fitness report cards were almost twice as likely to know or acknowledge that their child was actually overweight than those parents who did not get a report card [66]. They also were over twice as likely to plan weight-control activities for their overweight children.
A summary of prevention and intervention strategies is presented in Table 2.
Conclusion
Obesity is a chronic disorder that has multiple causes. Overweight and obesity in childhood have significant impact on both physical and psychological health. In addition, psychological disorders such as depression occur with increased frequency in obese children. Overweight children are more likely to have cardiovascular and digestive diseases in adulthood as compared with those who are lean. It is believed that both over-consumption of calories and reduced physical activity are mainly involved in childhood obesity.
Apparently, primary or secondary prevention could be the key plan for controlling the current epidemic of obesity and these strategies seem to be more effective in children than in adults. A number of potential effective plans can be implemented to target built environment, physical activity, and diet. These strategies can be initiated at home and in preschool institutions, schools or after-school care services as natural setting for influencing the diet and physical activity and at home and work for adults. Both groups can benefit from an appropriate built environment. However, further research needs to examine the most effective strategies of intervention, prevention, and treatment of obesity. These strategies should be culture specific, ethnical, and consider the socio-economical aspects of the targeting population.
Abbreviations
NHANES National Health and Nutrition Examination Survey
MONICA Multinational Monitoring of trends and determinants in cardiovascular disease
BF Body Fat
BMI Body Mass Index
BIA Bioelectrical Impedance Analysis
MRI Magnetic Resonance Imaging
TBW Total Body Water
FFM Fat-Free Mass
USDA United State Department of Agriculture
Authors' contributions
All authors had equal contribution in writing this manuscript.
Acknowledgements
We would like to thank Claire Vayalumkal for her helpful comments and careful reading of the final manuscript.
==== Refs
Nicklas TA T. B K.W. C G. B Eating Patterns, Dietary Quality and Obesity Journal of the American College of Nutrition 2001 20 599 608 11771675
Parsons TJ Power C Logan S Summerbell CD Childhood predictors of adult obesity: a systematic review International Journal of Obesity 1999 23 S1 S107 10641588
Whitaker RC Wright JA Pepe MS Seidel KD Dietz WH Predicting obesity in young adulthood from childhood and parental obesity New England Journal of Medicine 1997 337 869 873 9302300 10.1056/NEJM199709253371301
Livingstone MB Childhood obesity in Europe: a growing concern Public Health Nutr 2001 4 109 116 11255500
James PT Obesity: The worldwide epidemic Clinics in Dermatology 2004 22 276 280 15475226 10.1016/j.clindermatol.2004.01.010
Kelishadi R Pour MH Sarraf-Zadegan N Sadry GH Ansari R Alikhassy H Bashardoust N Obesity and associated modifiable environmental factors in Iranian adolescents: Isfahan Healthy Heart Program - Heart Health Promotion from Childhood Pediatr Int 2003 45 435 442 12911481 10.1046/j.1442-200X.2003.01738.x
AlNuaim AR Bamgboye EA AlHerbish A The pattern of growth and obesity in Saudi Arabian male school children International Journal of Obesity 1996 20 1000 1005 8923156
McCarthy HD Ellis SM Cole TJ Central overweight and obesity in British youth aged 11-16 years: cross sectional surveys of waist circumference BMJ 2003 326 624 12649234 10.1136/bmj.326.7390.624
Ruxton CH Reilly JJ Kirk TR Body composition of healthy 7-and 8-year-old children and a comparison with the 'reference child' International Journal of Obesity 1999 23 1276 1281 10643684 10.1038/sj.ijo.0801067
Daniels SR Arnett DK Eckel RH Gidding SS Hayman LL Kumanyika S Robinson TN Scott BJ St Jeor S Williams CL Overweight in children and adolescents: pathophysiology, consequences, prevention, and treatment Circulation 2005 111 1999 2012 15837955 10.1161/01.CIR.0000161369.71722.10
Mossberg HO 40-Year Follow-Up of Overweight Children Lancet 1989 2 491 493 2570196 10.1016/S0140-6736(89)92098-9
Must A Jacques PF Dallal GE Bajema CJ Dietz WH Long-Term Morbidity and Mortality of Overweight Adolescents - A Follow-Up of the Harvard Growth Study of 1922 to 1935 New England Journal of Medicine 1992 327 1350 1355 1406836
Flegal KM Carroll MD Ogden CL Johnson CL Prevalence and trends in obesity among US adults, 1999-2000 JAMA 2002 288 1723 1727 12365955 10.1001/jama.288.14.1723
Kuczmarski RJ Flegal KM Criteria for definition of overweight in transition: background and recommendations for the United States Am J Clin Nutr 2000 72 1074 1081 11063431
Williams DP Going SB Lohman TG Harsha DW Srinivasan SR Webber LS Berenson GS Body Fatness and Risk for Elevated Blood-Pressure, Total Cholesterol, and Serum-Lipoprotein Ratios in Children and Adolescents American Journal of Public Health 1992 82 358 363 1536350
Flegal KM Wei R Ogden C Weight-for-stature compared with body mass index-for-age growth charts for the United States from the Centers for Disease Control and Prevention American Journal of Clinical Nutrition 2002 75 761 766 11916765
Himes JH Dietz WH Guidelines for Overweight in Adolescent Preventive Services - Recommendations from An Expert Committee American Journal of Clinical Nutrition 1994 59 307 316 8310979
Flodmark CE Lissau I Moreno LA Pietrobelli A Widhalm K New insights into the field of children and adolescents' obesity: the European perspective (vol 28, pg 1189, 2004) International Journal of Obesity 2004 28
Lazarus R Baur L Webb K Blyth F Body mass index in screening for adiposity in children and adolescents: systematic evaluation using receiver operating characteristic curves Am J Clin Nutr 1996 63 500 506 8599312
Danford LC Schoeller DA Kushner RF Comparison of two bioelectrical impedance analysis models for total body water measurement in children Ann Hum Biol 1992 19 603 607 1476416
Deurenberg P van der KK Paling A Withagen P Assessment of body composition in 8-11 year old children by bioelectrical impedance Eur J Clin Nutr 1989 43 623 629 2606093
Deurenberg P Kusters CS Smit HE Assessment of body composition by bioelectrical impedance in children and young adults is strongly age-dependent Eur J Clin Nutr 1990 44 261 268 2364915
Deurenberg P Pieters JJ Hautvast JG The assessment of the body fat percentage by skinfold thickness measurements in childhood and young adolescence Br J Nutr 1990 63 293 303 2334665 10.1079/BJN19900116
Phillips SM Bandini LG Compton DV Naumova EN Must A A longitudinal comparison of body composition by total body water and bioelectrical impedance in adolescent girls Journal of Nutrition 2003 133 1419 1425 12730432
Stevens J Obesity, fat patterning and cardiovascular risk Adv Exp Med Biol 1995 369 21 27 7598009
Hill JO Peters JC Environmental contributions to the obesity epidemic Science 1998 280 1371 1374 9603719 10.1126/science.280.5368.1371
Goodrick GK Poston WS Foreyt JP Methods for voluntary weight loss and control: update 1996 Nutrition 1996 12 672 676 8936489 10.1016/S0899-9007(96)00243-2
Eckel RH Krauss RM American Heart Association call to action: obesity as a major risk factor for coronary heart disease. AHA Nutrition Committee Circulation 1998 97 2099 2100 9626167
Grundy SM Multifactorial causation of obesity: implications for prevention Am J Clin Nutr 1998 67 563S 572S 9497171
Link K Moell C Garwicz S Cavallin-Stahl E Bjork J Thilen U Ahren B Erfurth EM Growth hormone deficiency predicts cardiovascular risk in young adults treated for acute lymphoblastic leukemia in childhood J Clin Endocrinol Metab 2004 89 5003 5012 15472198 10.1210/jc.2004-0126
Styne DM Obesity in childhood: what's activity got to do with it? American Journal of Clinical Nutrition 2005 81 337 338 15699218
Willett W Food Frequency Methods Nutritional Epidemiology 1998 5 2 Oxford University Press 74
Nicklas TA Dietary Studies of Children - the Bogalusa Heart-Study Experience Journal of the American Dietetic Association 1995 95 1127 1133 7560684 10.1016/S0002-8223(95)00305-3
Prentice AM Jebb SA Obesity in Britain - Gluttony Or Sloth British Medical Journal 1995 311 437 439 7640595
Bellisle F Rolland-Cachera MF Deheeger M Guilloud-Bataille M Obesity and food intake in children: evidence for a role of metabolic and/or behavioral daily rhythms Appetite 1988 11 111 118 3239962
Griffiths M Payne PR Energy expenditure in small children of obese and non-obese parents Nature 1976 260 698 700 1264240 10.1038/260698a0
Maffeis C Zaffanello M Pinelli L Schutz Y Total energy expenditure and patterns of activity in 8-10-year-old obese and nonobese children J Pediatr Gastroenterol Nutr 1996 23 256 261 8890075 10.1097/00005176-199610000-00009
Troiano RP Briefel RR Carroll MD Bialostosky K Energy and fat intakes of children and adolescents in the united states: data from the national health and nutrition examination surveys Am J Clin Nutr 2000 72 1343S 1353S 11063476
Wright JD Kennedy-Stephenson J Wang CY McDowell MA Johnson CL Trends in intake of energy and macronutrients - United States, 1971-2000 (Reprinted from MMWR, vol 53, pg 80-82, 2004) Journal of the American Medical Association 2004 291 1193 1194 10.1001/jama.291.18.2194
Gregory JW Lowe S National Diet and Nutrition Survery: Young People Aged 4 to 18 Years : Report of the Diet and Nutrition Survey 2000 London, The Stationery Office.
Maffeis C Pinelli L Schutz Y Fat intake and adiposity in 8 to 11 year-old obese children International Journal of Obesity 1996 20 170 174 8646254
Tucker LA Seljaas GT Hager RL Body fat percentage of children varies according to their diet composition Journal of the American Dietetic Association 1997 97 981 986 9284875 10.1016/S0002-8223(97)00237-X
Heaney RP Davies KM Barger-Lux MJ Calcium and weight: clinical studies J Am Coll Nutr 2002 21 152S 155S 11999544
Pereira MA Jacobs DRJ Van Horn L Slattery ML Kartashov AI Ludwig DS Dairy consumption, obesity, and the insulin resistance syndrome in young adults: the CARDIA Study JAMA 2002 287 2081 2089 11966382 10.1001/jama.287.16.2081
Carruth BR Skinner JD The role of dietary calcium and other nutrients in moderating body fat in preschool children Int J Obes Relat Metab Disord 2001 25 559 566 11319662 10.1038/sj.ijo.0801562
Skinner JD Bounds W Carruth BR Ziegler P Longitudinal calcium intake is negatively related to children's body fat indexes J Am Diet Assoc 2003 103 1626 1631 14647089 10.1016/j.jada.2003.09.018
Putnam JJ Allshouse JE Food consumption, prices, and expenditures, 1970-97 1999 Washington,D.C., Food and Consumers Economics Division, Economic Research Service, US Department of Agriculture
Ludwig DS Peterson KE Gortmaker SL Relation between consumption of sugar-sweetened drinks and childhood obesity: a prospective, observational analysis Lancet 2001 357 505 508 11229668 10.1016/S0140-6736(00)04041-1
Gittelsohn J Wolever TM Harris SB Harris-Giraldo R Hanley AJ Zinman B Specific patterns of food consumption and preparation are associated with diabetes and obesity in a Native Canadian community J Nutr 1998 128 541 547 9482761
Heitmann BL Kaprio J Harris JR Rissanen A Korkeila M Koskenvuo M Are genetic determinants of weight gain modified by leisure-time physical activity? A prospective study of Finnish twins American Journal of Clinical Nutrition 1997 66 672 678 9280191
Swinburn B Egger G Preventive strategies against weight gain and obesity Obes Rev 2002 3 289 301 12458974 10.1046/j.1467-789X.2002.00082.x
Tremblay MS Willms JD Is the Canadian childhood obesity epidemic related to physical inactivity? Int J Obes Relat Metab Disord 2003 27 1100 1105 12917717 10.1038/sj.ijo.0802376
Gordon-Larsen P Griffiths P Bentley ME Ward DS Kelsey K Shields K Ammerman A Barriers to physical activity: qualitative data on caregiver-daughter perceptions and practices Am J Prev Med 2004 27 218 223 15450634 10.1016/j.amepre.2004.05.002
Carriere G Parent and child factors associated with youth obesity. 2003 Statistics Canada
Muller MJ Mast M Asbeck I Langnase K Grund A Prevention of obesity--is it possible? Obes Rev 2001 2 15 28 12119633 10.1046/j.1467-789x.2001.00012.x
Campbell K Crawford D Jackson M Cashel K Worsley A Gibbons K Birch LL Family food environments of 5-6-year-old-children: does socioeconomic status make a difference? Asia Pac J Clin Nutr 2002 11 Suppl 3 S553 S561 12492647 10.1046/j.0964-7058.2002.00346.x
Gortmaker SL Peterson K Wiecha J Sobol AM Dixit S Fox MK Laird N Reducing obesity via a school-based interdisciplinary intervention among youth: Planet Health Arch Pediatr Adolesc Med 1999 153 409 418 10201726
Robinson TN Reducing children's television viewing to prevent obesity: a randomized controlled trial JAMA 1999 282 1561 1567 10546696 10.1001/jama.282.16.1561
Dietz WH Gortmaker SL Preventing obesity in children and adolescents Annu Rev Public Health 2001 22 337 353 11274525 10.1146/annurev.publhealth.22.1.337
Stone EJ McKenzie TL Welk GJ Booth ML Effects of physical activity interventions in youth. Review and synthesis Am J Prev Med 1998 15 298 315 9838974 10.1016/S0749-3797(98)00082-8
Dwyer T Coonan WE Leitch DR Hetzel BS Baghurst RA An investigation of the effects of daily physical activity on the health of primary school students in South Australia Int J Epidemiol 1983 12 308 313 6629620
Guo X Popkin BM Mroz TA Zhai F Food price policy can favorably alter macronutrient intake in China J Nutr 1999 129 994 1001 10222391
Jacobson MF Brownell KD Small taxes on soft drinks and snack foods to promote health Am J Public Health 2000 90 854 857 10846500
Young L Swinburn B Impact of the Pick the Tick food information programme on the salt content of food in New Zealand Health Promot Int 2002 17 13 19 11847134 10.1093/heapro/17.1.13
Caterson ID Gill TP Obesity: epidemiology and possible prevention Best Pract Res Clin Endocrinol Metab 2002 16 595 610 12468409 10.1053/beem.2002.0228
Chomitz VR Collins J Kim J Kramer E McGowan R Promoting healthy weight among elementary school children via a health report card approach Archives of Pediatrics & Adolescent Medicine 2003 157 765 772 12912782 10.1001/archpedi.157.8.765
Freedman DS Srinivasan SR Valdez RA Williamson DF Berenson GS Secular increases in relative weight and adiposity among children over two decades: the Bogalusa Heart Study Pediatrics 1997 99 420 426 9041299 10.1542/peds.99.3.420
Zametkin AJ Zoon CK Klein HW Munson S Psychiatric aspects of child and adolescent obesity: a review of the past 10 years J Am Acad Child Adolesc Psychiatry 2004 43 134 150 14726719 10.1097/00004583-200402000-00008
Kotani K Nishida M Yamashita S Funahashi T Fujioka S Tokunaga K Ishikawa K Tarui S Matsuzawa Y Two decades of annual medical examinations in Japanese obese children: do obese children grow into obese adults? Int J Obes Relat Metab Disord 1997 21 912 921 9347410 10.1038/sj.ijo.0800492
Lobstein TJ James WP Cole TJ Increasing levels of excess weight among children in England Int J Obes Relat Metab Disord 2003 27 1136 1138 12917722 10.1038/sj.ijo.0802324
Moreno LA Sarria A Popkin BM The nutrition transition in Spain: a European Mediterranean country Eur J Clin Nutr 2002 56 992 1003 12373620 10.1038/sj.ejcn.1601414
Rolland-Cachera MF Deheeger M Thibault H [Epidemiologic bases of obesity] Arch Pediatr 2001 8 Suppl 2 287s 289s 11394090 10.1016/S0929-693X(01)80048-8
GE K T T C T T K Prevalence and trends in overweight and obesity among children and adolescents in Thessaloniki, Greece J Pediatr Endocrinol Metab 2005 14 1319 1365
|
16138930
|
PMC1208949
|
CC BY
|
2021-01-04 16:39:31
|
no
|
Nutr J. 2005 Sep 2; 4:24
|
utf-8
|
Nutr J
| 2,005 |
10.1186/1475-2891-4-24
|
oa_comm
|
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-191609823110.1186/1743-7075-2-19ResearchEffects of resistance training and protein supplementation on bone turnover in young adult women Mullins Nicole M [email protected] Wayne E [email protected] Department of Human Performance and Exercise Science, Youngstown State University, Youngstown, OH 44555, USA2 Exercise Science Laboratory, School of Exercise Leisure and Sport, Kent State University, Kent, OH 44242, USA2005 17 8 2005 2 19 19 11 8 2005 17 8 2005 Copyright © 2005 Mullins and Sinning; licensee BioMed Central Ltd.2005Mullins and Sinning; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The strength of aging bone depends on the balance between the resorption and formation phases of the remodeling process. The purpose of this study was to examine the interaction of two factors with the potential to exert opposing influences on bone turnover, resistance exercise training and high dietary protein intake. It was hypothesized that resistance training by young, healthy, untrained women with protein intakes near recommended levels (0.8 g·kg-1·d-1) would promote bone formation and/or inhibit bone resorption, and that subsequent supplementation to provide 2.4 g protein·kg-1·d-1 would reverse these effects.
Methods
Bone formation was assessed with serum bone-specific alkaline phosphatase (BAP) and osteocalcin (OC), and bone resorption with urinary calcium and deoxypyridinoline (DPD). Biochemical, strength, anthropometric, dietary, and physical activity data were obtained from 24 healthy, untrained, eumenorrheic women (18–29y) at baseline, after eight weeks of resistance training (3 d·wk-1, ~1 hr·d-1; 3 sets, 6–10 repetitions, 13 exercises, 75–85% maximum voluntary contraction), and after 12 weeks of resistance training and 10 days of protein/placebo supplementation. Subjects were randomized (double-blind) to either a high protein (HP) or training control (TC) group and, during the final 10 days, consumed either enough purified whey protein to bring daily protein intake to 2.4 g·kg-1·d-1, or an equivalent dose of isoenergetic, carbohydrate placebo.
Results
Strength, lean tissue mass, and DPD increased significantly in both groups over time, while percent body fat and BAP decreased (repeated measures ANOVA, p ≤ 0.05, Bonferroni correction). No significant changes were observed for serum OC or urinary calcium, and no significant group (TC, HP) × time (baseline, week 8, week 12) interactions emerged for any of the biochemical measures.
Conclusion
(1) Twelve weeks of high-intensity resistance training did not appear to enhance bone formation or inhibit bone resorption in young adult women, as assessed by biochemical markers of bone metabolism. (2) Subsequent maintenance of a high protein intake for 10 days in these regularly-training, calcium-replete women also showed no effects on bone metabolism.
==== Body
Background
Maximizing genetic bone mass potential during youth, and minimizing bone loss throughout adulthood constitute osteoporosis prevention at its most fundamental level. To these ends, a great many factors in the pathogenesis of osteoporosis must be identified, assessed, and, as much as possible, offset. Certainly some prominent risk factors, such as age, sex, race, and family history are unalterable. Others, however, including diet, physical inactivity, menstrual status, medication use, low sunlight exposure, alcohol consumption, and cigarette smoking are amenable to change, and are therefore particularly important targets for osteoporosis prevention.
The purpose of this study was to examine the interaction of two lifestyle factors with the potential to exert opposing influences on bone metabolism: resistance training and high dietary protein intake. Some research indicates that resistance training has osteogenic, or bone-building effects, and some, that the metabolism of high protein loads has osteopenic, or bone-weakening effects. However, while these notions are commonly discussed as though they were well-established facts, the majority of research into both areas has come from cross-sectional or methodologically-limited experimental studies. Indeed, research has confirmed both the beneficial effects of weight-bearing activity [1] and the deleterious effects of a lack of it [2,3], but relatively few studies have specifically investigated the capacity of resistance training to increase bone mineral density (BMD) and mass. Among the few that have, findings have been inconsistent and clouded by a number of noteworthy limitations, including a lack of randomization [4-9] and the use of highly variable training intensities {≤ 10 repetitions and/or ≥ 70–85% one repetition maximum (1RM) [4,10-12]; ≥ 10 repetitions and/or ≤ 70% 1RM [5,6,9]}, subject populations (young, healthy subjects [4,5,8,11,12]; older individuals [6,7,9,10,13]; patients [14]), and study durations. Furthermore, because most have lasted fewer than 12 months, precision error is an important limitation in those that used bone densitometry. Since the annual rate of BMD change generally lies within the precision error of bone densitometry [15], it is difficult to accurately assess BMD gains or losses within fewer than several years.
Regarding dietary protein and bone strength, several cross-sectional studies have supported an inverse relationship [16-18], while others have not [19-21]. Experimental trials have also generated mixed findings [22,23], which are further complicated by the use of diverse subject populations, methods of assessing skeletal responses, experimental controls (e.g., calcium supplementation, menstrual status, etc.), and types, levels, and durations of protein supplementation. Still, despite these limitations, some authors have plainly labeled dietary protein a "negative risk factor" [[24], p. 336].
There are several theoretical mechanisms through which high protein intakes might stimulate bone demineralization, but the major possibility centers on the skeleton's constitution of an alkali reserve that can be called upon to assist in buffering the acidic catabolites of high-protein foods [25]. A complete discussion of this mechanism is beyond the scope of this article, but the body is known to mobilize calcium and phosphorus salts in the presence of high acid loads, and experimental increases in protein intake have been shown to stimulate bone resorption and calcium excretion [22,25-27]. Thus, the core concern is whether individuals who maintain chronically high protein intakes over-rely on skeletal buffers and incur preventable losses of bone mineral.
Investigation into possible interactive effects of resistance training and dietary protein on bone seems warranted because of the tendency of some people to regularly consume extraordinary amounts of protein. Traditionally, strength and bodybuilder athletes have been the most likely to follow high-protein diets; now many in the general population are following suit, due largely to the popularity of the various high-protein, weight-loss diets. Regardless of the rationale, it is known that some people habitually consume from three to more than five times [28-34] the adult recommended dietary allowance (RDA) for protein, despite clear evidence that such quantities constitute a nutritional overload [35,36]. An overload of dietary protein is generally indicated by increased rates of amino acid oxidation and urea nitrogen excretion. When protein is consumed in excess of the body's needs to synthesize new and replace degraded tissue proteins, it is not simply stored as such, but is broken down into its constituent amino acids. The surplus amino carbon fragments are then either oxidized for energy, or converted to carbohydrate or fat and stored, while the amino nitrogen fragments are excreted, mainly as urea in the urine.
Resistance exercise is known to stimulate muscle protein synthesis [37-40], which can thereby raise individual protein needs above the RDA of 0.8 grams per kilogram of body weight per day (g·kg-1·d-1) [41], to as much as much as 1.6 to 1.8 g·kg-1·d [42-45]. However, because protein synthesis eventually plateaus despite increasing dietary protein supply [43-45], no evidence supports a need for the extreme levels consumed by some athletes and dieters. It must be noted that these data assume adequate overall energy intake, as it has long been known that hypoenergetic diets increase protein requirements [46].
Thus, this double-blind, randomized trial was designed to compare the effects of a normal (i.e., near RDA) versus a high protein intake on bone metabolism in young adult women participating in regular, high-intensity resistance training. The high intake was set at 2.4 g·kg-1·d-1 for several reasons: (1) several previous studies of protein intake and calcium balance have used values between 2.0 and 2.7 g·kg-1·d-1 [22,23,26,47,48]; (2) several studies of strength and bodybuilder athletes have reported habitual protein intakes up to 4.3 g·kg-1·d-1 [31-34]; (3) many people adhering to such plans as the Atkins Diet [28] and the Zone Diet [29] will consume well over 2.0 g·kg-1·d-1; and (4) it the first multiple of the RDA after 1.6 g·kg-1·d, which, as indicated, may be needed by some people [42-45]. A secondary objective was to examine bone turnover responses to the initiation of regular resistance training in untrained women with some remaining potential to add to skeletal mass (i.e., < 30 y) [49]. We hypothesized that the initiation of high-intensity resistance training in untrained, young adult women with protein intakes near recommended levels would stimulate osteogenic (i.e., increased bone formation) and/or anti-resorptive (i.e., reduced bone resorption) effects, and that a subsequent increase in protein intake would trigger osteopenic effects (i.e., increased bone resorption and/or reduced bone formation).
Methods
Experimental approach
The study was a two-group, 12-week, randomized trial. The first 10 1/2 weeks served as a control period, during which all of the subjects followed the same relative resistance training program, maintained their usual diets, and consumed a daily calcium supplement. Biochemical, strength, anthropometric, dietary, and physical activity data were collected at three time points: baseline (prior to both resistance training and protein/placebo supplementation), week 8 (after initial adaptation to resistance training, but prior to supplementation), and week 12 (after continued resistance training and 10 days of supplementation). This timeline was planned largely to control for menses-related variability in the biochemical measures and to fit within a semester at the university. Considering that both protein utilization [50] and bone biomarker concentrations [51] may vary throughout the menstrual cycle, blood and urine sampling at each time point was scheduled during the early follicular phase (days 1–8). This essentially left data collection opportunities for each woman at approximately weeks 4, 8, and 12. Week 4 was not used, to allow an initial period of adaptation to training to pass before applying the experimental treatment (i.e., protein supplement/placebo). While resistance training could initiate detectable changes in the biomarkers before week 8, the aim of the study was to investigate effects of mechanical loading and protein supplementation on the markers, not effects of the novel exercise. Since previous research has shown that the early stages of resistance training may increase dietary protein needs [34,42,45], the consumption of supplemental protein soon after the initiation of an exercise program might not induce as great an acid effect. Week 16 was not considered, because semester-end schedule changes and relocations were likely to prevent the continued participation of many subjects.
Subjects
Thirty healthy, eumenorrheic women between 18 and 30 years of age were recruited from a general university population to form a training control (TC) group and a high protein (HP) group. Exclusion criteria included having fewer than 10 normal menses during the previous year, engaging in resistance training within two years, using any drugs or medications known to affect bone metabolism within six months, having any known metabolic bone disease or chronic physical condition that would limit participation in an exercise program, and regularly smoking more than 10 cigarettes or drinking more than two alcoholic beverages per day. These criteria were established on the basis of research documenting their potential to affect bone [52] and their inclusion in related studies [11,12]. Race was not restricted because a repeated measures design was used, and because there is no evidence suggesting that women with different racial backgrounds would respond differently to training. The study was approved by the Kent State University Institutional Review Board, and informed consent was obtained from all subjects prior to data collection.
Data collection protocol
Of the more than 150 women who responded to the initial request for study volunteers, the first 30 to satisfy the eligibility criteria and agree to all procedures and responsibilities were accepted. Each subject reported to the Exercise Science Laboratory four times. During the initial visit, each completed the health history and physical activity questionnaires and underwent an anthropometric assessment. Preconditions for the anthropometry were explained via electronic mail or telephone prior to the initial visit, including appropriate attire and avoidance of any substances or activities that could significantly alter hydration status (specifically, heavy exercise for 12 hours, caffeine for 24 hours, alcohol for 48 hours). Each subject then received detailed instructions on keeping accurate seven-day diet records and collecting 24-hour urine specimens, and scheduled a second laboratory visit during her next anticipated follicular phase (visits were rescheduled if menses started earlier or later than anticipated). Finally, to exclude the potential effects of calcium deficiency, each subject was given a supply of calcium supplements (Tums Calcium for Life™-Bone Health, SmithKline Beecham, Pittsburgh, PA) to begin consuming for the duration of the study. Each was instructed to carry the calcium tablets in her purse or backpack, and was regularly reminded to consume one 500-mg tablet, twice per day.
For each subsequent laboratory visit, subjects reported between 6:00 and 9:00 a.m. in the overnight fasted state to control for circadian, dietary, and exercise influences on the biochemical measures. Of these, circadian variability appears to have the greatest impact, with most serum and urinary biomarker levels being unaffected by diet and only slightly affected by prior exercise [53]. Diet records and urine specimens were collected and a venous blood sample drawn for the determination of serum OC, BAP, E2, and P4. The serum was immediately separated from the blood cells, transferred into separate Cryovials® (Nalgene, Rochester, NY) for each biomarker, and stored at -120°C. Total urine volume was measured and several aliquots frozen at -120°C (DPD) and -20°C (calcium, creatinine, pH, urea nitrogen).
Questionnaires
Questionnaires were administered to assess general health, menstrual, and physical activity histories and to monitor ongoing health status and physical activity levels. The health questionnaire was a standard form of the Exercise Science Laboratory, supplemented with questions on menstrual history and eating behaviors. The physical activity questionnaire (PAQ) was based on the Seven-Day Activity Recall of Sallis et al. [54], and used to determine whether the subjects' participation in resistance training affected their involvement in other types of physical activity.
Anthropometry
Skinfold thicknesses (mm) were measured with Holtain® calipers (Holtain Ltd., Wales, UK) at the subscapular, triceps, chest, midaxillary, anterior suprailiac, abdominal, and mid-thigh sites, and used to estimate body composition. Percent body fat (%BF), fat mass (FM), and fat-free mass (FFM) were derived using the Jackson et al. [55] seven-site skinfold equation for adult women and the body density formula of Brozek et al. [56]. Girths were obtained with a fiberglass measuring tape at the neck, shoulders, chest, waist, umbilicus, hips, gluteal thigh, mid-thigh, calf, mid-arm, and forearm to evaluate the effects of resistance training on muscle mass. All values were obtained in triplicate according to Lohman et al. [57], by one experienced anthropometrist (N. Mullins). Median values were used as data.
Diet analyses
All subjects completed seven-day diet records prior to laboratory visits 2–4. Baseline and week 8 records were used to assess habitual dietary intake, consistency, and any between-groups differences. Week 12 records were used to confirm the subjects' continued maintenance of their usual diets. Estimates of average daily intake were computed for total kilocalories (kcals), percent kcals (%kcals) from protein, carbohydrate, fat, and alcohol, grams of protein per kilogram of body weight (g·kg-1), and calcium (mg), phosphorus (mg), vitamin D (μg), sodium (mg), magnesium (mg), caffeine (mg), and fiber (g) using diet analysis software (Diet Analysis Plus©, Version 4.0, ESHA Research, Salem, OR).
Biochemical measures
Serum osteocalcin (OC) and serum bone-specific alkaline phosphatase (BAP) were used as indicators of bone formation, and urinary deoxypyridinoline (DPD) and urinary calcium as indicators of bone resorption. Detailed discussions of the individual markers are available in several reviews [53,58-60]. The bone markers and urinary creatinine (correction for urinary DPD) were measured at all three time points, and serum estradiol (E2) and progesterone (P4) at baseline to confirm menstrual status.
Serum OC (Metra™ Osteocalcin), serum BAP (Metra™ BAP), and urinary DPD (Metra™ DPD) were measured with enzyme-linked immunosorbant assay kits (Quidel Corporation, Santa Clara, CA), and serum E2 (Coat-A-Count® Estradiol) and P4 (Coat-A-Count® Progesterone) via radioimmunassay (Diagnostics Products Corporation, Los Angeles, CA). Urinary calcium and creatinine were measured using automated procedures at Suburban Medical Laboratory (Cuyahoga Falls, OH). To minimize interassay variation, all serum and urine samples remained in frozen storage until all data were collected and assays could be run simultaneously. Duplicate values for all biochemical measures were obtained and mean values used as data.
Strength testing and training
Between the first and second laboratory visits, each subject was familiarized with the resistance training facility (in the same building as the laboratory) and given individual instruction on training safety and techniques. Each completed two familiarization workouts under one-on-one supervision, after which baseline strength testing was scheduled (at least 48 hours later). Strength was evaluated using 1RM procedures for select exercises (Table 1), and also using isometric dynamometry, to provide strength scores independent of the training program.
Table 1 Exercises used for resistance training and strength testing
MUSCLE GROUP EXERCISES MUSCLE GROUP EXERCISES
Back Lat pull-downa
Bent-over dumbbell row
Seated row
Underhand pull-down Biceps Seated dumbbell curla
Concentration curl
Preacher curl
Hammer curl
Quadriceps Leg extensiona
Seated leg press
Dumbbell squats
Dumbbell lunges Triceps Triceps pressdowna
Triceps kickback
Overhead triceps extension
Triceps dips
Chest Chest pressa
Dumbbell flyes
Dumbbell bench press
Dumbbell pullover Calves Standing calf raisea
Seated plantar flexion
Single-leg standing calf raise
Hamstrings Lying hamstring curla
Deadlifts Upper Body Push-upsb
Assisted chin-ups Assisted dips
Shoulders Overhead pressa
Lateral raises
Front raises
Shrugs Lower Back Machine back extension
Roman chair
Hip Standing hip abductiona
Seated hip abduction
Standing hip extension Abdominals Basic crunches
Reverse crunches
Elbow crunches
Machine crunches
Hanging knee-ups Oblique knee-ups
Hip Standing hip adductiona
Seated hip adduction
Standing hip flexion
All underlined exercises were used during weeks 1, 2, 5, 8, 11, 12, with alternative exercises used during the other weeks to provide variety and enhance adherence.
aExercises used in 1-RM testing
b Tested for maximum number of repetitions
Peak isometric flexion and extension force of the right arm and leg were determined using an electronic dynamometer, designed and built by Karpovich and Karpovich [61] and interfaced with a Biopak® computer system and AcqKnowledge® software (Biopak® Systems Inc, Santa Barbara, CA). Voltage readings from the dynamometer were converted to force units (kg) by the software, and peak force values easily identified from force curves. Subject instruction, encouragement and positioning, including posture, joint angle and limb stabilization, were closely controlled. The subjects' limbs were positioned at the joint angles shown to permit maximum force production: 115° for elbow flexion, 40° for elbow extension, 165° for knee flexion, and 115° for leg extension [62]. Three trials for each movement were performed, and the median values used as data.
Once baseline strength and all other baseline measures were obtained (lab visit 2), each subject began her 12-week program of high-intensity resistance training targeting the major muscle groups (3 d·wk-1, ~1 hr·d-1, 3 sets of 6–10 repetitions for 13 different exercises at 75–85% of 1-RM values; Table 1). All training sessions were fully supervised (subject:instructor ratios between 1:1 and 4:1), and subjects' training loads continually adjusted to maintain proper intensity levels.
Protein/placebo supplementation
During week 10, all subjects were randomized (double-blind) to either the TC or HP group. HP subjects were provided with a 10-day supply of purified whey protein (Extreme Pure Pro™, American Body Building Products®, Waterboro, SC), and TC subjects with a 10-day supply of carbohydrate placebo (Maltrin® M100 Maltodextrin, Grain Processing Corporation, Muscatine, IA). Treatment dosages were measured, packaged, and distributed by a single, third party technician. Daily rations were prepared to bring each HP subject's average protein intake to 2.4 g·kg-1·d-1, or to supply each TC subject with an equivalent dose of placebo. They were provided in three to five small containers, with the recommendation to consume one drink with each daily meal, and any additional supplement ad libitum. Since the protein supplement was a fruit punch flavored concentrate and the placebo an unflavored powder, the placebo was mixed with enough fruit drink powder (carbohydrate only) to make it similar in taste, appearance, and energy content to the protein supplement. The subjects had only to add water to prepare either substance. Each subject consumed the assigned daily ration during the final 10 days of her training program, and reported for week 12 data collection immediately afterwards (day 11).
Statistical Analyses
Descriptive statistics were computed for all data and select health questionnaire responses were evaluated in terms of frequency and percent frequency. All data were checked for normality and nonparametric tests were used where appropriate. Data were analyzed for the main effects of group and time and the interaction of group by time with a two (group: HP, TC) × three (test period: baseline, week 8, week 12) repeated-measures analysis of variance (ANOVA). Group differences in diet composition and exercise adherence were examined using Mann-Whitney tests, and Wilcoxon Signed Rank tests were used to test within-groups consistency in dietary intake over time. Significance was set at the p ≤ 0.05 level for all comparisons, and the Bonferroni correction used where multiple comparisons were made. Data were analyzed using SPSS for Windows, Version 9.0 (SPSS Inc., Chicago, IL, USA).
Results
Subject Characteristics and Exercise Adherence
Six volunteers dropped out within the first week of the study (three due to schedule conflicts and three due to unrelated health concerns), leaving a total of 24 subjects who completed the study (12 TC, 12 HP). Baseline subject characteristics for the TC and HP groups are shown in Table 2. There were no initial, between-groups differences in age, height, weight, %BF, FFM, FM, or serum E2 or P4. All subjects had baseline serum E2 and P4 levels within normal follicular phase ranges (i.e., E2 10–200 pg·ml-1; P4 0.15–1.4 ng·ml-1; Diagnostics Products Corporation, Los Angeles, CA) and maintained menstrual function throughout the study. The same number of women in each group used oral contraceptives (n = 7). Attendance at the resistance training sessions was excellent and similar for both groups (TC 92%, HP 94%, p = 0.86). On average, subjects in both groups missed only three of approximately 36 total training sessions. The total possible number of training sessions slightly exceeded 36 (36.6), as it was important to maintain the trained state while scheduling final data collection around slight menstrual cycle delays.
Table 2 Baseline subject characteristics (mean ± SE) of the training control (TC) and high-protein (HP) groups
VARIABLE TC (n = 12) HP (n = 12)
Age (yr) 22.7 ± 1.07 22.8 ± 0.85
Height (cm) 164.9 ± 1.84 167.1 ± 1.75
Weight (kg) 64.8 ± 2.31 64.2 ± 2.48
Body fat (%) 27.9 ± 1.70 26.0 ± 1.96
FFM (kg) 46.4 ± 1.00 47.1 ± 1.25
FM (kg) 18.4 ± 1.68 17.1 ± 1.81
* Significant (p ≤ 0.05) difference, independent samples t-tests
Dietary Intake
Both groups maintained similar and consistent diets over time, with no significant differences between groups at any time point, and no significant changes over time in any of the assessed dietary variables. Baseline and week 8 dietary data were pooled to provide the basis for calculating supplement dosages (Table 3).
Table 3 Mean† daily nutrient intake (mean ± SE) for the training control (TC) and high-protein (HP) groups
VARIABLE TC (n = 12) HP (n = 12)
Energy (kcals) 1632 ± 141 1792 ± 103
Protein (g·kg-1) 0.96 ± 0.05 0.91 ± 0.07
Protein (% kcals) 15.4 ± 1.0 12.9 ± 0.5
Carbohydrate (% kcals) 56.2 ± 2.2 55.9 ± 1.6
Fat (% kcals) 27.9 ± 1.5 27.2 ± 1.4
Alcohol (% kcals) 1.1 ± 0.5 3.2 ± 1.0
Calcium (mg) 685 ± 84 682 ± 64
Phosphorus (mg) 873 ± 75 876 ± 56
Vitamin D (μg) 3.37 ± 0.53 2.50 ± 0.38
Magnesium (mg) 182 ± 17 209 ± 13
Fiber (g) 12.4 ± 4.5 15.1 ± 5.9
Sodium (mg) 2428 ± 211 2781 ± 158
Caffeine (mg) 56 ± 16 46 ± 16
* Significant (p ≤ 0.05) difference, Mann-Whitney tests
† 14-day means
Physical Activity Questionnaire Data
No significant group by time interactions emerged from the PAQ data (not shown). Logically, there was a significant main effect for time (p ≤ 0.001) in the number of weekly hours spent performing strength exercise, but time spent engaged in all other types and levels of physical activity did not change significantly (moderate, hard or very hard physical activity, flexibility exercise, sleeping). There was a significant group effect in moderate physical activity (p = 0.02), such that the TC group reportedly engaged in more moderate-intensity physical activity than the HP group, but at all time points and not as a result of the resistance training or dietary protein interventions.
Strength and Body Composition
Large increases in voluntary strength (Figure 1) and changes in body composition (Figure 2) confirm the effectiveness of the training program. Since all strength and body composition measures were initially similar between groups and progressed similarly throughout the study, these data are presented as total sample means. Percent increases in strength over the course of the 12-week program ranged from 26–143% for upper body measures, and from 25–83% for lower body measures. Strength improvements were significant between all time points for all strength measures, except for those of isometric arm and leg flexion, which did not increase significantly between weeks 8 and 12. With regard to body composition, both groups showed similar, significant reductions in percent body fat (TC -6.5%, HP -7.3%, p ≤ 0.001) and gains in lean tissue mass (TC 2.7%, HP 3.9%, p ≤ 0.001).
Figure 1 Strength changes over time (mean ± SE) for the training control (TC) and high-protein (HP) groups combined (n = 24). Absolute strength was similar between TC and HP at all time points, and strength increases were significant (p ≤ 0.05) between all time points for all measures, except for those of isometric arm and leg flexion, which were not significantly different between weeks 8 and 12 (NS). † One-repetition (1-RM) maximum (kg) ‡ Isometric dynamometry (kg)
Figure 2 Anthropometric and body composition measures (mean ± SE) for the training control (TC) and high-protein (HP) groups combined (n = 24). All measures were similar between TC and HP at all time points, with both groups showing similar, significant (p ≤ 0.05) reductions in percent body fat and gains in lean tissue mass over time. ab Like letters are significantly different.
Biochemical Measures
The effects of resistance training and protein intake on the biochemical measures are shown in Figures 3, 4, 5, 6. Significant time effects were observed for serum BAP and urinary DPD in both groups, such that BAP was significantly lower at week 12 than at baseline (p ≤ 0.001), while DPD was significantly higher at week 12 than at both baseline and week 8 (p ≤ 0.001). A significant group effect emerged for urinary calcium, with the HP group showing greater calcium excretion than the TC group (p = 0.02), but at all time points and not as a result of training or supplementation. Serum OC showed neither time, nor group effects, and no significant group by time interactions emerged for any of the biochemical measures.
Figure 3 Serum bone-specific alkaline phosphatase (BAP) concentrations (mean ± SE) for the training control (TC) and high-protein (HP) groups from baseline through week 12. For both groups, BAP levels were significantly (p ≤ 0.05) lower at week 12 than at baseline. * Time point values are significantly different.
Figure 4 Serum osteocalcin (OC) concentrations (mean ± SE) for the training control (TC) and high-protein (HP) groups from baseline through week 12. Differences were not significant (p ≤ 0.05) for group, for time, or for group × time effects.
Figure 5 Urinary calcium concentrations (mean ± SE) for the training control (TC) and high-protein (HP) groups from baseline through week 12. Calcium levels were significantly (p ≤ 0.05) greater in TC than HP at all time points. * Group values are significantly different.
Figure 6 Urinary deoxypyridinoline (DPD) concentrations (mean ± SE) for the training control (TC) and high-protein (HP) groups from baseline through week 12. For both groups, DPD levels were significantly (p ≤ 0.05) higher at week 12 than at both baseline and week 8. ab Like letters are significantly different.
Discussion
The strength of aging bone depends on the balance between the resorption and formation phases of the remodeling process. To our knowledge, this is the first study to examine the interaction of two lifestyle factors with the potential to exert opposing influences on bone turnover, resistance exercise training and high dietary protein intake. It was hypothesized that high-intensity resistance training would increase bone formation and reduce bone resorption activity, as indicated by biochemical markers of bone metabolism, and that subsequent consumption of a high protein intake would reverse these effects. The loading intensity was at least as great as that in previous studies which reported significant osteogenic responses to resistance exercise [10-12], and the protein intake was at least as great as that in previous studies which reported significant protein-induced calciuric effects [47,48,63].
The results, however, conflicted with several previous findings. Progressive resistance training did not prompt an increase in formation marker concentrations, with serum BAP declining and serum OC showing no change in both groups over time. In contrast, Fujimura et al. [4] reported significantly elevated BAP and OC levels in eight healthy, young men (23–31y) one month after initiating resistance training – levels that remained elevated through four months of training and that were not seen in seven age-matched, non-exercising controls. Menkes et al. [6] reported significant increases in BAP and OC after 12 and 16 weeks of resistance training, respectively, in 11 healthy, untrained, older men (59 ± 2y), but not in seven non-training controls. Rockwell et al. [8] and Nelson et al. [10] reported significantly elevated OC levels after initiating resistance training in eumenorrheic, premenopausal (mean ± SE, 36.2 ± 1.3 yr) and postmenopausal (50–70 y) women, respectively. While the exercise protocols and subject samples varied, these studies all supported the possibility that resistance training might stimulate osteogenic effects, detectable through changes in bone formation marker concentrations.
Though explanations for the discrepant findings are not clear, for several reasons, we are confident that neither exercise adherence, nor training intensity was a confounding factor. First, the subjects' high overall attendance at the training sessions (93%) and their significant strength and body composition changes throughout the study (Figures 1 and 2) support the effectiveness of the exercise program. Second, the training program was consistent with recently published recommendations by the American College of Sports Medicine for preserving bone health throughout adulthood [64]. Third, all workouts were fully supervised, resistances carefully monitored and adjusted, and personalized encouragement given to help each subject consistently train at target levels. If resistance training can stimulate osteogenic responses in young adult women, the present program should have provided an adequate stimulus to do so.
The few studies that have evaluated the effects of resistance training on both biomarker concentrations and BMD were particularly important to forming the hypothesis that high-intensity resistance training would stimulate osteogenic shifts in the bone metabolism of previously untrained subjects. An overview of the majority of these studies is presented in Table 4[4-11,13,65]. However, while importantly contributing to the literature, these studies are complicated by several limitations, including nonrandomized designs, short study durations, low-intensity training protocols, small and varied subject samples (age, sex, previous training), inadequate controls over important hormonal and nutritional factors (estrogen, calcium), and the use of different bone assessment techniques. In fact, six of the 10 studies cited allowed subjects to self-select membership to exercise or control groups – all six of which reported either BMD or biomarker changes that could be interpreted as beneficial effects of resistance training. Since there may be inherent differences that motivate some individuals to choose exercise over control participation, these results must be viewed cautiously.
Table 4 Longitudinal studies of resistance training (RT) effects on bone mineral density (BMD1) and biochemical markers of bone metabolism
Bemben et al., 2000 Subjects: 25 postmenopausal women (17 RT1, 8 C), 41–60 y
Eligibility criteria: No RT previous 6 months; no current or recent HRT
Design: Randomized (subjects matched for lumbar BMD, then randomly assigned to groups)
Training duration / frequency: 6 months / 3 d·wk-1 (~45 min·d-1)
Training intensity: 3 sets, 8–16 reps, 40–80% 1-RM, 12 exercises
Calcium supplementation: 600 mg·d-1+125 IU vit. D only for subjects with daily Ca2+ <1500 mg
Training supervised: Yes
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: Total body, lumbar, & femoral BMD changes – NS
Results-Biomarkers: OC & CTX changes – NS
Fujimura et al., 1997 Subjects: 15 adult men (8 RT, 7 C), 23–31 y
Eligibility criteria: No RT previous 2 y
Design: Not randomized
Training duration / frequency: 4 months / 4 d·wk-1 (~45 min·d-1)
Training intensity: 1 set, 10 reps, 60% 1-RM, & 2 sets, 10 reps, 80% 1-RM, 7–8 exercises
Calcium supplementation: 600 mg·d-1
Training supervised: Not indicated Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: Total body, lumbar, femoral, radial BMD changes – NS
Results-Biomarkers: * OC & BAP ↑ at 1, 2, 3, & 4 months in RT, but not in C
* PICP ↓ at 2 & 4 months in C, but not in RT (not measured at 3 months in C) DPD changes – NS
Gleeson et al., 1990 Subjects: 72 eumenorrheic women (34 RT, 38 C), 23–46 y
Eligibility criteria: No previous, regular RT; oral contraceptive users eligible
Design: Not randomized
Training duration / frequency: 12 months / 3 d·wk-1 (~30 min·d-1)
Training intensity: 2 sets, 20 reps, 60% 1-RM, 8 exercises
Calcium supplementation: 500 mg·d-1
Training supervised: Not indicated
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Percent lumbar BMD change (0.8% ↑ in RT vs. 0.5% ↓ in C)
Absolute lumbar (DPA) & calcaneal (SPA) BMD changes – NS
Results-Biomarkers: OC changes – NS
Lohman et al., 1995 Subjects: 56 eumenorrheic women (22 RT, 34 C), 28–39 y
Eligibility criteria: No RT previous 2 y; oral contraceptives users not eligible
Design: Randomized
Training duration / frequency: 18 months / 3 d·wk-1 (~1 hr·d-1)
Training intensity: 3 sets, 8–12 reps, 70–80% 1-RM, 12 exercises
Calcium supplementation: 500 mg·d-1
Training supervised: Yes
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Lumbar BMD ↑ at 5 (2.0%) & 12 (1.6%) months in RT, but not in C
* Trochanteric BMD ↑ at 5 (0.8%), 12 (2.4%), & 18 (1.5%) months in RT, but not in C Total body & radial BMD (SPA) – NS
Results-Biomarkers: * OC ↑ in RT at 5, 12, & 18 months, but not in C
Menkes et al., 1993 Subjects: 18 men (11 RT, 7 C), 50–70 y
Eligibility criteria: No RT previous 2 y
Design: Not randomized
Training duration / frequency: 4 months / 3 d·wk-1
Training intensity: 1 set, 15 reps, 8 upper body exercises, & 2 sets, 15 reps, 4 lower body exercises
Calcium supplementation: No: subjects instructed to follow diet containing 1000 mg·d-1 calcium
Training supervised: Yes
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Lumbar (2.0%) & femoral (3.8%) BMD ↑ in RT, but not in C
Total body BMD changes – NS
Results-Biomarkers: * OC ↑ at 3 & 4 months, & BAP ↑ at 4 months in RT, but not in C
TRAP changes – NS
Nelson et al., 1994 Subjects: 39 postmenopausal women (20 RT, 19 C), 50–70 y
Eligibility criteria: No current regular exercise; no HRT previous 12 months
Design: Randomized
Training duration / frequency: 12 months / 2 d·wk-1 (~45 min·d-1)
Training intensity: 3 sets, 8 reps, 80% 1-RM, 5 exercises
Calcium supplementation: No: subjects consuming < 800 mg·d-1 counseled to ↑ intake
Training supervised: Yes
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Lumbar (1.0%) & femoral (0.9%) BMD ↑ in RT, but ↓ in C (lumb. -1.8, fem. -2.5%)
Total body BMD changes – NS
Results-Biomarkers: * OC ↑ in RT, but ↓ in C
Pruitt et al., 1992 Subjects: 26 postmenopausal women (17 RT, 9 C), RT 53.6 ± 1.0 y, C 55.6 ± 0.9 y
Eligibility criteria: No RT previous 6 months; no previous HRT
Design: Not randomized (early respondents placed in RT group)
Training duration / frequency: 9 months / 3 d·wk-1 (~40 min·d-1)
Training intensity: 1 set, 10–15 reps, 11 exercises
Calcium supplementation: Not indicated
Training supervised: Not indicated
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Lumbar BMD (DPA) ↑ in RT (1.6%), but ↓ in C (-3.6%)
Femoral (DPA) & forearm (SPA) BMD changes – NS
Results-Biomarkers: * Baseline OC greater in RT than in C
OC, BAP, & HYP changes – NS
Pruitt et al., 1995 Subjects: 26 postmenopausal women (15 RT2, 11 C), 65–79 y
Eligibility criteria: No previous RT; no previous HRT or HRT ≥ 1 y
Design: Randomized
Training duration / frequency: 12 months / 3 d·wk-1 (~50–55 min·d-1)
Training intensity: 3 sets, 7–14 reps, 40–80% 1-RM, 10 exercises
Calcium supplementation: 500 mg·d-1+200 IU vitamin D
Training supervised: Yes
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: Lumbar & hip BMD changes – NS
Results-Biomarkers: OC changes – NS
Rockwell et al., 1990 Subjects: 17 eumenorrheic women (10 RT, 7 C), RT 36.2 ± 1.3 y, C 40.4 ± 1.6y
Eligibility criteria: No previous RT; oral contraceptive users not eligible
Design: Not randomized
Training duration / frequency: 9 months / 2 d·wk-1 (~45 min·d-1)
Training intensity: 2 sets, 12 reps, 70% 1-RM, 8 exercises
Calcium supplementation: 500 mg·d-1 +200 IU vitamin D
Training supervised: Not indicated
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Lumbar BMD ↓ in RT at 4.5 (-2.9%) & 9 months (-4.0%); no change in C
Femoral BMD changes – NS
Results-Biomarkers: * Baseline OC greater in RT than in C
* OC ↑ in RT & in C at 4.5 & 9 months
Ryan et al., 1994 Subjects: 37 men (21 RT, 16 C), 51–71 y
Eligibility criteria: No RT previous 6 months Design: Not randomized
Training duration / frequency: 4 months / 3 d·wk-1
Training intensity: 1 set, 15 reps using variable resistance machines, 14 exercises
Calcium supplementation: Not indicated
Training supervised: Not indicated
Results-Strength: * Strength ↑ in RT, but not in C
Results-BMD: * Femoral BMD ↑ in RT (2.8%), but not in C
Total body & lumbar BMD (DXA) changes – NS
Results-Biomarkers: * TRAP ↑ in RT & in C (no difference between groups)
Changes in OC & BAP – NS
BMD1 = all BMD measurements made using dual-energy x-ray absorptiometry (DXA) unless otherwise specified (DPA = dual-photon absorptiometry; SPA = single-photon absorptiometry)
RT = resistance training
C = control
RT1 = two training groups: 10 high-load RT & 7 high-repetition RT
HRT = hormone replacement therapy
* = statistically significant
NS = non-significant
OC = serum osteocalcin
CTX = urinary C-telopeptide
BAP = serum bone-specific alkaline phosphatase
DPD = urinary deoxypyridinoline
PICP = plasma carboxy-terminal pro-collagen 1 extension peptide
TRAP = serum tartrate resistant acid phosphatase
HYP = urinary hydroxyproline
RT2 = two training groups: 8 high-intensity RT & 7 low-intensity RT
Particularly notable among the first studies to use both biomarker technology and densitometry to examine resistance training effects is that of Lohman et al. [11], due to its inclusion of important controls for hormonal status, training intensity, and calcium intake. Lohman's group reported significantly elevated serum OC in 22 young adult women (28–39 y) after five, 12, and 18 months of high-intensity resistance training, but not in 34 age-matched, non-training controls. Since the present study used both a similar subject sample and training program (Lohman et al.: 1 h·d-1, 3 d·wk-1, 3 sets, 8–12 repetitions, 12 exercises, 70–80% 1RM; present study: ~1 h·d-1, 3 d·wk-1, 3 sets, 6–10 repetitions, 13 exercises, 75–85% 1RM), similar results were anticipated. The present study was shorter, but did produce significant strength and body composition changes and was markedly longer than other studies in which biomarker responses to experimental manipulations have been observed [22,23,66].
Even fewer studies have specifically addressed the possibility that resistance training may benefit bone by reducing bone resorption. Fujimura et al. [4], cited earlier, reported not only significant elevations in BAP and OC, but also non-significant, yet noticeable reductions in urinary DPD during the first three of four months of resistance training in healthy young men (45 minutes, 4 d·wk-1, 3 sets of 10 repetitions at 60–80% 1-RM for 7–8 exercises). Ashizawa et al. [67], examining more acute effects of resistance exercise, observed reductions in two resorption markers [urinary DPD and serum tartrate-resistant alkaline phosphatase (TRAP)] within two hours of an intense resistance training workout. Reductions in these markers reached significance (p ≤ 0.05) on the third and first post-exercise days, respectively. Much more research is needed to investigate the potential for resistance exercise to reduce bone resorption, as well as to determine whether acute shifts in blood and urine chemistry, as in the Ashizawa study, are meaningful in terms of overall bone health.
The present data also did not support the hypothesis that the consumption of 2.4 g protein·kg-1·d-1 by women with habitual intakes near the RDA for protein (0.9 g protein·kg-1·d-1; Table 3) would increase bone resorption or reduce bone formation. As noted previously, one formation marker (serum BAP) did decline with supplementation, but in both groups, while the other (serum OC) showed no change. Regarding bone resorption, it was hypothesized that urinary calcium and DPD would rise in the HP, but not the TC group. Instead, urinary calcium showed no change with supplementation (Figure 5), while DPD levels increased nearly 50% between weeks 8 and 12 in both groups (Figure 6). Ryan et al. [9] reported a similarly unexpected rise in a resorption marker after 16 weeks of resistance training in older men (61 ± 1 y, n = 21), but the use of a different biomarker (serum TRAP) and an older subject sample complicates the ability to draw comparisons.
The lack of a calciuric response by the HP group is perhaps the most puzzling observation, as the protein dose was comparable to that in previous studies that did precipitate such an effect – an effect that has been shown to be consistent, rapid, and sustained [68,69]. It is possible that the phosphorus content of the supplement and the subjects' habitual diet attenuated the hypothesized effects, as phosphorus has known hypocalciuric effects [68,70]. A few studies have demonstrated blunted calciuric effects when using meat with a substantial phosphorus content, rather than purified isolates, as the source of supplemental protein [71,72]. The present study used a commercially-available form of purified whey protein, but the supplement did contain a noteworthy amount of phosphorus. An exact value was not available, but an estimate provided by Weider International® indicated that the powder contained about 56 g of phosphorus for every 100 g of protein (personal communication, February 2002). In addition, the subjects' dietary phosphorus intake, exclusive of supplementation (Table 3), was 125% of the RDA for females 19 to 24 years of age [41].
It is also possible that calcium excretion did increase, but through the gut and not the kidney – an increase that would go undetected without measures of endogenous fecal calcium. Previous studies have examined fecal calcium after manipulating dietary protein and have not shown significant changes [68,73], but the majority has measured only total fecal calcium, without distinguishing the endogenous fraction. Heaney et al. [74] have recently generated strong support for possibility that significant fecal losses may have previously gone undetected in studies that did not measure the endogenous fraction. Heaney's group studied 191 adult, female inpatients in a metabolic ward, obtained complete calcium intake, absorption, and excretion data, and demonstrated a significant, inverse relationship between urinary and endogenous fecal calcium. For every 0.41 mg drop in urinary calcium, there was an associated 1.0 mg rise in endogenous fecal calcium. Thus, it may, after all, be critical to measure both urinary and fecal calcium excretion to accurately assess calcium balance.
Still another possibility is that the hypothesized effects on bone turnover may have been offset by the subjects' high calcium intake. Calcium supplementation was considered essential to excluding the effects of calcium deficiency, but the high supplemental dosage, in addition to the subjects' dietary calcium, may have provided enough exogenous alkali material to effectively neutralize any protein-induced acid effects. Shapses et al. [23] have supported this possibility, reporting that an increase in dietary calcium initiated a drop in urinary DPD, and that a large increase in dietary protein had no effect on urinary calcium or DPD when dietary calcium was very high (~1600 mg·d-1).
Heaney [75] has simply explained the mechanism through which high calcium intakes may counter the effects of protein excess. Essentially, increased dietary protein stimulates calcium excretion, which in turn stimulates the synthesis and activation of vitamin D to enhance intestinal calcium absorption. If dietary calcium is sufficient, intestinal calcium absorption can be up-regulated and bone resorption is not needed to preserve calcium balance. However, if dietary calcium is inadequate, then an up-regulation of absorption will do nothing to compensate for the increased calcium excretion and bone resorption will ensue.
Finally, it is possible that the relatively uncontrolled timing of the treatment dosages could have influenced the results. While it was recommended that the subjects consume one drink with each daily meal and any additional supplement ad libitum, standardization of the timing of supplementation was not possible with these free-living college students and working women. Future studies may want to examine this potential influence.
Conclusion
The present study provided no evidence that high-intensity resistance training stimulated osteogenic effects, as assessed with serum osteocalcin and bone-specific alkaline phosphatase. It is possible that other biomarkers may have produced different results, and that, given a longer time frame, bone densitometry could detect osteogenic effects. The present study also yielded no evidence that short-term protein supplementation would have osteopenic effects in young adult women. However, the supplementation period was brief and the subjects were healthy, eumenorrheic, calcium-replete women, regularly participating in high-intensity exercise. These characteristics, which may have additive, beneficial effects on bone, are unfortunately not often descriptive of American women, and thus these results must not be taken as justification to perpetuate the common 'more-is-better' mentality toward dietary protein.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NM conceived the study, carried out all described methods, and drafted the manuscript. WS assisted with the design of the study, statistical analyses, interpretation of data, and editing of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
We wish to thank GlaxoSmithKline (Pittsburgh, PA) for supplying the calcium supplement, Weider International® (Salt Lake City, UT) for supplying both the protein supplement and carbohydrate placebo, and the Ohio Association for Health, Physical Education, Recreation and Dance (OAHPERD) and the Kent State University Graduate Student Senate for providing funding for this research. Finally, we thank all of the subjects who participated in this study.
==== Refs
Block JE Smith R Genant HK Overview of exercise and bone mass Osteoporosis Update 1987 1987 Berkeley, CA: University Press 267 70
Bloomfield SA Changes in musculoskeletal structure with prolonged bed rest Med Sci Sports Exerc 1997 29 197 206 9044223
Holick MF Perspective on the impact of weightlessness on calcium and bone metabolism Bone 1998 105 111 10.1016/S8756-3282(98)00014-3
Fujimura R Ashizawa N Watanabe M Mukai N Amagai H Fukubayashi T Hayashi K Tokuyama K Suzuki M Effect of resistance exercise training on bone formation and resorption in young male subjects assessed by biomarkers of bone metabolism J Bone Miner Res 1997 12 656 662 9101378
Gleeson PB Protas EJ LeBlanc AD Schneider VS Evans HJ Effects of weight lifting on bone mineral density in premenopausal women J Bone Miner Res 1990 5 153 158 2316403
Menkes A Mazel S Redmond RA Koffler K Libanti CR Gundberg CM Zizic TM Hagberg JM Pratley RE Hurley BF Strength training increases regional bone mineral density and bone remodeling in middle-aged and older men J Appl Physiol 1993 74 2478 2484 8335581
Nelson ME Fiatarone MA Morganti CM Trice I Greenburg RA Evans WJ Effects of high-intensity strength training on multiple risk factors for osteoporotic fractures: a randomized controlled trial J Am Med Assoc 1994 272 1909 1914 10.1001/jama.272.24.1909
Pruitt LA Jackson RA Bartels RL Lehnhard HL Weight-training effects on bone mineral density in early postmenopausal women J Bone Miner Res 1992 7 179 185 1570762
Rockwell JC Sorensen DM Baker S Leahey D Stock JL Michaels J Baran DT Weight training decreases vertebral bone density in premenopausal women: a prospective study J Clin Endocrinol Metab 1990 71 988 993 2401722
Ryan AS Treuth MS Rubin MA Miller JP Nicklas BJ Landis DM Pratley RE Libanati CR Gundberg CM Hurley BF Effects of strength training on bone mineral density: hormonal and bone turnover relationships J Appl Physiol 1994 77 1678 1684 7836186
Lohman T Going S Pamenter R Hall M Boyden T Houtkooper L Ritenbaugh C Bare L Hill A Aickin M Effects of resistance training on regional and total bone mineral density in premenopausal women: a randomized prospective study J Bone Miner Res 1995 10 1015 1024 7484276
Snow-Harter C Bouxsein M Lewis B Carter D Marcus R Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial J Bone Miner Res 1992 7 761 769 1642145
Bemben DA Fetters NL Bemben MG Navabi N Koh ET Musculoskeletal responses to high- and low-intensity resistance training in early postmenopausal women Med Sci Sports Exerc 2000 32 1949 1957 11079527 10.1097/00005768-200011000-00020
Braith RW Mills RM Welsch MA Keller JW Pollock ML Resistance exercise training restores bone mineral density in heart transplant recipients J Am Coll Cardiol 1996 28 1471 1477 8917260 10.1016/S0735-1097(96)00347-6
National Osteoporosis Foundation Osteoporosis: review of the evidence for prevention, diagnosis, and treatment and cost effective analysis Osteoporos Int 1998 1 88
Abelow BJ Holford TR Insogna KL Cross-sectional association between dietary animal protein and hip fracture Calcif Tiss Int 1992 50 14 18 10.1007/BF00297291
Feskanich D Willett WC Stampfer MJ Colditz GA Protein consumption and bone fractures in women Am J Epidemiol 1996 143 472 479 8610662
Frassetto LA Todd KM Morris RC JrSebastian A Worldwide incidence of hip fracture in elderly women: relation to consumption of animal and vegetable foods J Gerontol A Biol Sci Med Sci 2000 55 M585 M592 11034231
Barr SI Prior JC Janelle KC Lentle BC Spinal bone mineral density in premenopausal vegetarian and nonvegetarian women: cross-sectional and prospective comparisons J Am Diet Assoc 1998 98 760 765 9664916 10.1016/S0002-8223(98)00172-2
Lau EM Kwok T Woo J Ho SC Bone mineral density in Chinese elderly female vegetarians, vegans, lacto-vegetarians and omnivores Europ J Clin Nutr 1998 52 60 64 10.1038/sj.ejcn.1600516
Munger RG Cerhan JR Chiu BC Prospective study of dietary protein intake and risk of hip fracture in postmenopausal women Am J Clin Nutr 1999 69 147 152 9925137
Kerstetter JE Mitnick ME Gundberg CM Caseria DM Ellison AF Carpenter TO Insogna KL Changes in bone turnover in young women consuming different levels of dietary protein J Clin Endocrinol Metab 1999 84 1052 1055 10084594 10.1210/jc.84.3.1052
Shapses SA Robins SP Schwartz EI Chowdhury H Short-term changes in calcium but not protein intake alter the rate of bone resorption in healthy subjects as assessed by urinary pyridinium cross-link excretion J Nutr 1995 125 2814 2821 7472661
Nordin BE Need AG Steurer T Morris HA Chatterton BE Horowitz M Nutrition, osteoporosis, and aging Ann NY Acad Sci 1998 854 336 351 9928442
Barzel US The skeleton as an ion exchange system: implications for the role of acid-base imbalance in the genesis of osteoporosis J Bone Miner Res 1995 10 1431 1436 8686497
Kerstetter JE Allen LH Protein intake and calcium homeostasis Adv Nutr Res 1994 9 167 181 7747665
Barzel US Massey LK Excess dietary protein can adversely affect bone J Nutr 1998 128 1051 1053 9614169
Atkins RC Dr Atkins' New Diet Revolution 2001 New York: Morrow, William and Company
Sears B Lawren B Enter the zone: a dietary road map 1995 New York: Harper Collins
Steen SN Precontest strategies of a male bodybuilder Int J Sport Nutr 1991 1 69 78 1844404
Faber M Benade AJ Nutrient intake and dietary supplementation in bodybuilders S Afr Med J 1987 72 831 834 3424028
Grandjean AC Diets of elite athletes: has the discipline of sports nutrition made an impact? J Nutr 1997 874 877
Short SH Short WR Four-year study of university athletes' dietary intake J Am Diet Assoc 1983 82 632 645 6853938
Tarnopolsky MA MacDougall JD Atkinson SA Influence of protein intake and training status on nitrogen balance and lean body mass J Appl Physiol 1988 64 187 193 3356636
Fern EB Bielinski RN Schutz Y Effects of exaggerated amino acid and protein supply in man Experientia 1991 47 168 172 2001721 10.1007/BF01945420
Lemon PWR Is increased dietary protein necessary or beneficial for individuals with a physically active lifestyle? Nutr Rev 1996 54 S169 S175 8700446
Biolo G Maggi SP Williams BD Tipton KD Wolfe RR Increased rates of muscle protein turnover and amino acid transport after resistance exercise in humans Am J Physiol 1995 268 E514 E520 7900797
Chesley A MacDougall JD Tarnopolsky MA Atkinson SA Smith K Changes in human muscle protein synthesis after resistance exercise J Appl Physiol 1992 73 1383 1388 1280254
MacDougall JD Gibala MJ Tarnopolsky MA MacDonald JR Interisano SA Yarasheski KE The time course for elevated muscle protein synthesis following heavy resistance exercise C Can J Appl Physiol 1995 20 480 486
Phillips SM Tipton KD Aarsland A Wolf SE Wolfe RR Mixed muscle protein synthesis and breakdown after resistance exercise in humans Am J Physiol 1997 273 E99 E107 9252485
Food and Nutrition Board, Institute of Medicine Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids (Macronutrients) 2002 Washington DC: The National Academies Press
Lemon PWR Tarnopolsky MA MacDougall JD Atkinson SA Protein requirements and muscle mass/strength changes during intensive training in novice body builders J Appl Physiol 1992 73 767 775 1400008
Tarnopolsky MA Atkinson SA MacDougall JD Chesley A Philllips S Schwarcz HP Evaluation of protein requirements for trained strength athletes J Appl Physiol 1992 73 1986 1995 1474076
Lemon PW Beyond the zone: protein needs of active individuals J Am Coll Nutr 2000 513 521
Phillips SM Protein requirements in strength sports Nutrition 2004 20 689 695 15212752 10.1016/j.nut.2004.04.009
Munro HN Munro HN, Allison JB Introduction to part II Mammalian Protein Metabolism 1964 2 New York: Academic Press 1 39
Kerstetter JE O'Brien KO Insogna KL Dietary protein affects intestinal calcium absorption Am J Clin Nutr 1998 68 859 865 9771863
Kitano T Esashi T Azami S Effect of protein intake on mineral (calcium magnesium, and phosphorus) balance in Japanese males J Nutr Sci Vitaminol 1988 34 387 398 3236083
Center J Eisman J The epidemiology and pathogenesis of osteoporosis Baillieres Best Pract Res Clin Endocrinol Metab 1997 11 23 62
Calloway DH Kurzer MS Menstrual cycle and protein requirements of women J Nutr 1982 12 356 366 7057271
Zitterman A Schwarz I Scheld K Sudhop T Berthold HK von Bergmann K van der Ven H Stehle P Physiologic fluctuations of serum estradiol levels influence biochemical markers of bone resorption in young women J Clin Endocrin Metab 2000 85 95 101 10.1210/jc.85.1.95
Aurbach GD Marx SJ Spiegal AM Wilson JD, Foster DW Metabolic bone disease Williams' Textbook of Endocrinology 1992 8 Philadelphia: WB Saunders 1477 1510
Delmas PD Eastell R Garnero P Seibel MJ Stepan J The use of biochemical markers of bone turnover in osteoporosis Osteoporos Int 2000 2 17 10.1007/s001980070002
Sallis JF Haskell W Wood PD Fortmann SP Rogers T Blair SN Paffenbarger RS Jr Physical activity assessment methodology in the Five-City Project Am J Epidemiol 1985 121 91 106 3964995
Jackson AS Pollock ML Ward A Generalized equations for predicting body density of women Med Sci Sports Exerc 1980 12 175 182 7402053
Brozek JF Grande F Anderson JT Keys A Densitometric analyses of body composition: revision of some quantitative assumptions Ann NY Acad Sci 1963 110 113 140 14062375
Lohman TG Roche AF Martorell R Anthropometric Standardization Reference Manual 1988 Champaign, IL: Human Kinetics
Calvo MS Eyre DR Gundberg CM Molecular basis and clinical application of biological markers of bone turnover Endoc Rev 1996 17 333 368 10.1210/er.17.4.333
Christenson RH Biochemical markers of bone metabolism: an overview Clin Biochem 1997 30 573 593 9455610 10.1016/S0009-9120(97)00113-6
Seibel MJ Woitge HW Basic principles and clinical applications of biochemical markers of bone metabolism J Clin Densitom 1999 2 299 321 10548826 10.1385/JCD:2:3:299
Singh M Karpovich PV Isotonic and isometric forces of forearm flexors and extensors J Appl Physiol 1966 21 1435 1437 5916693
Clarke HH Clarke DH Developmental and Adapted Physical Education 1963 Englewood Cliffs, NJ: Prentice-Hall
Hegsted M Schuette SA Zemel MB Linkswiler HM Urinary calcium and calcium balance in young men as affected by level of protein and phosphorus intake J Nutr 1981 111 553 562 7205408
Kohrt WM Bloomfield SA Little KD Nelson ME Yingling VR American College of Sports Medicine position stand: physical activity and bone health Med Sci Sports Exerc 2004 36 1985 1996 15514517
Pruitt LA Taaffe DR Marcus R Effects of a one-year high-intensity versus low-intensity resistance training on bone mineral density in older women J Bone Miner Res 1995 10 1788 1795 8592957
Akesson K Lau KH Johnston P Imperio E Baylink DJ Effects of short-term calcium depletion and repletion on biochemical markers of bone turnover in young adult women J Clin Endocrinol Metab 1998 83 1921 1927 9626120 10.1210/jc.83.6.1921
Ashizawa N Ouchi G Fujimara R Yoshida Y Tokuyama K Suzuki M Effects of a single bout of resistance exercise on calcium and bone metabolism in untrained young males Calcif Tiss Int 1998 62 104 108 10.1007/s002239900402
Hegsted M Linkswiler HM Long-term effects of level of protein intake on calcium metabolism in young adult women J Nutr 1981 111 244 251 7463168
Schuette SA Zemel MB Linkswiler HM Studies on the mechanism of protein-induced hypercalciuria in older men and women J Nutr 1980 110 305 315 7354401
Heaney RP Excess dietary protein may not adversely affect bone J Nutr 1998 128 1054 1057 9614170
Moriguti JC Ferriolli E Marchnini JS Urinary calcium loss in elderly men on a vegetable:animal (1:1) high-protein diet Gerontology 1999 45 274 278 10460989 10.1159/000022101
Spencer H Kramer L DeBartolo M Norris C Osis D Further studies of the effect of a high protein diet as meat on calcium metabolism Am J Clin Nutr 1983 37 924 929 6846239
Lutz J Calcium balance and acid-base status of women as affected by increased protein intake and by sodium bicarbonate ingestion Am J Clin Nutr 1984 39 281 288 6320628
Heaney RP Recker RR Ryan RA Urinary calcium in perimenopausal women: normative values Osteoporos Int 1999 9 13 18 10367024 10.1007/s001980050110
Heaney RP Utilization of calcium and its relationship to protein in ensuring bone health 48th Annual Meeting of the American College of Sports Medicine: Baltimore 2001
|
16098231
|
PMC1208950
|
CC BY
|
2021-01-04 16:37:46
|
no
|
Nutr Metab (Lond). 2005 Aug 17; 2:19
|
utf-8
|
Nutr Metab (Lond)
| 2,005 |
10.1186/1743-7075-2-19
|
oa_comm
|
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-201612487510.1186/1743-7075-2-20ResearchEffects of dietary curcumin or N-acetylcysteine on NF-κB activity and contractile performance in ambulatory and unloaded murine soleus Farid Mehran [email protected] Michael B [email protected] Yi-Ping [email protected] Eric [email protected] William J [email protected] Pulmonary Medicine, Baylor College of Medicine, One Baylor Plaza, Houston, TX, 77030, USA2 Department of Physiology, College of Medicine, University of Kentucky, Lexington, KY,40506, USA2005 26 8 2005 2 20 20 21 6 2005 26 8 2005 Copyright © 2005 Farid et al; licensee BioMed Central Ltd.2005Farid et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Unloading of skeletal muscle causes atrophy and loss of contractile function. In part, this response is believed to be mediated by the transcription factor nuclear factor-kappa B (NF-κB). Both curcumin, a component of the spice turmeric, and N-acetylcysteine (NAC), an antioxidant, inhibit activation of NF-κB by inflammatory stimuli, albeit by different mechanisms. In the present study, we tested the hypothesis that dietary curcumin or NAC supplementation would inhibit unloading-induced NF-κB activity in skeletal muscle and thereby protect muscles against loss of mass and function caused by prolonged unloading.
Methods
We used hindlimb suspension to unload the hindlimb muscles of adult mice. Animals had free access to drinking water or drinking water supplemented with 1% NAC and to standard laboratory diet or diet supplemented with 1% curcumin. For 11 days, half the animals in each dietary group were suspended by the tail (unloaded) and half were allowed to ambulate freely.
Results
Unloading caused a 51–53% loss of soleus muscle weight and cross-sectional area relative to freely-ambulating controls. Unloading also decreased total force and force per cross-sectional area developed by soleus. Curcumin supplementation decreased NF-κB activity measured in peripheral tissues of ambulatory mice by gel shift analysis. In unloaded animals, curcumin supplementation did not inhibit NF-κB activity or blunt the loss of muscle mass in soleus. In contrast, NAC prevented the increase in NF-κB activity induced by unloading but did not prevent losses of muscle mass or function.
Conclusion
In conclusion, neither dietary curcumin nor dietary NAC prevents unloading-induced skeletal muscle dysfunction and atrophy, although dietary NAC does prevent unloading induced NF-κB activation.
==== Body
Background
Prolonged exposure to microgravity causes muscle atrophy, resulting in weakness and predisposition to fatigue. Such changes can markedly degrade astronauts' performance, especially upon return to Earth [1]. Atrophy starts as early as 4 days into space flight [2] and becomes more significant in long-term space missions [3]. In addition to atrophy, weight-bearing skeletal muscles lose their contractile force per cross-sectional area [4]. Since atrophy in space flight and disuse atrophy share similar pathophysiologic mechanisms [5], countermeasures to microgravity-induced muscle wasting may also be effective for the prevention of disuse atrophy and muscle weakness in long term bedridden patients. Thus, there is considerable interest in the development of countermeasures that oppose the complications of muscle disuse.
Hindlimb unloading, a commonly used model to study the effects of muscle disuse, activates the transcription factor nuclear factor-κB (NF-κB) in weight-bearing muscles [6,7]. This laboratory has recently reported that NF-κB activation increases ubiquitin conjugation [8], consistent with a role for NF-κB in the regulation of skeletal muscle protein breakdown [9]. In addition, activation of NF-κB in response to TNF-α alters myogenesis [10] and is essential for the loss of protein in TNF-α-stimulated skeletal muscle myotubes [11]. Stimuli that persistently activate NF-κB are associated with muscle protein loss and activation of the ubiquitin-proteasome pathway [12]. In vitro, selective inhibition of NF-κB can prevent loss of muscle protein [11]. In vivo, a recent study found that mice lacking specific NF-κB family members (p105/50 or bcl3) are resistant to unloading-induced atrophy and NF-κB activation [6], supporting a role for NF-κB as a robust in vivo catabolic stimulus during disuse atrophy.
The prevention of unloading-induced atrophy by ablation of genes for specific NF-κB family members [6] suggests that pharmacological inhibition of NF-κB might also reduce atrophy in response to unloading. In the present study, we tested the ability of two NF-κB inhibitors to inhibit disuse-induced skeletal muscle NF-κB activation. Curcumin (diferuloylmethane), a major component of the spice turmeric (Curcuma longa), has been reported to exert in vivo effects on serum lipids [13], aortic atherosclerosis [14], and various forms of cancer [15]. Curcumin also possesses anti-inflammatory [16] and antioxidant [14] properties in vivo and inhibits the canonical NF-κB activation pathway [17]. In the canonical activation pathway, the NF-κB inhibitor IκB is phosphorylated by IκB kinase (IKK), leading to its dissociation from NF-κB and subsequent translocation of NF-κB to the nucleus [6], where it binds to the promoter region of target genes. Curcumin acts at a proximal step in the canonical activation pathway, inhibiting IKK [17] phosphorylation of IκBα. N-acetylcysteine (NAC) is a naturally occurring compound found in several vegetables, including garlic [18], onion [18], peppers [19], and asparagus [19]. In addition to antioxidant, antiangiogenic, and anticarcinogenic properties [20], NAC inhibits NF-κB activation by several mechanisms acting distal to the activity of IKK [21-23].
NF-κB activation in response to muscle disuse has been reported to occur through a pathway distinct from the canonical activation pathway that does not involve IκB activation [6]. In the present study, we tested the hypothesis that curcumin, an inhibitor of the canonical activation pathway, and NAC, acting downstream of curcumin in the NF-κB activation pathway, would inhibit NF-κB and preserve muscle mass and function. In addition, as the in vivo effects of dietary curcumin on NF-κB activity are not well-studied, particularly under basal (non-induced) conditions, we measured NF-κB activity in liver, soleus, lung, and large intestine to test the hypothesis that dietary curcumin inhibits basal NF-κB activity in vivo.
Methods
Animals
Adult male ICR mice were purchased from Harlan (Indianapolis, IN) and housed under pathogen-free conditions. Animals were handled in accordance with the policies and guidelines for the care and use of laboratory animals of the National Institutes of Health. The protocol was approved by the Animal Protocol Review Committee at Baylor College of Medicine.
Food
Mice in control and treatment groups were fed a standard chow diet (Purina-Mills, 5053 diet, Labdiet, St Louis, MO). For the curcumin treatment group, the diet was modified by adding commercial curcumin (LKT laboratories, St. Paul, Minnesota) to 1% (w/w). Curcumin was uniformly incorporated into the food, which was pelleted to ensure uniform feed and curcumin intake. All control and experimental diets were stored at 4°C. All mice had free access to water during the dietary treatment period. For the NAC treatment group, 1% NAC (w/v) was included in the drinking water.
Methods
In the first study (Figure 1), ambulatory mice were studied after two days on either the control diet or the curcumin-supplemented diet, in order to test whether curcumin inhibited NF-κB in vivo. In the second study (Figure 1), the effects of curcumin were studied in three groups of mice. Two groups received the control diet throughout the 13-day study period, whereas the other group received the curcumin diet. After 2 days, the curcumin dietary group and one of the groups receiving the control diet were hindlimb-unloaded. In the two unloaded groups, the tail of each animal was taped to suspension wire using first aid tape in order to elevate the hindlimb of each animal. Mice remained unloaded for the remainder of the study period. Unloaded mice were able to ambulate using their forelegs; however, their hindlimbs were unable to touch the bottom or sides of the cage. In the third study (Figure 1), three groups were studied: one ambulatory group receiving normal drinking water, one hindlimb-unloaded group receiving normal drinking water, and one unloaded group receiving 1% NAC in the drinking water. All mice in the third study received the standard chow diet.
Figure 1 Unloading protocol. In study 1, ambulatory mice were fed either a control or a 1% curcumin diet for 2 days. In study 2, mice were divided into control and curcumin diet groups as in study 1 for the first two days, after which each diet group was further divided into ambulatory or unloaded groups.
Tissue collection
Mice were deeply anesthetized by methoxyflurane inhalation and soleus muscles were excised with their tendons intact. Following collection of the solei, animals were killed by cervical dislocation. In the first study, samples from the lungs, liver, and distal colon were taken immediately post-sacrifice. In the second and third studies, only solei were collected. All tissues were rinsed in Krebs-Ringer solution containing (mM) 137 NaCl, 5 KCl, 1 NaH2PO4, 24 NaHCO3, 2 CaCl2, and 1 MgSO4 and gently blotted dry prior to weighing and freezing. The solution was maintained at room temperature and aerated with 95% O2 + 5% CO2. In the second and third studies, solei were used for contractile studies prior to weighing and freezing (see below).
Contractile protocol
For contractile studies, one tendon of the soleus muscle was fixed to a glass bar with silk suture (size 5-0); the other tendon was tied to a force transducer (model BG 100, Kulite Instruments, Leonia, NJ) mounted on a micrometer. Solei were stimulated by field stimulation with platinum electrodes. Muscle length was adjusted to generate maximal twitch force (optimal length, L0). The muscle bath was warmed to 37°C after determination of Lo. This temperature was maintained by digital water bath throughout the remainder of the experiment. After a 30-min thermoequilibration period, twitch characteristics (including twitch force, time to peak force, and half-relaxation time) were measured, followed by determination of the force frequency relationship. To determine the force frequency relationship, tetanic contractions were stimulated at 2 min intervals (500-ms train duration, 17 volts). Between each intermediate frequency (15, 30, 50, 80, 120, 160, 250 Hz), a maximum tetanic contraction (P0, 300 Hz) was elicited to serve as a reference for changes in force over time. One minute after the last stimulation, the muscle was subjected to a fatiguing contraction protocol by stimulating the muscle at 40 Hz using a 1:4 duty cycle (0.5 trains/s, 500-ms train duration). Passive and developed forces were recorded on a strip chart for later analysis. At the end of the experiment, soleus length and weight were measured. Force measurements were normalized for functional cross section according to Close [24]. Results from contractile studies of mice receiving the control diet have been previously reported [7].
Preparation of tissue extracts
Cytoplasmic extracts were prepared by grinding tissue samples in a solution containing 10 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 10 mM KCl, 2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM DTT, and 2 mM PMSF. Samples were ground in a glass tissue grinder cooled in an ice water slurry. After grinding, samples were vortexed, incubated on ice for ≥ 10 minutes, vortexed again, and then frozen on dry ice. After thawing in cool water, samples were vortexed a final time and centrifuged for 10 s at 4°C. The supernatant, containing the cytosolic proteins, was then collected for subsequent analyses. Nuclear extracts were prepared by resuspending the pellet in a buffer containing 20 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 5% glycerol, 420 mM NaCl, 0.2 mM EDTA, 2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM DTT, and 2 mM PMSF, and incubating for 30 minutes at 4°C. During incubation, samples were vortexed ~10 s every 10 min. Following incubation, the samples were centrifuged 4 minutes at 4°C and the supernatant, containing the nuclear proteins, was collected.
Protein determination
For determination of cytosolic and nuclear protein concentrations, 999 μl of a 1:4 mixture of Bio-Rad concentrated dye reagent (#500-0006, Bio-Rad, Hercules, CA) and water were added to 1 μl of extract. The sample was vortexed and the absorbance at 595 nm was measured. Protein concentration was determined using a standard curve of absorbance vs. concentration of bovine serum albumin (BSA). Samples and standards were analyzed in triplicate.
Electrophoretic mobility shift assay (EMSA)
Binding reactions were performed using 3–10 μg nuclear protein combined with 1 ng of NF-κB-binding DNA probe (5'AGTTGAGGGGACTTTCCCAGGC-3'), labeled with [α-32P]dATP (Amersham-Pharmacia) using the Klenow fragment. Binding reactions were carried out in binding buffer containing 10 mM Tris·HCl (pH 7.5), 10% glycerol, 0.05% NP-40, 0.1 μg/μl BSA, 0.05 μg/μl poly (dI-dC), and 500 μM DTT. Samples were incubated on ice for 30 minutes. Supershifts with antibodies to the p50 and p65 subunits of NF-κB (Santa Cruz; Santa Cruz, CA), were added 15 minutes into the incubation. After incubation, samples were resolved on 5% polyacrylamide gels. After drying, gels were exposed to x-ray film and analyzed using commercial densitometry software (ImageQuant 5.2, Amersham Biosciences Corp., Piscataway, NJ). Results from studies of mice receiving the control diet have been previously reported [7].
Statistical analysis
Values are means ± SE. A one-way ANOVA (diet × frequency or diet × time) was used to compare force production among groups at different stimulation frequencies or time-points. For differences in single measurements in paired groups we utilized paired t-tests. P value was set < 0.05. Statistical analyses were carried out using SigmaStat, version 3.00.
Results
Body weight and diet
The curcumin and NAC treatments were well-tolerated by the mice; no overt changes in behavior, grooming, or response to handling were observed in the treated mice. To verify that the curcumin diet had systemic effects on NF-κB activity, as well as to screen for gross adverse effects of the curcumin diet, in the first study we measured protein concentrations and NF-κB activity in a panel of tissues after 2 days of feeding the curcumin-supplemented diet. Curcumin feeding did not affect cytosolic or nuclear protein concentrations in any of the tissues tested (data not shown). In addition, after 48 hours of feeding, neither absolute (control: 1.65 ± 0.11 g, curcumin: 1.67 ± 0.07 g) nor relative (control: 54 ± 3 mg/g body weight, curcumin: 56 ± 2 mg/g body weight) liver weight was affected by curcumin (P > 0.05).
In a second study, ambulatory and unloaded mice were fed either the control diet or the curcumin diet. Ambulatory mice gained weight, regardless of whether they received the curcumin or the control diet, whereas the weight of the unloaded mice in both diet groups did not change over the course of the experiment (Table 1). There was no effect of diet on animal weight. Unloading reduced water intake (Table 1), whereas diet had no effect on this variable (Table 1).
Table 1 Dietary intakes and body weights.
Curcumin study NAC study
Experimental groups (diet) Ambulatory Control Unloaded Control Unloaded Curcumin Ambulatory Control Unloaded Control Unloaded Curcumin
Number of mice 11 11 11 3 6 7
Body weight, Day 0 26.5 ± 0.8 26.8 ± 0.7 27.3 ± 0.9 22.8 ± 1.5 24.1 ± 1.0 23.1 ± 0.8
Day 11 30.0 ± 1.2 * 26.3 ± 0.7 26.5 ± 0.8 28.4 ± 1.5 21.4 ± 0.8 20.7 ± 0.7
NAC intake (mg/g/day) - - - - - 0.78 ± 0.04
Total curcumin intake (g/kg) - - 22 ± 2 - -
In the third part of this study, unloaded mice receiving either water or NAC (1% w/v) in the drinking water were compared with ambulatory mice receiving water. As in the curcumin study, ambulatory mice gained weight during the study period whereas unloaded mice neither gained nor lost weight, regardless of treatment (Table 1). Unloaded mice receiving NAC in their drinking water consumed less water than unloaded mice receiving water alone (data not shown).
Effect of curcumin on basal NF-κB activity
To evaluate the in vivo effects of oral curcumin ingestion on NF-κB activity in different tissues, in the first study we measured NF-κB activity in intestine, liver, lung, and soleus of ambulatory mice fed either the curcumin diet or the control diet for 2 days. Basal NF-κB activity was readily apparent in all of the tissues analyzed (Fig. 2). The curcumin diet reduced basal NF-κB activity by 22–25% in all tissues tested (Fig. 2).
Figure 2 Effect of 1% curcumin diet on basal NF-κB activity in various tissues. Mice were fed a 1% curcumin diet for 2 days and then sacrificed for measurement of NF-κB activity. (*P < 0.05 vs control, N = 6–8 comparisons). The right panel for each graph depicts results from a typical gel shift assay. A) Liver, B) Soleus, C) Lung, D) Large Intestine.
Effects of unloading and diet on soleus muscle
Unloading reduced soleus weight by ~40–50%, with no effect of either curcumin or NAC on this response (Table 2). Cross sectional area of the soleus was also smaller in unloaded mice than in ambulatory mice; neither curcumin nor NAC modulated this response (Table 2). Soleus NF-κB activity was not different in unloaded mice receiving curcumin and those receiving the normal diet (Fig. 3). In contrast, oral NAC completely prevented the unloading-induced increase in soleus NF-κB activity (Fig. 3). Neither unloading nor diet altered protein concentrations in the soleus (data not shown).
Table 2 Effects of conditioning and treatment on muscle properties.
Ambulatory Unloaded
Parameter Control Curcumin Control Curcumin Control NAC
Soleus weight (mg) 7.3 ± 0.3 6.9 ± 0.4 3.4 ± 0.0a 3.3 ± 0.2a 3.6 ± 0.1a 3.8 ± 0.2a
Muscle length (cm) 1.10 ± 0.02 1.13 ± 0.02 1.02 ± 0.03b 1.09 ± 0.03b 0.96 ± 0.03 0.92 ± 0.03
Cross sectional area (cm2) 6.3 ± 0.0 5.8 ± 0.0 3.2 ± 0.0c 3.0 ± 0.0c 3.5 ± 0.1 3.9 ± 0.1
Figure 3 Effect dietary treatment on NF-κB activity in unloaded soleus. Panel A. Typical gel shift results showing effect of unloading on NF-κB activity. Panel B. Curcumin effects on NF-κB activity in unloaded soleus muscle. The right panel depicts results from a typical gel shift assay. Panel C. NAC effects on NF-κB activity in unloaded soleus muscle. The right panel depicts results from a typical gel shift assay.
Effects of unloading and dietary treatment on contractile performance
Hindlimb unloading reduced force production; neither curcumin nor NAC prevented this response (Fig. 4). Twitch stress, time to peak, and half relaxation time, as well as fatigue, were similar in all groups (data not shown).
Figure 4 Force frequency relationship in mouse solei. A. Curcumin effects on force-frequency relationship in unloaded mice. B. NAC effects on force-frequency relationship in ambulatory and unloaded mice treated with either the control or the NAC-supplemented diet. Values are mean ± SE of force per unit area (N/cm2).
Discussion
NF-κB activation is associated with stimulation of the ubiquitin-proteasome pathway and loss of skeletal muscle protein [6,8,12], suggesting that unloading-induced NF-κB activation may contribute to the catabolic response to muscle unloading [6]. In the present study, we tested the hypothesis that inhibition of NF-κB would prevent unloading-induced skeletal muscle atrophy and weakness. We report two new findings: 1) dietary curcumin, an inhibitor of the classical NF-κB pathway, does not inhibit unloading-induced NF-κB activation, atrophy, or weakness, in agreement with a previous study suggesting unloading-induced NF-κB activation does not occur by the classical pathway [6], and 2) dietary NAC, which inhibits NF-κB activation by both classical and nonclassical pathways, prevents unloading-induced NF-κB activation but does not inhibit unloading-induced atrophy or weakness, suggesting that NF-κB inhibition alone is insufficient to prevent these responses.
A recent study found that unloading-induced atrophy either did not occur or was greatly reduced in mice lacking genes for the NF-κB family members Nfkb1 or Bcl3 [6]. Taken together with the results from the present study, it appears that if it is possible to inhibit unloading-induced atrophy by pharmacological NF-κB inhibition it will require greater or more targeted (e.g. specific for p105/50 and/or bcl3) inhibition than was achieved in the present study using NAC.
Unloading increases skeletal muscle reactive oxygen levels [25,26], which are a stimulus for NF-κB activation [12]. Our finding that NAC prevented unloading-induced NF-κB activation suggests this response is secondary to increased levels of reactive oxygen species. The finding that NAC did not prevent unloading-induced muscle atrophy and contractile dysfunction is consistent with a previous study that found antioxidant supplementation (including NAC) during unloading did not prevent these responses [25].
In addition to stimulating muscle protein breakdown, skeletal muscle unloading inhibits muscle protein synthesis [27,28]. In the present study, inhibition of NF-κB activity by NAC did not prevent the atrophic response to unloading. This may reflect that reduced protein synthesis is the dominant factor responsible for unloading-induced atrophy. It is also possible that NF-κB inhibition alone is not sufficient to prevent unloading induced muscle protein breakdown.
Curcumin is an established inhibitor of NF-κB [17]. However, most of the studies examining the effects of curcumin on NF-κB activity have been performed in vitro. Thus, in the present study we determined the effect of short-term (2 days) dietary curcumin on basal NF-κB activity in soleus, lung, intestine, and liver, to determine whether dietary curcumin had in vivo effects. Curcumin ingestion slightly inhibited basal NF-κB activity in all tissues under basal conditions but did not inhibit the NF-κB activation that occurs in skeletal muscle in response to unloading.
Despite the fact that curcumin did not inhibit unloading-induced NF-κB activation, the finding that basal muscle NF-κB activity was inhibited by curcumin is noteworthy, as NF-κB is emerging as an important regulator of skeletal muscle metabolism. Activation of NF-κB in skeletal muscle myotubes increases ubiquitin conjugation [8], consistent with a role for NF-κB in muscle protein catabolism [9]. NF-κB activation is also thought to contribute to cancer-induced cachexia, in which muscle loss is often dramatic and is a strong predictor of a poor outcome [29]. In addition, recent studies have suggested a possible role for NF-κB in the development of insulin resistance [30].
Conclusion
In conclusion, this study shows that oral curcumin as a dietary supplement inhibits NF-κB activity similarly in a variety of tissues in ambulatory mice. However, it does not inhibit NF-κB activity in the solei of unloaded mice and, contrary to our original hypothesis, does not prevent atrophy of unloaded soleus muscles. Similarly, oral NAC treatment had no effect on unloading-induced atrophy although, in contrast to curcumin, it did prevent unloading-induced NF-κB activation.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MF carried out the curcumin feeding and contractile studies, contributed to the design of the study, and assisted in the writing of the manuscript. EG carried out the NAC feeding and contractile studies, contributed to the design of the study, and assisted in the writing of the manuscript. MBR conceived of the study and design, contributed to the data interpretation, and participated in manuscript preparation. YPL contributed to data analysis and interpretation and participated in manuscript preparation. WJD performed the protein and gel shift assays, contributed to study design and data interpretation, and participated in manuscript preparation.
Acknowledgements
This research was supported by grants from the National Institutes of Health (NIH HL59878 to MBR and NIH AR49022 to YPL) and the National Space Biomedical Research Institute (to MBR)
==== Refs
Widrick JJ Knuth ST Norenberg KM Romatowski JG Bain JL Riley DA Karhanek M Trappe SW Trappe TA Costill DL Fitts RH Effect of a 17 day spaceflight on contractile properties of human soleus muscle fibres J Physiol 1999 516 915 930 10200437 10.1111/j.1469-7793.1999.0915u.x
Jiang B Roy RR Navarro C Edgerton VR Absence of a growth hormone effect on rat soleus atrophy during a 4-day spaceflight J Appl Physiol 1993 74 527 531 8458766
Fitts RH Riley DR Widrick JJ Physiology of a microgravity environment invited review: microgravity and skeletal muscle J Appl Physiol 2000 89 823 839 10926670
Caiozzo VJ Haddad F Baker MJ Herrick RE Prietto N Baldwin KM Microgravity-induced transformations of myosin isoforms and contractile properties of skeletal muscle J Appl Physiol 1996 81 123 132 8828654
Kondo H Sen CK, Packer L, Hanninen O Oxidative stress in muscular atrophy Handbook of Oxidants and Antioxidants in Exercise 2000 New York, Elsevier 631 653
Hunter RB Kandarian SC Disruption of either the Nfkb1 or the Bcl3 gene inhibits skeletal muscle atrophy J Clin Invest 2004 114 1504 1511 15546001 10.1172/JCI200421696
Durham WJ Li YP Gerken E Farid M Arbogast S Wolfe RR Reid MB Fatiguing exercise reduces DNA-binding activity of NF-κB in skeletal muscle nuclei J Appl Physiol 2004 97 1740 1745 Epub 2004 Jun 18 15208298 10.1152/japplphysiol.00088.2004
Li YP Lecker SH Chen Y Waddell ID Goldberg AL Reid MB TNF-α increases ubiquitin conjugating activity in skeletal muscle by upregulating UbcH2/E220k FASEB J 2003 17 1048 57 12773487 10.1096/fj.02-0759com
Reid MB Li YP Tumor necrosis factor-alpha and muscle wasting: a cellular perspective Respir Res 2001 2 269 272 11686894 10.1186/rr67
Li YP Schwartz RJ TNF-α regulates early differentiation of C2C12 myoblasts in an autocrine fashion FASEB J 2001 15 1413 5 11387241
Li YP Reid MB NF-kappaB mediates the protein loss induced by TNF-alpha in differentiated skeletal muscle myotubes Am J Physiol Regul Integr Comp Physiol 2000 279 R1165 R1170 11003979
Li YP Schwartz RJ Waddell ID Holloway BR Reid MB Skeletal muscle myocytes undergo protein loss and reactive oxygen- mediated NF-kappaB activation in response to tumor necrosis factor alpha FASEB J 1998 12 871 880 9657527
Singh S Aggarwal BB Activation of transcription factor NF-kappa B is suppressed by curcumin (diferuloylmethane) [corrected] J Biol Chem 1995 270 24995 25000 7559628 10.1074/jbc.270.25.14867
Satoskar RR Shah SJ Shenoy SG Evaluation of anti-inflammatory property of curcumin (diferuloyl methane) in patients with postoperative inflammation Int J Clin Pharmacol Ther Toxicol 1986 24 651 654 3546166
Quiles JL Mesa MD Ramirez-Tortosa CL Aguilera CM Battino M Gil A Ramirez-Tortosa MC Curcuma longa extract supplementation reduces oxidative stress and attenuates aortic fatty streak development in rabbits Arterioscler Thromb Vasc Biol 2002 22 1225 1231 12117742 10.1161/01.ATV.0000020676.11586.F2
Asai A Miyazawa T Dietary curcuminoids prevent high-fat diet-induced lipid accumulation in rat liver and epididymal adipose tissue J Nutr 2001 131 2932 2935 11694621
Bharti AC Aggarwal BB Nuclear factor κB and cancer: its role in prevention and therapy Biochem Pharmacol 2002 64 883 8 12213582 10.1016/S0006-2952(02)01154-1
Hsu CC Huang CN Hung YC Yin MC Five cysteine-containing compounds have antioxidative activity in Balb/cA mice J Nutr 2004 134 149 152 14704308
Demirkol O Adams C Ercal N Biologically important thiols in various vegetables and fruits J Agric Food Chem 2004 52
Tosetti F Ferrari N De Flora S Albini A Angioprevention: angiogenesis is a common and key target for cancer chemopreventive agents FASEB J 2002 16 2 14 11772931 10.1096/fj.01-0300rev
Qanungo S Wang M Nieminen AL N-Acetyl-L-cysteine enhances apoptosis through inhibition of nuclear factor-kappaB in hypoxic murine embryonic fibroblasts J Biol Chem 2004 279 50455 50464 15375156 10.1074/jbc.M406749200
Pajonk F Riess K Sommer A McBride WH N-acetyl-L-cysteine inhibits 26S proteasome function: implications for effects on NF-kappaB activation Free Rad Biol Med 2002 32 536 543 11958954 10.1016/S0891-5849(02)00743-8
Schubert SY Neeman I Resnick N A novel mechanism for the inhibition of NF-kappaB activation in vascular endothelial cells by natural antioxidants FASEB J 2002 16 1931 3 Epub 2002 Oct 4 12368228
Close RI Dynamic properties of mammalian skeletal muscle Physiol Rev 1972 52 129 197 4256989
Koesterer TJ Dodd SL Powers S Increased antioxidant capacity does not attenuate muscle atrophy caused by unweighting J Appl Physiol 2002 93 1959 65 12391085
Lawler JM Song W Demaree SR Hindlimb unloading increases oxidative stress and disrupts antioxidant capacity in skeletal muscle Free Rad Biol Med 2003 35 9 16 12826251 10.1016/S0891-5849(03)00186-2
Hornberger TA Hunter RB Kandarian SC Esser KA Regulation of translation factors during hindlimb unloading and denervation of skeletal muscle in rats Am J Physiol Cell Physiol 2001 281 C179 87 11401840
Fluckey JD Dupont-Versteegden EE Knox M Gaddy D Tesch PA Peterson CA Insulin facilitation of muscle protein synthesis following resistance exercise in hindlimb-suspended rats is independent of a rapamycin-sensitive pathway Am J Physiol Endocrinol Metab 2004 287 E1070 5 Epub 2004 Aug 10 15304378 10.1152/ajpendo.00329.2004
Hasselgren PO Fischer JE Muscle cachexia: current concepts of intracellular mechanisms and molecular regulation Ann Surg 2001 233 9 17 11141219 10.1097/00000658-200101000-00003
Itani SI Ruderman NB Schmieder F Boden G Lipid-induced insulin resistance in human muscle is associated with changes in diacylglycerol, protein kinase C, and IκBα Diabetes 2002 51 2005 11 12086926
|
16124875
|
PMC1208951
|
CC BY
|
2021-01-04 16:37:46
|
no
|
Nutr Metab (Lond). 2005 Aug 26; 2:20
|
utf-8
|
Nutr Metab (Lond)
| 2,005 |
10.1186/1743-7075-2-20
|
oa_comm
|
==== Front
Nutr Metab (Lond)Nutrition & Metabolism1743-7075BioMed Central London 1743-7075-2-201612487510.1186/1743-7075-2-20ResearchEffects of dietary curcumin or N-acetylcysteine on NF-κB activity and contractile performance in ambulatory and unloaded murine soleus Farid Mehran [email protected] Michael B [email protected] Yi-Ping [email protected] Eric [email protected] William J [email protected] Pulmonary Medicine, Baylor College of Medicine, One Baylor Plaza, Houston, TX, 77030, USA2 Department of Physiology, College of Medicine, University of Kentucky, Lexington, KY,40506, USA2005 26 8 2005 2 20 20 21 6 2005 26 8 2005 Copyright © 2005 Farid et al; licensee BioMed Central Ltd.2005Farid et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Unloading of skeletal muscle causes atrophy and loss of contractile function. In part, this response is believed to be mediated by the transcription factor nuclear factor-kappa B (NF-κB). Both curcumin, a component of the spice turmeric, and N-acetylcysteine (NAC), an antioxidant, inhibit activation of NF-κB by inflammatory stimuli, albeit by different mechanisms. In the present study, we tested the hypothesis that dietary curcumin or NAC supplementation would inhibit unloading-induced NF-κB activity in skeletal muscle and thereby protect muscles against loss of mass and function caused by prolonged unloading.
Methods
We used hindlimb suspension to unload the hindlimb muscles of adult mice. Animals had free access to drinking water or drinking water supplemented with 1% NAC and to standard laboratory diet or diet supplemented with 1% curcumin. For 11 days, half the animals in each dietary group were suspended by the tail (unloaded) and half were allowed to ambulate freely.
Results
Unloading caused a 51–53% loss of soleus muscle weight and cross-sectional area relative to freely-ambulating controls. Unloading also decreased total force and force per cross-sectional area developed by soleus. Curcumin supplementation decreased NF-κB activity measured in peripheral tissues of ambulatory mice by gel shift analysis. In unloaded animals, curcumin supplementation did not inhibit NF-κB activity or blunt the loss of muscle mass in soleus. In contrast, NAC prevented the increase in NF-κB activity induced by unloading but did not prevent losses of muscle mass or function.
Conclusion
In conclusion, neither dietary curcumin nor dietary NAC prevents unloading-induced skeletal muscle dysfunction and atrophy, although dietary NAC does prevent unloading induced NF-κB activation.
==== Body
Background
Prolonged exposure to microgravity causes muscle atrophy, resulting in weakness and predisposition to fatigue. Such changes can markedly degrade astronauts' performance, especially upon return to Earth [1]. Atrophy starts as early as 4 days into space flight [2] and becomes more significant in long-term space missions [3]. In addition to atrophy, weight-bearing skeletal muscles lose their contractile force per cross-sectional area [4]. Since atrophy in space flight and disuse atrophy share similar pathophysiologic mechanisms [5], countermeasures to microgravity-induced muscle wasting may also be effective for the prevention of disuse atrophy and muscle weakness in long term bedridden patients. Thus, there is considerable interest in the development of countermeasures that oppose the complications of muscle disuse.
Hindlimb unloading, a commonly used model to study the effects of muscle disuse, activates the transcription factor nuclear factor-κB (NF-κB) in weight-bearing muscles [6,7]. This laboratory has recently reported that NF-κB activation increases ubiquitin conjugation [8], consistent with a role for NF-κB in the regulation of skeletal muscle protein breakdown [9]. In addition, activation of NF-κB in response to TNF-α alters myogenesis [10] and is essential for the loss of protein in TNF-α-stimulated skeletal muscle myotubes [11]. Stimuli that persistently activate NF-κB are associated with muscle protein loss and activation of the ubiquitin-proteasome pathway [12]. In vitro, selective inhibition of NF-κB can prevent loss of muscle protein [11]. In vivo, a recent study found that mice lacking specific NF-κB family members (p105/50 or bcl3) are resistant to unloading-induced atrophy and NF-κB activation [6], supporting a role for NF-κB as a robust in vivo catabolic stimulus during disuse atrophy.
The prevention of unloading-induced atrophy by ablation of genes for specific NF-κB family members [6] suggests that pharmacological inhibition of NF-κB might also reduce atrophy in response to unloading. In the present study, we tested the ability of two NF-κB inhibitors to inhibit disuse-induced skeletal muscle NF-κB activation. Curcumin (diferuloylmethane), a major component of the spice turmeric (Curcuma longa), has been reported to exert in vivo effects on serum lipids [13], aortic atherosclerosis [14], and various forms of cancer [15]. Curcumin also possesses anti-inflammatory [16] and antioxidant [14] properties in vivo and inhibits the canonical NF-κB activation pathway [17]. In the canonical activation pathway, the NF-κB inhibitor IκB is phosphorylated by IκB kinase (IKK), leading to its dissociation from NF-κB and subsequent translocation of NF-κB to the nucleus [6], where it binds to the promoter region of target genes. Curcumin acts at a proximal step in the canonical activation pathway, inhibiting IKK [17] phosphorylation of IκBα. N-acetylcysteine (NAC) is a naturally occurring compound found in several vegetables, including garlic [18], onion [18], peppers [19], and asparagus [19]. In addition to antioxidant, antiangiogenic, and anticarcinogenic properties [20], NAC inhibits NF-κB activation by several mechanisms acting distal to the activity of IKK [21-23].
NF-κB activation in response to muscle disuse has been reported to occur through a pathway distinct from the canonical activation pathway that does not involve IκB activation [6]. In the present study, we tested the hypothesis that curcumin, an inhibitor of the canonical activation pathway, and NAC, acting downstream of curcumin in the NF-κB activation pathway, would inhibit NF-κB and preserve muscle mass and function. In addition, as the in vivo effects of dietary curcumin on NF-κB activity are not well-studied, particularly under basal (non-induced) conditions, we measured NF-κB activity in liver, soleus, lung, and large intestine to test the hypothesis that dietary curcumin inhibits basal NF-κB activity in vivo.
Methods
Animals
Adult male ICR mice were purchased from Harlan (Indianapolis, IN) and housed under pathogen-free conditions. Animals were handled in accordance with the policies and guidelines for the care and use of laboratory animals of the National Institutes of Health. The protocol was approved by the Animal Protocol Review Committee at Baylor College of Medicine.
Food
Mice in control and treatment groups were fed a standard chow diet (Purina-Mills, 5053 diet, Labdiet, St Louis, MO). For the curcumin treatment group, the diet was modified by adding commercial curcumin (LKT laboratories, St. Paul, Minnesota) to 1% (w/w). Curcumin was uniformly incorporated into the food, which was pelleted to ensure uniform feed and curcumin intake. All control and experimental diets were stored at 4°C. All mice had free access to water during the dietary treatment period. For the NAC treatment group, 1% NAC (w/v) was included in the drinking water.
Methods
In the first study (Figure 1), ambulatory mice were studied after two days on either the control diet or the curcumin-supplemented diet, in order to test whether curcumin inhibited NF-κB in vivo. In the second study (Figure 1), the effects of curcumin were studied in three groups of mice. Two groups received the control diet throughout the 13-day study period, whereas the other group received the curcumin diet. After 2 days, the curcumin dietary group and one of the groups receiving the control diet were hindlimb-unloaded. In the two unloaded groups, the tail of each animal was taped to suspension wire using first aid tape in order to elevate the hindlimb of each animal. Mice remained unloaded for the remainder of the study period. Unloaded mice were able to ambulate using their forelegs; however, their hindlimbs were unable to touch the bottom or sides of the cage. In the third study (Figure 1), three groups were studied: one ambulatory group receiving normal drinking water, one hindlimb-unloaded group receiving normal drinking water, and one unloaded group receiving 1% NAC in the drinking water. All mice in the third study received the standard chow diet.
Figure 1 Unloading protocol. In study 1, ambulatory mice were fed either a control or a 1% curcumin diet for 2 days. In study 2, mice were divided into control and curcumin diet groups as in study 1 for the first two days, after which each diet group was further divided into ambulatory or unloaded groups.
Tissue collection
Mice were deeply anesthetized by methoxyflurane inhalation and soleus muscles were excised with their tendons intact. Following collection of the solei, animals were killed by cervical dislocation. In the first study, samples from the lungs, liver, and distal colon were taken immediately post-sacrifice. In the second and third studies, only solei were collected. All tissues were rinsed in Krebs-Ringer solution containing (mM) 137 NaCl, 5 KCl, 1 NaH2PO4, 24 NaHCO3, 2 CaCl2, and 1 MgSO4 and gently blotted dry prior to weighing and freezing. The solution was maintained at room temperature and aerated with 95% O2 + 5% CO2. In the second and third studies, solei were used for contractile studies prior to weighing and freezing (see below).
Contractile protocol
For contractile studies, one tendon of the soleus muscle was fixed to a glass bar with silk suture (size 5-0); the other tendon was tied to a force transducer (model BG 100, Kulite Instruments, Leonia, NJ) mounted on a micrometer. Solei were stimulated by field stimulation with platinum electrodes. Muscle length was adjusted to generate maximal twitch force (optimal length, L0). The muscle bath was warmed to 37°C after determination of Lo. This temperature was maintained by digital water bath throughout the remainder of the experiment. After a 30-min thermoequilibration period, twitch characteristics (including twitch force, time to peak force, and half-relaxation time) were measured, followed by determination of the force frequency relationship. To determine the force frequency relationship, tetanic contractions were stimulated at 2 min intervals (500-ms train duration, 17 volts). Between each intermediate frequency (15, 30, 50, 80, 120, 160, 250 Hz), a maximum tetanic contraction (P0, 300 Hz) was elicited to serve as a reference for changes in force over time. One minute after the last stimulation, the muscle was subjected to a fatiguing contraction protocol by stimulating the muscle at 40 Hz using a 1:4 duty cycle (0.5 trains/s, 500-ms train duration). Passive and developed forces were recorded on a strip chart for later analysis. At the end of the experiment, soleus length and weight were measured. Force measurements were normalized for functional cross section according to Close [24]. Results from contractile studies of mice receiving the control diet have been previously reported [7].
Preparation of tissue extracts
Cytoplasmic extracts were prepared by grinding tissue samples in a solution containing 10 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 10 mM KCl, 2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM DTT, and 2 mM PMSF. Samples were ground in a glass tissue grinder cooled in an ice water slurry. After grinding, samples were vortexed, incubated on ice for ≥ 10 minutes, vortexed again, and then frozen on dry ice. After thawing in cool water, samples were vortexed a final time and centrifuged for 10 s at 4°C. The supernatant, containing the cytosolic proteins, was then collected for subsequent analyses. Nuclear extracts were prepared by resuspending the pellet in a buffer containing 20 mM HEPES-KOH (pH 7.9), 1.5 mM MgCl2, 5% glycerol, 420 mM NaCl, 0.2 mM EDTA, 2 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM DTT, and 2 mM PMSF, and incubating for 30 minutes at 4°C. During incubation, samples were vortexed ~10 s every 10 min. Following incubation, the samples were centrifuged 4 minutes at 4°C and the supernatant, containing the nuclear proteins, was collected.
Protein determination
For determination of cytosolic and nuclear protein concentrations, 999 μl of a 1:4 mixture of Bio-Rad concentrated dye reagent (#500-0006, Bio-Rad, Hercules, CA) and water were added to 1 μl of extract. The sample was vortexed and the absorbance at 595 nm was measured. Protein concentration was determined using a standard curve of absorbance vs. concentration of bovine serum albumin (BSA). Samples and standards were analyzed in triplicate.
Electrophoretic mobility shift assay (EMSA)
Binding reactions were performed using 3–10 μg nuclear protein combined with 1 ng of NF-κB-binding DNA probe (5'AGTTGAGGGGACTTTCCCAGGC-3'), labeled with [α-32P]dATP (Amersham-Pharmacia) using the Klenow fragment. Binding reactions were carried out in binding buffer containing 10 mM Tris·HCl (pH 7.5), 10% glycerol, 0.05% NP-40, 0.1 μg/μl BSA, 0.05 μg/μl poly (dI-dC), and 500 μM DTT. Samples were incubated on ice for 30 minutes. Supershifts with antibodies to the p50 and p65 subunits of NF-κB (Santa Cruz; Santa Cruz, CA), were added 15 minutes into the incubation. After incubation, samples were resolved on 5% polyacrylamide gels. After drying, gels were exposed to x-ray film and analyzed using commercial densitometry software (ImageQuant 5.2, Amersham Biosciences Corp., Piscataway, NJ). Results from studies of mice receiving the control diet have been previously reported [7].
Statistical analysis
Values are means ± SE. A one-way ANOVA (diet × frequency or diet × time) was used to compare force production among groups at different stimulation frequencies or time-points. For differences in single measurements in paired groups we utilized paired t-tests. P value was set < 0.05. Statistical analyses were carried out using SigmaStat, version 3.00.
Results
Body weight and diet
The curcumin and NAC treatments were well-tolerated by the mice; no overt changes in behavior, grooming, or response to handling were observed in the treated mice. To verify that the curcumin diet had systemic effects on NF-κB activity, as well as to screen for gross adverse effects of the curcumin diet, in the first study we measured protein concentrations and NF-κB activity in a panel of tissues after 2 days of feeding the curcumin-supplemented diet. Curcumin feeding did not affect cytosolic or nuclear protein concentrations in any of the tissues tested (data not shown). In addition, after 48 hours of feeding, neither absolute (control: 1.65 ± 0.11 g, curcumin: 1.67 ± 0.07 g) nor relative (control: 54 ± 3 mg/g body weight, curcumin: 56 ± 2 mg/g body weight) liver weight was affected by curcumin (P > 0.05).
In a second study, ambulatory and unloaded mice were fed either the control diet or the curcumin diet. Ambulatory mice gained weight, regardless of whether they received the curcumin or the control diet, whereas the weight of the unloaded mice in both diet groups did not change over the course of the experiment (Table 1). There was no effect of diet on animal weight. Unloading reduced water intake (Table 1), whereas diet had no effect on this variable (Table 1).
Table 1 Dietary intakes and body weights.
Curcumin study NAC study
Experimental groups (diet) Ambulatory Control Unloaded Control Unloaded Curcumin Ambulatory Control Unloaded Control Unloaded Curcumin
Number of mice 11 11 11 3 6 7
Body weight, Day 0 26.5 ± 0.8 26.8 ± 0.7 27.3 ± 0.9 22.8 ± 1.5 24.1 ± 1.0 23.1 ± 0.8
Day 11 30.0 ± 1.2 * 26.3 ± 0.7 26.5 ± 0.8 28.4 ± 1.5 21.4 ± 0.8 20.7 ± 0.7
NAC intake (mg/g/day) - - - - - 0.78 ± 0.04
Total curcumin intake (g/kg) - - 22 ± 2 - -
In the third part of this study, unloaded mice receiving either water or NAC (1% w/v) in the drinking water were compared with ambulatory mice receiving water. As in the curcumin study, ambulatory mice gained weight during the study period whereas unloaded mice neither gained nor lost weight, regardless of treatment (Table 1). Unloaded mice receiving NAC in their drinking water consumed less water than unloaded mice receiving water alone (data not shown).
Effect of curcumin on basal NF-κB activity
To evaluate the in vivo effects of oral curcumin ingestion on NF-κB activity in different tissues, in the first study we measured NF-κB activity in intestine, liver, lung, and soleus of ambulatory mice fed either the curcumin diet or the control diet for 2 days. Basal NF-κB activity was readily apparent in all of the tissues analyzed (Fig. 2). The curcumin diet reduced basal NF-κB activity by 22–25% in all tissues tested (Fig. 2).
Figure 2 Effect of 1% curcumin diet on basal NF-κB activity in various tissues. Mice were fed a 1% curcumin diet for 2 days and then sacrificed for measurement of NF-κB activity. (*P < 0.05 vs control, N = 6–8 comparisons). The right panel for each graph depicts results from a typical gel shift assay. A) Liver, B) Soleus, C) Lung, D) Large Intestine.
Effects of unloading and diet on soleus muscle
Unloading reduced soleus weight by ~40–50%, with no effect of either curcumin or NAC on this response (Table 2). Cross sectional area of the soleus was also smaller in unloaded mice than in ambulatory mice; neither curcumin nor NAC modulated this response (Table 2). Soleus NF-κB activity was not different in unloaded mice receiving curcumin and those receiving the normal diet (Fig. 3). In contrast, oral NAC completely prevented the unloading-induced increase in soleus NF-κB activity (Fig. 3). Neither unloading nor diet altered protein concentrations in the soleus (data not shown).
Table 2 Effects of conditioning and treatment on muscle properties.
Ambulatory Unloaded
Parameter Control Curcumin Control Curcumin Control NAC
Soleus weight (mg) 7.3 ± 0.3 6.9 ± 0.4 3.4 ± 0.0a 3.3 ± 0.2a 3.6 ± 0.1a 3.8 ± 0.2a
Muscle length (cm) 1.10 ± 0.02 1.13 ± 0.02 1.02 ± 0.03b 1.09 ± 0.03b 0.96 ± 0.03 0.92 ± 0.03
Cross sectional area (cm2) 6.3 ± 0.0 5.8 ± 0.0 3.2 ± 0.0c 3.0 ± 0.0c 3.5 ± 0.1 3.9 ± 0.1
Figure 3 Effect dietary treatment on NF-κB activity in unloaded soleus. Panel A. Typical gel shift results showing effect of unloading on NF-κB activity. Panel B. Curcumin effects on NF-κB activity in unloaded soleus muscle. The right panel depicts results from a typical gel shift assay. Panel C. NAC effects on NF-κB activity in unloaded soleus muscle. The right panel depicts results from a typical gel shift assay.
Effects of unloading and dietary treatment on contractile performance
Hindlimb unloading reduced force production; neither curcumin nor NAC prevented this response (Fig. 4). Twitch stress, time to peak, and half relaxation time, as well as fatigue, were similar in all groups (data not shown).
Figure 4 Force frequency relationship in mouse solei. A. Curcumin effects on force-frequency relationship in unloaded mice. B. NAC effects on force-frequency relationship in ambulatory and unloaded mice treated with either the control or the NAC-supplemented diet. Values are mean ± SE of force per unit area (N/cm2).
Discussion
NF-κB activation is associated with stimulation of the ubiquitin-proteasome pathway and loss of skeletal muscle protein [6,8,12], suggesting that unloading-induced NF-κB activation may contribute to the catabolic response to muscle unloading [6]. In the present study, we tested the hypothesis that inhibition of NF-κB would prevent unloading-induced skeletal muscle atrophy and weakness. We report two new findings: 1) dietary curcumin, an inhibitor of the classical NF-κB pathway, does not inhibit unloading-induced NF-κB activation, atrophy, or weakness, in agreement with a previous study suggesting unloading-induced NF-κB activation does not occur by the classical pathway [6], and 2) dietary NAC, which inhibits NF-κB activation by both classical and nonclassical pathways, prevents unloading-induced NF-κB activation but does not inhibit unloading-induced atrophy or weakness, suggesting that NF-κB inhibition alone is insufficient to prevent these responses.
A recent study found that unloading-induced atrophy either did not occur or was greatly reduced in mice lacking genes for the NF-κB family members Nfkb1 or Bcl3 [6]. Taken together with the results from the present study, it appears that if it is possible to inhibit unloading-induced atrophy by pharmacological NF-κB inhibition it will require greater or more targeted (e.g. specific for p105/50 and/or bcl3) inhibition than was achieved in the present study using NAC.
Unloading increases skeletal muscle reactive oxygen levels [25,26], which are a stimulus for NF-κB activation [12]. Our finding that NAC prevented unloading-induced NF-κB activation suggests this response is secondary to increased levels of reactive oxygen species. The finding that NAC did not prevent unloading-induced muscle atrophy and contractile dysfunction is consistent with a previous study that found antioxidant supplementation (including NAC) during unloading did not prevent these responses [25].
In addition to stimulating muscle protein breakdown, skeletal muscle unloading inhibits muscle protein synthesis [27,28]. In the present study, inhibition of NF-κB activity by NAC did not prevent the atrophic response to unloading. This may reflect that reduced protein synthesis is the dominant factor responsible for unloading-induced atrophy. It is also possible that NF-κB inhibition alone is not sufficient to prevent unloading induced muscle protein breakdown.
Curcumin is an established inhibitor of NF-κB [17]. However, most of the studies examining the effects of curcumin on NF-κB activity have been performed in vitro. Thus, in the present study we determined the effect of short-term (2 days) dietary curcumin on basal NF-κB activity in soleus, lung, intestine, and liver, to determine whether dietary curcumin had in vivo effects. Curcumin ingestion slightly inhibited basal NF-κB activity in all tissues under basal conditions but did not inhibit the NF-κB activation that occurs in skeletal muscle in response to unloading.
Despite the fact that curcumin did not inhibit unloading-induced NF-κB activation, the finding that basal muscle NF-κB activity was inhibited by curcumin is noteworthy, as NF-κB is emerging as an important regulator of skeletal muscle metabolism. Activation of NF-κB in skeletal muscle myotubes increases ubiquitin conjugation [8], consistent with a role for NF-κB in muscle protein catabolism [9]. NF-κB activation is also thought to contribute to cancer-induced cachexia, in which muscle loss is often dramatic and is a strong predictor of a poor outcome [29]. In addition, recent studies have suggested a possible role for NF-κB in the development of insulin resistance [30].
Conclusion
In conclusion, this study shows that oral curcumin as a dietary supplement inhibits NF-κB activity similarly in a variety of tissues in ambulatory mice. However, it does not inhibit NF-κB activity in the solei of unloaded mice and, contrary to our original hypothesis, does not prevent atrophy of unloaded soleus muscles. Similarly, oral NAC treatment had no effect on unloading-induced atrophy although, in contrast to curcumin, it did prevent unloading-induced NF-κB activation.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MF carried out the curcumin feeding and contractile studies, contributed to the design of the study, and assisted in the writing of the manuscript. EG carried out the NAC feeding and contractile studies, contributed to the design of the study, and assisted in the writing of the manuscript. MBR conceived of the study and design, contributed to the data interpretation, and participated in manuscript preparation. YPL contributed to data analysis and interpretation and participated in manuscript preparation. WJD performed the protein and gel shift assays, contributed to study design and data interpretation, and participated in manuscript preparation.
Acknowledgements
This research was supported by grants from the National Institutes of Health (NIH HL59878 to MBR and NIH AR49022 to YPL) and the National Space Biomedical Research Institute (to MBR)
==== Refs
Widrick JJ Knuth ST Norenberg KM Romatowski JG Bain JL Riley DA Karhanek M Trappe SW Trappe TA Costill DL Fitts RH Effect of a 17 day spaceflight on contractile properties of human soleus muscle fibres J Physiol 1999 516 915 930 10200437 10.1111/j.1469-7793.1999.0915u.x
Jiang B Roy RR Navarro C Edgerton VR Absence of a growth hormone effect on rat soleus atrophy during a 4-day spaceflight J Appl Physiol 1993 74 527 531 8458766
Fitts RH Riley DR Widrick JJ Physiology of a microgravity environment invited review: microgravity and skeletal muscle J Appl Physiol 2000 89 823 839 10926670
Caiozzo VJ Haddad F Baker MJ Herrick RE Prietto N Baldwin KM Microgravity-induced transformations of myosin isoforms and contractile properties of skeletal muscle J Appl Physiol 1996 81 123 132 8828654
Kondo H Sen CK, Packer L, Hanninen O Oxidative stress in muscular atrophy Handbook of Oxidants and Antioxidants in Exercise 2000 New York, Elsevier 631 653
Hunter RB Kandarian SC Disruption of either the Nfkb1 or the Bcl3 gene inhibits skeletal muscle atrophy J Clin Invest 2004 114 1504 1511 15546001 10.1172/JCI200421696
Durham WJ Li YP Gerken E Farid M Arbogast S Wolfe RR Reid MB Fatiguing exercise reduces DNA-binding activity of NF-κB in skeletal muscle nuclei J Appl Physiol 2004 97 1740 1745 Epub 2004 Jun 18 15208298 10.1152/japplphysiol.00088.2004
Li YP Lecker SH Chen Y Waddell ID Goldberg AL Reid MB TNF-α increases ubiquitin conjugating activity in skeletal muscle by upregulating UbcH2/E220k FASEB J 2003 17 1048 57 12773487 10.1096/fj.02-0759com
Reid MB Li YP Tumor necrosis factor-alpha and muscle wasting: a cellular perspective Respir Res 2001 2 269 272 11686894 10.1186/rr67
Li YP Schwartz RJ TNF-α regulates early differentiation of C2C12 myoblasts in an autocrine fashion FASEB J 2001 15 1413 5 11387241
Li YP Reid MB NF-kappaB mediates the protein loss induced by TNF-alpha in differentiated skeletal muscle myotubes Am J Physiol Regul Integr Comp Physiol 2000 279 R1165 R1170 11003979
Li YP Schwartz RJ Waddell ID Holloway BR Reid MB Skeletal muscle myocytes undergo protein loss and reactive oxygen- mediated NF-kappaB activation in response to tumor necrosis factor alpha FASEB J 1998 12 871 880 9657527
Singh S Aggarwal BB Activation of transcription factor NF-kappa B is suppressed by curcumin (diferuloylmethane) [corrected] J Biol Chem 1995 270 24995 25000 7559628 10.1074/jbc.270.25.14867
Satoskar RR Shah SJ Shenoy SG Evaluation of anti-inflammatory property of curcumin (diferuloyl methane) in patients with postoperative inflammation Int J Clin Pharmacol Ther Toxicol 1986 24 651 654 3546166
Quiles JL Mesa MD Ramirez-Tortosa CL Aguilera CM Battino M Gil A Ramirez-Tortosa MC Curcuma longa extract supplementation reduces oxidative stress and attenuates aortic fatty streak development in rabbits Arterioscler Thromb Vasc Biol 2002 22 1225 1231 12117742 10.1161/01.ATV.0000020676.11586.F2
Asai A Miyazawa T Dietary curcuminoids prevent high-fat diet-induced lipid accumulation in rat liver and epididymal adipose tissue J Nutr 2001 131 2932 2935 11694621
Bharti AC Aggarwal BB Nuclear factor κB and cancer: its role in prevention and therapy Biochem Pharmacol 2002 64 883 8 12213582 10.1016/S0006-2952(02)01154-1
Hsu CC Huang CN Hung YC Yin MC Five cysteine-containing compounds have antioxidative activity in Balb/cA mice J Nutr 2004 134 149 152 14704308
Demirkol O Adams C Ercal N Biologically important thiols in various vegetables and fruits J Agric Food Chem 2004 52
Tosetti F Ferrari N De Flora S Albini A Angioprevention: angiogenesis is a common and key target for cancer chemopreventive agents FASEB J 2002 16 2 14 11772931 10.1096/fj.01-0300rev
Qanungo S Wang M Nieminen AL N-Acetyl-L-cysteine enhances apoptosis through inhibition of nuclear factor-kappaB in hypoxic murine embryonic fibroblasts J Biol Chem 2004 279 50455 50464 15375156 10.1074/jbc.M406749200
Pajonk F Riess K Sommer A McBride WH N-acetyl-L-cysteine inhibits 26S proteasome function: implications for effects on NF-kappaB activation Free Rad Biol Med 2002 32 536 543 11958954 10.1016/S0891-5849(02)00743-8
Schubert SY Neeman I Resnick N A novel mechanism for the inhibition of NF-kappaB activation in vascular endothelial cells by natural antioxidants FASEB J 2002 16 1931 3 Epub 2002 Oct 4 12368228
Close RI Dynamic properties of mammalian skeletal muscle Physiol Rev 1972 52 129 197 4256989
Koesterer TJ Dodd SL Powers S Increased antioxidant capacity does not attenuate muscle atrophy caused by unweighting J Appl Physiol 2002 93 1959 65 12391085
Lawler JM Song W Demaree SR Hindlimb unloading increases oxidative stress and disrupts antioxidant capacity in skeletal muscle Free Rad Biol Med 2003 35 9 16 12826251 10.1016/S0891-5849(03)00186-2
Hornberger TA Hunter RB Kandarian SC Esser KA Regulation of translation factors during hindlimb unloading and denervation of skeletal muscle in rats Am J Physiol Cell Physiol 2001 281 C179 87 11401840
Fluckey JD Dupont-Versteegden EE Knox M Gaddy D Tesch PA Peterson CA Insulin facilitation of muscle protein synthesis following resistance exercise in hindlimb-suspended rats is independent of a rapamycin-sensitive pathway Am J Physiol Endocrinol Metab 2004 287 E1070 5 Epub 2004 Aug 10 15304378 10.1152/ajpendo.00329.2004
Hasselgren PO Fischer JE Muscle cachexia: current concepts of intracellular mechanisms and molecular regulation Ann Surg 2001 233 9 17 11141219 10.1097/00000658-200101000-00003
Itani SI Ruderman NB Schmieder F Boden G Lipid-induced insulin resistance in human muscle is associated with changes in diacylglycerol, protein kinase C, and IκBα Diabetes 2002 51 2005 11 12086926
|
16135250
|
PMC1208952
|
CC BY
|
2021-01-04 16:37:45
|
no
|
Nutr Metab (Lond). 2005 Aug 31; 2:21
|
latin-1
|
Nutr Metab (Lond)
| 2,005 |
10.1186/1743-7075-2-21
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-921609296110.1186/1465-9921-6-92ResearchHealth-related quality of life in patients with pulmonary arterial hypertension Taichman Darren B [email protected] Jennifer [email protected] Laryssa [email protected] Christine [email protected] Sandra [email protected] Jeffery S [email protected] Robert [email protected] Jason [email protected] John [email protected] Harold [email protected] Pulmonary, Allergy and Critical Care Division, University of Pennsylvania School of Medicine, University of Pennsylvania Medical Center-Presbyterian, Philadelphia, PA 19104, USA2005 10 8 2005 6 1 92 92 21 6 2005 10 8 2005 Copyright © 2005 Taichman et al; licensee BioMed Central Ltd.2005Taichman et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Improved outcomes with expanding treatment options for patients with pulmonary arterial hypertension present the opportunity to consider additional end-points in approaching therapy, including factors that influence health-related quality of life. However, comparatively little is known about health-related quality of life and its determinants in patients with pulmonary arterial hypertension.
Methods
Health-related quality of life was evaluated in a cross sectional study of 155 outpatients with pulmonary arterial hypertension using generic and respiratory-disease specific measurement tools. Most patients had either World Health Organization functional Class II or III symptoms. Demographic, hemodynamic and treatment variables were assessed for association with health-related quality of life scores.
Results
Patients with pulmonary arterial hypertension suffered severe impairments in both physical and emotional domains of health-related quality of life. Patients with idiopathic ("primary") pulmonary arterial hypertension had the best, and those with systemic sclerosis the worst health-related quality of life. Greater six-minute walk distance correlated with better health-related quality of life scores, as did functional Class II versus Class III symptoms. Hemodynamic measurements, however, did not correlate with health-related quality of life scores. No differences in health-related quality of life were found between patients who were being treated with calcium channel antagonists, bosentan or continuously infused epoprostenol at the time of quality of life assessment.
Conclusion
Health-related quality of life is severely impaired in patients with pulmonary arterial hypertension and is associated with measures of functional status. Specific associations with impaired health-related quality of life suggest potential areas for targeted intervention.
==== Body
Introduction
Pulmonary arterial hypertension (PAH) is a devastating disease, characterized by progressive dyspnea and exercise limitation. If not effectively controlled, PAH often progresses to right heart failure and premature death [1-3]. Fortunately, recent dramatic advances in pharmacological treatment have brought about significant improvements in physical functioning and survival, and new therapeutic options are emerging at a rapid pace [4]. Improved outcomes with expanding treatment options present the opportunity to consider additional end-points in choosing approaches to therapy, including factors influencing health-related quality of life.
Presently, little is known about the determinants of health-related quality of life (HRQOL) in patients with PAH. An improved understanding of these determinants will be essential as HRQOL becomes an important outcome in research on therapeutics for PAH [5,6].We studied health-related quality of life in a large population of PAH patients and performed an initial assessment of the clinical factors associated with better (or worse) HRQOL.
Methods
Study Design and Patients
We conducted a cross sectional survey of HRQOL in patients cared for in the Pulmonary Vascular Disease Program at the University of Pennsylvania Health System. Previously established as well as newly referred patients with PAH were asked to complete HRQOL questionnaires over one year (October, 2002 – September, 2003). PAH was defined according to standard criteria, including a mean pulmonary artery pressure >25 mmHg at rest or 30 mmHg with exertion, and the absence of significant left heart dysfunction [7,8]. Patients completed questionnaires without assistance at regularly scheduled appointments prior to the physician evaluation. Physicians and nurses were not aware of HRQOL survey responses.
Patient demographics, symptoms, medications, and physical exam findings on the day of HRQOL evaluations were obtained by chart review, as were demographic values. Test results within three months of HRQOL assessment were recorded (e.g. hemodynamic values from cardiac catheterization in 103 patients and six-minute walk distance in 33 patients).
Health-Related Quality of Life Measurements
We administered a generic and a respiratory disease-specific quality of life questionnaire. The Medical Outcome Study 36 Item Short Form health survey (SF-36, standard form, version 2; Quality Metric, Lincoln, RI) yields scaled scores in 8 physical and mental health areas affected by health and disease conditions, in addition to physical component and mental component summary scores [9]. The St. George's Respiratory Questionnaire (SGRQ) is a well validated respiratory disease specific HRQOL measure, yielding scores related to symptoms (concerning the frequency and severity of respiratory symptoms), activity (the degree to which activities are limited by breathlessness), and impact (aspects of social and psychological function affected by respiratory disease). In addition, a total score summarizes disease impact on overall health status [10].
Statistical Analysis
Scores for the 8 domains and 2 summary measures of the SF-36, and the 4 scores of the SGRQ were calculated electronically according to published guidelines, including imputation for missing responses [9]. Higher SF-36 scores indicate better quality of life, whereas higher SGRQ scores indicate poorer HRQOL. SF-36 scores are normalized to a mean of 50 and standard deviation of 10, based on the normal US population. Mean raw scores on the SGRQ were compared with established normal population scores [11].
Data were entered into a Microsoft Access 5.1 database (Microsoft Corporation) and statistical analysis performed with SAS 8.2 software (SAS Institute, Cary, NC). Data were analyzed for differences using an independent t-test for binary predictors, ANOVA for categorical predictors, and correlation coefficients for continuous measures. Comparisons between patient groups were limited to the physical and mental component summary scores of the SF-36, and significant associations defined as those having a p ≤ 0.05.
Normalized physical and mental component summary scores were calculated from published studies of populations with various chronic diseases using reported individual SF-36 domain scores and an on-line scoring program .
This study was approved by the Institutional Review Board of the University of Pennsylvania.
Results
Patient Characteristics
Health-related quality of life was examined in 155 adult outpatients. The population was 81% female with an average age of 53 years, ranging from 18 to 84. Other physical and social characteristics of the study population are shown in Table 1. Thirty-three patients (21% of the sample) completed HRQOL questionnaires at the time of initial evaluation in a pulmonary vascular disease specialty program. Patients who had been followed for treatment of PAH for up to 1, 2 or 3 years accounted for 19, 25 and 14 percent of evaluations, respectively; the remaining 21% of patients had been followed longer than 3 years. Approximately one third of patients had idiopathic ("primary") pulmonary arterial hypertension and most had either World Health Organization Class II or III symptoms. The majority were married and only 14 percent lived alone. Thirty-five percent of the patients were employed.
Table 1 Baseline Characteristics
Physical Characteristics Social Characteristics
Patients, n 155 Marital Status
Age, years (mean ± SD) 53 ± 13 Single 25 (16)
Female, n (%) 126 (81) Married 100 (65)
Caucasian 106 (68) Separated 4 (3)
African-American 34 (22) Divorced 11 (7)
Hispanic 7 (5) Widowed 15 (10)
Other 8 (5) Lives Alone 21 (14)
PAH Diagnosis Currently employed 55 (35)
Idiopathic ("Primary") 63 (41) Occupation
Familial 2 (1) Prof/Exec 22 (14)
Systemic Sclerosis 32 (21) Manager 17 (11)
Other CVD 13 (8) Clerical 45 (29)
Portal Hypertension 11 (7) Skilled Labor 4 (3)
Anorectic Agent Use 7 (5) Unskilled Labor 16 (10)
Other (i.e., HIV, PVOD) 27 (17) Homemaker 12 (8)
Other 9 (6)
WHO I, % 3 (2) Not Available 30 (19)
WHO II, % 80 (52) Substance Abuse
WHO III, % 65 (42) Prior Alcohol 11 (7)
WHO IV, % 7 (5) Current Alcohol 0 (0)
Prior Smoking 66 (43)
Mean RA pressure (mmHg) 8.7 ± 2.6 Current Smoking 12 (8)
Mean PA pressure (mmHg) 47 ± 17
Cardiac output (liters/min) 4.9 ± 1.8
PVR (dynes·sec·cm-5) 630 ± 425 Treatments
Symptoms Epoprostenol 50 (32)
Dyspnea 105 (68) Bosentan 89 (57)
Fatigue, Weakness 55 (35) Calcium Channel Blocker 52 (34)
Lightheadedness/Pre-syncope 31 (20) Sildenafil 3 (2)
Chest Pain 31 (20) Spironolactone 40 (26)
Cough 27 (17) Other Diuretic(s) 76 (49)
Leg Pain 26 (15) Warfarin 74 (48)
Jaw Pain 15 (10) Digoxin 47 (30)
Depression 15 (10) Oxygen, continuous 29 (19)
Anxiety 14 (9) Oxygen, nocturnal 17 (11)
Palpitations 15 (10) Oxygen, as needed 15 (10)
Diarrhea 22 (14)
Abdominal Complaints 7 (6)
Leg Weakness 7 (6)
Difficulty Sleeping 3 (2)
Health-Related Quality of Life Scores
Patients with PAH had significantly depressed physical and mental health-related quality of life (Figure 1). On the SF-36, a "generic" measure of HRQOL applicable to a wide range of populations and disease states, HRQOL was impaired in every domain, representing a broad range of quality of life concepts (p < 0.001 for each). Mean scores were particularly depressed for the general health, physical functioning, role physical and emotional domains. In addition, the physical component summary (PCS) and mental component summary (MCS) scores, which broadly measure overall effects on HRQOL, were significantly decreased (p < 0.001 for each).
Figure 1 Health-Related Quality of Life Scores for Patients with Pulmonary Arterial Hypertension. Shown are mean (±SE) scores on each domain and summary component score of the SF36. Numerically higher scores indicate better health-related quality of life. All domain and summary scores are significantly lower than the US population normal score of 50 (p < 0.001 for each).
Quality of life was also assessed using a respiratory-disease specific instrument for comparison. Scores of patients with PAH were similarly abnormal on each component of the Saint George's Respiratory Questionnaire (Figure 2). Abnormally elevated scores (indicating a worse HRQOL) were seen in assessments of patient symptoms, activity, and the impact of disease on social and psychological function (p < 0.0001 for the comparison of each with normal). Further, PAH patients' mean total score, summarizing the effect of respiratory disease on overall health status, was markedly abnormal (p < 0.001).
Figure 2 Respiratory-Disease Specific Health-Related Quality of Life Scores in Patients with Pulmonary Arterial Hypertension. Mean (±SE) scores on the Saint George's Respiratory Questionnaire for patients with pulmonary arterial hypertension are compared with the normal population. Numerically lower scores indicate better health-related quality of life. P < 0.0001 for the comparison between normal scores and PAH patients for each.
Factors Associated with Impaired Quality of Life
Idiopathic pulmonary arterial hypertension (IPAH) was associated with better health-related QOL than other causes of PAH (Figure 3). Physical component summary scores of the SF-36 were highest for patients with IPAH. Patients with disease associated with systemic sclerosis, on the other hand, had the poorest physical QOL scores. Mental component summary QOL scores differed only for patients with anorectic agent associated PAH, whose scores were lowest (p = 0.02; not shown). World Health Association (WHO) Class II patients had better physical component summary scores than those with Class III symptoms (p < 0.001); small numbers of patients with Class I and IV symptoms precluded further comparisons. Active employment was the only additional demographic variable associated with better physical component summary scores (p < 0.0001 for the comparison with patients who were not employed). HRQOL scores were not associated with gender, race or age. Correlation coefficients for age with physical and mental component scores were -0.148 (p = 0.08) and 0.05 (p = 0.58), respectively.
Figure 3 Box and whisker plots of scores on the physical and mental component summary measures of the SF36 according to PAH diagnosis and World Health Organization (WHO) Functional Class. ** indicates p < 0.0001 for the difference between patients with idiopathic pulmonary arterial hypertension (IPAH) and systemic sclerosis (SSc) related PAH, and for the difference between WHO Class II and III patients.
Certain symptoms were associated with worsened HRQOL. Chest pain and pre-syncope were each associated with poorer physical component scores (p ≤ 0.02 for each), generalized fatigue with lower physical and mental component scores (p ≤ 0.03, each). Common side effects of epoprostenol therapy (diarrhea, jaw pain) were not related to further impairment in HRQOL. Patients who reported abdominal discomfort had significantly worse physical and mental component summary scores, and the presence of abdominal tenderness on physical exam was associated with worsened MCS scores (p ≤ for each). The only additional finding on physical examination associated with poorer HRQOL was the presence of peripheral edema, with which lower physical component summary scores (p ≤ 0.0001) and mental component scores (p = 0.02) were seen.
Therapies were also assessed. No differences in summary physical or mental component HRQOL scores were found between patients who were being treated with calcium channel antagonists, intravenous epoprostenol or bosentan at the time of HRQOL assessment (Figure 4). The use of diuretics or continuous oxygen was each associated with worsened physical summary scores. No differences were observed for the use/non-use of either warfarin or digoxin (not shown).
Figure 4 Box and whisker plots of scores on the physical and mental component summary measures of the SF36 according to treatments received at the time HRQOL assessment was performed. * indicates p ≤ 0.001 for the comparison between patients receiving or not receiving diuretics; ** p = 0.03 for the comparison between patients receiving or not receiving continuous oxygen therapy.
The distance walked in six minutes was significantly correlated with the physical component summary HRQOL score and approached significance in correlation with the mental component summary scores (Figure 5). The degree of perceived exertion reported by patients during exercise testing (Borg dyspnea index; higher reported exertion indicating more severe dyspnea) was inversely correlated with the physical component score (correlation coefficient -0.46; p = 0.02). Oxyhemoglobin saturation at rest also correlated with physical component summary scores (correlation 0.24, p = 0.005), but not mental component scores. No correlation was observed between hemodynamic measurements and HRQOL scores (Figure 6). Neither physical nor mental component summary scores correlated with the mean right atrial or pulmonary artery pressures, cardiac output or pulmonary vascular resistance.
Figure 5 Correlation of six-minute walk distance and health-related quality of life scores. Y-axes indicate scores on the physical and mental component summary scores of the SF-36.
Figure 6 Lack of correlation of hemodynamic values and health-related quality of life scores. RA, right atrial pressure (mmHg); PA, mean pulmonary artery pressure (mmHg); PVR, pulmonary vascular resistance (dyne·sec·m5), Cardiac output = liters/minute. Y-axes indicate the physical and mental component summary scores of the SF-36.
Discussion
Changes in health-related quality of life have been reported in trials of medications for pulmonary arterial hypertension. Improvements over pre-treatment scores have been seen in association with increased exercise capacity resulting from various therapies [12-16]. Focused assessments of health-related quality of life itself, however, are lacking and a systematic evaluation of the factors influencing it has not been previously reported in this patient population. A recent cross-sectional study of 53 patients with PAH reported moderate to severe impairments in multiple domains of HRQOL, both physical and emotional [17].
In the present study we have assessed a large population of patients with PAH and have identified demographic, symptom and treatment factors associated with better or poorer health-related QOL. Significantly impaired HRQOL was found in each of the eight domains of the SF-36, representing those physical, social and emotional characteristics generally accepted as most directly affected by health and disease [18,19]. Similarly, impaired HRQOL was observed on all scales of a respiratory-specific assessment tool (the Saint George's Respiratory Questionnaire). We found HRQOL to be best for patients with idiopathic PAH and worst for those with systemic sclerosis, unrelated to the type of vasodilator therapy used, and to be correlated with functional but not hemodynamic assessments.
While we are not surprised to find reduced health-related quality of life in patients with PAH, the severity of impairment is remarkable. Impaired HRQOL – both physical and emotional – is associated with many chronic and physically debilitating conditions. The impairments observed in our population of PAH patients are as severe (and in many respects more so) than those reported in studies of patients with other severely debilitating and life-threatening conditions such as spinal cord injury, interstitial lung disease or cancer unresponsive to therapy (Figure 7) [20-26]. Indeed, Shafazand et al. found their patients with PAH unhappy enough with their condition as to be willing to accept a significant risk of death in exchange for a potential cure [17].
Figure 7 Comparison of health-related quality of life scores between disease states. Shown are mean population scores for the physical and mental component summary scores of the SF-36. Scores shown are derived from previous reports of cardiac rehabilitation [20], metastatic prostate cancer unresponsive to therapy [21], bone marrow transplantation for breast cancer [22], survivors of acute respiratory distress syndrome [23], interstitial lung disease [24], chronic obstructive lung disease [25] and spinal cord injury [26]. Previously published scores were normalized for comparison, as described in methods. Data shown for patients with pulmonary arterial hypertension are from the present study (as in Figure 1), shown here for comparison. Y-axis indicates the physical and mental component summary scores of the SF-36.
Several important clinical parameters were associated with the degree of impairment in quality of life we observed, including diagnosis and exercise capacity. In addition to their poorer overall prognosis [27,28], we found that patients with PAH associated with systemic sclerosis experience worse health-related QOL than patients with other forms of the disease. Despite clear improvements in hemodynamics and exercise capacity as compared to untreated patients, those with systemic sclerosis do not benefit to the same degree as patients with IPAH in response to PAH-specific therapies [29,30]. Further investigations are required to determine whether these patients derive less benefit in HRQOL in response to these therapies, or whether other specific interventions can result in improvements.
The results of a six-minute walk test have been shown to correlate with each the WHO functional Class and patient survival [31]. In our sample, the distance walked in six minutes correlated with health-related quality of life. Scores worsened as exercise capacity declined. Our data similarly indicated worse scores in patients with WHO Class III symptoms as compared with Class II. The low number of patients with WHO Class I or IV symptoms in our population, however, limits further comparisons. By contrast, it is interesting that our data did not indicate a correlation with hemodynamic values. This suggests that a patient's overall functional status is of greater importance than actual hemodynamic values in determining HRQOL. Such findings, if confirmed, emphasize the importance of such functional endpoints in clinical trials, and may help direct choices of therapy so as to bring about the greatest improvement in a patient's sense of well-being. These findings, however, require confirmation with larger data sets that allow for HRQOL assessments both before and after cardiac catheterization, the results of which frequently change in response to therapy.
That abdominal complaints were associated with impaired HRQOL may be an important finding among patients with PAH. While abdominal discomfort might be caused by impaired splanchnic perfusion due to poor cardiac performance, it can also (perhaps more commonly) result from any of a number of processes common in non-PAH patients (e.g. gastroesophageal reflux and peptic ulcer disease). Even if not the result of PAH's effect on circulatory function itself, the significant association of abdominal discomfort with HRQOL noted here warrants further investigation. Similarly, while swelling and edema are a common symptom and finding among patients with advanced PAH and cor pulmonale, the additional impairment in HRQOL associated with diuretic use emphasizes the need to be mindful of appropriate titration to the degree required to prevent significant fluid collection.
No difference was found in HRQOL between patients who were receiving calcium channel antagonists, bosentan or epoprostenol therapies at the time HRQOL assessments were made. We cannot determine from the present study whether this reflects differences in the factors that determined the choice of therapy, or if these treatments do indeed result in equivalent HRQOL. While it might be expected that the need for continuous intravenous therapy would be associated with worse HRQOL, the inconveniences of such treatment might be off-set by the resultant substantial and sustained improvements in exercise capacity. Further, HRQOL reflects an individual's satisfaction with his or her life as it is affected by health. As such, HRQOL will be affected differently on the basis of individual desires, perceptions and expectations [17]. Epoprostenol therapy requires a motivated patient willing and able to provide a high degree of self-care. Such therapy may thereby select for individuals whose HRQOL is influenced to a greater degree by a sense of control over one's disease and its treatments. Additional prospective studies will be required to further evaluate these possibilities, and to determine whether choice of therapy can be expected to influence a patient's HRQOL.
There are several limitations of this study. Retrospective collection of patient symptoms and diagnoses may have resulted in misclassification bias and thereby influenced the observed effects of various factors on HRQOL. Missing items on HRQOL questionnaires may not be equally distributed among patients, and could therefore serve as an additional source of bias. Multi-centered studies will be required to be sure that the associations with reported symptoms and treatments found here do not overly reflect the practice styles of the physicians at this single specialized center. Larger prospective studies are required to better assess potential differences in HRQOL resulting from different treatments.
There is no currently available instrument designed specifically for the assessment of health-related quality of life in patients with pulmonary arterial hypertension [5]. We have used the SF-36, a non-disease specific instrument and the most widely used worldwide in studies of many individual disease populations [9]. While we have used the SGRQ (an instrument aimed a patients with respiratory disease) as a means to compare and provide some validation of the results seen with the SF-36, this too is limited in its ability to evaluate many items contributing to HRQOL in patients with pulmonary arterial hypertension. The SGRQ was developed to assess patients with COPD. As such, many of its questions address symptoms not typical of pulmonary arterial hypertension (e.g. wheezing and productive cough). Further, other symptoms attributable to cor pulmonale and commonly faced by patients with pulmonary arterial hypertension are not addressed in the SGRQ. While several HRQOL instruments have been developed for patients with left heart failure, these too have significant limitations in application to other populations and have not been validated for use in patients with cor pulmonale. Indeed, pulmonary arterial hypertension represents a hybrid of respiratory and cardiac symptoms, signs and treatments. Pepke-Zaba and colleagues have recently reported on the development of a PAH-specific HRQOL questionnaire [32]. The availability of such an instrument specifically developed for evaluation of patients with pulmonary arterial hypertension will be an important contribution to future studies aimed at better understanding and improving HRQOL [5].
Conclusion
Our study provides an initial systematic evaluation of health-related quality of life in a large population of patients with pulmonary arterial hypertension. We found profound impairment in all domains of HRQOL measured by both generic and respiratory-disease specific measures, and associations with assessments of functional status as well as specific demographic, symptom and treatment factors. As therapeutic options expand, the factors associated with impaired HRQOL identified here may suggest areas for further study and targeted intervention in efforts to improve outcomes for patients with pulmonary arterial hypertension.
List of Abbreviations Used
PAH: Pulmonary Arterial Hypertension
HRQOL: Health-Related Quality Of Life
SF-36: Medical Outcomes Study 36 Short Form
SGRQ: St. George's Respiratory Questionnaire
PCS: Physical Component Summary
MCS: Mental Component Summary
IPAH: Idiopathic Pulmonary Arterial Hypertension
WHO: World Health Organization.
Competing interests
Drs. Palevsky and Taichman have each served as consultants or on speaker's bureaus for Actelion Pharmaceuticals.
Authors' contributions
DBT, JS, LH and HP designed and SK coordinated this study. JS, LH and CAC performed chart reviews. DBT, RG, JHF, JSS and JC interpreted the data and DBT wrote the manuscript.
Acknowledgements
We thank Ms. Peggy Hegarty for expert assistance in the preparation of this manuscript. JS was supported by a medical student fellowship from FOCUS on Health & Leadership for Women at the University of Pennsylvania. DBT received support from a Development Partners' Junior Faculty Award from GSK Pharmaceuticals and from a Stephan A. Hansel Award from the Foundation for Pulmonary Hypertension.
==== Refs
Rubin LJ Primary pulmonary hypertension N Engl J Med 1997 336 111 117 8988890 10.1056/NEJM199701093360207
Rich S Dantzker DR Ayres SM Bergofsky EH Brundage BH Detre KM Fishman AP Goldring RM Groves BM Koerner SK Primary pulmonary hypertension. A national prospective study Ann Intern Med 1987 107 216 223 3605900
D'Alonzo GE Barst RJ Ayres SM Bergofsky EH Brundage BH Detre KM Fishman AP Goldring RM Groves BM Kernis JT Survival in patients with primary pulmonary hypertension. Results from a national prospective registry Ann Intern Med 1991 115 343 349 1863023
Rubin LJ Galie N Pulmonary arterial hypertension: a look to the future J Am Coll Cardiol 2004 43 89S 90S 15194184 10.1016/j.jacc.2004.02.030
Hoeper MM Oudiz RJ Peacock A Tapson VF Haworth SG Frost AE Torbicki A End points and clinical trial designs in pulmonary arterial hypertension: clinical and regulatory perspectives J Am Coll Cardiol 2004 43 48S 55S 15194178 10.1016/j.jacc.2004.02.010
Highland KB Strange C Mazur J Simpson KN Treatment of pulmonary arterial hypertension: a preliminary decision analysis Chest 2003 124 2087 2092 14665484 10.1378/chest.124.6.2087
McGoon M Gutterman D Steen V Barst R McCrory DC Fortin TA Loyd JE Screening, early detection, and diagnosis of pulmonary arterial hypertension: ACCP evidence-based clinical practice guidelines Chest 2004 126 14S 34S 15249493 10.1378/chest.126.1_suppl.14S
Barst RJ McGoon M Torbicki A Sitbon O Krowka MJ Olschewski H Gaine S Diagnosis and differential assessment of pulmonary arterial hypertension J Am Coll Cardiol 2004 43 40S 47S 15194177 10.1016/j.jacc.2004.02.032
Ware JEKMDJE How to Score Version 2 of the SF-36 Health Survey. 2000 Lincoln, RI, QualityMetric Incorporated 229 11302434
Jones PW Quirk FH Baveystock CM Littlejohns P A self-complete measure of health status for chronic airflow limitation. The St. George's Respiratory Questionnaire Am Rev Respir Dis 1992 145 1321 1327 1595997
Jones PW Spencer S Adie S The St. George's Respiratory Questionnaire Manual, Version 2.1 2003 London, St. George's Hospital Medical School
Olschewski H Simonneau G Galie N Higenbottam T Naeije R Rubin LJ Nikkho S Speich R Hoeper MM Behr J Winkler J Sitbon O Popov W Ghofrani HA Manes A Kiely DG Ewert R Meyer A Corris PA Delcroix M Gomez-Sanchez M Siedentop H Seeger W Inhaled iloprost for severe pulmonary hypertension N Engl J Med 2002 347 322 329 12151469 10.1056/NEJMoa020204
Barst RJ Rubin LJ Long WA McGoon MD Rich S Badesch DB Groves BM Tapson VF Bourge RC Brundage BH A comparison of continuous intravenous epoprostenol (prostacyclin) with conventional therapy for primary pulmonary hypertension. The Primary Pulmonary Hypertension Study Group N Engl J Med 1996 334 296 302 8532025 10.1056/NEJM199602013340504
Archibald CJ Auger WR Fedullo PF Channick RN Kerr KM Jamieson SW Kapelanski DP Watt CN Moser KM Long-term outcome after pulmonary thromboendarterectomy Am J Respir Crit Care Med 1999 160 523 528 10430723
Sastry BK Narasimhan C Reddy NK Raju BS Clinical efficacy of sildenafil in primary pulmonary hypertension: a randomized, placebo-controlled, double-blind, crossover study J Am Coll Cardiol 2004 43 1149 1153 15063421 10.1016/j.jacc.2003.10.056
Simonneau G Barst RJ Galie N Naeije R Rich S Bourge RC Keogh A Oudiz R Frost A Blackburn SD Crow JW Rubin LJ Continuous subcutaneous infusion of treprostinil, a prostacyclin analogue, in patients with pulmonary arterial hypertension: a double-blind, randomized, placebo-controlled trial Am J Respir Crit Care Med 2002 165 800 804 11897647
Shafazand S Goldstein MK Doyle RL Hlatky MA Gould MK Health-related quality of life in patients with pulmonary arterial hypertension Chest 2004 126 1452 1459 15539712 10.1378/chest.126.5.1452
Ware JEJ Gandek B Overview of the SF-36 Health Survey and the International Quality of Life Assessment (IQOLA) Project J Clin Epidemiol 1998 51 903 912 9817107 10.1016/S0895-4356(98)00081-X
Ware JEJ Sherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Cohen RA Moser DJ Clark MM Aloia MS Cargill BR Stefanik S Albrecht A Tilkemeier P Forman DE Neurocognitive functioning and improvement in quality of life following participation in cardiac rehabilitation Am J Cardiol 1999 83 1374 1378 10235098 10.1016/S0002-9149(99)00103-4
Albertsen PC Aaronson NK Muller MJ Keller SD Ware JEJ Health-related quality of life among patients with metastatic prostate cancer Urology 1997 49 207 16; discussion 216-7 9037282 10.1016/S0090-4295(96)00485-2
Hann DM Jacobsen PB Martin SC Kronish LE Azzarello LM Fields KK Quality of life following bone marrow transplantation for breast cancer: a comparative study Bone Marrow Transplant 1997 19 257 264 9028556 10.1038/sj.bmt.1700651
Davidson TA Caldwell ES Curtis JR Hudson LD Steinberg KP Reduced quality of life in survivors of acute respiratory distress syndrome compared with critically ill control patients JAMA 1999 281 354 360 9929089 10.1001/jama.281.4.354
Chang JA Curtis JR Patrick DL Raghu G Assessment of health-related quality of life in patients with interstitial lung disease Chest 1999 116 1175 1182 10559073 10.1378/chest.116.5.1175
Mahler DA Mackowiak JI Evaluation of the short-form 36-item questionnaire to measure health-related quality of life in patients with COPD Chest 1995 107 1585 1589 7781351
Lucke KT Coccia H Goode JS Lucke JF Quality of life in spinal cord injured individuals and their caregivers during the initial 6 months following rehabilitation Qual Life Res 2004 13 97 110 15058792 10.1023/B:QURE.0000015284.95515.17
Kawut SM Taichman DB Archer-Chicko CL Palevsky HI Kimmel SE Hemodynamics and survival in patients with pulmonary arterial hypertension related to systemic sclerosis Chest 2003 123 344 350 12576350 10.1378/chest.123.2.344
Kuhn KP Byrne DW Arbogast PG Doyle TP Loyd JE Robbins IM Outcome in 91 consecutive patients with pulmonary arterial hypertension receiving epoprostenol Am J Respir Crit Care Med 2003 167 580 586 12446266 10.1164/rccm.200204-333OC
Badesch DB Tapson VF McGoon MD Brundage BH Rubin LJ Wigley FM Rich S Barst RJ Barrett PS Kral KM Jobsis MM Loyd JE Murali S Frost A Girgis R Bourge RC Ralph DD Elliott CG Hill NS Langleben D Schilz RJ McLaughlin VV Robbins IM Groves BM Shapiro S Medsger TAJ Continuous intravenous epoprostenol for pulmonary hypertension due to the scleroderma spectrum of disease. A randomized, controlled trial Ann Intern Med 2000 132 425 434 10733441
Rubin LJ Badesch DB Barst RJ Galie N Black CM Keogh A Pulido T Frost A Roux S Leconte I Landzberg M Simonneau G Bosentan therapy for pulmonary arterial hypertension N Engl J Med 2002 346 896 903 11907289 10.1056/NEJMoa012212
Miyamoto S Nagaya N Satoh T Kyotani S Sakamaki F Fujita M Nakanishi N Miyatake K Clinical correlates and prognostic significance of six-minute walk test in patients with primary pulmonary hypertension. Comparison with cardiopulmonary exercise testing Am J Respir Crit Care Med 2000 161 487 492 10673190
Doughty NMKSPMDCFPZJ The CAMPHOR: Correlations with Objective Measures of Severity of Pulmonary Hypertension. Proceedings of the American Thoracic Society 2005 2 A801
|
16092961
|
PMC1208953
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Aug 10; 6(1):92
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-92
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-961610721510.1186/1465-9921-6-96ResearchFunctional, radiological and biological markers of alveolitis and infections of the lower respiratory tract in patients with systemic sclerosis De Santis Maria [email protected] Silvia [email protected] Torre Giuseppe [email protected] Anna [email protected] Barbara [email protected] Gabriella [email protected] Riccardo [email protected] Francesco Maria [email protected] Angelo [email protected] Gianfranco [email protected] Department of Rheumatology, Institute of Internal Medicine and Geriatrics, Catholic University of the Sacred Heart, 00168 Rome, Italy2 Unit of Epidemiology and Biostatistics, Institute of Hygiene, Catholic University of the Sacred Heart, 00168 Rome, Italy3 Department of Pulmonary Medicine, Institute of Internal Medicine and Geriatrics, Catholic University of the Sacred Heart, 00168 Rome, Italy4 Institute of Radiology, Catholic University of the Sacred Heart, 00168 Rome, Italy2005 17 8 2005 6 1 96 96 18 6 2005 17 8 2005 Copyright © 2005 De Santis et al; licensee BioMed Central Ltd.2005De Santis et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
A progressive lung disease and a worse survival have been observed in patients with systemic sclerosis and alveolitis. The objective of this study was to define the functional, radiological and biological markers of alveolitis in SSc patients.
Methods
100 SSc patients (76 with limited and 24 with diffuse disease) underwent a multistep assessment of cardiopulmonary system: pulmonary function tests (PFTs) every 6–12 months, echocardiography, high resolution computed tomography (HRCT) and bronchoalveolar lavage (BAL), if clinically advisable. Alveolar and interstitial scores on HRCT and IL-6 plasma levels were also assessed as lung disease activity indices.
Results
90 SSc patients with abnormal PFTs and 3 with signs and/or symptoms of lung involvement and normal PFTs underwent HRCT and echocardiography. HRCT revealed evidence of fibrosis in 87 (93.5%) patients, with 55 (59.1%) showing both ground glass attenuation and fibrosis. In 42 patients who had exhibited ground glass on HRCT and consented to undergo BAL, 16 (38.1%) revealed alveolitis. 12 (75%) of these patients had restrictive lung disease (p < 0.0001) and presented diffuse skin involvement (p = 0.0009). IL-6 plasma levels were higher in patients with alveolitis than in patients without (p = 0.041). On logistic regression model the best independent predictors of alveolitis were diffuse skin involvement (OR(95%CIs):12.80(2.54–64.37)) and skin score > 14 (OR(95%CIs):7.03(1.40–34.33)). The alveolar score showed a significant correlation with IL-6 plasma levels (r = 0.36, p = 0.001) and with the skin score (r = 0.33, p = 0.001). Cultures of BAL fluid resulted positive in 10 (23.8%) of the 42 patients that underwent BAL and after one year a deterioration in PFTs occurred in 8 (80%) of these patients (p = 0.01). Pulmonary artery systolic pressure ≥ 40 mmHg was found in 6 (37.5%) patients with alveolitis.
Conclusion
We found alveolitis only in 38.1% of the patients who had exhibited ground glass on HRCT and then underwent BAL, probably because the concomitant fibrosis influenced results. A diffuse skin involvement and a restrictive pattern on PFTs together with ground glass on HRCT were judged possible markers of alveolitis, a BAL examination being indicated as the next step. Nevertheless BAL would be necessary to detect any infections of the lower respiratory tract that may cause further deterioration in lung function.
==== Body
Background
There are two types of lung disease in systemic sclerosis (SSc): pulmonary interstitial fibrosis and pulmonary arterial hypertension (PAH) [1]. Additionally there are two subgroups of SSc patients with interstitial lung disease: patients whose lung function deterioration is either stable or shows slow progress and patients with progressive lung disease, frequent secondary vascular involvement and worse survival. Bouros et al reported in a large cohort of SSc patients that the most common histopathologic patterns of SSc lung, non specific interstitial pneumonia (NSIP) and usual interstitial pneumonia (UIP), showed few differences in five-year survival and concluded that the outcome in SSc patients was linked more closely to disease severity at presentation than to histopathologic findings [2]. Morgan et al reported the prognostic value of functional lung indices at the onset of disease and pointed out abnormal forced vital capacity (FVC) and diffusion capacity for carbon monoxide (DLCO) in the early stage of SSc as predictors of end-stage lung disease [3]. Bronchoalveolar lavage (BAL) showed a prognostic value in predicting increased mortality in SSc patients and can identify patients with alveolitis before extensive lung disease has developed allowing earlier intervention [4]. In addiction BAL procedure entails fewer risks and lower costs than lung biopsy and it is the recommended method for obtaining specimens from the lower airways [5,6]. It has been reported that BAL quantitative cultures can discriminate between subjects with and without lung infection with a power comparable or superior to all of the commonly accepted diagnostic tests [6].
We applied a multistep approach to define cardiopulmonary involvement in a cohort of 100 SSc patients: pulmonary function tests (PFTs) every 6–12 months and echocardiography, high resolution computed tomography (HRCT) and bronchoalveolar lavage (BAL) if they are clinically advisable. This study summarises our results in defining a diagnostic approach and analysing the characteristics of patients with alveolitis with the aim of identifying those factors that will potentially act as markers of alveolitis. An alveolar and interstitial score system on HRCT was applied and IL-6 plasma levels were also assessed to better characterize inflammatory and fibrotic lung involvement.
Methods
Patients
One hundred (92 females and 8 males) Italian SSc patients attending the outpatient clinic of the Division of Rheumatology of the Catholic University (Rome, Italy) in the last ten years were included in the study. Age (mean ± sd) of the SSc patients was 55.4 ± 11.9 years. The median disease duration was 6 years (range 3–13), duration being calculated as the time from the onset of the first clinical event that was a clear manifestation of SSc (other than Raynaud's phenomenon) to the time of data collection. All patients fulfilled the criteria proposed by the American College of Rheumatology [7], and were grouped according to the classification system proposed by LeRoy et al. [8] in patients with limited or diffuse skin involvement. Modified Rodnan Skin Score was performed for all patients [9]. ANA (antinuclear antibodies) were determined by indirect immunofluorescence using Hep-2 cells as substrates and autoantibodies specificities were further assessed by enzyme-linked immunosorbent assay (ELISA) (Shield, Dundee, UK) [10]. Plasma levels of IL-6 were examined by ELISA method, as described by manufacturer (Biosource, Nivelles Belgium); blood samples were taken from all the patients when HRCT or PFTs were carried out; IL-6 plasma levels in 32 healthy blood donors, matched for age and sex, were 0.31 ± 0.93 pg/ml. Patients were categorized as non-smokers (68%), current smokers (14%) or ex-smokers (18%), ex-smokers being defined as patients who had smoked a minimum of one cigarette a day for a minimum of one year and then stopped at least one year before presentation (Table 1).
Table 1 Demographic, clinical and immunological characteristics of 100 SSc patients
SSc patients
Age (years), mean ± sd 55.4 ± 11.9
Disease duration (years), median (I.Q. range) 6 (3–13)
Sex n (%) 92 F (92%) 8 M (8%)
Skin involvement:
dSSc n (%) 24 (24%)
lSSc n (%) 76 (76%)
Autoantibodies pattern:
ACA n (%) 41 (41%)
AntiScl 70 n (%) 38 (38%)
Antinucleolus n (%) 10 (10%)
AntiRNP n (%) 6 (6%)
ACA/Scl 70/nucleolus/RNP negative n (%) 5 (5%)
Smokers n (%) 14 (14%)
Ex-smokers n (%) 18 (18%)
SSc: systemic sclerosis; dSSc: diffuse skin involvement; lSSc: limited skin involvement; ACA: anticentromere antibodies; antiScl 70: antitopoisomerasi I antibodies; antiRNP: antiribonucleoproteins antibodies.
Therapy
All patients received Iloprost (by an infusion of 0.5–2 ng/kg/minute, lasting 6 hours, for 5 days every six months), Ca-channel blockers (nifedipine 20–40 mg/day) and D-Penicillamine (150 mg/day) from the moment of the medical diagnosis. A wash out period of three months elapsed between an Iloprost course and either PFTs or echocardiography or BAL.
Pulmonary function tests
All SSc patients underwent PFTs every 6–12 months. PFTs were performed to define FVC and DLCO. FVC was measured using a light bell spirometer in sitting patients wearing nose clip. DLCO was measured using the single breath technique, with 10 seconds breath holding time. All measurements were performed according to the American Thoracic Society recommendations [11] and expressed as percent of predicted values based on age, sex and height [12-14]. Lung involvement evaluated with PFTs was defined as "normal" when FVC and DLCO were ≥ 80%, "mild" when FVC was 70–79% / DLCO was 70–79%, "moderate" when FVC was 50–69% / DLCO was 50–69% and "severe" when FVC was < 50% / DLCO was < 50%, using the assessment of disease severity and prognosis of SSc patients proposed by Medsger et al [15]. Clinically significant restrictive lung disease was defined when an abnormal FVC with normal FEV1/FVC was observed. At one year follow-up we considered a reduction in FVC and/or DLCO >10% as a deterioration in PFTs.
High resolution computed tomography score system
Patients with abnormal PFTs and/or signs or symptoms of lung involvement (persistent cough, dyspnea on exertion, low degree fever or bilateral crackles) underwent HRCT and echocardiography. PFTs, echocardiography and HRCT examinations were performed in a time frame of three months each from the others. HRCT was performed with 1.0 mm thick sections taken at 10 mm intervals throughout the entire thorax and reconstructed using a spatial frequency algorithm. All images were obtained at the suspended end-inspiratory volume in the supine position. In a limited number of cases showing opacity in the postero-basal segments, sections were acquired also with the patient prone, to ensure that results were not affected by gravity. Three independent readers scored ground glass opacity (alveolar score) and honeycombing (interstitial score) as reported by Kazerooni et al [16] on a scale of 0–5 in the three lobes of both lungs as follows: 0- no alveolar disease, 1- ground glass involving < 5% of the lobe, 2- ground glass involving up to 25% of the lobe, 3- ground glass involving 25–49% of the lobe, 4- ground glass involving 50–75% of the lobe, 5- ground glass involving > 75% of the lobe for alveolar score; 0- no interstitial disease, 1- septal thickening without honeycombing, 2- honeycombing involving up to 25% of the lobe, 3- honeycombing involving 25–49% of the lobe, 4- honeycombing involving 50–75% of the lobe, 5- honeycombing involving > 75% of the lobe for interstitial score. Each observer assessed the extent of involvement in each of 3 defined regions: above aortic arch, between arch and inferior pulmonary veins and between inferior pulmonary veins and lung base. The mean estimate of the three readers was used to define the interstitial and alveolar score for each lobe. The scores were also summed into an overall interstitial and alveolar score. Alveolar or interstitial score ≥ 2 was used to define lung involvement.
Echocardiography
Pulmonary artery systolic pressure (PASP) was assumed to be equal to the right ventricular systolic pressure (RVSP) when there was no obstruction of the right ventricular outflow. RVSP was calculated with the simplified Bernoulli equation using the maximum peak of tricuspid valve regurgitation velocity (V) and right atrial pressure (RAP) assumed to be 10 mmHg (RVSP = 4V2+RAP). To account for the expected increase in PASP with aging, PAH was considered present if PASP exceeded 40 mmHg [17]. PAH was considered secondary to interstitial lung disease in patients with restrictive pattern on PFTs and/or interstitial score ≥ 2 in at least one lobe on lung HRCT.
Bronchoalveolar lavage analysis
If patients had FVC and/or DLCO ≤ 79% and an alveolar score ≥ 2, they were asked to consent to a BAL examination. The site chosen was generally the one that appeared the most affected on the HRCT analysis. BAL was performed under topical anesthesia (lidocaine 2%, 5–10 ml) without premedication. Four 60 ml aliquots of saline fluid (37°C) were sequentially instilled. Total cell count was performed in a Burker chamber using an uncentrifuged specimen and the result expressed as cells/ml of recovered fluid. BAL fluid was cytocentrifuged for 5 minutes at 500 rpm and differential cell count was performed by light microscope examination of 500 nonephitelial cells after staining with May-Grunwald-Giemsa. The proportions of alveolar macrophages, lymphocytes, neutrophils and eosinophils were recorded. Alveolitis was diagnosed when the percentage of lymphocytes was ≥ 15% and/or neutrophils ≥ 5% and/or eosinophils ≥ 5% [18]. The first aliquot of BAL fluid was used for microbiological studies (bacteria, mycobacteria, fungi, parasites) [5]. 104 colony forming units/ml of BAL were considered significant amounts of bacterial growth [6].
Statistical analysis
Data were analyzed using SPSS 11.0 (SPSS. Chicago. IL-USA) and Prism software (Graph-Pad, S. Diego, CA 92121-USA). Categorical and quantitative variables were respectively described as numbers, percentage (%) and mean ± standard deviation (sd) or median and I.Q. range, according to data distribution. Mann-Whitney's test was used to compare continuous variable. Categorical variables were analysed using χ2 test or Fisher's test, depending on sample size restrictions and the Odd ratios (OR) with 95% confidence interval (95% CIs) were calculated. Spearman's rank correlation was used to evaluate the relationship between different disease parameters. A logistic regression model was used in order to determine the influence on the dependent variable "having alveolitis" by the independent variables that reached the value of p <0.25 at the univariate analysis. The values are expressed as OR (95% CI). The diagnostic values of the clinical variables were assessed by calculating the areas under the receiver operating characteristics (ROC) curves, which were used to assess the best cut-off points to identify the presence of alveolitis. The diagnostic accuracy was calculated by sensitivity and specificity. We used a stepwise procedure (backward elimination), following the method suggested by Hosmer and Lemeshow. The chi-square test and the Hosmer-Lemeshow test were used in order to assess the fitting of the model. A value of p < 0.05 was considered statistically significant.
Results
Clinical and immunological data
76 (76.0%) patients had limited (lSSc) and 24 (24.0%) diffuse skin involvement (dSSc). Anticentromere (ACA) was present in 41 patients (41.0%), antitopoisomerasi I (antiScl 70) in 38 (38.0%). 10 (10.0%) patients were antinucleolus positive, 6 (6.0%) were antiribonucleoproteins (antiRNP) positive and 5 (5.0%) were ACA/antiScl 70/antinucleolus/antiRNP negative (Table 1).
PFTs results
PFTs results showed that patients could be divided into 3 groups: patients with restrictive pattern, patients with isolated reduction in DLCO and patients with normal PFTs. Characteristics of the 3 groups are detailed in Table 2. 16 (16.0%) patients had FVC and DLCO ≤ 79%: 12 of these patients had FVC and DLCO ≤ 69% and 1 had FVC and DLCO ≤ 49%; 74 (74.0%) patients had FVC > 80% and DLCO ≤ 79%: 67 of these patients had DLCO ≤ 69% and 16 had DLCO ≤ 49%; 10 (10.0%) patients had normal PFTs. FVC and DLCO ≤ 79% was observed in 12 (50%) of the 24 dSSc patients and in 10 (41.7%) of these patients FVC and DLCO ≤ 69% occurred. Of the 76 lSSc patients, FVC and DLCO ≤ 79% was observed in 4 (5.3%) and FVC and DLCO ≤ 69% occurred in 2 (2.6%) (data not reported in table).
Table 2 PFTs results of 100 SSc patients
FVC ≤ 79%
DLCO ≤ 79%
16 patients FVC ≥ 80%
DLCO ≤ 79%
74 patients FVC ≥ 80%
DLCO ≥ 80%
10 patients
Age (years), mean ± sd 55.1 ± 12.4 56.1 ± 11.9 50.7 ± 11.4
Disease duration (years), median (I.Q. range) 7 (4 – 18) 6 (3 – 13) 4.5 (1 – 14.5)
Sex n (%) 14 F (87.5%)
2 M (12.5%) 68 F (91.9%)
6 M (8.1%) 10 F (100.0%)
0 M (0.0%)
Skin involvement:
dSSc n (%) 12 (75.0%) 12 (16.2%) 0 (0.0%)
lSSc n (%) 4 (25.0%) 62 (83.8%) 10 (100.0%)
Skin score mean ± sd 14.8 ± 10.7 9.9 ± 7.4 6.7 ± 4.6
Autoantibodies pattern:
ACA n (%) 2 (12.5%) 32 (43.2%) 7 (70.0%)
AntiScl 70 n (%) 11 (68.7%) 25 (33.8%) 2 (20.0%)
Antinucleolus n (%) 0 (0.0%) 9 (12.2%) 1 (10.0%)
AntiRNP n (%) 1 (6.3%) 5 (6.8%) 0 (0.0%)
ACA/Scl 70/nucleolus/RNP negative n (%) 2 (12.5%) 3 (4.0%) 0 (0.0%)
FVC (%) mean ± sd 65.0 ± 8.8 104.8 ± 16.3 122.0 ± 13.3
≤ 69% n (%) 12 (75.0%) 0 (0.0%) 0 (0.0%)
≤ 49% n (%) 1 (6.3%) 0 (0.0%) 0 (0.0%)
DLCO (%) mean ± sd 39.8 ± 11.8 59.0 ± 10.4 94.2 ± 12.6
≤ 69% n (%) 16 (100%) 67 (90.5%) 0 (0.0%)
≤ 49% n (%) 13 (81.3%) 16 (21.6%) 0 (0.0%)
FVC and DLCO ≤ 69% 12 (75.0%) 0 (0.0%) 0 (0.0%)
FVC and DLCO ≤ 49% 1 (6.3%) 0 (0.0%) 0 (0.0%)
PASP ≥ 40 mmHg n (%) 6 (37.5%) 5 (6.7%) 0 (0.0%)
≥ 60 mmHg n (%) 1 (6.3%) 1 (1.3%) 0 (0.0%)
PFTs: pulmonary function tests; FVC: forced vital capacity; DLCO: diffusing capacity for carbon monoxide; dSSc: diffuse skin involvement; lSSc: limited skin involvement; ACA: anticentromere antibodies; antiScl 70: antitopoisomerasi I antibodies; antiRNP: antiribonucleoproteins antibodies; PASP: pulmonary artery systolic pressure.
Functional, radiological and BAL analysis
90 patients with abnormal PFTs and 3 patients with clinical signs and/or symptoms of lung involvement and normal PFTs underwent echocardiography and HRCT (Table 3). PASP ≥ 40 mmHg was found in 11 (11.8%) patients, 2 patients had PASP ≥ 60 mmHg. In 7 (7.5%) patients elevated PASP was considered secondary to interstitial lung disease (see also Additional file 1). HRCT revealed evidence of fibrosis in 87 (93.5%) patients, while 55 (59.1%) patients had both ground glass attenuation and fibrosis. 42 patients, whose HRCT had exhibited ground glass attenuation, gave their informed consent to a BAL examination. Alveolitis, detected by the BAL cell criteria as reported above, was found in 16 (38.1%) of these patients. Percentage of neutrophils was ≥ 5% in all 16 patients with alveolitis. In 2 cases there was also a percentage of lymphocytes ≥ 15%, in 2 more cases percentage of eosinophils ≥ 5% was also present and in 1 case the percentage of the three cell lines was increased. Cultures of BAL fluid resulted positive in 10 (23.8%) patients. Microbiological analysis revealed: Candida in 3 cases, Candida, Aspergillus and Stenotrophomonas in 1 case, Candida and Haemophilus in 1 case, Haemophilus in 1 case, Fusarium oxysporium in 1 case, Neisseria in 1 case, Enterococcus faecalis in 1 case, Streptococcus pneumoniae in 1 case.
Table 3 Functional, radiological and BAL analysis.
100 SSc patients
FVC % mean ± sd 100.1 ± 22.1
≤ 79% n (%) 16 (16.0%)
≤ 69% n (%) 12 (12.0%)
≤ 49% n (%) 1 (1.0%)
DLCO % mean ± sd 59.4 ± 17.3
≤ 79% n (%) 90 (90.0%)
≤ 69% n (%) 83 (83.0%)
≤ 49% n (%) 29 (29.0%)
PASP ≥ 40 mmHg n (%)* 11 (11.8%)*
≥ 60 mmHg n (%)* 2 (2.2%)*
HRCT: interstitial score ≥ 2, n (%)* 87 (93.5%)*
mean total score ± sd 3.1 ± 4.2
HRCT: alveolar score ≥ 2, n (%)* 55 (59.1%)*
mean total score ± sd 5.7 ± 2.6
BAL: alveolitis n (%)** 16 (38.1%)**
N ≥ 5 % n (%)** 11 (68.8%)**
N ≥ 5 % and L ≥ 15 % n (%)** 2 (12.5%)**
N ≥ 5 % and E ≥ 5 % n (%)** 2 (12.5%)**
N ≥ 5 %, L ≥ 15 % and E ≥ 5 % n (%)** 1 (6.2%)**
N < 5 % and L ≥ 15 % and/or E ≥ 5 % n (%)** 0 (0.0%)**
BAL with positive cultures n (%)** 10 (23.8%)**
BAL: bronchoalveolar lavage; SSC: systemic sclerosis; FVC: forced vital capacity; DLCO: diffusing capacity for carbon monoxide; PASP: pulmonary artery systolic pressure; HRCT: high resolution computed tomography; N: neutrophils; L: lymphocytes; E: eosinophils.
*Echocardiography and HRCT performed in 93 SSc patients
** BAL performed in 42 SSc patients giving their informed consent
Characteristics of patients with alveolitis
Patients with alveolitis did not show significant differences in demographic characteristics and autoantibodies when compared to patients without alveolitis (Table 4). Diffuse skin involvement was present in 12 (75.0%) patients with alveolitis: patients with diffuse skin involvement and ground glass attenuation on HRCT were at higher risk of having alveolitis than patients with limited skin disease and ground glass with an odd ratio of 12.60 (95% CIs = 2.83–56.15, p = 0.0009 Fisher's test). Moreover the mean modified Rodnan skin score was higher in patients with alveolitis than in patients without (p = 0.038). Modified Rodnan skin score > 14 [9] was observed in 8 (50.0%) patients with alveolitis and in 5 (19.2%) without alveolitis (p = 0.036).
Table 4 Characteristics of patients with alveolitis
BAL: alveolitis
16 patients BAL: inactive
26 patients P
Age (years) mean ± sd 57.0 ± 9.8 57.1 ± 11.8 ns
Disease duration (years) median (I.Q. range) 7.5 (3–16) 5 (1.5–12) ns
Sex n (%) 1 (6.25%) 2 (7.7%) ns
Autoantibodies pattern:
ACA n (%) 2 (12.5%) 8 (30.8%) ns
AntiScl 70 n (%) 12 (75.0%) 11 (42.3%) ns
Antinucleolus n (%) 0 (0.0%) 4 (15.4%) ns
AntiRNP n (%) 1 (6.25%) 1 (3.8%) ns
ACA/Scl 70/nucleolus/RNP negative n (%) 1 (6.25%) 2 (7.7%) ns
Skin involvement:
dSSc n (%) 12 (75.0%) 5 (19.2%) 0.0009
lSSc n (%) 4 (25.0%) 21 (80.8%) 0.0009
Modified Rodnan Skin Score mean ± sd 16.3 ± 9.4 10.1 ± 9.1 0.038
Modified Rodnan Skin score >14 n (%) 8 (50.0%) 5 (19.2%) 0.036
PFTs: FVC mean (%) ± sd 76.1 ± 27.6 101.9 ± 17.0 0.0005
DLCO mean (%) ± sd 43.9 ± 16.6 54.8 ± 11.0 0.0086
FVC and DLCO ≤ 79% n (%) 12 (75.0%) 2 (7.7%) < 0.0001
FVC and DLCO ≤ 49% n (%) 10 (62.5%) 2 (7.7%) < 0.0001
FVC ≥ 80% and DLCO ≤ 79% n (%) 3 (18.8%) 24 (92.3%) < 0.0001
FVC ≥ 80% and DLCO ≤ 49% n (%) 10 (62.5%) 9 (34.6%) ns
FVC and DLCO ≥ 80% n (%) 1 (6.3%) 0 (0.0%) -
PASP ≥ 40 mmHg n (%) 6 (37.5%) 1 (3.8%) 0.008
≥ 60 mm Hg n (%) 1 (6.3%) 0 (0.0%) -
HRCT: interstitial score mean ± sd 6.9 ± 3.5 7.3 ± 1.9 ns
alveolar score mean ± sd 9.1 ± 5.3 5.0 ± 3.1 0.0095
C reactive protein mean (mg/l) ± sd 8.6 ± 10.8 6.6 ± 6.7 ns
IL-6 plasma levels (pg/ml) mean ± sd 6.0 ± 10.8 2.4 ± 4.1 0.041
BAL with positive microbiological culture n (%) 5 (31.3%) 5 (19.2%) ns
Smokers n (%) 1 (6.3%) 4 (15.4%) ns
Ex-smokers n (%) 3 (18.8%) 5 (19.2%) ns
BAL: bronchoalveolar lavage; ACA: anticentromere antibodies; antiScl 70: antitopoisomerasi I antibodies; antiRNP: antiribonucleoproteins antibodies; dSSc: diffuse skin involvement; lSSc: limited skin involvement; FVC: forced vital capacity; DLCO: diffusing capacity for carbon monoxide; PASP: pulmonary artery systolic pressure; HRCT: high resolution computed tomography.
The mean FVC (%) and the mean DLCO (%) were significantly lower in patients with alveolitis (p = 0.0005 for FVC, p = 0.0086 for DLCO, respectively vs patients without alveolitis), moreover alveolitis was found in 12 (75.0%) patients with restrictive pattern on PFTs (p < 0.0001 vs patients with restrictive pattern but without alveolitis) and in 3 (18.8%) patients with an isolated reduction in DLCO that underwent BAL. One patient with normal PFTs underwent HRCT because of a persistent cough: BAL, assessed because of the presence of ground glass on HRCT, revealed alveolitis. Data on PFTs one year before BAL examination were available only for 10 patients with subsequent diagnosis of alveolitis and for 15 patients without alveolitis. However there were no differences in the percentage of patients with a clinically significant reduction in FVC or DLCO.
The mean alveolar score was significantly higher (9.1 ± 5.3) in patients with alveolitis vs patients without (5.0 ± 3.1; p = 0.0095), while no differences were seen in the mean interstitial score (p = ns). PASP ≥ 40 mmHg was found in 6 (37.5%) patients with alveolitis and in 1 (3.8 %) patients without (p = 0.008); in 1 (6.3%) patient PASP was ≥ 60 mmHg.
C reactive protein (CRP) values tended to be higher in patients with alveolitis (8.6 ± 10.8 mg/l) than in patients without alveolitis (6.6 ± 6.7 mg/l), but the difference between the two groups was not statistically significant. IL-6 plasma levels were significantly higher in patients with alveolitis (6.0 ± 10.8 pg/ml) than in patients without (2.4 ± 4.1 pg/ml; p = 0.041).
Skin score (>14), skin involvement extent, autoantibodies pattern, FVC (≤ 79%), DLCO (≤ 49%), alveolar score on HRCT (> 6), IL-6 plasma levels (> 0.75 pg/ml) and CRP (> 5 mg/l) were the components of the logistic regression model for alveolitis. The cut-off values of the dependent variables considered in the analysis were based on ROC curves and are reported in table 5. The best independent predictors of alveolitis were the skin score > 14 (OR (95%CIs): 7.03 (1.4–34.33)) and diffuse skin involvement (OR(95%CIs): 12.80 (2.54–64.37)). If the skin involvement was excluded from the model IL-6 was seen to be the best independent marker of alveolitis (OR (95%CIs): 6.22 (1.37–38.37)).
Table 5 Diagnostic accuracy of the predictors of alveolitis
AUC (95% CI) p Cut-off value Se(%) Sp (%)
FVC 0.145 (-0.003 – 0.292) 0.000 80.5 77.8 100.0
DLCO 0.199 (0.058 – 0.340) 0.001 50.0 72.2 70.8
IL-6 0.694 (0.524 – 0.864) 0.045 0.75 80.0 60.9
Skin score 0.650 (0.473 – 0.826) 0.109 14.5 58.8 78.3
Alveolar score on HRCT 0.742 (0.591 – 0.893) 0.008 6.5 66.7 75.0
AUC: area under curve ROC; FVC: forced vital capacity; DLCO: diffusing capacity for carbon monoxide; HRCT: high resolution computed tomography.
The microbiological analysis of BAL fluid was positive in 5 (31.3%) of the patients with alveolitis and in 5 (19.2%) of the patients without (p = ns). IL-6 plasma levels were higher in patients with positive BAL fluid cultures (5.2 ± 4.8 pg/ml) than in patients without infections (3.3 ± 7.8 pg/ml; p = 0.003), and a similar trend was found in CRP serum levels (12.6 ± 13.0 mg/l vs 5.7 ± 5.7 mg/l; p = ns). Patients with positive BAL fluid cultures showed a deterioration in PFTs after one year in 8 (80.0%) cases (4 with alveolitis and 4 without, p = ns) vs 10 (31.3%) patients without infection (3 with alveolitis and 7 without, p = ns) despite antimicrobial treatment, thus suggesting that infection is a poor prognostic factor (OR(95% CIs):8.8 (1.60–49.02), p = 0.01). Instead deterioration in PFTs after one year was observed in 7 (43.8%) patients with alveolitis vs 11 (42.3%) patients without (p = ns).
Patients with ground glass on HRCT that refused the BAL examination (13 patients) did not show significant differences in demographic, clinical and functional features when compared to patients who underwent BAL; alveolar and interstitial scores, IL-6 plasma levels and CRP were lower in patients that did not perform BAL (see Additional file 2).
HRCT scores correlations
We found a correlation between alveolar and interstitial scores, assessed by the HRCT score system proposed by Kazerooni et al, and FVC (r = -0.51, p < 0.0001 for alveolar score; r = -0.32, p = 0.0016 for interstitial score, respectively) and DLCO (r = -0.53, p < 0.0001 for alveolar score; r = -0.35, p = 0.0006 for interstitial score, respectively) (Figure 1). In addiction, the alveolar score showed a statistically significant correlation with IL-6 plasma levels (r = 0.36, p = 0.0012) and skin score (r = 0.33, p = 0.0021) (Figure 2 and 3 respectively).
Figure 1 Correlations between HRCT scores and PFTs in SSc patients. FVC showed a significant correlation with both the alveolar score (r = -0.51, p < 0.0001) and the interstitial score (r = - 0.32, p = 0.0016). Similarly DLCO showed a significant correlation with the alveolar score (r = -0.53, p < 0.0001) and the interstitial score (r = -0.35, p = 0.0006).
Figure 2 Correlation between alveolar score and IL-6. The alveolar score showed a statistically significant correlation with IL-6 plasma levels (r = 0.36, p = 0.0012).
Figure 3 Correlation between alveolar score and skin score. The alveolar score showed a statistically significant correlation with the skin score (r = 0.33, p = 0.0021).
Discussion
Detecting alveolitis is an important diagnostic clue in assessing disease severity in SSc patients. A greater deterioration in pulmonary function, a larger extent of lung fibrosis on HRCT over time and an increased mortality have been reported in patients with untreated alveolitis [4,19]. Great differences in the prevalence of alveolitis have been noted in past studies (from 48% to 72%), even considering those with a high number of patients [4,19-21] probably because of the different ratios of patients with diffuse and limited disease. Moreover no clear correlation has been reported between lung function indices, ground glass on HRCT and the presence of alveolitis, even though patients with alveolitis seemed to have worse FVC and DLCO than patients without [20-22] and it has long been demonstrated that ground glass attenuation on HRCT is the probable result of an inflammatory process [23-25].
Abnormal PFTs, especially a decreased DLCO, are a common finding in SSc patients: 90% of our cohort had abnormal DLCO with or without a decrease in FVC. Clinically significant restrictive lung disease was seen only in 16% of our cohort but in 50% of patients with diffuse skin involvement, while moderate-severe restrictive lung disease occurred in 12% of the whole cohort and in 41.7% of the patients with diffuse skin involvement. Despite differences between cohorts, this is confirmed by previous studies: the prevalence of restrictive lung disease among SSc patients varies between 25% and 35%, and between 30% and 70% in patients with diffuse disease [20,26-29]. In our study PASP ≥ 40 mmHg was found in 11.8% of the patients with abnormal PFTs and/or signs or symptoms of lung involvement: in 7.5% of cases elevated PASP could be secondary to interstitial lung disease, less than reported in a large cohort of scleroderma patients in which a restrictive ventilatory defect was observed in 22% of the patients and secondary PAH in 18% of the patients [30]. In our cohort 93.5% of patients with abnormal PFTs had an interstitial score ≥ 2 on HRCT and 59.1% showed an alveolar score ≥ 2, but only 38.1% of the patients that underwent BAL because of ground glass attenuation on HRCT had alveolitis defined by the BAL cell criteria detailed above. In our study all patients with ground glass attenuation showed concomitant signs of fibrosis on HRCT, as previously reported in a significant percentage of SSc patients [31]. Ground glass attenuation with concomitant signs of fibrosis, such as traction bronchiectasis or a reticular pattern, does not always lead to the identification of an inflammatory process and this could explain our data. It has been reported that fine intralobular fibrosis increases lung density on HRCT resulting in ground glass attenuation that is indistinguishable from the HRCT appearance found in alveolitis or in any other inflammatory process which results in accumulation of inflammatory cells or oedema in the alveolar septa and air spaces, as occurs in infections [32]. In addiction in SSc lung, especially in the early stages of the disease, histopathologic studies have shown hypercellularity of the alveolar wall [33], oedema [34] and over-development of microvessels that are abnormal in both shape (multi-bubbles and intervascular fusion) and size in the alveolar septa and interstitium [35] resulting in increased capillary blood volume which could also cause ground glass attenuation [32].
The functional significance of the two major radiographic patterns of interstitial lung involvement, ground glass and fibrosis, has not been clarified in previous studies. No relation has been found between HRCT findings and parameters of disease severity, such as a decrease in DLCO and FVC [36] or biological markers. Moreover, when the extent of lung involvement was assessed without distinguishing ground glass and reticular pattern on HRCT, no relationship with parameters of lung function was found, except with DLCO [37,38]. This suggests that DLCO fails to discriminate between inflammatory and fibrotic lung involvement. Instead, when the extent of ground glass and fibrotic patterns were assessed separately, an inverse correlation with FVC and DLCO was found, as reported by Ooi et al [29]. Similar results were found in our study where the interstitial score and the alveolar score were assessed as described by Kazerooni et al [16]. The alveolar score also showed a significant correlation with the modified Rodnan skin score and IL-6 plasma levels, thus suggesting that patients with a greater extent of ground glass attenuation on HRCT had a more aggressive disease. In addition, higher IL-6 plasma levels suggest that inflammation could explain the more aggressive pulmonary disease in patients with alveolitis. Previous studies have identified IL-6 serum levels as a useful index of disease activity in SSc patients because of the correlation with the skin score and suggested a pathogenic role of IL-6 in skin fibrosis [39,40]. In our study the significant correlation between IL-6 plasma levels and alveolar score and the higher values of IL-6 found in patients with alveolitis seem to confirm the helpful role of IL-6 as a disease activity index. Further studies could clarify the IL-6 role in inflammatory pulmonary involvement in SSc patients.
When PFTs are abnormal, HRCT is therefore an essential second step to assess the extent of interstitial disease and to detect the presence of inflammation in SSc lung involvement but the presence of ground glass alone can identify alveolitis in less than 40% of cases. In our study patients with diffuse skin involvement and ground glass attenuation on HRCT were twelve times more likely to have alveolitis compared to patients with limited skin disease. Moreover 75% patients with alveolitis presented restrictive lung disease. In the logistic regression model the extent of skin involvement appeared as the best predictor of alveolitis. In fact when restrictive pattern on PFTs was considered together with severe reduction in DLCO (≤ 49%), the association with alveolitis disappeared.
Patients with alveolitis showed a more aggressive lung disease as indicated by a worse lung function, a greater extent of pulmonary involvement on HRCT and a higher frequency of PAH. Nevertheless a greater deterioration in pulmonary function at the one year follow-up was observed in patients with positive BAL fluid cultures. In a normal host recovering an infectious agent from the lower respiratory tract does not necessarily mean infection although the recovery of certain organisms is believed to be almost always pathologic and quantitative criteria for bacterial BAL fluid cultures interpretation are standardized [41]. Nevertheless infections of the lower respiratory tract constitute a risk factor for deterioration of pulmonary function especially in patients with interstitial lung disease and may be under-diagnosed in SSc patients with lung involvement. In a few studies BAL fluid cultures were performed and specific infectious agents have been reported [21]. In our cohort 23.8% of the patients that underwent BAL had positive BAL fluid cultures and in 50% of cases fungi or polymicrobial colonization were found. Since BAL has the highest sensitivity for detecting deep-seated fungal infections, quantitative culture techniques have not been investigated and it has not always been possible to distinguish colonization from infection when clinical signs and symptoms are not specific [42]. Antigen tests and PCR test on BAL samples will aid the diagnosis of infections of the lower respiratory tract [42]. Nevertheless, patients with positive BAL fluid cultures seems to be at high risk of faster lung function deterioration, as observed at the one year follow-up in our study. These data probably indicate that even colonization by infectious agents may be a risk factor in worsening of patients with interstitial lung disease or that colonization/infection is present in those patients who are more likely to get worse. Higher IL-6 plasma levels in patients with positive cultures suggest that inflammation could be the worsening factor.
Conclusion
Considering the limits of functional and radiographic procedures in identifying alveolitis, BAL appears to be an essential tool in characterizing patients at high risk of severe lung disease. In our study not as many patients as expected consented to the BAL examination but despite this, it is one of the largest studies presenting data on PFTs, HRCT score system and BAL examination in systemic sclerosis [2,4,35]. The data obtained lead us to believe that diffuse skin involvement and a restrictive pattern on PFTs together with ground glass on HRCT are possible markers of alveolitis to be followed by a BAL examination. Our data also suggest the importance of detecting infection or colonization of the lower respiratory tract that may lead to an even faster lung function deterioration.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MDS: Conceived the study, participated in the design, performed the study and drafted the manuscript.
SB: Conceived the study, participated in the design, performed the study and helped draft the manuscript.
GLT: Performed statistical analysis.
AC: Participated in the design and performed the study.
BT: Performed laboratory analysis and performed statistical analysis.
GP: Participated in the design and performed the study.
RP: Participated in the design and performed the study.
FMD: Participated in the design and performed the study.
AZ: Participated in the design and helped draft the manuscript.
GF: Conceived the study, participated in the design and co-ordination of the study and helped draft the manuscript.
Supplementary Material
Additional File 1
Correlations between PASP and FVC, DLCO and interstitial score. PASP showed statistically significant correlation with both FVC (r = -0.36, p = 0.006) and DLCO (r = -0.38, p = 0.004) but not with the interstitial score (r = 0.26, p = 0.055).
Click here for file
Additional File 2
Characteristics of patients refusing BAL. Patients with ground glass on HRCT that refused the BAL examination (13 patients) did not show significant differences in demographic, clinical and functional features when compared to patients who underwent BAL; alveolar and interstitial scores, IL-6 plasma levels and CRP were lower in patients that did not perform BAL.
Click here for file
Acknowledgements
Part of this study was presented in abstract form at the EULAR 2005.
==== Refs
Steen V Predictors of end stage lung disease in systemic sclerosis Ann Rheum Dis 2003 62 97 9 12525376 10.1136/ard.62.2.97
Bouros D Wells AU Nicholson AG Colby TV Polychronopoulos V Pantelidis P Haslam PL Vassilakis DA Black CM du Bois RM Histopathologic subset of fibrosing alveolitis in patients with systemic sclerosis and their relationship to outcome Am J Respir Crit Care Med 2002 165 1581 86 12070056 10.1164/rccm.2106012
Morgan C Knight C Lunt M Black CM Silman AJ Predictors of end stage lung disease in a cohort of patients with scleroderma Ann Rheum Dis 2003 62 146 50 12525384 10.1136/ard.62.2.146
White B Moore WC Wigley FM Xiao HQ Wise RA Cyclophosphamide is associated with pulmonary function and survival benefit in patients with scleroderma and alveolitis Ann Intern Med 2000 132 947 54 10858177
Klech H Pohl W Technical recommendations and guidelines for bronchoalveolar lavage (BAL). Report of the European Society of Pneumology Task Group on BAL Eur Respir J 1989 2 561 85 2663535
Baker AM Bowton DL Haponik EF Decision making in nosocomial pneumonia. An analytic approach to the interpretation of quantitative bronchoscopic cultures Chest 1995 107 85 95 7813319
Subcommittee for scleroderma criteria of the American Rheumatism Association diagnostic and therapeutic criteria committee Preliminary criteria for the classification of systemic sclerosis (scleroderma) Arthritis Rheum 1980 23 581 90 7378088
LeRoy EC Black C Fleischmajer R Jablonska S Krieg T Medsger TA JrRowell N Wollheim F Classification, subset and pathogenesis J Rheumatol 1988 15 202 5 3361530
Valentini G D'Angelo S Della Rossa A Bencivelli W Bombardieri S European Scleroderma Study Group to define disease activity criteria for systemic sclerosis. IV. Assessment of skin thickening by modified Rodnan skin score Ann Rheum Dis 2003 62 904 5 12922969 10.1136/ard.62.9.904
Reveille JD Solomon DH ACR ad hoc committee on immunologic testing guidelines Evidence-based guidelines fort he use of immunologic tests: anticentromere, Scl-70 and nucleolar antibodies Arthritis Rheum 2003 49 399 412 12794797 10.1002/art.11113
American Thoracic Society Standardization of Spirometry-1994 Update Am J Respir Crit Care Med 1995 152 1107 36 7663792
American Thoracic Society Single breath Carbon Monoxide Diffusing Capacity (Transfer Factor). Recommendation for a standard technique-1995 Update Am J Respir Crit Care Med 1995 152 2185 98 8520796
Paoletti P Viegi G Pistelli G Di Pede F Fazzi P Polato R Saetta M Zambon R Carli G Giuntini C Reference equations for the single breath diffusing capacity: a cross sectional analysis and effect of body size and age Am Rev Respir Dis 1985 132 806 13 3876793
Quanjer PH Tammeling GJ Cotes JE Pedersen OF Peslin R Vernault JC Lung volumes and forced ventilatory flows. Report Working Party Standardization of lung function tests, European Community for Steel and Coal. Official statement of the European Respiratory Society Eur Respir J 1993 Suppl 16 5 40 8381090
Medsger TA Bombardieri S Czirjak L Scorza R Della Rossa A Bencivelli W Assessment of disease severity and prognosis in SSc Clin Exp Rheumatol 2003 21 S42 6 12889222 12889222
Kazerooni EA Martinez FJ Flint A Jamadar DA Gross BH Spizarny DL Cascade PN Whyte RI Lynch JP 3rdToews G Thin-section CT obtained at 10 mm increments versus three-level thin-section CT for idiopathic pulmonary fibrosis: correlation with pathologic scoring AJR Am J Roentgenol 1997 169 977 83 9308447
McQuillan BM Picard MH Leavitt M Weyman AE Clinical correlates and reference intervals for pulmonary artery systolic pressure among echocardiographically normal subjects Circulation 2001 104 2797 2802 11733397
The Joint statement of the ATS and ETS Idiopathic pulmonary fibrosis: diagnosis and treatment. International Consensus Statement Am J Respir Crit Care Med 2000 161 646 64 10673212
Behr J Vogelmeier C Beinert T Meurer M Krombach F Konig G Fruhmann G Bronchoalveolar lavage for evaluation and management of scleroderma disease of the lung Am J Respir Crit Care Med 1996 154 400 6 8756813
Witt C Borges AC John M Fietze I Baumann G Krause A Pulmonary involvement in diffuse cutaneous systemic sclerosis: bronchoalveolar fluid granulocytosis predicts progression of fibrosing alveolitis Ann Rheum Dis 1999 58 635 40 10491363
Clements PJ Goldin JG Kleerup EC Furst DE Elashoff RM Tashkin DP Roth MD Regional differences in bronchoalveolar lavage and thoracic high-resolution computed tomography results in dyspneic patients with systemic sclerosis Arthritis Rheum 2004 50 1909 17 15188367 10.1002/art.20265
Silver RM Scott Miller K Kinsella MB Smith EA Schabel SI Evaluation and management of scleroderma lung disease using bronchoalveolar lavage Am J Med 1990 88 470 6 2337105 10.1016/0002-9343(90)90425-D
Muller NL Staples CA Miller RR Vedal S Thurlbeck WM Ostrow DN Disease activity in idiopathic pulmonary fibrosis: CT and pathologic correlation Radiology 1987 165 731 4 3685351
Wells AU Hansell DM Corrin B Harrison NK Goldstraw P Black CM du Bois RM High resolution computed tomography as a predictor of lung histology in systemic sclerosis Thorax 1992 47 738 42 1440471 1440471
Remy-Jardin M Remy J Wallaert B Bataille D Hatron PY Pulmonary involvement in progressive systemic sclerosis: sequential evaluation with CT, pulmonary function test and bronchoalveolar lavage Radiology 1993 188 499 506 8327704
Jacobsen S Halberg P Ullman S Van Venrooij WJ Hoier-madsen M Wiik A Petersen J Clinical features and serum antinuclear antibodies in 230 danish patients with systemic sclerosis Br J Rheumatol 1998 37 39 45 9487249 10.1093/rheumatology/37.1.39
Della Rossa A Valentini G Bombardieri S Bencivelli W Silman AJ D'Angelo S Matucci Cerinic M Belch JF Black CM Becvar R Bruhlman P Cozzi F Czirjak l Drosos AA Dziankowska B Ferri C Gabrielli A Giacomelli R Hayem G Inanc M McHugh NJ Nielsen H Scorza R Tirri E van der Hoogen FHJ Vlachoyiannopoulos PG European multicentre study to define disease activity criteria for systemic sclerosis. I Clinical and epidemiological features of 290 patients from 19 centres Ann Rheum Dis 2001 60 585 91 11350847 10.1136/ard.60.6.585
Steen VD Conte C Owens GR Medsger TA Jr Severe restrictive lung disease in systemic sclerosis Arthritis Rheum 1994 37 1283 9 7945490
Ooi GC Mok MY Tsang WT Wong Y Khong PL Fung PCW Chan S Tse HF Wong RWS Lam WK Lau CS Interstitial lung disease in systemic sclerosis. An HRCT-clinical correlative study Acta Rad 2003 44 258 64 10.1034/j.1600-0455.2003.00058.x
Chang B Wigley FM White B Wise RA Scleroderma patients with combined pulmonary hypertension and interstitial lung disease J Rheumatol 2003 30 2398 405 14677184
Desai SR Veeraraghavan S Nikolakopolou A Nicholson AG Colby TV Denton CP Black CM du Bois RM Wells AU CT features of lung disease in patients with systemic sclerosis: comparison with idiopathic pulmonary fibrosis and non-specific interstitial pneumonia Radiology 2004 232 560 7 15286324
Remy-Jardin M Giraud F Remy J Copin MC Gosselin B Duhamel A Importance of ground-glass attenuation in chronic diffuse infiltrative lung disease: pathologic-CT correlation Radiology 1993 189 693 698 8234692
Spencer H Lung changes in progressive systemic sclerosis Pathology of the Lung 1977 Oxford: Pergamon Press 741 46
D'Angelo WA Fries JF Masi AT Shulman LE Pathologic observations in systemic sclerosis (scleroderma). A study of fifty-eight autopsy cases and fifty-eight matched controls Am J Med 1969 46 428 40 5780367 10.1016/0002-9343(69)90044-8
Beon M Harley RA Wessels A Silver RM Ludwicka-Bradley A Myofibroblast induction and microvascular alteration in scleroderma lung fibrosis Clin Exp Rheumatol 2004 22 733 42 15638048
Wells AU Hansell DM Rubens MB Cullinan P Black CM Du Bois RM The predictive value of appearances on thin section computed tomography in fibrosing alveolitis Am Rev Respir Dis 1993 148 1076 82 8214928
Wells AU Hansell DM Rubens MB King AD Cramer D Black CM du Bois RM Fibrosing alveolitis in systemic sclerosis. Indices of lung function in relation to extent of disease on computed tomography Arthritis Rheum 1997 40 1229 36 9214422
Warrick JH Bhalla M Schabel SI Silver RM High resolution computed tomography in early scleroderma lung disease J Rheumatol 1991 18 1520 8 1765976
Stuart RA Littlewood AJ Maddison PJ Hall ND Elevated serum interleukin-6 levels associated with active disease in systemic connective tissue disorders Clin Exp Rheumatol 1995 13 17 22 7774098
Sato S Hasegawa M Takehara K Serum levels of interleukin-6 and interleukin-10 correlate with total skin thickness score in patients with systemic sclerosis J Dermatol Sci 2001 27 2 140 6 11532378 10.1016/S0923-1811(01)00128-1
Baughman RP Conrado CE Diagnosis of lower respiratory tract infections. What we have and what would be nice Chest 1998 113 219S 223S 9515896
Baselski VS Wunderink RG Bronchoscopic diagnosis of pneumonia Clin Microbiol rev 1994 7 4 533 58 7834604
|
16107215
|
PMC1208954
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Aug 17; 6(1):96
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-96
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-971612021710.1186/1465-9921-6-97ResearchSurfactant Protein-A inhibits Aspergillus fumigatus-induced allergic T-cell responses Scanlon Seth Thomas [email protected] Tatyana [email protected] Sonja [email protected] Yang [email protected] Elena N [email protected] Yaniv [email protected] Scott J [email protected] Michael F [email protected] Angela [email protected] Pulmonary, Allergy and Critical Care Division, Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, USA2 Institute for Environmental Medicine, University of Pennsylvania School of Medicine, Philadelphia, USA2005 24 8 2005 6 1 97 97 30 11 2004 24 8 2005 Copyright © 2005 Scanlon et al; licensee BioMed Central Ltd.2005Scanlon et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The pulmonary surfactant protein (SP)-A has potent immunomodulatory activities but its role and regulation during allergic airway inflammation is unknown.
Methods
We studied changes in SP-A expression in the bronchoalveolar lavage (BAL) using a murine model of single Aspergillus fumigatus (Af) challenge of sensitized animals.
Results
SP-A protein levels in the BAL fluid showed a rapid, transient decline that reached the lowest values (25% of controls) 12 h after intranasal Af provocation of sensitized mice. Decrease of SP-A was associated with influx of inflammatory cells and increase of IL-4 and IL-5 mRNA and protein levels. Since levels of SP-A showed a significant negative correlation with these BAL cytokines (but not with IFN-γ), we hypothesized that SP-A exerts an inhibitory effect on Th2-type immune responses. To study this hypothesis, we used an in vitro Af-rechallenge model. Af-induced lymphocyte proliferation of cells isolated from sensitized mice was inhibited in a dose-dependent manner by addition of purified human SP-A (0.1–10 μg/ml). Flow cytometric studies on Af-stimulated lymphocytes indicated that the numbers of CD4+ (but not CD8+) T cells were significantly increased in the parental population and decreased in the third and fourth generation in the presence of SP-A. Further, addition of SP-A to the tissue culture inhibited Af-induced IL-4 and IL-5 production suggesting that SP-A directly suppressed allergen-stimulated CD4+ T cell function.
Conclusion
We speculate that a transient lack of this lung collectin following allergen exposure of the airways may significantly contribute to the development of a T-cell dependent allergic immune response.
==== Body
Background
Surfactant protein (SP)-A belongs to a family of innate host defense proteins termed collectins because of the presence of collagenous and also lectin-like domains [1]. SP-A has a potent immunomodulatory function with the N-terminal and C-terminal portions of the molecule capable of either activating or suppressing functions of cells of the immune system [2].
Allergen-induced asthma is characterized by the activation and steering of T-cells to a "helper" (Th) 2-type inflammation with eosinophilia, high serum immunoglobulin (Ig) E levels, reversible airway obstruction and hyper responsiveness (AHR) [3]. We have previously shown that allergic AHR is associated with disruption of surfactant biophysics, changes in the hydrophobic surfactant protein B and C [4] and upregulation of the lung collectin SP-D [5,6].
While SP-A is the most abundant and most extensively studied surfactant protein in the lung immune homeostasis [7-9], it is unclear how this molecule is regulated or functions during allergic inflammation. Published studies have shown that SP-A can inhibit lymphocyte function in vitro [10-12] and that administration of this protein in murine models ameliorated allergic changes in the lung [13-15], indicating a potential protective role. However, its direct regulatory action has not been detailed. The results of this study indicate specific changes in SP-A expression during development of an allergen induced immune response and show that SP-A can directly modulate the function of Th lymphocytes suggesting a regulatory link between innate and adaptive immunity in allergic airway inflammation.
Methods
Murine model of Af sensitization
Sensitized mice were injected intraperitoneally (i.p.) with 20 μg of Af (Hollister Stier, Elkhart, IN) and 20 mg Al(OH)3 (Imject Alum; Pierce, Rockford, IL) in PBS (100 μl) on days 1 and 14, followed by intranasal (i.n.) challenge on day 27 with 25 μl of allergen extract: (12.5 μg Af in 21% glycerol/ PBS). To analyze the kinetics of SP-A protein and mRNA changes BAL fluid and lung tissue were collected before (0 h) and 1, 6, 12, 24, 48 and 72 h after intranasal (i.n.) treatment, using a separate animal group at each time point (n = 6 in each) as described previously [4]. These time points were selected on the basis of our previous experience with acute models of locally elicited inflammation in mice [4,16,17]. Af sensitized and vehicle challenged mice (n = 8), non sensitized, Af challenged mice (n = 6) and non sensitized, glycerol challenged mice (n = 8) were studied 12 h after treatment. Naïve mice (n = 8) were used to control for the effects of i.p. sensitization and/or glycerol challenge. Lung function, BAL inflammatory cell and cytokine profile and SP-A levels were not significantly different between naïve animals and the groups that received sensitization plus glycerol challenge or glycerol challenge alone (not shown). All experimental subjects in this study were under a protocol approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania.
Analysis of SP-A protein and mRNA expression
Lungs were lavaged with sterile saline as previously described [4-6]. SDS-PAGE of the surfactant samples was carried out using NuPAGE 10% Bis-Tris gels (Novex, San Diego, CA). Western blots were performed using rabbit polyclonal anti-SP-A. Each lane was loaded with 10 μg of total protein. To investigate the absolute SP-A content in the BAL in addition to Western blot analysis, we used an ELISA protocol as described previously in detail [5]. Total RNA was isolated from lungs and specific SP-A mRNA content was determined by Northern blot analysis as described previously [6,18].
Assessment of airway inflammation following intranasal challenge of Af-sensitized mice
In vivo measurement of lung function after Af challenge was performed in conscious, unrestrained, spontaneously breathing mice in a whole-body plethysmograph (Buxco Electronics Inc., Troy, NY) as described previously [4]. Airway function was quantified using the Enhanced Pause (Penh).
Analysis of total and differential cell count and cytokine profile of the BAL was carried out after lungs were lavaged with sterile saline as previously published [4,6,19]. Cytokine levels were measured in the cell free supernatant of the BAL by ELISA using antibodies and recombinant cytokines from PharMingen (San Diego, CA).
Preparation of SP-A
Human SP-A was purified from human surfactant collected previously from patients with pulmonary alveolar proteinosis (PAP) as previously published [20]. The purified material was analyzed for protein concentration, subjected to Coomassie Blue staining (Figure 4A) and Western blot analysis (Figure 4B) and stored in aliquots at -80°C. The biological activity of the purified SP-A was tested in stimulated human peripheral blood monocytes (PBMC) (Figure 4C).
Figure 4 SP-A has a significant inhibitory effect on antigen specific lymphocyte proliferation from mice sensitized and challenged with Af. (A): Baseline (unstimulated) proliferation of splenic lymphocytes. Endotoxin in concentration equivalent to the endotoxin content of the 10 μg/ml SP-A samples did not affect cell proliferation. (B): In vitro Af stimulated proliferation of cells from mice sensitized and challenged with Af was inhibited by presence of SP-A, in a dose-dependent manner. (C): The Af dose-response curve of sensitized lymphocytes was abrogated in the presence of SP-A (10 μg/ml). (D) Presence of SP-A inhibited murine lymphocytes stimulated with PMA (2 ng/ml) and ionomycin (100 ng/ml). Background cpm (from wells that contain no cells): 0–16. Mean ± SEM of n = 4 independent experiments presented, each performed in triplicates. * p < 0.05: 0 vs. 10 μg/ml
Cell Proliferation Assays
Mice were sacrificed and spleens were harvested 24 hours after treatment. Spleens were homogenized in 10 mL of sterile PBS and layered over 5 mL of Histopaque (Histopaque-1077, Sigma) followed by density gradient centrifugation at 400 × g for 30 minutes at 18°C. Cells were then washed and cultured at 2 × 106 cells ml-1 in U-bottom 96-well plates (Nunclon Delta Surface, Nunc, Denmark).
Af and PMA-Ionomycin (PI, Sigma-Aldrich, St. Louis, MO) were used as mitogens for the lymphocyte proliferation assay. Varying amounts of SP-A were added to stimulated lymphocytes. Mouse splenocyte cultures were incubated for 37°C in a humidified atmosphere with 5% CO2 for 96 h. At the 72 h time point, each well was pulsed with [3H] thymidine (0.2 mCi ml-1) for an additional 24 h at 37°C. Cells were then harvested on a PHD Harvester (Cambridge Technology, Cambridge, MA). [3H] thymidine incorporation was determined via liquid scintillation counting (Beckman Instruments, New York, NY). Samples were plated in triplicates and the experiments were repeated n = 4 times unless otherwise stated.
Human peripheral blood mononuclear cells were isolated from healthy volunteers using a standard protocol approved by the IRB of the University of Pennsylvania.
Flow cytometric analysis of lymphocytes
The effects of SP-A on lymphocyte proliferation was further assessed using CFSE labeling as previously described [21]. Surface staining was performed at the time of harvest using Phycoerythin conjugated anti-mouse CD3, Fluorescin (FITC) conjugated anti-mouse CD4, FITC conjugated rat IgG2b isotype control (all from Pharmingen) and Allophycocyanin conjugated anti-human CD8 (Caltag, Burlingame, CA). Data were acquired on a four-color, dual laser FACSCalibur (Becton Dickinson, San Jose, CA). For FACS analysis the gated cell populations were divided into four different generations (G1, G2, G3 and G4) according to their CFSE content. Isotype controls were used to distinguish positive and negative populations. The SP-A-treated samples were compared with, and data were expressed as the percentage of the non-treated samples.
Interleukin (IL)-4, IL-5 and IFN-γ assays
Real time PCR
A relative quantitation of lung IL-4, IL-5, and IFN-γ mRNA expression was performed by real time PCR using TaqMan(R) Gene Expression Assay and TaqMan(R) Universal PCR Master Mix: Mm00445259, Mm00439646, Mm00434204, and Mm00801778, respectively, (Applied Biosystems) according to the manufactures instructions. In brief three RNA samples isolated from the lungs per time points, were reverse transcribed using Superscript(TM) III First Strand Synthesis System (Invitrogen). Approximately 100 ng of cDNA was used per singleplex PCR reaction in a total volume of 25 μl. Cycling was performed on an ABI SDS-7000 and ABI SDS-7500 respectively with an initial denaturation step at 95°C for 10 min and 40 cycles of 15 sec. at 95°C and 1 min. at 60°C. Every sample was run in triplicate and β-2-microglobin (Mm00437762) was used as endogenous control. The Comparative Ct Method was used to analyze the results using ABI PRISM(R) 7900 SDS software. Results are expressed as fold difference (with range incorporating the standard deviation of the ddCt value into the fold difference calculation) relative to naïve animals.
In vitro release of cytokines from cultured lymphocytes
Splenic lymphocytes were cultured for 48 h as described. The supernatant was then removed and frozen at -80°C pending analysis. OptEIA mouse IL-4, IL-5 and IFN-γ ELISA kits (Pharmingen) were used.
Data analysis
Data were expressed as mean ± SEM. Time courses and dose responses were compared using ANOVA. To test differences between individual groups Student's t test assuming equal variances were performed. Correlations were investigated by regression analysis. A p value of < 0.05 was considered as significant. Data were analyzed with the Sigma Stat standard statistical package (Jandel Scientific).
Results
Allergic sensitization and challenge of Balb/c mice with Af induced a transient fall in the levels of SP-A in the BAL fluid
To study the relationship between changes in SP-A levels and the sequence of inflammatory events we used a model of single allergenic airway provocation [4]. Western blots were analyzed from the cell-free supernatant of the BAL fluid, using naïve control samples as standards that were run on each gel. Decreases in SP-A levels began at 1 h (88 ± 19% of naïve control) and reached the lowest values 12 h (26 ± 5%) after Af challenge. Recovery was observed 72 h later (135 ± 40 %) (Fig. 1A and 1B). To verify that the alterations in SP-A levels were specific at the 12 h time point, sensitized and Af-challenged mice (26.5 ± 5%) were compared with mice that received vehicle (glycerol) treatment instead of antigen (90.0 ± 19%; p < 0.05 n = 8, Fig 1C).
Figure 1 Allergen challenge of sensitized mice with Af induced a transient decrease in the BAL SP-A protein levels. (A) SP-A Western blots from two representative samples of BAL obtained 1, 6, 12, 24, 48 and 72 h after Af challenge. (B): SP-A protein from BAL (open circles) and lung tissue mRNA (closed circles); % of the naïve control values; n = 4–6. (C): SP-A protein levels measured 12 h after Af challenge Mean ± SEM of n = 8 each; ** p < 0.01 (Af/Af vs. Af/gly).
SP-A protein decreases were not associated with commensurate mRNA changes. Northern blot analysis on RNA extracted from the lung tissue of each group of sensitized mice showed that SP-A mRNA actually increased 12 h after allergen challenge in comparison with the 0 h controls (160 ± 6% vs. 106 ± 16 p < 0.05, n = 5, Fig 1B). The absolute BAL SP-A content (determined by ELISA) was decreased from 378 ± 46 to 239 ± 25 μg/lung (p < 0.05 n = 8) at 12 h and to 222 ± 22 μg/lung (p < 0.05 n = 8) 48 h after Af challenge.
Decreased SP-A is associated with allergic airway inflammation to Af 12 h after allergen provocation of sensitized mice
In order to examine the proinflammatory changes that were linked with the greatest fall in SP-A concentrations, we chose to study the 12 h time point. Following allergen provocation there were significant Penh increases 12 h after Af-challenge of sensitized mice (1.91 ± 0.36, n = 8) in comparison with glycerol challenge (0.88 ± 0.06, n = 8, p < 0.01). The glycerol challenge controls were not different from mice that received no sensitization or challenge (Fig 2A). Levels of Penh returned to near normal by 72 h (1.36 ± 0.1; n = 8). The cellular composition of BAL showed that 12 h after intranasal Af challenge there was a significant influx of inflammatory cells predominated by neutrophils (500 × 103 cells, p < 0.01, n = 8) and eosinophils (110 × 103 cells p < 0.01, n = 8) in comparison with glycerol challenged controls. These cell numbers were also reduced (1 ± 1 and 31 ± 19, respectively, n = 8) by the 72 h time point.
Figure 2 Single intranasal provocation of sensitized mice with Af induced increase in Penh, inflammatory cell influx and Th2 but not Th1 cytokine release, 12 h After intranasal challenge. (A): Lung function measured by Penh. (B): BAL total and differential cell counts were used to calculate the absolute cell number. MP: macrophages EP: eosinophils, NP: neutrophils, LC: lymphocytes. (C): BAL Cytokine levels measured by ELISA (pg/ml). Mean ± SEM of n = 8 each; ** p < 0.01 (Af/Af vs. Af/gly) (D): Lung tissue cytokine mRNA levels measured by real-time PCR (fold increase in comparison to naïve control levels). ** p < 0.01 (vs. 0 h)
Similarly to our previously characterized ovalbumin and Af-induced models of allergic airway inflammation [4,6,19], i.p. sensitization and i.n. Af challenge induced a local production of predominantly Th2-type cytokines. In comparison with glycerol challenged controls there was a significant release of IL-4 (559 pg/ml vs. 3 pg/ml p < 0.01), IL-5 (283 pg/ml vs. 10 pg/ml p < 0.001; Figure 2C) and Eotaxin (173 pg/ml vs. 10 pg/ml p < 0.001) but not IFN-γ 12 h after allergen challenge (Figure 2C). Regression analysis using data pooled from the sensitized/Af challenged, sensitized glycerol challenged and non sensitized glycerol challenged mice revealed a significant, negative correlation with each of these cytokines (SP-A vs. IL-4: r = -0.61, p < 0.05; SP-A vs. IL-5: r = -0.82, p = 0.01; SP-A vs. Eotaxin: r = -0.75, p < 0.01). There was no correlation between SP-A and IFN-γ (not shown). Further, by the 72 h time point after allergen challenge IL-4 (8.7 ± 2.7 pg/ml), IL-5 (10.3 ± 7.3 pg/ml) and Eotaxin (9.72 ± 3.5 pg/ml) returned to close to baseline. To verify that the amount of airway SP-A is negatively associated with Th2 cytokine production, we also performed quantitative real-time PCR analysis of IL-4, IL-5 and IFN-γ mRNA. These results showed that mRNA levels for the Th2 cytokines increased approximately 35 fold in comparison with naïve controls 12 h after allergen challenge and returned to control levels by 72 h (Figure 2D).
SP-A inhibits allergen-driven proliferation of lymphocytes in vitro
To test the hypothesis that SP-A exerts an inhibitory effect on Th2 processes, we used a model of in vitro allergen re-challenge of lymphocytes isolated from sensitized and control mice and treated the cells with SP-A purified from human PAP material. Figure 3 shows that following purification and endotoxin depletion the structural integrity (Fig 3A) and immunogenicity (Fig 3B) of SP-A remained intact. Purified, LPS-depleted human SP-A exhibited significant biological activity as assessed by its effects on proliferation of human PBMC (Fig 3C).
Figure 3 Purification and biological activity of SP-A from human PAP material. (A): Coomassie Blue staining. (B): Western blot comparison of endotoxin-depleted (ED) SP-A preparations (1 μg) with human PAP material (1 μg) used as a positive control. (C): Effect of purified human SP-A on normal human PBMC proliferation. Cells were stimulated with PMA (2 ng/ml) and ionomycin (100 ng/ml). Stimulation Index (cpm of stimulated cells divided by cpm of corresponding unstimulated cells). Mean ± SEM of n = 12
Splenic lymphocytes isolated from naïve mice and mice sensitized with Af were re-challenged in vitro with Af in the presence or absence of SP-A. 3H-thymidine uptake was expressed as cpm (counts per minute). The background cpm throughout our experiments ranged between 0 and 16. The SP-A preparation had < 0.07 endotoxin units/ml. This level is consistent with other cell culture SP-A preparations [10]. Furthermore, culturing murine lymphocytes in up to twice the amount of LPS (0.14 endotoxin units/ml) had no significant effect on lymphocyte function (Fig 4A). Baseline proliferation of naïve lymphocytes was not affected by SP-A (Figure 4A). It appears that an activated state of the lymphocytes is a prerequisite for inhibition by SP-A since sensitized cells stimulated with Af were markedly inhibited by SP-A in a dose-dependent manner (Fig 4B). Indeed greater antigenic stimulation resulted in greater suppression of proliferation by SP-A at 10 μg/ml (Fig 4C).
Suppression of T cell proliferation was not a result of sequestering Af in the culture because lymphocytes stimulated with PMA (10 ng/ml) and ionomycin (500 ng/ml) were inhibited by SP-A (10 μg/ml) to a similar extent as the Af-stimulated lymphocytes by 66 ± 4% (Figure 4D), suggesting a direct inhibitory action on activated cells.
SP-A inhibits Af-induced CD3/CD4+ Th2 cell function
To further confirm the effects of SP-A on the proliferating cells, we added an intracellular fluorescent dye 5,6-carboxy- fluorescein diacetate succinimidyl ester (CFSE) to the cell cultures. The amount of this dye is halved as the Af-stimulated lymphocytes divide, allowing individual generations to be detected using FACS analysis. Mononuclear cells were gated for positive CD3/CD4 or CD3/CD8 expression and marked as G1 (the highest CFSE content), G2, G3 and G4 population according to decreasing levels of CFSE expression. Cells incubated in the presence of 10 μg/ml of SP-A were compared with equivalent populations of cells stimulated in the absence of SP-A and the results are expressed as a percentage of the SP-A-free (100%) control. In comparison with medium treatment, (represented by the middle line) addition of SP-A increased the proportion of cells that remained in the G1 population in the samples stimulated by Af (Figure 5A). The proportion of the G1 cells in the SP-A treated CD3/CD4 populations were 95%, 118%, and 163% of the medium treated (SP-A-free) controls at 0, 1 and 10 μg/ml Af concentration, respectively. Thus, cells exposed to greater antigen stimulation appeared to be more susceptible to inhibition by SP-A. We did not observe a similar inhibitory effect by SP-A in the CD3/CD8 cells (not shown).
Figure 5 SP-A inhibited Af-induced CD4+ T cell proliferation and release of IL-4 and IL-5. (A): SP-A (10 μg/ml) increased the number of cells in the G1 (black bars) and decreased them in the G3-G4 populations of CD3/CD4+ lymphocytes (% change from the medium-treated samples). Average of 3 independent experiments. (B): Af-induced production of IL-4 and IL-5 by sensitized lymphocytes (S/Medium) was reduced in the presence of 10μg/ml of SP-A (S/SP-A) to levels similar to naïve lymphocytes (N/Medium). Lymphocytes were stimulated by Af (1 μg/ml). IL-4 (grey bars), IL-5 (white bars) and IFN-γ (black bars). * p < 0.05 S/SP-A vs. S/Medium. Mean ± SEM of n = 4 experiments performed in triplicates..
Addition of SP-A (10 μg/ml) to sensitized lymphocytes significantly inhibited IL-4 and IL-5 production both in the Af-induced (Figure 5B) and PMA/ionomycin stimulated (not shown) cultures (p < 0.05, n = 4). SP-A apparently also inhibited IFN-γ levels in the non-stimulated and in the PMA/ionomycin stimulated samples indicating a non-selective activity on cytokine production by stimulated T cells (not shown).
Discussion
Emerging evidence suggests that collectins provide a link between innate and adaptive immunity. The relationship of SP-A production to inflammatory changes in the lung however is poorly defined. This study reports that a transient decrease in airway SP-A levels is associated with mRNA transcription and release of Th2-type cytokines following allergen provocation. The results also demonstrate that this lung collectin has a direct negative regulatory effect on allergen-induced lymphocyte activation in vitro.
Studies investigating allergen-induced airway inflammation previously described either significant increases [22,23] or decreases of this surfactant protein [24]. One reason for the conflicting data may be that levels of SP-A undergo transitional changes during development of allergic inflammation. Our time course study confirmed that levels of SP-A in the BAL dramatically decreased to 25% of the original 12 h after Af challenge but completely recovered 72 h later. The exact mechanisms that regulate the changes of SP-A concentrations in the airways are uncertain. Since reduction in the absolute amounts of SP-A was only 42%, the effect of dilution by the extravasated plasma proteins 12 h after allergen challenge could have significantly added to the fall in the fractional SP-A concentrations. Nevertheless, while further studies are needed to unravel its intricate regulation [25], it appears that an early decrease in the airway SP-A protein levels is not compensated immediately by enhanced synthesis or release, creating a relative, transient deficiency in this protein.
SP-A changes were comparable to that of SP-B and SP-C, described previously in a similar model [4]. Interestingly, while SP-B and SP-C protein levels in the BAL closely matched their respective mRNA recovered from the lung tissue, the fall SP-A protein was not paralleled by mRNA decreases. Thus (unlike in SP-B and SP-C), there is a dissociation between SP-A mRNA and protein regulation during the acute phase of an inflammatory response in the lung. Such discrepancy was also observed in a study looking at the effects of secretagogue-induced stimulation of SP-A synthesis [26] and in an in vitro model where SP-A mRNA but not protein expression was induced by dexamethasone, cyclic AMP and IL-4 [27]. In addition to not promptly following changes in mRNA production, SP-A did not require constitutive protein synthesis indicating a long intracellular half life [27]. Taken together, our previous and current work suggests that there is a significant lag between transcriptional, translational and/or secretory processes in SP-A regulation. Although the implications of such SP-A deficiency are unclear, the regulatory effects on the adaptive immune function during development of an inflammatory response could be significant.
The inverse association we observed between levels of SP-A and the inflammatory changes 12 h after allergen provocation was used to generate our hypothesis that SP-A negatively regulates Th2 cytokine production. This hypothesis was then tested by directly adding SP-A to allergen-stimulated T cells. These results are in support of previous reports showing that administration of SP-A alleviated allergic inflammatory changes in mice (reviewed in [15]), however the underlying mechanisms remain unclear. Binding of inhaled allergens such as particles of Af [28,29] and house-dust mite [30] may be one way to interfere with the allergic response. A second mechanism could be an interference with the function of inflammatory effector cells as indicated by inhibitory effects of SP-A on chemokine release by activated eosinophils [13,31]. Further, recent research has highlighted the role of SP-A on adaptive immune functions such as antigen presentation by dendritic cells [32] and T cell activation [10-12]. SP-A inhibited T-cell proliferation induced by various non-specific stimulators through both an IL-2 independent and an IL-2 and T cell receptor dependent manner [12].
The mechanism of T cell susceptibility to inhibition by SP-A needs further clarifications. While the inhibitory effects of SP-A have been demonstrated in antigen-stimulated human PBMC [33] and more recently, in a murine system of splenocytes co-cultured with ovalbumin-specific T cell hybridomas [12], no study has examined the effect of SP-A on T cell proliferation in a mouse model of allergen-induced asthma in vitro. Here we present for the first time the inhibitory effects of SP-A on generational progression of sensitized lymphocytes to Af re-challenge in vitro. We show that this inhibition was dose-dependent. Further, inhibition of lymphocyte activation by SP-A was not due to toxic effects since baseline proliferation of the lymphocytes was not affected. In addition, Figure 5A demonstrates that CD3+/CD4+ lymphocytes had significantly inhibited proliferative ability (as measured by the CFSE content of the cells) in the presence of SP-A. In this FACS analysis study the dead cells were excluded and only the live cells were analyzed (an equal number of 10,000 events collected in every sample). These results suggest that SP-A has a direct effect in suppressing proliferation of cells. Previous studies similarly showed that neither SP-A nor SP-D affects cell viability or apoptosis in PMA and ionomycin-stimulated cells [12]. Similarly to a study using cells from human asthmatics [33], we also showed that the inhibitory effect of SP-A did not depend on interference with the antigen-T cell receptor interaction since it suppressed both antigen- and mitogen-stimulated models of proliferation. Our studies however further indicated that in antigen-stimulated cultures sensitized cells exposed to greater antigen concentration and thus induced to proliferate more vigorously, were more susceptible to inhibition by SP-A. In addition, in the presence of SP-A there were significantly more sensitized CD4+ (not CD8+) lymphocytes in the parental (G1) population at higher Af concentrations, confirming that the antigen-responsive, stimulated T helper cell population is the target of SP-A.
The effects of SP-A treatment on inhibiting Th2 cytokines were previously studied ex vivo [34]. Here we show that greater amount of airway SP-A was associated with a lesser extent of Th2 cytokine release in vivo and that there was a significant, negative correlation between SP-A and the amount of IL-4, IL-5 and eotaxin supporting the hypothesis that this collectin exerts an inhibitory effect on Th2 processes. Indeed addition of SP-A abolished production of the Th2 cytokines IL-4 and IL-5 by Af-stimulated lymphocytes. Thus, SP-A may act as a natural immunosuppressant in the lung with direct inhibitory actions on allergen-stimulated CD4+ Th2 lymphocytes.
Conclusion
A transient decrease in airway levels of SP-A was associated with transcription of mRNA and release of the Th2-type cytokines IL-4 and IL-5 but not IFN-γ. On the other hand, addition of SP-A to lymphocyte cultures from sensitized mice suppressed allergen-induced Th2-type immune response. We speculate that this lung collectin may play a protective role to prevent allergen-inhalation provoked immune reactions and propose the novel concept that a transient SP-A deficiency contributes to the pathogenesis of the allergic airway response.
List of abbreviations
Af: Aspergillus fumigatus
AHR: Airway hyperresponsiveness
BAL: Bronchoalveolar lavage
CFSE: 5,6-carboxyfluorescein diacetate succinimidyl ester
cpm: Count per minute
ELISA: Enzyme linked immunosorbent assay
FACS: Fluorescence activated cell sorter
FITC: Fluorescein isothiocyanate
Ig: Immunoglobulin
IL: Interleukin
i.n.: Intranasal
i.p.: Intraperitoneal
LPS: Lipopolysaccharide
MBL: Mannose binding lectin
SP: Surfactant protein
PAP: Pulmonary Alveolar Proteinosis
PI: PMA and ionomycin
PMA: Phorbol 12-myristate 13-acetate
PBMC: Peripheral Blood Mononuclear Cells
Th: T helper
Authors' contributions
Seth Thomas Scanlon carried out the surfactant protein analysis and the protein and phospholipids measurements. Tatyana Milovanova carried out the CFSE studies and FACS analysis. Sonja Kierstein performed the cytokine mRNA analysis. Yang Cao participated in the surfactant protein analysis and in the animal experiments. Elena N. Atochina analyzed the SP-A Western and Northern blots. Yaniv Tomer carried out the animal experiments. Scott J. Russo performed the mRNA extraction and Northern blots. Michael F. Beers participated in the design of the study and contributed to the manuscript writing with comments. Angela Haczku conceived of, designed and coordinated the study, and participated in its writing. All authors approved the manuscript.
Acknowledgements
The authors are greatly indebted for Dr. Milton D. Rossman (University of Pennsylvania) for his advice and help with the CFSE studies. This work was supported by ALA RG-144N and R01-AI055593 (AH); R01-HL59867 and R01-HL64520 (MFB)
==== Refs
Holmskov U Thiel S Jensenius JC Collections and ficolins: humoral lectins of the innate immune defense Annu Rev Immunol 2003 21 547 578 12524383 10.1146/annurev.immunol.21.120601.140954
Gardai SJ Xiao YQ Dickinson M Nick JA Voelker DR Greene KE Henson PM By binding SIRPalpha or calreticulin/CD91, lung collectins act as dual function surveillance molecules to suppress or enhance inflammation Cell 2003 115 13 23 14531999 10.1016/S0092-8674(03)00758-X
Larche M Robinson DS Kay AB The role of T lymphocytes in the pathogenesis of asthma J Allergy Clin Immunol 2003 111 450 63; quiz 464 12642820 10.1067/mai.2003.169
Haczku A Atochina EN Tomer Y Cao Y Campbell C Scanlon ST Russo SJ Enhorning G Beers MF The late asthmatic response is linked with increased surface tension and reduced surfactant protein B in mice Am J Physiol Lung Cell Mol Physiol 2002 283 L755 65 12225952
Atochina EN Beers MF Tomer Y Scanlon ST Russo SJ Panettieri RAJ Haczku A Attenuated allergic airway hyperresponsiveness in C57BL/6 mice is associated with enhanced surfactant protein (SP)-D production following allergic sensitization Respir Res 2003 4 15 14748931 10.1186/1465-9921-4-15
Haczku A Atochina EN Tomer Y Chen H Scanlon ST Russo S Xu J Panettieri RAJ Beers MF Aspergillus fumigatus-induced allergic airway inflammation alters surfactant homeostasis and lung function in BALB/c mice Am J Respir Cell Mol Biol 2001 25 45 50 11472974
Crouch E Wright JR Surfactant proteins a and d and pulmonary host defense Annu Rev Physiol 2001 63 521 554 11181966 10.1146/annurev.physiol.63.1.521
Haagsman HP Structural and functional aspects of the collectin SP-A Immunobiology 2002 205 476 489 12396009 10.1078/0171-2985-00148
Haagsman HP Interactions of surfactant protein A with pathogens Biochim Biophys Acta 1998 1408 264 277 9813361
Borron P McCormack FX Elhalwagi BM Chroneos ZC Lewis JF Zhu S Wright JR Shepherd VL Possmayer F Inchley K Fraher LJ Surfactant protein A inhibits T cell proliferation via its collagen-like tail and a 210-kDa receptor Am J Physiol 1998 275 L679 86 9755099
Borron P Veldhuizen RA Lewis JF Possmayer F Caveney A Inchley K McFadden RG Fraher LJ Surfactant associated protein-A inhibits human lymphocyte proliferation and IL-2 production Am J Respir Cell Mol Biol 1996 15 115 121 8679215
Borron PJ Mostaghel EA Doyle C Walsh ES McHeyzer-Williams MG Wright JR Pulmonary surfactant proteins A and D directly suppress CD3+/CD4+ cell function: evidence for two shared mechanisms J Immunol 2002 169 5844 5850 12421966
Cheng G Ueda T Nakajima H Nakajima A Arima M Kinjyo S Fukuda T Surfactant protein A exhibits inhibitory effect on eosinophils IL-8 production Biochem Biophys Res Commun 2000 270 831 835 10772911 10.1006/bbrc.2000.2515
Kishor U Madan T Sarma PU Singh M Urban BC Reid KB Protective roles of pulmonary surfactant proteins, SP-A and SP-D, against lung allergy and infection caused by Aspergillus fumigatus Immunobiology 2002 205 610 618 12396019 10.1078/0171-2985-00158
Hohlfeld JM Erpenbeck VJ Krug N Surfactant proteins SP-A and SP-D as modulators of the allergic inflammation in asthma Pathobiology 2002 70 287 292 12771511 10.1159/000070744
Louahed J Zhou Y Maloy WL Rani PU Weiss C Tomer Y Vink A Renauld J Van Snick J Nicolaides NC Levitt RC Haczku A Interleukin 9 promotes influx and local maturation of eosinophils Blood 2001 97 1035 1042 11159534 10.1182/blood.V97.4.1035
Tsicopoulos A Hamid Q Haczku A Jacobson MR Durham SR North J Barkans J Corrigan CJ Meng Q Moqbel R Kinetics of cell infiltration and cytokine messenger RNA expression after intradermal challenge with allergen and tuberculin in the same atopic individuals J Allergy Clin Immunol 1994 94 764 772 7930311 10.1016/0091-6749(94)90185-6
Atochina EN Beck JM Scanlon ST Preston AM Beers MF Pneumocystis carinii pneumonia alters expression and distribution of lung collectins SP-A and SP-D J Lab Clin Med 2001 137 429 439 11385364 10.1067/mlc.2001.115220
Haczku A Takeda K Hamelmann E Loader J Joetham A Redai I Irvin CG Lee JJ Kikutani H Conrad D Gelfand EW CD23 exhibits negative regulatory effects on allergic sensitization and airway hyperresponsiveness Am J Respir Crit Care Med 2000 161 952 960 10712348
Fisher AB Dodia C Chander A Beers MF Bates SR Inhibition of Trimeresurus flavoviridis phospholipase A2 by lung surfactant protein A (SP-A) Biochim Biophys Acta 1994 1211 256 262 8130257
Milovanova TN Popma SH Cherian S Moore JS Rossman MD Flow cytometric test for beryllium sensitivity Cytometry 2004 60B 23 30 15221866 10.1002/cyto.b.20015
Cheng G Ueda T Numao T Kuroki Y Nakajima H Fukushima Y Motojima S Fukuda T Increased levels of surfactant protein A and D in bronchoalveolar lavage fluids in patients with bronchial asthma Eur Respir J 2000 16 831 835 11153579 10.1183/09031936.00.16583100
Haley KJ Ciota A Contreras JP Boothby MR Perkins DL Finn PW Alterations in lung collectins in an adaptive allergic immune response Am J Physiol Lung Cell Mol Physiol 2002 282 L573 84 11839554
Wang JY Shieh CC Yu CK Lei HY Allergen-induced bronchial inflammation is associated with decreased levels of surfactant proteins A and D in a murine model of asthma Clin Exp Allergy 2001 31 652 662 11359435 10.1046/j.1365-2222.2001.01031.x
Alcorn JL Mendelson CR Trafficking of surfactant protein A in fetal rabbit lung in organ culture Am J Physiol 1993 264 L27 35 8430814
Wali A Beers MF Dodia C Feinstein SI Fisher AB ATP and adenosine 3',5'-cyclic monophosphate stimulate the synthesis of surfactant protein A in rat lung Am J Physiol 1993 264 L431 7 8388646
Cao Y Tao JQ Bates SR Beers MF Haczku A IL-4 induces production of the lung collectin surfactant protein-D J Allergy Clin Immunol 2004 113 439 444 15007344 10.1016/j.jaci.2003.11.031
Allen MJ Harbeck R Smith B Voelker DR Mason RJ Binding of rat and human surfactant proteins A and D to Aspergillus fumigatus conidia Infect Immun 1999 67 4563 4569 10456901
Madan T Kishore U Shah A Eggleton P Strong P Wang JY Aggrawal SS Sarma PU Reid KB Lung surfactant proteins A and D can inhibit specific IgE binding to the allergens of Aspergillus fumigatus and block allergen-induced histamine release from human basophils Clin Exp Immunol 1997 110 241 249 9367408
Wang JY Kishore U Lim BL Strong P Reid KB Interaction of human lung surfactant proteins A and D with mite (Dermatophagoides pteronyssinus) allergens Clin Exp Immunol 1996 106 367 373 8918587 10.1046/j.1365-2249.1996.d01-838.x
Cheng G Ueda T Nakajima H Nakajima A Kinjyo S Motojima S Fukuda T Suppressive effects of SP-A on ionomycin-induced IL-8 production and release by eosinophils Int Arch Allergy Immunol 1998 117 Suppl 1 59 62 9758900 10.1159/000053574
Brinker KG Garner H Wright JR Surfactant protein A modulates the differentiation of murine bone marrow-derived dendritic cells Am J Physiol Lung Cell Mol Physiol 2003 284 L232 41 12388334
Wang JY Shieh CC You PF Lei HY Reid KB Inhibitory effect of pulmonary surfactant proteins A and D on allergen-induced lymphocyte proliferation and histamine release in children with asthma Am J Respir Crit Care Med 1998 158 510 518 9700129
Madan T Kishore U Singh M Strong P Clark H Hussain EM Reid KB Sarma PU Surfactant proteins A and D protect mice against pulmonary hypersensitivity induced by Aspergillus fumigatus antigens and allergens J Clin Invest 2001 107 467 475 11181646
|
16120217
|
PMC1208955
|
CC BY
|
2021-01-04 16:23:25
|
no
|
Respir Res. 2005 Aug 24; 6(1):97
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-97
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-221596702910.1186/1742-4682-2-22ResearchThe ups and downs of biological timers Rappaport Noa [email protected] Shay [email protected] Naama [email protected] Departments of Molecular Genetics and Physics of Complex systems, Weizmann Institute of Science, Rehovot, Israel2005 20 6 2005 2 22 22 3 4 2005 20 6 2005 Copyright © 2005 Rappaport et al; licensee BioMed Central Ltd.2005Rappaport et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The need to execute a sequence of events in an orderly and timely manner is central to many biological processes, including cell cycle progression and cell differentiation. For self-perpetuating systems, such as the cell cycle oscillator, delay times between events are defined by the network of interacting proteins that propagates the system. However, protein levels inside cells are subject to genetic and environmental fluctuations, raising the question of how reliable timing is maintained.
Results
We compared the robustness of different mechanisms for encoding delay times to fluctuations in protein expression levels. Gradual accumulation and gradual decay of a regulatory protein have an equivalent capacity for defining delay times. Yet, we find that the former is highly sensitive to fluctuations in gene dosage, while the latter can buffer such perturbations. In particular, a positive feedback where the degrading protein auto-enhances its own degradation may render delay times practically insensitive to gene dosage.
Conclusion
While our understanding of biological timing mechanisms is still rudimentary, it is clear that there is an ample use of degradation as well as self-enhanced degradation in processes such as cell cycle and circadian clocks. We propose that degradation processes, and specifically self-enhanced degradation, will be preferred in processes where maintaining the robustness of timing is important.
==== Body
Background
Protein levels within cells are subject to genetic and environmental variations, but mechanisms have evolved that buffer cellular processes against those fluctuations [1]. Quantitative analysis has indicated that the need to ensure robustness can largely restrict the design of the underlying network [2-4]. Maintaining a reliable sequence of events appears straightforward in cases where the completion of one event directly triggers the next [5]. Often, however, temporal cascades are propagated by a self-sustained biochemical network, which functions even in the absence of feedback signals [6]. For example, the cell cycle is governed by an autonomous oscillator, although this oscillator is executively sensitive to checkpoint signals that may halt its progression [7]. Similarly, while the circadian timing is synchronized by light or temperature, it oscillates as well under constant conditions [8]. The prevalence of self-sustained networks that coordinate temporal cascades suggests that at least in certain cases not only is the temporal order important, but also the relative timing of events needs to be maintained. However, whether mechanisms that code for delay times can also buffer those times against fluctuations has not yet been examined.
Strategies for coding delay times can be classified into two main categories, which are based on the accumulation or on the decay of some regulatory protein, or of its active form (Fig. 1A–B). A typical cascade employs both strategies, but it is not clear which is rate limiting for defining the timing. For example, in the budding yeast cell cycle, Cln3 accumulation is followed by Sic1 degradation, with both processes required for the G1/S transition and the initiation of START [9-12].
Figure 1 Strategies for coding delay times. A schematic description of each strategy is shown on the top panel, while the respective protein dynamics is shown at the bottom panel. The solid black line corresponds to some reference system, while the dashed black line corresponds to a system in which production rates were reduced two-fold. The time to reach the threshold (taken here as 10% of Pmax) is also shown. It can be seen that the delay time sensitivity is largest for accumulation and smallest for non-linear decay. Moreover, the location of the threshold is limited; threshold of 90% (light grey) will never be crossed by a perturbed system with η of less than 0.9. a, Accumulation strategy. In this case gene production is initiated at t = 0. Once a certain threshold is reached, downstream genes would be affected. b, Degradation strategy. In this case, protein production is stopped at t = 0. Once protein levels have decayed below a certain threshold, target genes would be affected. c, Same as b, except that degradation is non-linear with n = 2.
In this work, we compare the two strategies for coding delay times with respect to their capacity to buffer fluctuations in gene dosage.
Results and discussion
Consider a protein P that is present at a low level Plow. Accumulation of P can be initiated either by enhancing its production or by inhibiting its degradation. In either case, P will start accumulating toward a new steady state Pmax. Subsequent events will follow once P has passed some threshold PT, defined for example by its affinity to target genes promoters if P is a transcription activator. The corresponding delay time, T0, is defined by the time it takes to accumulate this threshold level PT of proteins. Alternatively, delay time can be encoded by an analogous system, where P decays from Pmax toward Plow, activating subsequent events once its levels are reduced below the threshold PT.
Comparing the robustness of time coding strategies
The above two strategies appear to be equivalent for defining delay times. For example, in the absence of feedback, where each protein is degraded independently, both accumulation and degradation follow analogous exponential profiles (Table 1). To keep the equivalence of the two profiles, we assume, for example, that the two situations are characterized by the same degradation rate α. Thus, the only difference between the accumulation and decay situations resides in the production rate, which is either enhanced or repressed at the onset of the respective dynamics1. Evidently, accumulation times are precisely the same as degradation times (Table 1); for example, the time to accumulate or degrade 90% of Pmax is given by α-1log (10). For simplicity we also assume that Plow = 0, although our results do not depend on this assumption as long as P low is much lower than the threshold level PT (Additional file 1).
Table 1 Comparison of models
Linear Modela Non Linear Modela,b
Accumulation Decay Non Linear Decayd
Model n = 2, 3, 4...
Solution P = Pmax (1 - e-αt) P = Pmax e-αt
T0
Unperturbed delay time
T1
Perturbedc delay time > η-1T0 (η < 1)
<η-1T0 (η > 1)
Delay time sensitivity ≥ |η-1 - 1|
a Pmax
b T0, T1 and δt are presented for n = 2
c Peturbation: v0 → ηv0
We examined the sensitivity of the delay times to fluctuations in the production rate of P. Since production rate correlates with gene dosage, it is likely to be mostly sensitive to gene-specific perturbations. Perturbation was implemented by changing the production rate of P, v0, by some factor η. Consequently, the delay time T0, coded by the time to accumulate or degrade the protein level from its initial value to the threshold level PT , is changed. We denote this perturbed time by T1. The sensitivity of the delay time to this change in production rate was defined by the relative change in the delay time:
Delay times encoded by decay display a significantly lower sensitivity
Despite their apparent equivalence, we found that the accumulation and decay strategies differ greatly in their capacities to buffer fluctuations in production rate. In fact, for most cases, delay times encoded by decay display a significantly lower sensitivity (Fig. 2 and Table 1). For example, while a two-fold reduction in production rate (η = 1/2) increases delay times by at least 100% in the case of accumulation, it will cause only a 15% (if PT = 0.01 Pmax) or 30% (if PT = 0.1 Pmax) decrease in the case of degradation.
Figure 2 Delay-time sensitivities for different η and different threshold positions. a, η = 1/2. It can be seen that for all threshold positions the sensitivity of the delay time is smallest for non-linear decay and largest for accumulation. b, η = 2. Also here the sensitivity of the delay time is smallest for non-linear decay and largest for accumulation for most threshold positions. Note that the situation is reversed for high threshold levels corresponding to high sensitivity in all cases (fig. 1). c-e, Delay time sensitivity as a function of η and PT for the cases of accumulation (c), decay (d) and non-linear decay (e). The logarithm of the delay time sensitivity is shown: log (|η - 1|/|δt|) for decay and log (|η-1 - 1|/|δt|) for accumulation. δt was normalized by η-1 for decay and by η-1-1 for accumulation, which correspond to δt in the non-buffered case in which T1 = ηT0 for decay and T1 = η-1T0 for accumulation. Thus, blue represents a non-buffered system. Red represents a buffered system.
The reason underlying this differential behaviour may not be immediately apparent. Within both strategies, changing protein production rate impacts the dynamics in two principal ways. First, it alters the initial rate (v0) by which a protein accumulates or degrades. Second, it modifies the maximal level Pmax (Fig. 1A–B). The key difference between the two systems resides in the initial conditions: in the case of accumulation, the initial condition, and thus the amount of protein that needs to accumulate in order to reach the threshold, remains fixed. Consequently, increased velocity necessarily shortens the time to reach the threshold. In contrast, in the case of degradation, the initial condition, and thus the distance to the threshold, is modified as well. Indeed, this change in initial condition partially balances the change in velocity.
Moreover, this combination of effects leads to a completely different behaviour of the delay time sensitivity, δt. Whereas in the case of accumulation, perturbation in production rate can be mathematically approximated by rescaling time by a constant factor, in the case of degradation such a perturbation is captured by introducing a constant shift in time (see expressions for T1 in Table 1 and Additional file: 1 2.1.2). This difference is due to the fact that production rate enters the equation explicitly in the case of accumulation, but only implicitly, through the initial conditions, in the case of decay. Importantly, this distinct behaviour is not restricted to the linear model, but is in fact applicable also to a general model that includes arbitrary feedback interactions (Additional file: 1 2.2). Consequently, within the accumulation strategy, the dependence of delay times on perturbation will be at best linear with perturbation size, irrespective of possible feedbacks.
Thus, despite the apparent equivalence of the accumulation and degradation strategies, they differ greatly in their capacity to buffer delay times against perturbations in gene dosage. Still, even in the case of degradation, buffering capacity in the absence of feedbacks is limited by realistic dynamical range of the degrading protein. For example, in order to achieve <8% sensitivity to a two-fold change, protein levels have to degrade over four orders of magnitude. This need to increase the dynamical range in order to improve robustness reflects the fact that delay time sensitivity depends on the degradation rate at early times: the faster this initial decay, the greater the robustness (Additional file: 1 3.2.1). However, in the absence of feedback, the system is characterized by a uniform decay rate, so that increasing this initial degradation implies an overall faster decay of P during the given delay time.
Non-linear degradation enhances robustness
One possible way to overcome this interplay between robustness and dynamical range is to introduce a feedback that enhances degradation specifically at early times, while maintaining moderate decay rate during the rest of the time. This will be the case, for example, if the degrading protein functions to enhance its own degradation, either directly or by changing the activity of a third protein. Indeed, a similar feedback mechanism was recently shown to enhance the spatial robustness of morphogen gradients [13].
To rigorously examine the possible impact of an auto-induced degradation on buffering capacity, we extended the linear model to include non-linear degradation (Table 1). In contrast to the exponential dynamics found in the linear system, here the system decays as a power-law in time. Examining the delay time sensitivity, we observed a significantly improved robustness (Fig. 2). For example, for moderate non-linearity, with n = 2, two-fold reduction in production rate will decrease the delay time by merely 1%, compared to 15% in the absence of feedback (for PT = 0.01 Pmax,) Moreover, increasing this coefficient of non-linearity further enhances the robustness (Fig. 2).
The robustness of timing requires fast initial degradation coupled with slower degradation afterwards. In particular, degradation needs to be rapid when protein concentration is above Pmax * η min. Nonlinear degradation enables, in principle, such flexible degradation rates. However, we note that there is an upper limit to degradation rate (see next section).
Auto-regulated degradation is implied in various stages of the cell cycle, such as the transition from S phase to mitosis and the exit from mitosis [14,15]. The degradation of the budding yeast Cdc20 exemplifies this. It begins degrading in late M phase, just before exit from mitosis, and continues throughout G1 phase [16]. This degradation is self-enhanced as Cdc20 itself is an activator of APC-dependent proteolysis through the subunits Cdc23 and Cdc27 [17]. However, degradation in those stages is commonly assumed to occupy only a small portion of the transition time, with most of the delay defined by protein accumulation. It may be that autonomous timing of those stages is less crucial, since the transitions are completely dependent on checkpoint mechanisms that survey the successful completion of the critical events occurring during those cell-cycle stages. Alternatively, protein decay may actually occupy a longer portion of the transition time, or other feedback mechanisms exist but have not yet been identified.
Cell degradation machinery sets a lower limit on time variability
For robust measurement of time, the time for degradation of protein concentrations above Pmax * η min (denoted by δ, different from the sensitivity δt) needs to be short. However, this degradation time is bounded from below by the maximum degradation rate of the cell machinery (in this section we assume that the protein is being degraded rather than being modified):
Where δguaranteed is worst case δ. ηmax , ηmin are worst cases for η and degmax is maximum degradation rate [molecules/s].
An order of magnitude estimate of this limit is given, based on a work by Shibatani and Ward [18], which assayed for 20S rat proteasome activity. The 20S proteasome complex is found in all eukaryotic cells and constitutes 0.5–1% of the soluble protein in the cell.
Shibatani and Ward have measured degradation rates in vitro, activating the proteasome with sodium dodecyl sulfate (SDS). The maximum degradation rate measured was 20 nmol/h for 0.07 nmol proteasome. I.e., each proteasome complex degraded roughly 300 molecules per hour. The proteasome composes 0.5–1% of the soluble proteins in the cell (by mass). It is a very heavy complex, about 700 kDa, 14-fold greater than average protein mass, which is roughly 50 kDa. Hence, there are about 1400–2800 proteins per each proteasome complex in the cell. Estimating protein number in the cell as 106-107 (molecules) gives ~ 103 proteasome units. Each unit is capable of degrading 300 molecules in 1 hour, giving degmax ~ 100 [proteins/s].
Assuming Pmax ~ 103 molecules, and fluctuations of the same order (e.g., η = 2), δguaranteed = 10 seconds. Processes sufficiently longer than δguaranteed can be measured accurately: even relaxing some of the assumptions (degradation dedicated to single protein, larger Pmaxetc.), will enable timing of many processes with good accuracy. For example, yeast cell cycle is about 120 minutes, circadian clock is 24 hours – 103-104 fold longer than δguaranteed.
Perturbation to production vs. degradation rates
Our discussion focused on the robustness to fluctuations in production rate (vo) while assuming degradation rate (α) to be relatively stable. Since the degradation machinery plays a crucial role in numerous cellular processes, it is reasonable to assume that its abundance is under a tight regulation, which also limits the noise in degradation rates of individual proteins. Moreover, gene dosage perturbations to production rate are of large magnitude compared to other sources of noise. We expect, as consequence, that mechanisms for buffering against production rate perturbations will be abundant.
Different buffering mechanisms will need to be utilized in the alternative situations where fluctuations in degradation rate dominate. One possible scheme that could reduce the effect of fluctuations in degradation relies on in-cis degradation, where each molecule promotes its own degradation. Such a mechanism was recently reported in the context of cell-cycle timing, where S-phase can only start after UbcH10 undergoes in-cis degradation [19]. Alternatively, delay time could be coded by the linear phase of accumulation, before degradation comes into effect. In this case, the delay time is given by To = PT/vo and does not depend on the degradation rate.
More generally, one may envision other noise characteristics, each dictating its own limitations; for example both production rate and degradation rate might be perturbed together (e.g. temperature effect). The threshold PT might be perturbed together with production rate (Additional file: 1 7) or any other perturbation characteristics. Different buffering mechanisms may need to be tuned for these different perturbations types, which could be analyzed using the framework presented in this paper.
Conclusions
Ensuring the robustness of timing may be of particular importance in order to support crosstalk amongst several processes that are executed in parallel. In such cases, not only the successful completion of events, but also maintaining the coordination, is important. This need may be of particular relevance during development of multicellular organisms, where multiple differentiation processes often proceed in parallel. Our identification of mechanisms that are able to maintain such robustness of timing may provide a new framework for examining the robustness of the long-range cascades that underlie those processes.
Our discussion focused on timing mechanisms that rely on the accumulation or degradation of a single protein component. While such mechanisms can serve as independent timers, more often they present an elementary unit in a more complex cascade. For example, models of cell cycle regulation propose that delay times are generated through the activation of some intermediate components, leading to a delayed negative feedback [20,21]. Further work is required to define how the properties of the full cascade are determined from the properties of its elementary units, and what additional constraints are required for proper coupling of different elementary units.
Methods
Figures were generated using Matlab simulations.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NR performed the analysis and drafted the manuscript. SW performed the analysis and drafted the manuscript. NB conceived of the study, participated in its design and drafted the manuscript. All authors read and approved the final manuscript.
Note
1 The results are unchanged if we keep production rate fixed and vary the degradation rate. This will become clear later (see expressions for delay time sensitivity in table 1).
Supplementary Material
Additional File 1
Supplementary Information
Click here for file
Acknowledgements
We thank members of our group for useful discussions. This work was supported by the Minerva and by the ISF. N. B. is the incumbent of the Soretta and Henry Shapiro career development chair.
==== Refs
Kerszberg M Noise, delays, robustness, canalization and all that Curr Opin Genet Dev 2004 14 440 445 15261662 10.1016/j.gde.2004.06.001
Eldar A Dorfman R Weiss D Ashe H Shilo BZ Barkai N Robustness of the BMP morphogen gradient in Drosophila embryonic patterning Nature 2002 419 304 308 12239569 10.1038/nature01061
Von Dassow G Odell GM Design and constraints of the Drosophila segment polarity module: Robust spatial patterning emerges from intertwined cell state switches J Exp Zool 2002 294 179 215 12362429 10.1002/jez.10144
Barkai N Leibler S Robustness in simple biochemical networks Nature 1997 387 913 917 9202124 10.1038/43199
Hartwell LH Weinert TA Checkpoints: controls that ensure the order of cell cycle events Science 1989 246 629 634 2683079
Murray AW Kirschner MW Dominoes and clocks: the union of two views of the cell cycle Science 1989 246 614 621 2683077
Haase SB Reed SI Evidence that a free-running oscillator drives G1 events in the budding yeast cell cycle Nature 1999 401 394 397 10517640 10.1038/43930
Edery I Circadian rhythms in a nutshell Physiol Genomics 2000 3 59 74 11015601
Stuart D Wittenberg C CLN3, not positive feedback, determines the timing of CLN2 transcription in cycling cells Genes Dev 1995 9 2780 2794 7590253
Tyers M Tokiwa G Futcher B Comparison of the Saccharomyces cerevisiae G1 cyclins: Cln3 may be an upstream activator of Cln1, Cln2 and other cyclins Embo J 1993 12 1955 1968 8387915
Verma R Annan RS Huddleston MJ Carr SA Reynard G Deshaies RJ Phosphorylation of Sic1p by G1 Cdk required for its degradation and entry into S phase Science 1997 278 455 460 9334303 10.1126/science.278.5337.455
Schwob E Bohm T Mendenhall MD Nasmyth K The B-type cyclin kinase inhibitor p40SIC1 controls the G1 to S transition in S. cerevisiae Cell 1994 79 233 244 7954792 10.1016/0092-8674(94)90193-7
Eldar A Rosin D Shilo BZ Barkai N Self-enhanced ligand degradation underlies robustness of morphogen gradients Dev Cell 2003 5 635 646 14536064 10.1016/S1534-5807(03)00292-2
King RW Deshaies RJ Peters JM Kirschner MW How proteolysis drives the cell cycle Science 1996 274 1652 1659 8939846 10.1126/science.274.5293.1652
Hoyt MA Eliminating all obstacles: regulated proteolysis in the eukaryotic cell cycle Cell 1997 91 149 151 9346231 10.1016/S0092-8674(00)80396-7
Goh PY Lim HH Surana U Cdc20 protein contains a destruction-box but, unlike Clb2, its proteolysisis not acutely dependent on the activity of anaphase-promoting complex Eur J Biochem 2000 267 434 449 10632713 10.1046/j.1432-1327.2000.01014.x
Prinz S Hwang ES Visintin R Amon A The regulation of Cdc20 proteolysis reveals a role for APC components Cdc23 and Cdc27 during S phase and early mitosis Curr Biol 1998 8 750 760 9651679 10.1016/S0960-9822(98)70298-2
Shibatani T Ward WF Sodium dodecyl sulfate (SDS) activation of the 20S proteasome in rat liver Arch Biochem Biophys 1995 321 160 166 7639516 10.1006/abbi.1995.1381
Rape M Kirschner MW Autonomous regulation of the anaphase-promoting complex couples mitosis to S-phase entry Nature 2004 432 588 595 15558010 10.1038/nature03023
Novak B Tyson JJ Numerical analysis of a comprehensive model of M-phase control in Xenopus oocyte extracts and intact embryos J Cell Sci 1993 106 1153 1168 8126097
Ciliberto A Petrus MJ Tyson JJ Sible JC A kinetic model of the cyclin E/Cdk2 developmental timer in Xenopus laevis embryos Biophys Chem 2003 104 573 589 12914904 10.1016/S0301-4622(03)00060-7
|
15967029
|
PMC1208956
|
CC BY
|
2021-01-04 16:39:26
|
no
|
Theor Biol Med Model. 2005 Jun 20; 2:22
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-22
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-241599627110.1186/1742-4682-2-24ReviewVasculature deprivation – induced osteonecrosis of the rat femoral head as a model for therapeutic trials Bejar Jacob [email protected] Eli [email protected] Jochanan H [email protected] Department of Pathology, The Bnai-Zion Medical Center and The Bruce Rapapport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel2 Department of Orthopaedic Surgery B, Rambam Medical Center, and the Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel2005 5 7 2005 2 24 24 13 2 2005 5 7 2005 Copyright © 2005 Bejar et al; licensee BioMed Central Ltd.2005Bejar et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Experimental Osteonecrosis
The authors' experience with experimentally produced femoral capital osteonecrosis in rats is reviewed: incising the periosteum at the base of the neck of the femur and cutting the ligamentum teres leads to coagulation necrosis of the epiphysis. The necrotic debris is substituted by fibrous tissue concomitantly with resorption of the dead soft and hard tissues by macrophages and osteoclasts, respectively. Progressively, the formerly necrotic epiphysis is repopulated by hematopoietic-fatty tissue, and replaced by architecturally abnormal and biomechanically weak bone. The femoral heads lose their smooth-surfaced hemispherical shape in the wake of the load transfer through the hip joint such that, together with regressive changes of the joint cartilage and inflammatory-hyperplastic changes of the articular membrane, an osteoarthritis-like disorder ensues.
Therapeutic Choices
Diverse therapeutic options are studied to satisfy the different opinions concerning the significance of diverse etiological and pathogenic mechanisms: 1. Exposure to hyperbaric oxygen. 2. Exposure to hyperbaric oxygen and non-weight bearing on the operated hip. 3. Medication with enoxaparin. 4. Reduction of intraosseous hypertension, putting to use a procedure aimed at core decompression, namely drilling a channel through the femoral head. 5. Medication with vascular endothelial growth factor with a view to accelerating revascularization. 6. Medication with zoledronic acid to decrease osteoclastic productivity such that the remodeling of the femoral head is slowed.
Glucocorticoid-related osteonecrosis appears to be apoptosis-related, thus differing from the vessel-deprivation-induced tissue coagulation found in idiopathic osteonecrosis. The quantities of TNF-α, RANK-ligand and osteoprotegerin are raised in glucocorticoid-treated osteoblasts so that the differentiation of osteoclasts is blocked. Moreover, the osteoblasts and osteocytes of the femoral cortex mostly undergo apoptosis after a lengthy period of glucocorticoid medication.
==== Body
Background
Osteonecrosis of the femoral head is of both clinical and economic interest, nearly 20,000 patients being hospitalized annually in the U.S.A. for treatment of this disease. Different risk factors have been discussed, yet the etiology and the pathogenesis of osteonecrosis are still uncertain [1]. Clinical trials of novel therapeutic modalities are hindered by the lack of a suitable experimental model of the human disease [2]. Osteonecrosis is either "idiopathic" in nature or incidental to one of a number of diseases. To discover the chain of events resulting in osteocytic death, be it by necrosis or apoptosis, experimental models ought to replicate a "circulatory-deprivation" mishap, implicit in the practice among physicians of applying the epithet "avascular" to the disease. The epiphysis of the head of the femur is at particular risk of ischemic damage because it is undersupplied with effectual collateral circulation. Indeed, blood supply and drainage are provided by functional end-vessels. Irrespective of where the blood flow is initially disrupted, i.e. at the level of arteries, veins, capillaries or sinusoids, the circulation in the arteries is ultimately arrested [3].
Rodents are frequently used in preclinical tests of novel therapeutic modalities. So it behooves the reader to notice that interrupting the circulation in the femoral head of rats, with their lifelong persisting physeal cartilage, mimics children's Legg-Calvé-Perthes disease more than it resembles adult osteonecrosis [4]. Irrespective of the rat's age, the reduced uptake of bone-seeking isotopes at the sites of the necrotic bone implicates the disruption of the blood supply in triggering all cases of osteonecrosis [5].
Osteonecrosis of the Femoral Head of the Rat
The effects of therapeutic interventions on the course of osteonecrosis of the femoral head may be studied using various models. The following model has been applied by the authors of this review: the blood supply and drainage of epiphysis are interrupted by cutting the ligamentum teres and incising the periosteum at the cervical base of the femoral head of 6 month-old rats. After the operation, the rats are placed in spacious cages such that their perambulation is almost unhindered. At the time of sacrifice, the femora are excised and fixed in formalin. The samples are embedded in paraffin after decalcification. Blocks are cut such that longitudinally orientated sections bisect the insertion of the ligamentum teres [6].
Necrosis of the adipose and hematopoietic cells is histologically evident as early as the 2nd postoperative day. Necrosis of the subchondral and trabecular bone first becomes overt on the 5th postoperative day. Repair begins soon afterwards with growth of viable tissue from the epiphyseal-capsular junction into the necrotic debris within the intertrabecular spaces. Residues of the eosinophilic, granular, necrotic marrow are no longer apparent after the 3rd week. Undifferentiated mesenchymal cells initially infiltrate the necrotic marrow and are later replaced by well-vascularized fibrous tissue, carrying with it macrophages, resorbing the dead soft tissues, and by osteoclasts, absorbing the necrotic bone. Beginning in the 3rd postoperative week, newly-formed intramembranous and appositional bone remodel the original osseous framework of the epiphysis. Unevenly contoured, recently formed bony beams crisscross the intertrabecular fibrous tissue, spanning between the viable osteoid seams and the dead trabeculae. Complete replacement of all the necrotic by living bone occurs at the 6-week interval or later. The marrow spaces are repopulated by hematopoietic-fatty tissue. The femoral heads collapse, flatten or are otherwise disfigured. The physeal cartilage is mostly unaffected. Fibrous tissue invades the joint cartilage wherever the continuity of the subchondral bone plate is disrupted. Chondroclasts erode the cartilaginous matrix. A fibrous pannus eventually covers the roughened and fibrillated surface cartilage. As judged by the lack of stainable chondrocytic nuclei, the articular cartilage is undergoing focal chondrolysis, resulting occasionally in delamination of a partly free-floating cartilaginous membrane. The tissue in the expanded joint capsule is contiguous with the pannus and fibrous tissue in the marrow spaces. A shortcoming of this model is the widespread necrosis of the rat femoral heads, sporadically extending to the articular and physeal cartilage [6].
Disposition of the Epiphyses to Undergo Necrosis
Why is the epiphysis of the femoral head frequently affected by ischemic insults, while the diaphysis and metaphysis are spared? According to Johnson and her colleagues, the limited blood circulation accounts for the clinically high incidence of osteonecrosis of the femoral head [7]. Blood supply and drainage of the diaphysis and metaphysis depend on the nutrient, metaphyseal and periosteal arteries, which enter the bone through the foramina of the cortex. Having entered the marrow, they ramify and widely anastomose with each other. On the other hand, there is no dual supply and drainage of blood to and from the epiphysis because the femoral head is covered by cartilage. Ascending fan-like to the surface of the joint, the vessels are functionally end-arteries. It follows that the osseous-hematopoietic-fatty tissues of the epiphysis as well as the articular and the physeal cartilages are particularly susceptible to obstruction of the blood flow [8,9].
The Fate of the Ischemia – Induced Necrotic Bone
The gradual substitution of necrotic by living bone is divided into phases, which nevertheless overlap. Oxygen- and nutrient-deprived osteocytes and marrow cells die to the nearest link with the collateral circulation. Neutrophilic infiltration characterizes the acute phase, which is rapidly followed by the chronic stage during which invasion of macrophages is dominant. Granulation tissue forms and, with time, the detritus is resorbed. The stage of repair starts with the lessening of inflammation and resorption of the dead tissues. It is set in motion by an influx of pluripotential mesenchymal cells. The environmental variations and stresses to which the cells are exposed induce the pluripotential cells to differentiate into fibroblasts, chondroblasts, osteoblasts or angioblasts. The bulk of the cells involved in the reparative process infiltrate the necrotic femoral head from the hyperplastic subsynovial layer. Repair is associated with an ingress of capillary buds, which are recruited by vascular endothelial growth factor (VEGF) and diverse cytokines, which are abundantly synthesized by and released from the synovial fibroblasts residing within the hyperplastic subsynovial cell population [10,11].
The cues that monitor the behavior of the mesenchymal cells are probably derived from the microenvironment. To exemplify, cartilage and bone are produced in areas of low and high oxygen tension, respectively. Afterwards, the cartilage is transformed by endochondral ossification into bone [12]. Eventually, biomechanically redundant bone is resorbed during the remodeling stage and the newly deposited bony trabeculae are positioned along the lines of stress, as first postulated by Wolff in 1892, in so far as the skeletal architecture is adapted to biomechanical demands [13]. Concomitantly with the osteoclastic resorption of nonessential and poorly placed osseous beams, osteogenesis of trabeculae that fit the lines of force takes place. The tissue module regulating these events is the bone multicellular unit (BMU). The BMU is made up of an intraosseous, dissecting bulge of fibrous tissue with osteoclasts positioned at its closed side and osteoblasts situated along both bony surfaces. The remodeling compartment of the BMU at the fibrous tissue-bone interface is covered by flat cells facing the marrow, and by refashioning cells, i.e. osteoblasts, on its osseous side. The outspread marrow lining cells are continuous with the osteoblasts at the fringes of the remodeling compartment. The BMU's initial activity in remodeling of the cancellous bone is the digestion of the non-mineralized matrix. Given that the natural lifespan of both osteoclasts and osteoblasts is shorter than that of the BMUs, both these cell types have to be constantly replenished for remodeling to continue. The bone lining cells replace the marrow lining cells at the termination of each episode of osteogenesis such that the BMUs are sealed. The end product of BMU activity is a bone that differs from its original counterpart in that its modification is optimally adapted to perform the biomechanical functions demanded of it [14,15].
The above-described repair of the necrotic epiphyses might suggest that the healing process restores the femoral heads to their former selves. Yet unless the necrotic segment is small or restricted to the non-load bearing part of the femoral head, this is not the case in man. The clinical condition of untreated patients goes gradually downhill. Inasmuch as the reparative properties of healthy bone are excellent, this apparent discrepancy remains to be elucidated.
Fate of the Necrotic Femoral Head in Rats
Sevitt pioneered the prevailing explanation of avascular necrosis of the femoral head in the wake of a fractured femoral neck and the ensuing revascularization of the epiphysis [16]. In the context of the vascular-deprivation-induced model of osteonecrosis [6], analysis of the derivation of the tissues spreading into the necrotic marrow is of note. Invasion of fibrous tissue into the detritus proceeds from the hyperplastic tissue around the head and neck, remnants of living tissue, residues of the ligamentum teres, and the metaphysis (given that the physis has been breached). In view of the lifelong persistence of the rat physis, the necrotic epiphyses are mainly repopulated by tissue emanating from the expanded subsynovial compartment.
With the ingrowth of blood vessels, the reparative processes are launched by permeation of circulation-born monocytes throughout the necrotic debris. The emigrating monocytes proliferate and, having differentiated into macrophages and osteoclasts, resorb the dead tissues. Perivascular progenitor cells transform into osteoblasts, which deposit bone. In addition, the invading tissues are replete with undifferentiated mesenchymal cells, which stay dormant awaiting appropriate signals, upon which they are induced to proliferate and differentiate into fibroblasts, chondroblasts, osteoblasts, angioblasts and lipoblasts. This cascade of events, however, does not hold true for all species; thus, the undifferentiated cells in canine experimental osteonecrosis migrate first and foremost from adjacent living bone [17].
As stated above, recently deposited and mineralized bony matrix is biomechanically inferior to mature bone in respect of stiffness, strength, and toughness. Hence, the recently remodeled femoral heads do not withstand the transarticular stresses of daily activity-related loads without caving [18-20]. As rats' femoral heads always undergo total necrosis, this leads to a rapidly evolving osteoarthritis-like disorder [21]. Similarly, the revascularization-related restitution of the epiphyses by newly synthesized weak osseous trabeculae is blamed for the collapse of the femoral heads within only 4 to 6 weeks of disrupting the venous drainage of the femoral neck in minipigs [22].
The restorative activities begin during the 2nd postoperative week in the rat model. The near-hemispherical, smooth-contoured profile of the healthy femoral head is lost by the 2nd to the 3rd postoperative month. The femoral heads deviate structurally from normal shape in that they acquire a crescent-profiled, triangular or other aberrant form with rugged, murky brown joint cartilage. Not infrequently, there are no residues of dead tissues at this point in time. The amounts of newly deposited bone vary from one segment of the epiphysis to another and from one rat to another. Variously shaped and sized, lamellar-fibred or woven-fibred, newly formed trabeculae of bone crisscross the loosely to densely textured fibrous tissue that has permeated the intertrabecular spaces (Fig. 1). Cuboidal osteoblasts, often arrayed in multiple layers, abut on the lamellar-fibred or woven-fibred bone. Pseudocysts are sparsely scattered in the subchondral zone. The physis is focally or totally absent in a few cases such that the bony trabeculae of the epiphysis and metaphysis connect with one another by way of transphyseal bridges (Fig. 2). It seems that the physis is first broken up, then fibrous tissue and lastly bony beams replace the dead cartilage [9].
Figure 1 Several fissures (arrows) split the degenerated joint cartilage. The articular aspect of the femoral head is segmentally polished and eburnated (arrowheads). The intertrabecular spaces contain hematopoietic-fatty tissue (square) or hyalinized fibrous tissue (triangle). The physis is uninterrupted all along its path (asterisks). Inset: Residual necrotic bone within the fibrous tissue surrounded by some osteoblasts and an osteoclast (arrow).
Figure 2 Femoral head with AVN treated with alendronate after 42 days. The right operated femoral head of an alendronate-treated rat. There are just remnants of the physeal cartilage (●). The physis has been breached and epiphyseal-metaphyseal bridges (long thin arrow) join the epiphyseal and metaphyseal bony trabeculae with one another. The articular cartilage is of unequal thickness (thick arrow). Even so the hemispherical configuration is preserved. The height of epiphysis is within the standard range. Remnants of the ligamentum teres (■).
The articular aspect demonstrates a spectrum of changes ranging from a reduced content of glycosaminoglycan in the cartilage to a segmentally burnished and eburnated bony surface devoid of cartilage (Figs. 1 and 2). The degenerated cartilage is usually covered by a vascularized or avascular fibrous pannus. By and large, the scene at or about the 3rd postoperative month is that of osteoarthritis portraying distorted anatomical landmarks due to inappropriate repair of the epiphyseal hard and soft tissues and articular cartilage [19], matching Sokoloff's concept of degenerative joint disease as a deranged tricompartmental articulation [23].
Dead bone retains its rigid qualities for quite a long time unless it is substituted by newly formed osseous tissues. Non-remodeled necrotic bone should theoretically retain its properties of resistance to load-bearing and bending strains. There is consequently no biomechanical basis for evolving alterations of the conformation of the necrotic femoral head in the immediate period after an ischemic injury. In fact, descriptions in the literature of both functional and morphological deviations from the norm of the postosteonecrotic rat femoral head pertain to the late stage of the disease.
Human analyzers adequately and competently interpret what they perceive, but they experience difficulties in quantifying what they observe [24]. This knowledge is crucial in view of the widely-accepted supposition that the rat femoral head flattens during the early post-necrotic stage. Histomorphometrically, the height-to-width ratios and the values of the shape factor of femoral heads in rats killed 18 days after an ischemic insult differ statistically from those of rats sacrificed at an earlier time. These quantitatively-gauged statistics of remodeled femoral heads refute other authors' notions with respect to the purportedly consistent flattening, or collapse, of rat femoral heads. As a matter of fact, postnecrotic femoral heads evidently transmute into any of a number of forms during the repair stage, including femoral heads that are higher than those of healthy rats [25].
The distortion of an infarcted femoral head depends on the extent of structural degradation of its cancellous bone [26]. Because the repair processes are set in motion during the 2nd post-operative week, there is apparently no deterioration in the biomechanical properties of the femoral heads at the early stages. The differences in yield and maximum stress between the necrotic and adjacent vital bone are insignificant at the pre-deformation stage. Both parameters begin to decline with the initiation of osteoneogenesis such that they are, at this time, lower in the dead than in the contiguous living bone. The maximum stress of the adapted-sclerotic bone is higher than that of the subjacent uninvolved bone, explaining the aspherical distortion and secondary osteoarthritis of the hip at late stages of the disease [27-29].
The maximal deficit in material properties manifests itself during the mid- to late-stages of the repair phase [30], which in rats occurs a fortnight or so after the ischemic episode. Healing of the rats' injured tissues is speedy in comparison with the prolonged repair in large animal species [6]. In agreement with this paradigm, the height-to-width ratios of the femoral heads of rats killed on the 18th postoperative day clearly deviate from those of non-operated rats. Nevertheless, the direction of the shift in height-to-width ratios is unpredictable. Ratios greater than 0.4 are not encountered in non-operated rats. In contrast, height-to-width ratios greater than 0.4 are often detected on the 18th postoperative day, values ranking as high as 0.9 being occasionally encountered. There is no equivalent information in the Medline database with which these structural changes in remodeling femoral heads of rats could be compared.
Interestingly, about one third of the femoral heads of children with Perthes disease round up [31]. The epiphyseal index assigns a rank to the height-to-width ratios of femoral head contours measured by magnetic resonance imaging. Indices within the normal range are measured in children with stage I Perthes disease. These indices decrease in patients with stage II and III disease. The loss of sphericity and congruence of the femoral heads and acetabula in children with stage II and III disease coincides with flattening and widening of the epiphyses as well as with an increase in femoral head size [32]. On theoretical grounds, some authors have challenged the cascade of events mentioned above. They have postulated that the distortion of the architecture of the remodeled femoral heads in Perthes disease is secondary to the combined effects of collapse, asymmetric growth and disturbed endochondral ossification [33].
Contrary to the universally accepted paradigm, the mode by which the rats' vessel-deprived necrotic femoral heads remodel is unanticipated. The height-to-width ratios of numerous epiphyses obtained 18 and 36 days postoperatively are in fact greater than those of the control epiphyses. The adaptive reshaping of osseous tissues is responsive to alterations in the distribution and magnitude of the strain generated within the bone [34]. Comprising immature and malleable bone at the early stages after necrosis, it is hypothesized that the rat femoral heads are forced into atypical shapes by protruding from the acetabulum, or by other as yet unidentified mechanisms, such that they expand in the longitudinal direction.
Curetting the core of the necrotic epiphysis (thus stimulating osteoneogenesis) is assumed to prevent the collapse of the joint surface following blending with the subchondral bone plate of a cancellous bone-augmented vascularized fibular graft [35]. Likewise, buttressing of the remodeled epiphyses by the recently formed thick osseous trabeculae may reinforce the joint surface prior to the load-induced cave-in of the femoral heads. This mechanism possibly accounts for the protruding, rather than flattening, of the uppermost faces of the femoral head. The observation by Carter his coworkers that the perimeter of the tarsocrural joints in methylprednisolone-treated and exercised horses with a full-thickness osteochondral lesion increases within a few weeks of the operation [36] is crucial in the context of the hypothesis that post-necrotic repair processes may enlarge the articular structures. Notwithstanding the rather few and widely spaced trabeculae making up the osseous framework of remodeled rat femoral heads, these broad trabeculae (Fig. 2) seem biomechanically to equal the augmented bone volume fraction of osteoarthritic joints [37].
It is currently conjectured that, firstly, vascular impediment and defective repair capacity act in concert in causing non-traumatic variants of osteonecrosis and, secondly, the replicative potential of the osteoblasts is reduced in the living parts of the femoral head, supporting the pathogenetic role in osteonecrosis of malfunctioning of the bone-forming cells [38,39].
Therapeutic Trials
1. Reduction of Intraosseous Pressure
Taking for granted the accuracy of the paradigm of the pathogenetic role of vascular deprivation and anoxia in bringing about necrosis of the femoral head, revascularization and oxygenation ought to be the paramount therapeutic modalities. As a matter of fact, both core decompression and implantation of a vascularized bone graft have met with success in rescuing patients' necrotic femoral heads. This success is attributed, at least partly, to the encouragement of ingrowth of well-vascularized fibrous tissue into the necrotic bone. The size of the necrotic zone dictates the fate of necrotic femoral heads. In rats, resorption of the epiphysis takes place at all times because their femoral heads undergo total necrosis because the blood inflow and outflow at the cervical level and ligamentum teres are completely severed [40,41]. Core decompression is assumed to decrease the intraosseous hypertension that causes destruction. True, core decompression provides relief of pain for the patient, but in the long run its effectiveness in preventing the progressive distortion of the epiphysis is, at best, debatable [42].
2. Intraosseous Conduit as a Model of Core Decompression
In the experience of Simank et al., drilling a sheep's epiphysis (their model of core decompression) encouraged healing of the necrotic femoral heads [43]. The authors of the present review used a rat model to study the fate of the necrotic epiphysis after creating an intraosseous conduit through the femoral head. After incising the periosteum at the cervical base and cutting the ligamentum teres, a 21-gauge needle was lanced into the foveola via the residue of the ligament and pushed in the direction of the neck up to the opposite cortical bone. Hypercellular fibrous tissue with crowding sinusoidal blood vessels replaced the hematopoietic-adipose marrow 4 to 6 weeks after the operation. Clustered osteoblasts blended with undifferentiated mesenchymal cells. Osteoclasts abutted on to the necrotic trabecular and subchondral bone. Osteoclast-type cells were also scattered in the fibrous tissue, and when mingled with the mononuclear cell infiltrate, presented a giant cell granuloma-like appearance. Excessive osteogenesis resulted in the formation of compacta-like features and epiphyseal-metaphyseal bony bridges. Fibrous tissue occasionally extended upwards, replacing the joint cartilage, or downwards into the metaphysis. Dents, deeply permeating tunnels and large circular or polycyclic cavities at the surface of the femoral heads were found by analysis of serial sections to consist of cuts through the drilling channels. The joint cartilage showed severe degenerative changes. It is noteworthy that the disfigurement of the epiphyses was more prominent in this than in the authors' other models of attempts at therapy. The myriad sinusoidal vessels and their proximity to one another indicate that the intraosseous conduits support an exaggerated revascularization of the formerly avascular femoral heads. To conclude, the above alterations are unmistakably exclusive to the healing phase of osteonecrosis of the femoral head in the presence of an intraosseous conduit [6,44].
Lancing the epiphysis with a 21-gauge needle is not expected to weaken the bone. An explanation for the conduit-related intensification of remodeling should, therefore, be sought elsewhere. 1. Conceding that a conduit accelerates the healing process as a result of its tension-lowering effect and opening up a path for vascular ingrowth, the rapid replacement of dead by living bone leads to the deposition of a weak osseous structure that is unlikely to carry weight-bearing loads without collapsing. 2. The conduit hastens the development of osteoarthritis since the osteochondral junction is inadequately reconstructed. 3. The insertion of a needle through the foveola into the epiphysis creates an inlet that permits the synovial fluid to spill from the joint cavity into the intertrabecular spaces, thus delaying the repair of bone defects. 4. The synovial fibroblasts in the distended joint capsule of rats with vessel-deprived osteonecrosis of the femoral head are jam-packed with vascular endothelial growth factor. The overexpression of this and other intermediates probably accounts for the enhanced ingrowth of blood vessels after the creation of an intraosseous conduit in the necrotic femoral heads [11,45,56].
3. Heparin and Low Molecular Weight Heparin
The expectation that anticoagulation would thwart osteonecrosis of the femoral head goes back to the early 1970s, when Fahlström et al. established that the incidence of osteonecrosis complicating fractures of the femoral neck was reduced nearly fourfold in patients on a daily heparin regimen as compared to a control group of untreated patients [47]. Study of the impact of heparins on revascularization and stromal cells is germane in view of the current vogue for anticoagulation of patients with osteonecrosis. In contrast to untreated rats with vessel-deprived necrotic femoral heads, nearly all the necrotic bone is resorbed in less than a month in animals receiving a daily intramuscular injection of enoxaparin at a dose of 1 mg/kg. The differences between enoxaparin-treated and untreated rats in quantities of necrotic and newly formed bone, extent of remodeling and degeneration of the articular cartilage during the repair stage are statistically significant. Indeed, slowing of the progression towards an osteoarthritis-like phenotype is a major effect of enoxaparin therapy. In vitro, heparin makes the mitogenic effect of fibroblast growth factors on endothelial cells more efficacious, stabilizes as well as protects these factors from inactivation, acts as a receptor segregating basic fibroblast growth factor, and promotes the interaction with high affinity signaling receptors on the cell surfaces. VEGF and basic fibroblast growth factor support the spread of the vasculature. These factors, which are preferentially attracted to the heparin, increase the proliferation and migration of cells associated with neovascularization. In as much as enoxaparin suppresses the reactive leukocytic response, it favors bone healing because osteogenesis is inhibited by inflammation [48-56].
4. Hyperoxygenation
A series of hyperbaric oxygen-treated patients with osteonecrosis of the femoral head was reported in 1990 at the 10th International Congress of Hyperbaric Medicine [57]. However, the first publication in a peer-reviewed journal about the therapeutic effects of hyperbaric oxygen (HBO) on patients with avascular osteonecrosis of the femoral head appeared belatedly 13 years later [58].
Daily exposure of patients with Steinberg stage-I osteonecrosis to HBO for 100 days reportedly results in the return to a normal MRI pattern in ~80% of cases. This cure rate compares favorably with a ~80% rate of collapse of the femoral heads in untreated patients within 4 years of the onset of the disease [59]. Yet therapeutic investigations show that hyperoxygenation has few beneficial effects on rats with necrosis of the femoral heads. This may be explained by the toxic effects of HBO or an unbalanced stimulation of cells from different lineages when a very high dose of O2 is employed. The in vitro upregulation of osteoclastic activity may be related to the extended exposure to O2 radicals. In vivo, sustained hyperoxygenation results in the production of a repair tissue replete with structurally weak collagen fibers [60-62]. Too long or too frequent exposure to HBO impacts negatively on both the structure and the mechanical properties of the bone. For instance, extensive osteolysis of living and dead bone ensues in the femoral heads of rabbits after 2 daily sessions of one hour at 2 atmospheres absolute (ATA) followed by one daily session of 3 hours at 1 ATA for 16 days and finally 2 daily sessions of 3 hours for a further 12 days at 1 ATA [63]. The breaking strength of rat bones decreases when daily exposures to HBO are extended from 4 to 6 hours [64]. Ingrowth of vessels into metaphyseal cortical defects in rats is accelerated after one daily HBO session, but is retarded when two sessions are allotted [65]. To sum up, optimal healing of a bony lesion is achieved only if exposure to HBO is restricted within an auspicious dose range.
Daily 90 minute exposures to HBO in a monoplace hyperbaric chamber enhances osteogenesis in rats after ischemic damage of the femoral heads. Hyperoxygenation is intended to uphold the innate re-establishment of well-being, and to enhance fibrogenesis, appositional and intramembranous osteogenesis, resorption of necrotic soft tissues and osteoclastic osteolysis during the late phase of osteonecrosis [66,67].
Histomorphometric parameters indicate that exposure to HBO modifies the architectural distortion of the femoral heads [63]. The HBO-mediated intensification of fibrogenesis and angiogenesis prepare the ground for the restoration of the osseous framework in the necrotic femoral heads. Unfortunately, the betterment of healing comes at the expense of an architectural disarray of the healing epiphyses with biomechanically weak bone being produced after "too great amounts" of necrotic bone are "too rapidly" replaced by immature and weak bone, so that the femoral head undergoes structural disfigurement on weight-bearing [64-67].
Exposure to HBO provides an optimal environment for repair processes as the additional oxygen carried by the circulation to ischemic sites raises the oxygen tension in the tissues. The hyperoxygenation-mediated relief of ischemia boosts the activities of fibroblasts, osteoblasts and osteoclasts in addition to supplying the extra oxygen that is indispensable for meeting the increased metabolic demands of regenerating tissues. Given that vascularization of the ischemic site is sufficient, exposure to HBO within the first 4 to 6 hours after injury achieves the optimum results [68-70]. Shifting the homeostatic environment by affecting the functions of the bone cells and mineralization of the osteoid, exposure to HBO reduces the healing time of bone fractures and beneficially influences, among other factors, the healing of non-unions. In rats, intermittent exposure to HBO hastens callus formation in fractured bones [71,72]. Treatment of spontaneous hypertensive rats with HBO averts osteonecrosis of the femoral heads [73].
The prognosis after conservative therapy of femoral capital osteonecrosis is mostly poor, osteoarthritis more often than not evolving within 2 to 3 years of the diagnosis [74]. A perfect therapeutic modality would boost the substitution of new bone in the necrotic femoral head at a pace at least as rapid as the resorption of the dead bone, such that loss of structural integrity and biomechanical adequacy would not be below the capacity of the femoral head-acetabulum couple for functionally effective load-carrying without collapse of the epiphysis [75,76]. Diverse therapeutic options are proposed to achieve this goal; for instance, bone grafting, implantation of a vascularized bone graft, core decompression, electrical stimulation and hyperbaric oxygenation.
5. Hyperoxygenation and Non-Weight Bearing
It is now close to half a century since HBO was first acclaimed as a beneficial adjunct to conventional therapy for miscellaneous illnesses [77-80]. An interesting proposition is to combine non-weight bearing (NWB) on the necrotic femoral head with exposure to HBO [77]. The rationale is founded on the reduction of bone marrow edema and lessening of intramedullary ischemia by elevating the arterial oxygen tension by exposing the patients with osteonecrosis to HBO. Both ischemia and edema of the marrow are critical factors in the survival of bony tissues confined by non-yielding boundaries, to wit, the rigid cortex. Ischemia and edema bring about metabolic conditions that counteract an effective osteolysis of the dead bone on the one hand and osteogenesis on the other [78,79]. Exposure to HBO enhances angiogenesis, maturation of collagen and proliferation of fibroblasts, osteoblasts and osteoclasts, all of which contribute to the speedy repair of bone lesions [80,81]. While the advantages of exposing a damaged bone to HBO are well founded, the clinical implementation of NWB as a monotherapy does not prevent collapse of the necrotic femoral head [82,83].
In a study of the combined effect of exposure to HBO and NWB on the repair of necrotic femoral heads, rats were housed in an enclosed 2 × 2 × 1.3 feet Plexiglas space, in which the hind limbs were suspended by tail traction so that the hip joints were not loaded. The trailing end of a Velcro strip, wrapped around the rats' tails, was fixated to a crossbar with a wheel and swivel assembly riding on opposite walls of the cage. Thus, the rats had freedom of movement in the longitudinal and orthogonal directions and access to food and water at all times. From the 5th postoperative day, the rats were exposed to 100% oxygen at 2.5 ATA over 22 or 32 sessions, each lasting 90 min. Control animals were treated only by NWB. The rats were killed 30 or 42 days postoperatively. There were no changes in the femoral heads of sham operated (control) rats that had been subjected to NWB, HBO, or both. The gamut of post-osteonecrotic repair activities was enhanced in rats on the HBO plus NWB regimen: osteogenesis, florid osteoblastic rimming, preosteoblasts abutting on necrotic or lately deposited bone, clustered undifferentiated mesenchymal cells in hypercellular fibrous tissue, osteoclastic osteolysis of viable and necrotic bone, chondroclastic chondrolysis and degeneration of the joint cartilage were significantly more advanced than in other reported models of therapy. Severe distortion of the femoral heads ensued in almost a third of the rats. The structural deformations manifested various configurations affecting the shape, symmetry, organization of the hard and soft tissues, and the height as well as the width of the epiphysis. The irregularly shaped femoral heads had jagged surfaces subsequent to asymmetrical resorption of the necrotic bone and erratic substitution by thriving, recently formed bone. In place of the innate, smoothly surfaced hemispherical outline of the femoral head, any of a myriad geometric configurations evolved. Loss of tissue led to localized surface depressions, which were lined by a layer of synovial-like cells several cells thick. In other instances, exuberant tissue proliferation resulted in an elevation of the articular aspect. The sporadically decreased epiphyseal height signified flattening of the bony compartments of the femoral heads. Even though remodeling and distortion often coincided, the hemispherical profile of the femoral head was every so often preserved. Where sizable parts of the epiphysis had been replaced, the cartilage, the bone, the fibrous tissue, or all of these were always accompanied by peculiar architectural modifications. The semiquantitatively gauged parameters indicating deformation were statistically less significant on the 30th postoperative day in rats treated by the combined NWB plus HBO regimen than in the rats treated with either NWB or HBO alone [6-8,12,19,22,26,44,59,77]. Yet the management of patients with osteonecrosis of the femoral head or Perthes disease by NWB is at best debatable in so far as improvement of the functionality of the hip joint is concerned [2,84-87].
6. Advantages and Disadvantages of Hyperoxygenation
Several studies have established the favorable effects of HBO therapy on the course of certain ischemia-induced conditions, but there is no consensus about its therapeutic value in osteonecrosis of the femoral head [88].
Vessel-deprived epiphyseal osteonecrosis in rats does not fully imitate all the clinical, humoral and metabolic conditions that precede the disease in man. Nevertheless, the causal pathway of impeded blood supply and drainage is embodied in most experimental models of the disorder [6,89]. The versatile HBO therapy opposes the progression of necrosis and expedites reparative processes. Theoretically, the fibrous tissue enclosing the bone acts as a barrier that prevents oxygenation of the vessel-deprived region [90]. Practically, this barrier is overcome by the large amount of serum-dissolved O2 which, after HBO medication, increases the diffusion distance notwithstanding the fibrous tissue enclosure. The hyperoxygenation-induced relief of marrow edema is a spin-off of HBO exposure; it is the byproduct of reflex vasoconstriction and oxygen-induced osmosis, which reverses the usual pumping mode of interstitial fluids, i.e. from the tissues back into the circulation. Hyperoxygenation also induces the precursors of the multipotential mesenchymal cells to mature into osteoblasts and at the same time encourages osteoclastic osteolysis such that remodeling is enhanced overall [90-96]. Finally, HBO-induced suppression of the inflammatory response promotes osteogenesis [97]. Acting in concert, these consequences of HBO therapy influence the cascade of events so that bone turnover is accelerated. Alas, all the advantages gained by HBO exposure come at a price. True, hyperoxygenation results in rapid removal of the necrotic debris and a speedy rebuilding of a viable bone; but having been just lately deposited and mineralized, this bone is biomechanically weak. In fact, daily ambulation suffices to distort the architecture of the femoral head, and the evolution of an osteoarthritis-like disorder is just a matter of time [29].
7. Medication with Vascular Endothelial Growth Factor
VEGF stimulates angiogenesis, recruitment and migration of osteoblasts and activation of osteoclasts. So far so good; but medication with VEGF would also enhance the removal of dead bone and increase the formation of a mechanically weak intramembranous bone, two events that ought to be avoided at all costs. In the context of fracture healing, a slow VEGF-releasing device is an effective therapeutic mode [98-102], but its efficacy in the treatment of femoral capital osteonecrosis is doubtful, considering that the para-articular apparatus is already jam-packed with VEGF-containing synovial fibroblasts [11]. Contrary to the widely accepted goal of supporting angiogenesis, the authors of this review are convinced that release of VEGF should be inhibited [103]. Given that the ingrowth of blood vessels into the necrotic epiphysis sets in motion a cascade of events terminating in the destruction of the femoral head, whether partial or total, arresting the release or activity of VEGF may possibly slow down the rapid impairment of the biomechanical properties of healing bone. Åstrand and Aspenberg have arrived at a similar conclusion, albeit in a different model. During the ingrowth of osseous tissues into a bone graft placed in a bone chamber, the necrotic debris was not resorbed in rats treated with alendronate but was more or less removed in their untreated counterparts [104]. By analogy, the structural failure of necrotic femoral heads in patients begins with the resorption of dead bone during the revascularization phase prior to the point in time at which sufficient new osseous matrix has been synthesized and mineralized, i.e. that the skeleton has been adequately reinforced. Otherwise, daily load-bearing of the hip would deform the femoral head. If the early resorption of necrotic subchondral and trabecular bone could be minimized, premature structural breakdown of the femoral head should be averted and the ensuing osteoarthritis may be prevented or at least slowed down [21,105].
Lieberman et al. recommended combining core decompression with VEGF medication so as to strengthen "patients' angiogenic potential" [106]. This proposal is diametrically opposed to the concepts of the authors of this review. Firstly, the cells of the hyperplastic para-articular apparatus of rats with osteonecrosis are loaded with VEGF. Secondly, an additional hastening of the already hurried revascularization and remodeling of the necrotic femoral head would speed up the structural and mechanical deterioration of the hip joint. On the contrary, it is mandatory to slow down the repair process as far as is feasible in order to conserve the greatest amounts of innate and biomechanically sufficient (albeit necrotic) epiphyseal bone for as long as possible, because accelerated bone turnover causes production of a mechanically frail osseous framework. Bone turnover should, therefore, be halted by medication with inhibitor(s) of VEGF, the prime intermediate in recruiting endothelial cell progenitors [102].
8. Medication with Zoledronic Acid
Little et al. have carried out a proficient series of experiments on the medication of rats with zoledronic acid (ZA) after surgically inducing osteonecrosis of the femoral head. They hypothesize that this bisphosphonate may preserve the structure of the femoral head while, at the same time, allowing incremental bone repair. Indeed, treatment and prophylaxis with ZA improve the sphericity and maintain the architecture of the necrotic femoral head. They have studied rats medicated subcutaneously with saline, ZA at one and 4 weeks after the operation (ZA-post), and ZA at 2 weeks pre-operation and at 1 and 4 weeks post-operation (ZA-pre-post). Six weeks postoperatively, 71% of the femoral heads of the saline-treated rats were aspherical. This contrasts with 13% and 0% aspherical femoral heads 6 weeks postoperatively in the ZA-post and ZA-pre-post animals (p < 0.05). Histomorphometrically, the bone volume was decreased by 12% in the saline group and close to 20% in the ZA-post and ZA-pre-post groups (p < 0.05). The retention of necrotic bone in the epiphyses of the treated rats was unambiguous. The difference between the non-treated and treated rats is explicitly due to the reduction in bone turnover [107].
9. Post-osteonecrotic Osteoarthritis-Like Disorder
Hip osteoarthritis is the leading treatment failure in children with Perthes disease and in adults with osteonecrosis. It results from the abnormal load transfer from the acetabulum to the femur across a remodeled and deformed femoral head. Contrary to clinicians' precepts, therapy that minimizes or hinders the remodeling processes delays the progressive deterioration of the articular structures. A balance between osteolysis and osteogenesis in the appropriate ratio is decisive in forestalling the collapse of the epiphysis, as preservation of the hemispherical shape of the femoral head is crucial in averting the development of osteoarthritis [61,62].
10. Concluding remarks
The means of treating osteonecrosis of the femoral head appraised in this review have been under experimental and clinical analyses for a few decades. Each of them has been praised at one time or another for providing the solution to the orthopaedic surgeons' frustrating deadlock in respect of restoring the necrotic femoral head to its earlier physical condition. In the rat model of vessel deprivation-induced osteonecrosis of the femoral head, medication with enoxaparin, construction of an intraosseous conduit, exposure to HBO and exposure to HBO plus NWB have been shown to hasten those reparative activities that are conventionally accepted as the epitome of revitalization of avascular dead bone. Investigators have a priori endeavored to enable vascular ingrowth. Accepting that osteonecrosis is caused by lack of blood supply, it is reasoned that the sooner the vasculature is reinstituted and delivery of oxygen and nutrients is returned to normal, the faster and more comprehensive would be the reconstruction of a living and mechanically well-performing femoral head. In mild cases, the femoral heads more or less retain their hemispherical profile. In more advanced cases, they are somewhat flattened or otherwise deviate from the hemispherical shape. Lastly, in the most severe cases, the femoral heads acquire any of a number of bizarre geometric forms. All repair processes are accelerated in rats treated by the above-mentioned means, including amassing undifferentiated mesenchymal cells, and profuse fibrogenesis, vasculogenesis, chondrogenesis and osteogenesis. At first sight it appears that all the clinically desired goals are attained. Alas, the profile of many a treated rat femoral head is disfigured to such a degree that smooth functioning of the hip joint is out of the question. The rising array of deformations correlates with an increasing extent of repair, indicating an inverse relationship between the degree of reconstruction and extent of revascularization on the one hand, and the magnitude of distortion of the femoral heads on the other. This explains the dictum that rats with maximally reconditioned necrotic femoral heads have the worst of it [103].
Brown et al. have given an account of the biomechanical properties of cancellous bone samples obtained from middle-stage and late-stage osteonecrosis of adult necrotic femoral heads. Compared to specimens retrieved from femoral heads of healthy individuals, samples removed from infarcted zones exhibit low yield strength, a much reduced elastic modulus and a modestly increased strain-to-failure. It is noteworthy that minor deviations in the strength and stiffness of bone taken from the affected regions are associated with large differences in the pattern of collapse and revascularization of the femoral heads [30]. An orthopaedic surgeon's dilemma is in which way to sway the modification of the remodeling necrotic bone without the usually-occurring decline in biomechanical properties, so that the structural distortion of the femoral heads is kept to a minimum.
The clinical relevance of animal experiments utilizing hyperoxygenation as the exclusive mode of treatment may be criticized because HBO in the clinical setting constitutes an adjunct to other therapeutic modalities. Incidentally, the outcome of studies of exposure of spontaneously hypertensive rats to HBO is irrelevant to our subject matter, because hyperoxygenation is utilized prophylactically [73]. As a rule, treatment in clinical practice commences after the symptoms and signs are overt, i.e. at a point in time when osteonecrosis is already comparatively advanced. In the studies cited herein, exposure to HBO was begun late in the course of the disease when the signs of osteonecrosis were already well developed.
Rats with vessel-deprived osteonecrosis of the femoral heads do not gain markedly from a NWB regimen. This concurs with the almost predictable collapse of the necrotic femoral heads in patients managed by restricted weight bearing [107]. While NWB by itself does not avert deformation of the femoral heads, the institution of HBO therapy in non-weight bearing rats often brings about a favorable outcome after 22 sessions of exposure to HBO. High oxygen tension is essential for osteogenesis to take place. Based on the documentation of raised osteolysis in mouse calvariae and rabbit femoral heads exposed to HBO, there is concern as to the biomechanical strength of the femoral heads after healing expedited by excess O2, in so far as too much osteolysis in too short a time may result in an untimely collapse of the femoral head [57,84,108,109]. Be this as it may, the deformation of the dead femoral heads in rats under weight bearing and exposure to HBO is less than that under NWB alone. These results are reminiscent of the enhanced mineralization and greater breaking strength of fractured femora in rats exposed to HBO and the greater mineral density of the bones and torsional strength of the tibiae of HBO-treated rabbits subjected to distraction osteogenesis [110,111].
The experimental mimickers of osteonecrosis of patient femoral heads possess certain distinctive traits, which differ from the disease as witnessed at the bedside. Osteonecrosis, with only a few exceptions, affects only part of the femoral head, while the epiphysis of rats virtually always undergoes total necrosis. Also, glucocorticoid-induced osteonecrosis in patients does not duplicate the coagulation-type death of the vascular deprivation-induced disorder in rats, but rather evinces an apoptotic process [112]. However, Glueck and coworkers have indicated that thrombophilia may cause thrombotic occlusion of veins, accompanied by venous hypertension of the marrow and hypoxic bone death, so that Perthes disease ensues [113].
The Role of Glucocorticoids, Rank-Ligand and Apoptosis
It is well known that glucocorticoids suppress osteogenesis and cause calcium loss. Yet the mechanism(s) linking glucocorticoid medication to osteonecrosis of the femoral head remain mystifying. It is fascinating that idiopathic and glucocorticoid-induced osteonecrosis cannot be distinguished histologically, maybe because at least some intermediate events are common to both [114]. Fat embolization may lead to osteonecrosis in patients with lipid disorders as well as those with glucocorticoid-induced hyperlipidemias [115]. Jones has described the triad of intraosseous fat embolism, intravascular coagulation and osteonecrosis in man. He has suggested that the mediatory pathway of fat emboli, hypercoagulability, stasis and endothelial damage by free fatty acids induces osteonecrosis by gradually occluding subchondral capillaries and sinusoids with fibrin-platelet thrombi [116]. Drescher and coworkers have proposed that the declining epiphyseal blood flow and the raised plasma fibrinogen after an infusion of high doses of methylprednisolone is causal in the early stage of steroid-induced osteonecrosis in the conscious pig [117].
The ratio of cytokines promoting the differentiation of osteoclasts to those blocking this process is raised in glucocorticoid-treated osteoblasts. These cytokines include the TNF-α family, RANKL and osteoprotegerin [115,118]. It is yet undecided how they are linked to the site-specific damage caused by glucocorticoids. Osteoblasts and osteocytes of the femoral cortex in mice, and the iliac bones of patients prescribed glucocorticoids for some years, undergo apoptosis [119]. Nevertheless, isolated osteoblasts behave erratically in that they often do not display the expected glucocorticoid-induced apoptosis. In fact, glucocorticoids have been employed to stimulate precursor cells to differentiate into osteoblasts [120].
It is challenging to determine how glucocorticoids bring about widespread osteolysis on the one hand, and cause damage to distinctive skeletal sites – say, the femoral head – on the other. Experimental data imply that glucocorticoid-induced apoptosis of osteocytes coincides with vascular blockage-independent osteonecrosis [121]. By itself, a positive TUNEL reaction does not discriminate among the manifold causes of cell death. It is often associated with a synchronized pattern of clustered dead osteocytes and changes in the osseous matrix. The hyperpermeability of the matrix (confirmed by the lamellar adsorption of tetracycline) recalls the fact that osteocytes before and throughout the apoptotic process participate in the degradation of the environment. Osteocytes produce collagenases in vitro, but their capacity to release proteolytic enzymes in vivo is as yet unknown [122].
Glucocorticoid-promoted expression of the RANK-ligand by the osteoblasts represents a central pathway in osteoclast maturation [123]. The activation of the RANK-ligand is linked to the injured bony matrix, increased permeability of the matrix, and fragmentation of osteoblast and osteocyte DNA. The differentiation factors accelerating the osteoclastic activities seem to originate from the residual living stromal cells surrounding the apoptotic bone cells. The stromal cells of the marrow produce RANK-ligand and other key osteoclast-modifying cytokines, including those that stimulate bone turnover [124]. The in vivo survival and differentiation of osteoblasts depend on high glucocorticoid levels. Long-term corticosteroid medication-induced apoptosis occurs in murine femoral cortex and patients' iliac bones. Fewer than 1% of the trabeculae-lining osteoblasts are TUNEL-positive in healthy bone. While the cortex is normal in methylprednisolone-treated rats, the trabecular bone is undergoing resorption and a large fraction of its osteoblasts and osteocytes are TUNEL-positive. It should not go unnoticed that the TUNEL technique has an important shortcoming in so far as it indicates DNA fragmentation, which is not only a late occurrence in the apoptotic cascade but might also reflect other mechanisms of cell death [119,125].
Glucocorticoids play a prominent physiological role in the turnover of healthy bone. The process of osteoblastic maturation in tissue cultures is stimulated by glucocorticoids at the proper concentration [122,126]. In view of experimental data from rabbits, Eberhardt et al. have reasoned that apoptosis of the bone cells necessarily entails excessive stimulation of a crucial receptor [127].
Trabecular bone bends under a load to a greater extent than cortical bone. Hence, the especial sensitivity of trabecular osteoblasts and osteocytes possibly reflects variations in the microenvironment. Lastly, proapoptotic signals such as NO2, generated by mechanically strained osteoblasts and osteocytes, are likely to affect the response to glucocorticoids [128]. Based on their experimental data, Weinstein and his colleagues hypothesize that bone in the subarticular trabeculae of the femoral head is especially sensitive to glucocorticoid-induced apoptosis because it has a large active surface area under constant high stress load [119].
Acknowledgements
Approval of the Institutional Review Board of the Rappaport Faculty of Medicine of the Technion, Technological Institute of Israel, was obtained prior to initiation of the experiments performed by the authors referred to herein.
==== Refs
Lavernia CJ Sierra RJ Grieco FR Osteonecrosis of the femoral head J Am Acad Orthop Surg 1999 7 250 261 10434079
Mont MA Hungerford D Non-traumatic avascular necrosis of the femoral head J Bone Joint Surg 1995 77A 459 474 7890797
Downey DJ Simkin PA Lanzer WL Matsen FA III Hydraulic resistance: a measure of vascular outflow obstruction in osteonecrosis J Orthop Res 1988 6 272 278 3343632 10.1002/jor.1100060215
Catterall A Legg-Calvé-Perthes syndrome Clin Orthop 1981 158 41 52 7023775
Winet H Hsieh A Bao JY Approaches to study of ischemia in bone J Biomed Mater Res 1988 43 410 421 9855199 10.1002/(SICI)1097-4636(199824)43:4<410::AID-JBM8>3.0.CO;2-B
Norman D Reis D Zinman C Misselevich I Boss JH Vascular deprivation-induced necrosis of the femoral head of the rat. An experimental model of avascular necrosis in the skeletally immature individual or Legg-Perthes disease Int J Exp Path 1998 79 173 181 9741359 10.1046/j.1365-2613.1998.00063.x
Johnson EO Soultanis K Soucacos PN Vascular anatomy and microcirculation of skeletal zones vulnerable to osteonecrosis: vascularization of the femoral head Orthop Clin North Am 2004 35 285 291 15271536 10.1016/j.ocl.2004.03.002
Boss JH Chondrolysis in rats with vascular deprivation-induced osteonecrosis of the femoral head Eur J Orthop Surg Traumatol 2001 11 15 20 10.1007/BF01706656
Peskin B Shupak A Misselevich I Zinman C Levin D Jacob Z Reis DN Boss JH Transphyseal osseous bridges in experimental osteonecrosis of the femoral head of the rat. Histologic study of the bony bridges connecting the epiphyseal with the metaphyseal bony trabeculae through gaps in the physeal cartilage J Pediatr Orthop 2001 10B 214 218
Newton CD Nunamaker DM Eds Textbook of Small Animal Orthopaedics 1985 Ithaca: International Veterinary Information Service
Vadasz Z Misselevich I Norman N Peled E Boss JH Localization of Vascular endothelial growth factor during the early reparative phase of the rats' vessels deprivation-induced osteonecrosis of the femoral heads Exp Mol Path 2004 77 145 148 15351238 10.1016/j.yexmp.2004.06.002
Levin D Norman D Zinman C Misselevich I Reis NR Boss JH Osteoarthritis-like disorder in rats with vascular deprivation-induced necrosis of the femoral head Pathol Res Prac 1999 195 637 642
Wolff J Das Gesetz der Transformation Transformation der Knochen 1892 Berlin: Hirschwald
Jilka RL Biology of the basic multicellular unit and the pathophysiology of osteoporosis Med Pediatr Oncol 2003 41 182 185 12868116 10.1002/mpo.10334
Hauge EM Qvesel D Eriksen EF Mosekilde L Melsen F Cancellous bone remodeling occurs in specialized compartments lined by cells expressing osteoblastic markers J Bone Miner Res 2001 16 1575 1582 11547826
Sevitt S Avascular necrosis and revascularisation of the femoral head after intracapsular fractures. A combined arteriographic and histological necropsy study J Bone Joint Surg 1964 46B 270 296
Malizos KN Quarles LD Seaber AV Rizk WS Urbaniak JR An experimental canine model of osteonecrosis: Characterization of the repair process J Orthop Res 1993 11 350 357 8326441 10.1002/jor.1100110306
Kenzora JE Glimcher MJ Pathogenesis of idiopathic osteonecrosis: The ubiquitous crescent sign Orthop Clin North Am 1985 16 681 696 4058896
Claes L The mechanical and morphological properties of bone beneath internal fixation plates of differing rigidity J Orthop Res 1989 7 170 177 2918416 10.1002/jor.1100070203
Hirano T Iwasaki K Sagara K Nishimura Y Kumashiro T Necrosis of the femoral head in growing rats. Occlusion of lateral epiphyseal vessels Acta Orthop Scand 1989 60 407 410 2816316
Boss JH Misselevich I Review. Experimental avascular osteonecrosis Cur Orthop 2001 15 57 67 10.1054/cuor.2001.0140
Swiontkowski MF Tepic S Rahn BA Cordey J Perren SM The effect of fracture on femoral head blood flow. Osteonecrosis and revascularization studied in miniature swine Acta Orthop Scand 1993 64 196 202 8498186
Sokoloff L Osteoarthritis as a remodeling process J Rheumatol 1987 14 7 10 3114489
Loebel J Image Analysis, Principles and Practice A Technical Handbook 1992 London: Vickers Company 1–2 79
Sabo E Peskin B Misselevich I Zinman C Levin D Norman D Reis DN Boss JH Computer-assisted image analysis of the rat postosteonecrotic remodeled femoral head Exp Molec Path 2001 71 256 264 11733951 10.1006/exmp.2001.2394
Brown TD Baker KJ Brand RA Structural consequences of subchondral bone involvement in segmental osteonecrosis of the femoral head J Orthop Res 1992 10 79 82 1727938 10.1002/jor.1100100110
Kim YM Lee SH Lee FY Koo KH Cho KH Morphologic and biomechanical study of avascular necrosis of the femoral head Orthopedics 1991 14 1111 1116 1946047
Leung PC Femoral head reconstruction and revascularization. Treatment for ischemic necrosis Clin Orthop 1996 323 139 145 67 8625571 10.1097/00003086-199602000-00019
Boss JH Misselevich I Peskin B Zinman C Norman D Reis DN Postosteonecrotic Osteoarthritis-like of the Femoral Head of Rats J Comp Path 2003 129 235 239 12921731 10.1016/S0021-9975(03)00031-8
Brown TD Way ME Ferguson AB Jr Mechanical characteristics of bone in femoral capital aseptic necrosis Clin Orthop 1981 156 240 247 7226659
Herring JA Williams JJ Neustadt JN Early JS Evolution of femoral head deformity during the healing phase of Legg-Calvé-Perthes disease J Pediatr Orthop 1993 13 41 45 8416352
Kumasaka Y Harada K Watanabe H Higashihara T Kishimoto H Sakurai K Kozuka T Modified epiphyseal index for MRI in Legg-Calvé-Perthes disease (LCPD) Pediatr Radiol 1991 21 208 210 2047161 10.1007/BF02011050
Mickelson MR McCurnin DM Awbrey BJ Maynard JA Martin RK Legg-Calvé-Perthes disease in dogs: a comparison to human Legg-Calvé-Perthes disease Clin Orthop 1981 157 287 300 7249458
Rubin CT Lanyon LE Osteoregulatory nature of mechanical stimuli: function as a determinant for adaptive remodeling in bone J Orthop Res 1987 5 300 310 3572599 10.1002/jor.1100050217
Malizos KN Soucacos PN Beris AE Osteonecrosis of the femoral head. Hip salvaging with implantation of a vascularized fibular graft Clin Orthop 1995 314 67 75 7634653
Carter BG Bertone AL Weisbrode SE Bailey MQ Andrews JM Palmer JL Influence of methylprednisolone acetate on osteochondral healing in exercised tarsocrural joints of horses Am J Vet Res 1996 57 914 922 8725823
Kamibayashi L Wyss UP Cooke TD Zee B Trabecular microstructure in the medial condyle of the proximal tibia of patients with knee osteoarthritis Bone 1995 17 27 35 7577155 10.1016/8756-3282(95)00137-3
Assouline-Dayan Y Chang C Greenspan A Shoenfeld Y Gershwin ME Pathogenesis and natural history of osteonecrosis Semin Arthritis Rheum 2002 32 94 124 12430099
Gangji V Hauzeur JP Schoutens A Hinsenkamp M Appelboom T Egrise D Abnormalities in the replicative capacity of osteoblastic cells in the proximal femur of patients with osteonecrosis of the femoral head J Rheumatol 2003 30 348 351 12563694
Lieberman JR Berry DJ Mont MA Aaron RK Callaghan JJ Rajadhyaksha AD Urbaniak JR Osteonecrosis of the hip: management in the 21st century Instr Course Lect 2003 52 337 55 12690862
Plenk H JrGstettner M Grossschmidt K Breitenseher M Urban M Hofmann S Magnetic resonance imaging and histology of repair in femoral head osteonecrosis Clin Orthop 2001 386 42 53 11347847
Urbaniak JR Harvey EJ Revascularization of the femoral head in osteonecrosis J Am Acad Orthop Surg 1998 6 44 54 9692940
Simank HG Graf J Kerber A Wiedmaier S Long-term effects of core decompression by drilling. Demonstration of bone healing and vessel ingrowth in an animal study Acta Anat (Basel) 1997 158 185 191 9394955
Norman D Miller Y Sabo E Misselevich I Peskin B Zinman C Levin DN Reis DN Boss JH The effects of enoxaparin on the reparative processes in experimental osteonecrosis of the femoral head of the rat Acta Path Microbiol Immunol Scand 2002 110 221 228
Salas A Sans M Soriano A Reverter JC Anderson DC Pique JM Panes J Long-term effects of core decompression by drilling. Demonstration of bone healing and vessel ingrowth in an animal study Acta Anat (Basel) 1997 158 185 191 9394955
Mont MA Carbone JJ Fairbank AC Core decompression versus nonoperative management for osteonecrosis of the hip Clin Orthop 1996 324 169 178 8595753 10.1097/00003086-199603000-00020
Fahlström G Öhman U Backman LE Björkegren NÅ Intramuscular heparin in femoral neck fracture: Reduced incidence of avascular necrosis Acta Chir Scand 1971 137 227 231 5149080
Goldhaber P The effect of hyperoxia on bone resorption in tissue culture Arch Pathol 1958 66 635 641
Muir JM Andrew M Hirsh J Weitz JI Young E Deschamps P Shaughnessy SG Histomorphometric analysis of the effects of standard heparin on trabecular bone in vivo Blood 1996 88 1314 1320 8695849
Fuller K Chambers TJ Gallagher AC Heparin augments osteoclast resorption-stimulating activity in serum J Cell Physiol 1991 147 208 214 2040655 10.1002/jcp.1041470204
Hall TJ Chambers TJ Molecular aspects of osteoclast function Inflamm Res 1996 45 1 9 8821771
Muir JM Andrew M Hirsh J Weitz JI Young E Deschamps P Shaughnessy SG Histomorphometric analysis of the effects of standard heparin on trabecular bone in vivo Blood 1996 88 1314 1320 8695849
Goldhaber P The effect of hyperoxia on bone resorption in tissue culture Arch Pathol 1958 66 635 641
Wray JB Rogers LS Effect of hyperbaric oxygenation upon fracture healing in the rat J Surg Res 1968 8 373 378 5667861 10.1016/0022-4804(68)90050-4
Barth E Sullivan T Berg E Animal model for evaluating bone repair with and without adjunctive hyperbaric oxygen therapy (HBO): comparing dose schedules J Invest Surg 1990 3 387 392 2291896
Jones JP Coagulopathies and osteonecrosis Acta Orthop Belg 1999 65 5 46 10084210
Hutter CD Dysbaric osteonecrosis: a reassessment and hypothesis Med Hypotheses 2000 54 585 590 10859644 10.1054/mehy.1999.0901
Jones JP JrLewis RH Lewis T Faugere MC Malluche HH The effect of hyperbaric oxygen on osteonecrosis Orthop Trans 1991 15 588 589
Wray JB Rogers LS Effect of hyperbaric oxygenation upon fracture healing in the rat J Surg Res 1968 8 373 378 5667861 10.1016/0022-4804(68)90050-4
Barth E Sullivan T Berg E Animal model for evaluating bone repair with and without adjunctive hyperbaric oxygen therapy (HBO): comparing dose schedules J Invest Surg 1990 3 387 392 2291896
Levin D Norman D Zinman C Rubinstein L Sabo E Misselevich I Reis D Boss JH Treatment of experimental avascular necrosis of the femoral head with hyperbaric oxygen in rats: Histological evaluation of the femoral heads during the early phase of the reparative process Exp Molec Path 1999 67 99 108 10527761 10.1006/exmp.1999.2273
Urbaniak JR Harvey EJ Revascularization of the femoral head in osteonecrosis J Am Acad Orthop Surg 1998 6 44 54 9692940
Iapicca M Turati A Longone C Necrosis of the femoral head and hyperbaric oxygen: A therapeutic approach Proc 10th Int Congr Hyperbaric Med: Amsterdam 1990 168 170
Reis ND Schwartz O Militianu D Ramon Y Levin D Norman D Melamed Y Shupak A Goldsher D Zinman C Hyperbaric oxygen therapy as a treatment for stage-I avascular necrosis of the femoral head J Bone Joint Surg 2003 85B 371 375 10.1302/0301-620X.85B3.13237
Yrjonen T Long-term prognosis of Legg-Calvé-Perthes disease: a meta-analysis J Pediatr Orthop 1999 8B 169 172
Iapicca M Turati A Longone C Necrosis of the femoral head and hyperbaric oxygen: A therapeutic approach Proc 10th Int Congr Hyperbaric Med: Amsterdam 1990 168 170
Yrjonen T Long-term prognosis of Legg-Calvé-Perthes disease: a meta-analysis J Pediatr Orthop 1999 8B 169 172
Rosenwasser MP Garino JP Kiernan HA Michelsen CB Long term followup of thorough debridement and cancellous bone grafting of the femoral head for avascular necrosis Clin Orthop 1994 306 17 27 8070190
Davis JC Landeen JM Levine RA Pyoderma gangrenosum: Skin grafting after preparation with hyperbaric oxygen Plast Reconstr Surg 1987 79 200 207 3543980
Bluemke DA Zerhouni EA MRI of avascular necrosis of bone Top Magn Reson Imaging 1996 8 231 246 8870181
Morykwas MJ Argenta LC No surgical modalities to enhance healing and care of soft tissue wounds J South Orthop Assoc 1997 6 279 288 9434249
Brown DH Evans AW Sandor GK Hyperbaric oxygen therapy in the management of osteoradionecrosis of the mandible Adv Otorhinolaryngol 1998 54 14 32 9547875
London SD Park SS Gampper T Hoard MA Hyperbaric oxygen for the management of radionecrosis of bone and cartilage Laryngoscope 1998 108 1291 1296 9738743 10.1097/00005537-199809000-00005
Kindwall EP Hyperbaric oxygen: More indications than many doctors realise Brit Med J 1993 307 515 516 8400966
Morykwas MJ Argenta LC Nonsurgical modalities to enhance healing and care of soft tissue wounds J South Orthop Assoc 1997 6 279 288 9434249
Neovius EB Lind MG Lind FG Hyperbaric oxygen therapy for wound complications after surgery in the irradiated head and neck: A review of the literature and a report of 15 consecutive patients Head Neck 1997 19 315 322 9213110 10.1002/(SICI)1097-0347(199707)19:4<315::AID-HED10>3.0.CO;2-8
Staples J Clement D Hyperbaric oxygen chambers and the treatment of sports injuries Sports Med 1996 22 219 227 8898456
Jones JP Jr Etiology and pathogenesis of osteonecrosis Semin Arthroplasty 1991 2 160 168
Kataoka Y Hasegawa Y Iwata H Matsuda T Genda E Miura T Takahashi H Effect of hyperbaric oxygenation on femoral head osteonecrosis in spontaneously hypertensive rats Acta Orthop Scand 1992 63 527 530 1441949
Mont MA Jones LC Sotereanos DG Amstutz HC Hungerford DS Understanding and treating osteonecrosis of the femoral head Instr Course Lect 2000 49 169 85 10829173
Peskin B Shupak A Levin D Norman D Jacob Z Boss JH Misselevich I Reis DN Zinman C Effects of non-weight bearing and hyperbaric oxygen therapy in vascular deprivation-induced osteonecrosis of the rat femoral head Undersea Hyper Med 2001 28 187 194
Esmond WG Attar S Cowley RA Hyperbaric oxygenation in experimental hemorrhagic shock: experimental chamber design and operation Trans Am Soc Artif Intern Organs 1962 8 384 93 13890773
Niinikoski J Hunt TK Oxygen tensions in healing bone Surg Gynecol Obstet 1972 134 746 750 5031484
Yablon IG Cruess RL The effect of hyperbaric oxygen on fracture healing in rats J Trauma 1968 8 186 202 5645246
Sirsjo A Lehr HA Nolte D Haapaniemi T Lewis DH Nylander G Messmer K Hyperbaric oxygen treatment enhances the recovery of blood flow and functional capillary density in postischemic striated muscle Circ Shock 1993 40 9 13 7686827
McKenzie MR Wong FL Epstein JB Lepawsky M Hyperbaric oxygen and post-radiation osteonecrosis of the mandible Eur J Cancer B Oral Oncol 1993 29B 201 207 8298424 10.1016/0964-1955(93)90023-8
Lavernia CJ Sierra RJ Grieco FR Osteonecrosis of the femoral head J Am Acad Orthop Surg 1999 7 250 261 10434079
Griffin PP Green NE Beauchamp RD Legg-Calve-Perthes disease: treatment and prognosis Orthop Clin North Am 1980 11 127 39 7360497
Iwata H Torii S Hasegawa Y Itoh H Mizuno M Genda E Kataoka Y Indications and results of vascularized pedicle iliac bone graft in avascular necrosis of the femoral head Clin Orthop 1993 295 281 288 8403663
Strauss MB Dvorak T Kindwall EP, Whelan HT Femoral head necrosis and hyperbaric oxygen therapy Hyperbaric Medicine Practice 1999 Second Flagstaff: AZ, Best Publishing Company 909 928
Atsumi T Kuroki Y Role of impairment of blood supply of the femoral head in the pathogenesis of idiopathic osteonecrosis Clin Orthop 1992 277 22 30 1372850
Schroer WC Current concepts on the pathogenesis of osteonecrosis of the femoral head Orthop Rev 1994 23 487 497 8065806
Marcus ND Enneking WF Massam RA The silent hip in idiopathic aseptic necrosis. Treatment by bone-grafting J Bone Joint Surg 1973 55A 1351 1366 93 4758708
Tompach PC Lew D Stoll JL Cell response to hyperbaric oxygen treatment Int J Oral Maxillofac Surg 1997 26 82 86 9151158
Schroer WC Current concepts on the pathogenesis of osteonecrosis of the femoral head Orthop Rev 1994 23 487 497 8065806
Hayes CW Conway WF Daniel WW MR imaging of bone marrow edema pattern: transient osteoporosis, transient bone marrow edema syndrome, or osteonecrosis Radiographics 1993 13 1001 1011 8210586
Lundgren CEG Harabin A Bennett PB Van Liew HD Thalmann ED Fregly MJ, Blatteis CM Gas physiology in diving Handbook of Physiology Section 4: Environmental Physiology 1996 II New York: Oxford University Press 999 1019
Hills BA A role for oxygen-induced osmosis in hyperbaric oxygen therapy Med Hypotheses 1999 52 259 263 10362286 10.1054/mehy.1997.0640
Vande Berg BC Malghem JJ Lecouvet FE Jamart J Maldague BE Idiopathic bone marrow edema lesions of the femoral head: predictive value of MR imaging findings Radiology 1999 212 527 535 10429713
Hills BA A role for oxygen-induced osmosis in hyperbaric oxygen therapy Med Hypotheses 1999 52 259 263 10362286 10.1054/mehy.1997.0640
Bassett CAL Herrmann I Influence of oxygen concentration and mechanical factors on differentiation of connective tissues in vitro Nature 1961 190 460 461 13687643
Goldhaber P The effect of hyperoxia on bone resorption in tissue culture Arch Pathol 1958 66 635 41
Lindstrom T Gullichsen E Lertola K Niinikoski J Effects of hyperbaric oxygen therapy on perfusion parameters and transcutaneous oxygen measurements in patients with intramedullary nailed tibial shaft fractures Undersea Hyperb Med 1998 25 87 291 9670433
Boss JH Shajrawi I Alperson M Mendes DG Inhibition of osteogenesis by foreign body granulomatous response to bony debris Orthopaedics Int Edn 1994 2 447 453
Camporesi EM Mascia MF Thom SR Physiological principles of hyperbaric oxygen Handbook on Hyperbaric Medicine 1996 NY: Springer-Verlag 35 58
Camp JF Colwell CW Jr Core decompression of the femoral head for osteonecrosis J Bone Joint Surg 1986 68A 1313 1319 3782202
Saito S Ohzono K Ono K Joint-preserving operations for idiopathic avascular necrosis of the femoral head J Bone Joint Surg 1988 70B 78 84
Street J Bao M deGuzman L Bunting S Peale FV JrFerrara N Steinmetz H Hoeffel J Cleland JL Daugherty A van Bruggen N Redmond HP Carano RA Filvaroff EH Vascular endothelial growth factor stimulates bone repair by promoting angiogenesis and bone turnover Proc Natl Assoc Sci 2002 99 9656 9661 10.1073/pnas.152324099
Boss JH Misselevich I Bejar J Norman D Zinman C Reis DN Experimentally gained insight-based proposal apropos the treatment of osteonecrosis of the femoral head Med Hypotheses 2004 62 958 965 15142657 10.1016/j.mehy.2003.12.036
Åstrand J Aspenberg P Systemic alendronate prevents resorption of necrotic bone during revascularization. A bone chamber study in rats BMC Musculoskelet Disord 2002 7 19 10.1186/1471-2474-3-19
Boss JH Misselevich I Osteonecrosis of the femoral head of laboratory animals: the lessons learned from a comparative study of osteonecrosis in man and experimental animals Vet Pathol 2003 40 345 354 12824505 10.1354/vp.40-4-345
Lieberman JR Berry DJ Mont MA Aaron RK Callaghan JJ Rajadhyaksha AD Urbaniak JR Osteonecrosis of the hip: Management in the twenty-first century Instr Course Lect 2003 52 337 355 12690862
Little DG Peat RA Mcevoy A Williams PR Smith EJ Baldock PA Zoledronic acid treatment results in retention of femoral head structure after traumatic osteonecrosis in young Wistar rats J Bone Miner Res 2003 18 2016 2022 14606515
Lavernia CJ Sierra RJ Grieco FR Osteonecrosis of the femoral head J Am Acad Orthop Surg 1999 7 250 261 10434079
Wilcox JW Koloding SC Acceleration of healing of maxillary and mandibular osteotomies by use of hyperbaric oxygen: A preliminary report J Oral Maxillofac Surg 1976 34 370 375
Coulson DB Ferguson AB JrDiehl RC Jr Effect of hyperbaric oxygen on the healing femur of the rat Surg Forum 1966 17 449 450 5921003
Ueng SWN Lee SS Lin SS Wang CR Liu SJ Yang HF Tai CL Shih CH Bone healing of tibial lengthening is enhanced by hyperbaric oxygen therapy: a study of bone mineral density and torsional strength on rabbits J Trauma 1998 44 676 681 9555841
Weinstein RS Nicholas RW Manolagas SC Apoptosis of osteocytes in glucocorticoid-induced osteonecrosis of the hip J Clin Endocrinol Metab 2000 85 2907 2912 10946902 10.1210/jc.85.8.2907
Glueck CJ Brandt G Gruppo R Crawford A Roy D Tracy T Stroop D Wang P Becker A Resistance to activated protein C and Legg-Perthes disease Clin Orthop 1997 338 139 152 9170375
Chernetsky SG Mont MA LaPorte DM Jones LC Hungerford DS McCarthy F Pathologic features in steroid and nonsteroid associated osteonecrosis Clin Orthop 1999 368 149 161 10613163
Chang CC Greenspan A Gershwin ME Osteonecrosis: current perspectives on pathogenesis and treatment Semin Arthritis Rheum 1993 23 47 69 8235665
Jones JP Jr Fat embolism, intravascular coagulation, and osteonecrosis Clin Orthop 1993 292 294 308 8519124
Drescher W Weigert KP Bunger MH Ingerslev J Bunger C Hansen ES Femoral head blood flow reduction and hypercoagulability under 24 h megadose steroid treatment in pigs J Orthop Res 2004 22 501 508 15099627 10.1016/j.orthres.2003.10.002
Cruess RL Ross D Crawshaw E The etiology of steroid-induced avascular necrosis of bone. A laboratory and clinical study Clin Orthop 1975 113 178 183 1192666
Weinstein RS Jilka RL Parfitt AM Manolagas SC Inhibition of osteoblastogenesis and promotion of apoptosis of osteoblasts and osteocytes by glucocorticoids. Potential mechanisms of their deleterious effects on bone J Clin Invest 1998 102 274 282 9664068
Nakashima T Sasaki H Tsuboi M Kawakami A Fujiyama K Kiriyama T Eguchi K Ichikawa M Nagataki S Inhibitory effect of glucocorticoid for osteoblast apoptosis induced by activated peripheral blood mononuclear cells Endocrinol 1998 139 2032 2040 10.1210/en.139.4.2032
Stewart K Walsh S Screen J Jefferiss CM Chainey J Jordan GR Beresford JN Further characterization of cells expressing STRO-1 in cultures of adult human bone marrow stromal cells J Bone Miner Res 1999 14 1345 1356 10457267
Winchester SK Bloch SR Fiacco GJ Partridge NC Regulation of expression of collagenase-3 in normal, differentiating rat osteoblasts J Cell Physiol 1999 181 479 488 10528234 10.1002/(SICI)1097-4652(199912)181:3<479::AID-JCP12>3.0.CO;2-D
Yamashita T Asano K Takahashi N Akatsu T Udagawa N Sasaki T Martin TJ Suda T Cloning of an osteoblastic cell line involved in the formation of osteoclast-like cells J Cell Physiol 1990 145 587 595 1703173 10.1002/jcp.1041450327
Atkins GJ Haynes DR Geary SM Loric M Crotti TN Findlay DM Coordinated cytokine expression by stromal and hematopoietic cells during human osteoclast formation Bone 2000 26 653 661 10831938 10.1016/S8756-3282(00)00280-5
Dempster DW Moonga BS Stein LS Horbert WR Antakly T Glucocorticoids inhibit bone resorption by isolated rat osteoclasts by enhancing apoptosis J Endocrinol 1997 154 397 406 9379116 10.1677/joe.0.1540397
Kim CH Cheng SL Kim GS Effects of dexamethasone on proliferation, activity, and cytokine secretion of normal human bone marrow stromal cells: possible mechanisms of glucocorticoid-induced bone loss J Endocrinol 1999 162 371 379 10467228 10.1677/joe.0.1620371
Eberhardt AW Yeager-Jones A Blair HC Alan W Regional Trabecular Bone Matrix Degeneration and Osteocyte Death in Femora of Glucocorticoid-Treated Rabbits Endocrinology 2001 142 1333 1340 11181552 10.1210/en.142.3.1333
Zaman G Pitsillides AA Rawlinson SC Suswillo RF Mosley JR Cheng MZ Platts LA Hukkanen M Polak JM Lanyon LE Mechanical strain stimulates nitric oxide production by rapid activation of endothelial nitric oxide synthase in osteocytes J Bone Miner Res 1999 14 1123 1131 10404012
|
15996271
|
PMC1208957
|
CC BY
|
2021-01-04 16:39:26
|
no
|
Theor Biol Med Model. 2005 Jul 5; 2:24
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-24
|
oa_comm
|
==== Front
Theor Biol Med ModelTheoretical Biology & Medical Modelling1742-4682BioMed Central London 1742-4682-2-291609552710.1186/1742-4682-2-29ReviewThree subsets of sequence complexity and their relevance to biopolymeric information Abel David L [email protected] Jack T [email protected] Director, The Gene Emergence Project, The Origin-of-Life Foundation, Inc., 113 Hedgewood Dr., Greenbelt, MD 20770-1610 USA2 Professor, Department of Environmental Biology, University of Guelph, Rm 3220 Bovey Building, Guelph, Ontario, N1G 2W1, Canada2005 11 8 2005 2 29 29 23 5 2005 11 8 2005 Copyright © 2005 Abel and Trevors; licensee BioMed Central Ltd.2005Abel and Trevors; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Genetic algorithms instruct sophisticated biological organization. Three qualitative kinds of sequence complexity exist: random (RSC), ordered (OSC), and functional (FSC). FSC alone provides algorithmic instruction. Random and Ordered Sequence Complexities lie at opposite ends of the same bi-directional sequence complexity vector. Randomness in sequence space is defined by a lack of Kolmogorov algorithmic compressibility. A sequence is compressible because it contains redundant order and patterns. Law-like cause-and-effect determinism produces highly compressible order. Such forced ordering precludes both information retention and freedom of selection so critical to algorithmic programming and control. Functional Sequence Complexity requires this added programming dimension of uncoerced selection at successive decision nodes in the string. Shannon information theory measures the relative degrees of RSC and OSC. Shannon information theory cannot measure FSC. FSC is invariably associated with all forms of complex biofunction, including biochemical pathways, cycles, positive and negative feedback regulation, and homeostatic metabolism. The algorithmic programming of FSC, not merely its aperiodicity, accounts for biological organization. No empirical evidence exists of either RSC of OSC ever having produced a single instance of sophisticated biological organization. Organization invariably manifests FSC rather than successive random events (RSC) or low-informational self-ordering phenomena (OSC).
Self-organizationself-assemblyself-orderingself-replicationgenetic code origingenetic informationself-catalysis.
==== Body
Background
"Linear complexity" has received extensive study in many areas relating to Shannon's syntactic transmission theory [1-3]. This theory pertains only to engineering. Linear complexity was further investigated by Kolmogorov, Solomonoff, and Chaitin [4-8]. Compressibility became the measure of linear complexity in this school of thought. Hamming pursued Shannon's goal of noise-pollution reduction in the engineering communication channel through redundancy coding [9].
Little progress has been made, however, in measuring and explaining intuitive information. This is especially true regarding the derivation through natural process of semantic instruction. The purely syntactic approaches to sequence complexity of Shannon, Kolmogorov, and Hamming have little or no relevance to "meaning." Shannon acknowledged this in the 3rd paragraph of his first famous paper right from the beginning of his research [2]. The inadequacy of more recent attempts to define and measure functional complexity [10-45] will be addressed in a separate manuscript.
Nucleic acid instructions reside in linear, digital, resortable, and unidirectionally read sequences [46-49]. Replication is sufficiently mutable for evolution, yet conserved, competent, and repairable for heritability [50]. An exception to the unidirectionality of reading is that DNA can occasionally be read from both directions simultaneously. For example, the circular bacterial chromosome can be replicated in both directions at the same time [51] But the basic principle of unidirectionality of the linear digital flow of information nonetheless remains intact.
In life-origin science, attention usually focuses on a theorized pre-RNA World [52-55]. RNA chemistry is extremely challenging in a prebiotic context. Ribonucleotides are difficult to activate (charge). And even oligoribonucleotides are extremely hard to form, especially without templating. The maximum length of such single strands in solution is usually only eight to ten monomers (mers). As a result, many investigators suspect that some chemical RNA analog must have existed [56,57]. For our purposes here of discussing linear sequence complexity, let us assume adequate availability of all four ribonucleotides in a pre-RNA prebiotic molecular evolutionary environment. Any one of the four ribonucleotides could be polymerized next in solution onto a forming single-stranded polyribonucleotide. Let us also ignore in our model for the moment that the maximum achievable length of aqueous polyribonucleotides seems to be no more than eight to ten monomers (mers). Physicochemical dynamics do not determine the particular sequencing of these single-stranded, untemplated polymers of RNA. The selection of the initial "sense" sequence is largely free of natural law influences and constraints. Sequencing is dynamically inert [58]. Even when activated analogs of ribonucleotide monomers are used in eutectic ice, incorporation of both purine and pyrimidine bases proceed at comparable rates and yields [59]. Monnard's paper provides additional evidence that the sequencing of untemplated single-stranded RNA polymerization in solution is dynamically inert – that the sequencing is not determined or ordered by physicochemical forces. Sequencing would be statistically unweighted given a highly theoretical "soup" environment characterized by 1) equal availability of all four bases, and 2) the absence of complementary base-pairing and templating (e.g., adsorption onto montmorillonite).
Initial sequencing of single-stranded RNA-like analogs is crucial to most life-origin models. Particular sequencing leads not only to a theorized self- or mutually-replicative primary structure, but to catalytic capability of that same or very closely-related sequence. One of the biggest problems for the pre-RNA World model is finding sequences that can simultaneously self-replicate and catalyze needed metabolic functions. For even the simplest protometabolic function to arise, large numbers of such self-replicative and metabolically contributive oligoribonucleotides would have to arise at the same place at the same time.
Little empirical evidence exists to contradict the contention that untemplated sequencing is dynamically inert (physically arbitrary). We are accustomed to thinking in terms of base-pairing complementarity determining sequencing. It is only in researching the pre-RNA world that the problem of single-stranded metabolically functional sequencing of ribonucleotides (or their analogs) becomes acute. And of course highly-ordered templated sequencing of RNA strands on natural surfaces such as clay offers no explanation for biofunctional sequencing. The question is never answered, "From what source did the template derive its functional information?" In fact, no empirical evidence has been presented of a naturally occurring inorganic template that contains anything more than combinatorial uncertainty. No bridge has been established between combinatorial uncertainty and utility of any kind.
It is difficult to polymerize even activated ribonucleotides without templating. Eight to ten mers is still the maximum oligoribonucleotide length achievable in solution. When we appeal to templating as a means of determining sequencing, such as adsorption onto montmorillonite, physicochemical determinism yields highly ordered sequencing (e.g., polyadenines) [60]. Such highly-ordered, low-uncertainty sequences retain almost no prescriptive information. Empirical and rational evidence is lacking of physics or chemistry determining semantic/semiotic/biomessenger functional sequencing.
Increased frequencies of certain ribonucleotides, CG for example, are seen in post-textual reference sequences. This is like citing an increased frequency of "qu" in post-textual English language. The only reason "q" and "u" have a higher frequency of association in English is because of arbitrarily chosen rules, not laws, of the English language. Apart from linguistic rules, all twenty-six English letters are equally available for selection at any sequential decision node. But we are attempting to model a purely pre-textual, combinatorial, chemical-dynamic theoretical primordial soup. No evidence exists that such a soup ever existed. But assuming that all four ribonucleotides might have been equally available in such a soup, no such "qu" type rule-based linkages would have occurred chemically between ribonucleotides. They are freely resortable apart from templating and complementary binding. Weighted means of each base polymerization would not have deviated far from p = 0.25.
When we introduce ribonucleotide availability realities into our soup model, we would not expect hardly any cytosine to be incorporated into the early genetic code. Cytosine is extremely difficult even for highly skilled chemists to generate [61,62]. If an extreme paucity of cytosine existed in a primordial environment, uncertainty would have been greatly reduced. Heavily weighted means of relative occurrence of the other three bases would have existed. The potential for recordation of prescriptive information would have been reduced by the resulting low uncertainty of base "selection." All aspects of life manifest extraordinarily high quantities of prescriptive information. Any self-ordering (law-like behavior) or weighted-mean tendencies (reduced availability of certain bases) would have limited information retention.
If non-templated dynamic chemistry predisposes higher frequencies of certain bases, how did so many highly-informational genes get coded? Any programming effort would have had to fight against a highly prejudicial self-ordering dynamic redundancy. There would have been little or no uncertainty (bits) at each locus. Information potential would have been severely constrained.
Genetic sequence complexity is unique in nature
"Complexity," even "sequence complexity," is an inadequate term to describe the phenomenon ofgenetic "recipe." Innumerable phenomena in nature are self-ordered or complex without being instructive (e.g., crystals, complex lipids, certain polysaccharides). Other complex structures are the product of digital recipe (e.g., antibodies, signal recognition particles, transport proteins, hormones). Recipe specifies algorithmic function. Recipes are like programming instructions. They are strings of prescribed decision-node configurable switch-settings. If executed properly, they become like bug-free computer programs running in quality operating systems on fully operational hardware. The cell appears to be making its own choices. Ultimately, everything the cell does is programmed by its hardware, operating system, and software. Its responses to environmental stimuli seem free. But they are merely pre-programmed degrees of operational freedom.
The digital world has heightened our realization that virtually all information, including descriptions of four-dimensional reality, can be reduced to a linear digital sequence. Most attempts to understand intuitive information center around description and knowledge [41,63-67]. Human epistemology and agency invariably get incorporated into any model of semantics. Of primary interest to The Gene Emergence Project, however, is the derivation through natural process of what Abel has called prescriptive information (semantic instruction; linear digital recipe; cybernetic programming) [68-71]. The rise of prescriptive information presumably occurred early in the evolutionary history of life. Biopolymeric messenger molecules were instructing biofunction not only long before Homo sapiens existed, but also long before metazoans existed. Many eubacteria and archaea depend upon nearly 3,000 highly coordinated genes. Genes are linear, digital, cybernetic sequences. They are meaningful, pragmatic, physically instantiated recipes.
One of the requirements of any semantic/semiotic system is that the selection of alphanumeric characters/units be "arbitrary"[47]. This implies that they must be contingent and independent of causal determinism. Pattee [72-74] and Rocha [58] refer to this arbitrariness of sequencing as being "dynamically inert." "Arbitrary" does not mean in this context "random," but rather "unconstrained by necessity." Contingent means that events could occur in multiple ways. The result could just as easily have been otherwise. Unit selection at each locus in the string is unconstrained. The laws of physics and chemistry apply equally to whatever sequencing occurs. The situation is analogous to flipping a "fair coin." Even though the heads and tails side of the coin are physically different, the outcome of the coin toss is unrelated to dynamical causation. A heads result (rather than a tails) is contingent, unconstrained by initial conditions or law.
No law of physics has utility without insertion of a symbolic representation of the initial conditions. This usually comes in the form of measurement or graph coordinates. The initial physical conditions themselves cannot be inserted into a mathematical formula. Only a mathematical representation can be inserted. Physicist Howard Pattee refers to this as a "description" of initial conditions. The "epistemic cut" [75,76], "Complementarity" [77-81], and "Semantic Closure" [82-85] must occur between physicality and any description of dynamics such as the tentative formal generalizations we call laws.
Pattee's Epistemic Cut, Complementarily, and Semantic Closure apply equally well to sequences of physical symbol vehicles [72-75,77-80,84,86-89]. Nucleotides and their triplet-codon "block codes" represent each amino acid. Genes are informational messenger molecules specifically because codons function as semantic physical symbol vehicles. A codon "means" a certain amino acid. The instantiation of prescriptive information into biopolymers requires an arbitrary reassortment potential of these symbol vehicles in the linear sequence. This means that sequencing is dynamically inert. If the sequence were ordered by law-like constraint, the sequence would manifest monotonous redundancy of monomer occurrence. There would be little or no uncertainty at each decision node. Uncertainty (contingency: freedom from necessity) is required in a physical matrix for it to serve as a vehicle of descriptive or prescriptive information.
Sequence complexity falls into three qualitative categories
1. Random Sequence Complexity (RSC),
2. Ordered Sequence Complexity (OSC), and
3. Functional Sequence Complexity (FSC)
Sequence order and complexity are at opposite ends of a bi-directional vector (Fig. 1). The most complex sequence is a random sequence with no recognizable patterns or order. Shannon uncertainty is a function of -log2 p when decision nodes offer equiprobable and independent choice opportunities. Maximum sequence order has a probability of 1.0 at each locus in the string. A polyadenine, for example, has a probability of nearly 1.0 of having an adenine occur at any given four-way decision-node locus in the string. P = 1.0 represents 0 uncertainty. Minimum sequence order (maximum complexity; sequence randomness) has a probability of 0.5 at each binary node. In a binary system, P = 0.5 represents maximum uncertainty (1.0 bit at that binary decision node). The above points have been clearly established by Gregory Chaitin [6,90,91] and Hubert Yockey [46-49,92-96].
Figure 1 The inverse relationship between order and complexity as demonstrated on a linear vector progression from high order toward greater complexity (modified from [93]).
Random Sequence Complexity (RSC)
A linear string of stochastically linked units, the sequencing of which is dynamically inert, statistically unweighted, and is unchosen by agents; a random sequence of independent and equiprobable unit occurrence.
Random sequence complexity can be defined and measured solely in terms of probabilistic combinatorics. Maximum Shannon uncertainty exists when each possibility in a string (each alphabetical symbol) is equiprobable and independent of prior options. When possibilities are not equiprobable, or when possibilities are linked (e.g., paired by association, such as "qu" in the rules of English language), uncertainty decreases. The sequence becomes less complex and more ordered because of redundant patterning, or because of weighted means resulting from relative unit availability. Such would be the case if nucleotides were not equally available in a "primordial soup." This is demonstrated below under the section labeled "Ordered Sequence Complexity (OSC)."
Random sequence complexity (RSC) has four components:
1. The number of "symbols" in the "alphabet" that could potentially occupy each locus of the sequence (bit string)
(e.g., four potential nucleotide "alphabetical symbols" could occupy each monomeric position in a forming polynucleotide. In the English language, there are 26 potential symbols excluding case and punctuation.)
2. Equal probabilistic availability (often confused with post-selection frequency) of each "symbol" to each locus
(e.g., the availability of adenine was probably not the same as that of guanine, cytosine, or uracil to each position in a randomly forming primordial oligoribonucleotide. When each possibility is not equiprobable, weighted means must be used to calculate uncertainty. See equation 1)
3. The number of loci in the sequence
(e.g., the number of ribonucleotides must be adequate for a ribozyme to acquire minimal happenstantial function. A minimum of 30–60 "mers" has been suggested [97,98]
4. Independence of each option from prior options.
(e.g., in the English language, the letters "qu" appear together with much higher frequency than would be expected from independent letter selections where P = 1/26. Thus, if the generation of the signal were viewed as a stationary Markov process [as Shannon transmission theory does], conditional probabilities would have to be used to calculate the uncertainty of the letter "u".)
The Shannon uncertainty of random alphanumeric symbol sequences can be precisely quantified. No discussions of "aboutness" [12,13,99] or "before and after" differences of "knowledge" [100-104] are relevant to a measure of the Shannon uncertainty of RSC. Sequences can be quantitatively compared with respect to syntax alone.
In computer science, "bits" refer generically to "the number of binary switch-setting opportunities" in a computational algorithm. Options are treated as though they were equiprobable and independent combinatorial possibilities. Bits are completely nonspecific about which particular selection is made at any switch. The size of the program is measured in units of RSC. But the programming decisions at each decision node are anything but random.
Providing the information of how each switch is set is the very essence of what we want when we ask for instructions. The number of bits or bytes in a program fails to provide this intuitive meaning of information. The same is true when we are told that a certain gene contains X number of megabytes. Only the specific reference sequences can provide the prescriptive information of that gene's instruction. Measurements of RSC are not relevant to this task.
Ordered Sequence Complexity (OSC)
A linear string of linked units, the sequencing of which is patterned either by the natural regularities described by physical laws (necessity) or by statistically weighted means (e.g., unequal availability of units), but which is not patterned by deliberate choice contingency (agency).
Ordered Sequence Complexity is exampled by a dotted line and by polymers such as polysaccharides. OSC in nature is so ruled by redundant cause-and-effect "necessity" that it affords the least complexity of the three types of sequences. The mantra-like matrix of OSC has little capacity to retain information. OSC would limit so severely information retention that the sequence could not direct the simplest of biochemical pathways, let alone integrated metabolism.
Appealing to "unknown laws" as life-origin explanations is nothing more than an appeal to cause-and-effect necessity. The latter only produces OSC with greater order, less complexity, and less potential for eventual information retention (Figs. 1 and 2).
Figure 2 The adding of a second dimension to Figure 2 allows visualization of the relationship of Kolmogorov algorithmic compressibility to complexity. The more highly ordered (patterned) a sequence, the more highly compressible that sequence becomes. The less compressible a sequence, the more complex is that sequence. A random sequence manifests no Kolmogorov compressibility. This reality serves as the very definition of a random, highly complex string.
The Shannon uncertainty equation would apply if forming oligoribonucleotides were stochastic ensembles forming out of sequence space:
where M = 4 ribonucleotides in an imagined "primordial soup." Suppose the prebiotic availability pi for adenine was 0.46, and the pi 's for uracil, guanine, and cytosine were 0.40, 0.12, and 0.02 respectively. This is being generous for cytosine, since cytosine would have been extremely difficult to make in a prebiotic environment [62]. Using these hypothetical base-availability probabilities, the Shannon uncertainty would have been equal to Table 1
Table 1 Hypothetical pre-biotic base availabilities
Adenine 0.46 (- log2 0.46) = 0.515
Uracil 0.40 (- log2 0.40) = 0.529
Guanine 0
12 (- log2 0.12) = 0.367
Cytosine 0.02 (- log2 0.02) = 0.113
1.00 1.524 bits
Notice how unequal availability of nucleotides (a form of ordering) greatly reduces Shannon uncertainty (a measure of sequence complexity) at each locus of any biopolymeric stochastic ensemble (Fig. 1). Maximum uncertainty would occur if all four base availability probabilities were 0.25. Under these equally available base conditions, Shannon uncertainty would have equaled 2 bits per independent nucleotide addition to the strand. A stochastic ensemble formed under aqueous conditions of mostly adenine availability, however, would have had little information-retaining ability because of its high order.
Even less information-retaining ability would be found in an oligoribonucleotide adsorbed onto montmorillonite [60,97,105-108]. Clay surfaces would have been required to align ribonucleotides with 3' 5' linkages. The problem is that only polyadenines or polyuracils tend to form. Using clay adsorption to solve one biochemical problem creates an immense informational problem (e.g., high order, low complexity, low uncertainty, low information retaining ability, see Fig. 1). High order means considerable compressibility. The Kolmogorov [4] algorithmic compression program for clay-adsorbed biopolymers (Fig. 2) would read: "Choose adenine; repeat the same choice fifty times." Such a redundant, highly-ordered sequence could not begin to prescribe even the simplest protometabolism. Such "self-ordering" phenomena would not be the key to life's early algorithmic programming.
In addition to the favored RNA Word model [55,109], life origin models include clay life [110-113]; early three-dimensional genomes [114,115]; "Metabolism/Protein First" [116-119]; "Co-evolution" [120] and "Simultaneous nucleic acid and protein" [121-123]; and "Two-Step" models of life-origin [124-126]. In all of these models, "self-ordering" is often confused with "self-organizing." All known life depends upon genetic instructions. No hint of metabolism has ever been observed independent of an oversight and management information/instruction system. We use the term "bioengineering" for a good reason. Holistic, sophisticated, integrative processes such as metabolism don't just happen stochastically. Self-ordering in nature does. But the dissipative structures of Prigogine's chaos theory [127] are in a different category from the kind of "self-organization" that would be required to generate genetic instructions or stand-alone homeostatic metabolism. Semantic/semiotic/bioengineering function requires dynamically inert, resortable, physical symbol vehicles that represent time-independent, non-dynamic "meaning." (e.g., codons) [73,74,86,87,128-131].
No empirical or rational basis exists for granting to chemistry non-dynamic capabilities of functional sequencing. Naturalistic science has always sought to reduce chemistry to nothing more than dynamics. In such a context, chemistry cannot explain a sequencing phenomenon that is dynamically inert. If, on the other hand, chemistry possesses some metaphysical (beyond physical; beyond dynamics) transcendence over dynamics, then chemistry becomes philosophy/religion rather than naturalistic science. But if chemistry determined functional sequencing dynamically, sequences would have such high order and high redundancy that genes could not begin to carry the extraordinary prescriptive information that they carry.
Bioinformation has been selected algorithmically at the covalently-bound sequence level to instruct eventual three-dimensional shape. The shape is specific for a certain structural, catalytic, or regulatory function. All of these functions must be integrated into a symphony of metabolic functions. Apart from actually producing function, "information" has little or no value. No matter how many "bits" of possible combinations it has, there is no reason to call it "information" if it doesn't at least have the potential of producing something useful. What kind of information produces function? In computer science, we call it a "program." Another name for computer software is an "algorithm." No man-made program comes close to the technical brilliance of even Mycoplasmal genetic algorithms. Mycoplasmas are the simplest known organism with the smallest known genome, to date. How was its genome and other living organisms' genomes programmed?
Functional Sequence Complexity (FSC)
A linear, digital, cybernetic string of symbols representing syntactic, semantic and pragmatic prescription; each successive sign in the string is a representation of a decision-node configurable switch-setting – a specific selection for function.
FSC is a succession of algorithmic selections leading to function. Selection, specification, or signification of certain "choices" in FSC sequences results only from nonrandom selection. These selections at successive decision nodes cannot be forced by deterministic cause-and-effect necessity. If they were, nearly all decision-node selections would be the same. They would be highly ordered (OSC). And the selections cannot be random (RSC). No sophisticated program has ever been observed to be written by successive coin flips where heads is "1" and tails is "0."
We speak loosely as though "bits" of information in computer programs represented specific integrated binary choice commitments made with intent at successive algorithmic decision nodes. The latter is true of FSC, but technically such an algorithmic process cannot possibly be measured by bits (-log2 P) except in the sense of transmission engineering. Shannon [2,3] was interested in signal space, not in particular messages. Shannon mathematics deals only with averaged probabilistic combinatorics. FSC requires a specification of the sequence of FSC choices. They cannot be averaged without loss of prescriptive information (instructions).
Bits in a computer program measure only the number of binary choice opportunities. Bits do not measure or indicate which specific choices are made. Enumerating the specific choices that work is the very essence of gaining information (in the intuitive sense). When we buy a computer program, we are paying for sequences of integrated specific decision-node choice-commitments that we expect to work for us. The essence of the instruction is the enumeration of the sequence of particular choices. This necessity defines the very goal of genome projects.
Algorithms are processes or procedures that produce a needed result, whether it is computation or the end-products of biochemical pathways. Such strings of decision-node selections are anything but random. And they are not "self-ordered" by redundant cause-and-effect necessity. Every successive nucleotide is a quaternary "switch setting." Many nucleotide selections in the string are not critical. But those switch-settings that determine folding, especially, are highly "meaningful." Functional switch-setting sequences are produced only by uncoerced selection pressure. There is a cybernetic aspect of life processes that is directly analogous to that of computer programming. More attention should be focused on the reality and mechanisms of selection at the decision-node level of biological algorithms. This is the level of covalent bonding in primary structure. Environmental selection occurs at the level of post-computational halting. The fittest already-computed phenotype is selected.
We can hypothesize that metabolism "just happened," independent of directions, in a prebiotic environment billions of years ago. But we can hypothesize anything. The question is whether such hypotheses are plausible. Plausibility is often eliminated when probabilities exceed the "universal probability bound" [132]. The stochastic "self-organization" of even the simplest biochemical pathways is statistically prohibitive by hundreds of orders of magnitude. Without algorithmic programming to constrain (more properly "control") options, the number of possible paths in sequence space for each needed biopolymer is enormous. 1015 molecules are often present in one test tube library of stochastic ensembles. But when multiple biopolymers must all converge at the same place at the same time to collectively interact in a controlled biochemically cooperative manner, faith in "self-organization" becomes "blind belief." No empirical data or rational scientific basis exists for such a metaphysical leap. Certainly no prediction of biological self-organization has been realized apart from SELEX-like bioengineering. SELEX is a selection/amplification methodology used in the engineering of new ribozymes [133-135]. Such investigator interference hardly qualifies as "self-organization." All of the impressive selection-amplification-derived ribozymes that have been engineered in the last fifteen years have been exercises in artificial selection, not natural selection.
Random sequences are the most complex (the least compressible). Yet empirical evidence of randomness producing sophisticated functionality is virtually nonexistent. Neither RSC nor OSC possesses the characteristics of informing or directing highly integrative metabolism. "Bits" of complexity alone cannot adequately measure or prescribe functional ("meaningful") bioinformation. Shannon information theory does not succeed in quantifying the kind of information on which life depends. It is called "information," but in reality we are quantifying only reduced combinatorial probabilistic uncertainty. This presupposes RSC. It is true that sophisticated bioinformation involves considerable complexity. But complexity is not synonymous with genetic instruction. Bioinformation exists as algorithmic programs, not just random combinations. And these programs require an operating system context along with common syntax and semantic "meanings" shared between source and destination.
The sequence of decision-node selections matters in how the polymer will finally fold. Folding is central to biofunction whether in a cell or a buffer in a test tube. In theory, the same protein can fold and unfold an infinite number of times via an ensemble of folding pathways [136]. But its favored minimal-free-energy molecular conformation is sequence dependent in the cell or assay mixture. The molecular memory for the conformation is the translated sequence. This is not to say that multiple sequences out of sequence space cannot assume the same conformation.
Nucleotides are grouped into triplet Hamming block codes [47], each of which represents a certain amino acid. No direct physicochemical causative link exists between codon and its symbolized amino acid in the physical translative machinery. Physics and chemistry do not explain why the "correct" amino acid lies at the opposite end of tRNA from the appropriate anticodon. Physics and chemistry do not explain how the appropriate aminoacyl tRNA synthetase joins a specific amino acid only to a tRNA with the correct anticodon on its opposite end.
Genes are not analogous to messages; genes are messages. Genes are literal programs. They are sent from a source by a transmitter through a channel (Fig. 3) within the context of a viable cell. They are decoded by a receiver and arrive eventually at a final destination. At this destination, the instantiated messages catalyze needed biochemical reactions. Both cellular and extracellular enzyme functions are involved (e.g., extracellular microbial cellulases, proteases, and nucleases). Making the same messages over and over for millions to billions of years (relative constancy of the genome, yet capable of changes) is one of those functions. Ribozymes are also messages, though encryption/decryption coding issues are absent. The message has a destination that is part of a complex integrated loop of information and activities. The loop is mostly constant, but new Shannon information can also be brought into the loop via recombination events and mutations. Mistakes can be repaired, but without the ability to introduce novel combinations over time, evolution could not progress. The cell is viewed as an open system with a semi-permeable membrane. Change or evolution over time cannot occur in a closed system. However, DNA programming instructions may be stored in nature (e.g., in permafrost, bones, fossils, amber) for hundreds to millions of years and be recovered, amplified by the polymerase chain reaction and still act as functional code. The digital message can be preserved even if the cell is absent and non-viable. It all depends on the environmental conditions and the matrix in which the DNA code was embedded. This is truly amazing from an information storage perspective.
Figure 3 Shannon's original 1948 communication diagram is here modified with an oval superimposed over the limits of Shannon's actual research. Shannon never left the confines of this oval to address the essence of meaningful communication. Any theory of Instruction would need to extend outside of the oval to quantify the ideal function and indirect "meaning" of any message.
A noisy channel is one that produces a high corruption rate of the source's signal (Fig. 3). Signal integrity is greatly compromised during transport by randomizing influences. In molecular biology, various kinds of mutations introduce the equivalent of noise pollution of the original instructive message. Communication theory goes to extraordinary lengths to prevent noise pollution of signals of all kinds. Given this longstanding struggle against noise contamination of meaningful algorithmic messages, it seems curious that the central paradigm of biology today attributes genomic messages themselves solely to "noise."
Selection pressure works only on existing successful messages, and then only at the phenotypic level. Environmental selection does not choose which nucleotide to add next to a forming single-stranded RNA. Environmental selection is always after-the-fact. It could not have programmed primordial RNA genes. Neither could noise. Abel has termed this The GS Principle (Genetic Selection Principle) [137]. Differential molecular stability and happenstantial self- or mutual-replication are all that nature had to work with in a prebiotic environment. The environment had no goal or intent with which to "work." Wasted energy was just as good as "energy available for work" in a prebiotic world.
Denaturization factors like hydrolysis in water correspond to normal Second Law deterioration of the physical matrix of information retention. This results in the secondary loss of initial digital algorithmic integrity. This is another form of randomizing noise pollution of the prescriptive information that was instantiated into the physical matrix of nucleotide-selection sequences. But the particular physical matrix of retention should never be confused with abstract prescriptive information itself. The exact same message can be sent using completely different mass/energy instantiations. The Second Law operates on the physical matrix, not on the nonphysical conceptual message itself. The abstract message enjoys formal immunity from dynamic deterioration in the same sense that the mathematical laws of physics transcend the dynamics they model.
The purpose of biomessages is to produce and manage metabolic biofunctions, including the location, specificity, speed, and direction of the biochemical reactions. Any attempt to deny that metabolic pathways lack directionality and purpose is incorrect. Genes have undeniable "meaning" which is shared between source and destination (Fig. 3). Noise pollution of this "meaning" is greatly minimized by ideally optimized redundancy coding [9] and impressive biological repair mechanisms [138-143].
For prescriptive information to be conveyed, the destination must understand what the source meant in order to know what to do with the signal. It is only at that point that a Shannon signal becomes a bona fide message. Only shared meaning is "communication." This shared meaning occurs within the context of a relatively stable cellular environment, unless conditions occur that damage/injure or kill the cells. Considerable universality of "meaning" exists within biology since the Last Universal Common Ancestor (LUCA). For this reason, messages can be retrieved by bacteria even from the DNA of dead cells during genetic transformation [144]. The entire message is not saved, but significant "paragraphs" of recipe. The transforming DNA may escape restriction and participate in recombination events in the host bacterial cell. A small part of the entire genome message can be recovered and expressed. Evolution then proceeds without a final destination or direction.
Shannon's uncertainty quantification "H" is maximized when events are equiprobable and independent of each other. Selection is neither. Since choice with intent is fundamentally non probabilistic, each event is certainly not equiprobable. And the successive decision-node choice-commitments of any algorithm are never independent, but integrated with previous and future choices to collectively achieve functional success.
The "uncertainty" ("H") of Shannon is an epistemological term. It is an expression of our "surprisal" [145] or knowledge "uncertainty." But humans can also gain definite after-the-fact empirical knowledge of which specific sequences work. Such knowledge comes closer to "certainty" than "uncertainty." More often than not in everyday life, when we use the term "information," we are referring to a relative certainty of knowledge rather than uncertainty. Shannon equations represent a very limited knowledge system. But functional bioinformation is ontological, not epistemological. Genetic instructions perform their functions in objective reality independent of any knowers.
Stochastic ensembles could happenstantially acquire functional sequence significance. But a stochastic ensemble is more likely by many orders of magnitude to be useless than accidentally functional. Apart from nonrandom selection pressure, we are left with the statistical prohibitiveness of a purely chance metabolism and spontaneous generation. Shannon's uncertainty equations alone will never explain this phenomenon. They lack meaning, choice, and function. FSC, on the other hand, can be counted on to work. FSC becomes the objective object of our relative certainty. Its objective function becomes known empirically. Its specific algorithmic switch-settings are worth enumerating. We do this daily in the form of "reference sequences" in genome projects, applied pharmacology research, and genetic disease mapping. Specifically enumerated sequencing coupled with observed function is regarded as the equivalent of a proven "halting" program. This is the essence of FSC.
Symbols can be instantiated into physical symbol vehicles in order to manipulate dynamics to achieve physical utility. Symbol selections in the string are typically correlated into conceptually coordinated holistic utility by some externally applied operating system or language of arbitrary (dynamically inert) rules. But functional sequence complexity is always mediated through selection of each unit, not through chance or necessity.
The classic example of FSC is the nucleic acid algorithmic prescription of polyamino acid sequencing. Codon sequence determines protein primary structure only in a conceptual operational context. This context cannot be written off as a subjective epistemological mental construction of humans. Transcription, post-transcriptional editing, the translation operational context, and post-translational editing, all produced humans. The standard coding table has been found to be close to conceptually ideal given the relative occurrence of each amino acid in proteins [146]. A triplet codon is not a word, but an abstract conceptual block code for a protein letter [47]. Block coding is a creative form of redundancy coding used to reduce noise pollution in the channel between source and destination [9].
Testable hypotheses about FSC
What testable empirical hypotheses can we make about FSC that might allow us to identify when FSC exists? In any of the following null hypotheses [137], demonstrating a single exception would allow falsification. We invite assistance in the falsification of any of the following null hypotheses:
Null hypothesis #1
Stochastic ensembles of physical units cannot program algorithmic/cybernetic function.
Null hypothesis #2
Dynamically-ordered sequences of individual physical units (physicality patterned by natural law causation) cannot program algorithmic/cybernetic function.
Null hypothesis #3
Statistically weighted means (e.g., increased availability of certain units in the polymerization environment) giving rise to patterned (compressible) sequences of units cannot program algorithmic/cybernetic function.
Null hypothesis #4
Computationally successful configurable switches cannot be set by chance, necessity, or any combination of the two, even over large periods of time.
We repeat that a single incident of nontrivial algorithmic programming success achieved without selection for fitness at the decision-node programming level would falsify any of these null hypotheses. This renders each of these hypotheses scientifically testable. We offer the prediction that none of these four hypotheses will be falsified.
The fundamental contention inherent in our three subsets of sequence complexity proposed in this paper is this: without volitional agency assigning meaning to each configurable-switch-position symbol, algorithmic function and language will not occur. The same would be true in assigning meaning to each combinatorial syntax segment (programming module or word). Source and destination on either end of the channel must agree to these assigned meanings in a shared operational context. Chance and necessity cannot establish such a cybernetic coding/decoding scheme [71].
How can one identify Functional Sequence Complexity empirically? FSC can be identified empirically whenever an engineering function results from dynamically inert sequencing of physical symbol vehicles. It could be argued that the engineering function of a folded protein is totally reducible to its physical molecular dynamics. But protein folding cannot be divorced from the causality of critical segments of primary structure sequencing. This sequencing was prescribed by the sequencing of Hamming block codes of nucleotides into triplet codons. This sequencing is largely dynamically inert. Any of the four nucleotides can be covalently bound next in the sequence. A linear digital cybernetic system exists wherein nucleotides function as representative symbols of "meaning." This particular codon "means" that particular amino acid, but not because of dynamical influence. No direct physicochemical forces between nucleotides and amino acids exist.
The relationship between RSC, OSC, and FSC
A second dimension can be added to Figure 1, giving Figure 2, to visualize the relation of Kolmogorov algorithmic compression to order and complexity. Order and complexity cannot be combined to generate FSC. Order and complexity are at opposite ends of the same bi-directional vector. Neither has any direct relationship to cybernetic choices for utility. FSC cannot be visualized on the unidimensional vector of Figure 1. FSC cannot be visualized within Figure 2, either, despite its added dimension of Kolmogorov compressibility. At least a third dimension is required to visualize the functionality of each sequence. Figure 4 emphasizes the difference between algorithmic compression vs. algorithmic function produced by the sequence itself. Algorithmic compression is something we do to the sequence to shorten it. Algorithmic function is something the sequence itself does in an operational context.
Figure 4 Superimposition of Functional Sequence Complexity onto Figure 2. The Y1 axis plane plots the decreasing degree of algorithmic compressibility as complexity increases from order towards randomness. The Y2 (Z) axis plane shows where along the same complexity gradient (X-axis) that highly instructional sequences are generally found. The Functional Sequence Complexity (FSC) curve includes all algorithmic sequences that work at all (W). The peak of this curve (w*) represents "what works best." The FSC curve is usually quite narrow and is located closer to the random end than to the ordered end of the complexity scale. Compression of an instructive sequence slides the FSC curve towards the right (away from order, towards maximum complexity, maximum Shannon uncertainty, and seeming randomness) with no loss of function.
Compression of language is possible because of repetitive use of letter and word combinations. Words correspond to reusable programming modules. The letter frequencies and syntax patterns of any language constrain a writer's available choices from among sequence space. But these constraints are the sole product of arbitrary intelligent choice within the context of that language. Source and destination reach a consensus of communicative methodology before any message is sent or received. This methodology is called a language or an operating system. Abstract concept ("choice contingency") determines the language system, not "chance contingency," and not necessity (the ordered patterning of physical "laws.")
Choice contingency is very different from chance contingency. Uncoerced (free) programming decisions have little in common with the self-ordering necessity of natural regularities. Neither necessity (OSC) nor chance (RSC) empirically displays any ability to instruct organization. At best, RSC and OSC per Prigogine's chaos theory [147] occasionally display self-ordering properties. "Self-ordering" is not the same as "self-organizing" despite abundant confusion in the literature. [148]. Organization requires Functional Sequence Complexity (FSC), not RSC or OSC. FSC in turn requires choice with algorithmic intent (in biology, selection for optimized biofunction and survivability).
What about some yet-to-be discovered "law" of nature? Couldn't that eventually explain FSC and the origin of life? Such a hope is based on a clouded understanding of a law and the dynamic "necessity" that a law describes. Laws are basically compression algorithms. Laws reduce reams of dynamic data down to parsimonious mathematical formulae. This is possible only because the behavior pattern is highly ordered, fixed, redundant, and low-informational. Degrees of freedom are severely constrained by necessity. But degrees of freedom are exactly what is required to generate FSC necessary for living organisms. Engineering decisions require freedom of selection. Environmental selection requires the same freedom. No undiscovered law can logically provide this needed dynamic decoherence and decoupling from law [149]. Decision nodes and configurable switches must be dynamically inert [58] with reference to their cybernetic function.
Controls are not the same as constraints. Controls cannot arise from self-ordering necessity where probability approaches 1.0 and uncertainty approaches 0 bits. Selection at decision nodes must be uncoerced to generate significant utility. Choice contingency realizes its maximum utility when options are equally available. Bits are maximized with 50:50 possibilities (unweighted means). Any natural-law constraints on selection only reduces bits, complexity, and algorithmic potential. Governance by any law would render flexible genetic control impossible. The organism's genome would lack freedom to respond to environmental changes. As a result, natural selection at the organism level would not be possible.
Proteins have a large number of decision nodes at which most of the twenty amino acids can be selected without total loss of protein function. Doesn't this fact negate the argument that life's recipe is digital and algorithmic? The answer is no. These sections of the protein Turing tape are merely non-interactive bends and buried sections of the highly-knotted globular protein. We would not expect all sections of the knotted protein to be critical to metabolic function. There are numerous other sections of the protein's sequence where only one or a few of the twenty amino acids can be used at each decision node. These sections contain the bulk of instructions. The other sections are somewhat like DNA introns that were once considered breaks between valued gene-product programming. The linear spacing within primary structure, after editing, is still critical. The exact messages have to occur at the right places in the one long protein molecule for it to fold and function properly. So instruction is inherent in the fact that the message segments must occur at positions 21–29, 67–71, 109–111, for example. The spacing of catalytic segments is itself a critical part of the digital algorithmic program.
FSC quantitative units: The problem of measuring meaning
If there is any hope of quantifying meaning, we would need units with which to measure prescriptive information, not just weighted-mean combinatorial probabilities and mutual uncertainties. The "laws" of physics and chemistry are mathematical relationships made possible by fixed units of measure in equations scaled by constants. With Shannon uncertainty as applied to computer science, it is possible to have a fixed unit of measure only because each unit represents the constant value of one binary choice opportunity. Computer bits and bytes provide an additive function measuring the number of binary decision nodes. A bit does not represent "a choice" as is commonly believed. If that were true, it would not be a unit of constant measure. "Yes" does not equal "No." "On' does not have the same meaning value as "off." If it did, computation would be impossible. Choices at decision nodes matter only because they are different. So we know immediately that no one unit of measure could possibly be assigned to a decision-node switch-setting. At present, we can only quantify the number of decision nodes irrespective of which choice is made at each decision node.
We get away with the sloppy definition of "bit" in computer science (a binary "choice") only because we are measuring the "space" requirements needed for any program with that number of binary decisions. Bits measure averages, never specific choice commitments made with intent. The latter is the essence of algorithmic programming. Bit measurements are generic. They tell us nothing about which choice was made at each decision node. Bit measurements cannot tell us whether a program has a bug, or computes at all.
We live under the illusion that bits measure choices because we fail to recognize the role of background knowledge that we bring to the strict mathematical measurement of "bits." Our minds assign added information to the bit computation. This background information tells us that this particular bit value is for a certain program with a certain function. But the computation itself knows nothing of meaning or function. Knowledge about the particular message and its purpose is totally external to its bit value. Shannon knew this well. Not only did he carefully exclude all discussion of meaning and specific message function from his research [2], but later warned against regarding his probabilistic combinatorial "uncertainty" measures as "information theory" [150].
The function of an algorithmic program is often lost through attempts at reduction. Prescriptive information is lost in the process. Many computations are all-or-none ends in themselves. Just as random sequences cannot be compressed, computational algorithms cannot always be compressed.
FSC cannot have individual units of fixed value. Does this negate the reality of FSC? If so, we have a lot of explaining to do for fields such as engineering and computer science that depend squarely upon FSC. Language, rationality, and the scientific method itself all depend upon FSC, not RSC or OSC. Science must recognize that there are legitimate aspects of reality that cannot always be reduced or quantified.
Conclusion
In summary, Sequence complexity can be 1) random (RSC), 2) ordered (OSC), or functional (FSC). OSC is on the opposite end of the bi-directional vectorial spectrum of complexity from RSC. FSC is usually paradoxically closer to the random end of the complexity scale than the ordered end. FSC is the product of nonrandom selection. FSC results from the equivalent of a succession of integrated algorithmic decision node "switch settings." FSC alone instructs sophisticated metabolic function. Self-ordering processes preclude both complexity and sophisticated functions. Self-ordering phenomena are observed daily in accord with chaos theory. But under no known circumstances can self-ordering phenomena like hurricanes, sand piles, crystallization, or fractals produce algorithmic organization. Algorithmic "self-organization" has never been observed [70] despite numerous publications that have misused the term [21,151-162]. Bone fide organization always arises from choice contingency, not chance contingency or necessity.
Reduced uncertainty (misnamed "mutual entropy") cannot measure prescriptive information (information that specifically informs or instructs). Any sequence that specifically informs us or prescribes how to achieve success inherently contains choice controls. The constraints of physical dynamics are not choice contingent. Prescriptive sequences are called "instructions" and "programs." They are not merely complex sequences. They are algorithmically complex sequences. They are cybernetic. Random sequences are maximally complex. But they don't do anything useful. Algorithmic instruction is invariably the key to any kind of sophisticated organization such as we observe in any cell. No method yet exists to quantify "prescriptive information" (cybernetic "instructions").
Nucleic acid prescription of function cannot be explained by "order out of chaos" or by "order on the edge of chaos" [163]. Physical phase changes cannot write algorithms. Biopolymeric matrices of high information retention are among the most complex entities known to science. They do not and can not arise from low-informational self-ordering phenomena. Instead of order from chaos, the genetic code was algorithmically optimized to deliver highly informational, aperiodic, specified complexity [164]. Specified complexity usually lies closer to the noncompressible unordered end of the complexity spectrum than to the highly ordered end (Fig. 4). Patterning usually results from the reuse of programming modules or words. But this is only secondary to choice contingency utilizing better efficiency. Order itself is not the key to prescriptive information.
The current usage of the word "complexity" in the literature represents a quagmire of confusion. It is an ill-defined, nebulous, often self-contradictory concept. We have defined FSC in a way that allows us to differentiate it from random and self-ordering phenomena, to frame testable empirical hypotheses about it, and to identify FSC when it exists.
Science has often progressed through the formulation of null hypotheses. Falsification allows elimination of plausible postulates [165,166]. The main contentions of this paper are offered in that context. We invite potential collaborators to join us in our active pursuit of falsification of these null hypotheses.
Abbreviations used in this paper
Random Sequence Complexity (RSC); Ordered Sequence Complexity (OSC); Functional Sequence Complexity (FSC).
Acknowledgements
Biographical: Dr. David L. Abel is a theoretical biologist focusing on primordial biocybernetics. He is the Program Director of The Gene Emergence Project, an international consortium of scientists pursuing the derivation of initial biocybernetic/biosemiotic function. DLA is supported by grants from The Origin-of-Life Foundation, Inc. a 501-c-3 science foundation. Jack T. Trevors is Professor of Environmental Biology/Microbiologist at the University of Guelph. Professor Trevors is also an editor for the The Journal of Microbiological Methods, editor-in-chief of Water, Air and Soil Pollution and a Fellow of the American Academy of Microbiology. JTT is supported by an NSERC (Canada) Discovery Grant.
==== Refs
Shannon CE Weaver W The Mathematical Theory of Communication 1949 Urbana, IL , University of Illinois Press
Shannon C Part I and II: A mathematical theory of communication The Bell System Technical Journal 1948 XXVII 379 423
Shannon C Part III: A mathematical theory of communication The Bell System Technical Journal 1948 XXVII 623 656
Kolmogorov AN Three approaches to the definition of the concept "quantity of information" Problems Inform Transmission 1965 1 1 7
Li M Vitanyi P Gries DSFB An Introduction to Kolmogorov Complexity and Its Applications Graduate Texts in Computer Science 1997 2 New York , Springer-Verlag 637
Chaitin GJ Algorithmic information theory IBM Journal of Research and Development 1977 21 350 359
Chaitin GJ Computational complexity and Godel's incompleteness theorem and To a mathematical definition of life Monographs in computer science and computer applications, no 3/70 1970 [Rio de Janeiro, , Centro Tâecnico Cientâifico, Pontifâicia Universidade Catâolica do Rio de Janeiro 10, 12 l.
Chaitin GJ On the length of programs for computing finite binary sequences J ACM 1966 13 547 10.1145/321356.321363
Hamming RW Coding and Information Theory 1986 Englewood Cliffs, N.J. , Prentice Hall
MacKay DM Information, Mechanism and Meaning 1969 Cambridge, MA , M.I.T. Press
Badii R Politi A Complexity : hierarchical structures and scaling in physics Cambridge nonlinear science series ; 6 1997 Cambridge ; New York , Cambridge University Press xiii, 318 p.
Adami C Ofria C Collier TC Evolution of biological complexity Proc Natl Acad Sci U S A 2000 97 4463 4468 10781045 10.1073/pnas.97.9.4463
Adami C Cerf NJ Physical complexity of symbolic sequences Physica D 2000 137 62 69 10.1016/S0167-2789(99)00179-7
Adami C What is complexity? Bioessays 2002 24 1085 1094 12447974 10.1002/bies.10192
Allison L Stern L Edgoose T Dix TI Sequence complexity for biological sequence analysis Comput Chem 2000 24 43 55 10642879
Bennett DH Herken R Logical depth and physical complexity The Universal Turing Machine: a Half-Century Survey 1988 Oxford , Oxford University Pres
Bennett CH Zurek WH How to define complexity in physics, and why Complexity, Entropy and the Physics of Information, SFI studies in the Sciences of Complexity 1990 8 137 148
Dewey TG Algorithmic complexity of a protein Physical Review E Statistical Physics, Plasmas, Fluids, and Related Interdisciplinary Topics 1996 54 R39 R41. 9965170
Gell-Mann M What is complexity? Complexity 1995 1 16 19
Gell-Mann M Lloyd S Information measures, effective complexity, and total information Complexity 1996 2 44 52 10.1002/(SICI)1099-0526(199609/10)2:1<44::AID-CPLX10>3.0.CO;2-X
Kauffman S At Home in the Universe: The Search for the Laws of Self-Organization and Complexity 1995 New York , Oxford University Press 320
Konopka AK Owens J Complexity charts can be used to map functional domains in DNA Genet Anal Tech Appl 1990 7 35 38 2328140 10.1016/0735-0651(90)90010-D
Konopka AK Smith D Sequences and Codes: Fundamentals of Biomolecular Cryptology Biocomputing: Informatics and Genome Projects 1994 San Diego , Academic Press 119 174
Konopka AK Cooper DN Sequence complexity and composition. Nature Encyclopedia of the Human Genome Vol 5 2003 London , Nature Publishing Group Reference 217 224
Lempel A Ziv J On the complexity of finite sequences IEEE Trans Inform Theory 1976 22 75 10.1109/TIT.1976.1055501
Lenski RE Ofria C Collier TC Adami C Genome complexity, robustness and genetic interactions in digital organisms Nature 1999 400 661 664 10458160 10.1038/23245
Nicolaou KC Creating complexity – the beauty and logic of synthesis Chem Commun 2003 6 661 6664 10.1039/b212248k
Ricard J What do we mean by biological complexity? C R Biol 2003 Feb;326 133 140 12754932 10.1016/S1631-0691(03)00064-7
Weng G Bhalla US Iyengar R Complexity in biological signaling systems Science 1999 284 92 96 10102825 10.1126/science.284.5411.92
Zurek WH Thermodynamic cost of computation, algorithmic complexity, and the information metric Nature 1989 341 119 124 10.1038/341119a0
Pincus SM Irregularity and asynchrony in biologic network signals Methods Enzymol 2000 321 149 182 10909056
Conrad M Unity of measurement and motion Biosystems 2001 60 23 38 11325501 10.1016/S0303-2647(01)00105-8
Rosen R Essays On Life Itself 2000 New York , Columbia University Press
Rosen R Life Itself 1991 New York , Columbia University Press
Mikulecky DC The emergence of complexity: science coming of age or science growing old? Computers Chem 2001 25 341 348 10.1016/S0097-8485(01)00070-5
Mikulecky DC Network thermodynamics and complexity: a transition to relational systems theory Comput Chem 2001 25 369 391 11459352 10.1016/S0097-8485(01)00072-9
Lewontin R The Triple Helix: Gene, Organism and Environment 2000 Cambridge, MA , Harvard University Press 11127028
Lewontin RC Collado-Vides J, Smith T, Magasanik B Evolution as engineering Integrative Approaches to Molecular Biology 1996 Cambridge, MA , MIT Press
Gordon R Evolution Escapes Rugged Fitness Landscapes by Gene or Genome Doubling: the Blessing of Higher Dimensionality. Computers Chem 1994 18 325 3332 10.1016/0097-8485(94)85026-7
Konopka AK Grand metaphors of biology in the genome era Computers & Chemistry 2002 26 397 401 12144170 10.1016/S0097-8485(02)00024-4
Bar-Hillel Y Carnap R Semantic Information. British Journal for the Philosophy of Science 1953 4 147 157 10.1093/bjps/IV.14.147
Mayr E This Is Biology: The Science of the Living World 1997 Cambridge, MA , Harvard University Press
Bell GI Evolution of simple sequence repeats. Computers Chem 1996 20 41 448 10.1016/S0097-8485(96)80006-4
Reidys C Stadler PF Bio-molecular shapes and algebraic structures Computers Chem 1996 20 85 894 10.1016/S0097-8485(96)80010-6
Conrad M Origin of life and the underlying physics of the universe Biosystems 1997 42 177 190 9184763 10.1016/S0303-2647(97)01705-X
Yockey HP An application of information theory to the Central Dogma and the Sequence Hypothesis J Theor Biol 1974 46 369 406 4423458 10.1016/0022-5193(74)90005-8
Yockey HP Information Theory and Molecular Biology 1992 Cambridge , Cambridge University Press 408
Yockey HP Information theory, evolution and the origin of life Information Sciences 2002 141 219 225 10.1016/S0020-0255(02)00173-1
Yockey HP Origin of life on earth and Shannon's theory of communication Comput Chem 2000 24 105 123 10642883 10.1016/S0097-8485(99)00050-9
Maeshiro T Kimura M The role of robustness and changeability on the origin and evolution of genetic codes PNAS 1998 95 5088 5093 9560233 10.1073/pnas.95.9.5088
Snyder L Champness W Molecular Genetics of Bacteria 1997 Washington, DC, USA , ASM Press
Lazcano A Miller SL The origin and early evolution of life: prebiotic chemistry, the pre-RNA world, and time Cell 1996 85 793 798 8681375 10.1016/S0092-8674(00)81263-5
Kolb VM Dworkin JP Miller SL Alternative bases in the RNA world: the prebiotic synthesis of urazole and its ribosides J Mol Evol 1994 38 549 557 11539446 10.1007/BF00175873
Dworkin JP Lazcano A Miller SL The roads to and from the RNA world J Theor Biol 2003 222 127 134 12699739 10.1016/S0022-5193(03)00020-1
Gesteland RF Cech TR Atkins JF The RNA World 1999 2 Cold Spring Harbor , Cold Spring Harbor Laboratory Press
Monnard PA Deamer DW Preparation of vesicles from nonphospholipid amphiphiles Methods Enzymol 2003 372 133 151 14610811
Kozlov IA Zielinski M Allart B Kerremans L Van Aerschot A Busson R Herdewijn P Orgel LE Nonenzymatic template-directed reactions on altritol oligomers, preorganized analogues of oligonucleotides Chemistry 2000 6 151 155 10747399 10.1002/(SICI)1521-3765(20000103)6:1<151::AID-CHEM151>3.3.CO;2-8
Rocha LM Evolution with material symbol systems Biosystems 2001 60 95 121 11325506 10.1016/S0303-2647(01)00110-1
Monnard PA Kanavarioti A Deamer DW Eutectic phase polymerization of activated ribonucleotide mixtures yields quasi-equimolar incorporation of purine and pyrimidine nucleobases J Am Chem Soc 2003 125 13734 13740 14599212 10.1021/ja036465h
Huang W Ferris JP Synthesis of 35-40 mers of RNA oligomers from unblocked monomers. A simple approach to the RNA world Chem Commun (Camb) 2003 12 1458 1459 12841290 10.1039/b303134a
Shapiro R Comments on `Concentration by Evaporation and the Prebiotic Synthesis of Cytosine' Origins Life Evol Biosph 2002 32 275 278 10.1023/A:1016582308525
Shapiro R Prebiotic cytosine synthesis: a critical analysis and implications for the origin of life Proc Natl Acad Sci U S A 1999 96 4396 4401 10200273 10.1073/pnas.96.8.4396
Devlin K Logic and Information 1991 New York , Cambridge University Press
Barwise J Perry J Situations and Attitudes 1983 Cambridge, Ma , MIT Press
Floridi L Floridi L Information The Blackwell Guide to the Philosophy of Computing and Information 2003 Oxford , Blackwell
Floridi L Open problems in the philosphy of information Metaphilosophy 2003 In Press
Dretske F Knowledge and the Flow of Information 1981 Cambridge, Mass , MIT Press
Abel DL What is Life? 1997 (under Definitions); (Last accessed July 2005)
Abel DL Palyi G, Caglioti L, Zucchi C Is Life Reducible to Complexity? Workshop on Life: a satellite meeting before the Millenial World Meeting of University Professors; Modena, Italy 2000 University of Modena and Reggio emilia 3 4
Abel DL Palyi G, Caglioti L, Zucchi C Is Life Reducible to Complexity?: Modena, Italy.
Fundamentals of Life 2002 Paris: Elsevier 57 72
Trevors JT Abel DL Chance and necessity do not explain the origin of life Cell Biology International 2004 28 729 739 15563395 10.1016/j.cellbi.2004.06.006
Pattee HH Waddington CH Physical problems with heredity and evolution Towards a Theoretical Biology 2 1969 Edinburgh , Edinburgh University Press
Pattee HH Coombs M The physics of symbols and the evolution of semiotic controls Proc Workshop on Control Mechanisms for Complex Systems 1997 Addison-Wesley http://www.ssie.binghamton.edu/pattee/semiotic.html
Pattee HH Lang A How does a molecule become a message? Communication in Development; Twenty-eighth Symposium of the Society of Developmental Biololgy 1969 New York , Academic Press 1 16
Pattee HH The physics of symbols: bridging the epistemic cut Biosystems 2001 60 5 21 11325500 10.1016/S0303-2647(01)00104-6
Hoffmeyer J Code-duality and the epistemic cut Ann N Y Acad Sci 2000 901 175 186 10818569
Pattee HH The complementarity principle and the origin of macromolecular information Biosystems 1979 11 217 226 497371 10.1016/0303-2647(79)90013-3
Pattee HH Complementarity vs. reduction as explanation of biological complexity Am J Physiol 1979 236 R241 6 443402
Pattee HH The complementarity principle in biological and social structures Journal of Social and Biological Structure 1978 1 191 200 10.1016/S0140-1750(78)80007-4
Pattee HH Dynamic and linguistic modes of complex systems Int J General Systems 1977 3 259 266
Walter DO Invited comments on Pattee's reduction to complementarity Am J Physiol 236: 1979 236 R247 8
Vaneechoutte M The scientific origin of life. Considerations on the evolution of information, leading to an alternative proposal for explaining the origin of the cell, a semantically closed system Ann N Y Acad Sci 2000 901 139 147 10818565
Pattee HH Cell psychology: an evolutionary approach to the symbol-matter problem Cognition and Brain Theory 1982 5 325 341
Pattee HH Evolving Self-Reference: Matter, Symbols, and Semantic Closure Communication and Cognition-Artificial Intelligence 1995 12 9 28
Umerez J Moran F, Moreno JJ, Chacon P Semantic Closure: a guiding notion to ground Artificial Life Advances in Artifiical Life 1995 Berlin , Springer-Verlag 77 94
Pattee HH Waddington CH Laws and constraints, symbols and languages Towards a Theoretical Biology 1972 4 Edinburgh , University of Edinburgh Press 248 258
Pattee HH Locker A Physical problems of the origin of natural controls Biogenesis, Evolution, and Homeostasis 1973 Heidelberg , Springer-Verlag 41 49
Pattee HH Artificial Life Needs a Real Epistemology: Granada, Spain.
1995 1
Pattee HH Irreducible and complementary semiotic forms Semiotica 2001 134 341 358 10.1515/semi.2001.032
Chaitin GJ Algorithmic Information Theory 1988 Revised Second Printing Cambridge , Cambridge University Press 3181327
Chaitin GJ Information, Randomness and Incompleteness 1987 Singapore , World Scientific
Yockey HP A calculation of the probability of spontaneous biogenesis by information theory J Theor Biol 1977 67 377 398 198618 10.1016/0022-5193(77)90044-3
Yockey HP Palyi G, Zucchi C, Caglioti L Information theory, evolution, and the origin of life Fundamentals of Life 2002 Paris , Elsevier 335 348
Yockey HP Informatics, Information Theory, and the Origin of Life: Duke University; Research Triangle Park.
2002
Yockey HP Self organization origin of life scenarios and information theory J Theor Biol 1981 91 13 31 7300380 10.1016/0022-5193(81)90370-2
Yockey HP Can the central dogma be derived from information theory? J Theor Biol 1978 74 149 152 713565 10.1016/0022-5193(78)90295-3
Ferris JP Hill ARJ Liu R Orgel LE Synthesis of long prebiotic oligomers on mineral surfaces Nature 1996 381 59 61 8609988 10.1038/381059a0
Joyce GF Orgel LE Gesteland RF, Cech TR, Atkins JF Prospects for understanding the origin of the RNA World The RNA World 1999 Second Cold Spring Harbor, NY , Cold Spring Harbor Laboratory Press 49 78
Adami C Introduction to Artificial Life 1998 New York , Springer/Telos 374
Schneider TD Theory of molecular machines. I. Channel capacity of molecular machines J Theor Biol 1991 148 83 123 2016886 10.1016/S0022-5193(05)80466-7
Schneider TD Theory of molecular machines. II. Energy dissipation from molecular machines J Theor Biol 1991 148 125 137 2016881 10.1016/S0022-5193(05)80467-9
Schneider TD Information content of individual genetic sequences J Theor Biol 1997 189 427 441 9446751 10.1006/jtbi.1997.0540
Schneider TD Measuring molecular information J Theor Biol 1999 201 87 92 10534438 10.1006/jtbi.1999.1012
Schneider TD Evolution of biological information Nucleic Acids Res 2000 28 2794 2799 10908337 10.1093/nar/28.14.2794
Ferris JP Huang CH Hagan WJJ Montmorillonite: a multifunctional mineral catalyst for the prebiological formation of phosphate esters Orig Life Evol Biosph 1988 18 121 133 3368214 10.1007/BF01808786
Ferris JP Ertem G Oligomerization of ribonucleotides on montmorillonite: reaction of the 5'-phosphorimidazolide of adenosine Science 1992 257 1387 1389 1529338 10.1126/science.1529338
Ferris JP Catalysis and prebiotic RNA synthesis Orig Life Evol Biosph 1993 23 307 315 7509475 10.1007/BF01582081
Miyakawa S Ferris JP Sequence- and regioselectivity in the montmorillonite-catalyzed synthesis of RNA J Am Chem Soc 2003 125 8202 8208 12837090 10.1021/ja034328e
Gilbert W Origin of life -- the RNA World Nature 1986 319 618 10.1038/319618a0
Cairns-Smith AG Seven Clues to the Origin of Life 1990 Canto Cambridge , Cambridge University Press 130
Cairns-Smith AG The origin of life and the nature of the primitive gene J Theor Biol 1966 10 53 88 5964688 10.1016/0022-5193(66)90178-0
Cairns-Smith AG Takeover mechanisms and early biochemical evolution Biosystems 1977 9 105 109 907803 10.1016/0303-2647(77)90019-3
Cairns-Smith AG Walker GL Primitive metabolism Curr Mod Biol 1974 5 173 186 4407424
Segre D Ben-Eli D Lancet D Compositional genomes: prebiotic information transfer in mutually catalytic noncovalent assemblies Proc Natl Acad Sci U S A 2000 97 4112 4117 10760281 10.1073/pnas.97.8.4112
Segre D Lancet D Kedem O Pilpel Y Graded autocatalysis replication domain (GARD): kinetic analysis of self-replication in mutually catalytic sets Orig Life Evol Biosph 1998 28 501 514 10.1023/A:1006583712886
Guimaraes RC Linguistics of biomolecules and the protein-first hypothesis for the origins of cells Journal of Biological Physics 1994 20 193 199 10.1007/BF00700436
Shapiro R A replicator was not involved in the origin of life IUBMB Life 2000 49 173 176 10868906 10.1080/152165400306160
Freeland SJ Knight RD Landweber LF Do proteins predate DNA? Science 1999 286 690 692 10577226 10.1126/science.286.5440.690
Rode BM Peptides and the origin of life Peptides 1999 20 773 786 10477135 10.1016/S0196-9781(99)00062-5
Wong JT A co-evolution theory of the genetic code Proc Natl Acad Sci U S A 1975 72 1909 1912 1057181 10.1073/pnas.72.5.1909
Zhao YF Cao P Phosphoryl amino acids: Common origin for nucleic acids and protein Journal of Biological Physics 1994 20 283 287 10.1007/BF00700446
Zhou W Ju Y Zhao Y Wang Q Luo G Simultaneous formation of peptides and nucleotides from N-phosphothreonine Orig Life Evol Biosph 1996 26 547 560 9008881 10.1007/BF01808220
Nashimoto M The rna/protein symmetry hypothesis: experimental support for reverse translation of primitive proteins. J Theor Biol 2001 209 181 187 11401460 10.1006/jtbi.2000.2253
Dyson FJ Origins of Life 1998 2nd Cambridge , Cambridge University Press
Dyson F Life in the Universe: Is Life Digital or Analog?: Greenbelt, MD.
1999
Dyson FJ A model for the origin of life J Mol Evol 1982 18 344 350 7120429 10.1007/BF01733901
Prigogine I From Being to Becoming 1980 San Francisco , W. H. Freeman and Co
Pattee HH Waddington CH The physical basis of coding and reliabiity in biological evolution Prolegomena to Theoretical Biology 1968 Edinburgh , University of Edinburgh
Pattee HH Casti JL, Karlqvist A Universal principles of measurement and language functions in evolving systems. Complexity, Language, and Life: Mathematical Approaches 1986 Berlin , Springer-Verlag 579 581
Pattee HH On the origin of macromolecular sequences Biophys J 1961 1 683 710 14484391
Pattee HH Rosen R The nature of hierarchichal controls in living matter Foundations of Mathematical Biology 1971 1 New York , Academic Press 1 22
Borel E Probabilities and Life 1962 New York , Dover
Ellington AD Szostak JW In vitro selection of RNA molecules that bind specific ligands Nature 1990 346 818 822 1697402 10.1038/346818a0
Tuerk C Gold L Systematic evolution of ligands by exponential enrichment -- RNA ligands to bacteriophage - T4 DNA-polymerase Science 1990 249 505 510 2200121 10.1126/science.2200121
Robertson DL Joyce GF Selection in virtro of an RNA enzyme that specifically cleaves single-stranded DNA Nature 1990 344 467 468 1690861 10.1038/344467a0
Rhoades E Gussakovsky E Haran G Watching proteins fold one molecule at a time PNAS 2003 100 3197 3202 12612345 10.1073/pnas.2628068100
Abel DL Trevors JT More than metaphor: Genomes are objective sign systems. Journal of Biosemiotics 2005 (In press)
Bernardi F Ninio J The accuracy of DNA replication Biochimie 1978 60 1083 1095 367448
Liebovitch LS Tao Y Todorov AT Levine L Is there an error correcting code in the base sequence in DNA? Biophys J 1996 71 1539 1544 8874027
Ninio J Kinetic devices in protein synthesis, DNA replication, and mismatch repair Cold Spring Harb Symp Quant Biol 1987 52 639 646 3454281
Ninio J Illusory defects and mismatches: why must DNA repair always be (slightly) error prone? Bioessays 2000 22 396 401 10723037 10.1002/(SICI)1521-1878(200004)22:4<396::AID-BIES10>3.0.CO;2-K
Oefner PJ Sequence variation and the biological function of genes: methodological and biological considerations J Chromatogr B Analyt Technol Biomed Life Sci 2002 782 3 25 12457993 10.1016/S1570-0232(02)00699-2
Sullenger BA Cech TR Ribozyme-mediated repair of defective mRNA by targeted, trans-splicing Nature 1994 371 619 622 7935797 10.1038/371619a0
Timms-Wilson TM Van Overbeek LS Bailey MJ Trevors JT Van Elsas JD C. J. Hurst GRKMJMILDSMVW Quantification of gene transfer in soil and rhizosphere. Manual of Environmental Microbiology 2001 Washington, D.|C. , American Society for Microbiology, ASM Press, Washington, D.C Chapter 61
Tribus M Thermostatics and Thermodynamics. 1961 Princeton, N. J. , D. van Nostrand Company, Inc.,
Bradley D Informatics. The genome chose its alphabet with care Science 2002 297 1789 1791 12228695 10.1126/science.297.5588.1789b
Prigogine I Stengers I Order Out of Chaos 1984 London , Heinemann
Denton MJ Marshall CJ Legge M The Protein Folds as Platonic Forms: New Support for the Pre-Darwinian Conception of Evolution by Natural Law J Theor Biol 2002 219 325 342 12419661 10.1006/jtbi.2002.3128
Rocha MR Hordijk W Material representations: From the genetic code to the evolution of cellular automata Artificial Life 2005 In Press
Shannon EC Intersymbol influence reduces information content Bell Sys Tech J 1951 30 50
Eigen M Self-organization of matter and the evolution of biological macromolecules Naturwissenschaften 1971 58 465 523 4942363 10.1007/BF00623322
Kauffman SA The Origins of Order: Self-Organization and Selection in Evolution 1993 Oxford , Oxford University Press
Surrey T Nedelec F Leibler S Karsenti E Physical properties determining self-organization of motors and microtubules Science 2001 292 1167 1171 11349149 10.1126/science.1059758
Fuchslin RM McCaskill JS Evolutionary self-organization of cell-free genetic coding PNAS USA 2001 98 9185 9190 11470896 10.1073/pnas.151253198
Nicolis G Prigogine I Self-Organization in Nonequilibirum Systems: From Dissipative Structures to Order Through Fluctuations. 1977 New York , Wiley-Interscience
Atamas SP Self-organization in computer simulated selective systems Biosystems 1996 39 143 151 8866051 10.1016/0303-2647(96)01612-7
Martin W Russell MJ On the origins of cells: a hypothesis for the evolutionary transitions from abiotic geochemistry to chemoautotrophic prokaryotes, and from prokaryotes to nucleated cells Philos Trans R Soc Lond B Biol Sci 2003 358 59 83; discussion 83-85 12594918 10.1098/rstb.2002.1183
Eigen M Molecular self-organization and the early stages of evolution Experientia 1971 27 149 212 5134320
Eigen M Schuster P The hypercycle. A principle of natural self-organization. Part A: Emergence of the hypercycle Naturwissenschaften 1977 64 541 565 593400 10.1007/BF00450633
Epstein IR Eigen M Selection and self-organization of self-reproducing macromolecules under the constraint of constant flux Biophys Chem 1979 10 153 160 486701 10.1016/0301-4622(79)85035-8
Cariani P van de Vijver G, Salthe S, Delpos M Towards an evolutionary semiotics: the emergence of new sign-functions in organisms and devices Evolutionary Systems: Biological Perspectives on Selection and Self-Organization 1998 The Netherlands , Kluwer 359 376
Rocha LM Salthe S, van de Vijver G, Delpos M Selected self-organization and the semiotics of evolutionary systems Evolutionary Systems: Biological and Epistemological Perspectives on Selection and Self-Organization 1998 The Netherlands , Kluwer 341 358
Kauffman SA Investigations 2000 New York , Oxford University Press 286
Orgel LE The Origins of Life: Molecules and Natural Selection, p 189 1973 New York , John Wiley
Popper K The Logic of Scientific Discovery 1934 New York: Harper Reprinted 1959 edn
Popper K Conjectures and Refutations 1963 New York: Harper
|
16095527
|
PMC1208958
|
CC BY
|
2021-01-04 16:39:26
|
no
|
Theor Biol Med Model. 2005 Aug 11; 2:29
|
utf-8
|
Theor Biol Med Model
| 2,005 |
10.1186/1742-4682-2-29
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-571602662110.1186/1743-422X-2-57ResearchInhibition of Henipavirus fusion and infection by heptad-derived peptides of the Nipah virus fusion glycoprotein Bossart Katharine N [email protected] Bruce A [email protected] Gary [email protected] Lin-Fa [email protected] Bryan T [email protected] Christopher C [email protected] CSIRO Livestock Industries, Australian Animal Health Laboratory, Geelong, Victoria 3220, Australia2 Department of Microbiology and Immunology, Uniformed Services University, Bethesda, MD 20814, USA2005 18 7 2005 2 57 57 24 5 2005 18 7 2005 Copyright © 2005 Bossart et al; licensee BioMed Central Ltd.2005Bossart et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The recent emergence of four new members of the paramyxovirus family has heightened the awareness of and re-energized research on new and emerging diseases. In particular, the high mortality and person to person transmission associated with the most recent Nipah virus outbreaks, as well as the very recent re-emergence of Hendra virus, has confirmed the importance of developing effective therapeutic interventions. We have previously shown that peptides corresponding to the C-terminal heptad repeat (HR-2) of the fusion envelope glycoprotein of Hendra virus and Nipah virus were potent inhibitors of both Hendra virus and Nipah virus-mediated membrane fusion using recombinant expression systems. In the current study, we have developed shorter, second generation HR-2 peptides which include a capped peptide via amidation and acetylation and two poly(ethylene glycol)-linked (PEGylated) peptides, one with the PEG moity at the C-terminus and the other at the N-terminus. Here, we have evaluated these peptides as well as the corresponding scrambled peptide controls in Nipah virus and Hendra virus-mediated membrane fusion and against infection by live virus in vitro.
Results
Unlike their predecessors, the second generation HR-2 peptides exhibited high solubility and improved synthesis yields. Importantly, both Nipah virus and Hendra virus-mediated fusion as well as live virus infection were potently inhibited by both capped and PEGylated peptides with IC50 concentrations similar to the original HR-2 peptides, whereas the scrambled modified peptides had no inhibitory effect. These data also indicate that these chemical modifications did not alter the functional properties of the peptides as inhibitors.
Conclusion
Nipah virus and Hendra virus infection in vitro can be potently blocked by specific HR-2 peptides. The improved synthesis and solubility characteristics of the second generation HR-2 peptides will facilitate peptide synthesis for pre-clinical trial application in an animal model of Henipavirus infection. The applied chemical modifications are also predicted to increase the serum half-life in vivo and should increase the chance of success in the development of an effective antiviral therapy.
ParamyxovirusHendra virusNipah virusenvelope glycoproteinfusioninfectioninhibitionantiviral therapies
==== Body
Background
Two novel zoonotic paramyxoviruses have emerged to cause disease in the past decade, Hendra virus (HeV) in Australia in 1994–5 [1], and Nipah virus (NiV) in Malaysia in 1999 [2]. HeV and NiV caused severe respiratory and encephalitic disease in animals and humans (reviewed in [3,4]), HeV was transmitted to humans by close contact with infected horses; NiV was passed from infected pigs to humans. Both are unusual among the paramyxoviruses in their ability to infect and cause potentially fatal disease in a number of host species, including humans. Both viruses also have an exceptionally large genome and are genetically closely related yet distinct from all other paramyxovirus family members. Due to their unique genetic and biological properties, HeV and NiV have been classified as prototypic members of the new genus Henipavirus, in the family Paramyxoviridae [5,6]. Serological surveillance and virus isolation studies indicated that HeV and NiV reside naturally in flying foxes in the genus Pteropus (reviewed in [7]). Investigation of possible mechanisms precipitating their emergence indicates ecological changes resulting from deforestation, human encroachment into bat habitats and high intensity livestock farming practices as the likely primary factors [7]. Because these viruses are harboured in a mammalian reservoir whose range is vast, both HeV and NiV have the capability to cause disease over a large area and in new regions where disease was not seen previously. There have been several other suspected NiV occurrences since its recognition in 1999. Recently two confirmed outbreaks in 2004 in Bangladesh caused fatal encephalitis in humans and for the first time, person-to-person transmission appeared to have been a primary mode of spread [8-12]. In addition, there appeared to be direct transmission of the virus from the flying fox to humans, and the case mortality rate was ~70%; significantly higher than any other NiV outbreak to date. Currently, HeV and NiV are categorized as biological safety level-4 (BSL-4) pathogens, and NiV has also been classified as a category C priority pathogen. Category C agents include emerging pathogens that could be engineered for mass dissemination in the future because of availability; ease of production and dissemination; and potential for high morbidity and mortality and major health impact. All of the above reasons illustrate why an effective antiviral therapy is needed for henipaviruses.
Paramyxoviruses contain two membrane-anchored glycoproteins that are required for virion attachment to and fusion with the membrane of the host cell. Depending on the biological properties of the virus, the attachment protein is referred to as either the hemagglutinin-neuraminidase (HN), the hemagglutinin (H), or the G glycoprotein which lacks hemagglutinating and neuraminidase activities. Whereas most paramyxoviruses employ sialic acid moieties as receptors, HeV and NiV make use of a cell-surface expressed protein and their G glycoprotein binds to ephrin-B2 on host cells [13]. The fusion protein (F) facilitates the fusion of virion and host cell membranes during virus infection, and is an oligomeric homotrimer [14,15]. The biologically active F protein consists of two disulfide linked subunits, F1 and F2, which are generated by the proteolytic cleavage of a precursor polypeptide known as F0 (reviewed in [16,17]). In all cases the membrane-anchored subunit, F1, contains a new amino terminus that is hydrophobic and highly conserved across virus families and referred to as the fusion peptide (reviewed in [18]). There have been considerable advances in the understanding of the structural features and development of mechanistic models of how several viral envelope glycoproteins function in driving the membrane fusion reaction (reviewed in [19-21]). One important feature of many of these fusion glycoproteins are two α-helical domains referred to as heptad repeats (HR) that are involved in the formation of a trimer-of-hairpins structure [22,23]. HR-1 is located proximal to the amino (N)-terminal fusion peptide and HR-2 precedes the transmembrane domain near the carboxyl (C)-terminus [22,24-26]. For many viral fusion glycoproteins the N-terminal HR-1 forms an interior, trimeric coiled-coil surrounded by three anti-parallel helices formed from HR-2 (reviewed in [18]). Both the HeV and NiV F glycoprotein HR domains have been shown to interact with each other and form the typical 6-helix coiled-coil bundles [24,27].
Peptide sequences from either HR domain of the F glycoprotein of several paramyxoviruses, including HeV and NiV have been shown to be inhibitors of fusion [25,28-35]. Targeting this membrane fusion step of the viral infection process has garnered much attention, primarily lead by work on human immunodeficiency virus type 1 (HIV-1) (reviewed in [36]). Indeed, the HIV-1 envelope derived peptide, enfuvirtide (Fuzeon™, formerly T-20), has been clinically successful [37,38]. Enfuvirtide is a 36-amino acid peptide corresponding to a portion of the C-terminal HR-2 domain of the gp41 subunit of the envelope glycoprotein. Approved by the FDA in March 2003, enfuvirtide has been shown to be comparable to other anti-retroviral therapeutics in terms of reducing viral load, and is generally well tolerated despite its parenteral administration, and enfuvirtide has added significantly to optimized combination therapy in a growing number of patients with multiple HIV-1 resistance to the currently available antiretroviral drugs [39].
No therapeutic treatments are currently available for HeV or NiV infection. In our previous studies, we demonstrated that peptides derived from the HR-2 of either the HeV or NiV F were potent inhibitors of fusion [34]. However, although these peptides were effective, their specific properties such as overall length where not optimized, and they were large and somewhat insoluble making synthesis and purification problematic. In preparation to evaluate these peptides as potential therapeutic fusion inhibitors against NiV and HeV infection, second generation versions were designed with changes aimed at improving their solubility and in vivo half-life when administered to animals. In the current study, we have produced shorter 36 amino acid capped peptides by amidation at the N-terminus and acetylation at the carboxyl-terminus. In addition, two alternate peptide versions were made with the addition of a poly(ethylene glycol) moiety to either the C-terminus or the N-terminus. Here we report on the biological activity of these modified peptides and demonstrate that chemical modification increased solubility significantly without altering their biological properties of inhibiting membrane fusion. Further, all three versions were capable of blocking both fusion as well as live HeV and NiV infection with IC50 concentrations in the nM range, similar to those reported with other viral systems.
Results
Heptad peptide inhibition of Hendra virus and Nipah virus-mediated cell-cell fusion
Hypothetical models of the transmembrane (F1) glycoproteins of HeV and NiV are shown in Fig. 1. The models are derived by homology modeling with the known structure of the F protein of Newcastle disease virus [40]. These models are consistent protein structures predicted by the computer algorithms PHDsec [41] and TMpred [42]. Overall, the structures of the HeV and NiV F1 transmembrane subunit, including the heptad repeats (HR-1 and HR-2 helices), closely resemble that of the gp41 subunit of the HIV-1 envelope glycoprotein [43-45]. The depicted circle in the background represents the F2 subunit of NiV F. Due to the structural similarities and clinical success of the gp41 heptad peptides, we hypothesized that peptides derived from the HR-2 of HeV or NiV F would be effective antiviral therapies for henipavirus infection. In previous studies we evaluated the inhibition properties of 42 amino acid length peptides derived from both the N and C-terminal heptad repeats (HR-1 and HR-2) of HeV and NiV F in a vaccinia virus-based reporter gene assay that quantitatively measured cell-cell fusion mediated by the envelope glycoproteins of HeV and NiV [25,34]. Although both HR-1 and HR-2 derived peptides exhibited fusion inhibitory activity, the HR-2 peptide (residues 447–489) was more potent and more soluble. The HeV and NiV HR-2 peptides differed at three locations (amino acids 450, 479 and 480) with phenylalanine, arginine and leucine in NiV replaced by tyrosine, lysine and isoleucine in HeV [6,46]. These differences in the sequence of either peptide did not alter their homologous or heterologous inhibitory activity, suggesting that either peptide possessed potential therapeutic activity to both HeV and NiV. Here, we designed second generation versions of the NiV based HR-2 derived peptide with changes aimed at improving their solubility and in vivo half-life when administered to animals. Shorter, 36 amino acid capped peptides were synthesized (sequence denoted as FC2 in Fig. 1) by amidation at the N-terminus and acetylation at the carboxyl-terminus, modifications known to have improved in vivo half-life of Fuzeon™ (Thomas Matthews, Trimeris Inc., personal communication). In addition, two alternate peptide versions were made with the addition of a poly(ethylene glycol) moiety to either the C-terminus or the N-terminus which improved peptide solubility during preparation, and may also potentially improve the pharmacokinetics in vivo [47,48].
Figure 1 Hypothetical models of the transmembrane (F1) glycoproteins of Hendra virus and Nipah virus. The models are derived by homology modeling with the known structure of the F protein of Newcastle disease virus [40]. These models are consistent protein structures predicted by the computer algorithms PHDsec [41] and TMpred [42] as described in the Methods. The heptad repeats are indicated as HR-1 (grey) and HR-2 (yellow/orange), transmembrane anchor (blue). The F2 subunit is represented by the circle behind the F1 subunit. The 36 amino acid fusion inhibitor peptide sequence used in the present study is designated as FC2 and is boxed (yellow). The equivalent location of FC2 in the HeV F1 subunit is shown for comparison.
First, we examined the activity of the capped peptides on HeV and NiV-mediated membrane fusion. In previous studies, un-capped heptad-derived peptides had to be dissolved initially in 100% DMSO at concentrations between 50 and 500 μg/ml and then diluted in medium in order to maintain solubility. Here, the capped heptad-derived peptide (capped-NiV FC2) was completely soluble and dissolved in cell culture medium at concentrations as high as 10 mg/ml. For cell-cell fusion, envelope expressing-effector cells were added to peptides prior to the addition of target cells. Shown in Fig. 2 are the dose-dependent inhibition profiles of HeV (column one) and NiV-mediated (column 2) cell-cell fusion mediated by the capped-NiV FC2 peptide in Vero (Fig. 2A), U373 (Fig. 2B), and PCI 13 (Fig. 2C) cell lines. The scrambled, capped, control peptide (capped-ScNiV FC2) had no inhibitory effect, over the same concentration range, on the cell-fusion mediated by either virus in any of the three cell lines. NiV-mediated fusion appeared to be slightly more sensitive to peptide inhibition in comparison to the cell-fusion activity of HeV, although the calculated IC50 concentrations in each were comparable (Table 1). Importantly, the IC50 values of the capped version of NiV FC2 in these in vitro cell-cell fusion assays were within the 13–27 nM range, similar to what was observed in prior studies utilizing un-capped versions of the 42 amino acid heptad-derived peptides which yielded IC50 values between 5.2 and 5.8 nM [34].
Figure 2 Inhibition of Hendra virus and Nipah virus-mediated cell-cell fusion by capped C-terminal heptad peptide NiV FC2. HeLa cells were infected with vaccinia recombinants encoding HeV F and HeV G or NiV F and NiV G glycoproteins, along with a vaccinia recombinant encoding T7 RNA polymerase (effector cells). Each designated target cell type was infected with the E. coli LacZ-encoding reporter vaccinia virus vCB21R. Each target cell type (1 × 105) was plated in duplicate wells of a 96-well plate. Inhibition was carried out using either capped NiV FC2 or ScNiV FC2 (control) heptad peptide. Peptides were added to the HeV or NiV glycoprotein-expressing cells (1 × 105), incubated for 30 min at 37°C, and then each target cell type was added. The cell fusion assay was performed for 2.5 hr at 37°C, followed by lysis in Nonidet P-40 (1%) and β-Gal activity was quantified.
Table 1 Summary of 50% inhibitory concentration values of peptide fusion inhibitors in cell-cell fusion and virus infection assays.
Virus Cell line IC50* Capped NiV FC2 (nM) IC50 N-PEG NiV FC2 (nM) IC50 C-PEG NiV FC2 (nM)
Fusion Inhibition HeV Vero 17.59 6.54 142.4
NiV Vero 13.08 3.66 98.05
HeV U373 23.91 9.71 78.07
NiV U373 16.28 4.85 79.19
HeV PCI 13 27.54 6.18 147.2
NiV PCI 13 17.79 5.04 93.32
Live virus Inhibition HeV Vero 4.17 0.46 14.28
NiV Vero 11.42 1.36 43.76
HeV PCI 13 53.51 2.05 11.94
NiV PCI 13 2.70 1.26 55.57
*All IC50s were calculated using the non-linear regression function of GraphPad Prism software.
Using the cell-cell fusion assay we next examined the PEG-modified versions of NiV FC2. As predicted, these pegylated heptad peptides also possessed increased solubility characteristics and could be readily prepared at concentrations up to 10 mg/ml. The dose-response inhibition results of the N-PEG-NiV FC2 and C-PEG-NiV FC2 peptides are shown in Fig. 3, and inhibition was demonstrated in Vero (Fig. 3A), U373 (Fig. 3B), and PCI 13 (Fig. 3C) cell lines. Both pegylated versions of NiV FC2 were capable of blocking NiV and HeV-mediated cell-fusion, while the scrambled PEG-control peptide (C-PEG-ScNiV FC2) had no inhibitory activity. Because of the required specificity of the heptad peptide amino acid sequence to convey fusion inhibitory activity, as well as the high cost of peptide synthesis, we chose to only synthesize one version of the scrambled peptide as a pegylated control with the PEG10 moiety linked to the C-terminus. It was also noted that the NiV FC2 peptide with the PEG10 moiety added to the C-terminus had significantly reduced inhibitory capacity, as compared to PEG10 added to the N-terminus, against both NiV and HeV-mediated cell-fusion in all three cell lines tested. The reduction of C-PEG-NiV FC2 activity versus N-PEG-NiV FC2 was approximately 20-fold in all cases (Table 1) with the exception of HeV-mediated cell-fusion with the U373 cell line (Fig. 3B). Importantly, in all cases, the N-PEG-NiV FC2 demonstrated very similar IC50s (3–10 nM) to what was observed in prior studies utilizing un-capped versions of the 42 amino acid heptad-derived peptides (5–6 nM).
Figure 3 Inhibition of Hendra virus and Nipah virus-mediated cell-cell fusion by N-terminal and C-terminal (PEG10) pegylated heptad peptide NiV FC2. HeLa cells were infected with vaccinia recombinants encoding HeV F and HeV G or NiV F and NiV G glycoproteins, along with a vaccinia recombinant encoding T7 RNA polymerase (effector cells). Each designated target cell type was infected with the E. coli LacZ-encoding reporter vaccinia virus vCB21R. Each target cell type (1 × 105) was plated in duplicate wells of a 96-well plate. Inhibition was carried out using either the N-terminal (N-PEG-NiV FC2) or C-terminal (C-PEG-NiV FC2) pegylated and capped heptad peptides or C-terminal pegylated scrambled control peptide (C-PEG-ScNiV FC2). Peptides were added to the HeV or NiV glycoprotein-expressing cells (1 × 105), incubated for 30 min at 37°C, and then each target cell type was added The cell fusion assay was performed for 2.5 hr at 37°C, followed by lysis in Nonidet P-40 (1%) and β-Gal activity was quantified.
Heptad peptide inhibition of Hendra virus and Nipah virus infection
We next sought to confirm the inhibitory activity of Nipah virus heptad-derived peptides on the infection of live HeV and NiV in cell culture. We routinely employ Vero cell culture to perform live henipavirus infection assays, as well as in the propagation of virus stocks. The infection of Vero cells with HeV or NiV produced characteristic syncytial morphologies for each virus [49]. HeV reproducibly incorporated surrounding cells in the culture monolayer into each syncytium with the cell nuclei and viral proteins spread throughout the majority of the giant cell. In contrast, NiV infected syncytia initially demonstrated a similar appearance to their HeV counterparts, but characteristically both cell nuclei and viral protein were later sequestered around the periphery of each giant cell leaving the central region largely empty. In order to assess the extent of viral infection, we have developed an assay that will detect viral protein by immunofluorescence staining and localization of the P protein using a cross-reactive anti-P peptide-specific antiserum. Using this syncytia-based immunofluorescence infection assay, we initially tested the N-PEG NiV FC2 peptide for its ability to block virus infection and results are shown in Fig. 4. In the absence of peptide, the different syncytial morphologies of HeV and NiV- infected cells were clearly evident. In the HeV-infected syncytia (Fig. 4A), the viral P protein was spread throughout the majority of the giant cell; whereas, the NiV-infected syncytia (Fig. 4D) were circular structures delineated by a ring of the viral antigen. Incubation of 500 nM N-PEG-NiV FC2 with either HeV (Fig. 4B) or NiV (Fig. 4E) infected cells resulted in a dramatic and robust reduction in syncytial size although the number of syncytia per cell monolayer remained largely unchanged. In parallel, the incubation of 500 nM C-PEG-ScNiV FC2 control peptide with HeV or NiV-infected cells (Fig. 4C and 4F respectively) revealed a syncytial morphology and size identical to those observed in the absence of any peptide.
Figure 4 Immunofluorescence-based syncytia assay of Hendra virus and Nipah virus infection. Vero cells were plated into 96 well plates and grown to 90% confluence. Cells were pre-treated with heptad peptides for 30 min at 37°C prior to infection with 1.5 × 103 TCID50/ml and 7.5 × 102 TCID50/ml of live HeV or NiV (combined with peptide). Cells were incubated for 24 hours, fixed in methanol and immunofluorescently stained for P protein prior to digital microscopy. Images were obtained using an Olympus IX71 inverted microscope coupled to an Olympus DP70 high resolution color camera and all images were obtained at an original magnification of 85×. Representative images of FITC immunofluorescence of anti-P labeled HeV and NiV syncytia are shown. A: HeV without peptide. B: HeV with C-PEG-NiV FC2. C: HeV with N-PEG-ScNiV FC2. D: NiV without peptide. E: NiV with N-PEG-NiV FC2. F: NiV with N-PEG-ScNiV FC2.
We next used the syncytia-based immunofluorescence infection assay to examine all of the peptides over a range of concentrations in two different cell lines. We further preformed a quantitative analysis of syncytial areas based on immunofluorescence detection of viral antigen for HeV and NiV (see Materials and Methods) and revealed a grading of syncytial area inversely proportional to peptide concentration. Shown in Fig. 5 is the quantitative analysis of the syncytial area observed in HeV and NiV infection of both Vero (Fig. 5A and 5B) and PCI 13 (Fig. 5C and 5D) cell cultures over a range of concentrations of the capped-NiV FC2 peptide. In all cases significant inhibition of HeV and NiV infection and spread is observed in comparison to the scrambled capped control peptide (capped-ScNiV FC2). Similarly, shown in Fig. 6, both the N-PEG and C-PEG NiV FC2 peptides possessed potent inhibitory activity on HeV and NiV infection in Vero (Fig. 6A and 6B) and PCI 13 (Fig. 6C and 6D) cell cultures. Again, the scrambled C-PEG control peptide (C-PEG-ScNiV FC2) had no effect at any concentration tested. As was observed in the cell-cell fusion assays, in all cases, the C-PEG-NiV FC2 peptide exhibited weaker inhibitory activity in blocking virus infection, spread and syncytial size in comparison to the N-PEG-NiV FC2. The N-PEG-NiV FC2 peptide had considerable potency against both NiV and HeV and the calculated IC50 values for inhibiting either virus on both cell lines ranged from 0.46 nM to 2.05 nM (Table 1).
Figure 5 Inhibition of Hendra virus and Nipah virus infection by capped heptad peptides. Vero cells or PCI 13 cells were plated into 96 well plates and grown to 90% confluence. Cells were pre-treated with the indicated peptide for 30 min at 37°C prior to infection with 1.5 × 103 TCID50/ml and 7.5 × 102 TCID50/ml of live HeV or NiV (combined with peptide). Cells were incubated for 24 hours, fixed in methanol and immunofluorescently labeled for P protein prior to digital microscopy and image analysis to determine the relative area of each syncytium (see Methods). The figure shows the relative syncytial area (pixel2) versus the indicated peptide concentration for HeV and NiV.
Figure 6 Inhibition of Hendra virus and Nipah virus infection by N-terminal and C-terminal pegylated heptad peptides. Vero cells or PCI 13 cells were plated into 96 well plates and grown to 90% confluence. Cells were pre-treated with the indicated peptide for 30 min at 37°C prior to infection with 1.5 × 103 TCID50/ml and 7.5 × 102 TCID50/ml of live HeV or NiV (combined with peptide). Cells were incubated for 24 hours, fixed in methanol and immunofluorescently labeled for P protein prior to digital microscopy and image analysis to determine the relative area of each syncytium (see Methods). The figure shows the relative syncytial area (pixel2) versus the indicated peptide concentration for HeV and NiV.
Discussion
Both NiV and HeV continue to re-emerge, and in early 2004 two NiV outbreaks in Bangladesh have been confirmed totalling some 53 human cases of infection, and HeV has reappeared in Northern Australia in late 2004 with two cases of fatal infection in horses and one non-fatal human case [50]. The most recent NiV occurrence has again appeared in Bangladesh in January of 2005 [51]. Several important observations in these most recent outbreaks of NiV have been made, including a higher incidence of acute respiratory distress syndrome, person-to-person transmission occurring in the majority of cases, and significantly higher case fatality rates (60–75%), and no direct link to infected livestock or domestic animals [8-12,51]. In particular, the availability of NiV in the environment and the ability to grow the virus to high titer in the laboratory, it is also now considered a potential biological terror agent. Taken together these observations highlight the need to explore therapeutic strategies for henipaviruses. While there is some evidence that ribavirin therapy may be of clinical benefit [52], there are currently no other specific treatment options and only supportive care is indicated.
Paramyxoviruses, like retroviruses, possess a class I membrane fusion mechanism, and there have been major recent advances in the understanding of the structural requirements and mechanisms involved in the fusion process mediated by these viruses (reviewed in [19,53-55]). The present model of class I membrane fusion describes the formation of a trimer-of-hairpins structure whose oligomeric coiled-coil formation is mediated by the 2 α-helical heptad repeat domains of the fusion glycoprotein which drives membrane fusion. Peptides corresponding to either of these heptad domains block fusion by interfering with the formation of the trimer-of-hairpins structure, first noted with sequences derived from the gp41 subunit of the HIV-1 envelope glycoprotein [56,57]. HIV-1 heptad-peptides have now met with clinical success and are the first approved fusion inhibitor therapeutics for a viral infection. Peptide sequences from either the N or C heptads of the F glycoprotein from a variety of paramyxoviruses have also been shown to inhibit fusion [28-33,58]. Previously, we demonstrated that fusion-inhibiting peptides corresponding to the C-terminal heptad repeat domain of the F glycoprotein of either HeV or NiV could potently inhibit the membrane fusion activity of either virus [25,34]. Because the peptides derived from the HR-2 of NiV F could inhibit both HeV and NiV-mediated fusion, in this study we only pursued peptides derived from NiV F. Furthermore, we have refined our initial peptide fusion inhibitors by reducing their length and chemically modifying their amino and carboxyl termini either by amidation or acetylation or through the addition of a PEG10 moiety, and have examined these new peptides in both membrane fusion and virus infection assays.
We have demonstrated that Henipavirus-mediated fusion and infection can be potently inhibited by these chemically modified peptides in vitro in a dose-dependent fashion. Overall, the IC50 concentrations of the peptides in the present study were similar to our previous observations on un-capped 42-mer peptides against HeV and NiV-mediated cell-cell fusion as well as to those observed in other paramyxovirus and retrovirus systems. However, we found that the N-terminal pegylated NiV FC2 peptide used here to be particularly potent with overall IC50 values of <10 nM for both HeV and NiV cell-cell fusion and virus infection. The present results indicate that both the capped and pegylated peptides are equally as effective as the unmodified first generation fusion-inhibiting peptides. Interestingly, peptides with the PEG10 moiety linked to the C-terminus were slightly, yet reproducibly, less effective than N-terminal pegylated peptides. We speculate that this could reflect some process of steric hindrance effect by the PEG10 moiety in interacting with the F glycoprotein during its conformational alteration leading to 6-helix bundle formation.
These same chemical modifications also improved the solubility characteristics of the heptad-derived peptides, and also significantly increased the yield during synthesis and purification (data not shown). The primary objectives of the present study were to demonstrate that these peptides possessed potent inhibitory activity in surrogate viral glycoprotein-mediated membrane fusion assays as well as in live virus infection assays, and improve peptide solubility and synthesis yields. The specific chemical modifications were chosen, especially pegylation, to help improve the plasma half-life and thus enhance therapeutic success. Covalent coupling of PEG to proteins or "pegylation" is currently considered one of the most successful techniques to prolong the residence time of protein drugs in the bloodstream [47,59-61]. PEG is a water soluble polymer that when covalently linked to molecules, conveys its physico-chemical properties and therefore modifies the biodistribution and solubility of peptide and non-peptide drugs. Additionally, pegylation masks the peptide's surface and increases the molecular size of the polypeptide, thus reducing its renal ultrafiltration. PEG modification can also prevent the approach of antibodies or antigen processing cells and reduce their degradation by proteolytic enzymes [62].
Conclusion
The isolation of four new members of the family Paramyxoviridae in the past 10 years in addition to several largely uncharacterised paramyxoviruses recovered from historical rodent and snake sampling may indicate a much larger than previously thought, reservoir of paramyxoviruses. Given the success of heptad-derived peptide inhibition of paramyxovirus fusion, many of these new and yet to be discovered viruses may well be inherently treatable, such that the possibility of antiviral therapy may be available as soon as a sequence has been obtained for the respective fusion envelope glycoprotein. We are presently evaluating the properties of these anti-HeV and NiV fusion inhibiting peptides as well as their potential therapeutic value with an in vivo model of virus infection.
Methods
Cells and Culture conditions
HeLa cells (ATCC CCL 2) and African green monkey (Vero) cells (ATCC CCL 81) were obtained from the American Type Culture Collection. A HeLa cell line derivative (HeLa-USU) which does not express the NiV and HeV receptor, ephrin-B2 [13] was provided by Anthony Maurelli, USUHS, Bethesda, MD. The human glioblastoma cell line U373-MG was provided by Adam P. Geballe, Fred Hutchinson Cancer Research Center, Seattle, WA [63]. The human head and neck carcinoma PCI 13 cell line was the kind gift of Ernest Smith, Vaccinex, Inc. HeLa and U373 cell monolayers were maintained in Dulbecco's modified Eagle's medium supplemented with 10% cosmic calf serum (CCS) (Hyclone, Logan, UT) and 2 mM L-glutamine (DMEM-10). PCI 13 cell monolayers were maintained in DMEM-10 supplemented with 1 mM HEPES. Vero cells were maintained in the absence of antibiotics in Minimal Essential Medium containing Earle's salts and 10% fetal calf serum (EMEM-10). All cell cultures were maintained at 37°C under a humidified 5% CO2 atmosphere.
Viruses
For expression of recombinant HeV and NiV F and G glycoproteins, the following recombinant vaccinia viruses were employed: vKB7 (NiV F), vKB6 (NiV G), vKB1 (HeV F), vKB2 (HeV G) [25,34,64]. Bacteriophage T7 RNA polymerase was produced by infection with vTF7-3 which contains the T7 RNA polymerase gene linked to a vaccinia virus promoter [65]. The E. coli lacZ gene linked to the T7 promoter was introduced into cells by infection with vaccinia virus recombinant vCB21R-LacZ, which was described previously [66]. HeV stock virus (titer 1 × 108 TCID50/ml) was prepared as described [67]. NiV stock virus (titer 3 × 107 TCID50/ml) was prepared as described [68].
Cell-fusion assays
Fusion between envelope glycoprotein-expressing and target cells was measured by a reporter gene assay in which the cytoplasm of one cell population contained vaccinia virus-encoded T7 RNA polymerase and the cytoplasm of the other contained the E. coli lacZ gene linked to the T7 promoter; β-galactosidase (β-Gal) is synthesized only in fused cells [35,69]. Vaccinia virus-encoded proteins were produced by infecting cells (moi = 10) and incubating infected cells at 31°C overnight. Cell-fusion reactions were conducted with the various cell mixtures in 96-well plates at 37°C. Typically, the ratio of envelope glycoprotein-expressing cells to target cells was 1:1 (2 × 105 total cells per well, 0.2-ml total volume). Cytosine arabinoside (40 μg/ml) was added to the fusion reaction mixture to reduce non-specific β-Gal production [35]. For quantitative analyses, Nonidet P-40 was added (0.5% final) at 2.5 h and aliquots of the lysates were assayed for β-Gal at ambient temperature with the substrate chlorophenol red-D-galactopyranoside (CPRG; Roche Diagnostics Corp.). For inhibition by peptides, serial dilutions of peptides were performed and added to effector cell populations prior to the addition of target cell populations. All assays were performed in duplicate and fusion results were calculated and expressed as rates of β-Gal activity (change in optical density at 570 nm per minute × 1,000).
Virus infection assay and immunofluorescence
Vero cells were seeded into 96 well plates at 6 × 104 cells/300 μl and grown to 90% confluence in EMEM-10 at 37°C under a humidified 5% CO2 atmosphere. Peptides were diluted 4-fold in EMEM. Under biohazard level 4 conditions, media were discarded and 100 μl of diluted virus was added to each well and incubated at 37°C for 30 min. Virus dilutions were chosen to generate 50 plaques under these adsorption conditions. Virus inoculum was removed and 200 μl of diluted peptide was added to each well and incubated at 37°C for 18 h. The culture medium was discarded and plates were immersed in ice-cold absolute methanol for at least 20 min prior to air-drying outside the biohazard level 4 facility. Fixed plates were immunolabeled with anti-P monospecific antisera [70]. Briefly, slides were washed in 0.01 M phosphate-buffered saline (PBS), pH 7.2 containing 1% BSA for 5 min. 40 μl of anti-P antiserum (1:200 in PBS-BSA) was applied to each well and incubated at 37°C for 30 min. Slides were rinsed with PBS containing 0.05% Tween 20 (PBS-T) and washed for 5 min in PBS-BSA. 40 μl of FITC labeled goat anti-rabbit antiserum (ICN Pharmaceuticals, Costa Mesa, USA) diluted 1:100 in PBS-BSA was then applied to each well and incubated at 37°C for 30 min. Slides were rinsed again with PBS containing 0.05% Tween 20 (PBS-T) and washed for 5 min in PBS-BSA. Wells were overlaid with glycerol/PBS (1:1) containing DABCO (25 ug/ml) and stored in the dark prior to imaging.
FITC immunofluorescence was visualized using an Olympus IX71 inverted microscope (Olympus Australia, Mt. Waverley, Australia) coupled to an Olympus DP70 high resolution color camera. Image analysis was performed using AnalySIS® image analysis software (Soft Imaging System GmbH, Munster, Germany). Briefly, individual virus syncytia were detected by threshold analysis followed by "hole filling" and subsequently measured to determine the area of each syncytium. To ensure repeatability between images, all procedures were performed as a macro function with fixed parameters. Nine images were analysed for each peptide concentration resulting in the collation of syncytial area data for between 9–36 foci per peptide concentration (average ~15). Measurements were collated and non-linear regression analysis performed using GraphPad Prism software (GraphPad Software, San Diego, CA USA) to determine the IC50.
Peptide synthesis
The following peptide sequence, corresponding to the C-terminal α-helical heptad repeat domain (HR-2) of the NiV F glycoprotein, was chosen for synthesis: KVDISSQISSMNQSLQQSKDYIKEAQRLLDTVNPSL (NiV FC2). A scrambled version of the 36-amino-acid peptide was also synthesized for use as a negative control KQSSMISLQSQKSINSLPSQIRDYVQKTVLLAEDND (ScNiV FC2).
All peptides were synthesized utilizing the Fmoc/tBu protection scheme. The peptides with PEG(10) on the N-terminus were synthesized on a PS3 automated synthesizer (Protein Technologies Inc., Tucson, AZ) using NovaSYN® TGR Resin (Nova Biochem, EMD Biosciences, Inc. La Jolla, CA). The peptides with PEG(10) on the C-terminus were synthesized on an ABI433 automated synthesizer (Applied Biosystems, Foster City, CA) using 2-Chlorotrityl resin (Nova Biochem). The protected amino acids were incorporated into the peptide via active ester formation using 2-(6-Chloro-1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HCTU) (Nova Biochem). All Fmoc protected amino acids were supplied by Nova Biochem. The protecting groups used were as follows: sidechains of Asn, Cys, His, and Gln were protected with Trityl (trt), Glu and Ser were protected with tert-Butyl (tBu), Lys was protected with tert-Butyloxycarbonyl (Boc),Arg was protected with 2,2,4,6,7-pentamethyldihydrobenzopfuran-5-sulfonyl (Pbf). Ala, Leu, Phe, Val, and Gly were used without sidechain protection. PEG(10) was incorporated using O-(N-Fmoc-2-aminoethyl)-O'-(2-carboxyethyl)-undecaethyleneglycol (Nova Biochem) and the peptides were acetylated by treatment with acetic anhydride (Sigma-Aldrich). Peptides were cleaved from the solid support using 92% trifluoacetic acid (Halocarbon), 2% anisole, 2% ethanedithiol, 2% triisopropylsilane (all Sigma-Aldrich), and 2% water. Peptides were purified on a Waters 600e semi-prep HPLC system using a grace Vydac 300. Diphenyl column and solvents 0.1%TFA/water (A) and 0.1%TFA/acetonitrile (ACN) (B). Analytical HPLC analysis of all fractions was performed using a Waters Alliance 2695 with a 2.1 × 30 mm Symmetry Shield™ RP18 3.5 m column. Matrix assisted laser desorption/ionization time of flight (MALDI-ToF) mass spec analysis of the crude and pure peptides was performed using an ABI Voyager DE Pro system. Crude peptide from each synthesis and pure peptide was dissolved in 50%ACN/water and spotted with α-Cyano-4-Hydroxycinnimic Acid matrix (Sigma-Aldrich). Positive ions were detected using the linear detector, which is calibrated with Bradykinin and Angiotensin standards.
Proteomics computational methods
Methods to derive general models of surface glycoproteins have been described previously [43]. Homology modelling of Hendra virus and Nipah virus F was based on the structure of the F protein of Newcastle disease virus, another member of the Paramyxoviridae, determined by x-ray crystallography [40]. MacMolly (Soft Gene GmbH, Berlin) was used to locate areas of sequence similarity and to perform alignments. PHDsec (Columbia University Bioinformatics Center, /) was used for secondary structure prediction [41]. PHDsec predicts secondary structure from multiple sequence alignments by a system of neural networks, and is rated at an expected average accuracy of 72% for three states, helix, strand and loop. Domains with significant propensity to form transmembrane helices were identified with TMpred (ExPASy, Swiss Institute of Bioinformatics, ). TMpred is based on a statistical analysis of TMbase, a database of naturally occurring transmembrane glycoproteins [42].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
KNB conceived and contributed to the design and use of heptad derived peptides as fusion inhibitors henipaviruses, designed and carried out all cell-fusion assays, interpreted data, and edited and corrected the manuscript. BAM and GC developed and carried out all live virus infections and peptide inhibition assays, interpreted data and edited and corrected the manuscript. LFW provided financial support, corrected the manuscript and provided supervision of KNB. BTE provided expertise for conducting the live virus infection experiments, financial support, corrected the manuscript and provided supervision of BAM and GC. CCB conceived and contributed to design and use of heptad derived peptides as fusion inhibitors for henipaviruses and PEG-linked versions of peptides, provided overall supervision and financial support and prepared the final versions of the manuscript.
Acknowledgements
We wish to especially thank Robert F. Garry, Tulane University Health Sciences Center, and William R Gallaher, Louisiana State University Medical Center, for modeling the Nipah and Hendra virus F1 glycoprotein. We also wish to acknowledge the excellent technical assistance and service provided by John Phipps and Ron Donoho of Global Peptide Services, Inc. We also especially thank Thomas Matthews of Trimeris Inc., for advice and consultation and whose help will be particularly missed due to his untimely death. The views expressed in the manuscript are solely those of the authors, and they do not represent official views or opinions of the Department of Defence or the Uniformed Services University. This study was supported by NIH AI056423 grant to C.C.B.
==== Refs
Murray K Selleck P Hooper P Hyatt A Gould A Gleeson L Westbury H Hiley L Selvey L Rodwell B A morbillivirus that caused fatal disease in horses and humans Science 1995 268 94 97 7701348
Chua KB Goh KJ Wong KT Kamarulzaman A Tan PS Ksiazek TG Zaki SR Paul G Lam SK Tan CT Fatal encephalitis due to Nipah virus among pig-farmers in Malaysia [see comments] Lancet 1999 354 1257 1259 10520635 10.1016/S0140-6736(99)04299-3
Barclay AJ Paton DJ Hendra (equine morbillivirus) Vet J 2000 160 169 176 11061954 10.1053/tvjl.2000.0508
Wong KT Shieh WJ Kumar S Norain K Abdullah W Guarner J Goldsmith CS Chua KB Lam SK Tan CT Goh KJ Chong HT Jusoh R Rollin PE Ksiazek TG Zaki SR Nipah virus infection: pathology and pathogenesis of an emerging paramyxoviral zoonosis Am J Pathol 2002 161 2153 2167 12466131
Wang LF Yu M Hansson E Pritchard LI Shiell B Michalski WP Eaton BT The exceptionally large genome of Hendra virus: support for creation of a new genus within the family Paramyxoviridae J Virol 2000 74 9972 9979 11024125 10.1128/JVI.74.21.9972-9979.2000
Wang L Harcourt BH Yu M Tamin A Rota PA Bellini WJ Eaton BT Molecular biology of Hendra and Nipah viruses Microbes Infect 2001 3 279 287 11334745 10.1016/S1286-4579(01)01381-8
Field H Young P Yob JM Mills J Hall L Mackenzie J The natural history of Hendra and Nipah viruses Microbes Infect 2001 3 307 314 11334748 10.1016/S1286-4579(01)01384-3
Enserink M Emerging infectious diseases. Nipah virus (or a cousin) strikes again Science 2004 303 1121 14976284 10.1126/science.303.5661.1121b
Anonymous Nipah virus outbreak(s) in Bangladesh, January-April 2004 Wkly Epidemiol Rec 2004 79 168 171 15132054
Butler D Fatal fruit bat virus sparks epidemics in southern Asia Nature 2004 429 7 15129247 10.1038/429007b
Anonymous Vol. 2 No. 2 June 2004
Person-to-person transmission of Nipah virus during outbreak in
Faridpur District, 2004 Health and Science Bulletin (ICDDR,B) 2004 2 5 9
Hsu VP Hossain MJ Parashar UD Ali MM Ksiazek TG Kuzmin I Niezgoda M Rupprecht C Bresee J Breiman RF Nipah Virus Encephalitis Reemergence, Bangladesh Emerg Infect Dis 2004 10 2082 2087 15663842
Bonaparte MI Dimitrov AS Bossart KN Crameri G Mungall BA Bishop KA Choudhry V Dimitrov DS Wang LF Eaton BT Broder CC Ephrin-B2 Ligand is a Functional Receptor for Hendra Virus and Nipah Virus Proc Natl Acad Sci U S A 2005 In Press
Baker KA Dutch RE Lamb RA Jardetzky TS Structural basis for paramyxovirus-mediated membrane fusion Mol Cell 1999 3 309 319 10198633 10.1016/S1097-2765(00)80458-X
Lamb RA Kolakofsky D Knipe DM and Howley PM Paramyxoviridae: The viruses and their replication. Fields Virology 2001 4 Philadelphia, Lippincott Williams & Wilkins 1305 1340
Scheid A Choppin PW Two disulfide-linked polypeptide chains constitute the active F protein of paramyxoviruses Virology 1977 80 54 66 195398 10.1016/0042-6822(77)90380-4
Klenk HD Garten W Host cell proteases controlling virus pathogenicity Trends Microbiol 1994 2 39 43 8162439 10.1016/0966-842X(94)90123-6
Earp LJ Delos SE Park HE White JM The many mechanisms of viral membrane fusion proteins Curr Top Microbiol Immunol 2005 285 25 66 15609500
Weissenhorn W Dessen A Calder LJ Harrison SC Skehel JJ Wiley DC Structural basis for membrane fusion by enveloped viruses Mol Membr Biol 1999 16 3 9 10332732 10.1080/096876899294706
Chan DC Kim PS HIV entry and its inhibition Cell 1998 93 681 684 9630213 10.1016/S0092-8674(00)81430-0
Skehel JJ Wiley DC Coiled coils in both intracellular vesicle and viral membrane fusion Cell 1998 95 871 874 9875840 10.1016/S0092-8674(00)81710-9
Singh M Berger B Kim PS LearnCoil-VMF: computational evidence for coiled-coil-like motifs in many viral membrane-fusion proteins J Mol Biol 1999 290 1031 1041 10438601 10.1006/jmbi.1999.2796
Hughson FM Enveloped viruses: a common mode of membrane fusion? Curr Biol 1997 7 R565 9 9285698 10.1016/S0960-9822(06)00283-1
Xu Y Gao S Cole DK Zhu J Su N Wang H Gao GF Rao Z Basis for fusion inhibition by peptides: analysis of the heptad repeat regions of the fusion proteins from Nipah and Hendra viruses, newly emergent zoonotic paramyxoviruses Biochem Biophys Res Commun 2004 315 664 670 14975752 10.1016/j.bbrc.2004.01.115
Bossart KN Wang LF Eaton BT Broder CC Functional expression and membrane fusion tropism of the envelope glycoproteins of Hendra virus Virology 2001 290 121 135 11882997 10.1006/viro.2001.1158
Chambers P Pringle CR Easton AJ Heptad repeat sequences are located adjacent to hydrophobic regions in several types of virus fusion glycoproteins J Gen Virol 1990 71 3075 3080 2177097
Xu Y Lou Z Liu Y Cole DK Su N Qin L Li X Bai Z Rao Z Gao GF Crystallization and preliminary crystallographic analysis of the fusion core from two new zoonotic paramyxoviruses, Nipah virus and Hendra virus Acta Crystallogr D Biol Crystallogr 2004 60 1161 1164 15159588 10.1107/S0907444904009515
Lambert DM Barney S Lambert AL Guthrie K Medinas R Davis DE Bucy T Erickson J Merutka G Petteway SRJ Peptides from conserved regions of paramyxovirus fusion (F) proteins are potent inhibitors of viral fusion Proc Natl Acad Sci U S A 1996 93 2186 2191 8700906 10.1073/pnas.93.5.2186
Joshi SB Dutch RE Lamb RA A core trimer of the paramyxovirus fusion protein: parallels to influenza virus hemagglutinin and HIV-1 gp41 Virology 1998 248 20 34 9705252 10.1006/viro.1998.9242
Wild TF Buckland R Inhibition of measles virus infection and fusion with peptides corresponding to the leucine zipper region of the fusion protein J Gen Virol 1997 78 107 111 9010292
Young JK Li D Abramowitz MC Morrison TG Interaction of peptides with sequences from the Newcastle disease virus fusion protein heptad repeat regions J Virol 1999 73 5945 5956 10364347
Young JK Hicks RP Wright GE Morrison TG Analysis of a peptide inhibitor of paramyxovirus (NDV) fusion using biological assays, NMR, and molecular modeling Virology 1997 238 291 304 9400602 10.1006/viro.1997.8834
Rapaport D Ovadia M Shai Y A synthetic peptide corresponding to a conserved heptad repeat domain is a potent inhibitor of Sendai virus-cell fusion: an emerging similarity with functional domains of other viruses Embo J 1995 14 5524 5531 8521809
Bossart KN Wang LF Flora MN Chua KB Lam SK Eaton BT Broder CC Membrane fusion tropism and heterotypic functional activities of the nipah virus and hendra virus envelope glycoproteins J Virol 2002 76 11186 11198 12388678 10.1128/JVI.76.22.11186-11198.2002
Bossart KN Broder CC Viral glycoprotein-mediated cell fusion assays using vaccinia virus vectors Methods Mol Biol 2004 269 309 332 15114023
Weiss CD HIV-1 gp41: mediator of fusion and target for inhibition AIDS Rev 2003 5 214 221 15012000
Kilby JM Hopkins S Venetta TM DiMassimo B Cloud GA Lee JY Alldredge L Hunter E Lambert D Bolognesi D Matthews T Johnson MR Nowak MA Shaw GM Saag MS Potent suppression of HIV-1 replication in humans by T-20, a peptide inhibitor of gp41-mediated virus entry Nat Med 1998 4 1302 1307 9809555 10.1038/3293
Kilby JM Lalezari JP Eron JJ Carlson M Cohen C Arduino RC Goodgame JC Gallant JE Volberding P Murphy RL Valentine F Saag MS Nelson EL Sista PR Dusek A The safety, plasma pharmacokinetics, and antiviral activity of subcutaneous enfuvirtide (T-20), a peptide inhibitor of gp41-mediated virus fusion, in HIV-infected adults. AIDS Res Hum Retroviruses 2002 18 685 693 12167274 10.1089/088922202760072294
Rockstroh JK Mauss S Clinical perspective of fusion inhibitors for treatment of HIV J Antimicrob Chemother 2004 53 700 702 15044424 10.1093/jac/dkh161
Chen L Gorman JJ McKimm-Breschkin J Lawrence LJ Tulloch PA Smith BJ Colman PM Lawrence MC The structure of the fusion glycoprotein of Newcastle disease virus suggests a novel paradigm for the molecular mechanism of membrane fusion Structure (Camb) 2001 9 255 266 11286892 10.1016/S0969-2126(01)00581-0
Rost B Liu J The PredictProtein server Nucleic Acids Res 2003 31 3300 3304 12824312 10.1093/nar/gkg508
Hofmann K Stoffel W TMbase - A database of membrane spanning proteins segments Biol Chem Hoppe-Seyler 1993 374 166
Gallaher WR Ball JM Garry RF Griffin MC Montelaro RC A general model for the transmembrane proteins of HIV and other retroviruses [see comments] AIDS Res Hum Retroviruses 1989 5 431 440 2788443
Gallaher WR Ball JM Garry RF Martin-Amedee AM Montelaro RC A general model for the surface glycoproteins of HIV and other retroviruses AIDS Res Hum Retroviruses 1995 11 191 202 7742034
Earl PL Broder CC Doms RW Moss B Epitope map of human immunodeficiency virus type 1 gp41 derived from 47 monoclonal antibodies produced by immunization with oligomeric envelope protein J Virol 1997 71 2674 2684 9060620
Harcourt BH Tamin A Ksiazek TG Rollin PE Anderson LJ Bellini WJ Rota PA Molecular characterization of Nipah virus, a newly emergent paramyxovirus Virology 2000 271 334 349 10860887 10.1006/viro.2000.0340
Harris JM Martin NE Modi M Pegylation: a novel process for modifying pharmacokinetics Clin Pharmacokinet 2001 40 539 551 11510630
Francis GE Fisher D Delgado C Malik F Gardiner A Neale D PEGylation of cytokines and other therapeutic proteins and peptides: the importance of biological optimisation of coupling techniques Int J Hematol 1998 68 1 18 9713164 10.1016/S0925-5710(98)00039-5
Hyatt AD Zaki SR Goldsmith CS Wise TG Hengstberger SG Ultrastructure of Hendra virus and Nipah virus within cultured cells and host animals Microbes Infect 2001 3 297 306 11334747 10.1016/S1286-4579(01)01383-1
Anonymous Hendra Virus - Australia (Queensland) Pro-med 2004 , International Society for Infectious Diseases
Anonymous Emerging Infections update: November 2004 to January 2005 Communicable Disease Report Weekly (CDR Weekly) 2005 15
Chong HT Kamarulzaman A Tan CT Goh KJ Thayaparan T Kunjapan SR Chew NK Chua KB Lam SK Treatment of acute Nipah encephalitis with ribavirin Ann Neurol 2001 49 810 813 11409437 10.1002/ana.1062
Dimitrov DS Cell biology of virus entry Cell 2000 101 697 702 10892741 10.1016/S0092-8674(00)80882-X
Russell CJ Jardetzky TS Lamb RA Membrane fusion machines of paramyxoviruses: capture of intermediates of fusion Embo J 2001 20 4024 4034 11483506 10.1093/emboj/20.15.4024
Eckert DM Kim PS Mechanisms of Viral Membrane Fusion and Its Inhibition Annu Rev Biochem 2001 70 777 810 11395423 10.1146/annurev.biochem.70.1.777
Wild CT Shugars DC Greenwell TK McDanal CB Matthews TJ Peptides corresponding to a predictive alpha-helical domain of human immunodeficiency virus type 1 gp41 are potent inhibitors of virus infection Proc Natl Acad Sci U S A 1994 91 9770 9774 7937889
Jiang S Lin K Strick N Neurath AR HIV-1 inhibition by a peptide Nature 1993 365 113 8371754 10.1038/365113a0
Young JK Hicks RP Wright GE Morrison TG The role of leucine residues in the structure and function of a leucine zipper peptide inhibitor of paramyxovirus (NDV) fusion Virology 1998 243 21 31 9527912 10.1006/viro.1998.9044
Delgado C Pedley RB Herraez A Boden R Boden JA Keep PA Chester KA Fisher D Begent RH Francis GE Enhanced tumour specificity of an anti-carcinoembrionic antigen Fab' fragment by poly(ethylene glycol) (PEG) modification Br J Cancer 1996 73 175 182 8546903
Delgado C Malik F Selisko B Fisher D Francis GE Quantitative analysis of polyethylene glycol (PEG) in PEG-modified proteins/cytokines by aqueous two-phase systems J Biochem Biophys Methods 1994 29 237 250 7699201 10.1016/0165-022X(94)90035-3
Caliceti P Veronese FM Pharmacokinetic and biodistribution properties of poly(ethylene glycol)-protein conjugates Adv Drug Deliv Rev 2003 55 1261 1277 14499706 10.1016/S0169-409X(03)00108-X
Veronese FM Peptide and protein PEGylation: a review of problems and solutions Biomaterials 2001 22 405 417 11214751 10.1016/S0142-9612(00)00193-9
Harrington RD Geballe AP Cofactor requirement for human immunodeficiency virus type 1 entry into a CD4-expressing human cell line J Virol 1993 67 5939 5947 7690415
Bossart KN Crameri G Dimitrov AS Mungall BA Feng YR Patch JR Choudhary A Wang LF Eaton BT Broder CC Receptor binding, fusion inhibition, and induction of cross-reactive neutralizing antibodies by a soluble g glycoprotein of hendra virus J Virol 2005 79 6690 6702 15890907 10.1128/JVI.79.11.6690-6702.2005
Fuerst TR Niles EG Studier FW Moss B Eukaryotic transient-expression system based on recombinant vaccinia virus that synthesizes bacteriophage T7 RNA polymerase Proc Natl Acad Sci USA 1986 83 8122 8126 3095828
Feng Y Broder CC Kennedy PE Berger EA HIV-1 entry cofactor: functional cDNA cloning of a seven-transmembrane, G protein-coupled receptor Science 1996 272 872 877 8629022
Wang LF Michalski WP Yu M Pritchard LI Crameri G Shiell B Eaton BT A novel P/V/C gene in a new member of the Paramyxoviridae family, which causes lethal infection in humans, horses, and other animals J Virol 1998 72 1482 1490 9445051
Crameri G Wang LF Morrissy C White J Eaton BT A rapid immune plaque assay for the detection of Hendra and Nipah viruses and anti-virus antibodies J Virol Methods 2002 99 41 51 11684302 10.1016/S0166-0934(01)00377-9
Nussbaum O Broder CC Berger EA Fusogenic mechanisms of enveloped-virus glycoproteins analyzed by a novel recombinant vaccinia virus-based assay quantitating cell fusion- dependent reporter gene activation J Virol 1994 68 5411 5422 8057423
Michalski WP Crameri G Wang L Shiell BJ Eaton B The cleavage activation and sites of glycosylation in the fusion protein of Hendra virus Virus Res 2000 69 83 93 11018278 10.1016/S0168-1702(00)00169-6
|
16026621
|
PMC1208959
|
CC BY
|
2021-01-04 16:39:00
|
no
|
Virol J. 2005 Jul 18; 2:57
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-57
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-641610517910.1186/1743-422X-2-64CommentaryInternational Committee on Taxonomy of Viruses and the 3,142 unassigned species Fauquet CM [email protected] D [email protected] International Laboratory for Tropical Agricultural Biotechnology, Danforth Plant Science Center, 975 N. Warson Rd., St. Louis, MO 63132 USA2 Institut de Recherche pour le Developpement, BP 64501, 34394 Montpellier cedex 5, France2005 16 8 2005 2 64 64 8 7 2005 16 8 2005 Copyright © 2005 Fauquet and Fargette; licensee BioMed Central Ltd.2005Fauquet and Fargette; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In 2005, ICTV (International Committee on Taxonomy of Viruses), the official body of the Virology Division of the International Union of Microbiological Societies responsible for naming and classifying viruses, will publish its latest report, the state of the art in virus nomenclature and taxonomy. The book lists more than 6,000 viruses classified in 1,950 species and in more than 391 different higher taxa. However, GenBank contains a staggering additional 3,142 "species" unaccounted for by the ICTV report. This paper reviews the reasons for such a situation and suggests what might be done in the near future to remedy this problem, particularly in light of the potential for a ten-fold increase in virus sequencing in the coming years that would generate many unclassified viruses. A number of changes could be made both at ICTV and GenBank to better handle virus taxonomy and classification in the future.
==== Body
Introduction
The VIIIth ICTV Report [1] lists more than 6,000 viruses classified in 1,950 species, themselves classified in 391 taxa. Thus, it was surprising to discover that GenBank (part of the International Nucleotide Sequence Database Collaboration, comprising the DNA DataBank of Japan, the European Molecular Biology Laboratory, and GenBank, which is located at the National Center for Biotechnology Information (NCBI) in the U.S. National Institutes of Health) has collected sequences belonging to 3,142 "species" of viruses not present in the ICTV current master list of 2005.
For the past 20 years, GenBank has increased the number of sequences stored from one in 1982 to 42 million in 2002, and has increased the number of tools at the disposal of users to the point where today it is an absolutely necessary system for molecular biologists and virologists world-wide. In particular, GenBank has developed a complete database for taxonomy, including virus taxonomy, allowing virologists to search all viruses and to identify new sequences and new virus names. Were it not for the existence of this excellent system, we would not have known of the existence of these 3,142 unassigned "species" of viruses.
ICTV also has been very active in the last 20 years, incrementally increasing the number of taxa and virus names from 369 in 1985 to 7,881 in 2004. Not only have the numbers increased exponentially (about 20-fold), but the complexity of virus nomenclature and taxonomy has become tremendously complicated and controversial, even puzzling for some. However, the overall stability of this virus classification, established in 1962 [2], is quite remarkable in that, for example, names of all genera and families established in the 1980s are still in use in 2005. The advancement with the most impact was the definition of a virus species [3], which still is not fully understood by most virologists. Therefore, it is not surprising that its application in terms of names and concept causes problems such as the one discussed in this paper.
The listing by GenBank, but not by ICTV, of 3,142 unassigned "species" clearly demonstrates a general problem of the application of the definition of virus species and that ICTV and GenBank must work in concert to cope with the day-to-day reality of virology and virus classification, and that collectively we need to improve the system in such a way that we will rapidly classify so-called "species" of viruses and change the current system, so that we do not find ourselves in a similar situation in the future.
This paper will review the probable causes of the present situation and will identify actions that could remedy the situation both at GenBank and at ICTV. Finally a brief description of the Taxonomic Proposals Management System (TPMS), a new database to electronically handle all taxonomic proposals for ICTV, will be provided.
Discussion
What are the causes of such a situation?
Review of the species concept
A virus species, as with all other taxonomic levels, is a concept devised by humans to describe concrete (real) biological viral entities. The definition of a virus "species", established [4] and accepted by ICTV [3,5], is as follows: "A virus species is a polythetic class of viruses that constitutes a replicating lineage and occupies a particular ecological niche". This definition is broad enough to allow considerable flexibility. It is the responsibility of the virologists interested in each virus family to apply this definition to their favorite viruses and to decide how many and which species should comprise a particular genus of a particular family [6]. Typically, this is the work of ICTV study-groups, whose members are experts in particular virus families [7]. Virus sequences are physical realities that are not species but entities assigned to a particular species [8]. For example, the 24 bluetongue viruses are assigned to the species Bluetongue virus in the genus Orbivirus, family Reoviridae; thus, 24 viruses assigned to a single species. Therefore sequences entered in GenBank must be associated with the name of a virus isolate. The term virus isolate has no specific definition and can be applied to any virus, so long as that virus has existed for some time. Several terms are used by virologists to define a virus entity below the species level and are somewhat variable and confusing, but for virus sequences the term "virus isolate" seems the most appropriate, practical and useful [9].
Sequence and information collection
The intent of GenBank is to collect sequences but the author, the person who enters the sequences in the database, is responsible for the accuracy of the sequences as well as for the validity of the associated information about the sequences, which includes the taxonomic information about the virus in question. The primary taxonomic information pertains to an isolate of a previously recognized virus or of a newly recognized virus. Consequently, if the author errs in naming the virus or in its classification, the entry will be incorrect and might remain incorrect for a long period of time. GenBank has personnel qualified to review all entries and they effectively and immediately correct a large number of obvious mistakes. However, with regard to "new species", GenBank personnel do not have the means to determine whether this new taxon is justified and acceptable or not. Obviously, this policy of GenBank cannot be changed from without, but it is certainly possible for GenBank to slightly modify the Entrez form in order to build in safety nets to avoid such situations in the future.
GenBank Entrez form and software
The Entrez form of GenBank is a general form for all organisms, and one of the first questions asks the name of the organism. Whereas viruses are not organisms, this question immediately causes ambiguity because it is not clear whether the isolate name or the name of the species to which it belongs should be indicated, notwithstanding the fact that many investigators have not yet understood the difference between an isolate name, corresponding to a specific sequence of a virus, and the conceptual species name, to which a particular isolate has been assigned [10,8] (see above). The second drawback comes from the fact that the software dealing with Entrez and connected to the Taxonomy database of GenBank will by default assign any new virus name in the field "Organism" to a new "species" if it does not exist in the ICTV Master list, from which the current valid classification is derived. This implies that most of the 3,142 "species" are in fact virus isolates only (concrete entities), rather than species (conceptual). The third problem arises from the fact that many virus species names and virus names are identical in the words that compose them, except that species names are written in italics and are not abbreviated, while virus names are not italicized and may be abbreviated [10-13]; unfortunately, databases can only use ASCII characters and cannot accept italics. It is particularly unfortunate that we are in this situation and that different names have not been created for virus species, but this is a reality that cannot be changed, at least in the short term, and consequently we need to find a way to finesse these problems and still adhere to acceptable virus nomenclature. It seems that the only possibility is to redesign software to show taxonomy as it should be shown, and to restore the difference between the two names, if not by italics at least by some equivalent means. The fourth problem also is associated with the ASCII code and that is the impossibility of using Greek letters, which are incorporated in many phage names. For this, one solution might be to have ICTV change these names slightly by spelling out the Greek letters, c.f., "Enterobacteria phage lambda", instead of "Enterobacteria phage λ".
Exponential increase in sequencing
Because of lower costs and more efficient technology, sequencing is increasing exponentially. Known viral sequences already are in the hundreds of thousands and it is conceivable that millions more will become available in the coming years. The number of virus taxa and isolates also increases exponentially and is therefore linearly related to sequencing activities, as shown in Figure 1. It is therefore critical to have all data stored and named properly and with taxonomic updates in real time.
Figure 1 Linear correlation between the number of virus sequences recorded at GenBank between 1996 and 2005 and the number of virus taxa and virus isolates.
Taxonomic knowledge and consciousness of virologists
Not all virologists determining sequences are fully aware of or interested in virus taxonomy. These people are the primary source of information and they consequently are the primary sources of data entry errors. Left adrift, they will not make efforts to solve these particular sorts of problems, so other solutions should be put in place to remedy the problem. Insisting that they follow taxonomic rules seems incompatible with the general "author" principle of GenBank, but it is possible to "invite" them to enter the proper information directly on the Entrez form.
ICTV organization
The organization structure of ICTV, the generally accepted leader in viral taxonomy, can be conceived of as pyramidal [7]. The base of the pyramid comprises more than 500 virologists in six Subcommittees, divided into 82 specialty study-groups (SG), each representing a family of viruses or an unassigned genus (ICTVnet: ) (Fig. 2).
Each SG comprises from three to 15 experts selected for their expertise in recognized and new viruses and new sequences of viruses in each family; obviously this is an increasingly difficult task, given the rapid increase in available data. These 82 SGs present taxonomic proposals from the level of species to the level of order, the proposals are reviewed by the Executive Committee (EC) of the ICTV and, 18 months later, through a posting on ICTVnet and two rounds of discussions, are then ratified (or not) and voted upon (or not) by the ICTV membership at large. This system, although well organized and in full compliance with the international code for virus taxonomy and nomenclature [14], is neither adapted to a large number of new isolates and species nor to a quick acceptance, so that changes are envisioned. In addition, by statute, the ICTV does not have the mandate to deal with virus isolates, which are not taxa. This is a fundamental problem because no organization is responsible for certifying isolate names or for determining to which virus species they belong. However, ICTV, for historical reasons, lists a large number of unclassified virus isolates and SG members are the most knowledgeable people to deal with such an issue. Obviously, this should be considered by ICTV in order to solve this issue of unassigned viruses erroneously purported to be "species" and found in various databases.
Figure 2 Diagrammatic organization of the ICTV.
What are the solutions?
It is not the responsibility of the authors to decide whether their new virus isolates belong to an existing or a new species, not the responsibility of GenBank to police authors and organize virus taxonomy and nomenclature, and not the responsibility of ICTV to deal with non-taxonomic issues, such as isolate names, but it is our communal responsibility to work together to make the system functional; and it certainly is feasible. Here we call for collaboration of all parties to establish a generally satisfactory system for virus taxonomy and nomenclature of virus isolates, sequence data, and all other virological data. Obtaining solutions will necessitate the collaboration of all those involved in producing, collecting and classifying viruses.
Classifying the 3,142 unassigned virus names in GenBank
Placing the 3,142 unassigned "species" (which likely are simply "viruses") registered at GenBank will require manual classification by the ICTV SG members. Most of these "species" are probably isolates of recognized species and are in the GenBank database by default as new "species" because, whether they have been named or not, they have not been entered in the GenBank forms. Some of these GenBank "species", however, may actually be new species that have not yet been submitted to ICTV and, if so, they should be proposed through the normal channels of the Taxonomic Proposals (TaxoProps) for evaluation and official ratification before they can be recognized as new species.
Improving the management of taxonomic proposals
As a remedy to the general problem of virus classification, we suggest a two-tier plan. Firstly, GenBank would work with ICTV to provide to the authors rolling menus, or automatic writing, offering a list of all the official ICTV names to select from. This would help avoid spelling errors or the creation of incorrect new names by authors. In the case of an unaccepted (or provisional) virus name, the default should be "unassigned virus in the genus, or in the family, or in the order or in the virus kingdom". This assignation to an "unassigned category" would trigger an email to the relevant ICTV SG chair and to some EC members so that ICTV will be immediately aware of this new entry. This information also would be stored in a proposed new database called TaxoProp Management System (TPMS), with a pending status to be resolved promptly (see below). The second tier of the remedy would be to establish the TPMS database to handle all new TaxoProps containing information originating from an SG member, from any individual virologist or directly from GenBank.
Changing ICTV to work in real time
Resolving the mechanics of the system is only a partial solution and more details are needed in order for this to be functional in real time. ICTV has established a sophisticated and democratic system to receive, review, discuss, re-discuss and finally approve taxonomic proposals, whether they concern a new species or a new order [15]. This system is very sophisticated in principle but is too complex and takes too long to be compatible with modern virology work. There are, however, fundamental and practical differences between a TaxoProp for a new species and a TaxoProp for everything else, and that is that species are solely determined by specialty groups (ICTV SGs) and rarely by the public or the ICTV EC members. The only envisioned ICTV concern would be to make certain that the virus species name follows the ICTV International Rules [14]. It would therefore be possible to computerize the verification of the names in order to "bypass" ICTV EC scrutiny and only then obtain approval of the ICTV membership by electronic voting. This in turn would considerably reduce the turn-around time: essentially, as soon as a name would be proposed by an ICTV SG, it would be automatically verified and accepted if all the rules were followed.
Increasing relationships between ICTV and GenBank
In order to have taxonomic consistency, it is absolutely essential that we strengthen the relationships between ICTV and GenBank. This could be done with an electronic link between TPMS and GenBank, supplemented with improved and regular personal relations. For example, GenBank has appointed virus experts for a number of families of viruses. It would be more efficient to have GenBank somehow linked directly to the SG chairs of ICTV, so that GenBank would get taxonomic experts for advice in virus taxonomy. This, in turn, would provide a more active role for the SG members and would activate the overall system of virus taxonomy.
The TPMS database?
Due to the rapid increase in the number of TaxoProps and the increasing difficulty in manually handling several thousand taxonomic names, it is planned by ICTV to create a database to handle all ICTV taxa, from inception to acceptance, and from species to order (Fig. 3).
Figure 3 Diagrammatic organization of the TaxoProp Management System (TPMS) database.
Overall description of the TPMS
The goal of the TPMS database is to electronically handle all the virus taxa at large (including virus isolates), i.e., those that have been inherited from previous reports, those that have been approved between 1999 and 2005, those in the process of being approved and those that will be approved in the future.
At any time it will then be possible to:
1- certify the approval status of any TaxoProp
2- elaborate a complete list of approved taxa
3- establish an electronic link with GenBank and all other virus databases
Structure of the TPMS database
The database will be accessible on line to all ICTV members and to the public at large. The database will comprise a collection of individual files, each one corresponding to a single taxon (order, family, subfamily, genus, and species), and in addition will record isolates. Each TaxoProp will progress from the initial proposal to the approval of the taxon by the ICTV. The bulk of the database will be constituted by all the approved taxa and individual viruses and should be exportable to other databases for future use.
Approval process of the TaxoProps
The TaxoProp approval system of ICTV is depicted in Figure 4. When a TaxoProp form is filed on line, it is then considered by ICTV and becomes "Considered". A TaxoProp will consist of the proposal itself, of arguments provided to support the TaxoProp, and by attached documents to be downloaded by the reader. A unique TaxoProp number will be automatically generated when a form is completed or when a TaxoProp is modified by the author. After consideration, it becomes "Considered", and comments of the ICTV are added on line and available to the public at large. Comments can be added by any virologist and the author of the TaxoProp can respond on line. The change of level will be done through a voting system in which ICTV EC members can view the votes at any time and will be operated by the TPMS curator. When accepted by ICTV, the TaxoProp becomes "Accepted" and comments are no longer added. When the ICTV at large has voted to approve the TaxoProp, it becomes "Voted" and is de facto a ratified taxon of the ICTV.
Figure 4 Diagrammatic organization of the taxonomic proposals system at ICTV.
Filing TaxoProps
TaxoProps can be filed by any virologist, but it is likely that most will be filed by SG members. There will be a set of TaxoProps to choose from on line according to the sort of taxon proposed: an isolate (although not a formal taxon), a species, a type species, a genus, a subfamily, a family, an order, or any other taxon proposed by any virologist. Higher taxa cannot be proposed unless a lower level has been proposed and accepted. A genus must have a type species and at least one species or a list of species in the genus. Any proposed new taxa must have a name following the ICTV Code [14].
Link with species demarcation criteria lists
For each genus, it will be possible to list the species demarcation criteria of that genus for the benefit of the author and readers of the species TaxoProps. These lists will be extracted from the VIIIth Report and from subsequent updates. It is envisioned that these lists of species demarcation criteria will become useful tools for ICTV to search for a better means of making the overall virus taxonomy more uniform in the future.
Special case of TaxoProp isolates
Although isolates do not comprise formal taxa, recording and listing them are indispensable for the creation of species taxa and also for practical purposes. Virus species being defined from properties of virus isolates, ICTV records the existence of isolates in order to define species. There are no formal rules for naming isolates and, consequently, any name could potentially be accepted. However, it is common sense that these names follow previously established names and follow somewhat virus species names rules to avoid complete chaos [14,9].
Accepting TaxoProp species
TaxoProp species, after being proposed by the SGs, will automatically be reviewed to ensure that they follow the ICTV International Code and, if they do, will automatically be approved electronically. However, a vote by the ICTV membership at large will still be necessary for it to be considered as ratified by the ICTV; this also could be done electronically.
Approved taxa
All approved taxa, because they have been inherited, will be entered in the TPMS, so that the database will be updated automatically and in real time and can be linked to GenBank, ICTVdb (ICTV database) and other databases. Files for each taxon will archive all the comments and votes so that we can always return to this information if needed. However, public access to the TPMS for the ratified ICTV Master List will remain under control of the President of ICTV.
Generating the ratified list of taxa
Using TPMS, it will be possible to generate a complete list of approved taxa at any time, and this will be possible either alphabetically or taxonomically. Accepted taxa also will be presented according to the order of presentation of virus taxonomy: dsDNA, ssDNA, rtDNA, rtRNA, dsRNA, ssNRNA, ssPRNA, SAT (Satellites), VIR (Viroids), UN (unassigned). Within each category, the families will be classified according to the Order of Presentation of the Virus. Taxonomy Species will be assigned to a genus and classified alphabetically within the genus. The species taxa should comprise at least one isolate but include as many as described.
Searching the TPMS database
Any virologist will have access to the TPMS database on line anytime so that he or she can verify names, either with an alphabetical pull down menu system or by use of a search engine.
Linking the TPMS database to other databases
TPMS will de facto become the definitive reference for virus taxonomy and nomenclature. Once approved by the ICTV President, typically after each ICTV EC meeting or after a certain period of time (such as six months), the TPMS database will be electronically linked with GenBank, ICTVdb, and other associated databases to ensure complete homogeneity of the virus taxonomy and nomenclature. This tool will be particularly important for the members of the International Virus Database Network, who recognize the critical need for such a system, and to homogenize and update in real time the virus taxonomy and nomenclature.
Conclusion
Because sequencing is much less expensive and increasingly done these days, it is assumed that investigators will generate many more sequences of known and unknown viruses in the very near future. The number of sequences is increasing exponentially and so will the number of virus names and virus taxa, as they are directly correlated with sequencing activities. It is clear that past management of virus taxonomy will become inadequate commensurate with this increase and that more rapid and better-adapted management is required. The electronic connection between databases storing sequences, storing biological data or handling taxonomic proposals is the only solution for the future. Human intervention is indispensable, but should be restricted to the decision-making process; all other interventions might best be done electronically. This will mean better collaboration between the people responsible for the different types of databases, the people involved in the ICTV organization, and individual investigators themselves. Software will have to be adapted or devised to better fit virus taxonomy and nomenclature, ICTV will have to better appreciate day-to-day realities such as dealing with virus isolates, and scientists will have to be better educated in virus taxonomy. Regarding the decision-making process for virus species and isolates, a much greater role for the ICTV SGs will be necessary because the ultimate expertise resides with them in determining which new viruses fit into existing taxa and which must be placed in defined new taxa. Consequently, it would be best for ICTV to work more closely with GenBank to establish the electronic TPMS database to work in real time and to cope with an acutely burgeoning number of taxa.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
The authors contributed equally to the writing of this paper.
Acknowledgements
The opinions expressed in this paper reflect the personal opinions of the authors and should not be misconstrued as an official position of the ICTV or GenBank. A Taxonomic Proposal (TaxoProps) Management System (TPMS) database has been devised by the authors and will be officially proposed to ICTV. Its complete description will appear in Virology Division News, Archives of Virology.
We thank Dr. F. Nino de Medias-France for his contribution to the elaboration of structure of the TPMS database, and Dr. Andy Ball for his review of the manuscript and his useful suggestions. Dr. Charlie Calisher is warmly acknowledged for all the time he took to edit this paper and make comments well taken by the authors.
==== Refs
Fauquet CM, Mayo MA, Maniloff J, Desselberger U, Ball LA Virus Taxonomy, VIIIth Report of the ICTV 2005 London: Elsevier/Academic Press
Lwoff A Horne R Tournier P A system of viruses Cold Spring Harbor Symp Quant Biol 1962 7 51 55 13931895
van Regenmortel MHV Maniloff J Calisher C The concept of virus species Arch Virol 1991 120 313 314 1958131 10.1007/BF01310487
van Regenmortel MHV Virus species, a much overlooked but essential concept in virus classification Intervirology 1990 31 241 254 2272775
van Regenmortel MHV van Regenmortel MHV, Fauquet C., Bishop DHL, Carstens EB, Estes MK, Lemon SM, McGeogh DJ, Maniloff J, Mayo MA, Pringle CR, Wickner RB Introduction to the species concept in virus taxonomy Virus Taxonomy Seventh Report of the International Committee on Taxonomy of Viruses 2000 San Diego: Academic Press 3 16
van Regenmortel MHV Bishop DHL Fauquet CM Maniloff J Mayo MA Guidelines to the demarcation of virus species Arch Virol 1997 142 1505 1518 9267460
Mayo MA Fauquet CM The current composition of ICTV Arch Virol 2000 145 1497 1504 10963354 10.1007/s007050070107
van Regenmortel MHV Viruses are real, virus species are man-made, taxonomic constructions Arch Virol 2003 148 2483 2490
Fauquet CM Stanley J Revising the way we conceive and name viruses below the species level. A review of geminivirus taxonomy calls for new standardized isolate descriptors Arch Virol
Drebot MA Henchal E Hjelle B LeDuc JW Repik PM Roehrig JT Schmaljohn CS Shope RE Tesh RB Weaver SC Calisher CH Improved clarity of meaning from the use of both formal species names and common (vernacular) virus names in virological literature Arch Virol 2002 147 2465 2472 12491112 10.1007/s00705-002-0938-8
van Regenmortel MHV How to write the names of virus species Arch Virol 1999 144 1041 1042 10447501 10.1007/s007050050567
van Regenmortel MHV On the relative merits of italics, Latin and binomial nomenclature in virus taxonomy Arch Virol 2000 145 433 441 10752566 10.1007/s007050050036
Calisher CH Mahy BWJ Taxonomy: get it right or leave it alone (editorial) Am J Trop Med Hyg 2003 68 505 506 12812333
Mayo MA Horzinek M A revised version of the International Code of Virus Classification and Nomenclature Arch Virol 1998 143 1645 1654 10.1007/s007050050406
Mayo MA Fauquet CM Maniloff J Taxonomic proposals on the Web: new ICTV consultative procedures Arch Virol 2003 148 609 611 12607111 10.1007/s00705-003-0002-3
|
16105179
|
PMC1208960
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Aug 16; 2:64
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-64
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-671610917910.1186/1743-422X-2-67ResearchEvidence that spontaneous reactivation of herpes virus does not occur in mice Gebhardt Bryan M [email protected] William P [email protected] LSU Eye Center, Louisiana State University Health Sciences Center, New Orleans, LA 70112 USA2 Department of Veterinary Microbiology, Montana State University, Bozeman, MT 59718 USA2005 18 8 2005 2 67 67 17 6 2005 18 8 2005 Copyright © 2005 Gebhardt and Halford; licensee BioMed Central Ltd.2005Gebhardt and Halford; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Some species, including humans and rabbits, exhibit periodic viral reactivation and shed infectious virus at the infected end organ. Mice may be an exception, because spontaneous shedding of infectious virus rarely, if ever, occurs. However, spontaneous molecular reactivation, i.e., the expression of a few viral genes and the synthesis of the viral glycoproteins coded for by these genes, has been reported. This finding has prompted the assumption that molecular reactivation is an indicator of reactivation and the production of infectious virus. The goal of this study was to differentiate between viral gene expression during latency and the episodic production of infectious virus in mice.
Results
Viral reactivation and infection were not seen in herpes simplex virus type 1 (HSV-1) latent ganglion graft recipient BALB/c scid or immunocompetent BALB/c mice, which survived the 65-day observation period with no evidence of viral infection although the immunocompetent mice developed cellular and humoral immunity to HSV-1. In contrast, BALB/c scid recipients of ganglia containing reactivating virus invariably developed a local and, subsequently, systemic viral infection and died within 14 days. Immunocompetent BALB/c mice that received ganglion grafts containing reactivating virus survived the infection and became immune to the virus. Trigeminal ganglia removed from scid and immunocompetent recipient graft sites 5, 14, and 28 days after transplantation contained latent virus and viable neurons.
Conclusion
The results suggest that, within the limits of detection of the experiments, spontaneous episodic production of immunogenic viral antigens but not of infectious virus occurs in mouse neural ganglia during latency.
==== Body
Background
The infectious cycle of herpes simplex virus type 1 (HSV-1) in experimental animals is similar to that which occurs in humans, but there may be a significant difference as well. HSV-1 readily infects epithelial surfaces of most mammalian species, replicates in these cells, enters the nervous system, and achieves a latent state in neurons in the peripheral nervous system. A notable species difference is that the virus undergoes spontaneous, episodic reactivation with or without evidence of recurrent disease in humans and rabbits, whereas mice either do not undergo spontaneous reactivation or undergo spontaneous reactivation at such a low frequency that it is difficult to document [1].
Testing an end organ such as the eye or the site of viral latency, the sensory ganglia, for infectious virus during latency in mice fails to yield virus [2-5]. However, evidence of viral gene expression in the trigeminal ganglia of mice during latency has been reported [6,7]. In addition to the expression of the latency-associated transcript (LAT), the expression of other viral genes and their products has been found in a small number of ganglion cells. Feldman et al. [8] described "abundant" expression of viral genes and proteins and noted viral DNA synthesis in occasional neurons. This process was termed "spontaneous molecular reactivation"; no evidence of infectious virus was reported in this study [8].
Stevens and Cook [3] transplanted ganglia from latent mice into mice that were actively immunized with irradiated virus or passively immunized with anti-HSV antibody and concluded that antiviral antibody helped maintain viral latency. Tenser et al. [4] reported that viral reactivation occurred in ganglion transplants after ex vivo explantation. The occurrence of secondary latency was proposed as a consequence of viral reactivation and infection of "secondary" neurons in the grafts; however, infectious virus was not found in ganglion homogenates [4].
The current study was designed to differentiate between viral gene expression and the production of infectious virus in latent mouse ganglia in vivo. The experimental system was designed to assess for the production of small numbers of infectious viral particles which would lead to morbidity and, ultimately, mortality in the host mice. In the results reported here, molecular reactivation (i.e., expression of HSV-1 genes and production of glycoproteins during latency) did not proceed to the production of detectable infectious virus in immune-deficient mice. The results suggest that viral reactivation does not occur spontaneously and episodically in the mouse trigeminal ganglion in vivo.
Results
Absence of infectious virus in the trigeminal ganglion during latency
Infectious virus was present on the ocular surface and in both trigeminal ganglia of a group of five BALB/c mice sacrificed 5 days after topical ocular infection (Table 1). On days 10, 20, 30, 50, 70, and 100 after infection, both the ocular surface and the trigeminal ganglion homogenates of latently infected mice failed to yield infectious virus as evidenced by cytopathic effect on Vero cells (Table 1).
Table 1 Analysis of infectious virus in the eye and trigeminal ganglion during establishment of latency
Location Days after infectiona
5 10 20 30 50 70 100
Eye 5/5b 0/5 0/5 0/5 0/5 0/5 0/5
Trigeminal ganglia 10/10 0/10 0/10 0/10 0/10 0/10 0/10
aInfected mice (N = 5) were killed and eye swabs (both combined) and trigeminal ganglion homogenates (separately) tested for infectious virus.
bThe numbers indicate number of eye swabs and trigeminal ganglion homogenates containing infectious virus/number of each tested at each time.
Sensitivity of the ganglion assay
Assay of trigeminal ganglion homogenates for infectious virus immediately after microinjection of a known number of PFU of virus revealed that the limit of sensitivity for the in vitro assay was between 50 and 100 PFU per ganglion (Table 2). All ganglia injected with 100 PFU yielded plaques and 6 of 10 ganglia injected with 50 PFU yielded plaques (Table 2). Injection of smaller numbers of PFU did not reproducibly yield plaques (Table 2).
Table 2 Detection of virus injected into the trigeminal ganglion by plaque assaya
Number of PFU injected/ganglion Number of ganglia containing infectious virus/number of ganglia tested Mean PFU ± standard deviation
100 10/10 28 ± 7
50 6/10 7 ± 4
10 1/10 1
5 0/10 0
1 0/10 0
aIntact trigeminal ganglia were injected with the amounts of virus indicated and homogenized. The homogenate was tested for infectious virus by plaque assay.
Ten out of 10 of the BALB/c scid mice receiving ganglia injected with 100 PFU and 9 of 10 mice receiving ganglia injected with 50 PFU died from complications of viral pathogenesis within 12 days of receiving ganglion transplants (Table 3). Five of 10 BALB/c scid mice receiving ganglion grafts containing 10 PFU and 1 of 10 mice receiving grafts containing 5 PFU died within 14 days of receiving the grafts (Table 3). None of the 10 animals receiving the ganglion grafts containing 1 PFU gave evidence of viral infection and viral pathogenesis over a 35-day observation period.
Table 3 Detection of virus injected into the trigeminal ganglion by transplantation into scid micea
Number of PFU injected/ganglion Number of mice dead/number of mice grafted
100 10/10
50 9/10
10 5/10
5 1/10
1 0/10
aTrigeminal ganglia injected with the amount of virus indicated were transplanted to BALB/c scid mice and the animals observed for viral pathogenesis and death over a 35-day period.
Outcome of ganglion transplantation
Acute protocol
Results from three replicate experiments revealed that the BALB/c scid recipients of ganglion transplants from acutely infected BALB/c donors transplanted 3 days after infection (N = 19) all died, with a mean time to death of 14 days and a standard deviation of ± 2 days (Fig. 1). In contrast, all of the immunocompetent BALB/c recipients of acutely infected ganglia (N = 15) survived (Fig. 1).
Figure 1 Acute Protocol: Kaplan Meier analysis of the fate of ganglion transplant recipients. BALB/c scid mice (N = 19) and immunocompetent BALB/c mice (N = 15) were observed daily for evidence of infection and death. The day of death was recorded as the number of days after grafting. BALB/c scid recipients all died by day 16, whereas all of the immunocompetent BALB/c recipients survived throughout the entire observation period (P < 0.0001).
The cause of death in the BALB/c scid mice was not extensively examined in this study. As the animals became progressively moribund, it was evident that they were experiencing a neurological disease resembling encephalitis. In randomly chosen animals, virus was isolated from the ear graft site at the time that the animal died. Confirmation that virus was replicating at the transplant site is provided below.
Latent protocol
In a series of three experiments involving a total of 100 recipients (70 BALB/c scid, 30 immunocompetent BALB/c), only 1 of the 70 BALB/c scid recipients of a latent ganglion transplant died (28 days after grafting). Virus was not isolated from the graft site or the brain of this animal. As shown in Figure 2, the remainder of the BALB/c scid recipients survived without evidence of viral infection up to 65 days after grafting. All of the 30 immunocompetent BALB/c recipients of latent ganglion transplants survived for the duration of the experiment (Fig. 2). Mice in both of these groups were bled at 21, 45, and 65 days after grafting to test their sera for anti-HSV-1 antibodies, as described below.
Figure 2 Latent Protocol: Kaplan Meier analysis of the fate of BALB/c scid (N = 70) and immunocompetent BALB/c (N = 30) ganglion recipients. Mice were observed daily for evidence of infection and death. The day of death was recorded as the number of days after grafting. One of the BALB/c scid mice died on day28, but virus was not found in the graft site or brain of this animal. There was no significant difference between the survival of the BALB/c scid and immunocompetent BALB/c mice (P > 0.05).
Reactivation protocol
Ganglia from BALB/c mice latent for HSV-1 were passaged in tissue culture for 3 days and then transplanted into BALB/c scid recipients (N = 24) and immunocompetent BALB/c recipients (N = 16). All of the BALB/c scid recipients developed viral infections and viral-mediated neurologic disease and died, with a mean time to death of 14 days (Fig. 3). Analyses of the transplanted tissue and the local graft site supported the conclusion that the cause of death was the reactivated virus. In contrast, all of the immunocompetent BALB/c recipients of reactivated ganglia survived without evidence of disease (Fig. 3) and ultimately became immune to the virus.
Figure 3 Reactivation Protocol: Kaplan Meier analysis of the fate of BALB/c scid (N = 24) and immunocompetent BALB/c (N = 16) recipients of ganglion grafts. Mice were observed daily for infection and death. The day of death was recorded as the number of days after grafting. All of the BALB/c scid recipients were dead by day 18, whereas all of the immunocompetent BALB/c recipients survived throughout the entire observation period (P < 0.0001).
Serum anti-HSV-1 antibody responses in ganglion recipients
None of the BALB/c scid recipients of latently infected ganglia developed a serum IgG antibody response as measured by enzyme-linked immunosorbent assay (ELISA). Randomly chosen BALB/c scid mice tested at 21, 45, and 65 days after transplantation showed no evidence of having developed a humoral immune response to the virus (Fig. 4a). The BALB/c mice that received latent ganglion transplants did not exhibit an anti-HSV-1 antibody response on day 21 after transplantation, but had serum antibody on days 45 and 65 after grafting (Fig. 4a). The immunocompetent BALB/c recipients of acutely infected ganglia or of ganglia containing reactivating virus developed serum antibody IgG responses by 21 days after infection, which were present also at 45 and 65 days after transplantation (Fig. 4b).
Figure 4 Serum anti-HSV-1 antibody responses. (a) Serum anti-HSV-1 antibody responses of BALB/c scid (N = 8) and immunocompetent BALB/c (N = 8) mice in the Latent Protocol. The mice were grafted with ganglia containing latent virus and randomly chosen mice were bled on days 21, 45, and 65 after transplantation. The corrected optical density readings indicate that the BALB/c scid mice did not produce IgG antibody, whereas the immunocompetent BALB/c mice all had serum IgG anti-HSV-1 antibody on days 45 and 65 after transplantation. (b) Serum antibody responses of BALB/c mice receiving acutely infected ganglia (N = 4) or ganglia containing reactivating virus (N = 4). The optical density readings indicate that the immunocompetent mice in the Acute Protocol and the Reactivation Protocol developed serum anti-HSV-1 antibody titers by day 21 after grafting and that this antibody continued to be present on days 45 and 65 after transplantation. At each of the three time points, symbols representing one mouse each are spread out to avoid concealment by overlap.
Cell-mediated immunity in ganglion transplant recipients
BALB/c scid mice from the Latent Protocol that survived the 65-day observation period were tested by footpad swelling assay. None of the animals tested gave evidence of a delayed-type hypersensitivity response to viral antigens (Fig. 5). In contrast, all of the immunocompetent BALB/c recipients of acutely infected ganglia and recipients of ganglia undergoing viral reactivation exhibited delayed-type hypersensitivity responses on day 65 after ganglion transplantation (Fig. 5). Four of seven immunocompetent BALB/c recipients of latent ganglia also had positive delayed-type hypersensitivity responses on day 65 after transplantation (Fig. 5).
Figure 5 Footpad swelling responses to measure cell-mediated immunity. Footpad swelling responses in the BALB/c scid mice in the Latent Protocol (N = 9) and immunocompetent BALB/c recipients in the Latent (N = 7), Acute (N = 7), and Reactivation (N = 6) Protocols are shown. Included in the data sets are the footpad swelling responses of immune(N = 6) and nonimmune (N = 6) immunocompetent BALB/c mice. The BALB/c scid recipients of latent ganglia failed to develop a cell-mediated immune response, whereas four of the seven BALB/c wild-type recipients of latent ganglia showed a delayed-type hypersensitivity response. All of the immunocompetent BALB/c recipients of ganglia in the Acute and Reactivation Protocols exhibited cellular immune reactivity. Comparison of the footpad swelling response of the BALB/c immune mice, immunocompetent BALB/c recipients in the Reactivation Protocol and the Acute Protocol, and immunocompetent BALB/c recipients in the Latent Protocol with the BALB/c scid recipients in the Latent Protocol revealed that the response in the immunocompetent mice in each group was significantly greater than the response in the BALB/c scid mice (P < 0.001). Values are means ± SD.
Viability of the ganglion transplants
Latent ganglia that had been transplanted into BALB/c scid and immunocompetent BALB/c recipients were removed from the graft recipients on days 5, 14, or 28 after transplantation and placed into tissue culture. Seventeen of 18 latent ganglia removed from BALB/c scid animals underwent reactivation in vitro (Table 4), typically within 3 to 6 days after explantation. All of the latent ganglia recovered from immunocompetent BALB/c recipients underwent reactivation in tissue culture between days 6 and 10 after explantation.
Table 4 Recovery (reactivation) of virus in explanted ganglion graftsa
Source of explants Day of explantation relative to day of grafting
5 14 28
BALB/c scid Latent Protocol 5/5b 6/6 6/7
Immunocompetent BALB/c Latent Protocol 4/4 7/7 5/5
aGanglion grafts were placed into culture on the day indicated and the culture medium tested for infectious virus on days 1, 3, 5, 7, 10, 14, and 21 of incubation.
bNumber of ganglia from which virus reactivated/number of ganglia tested.
The histology of ganglion transplants was examined on days 5, 28, and 65 after transplantation. Although the architecture of transplanted ganglia was somewhat altered compared with that of freshly isolated ganglia (Fig. 6a), numerous large cells with the morphology of neurons were seen in latent ganglia transplanted to BALB/c scid (Fig. 6b) and immunocompetent BALB/c (Fig. 6c) recipients.
Figure 6 Histology of ganglion grafts. (a) Hematoxylin and eosin (H & E) stained section of a freshly isolated trigeminal ganglion. The large neuron cell bodies (arrows) interspersed among a field of nerve fibers present the typical histologic appearance of the trigeminal ganglion. (original magnification 400×) (b) H & E stained section of a latent ganglion graft removed from a BALB/c scid mouse 45 days after grafting. In this section, neuron cell bodies (arrows) with typical morphology can be seen. (original magnification 400×) (c) H & E stained section of alatent ganglion graft removed from an immunocompetent BALB/c recipient 45 days after transplantation. Clusters of neuron cell bodies (arrows) with typical morphology can be seen. (original magnification 400×)
Table 5 presents the results of the vital dye staining of cells isolated from ganglion transplants. Viable small (non-neuronal) and large (neurons) cells were found in roughly the same ratios on days 5, 14, 28, and 65 after transplantation. These ratios were similar to the ratios of small and large cells obtained from freshly isolated ganglia (Table 5).
Table 5 Viability of cells in ganglion graftsa
Source of ganglia Day of cell viability determination relative to day of grafting
5 14 28 65
Small Large Small Large Small Large Small Large
BALB/c scid Latent Protocol 328/416b (79) 54/72 (75) 477/519 (92) 38/49 (78) 622/705 (88) 88/97 (91) 219/279 (78) 41/54 (76)
Immuno-competent BALB/c Latent Protocol 789/885 (89) 66/81 (81) 413/500 (83) 75/82 (91) 917/998 (92) 31/48 (65) 261/308 (85) 87/101 (86)
Freshly isolated ganglia 419/524 (80) 88/110 (80) 523/567 (92) 101/123 (82) 816/911 (90) 23/38 (61) 377/408
(92) 99/122 (81)
aGanglion grafts or freshly isolated ganglia were dissociated and stained with vital DNA dye and trypan blue. The total numbers of small cells (10–50 μm) and large cells (larger than 50 μm), as well as the number of live cells in each size category, were determined by microscopic examination.
bNumber of viable cells of each size/total number of cells of each size counted (%).
Confirmation that viral reactivation and viral glycoprotein synthesis was occurring in ganglia transplanted during acute infection and in ganglia transplanted following reactivation was obtained by performing immunohistochemical staining for HSV-1 antigens in tissue sections. Staining of viral antigens was seen in the ganglion transplants and adjacent ear cells in BALB/c scid recipients of acutely infected (Fig. 7a) and reactivating (Fig. 7b) ganglia, but not in recipients of latent ganglion transplants (Fig. 7c).
Figure 7 Immunohistochemical staining for viral antigens. (a) Immunohistochemical staining of a tissue section through the ear graft site of a BALB/c scid mouse containing a ganglion from an acutely infected donor 7 days after grafting. In this immunoperoxidase-stained section, cells expressing HSV-1 antigens can be seen in the ganglion graft (arrows). (original magnification, 400×) (b) Immunohistochemical staining for HSV-1 antigens in the ganglion graft in a BALB/c scid mouse from the Reactivation Protocol. Cells expressing HSV-1 antigen are seen in both the graft and the surrounding ear cells at the graft site (arrows). (original magnification, 400×) (c) Immunohistochemical staining of a latent ganglion graft in a BALB/c scid recipient at 28 days after grafting. No evidence of cells expressing HSV-1antigens was seen in these tissue sections. (original magnification, 400×)
Discussion
The results of these experiments indicate that little, if any, infectious virus is produced during latency in mice. It has been proposed that some component of the immune system is necessary to induce HSV-1 into latency and prevent viral reactivation [9-15]. Sawtell [16] reported that immune cells in mouse ganglia do not inhibit viral reactivation. Thus, the role of antiviral immunity in the establishment and maintenance of latency is still being debated.
The immunocompetent BALB/c mice in the acute, latent, and reactivation protocols developed cellular and humoral immunity, indicating that there was an adequate amount of viral antigen produced in all three circumstances to sensitize the recipients. The finding that T cell-mediated and humoral responses developed and were sustained for 65 days suggests that viral antigen expression during latency has a role in this process. BALB/c scid mice lack an acquired immune system but have an intact innate immune system, including cells such as macrophages and natural killer (NK) cells and antiviral cytokines such as the types 1 and 2 interferons. NK cells alone cannot protect scid mice from HSV-1 infection and pathogenesis [17]. However, protection against viral-mediated death can be provided by T lymphocytes [18-21]. There is ample evidence to indicate that the interferons modulate the level of viral replication and spread, although this response is not known to protect scid mice from HSV-1-mediated death [22-24]. However, in the absence of an acquired immune system and, in particular, T lymphocytes, the virus evades the interferon response, enters the nervous system, and replicates in vital cells causing a fatal encephalitis.
The findings of the current study appear to imply that mouse neural tissue containing latent HSV-1 (e.g., the trigeminal ganglion) does not support periodic episodic viral reactivation. Although spontaneous molecular viral reactivation has been reported [8], the results presented here suggest that this molecular reactivation does not proceed to the production of detectable infectious virus. It may be that there are nonimmunological cellular or molecular factors that prevent spontaneous viral reactivation in mice.
These findings in vivo are particularly important since it is known that explanted mouse neural tissues latent for HSV-1 demonstrate viral reactivation in culture. This suggests that explantation itself or factors in tissue culture that we do not understand may suppress or destroy the in vivo factors that maintain viral latency.
A number of reports describe the induction of viral reactivation from latency in mice using a variety of stimuli such as immunosuppressive drugs, UV irradiation, and thermal stress [25-29]. These induction protocols yield a variable frequency of viral reactivation. There are no reports confirming spontaneous episodic shedding of virus at epithelial surfaces of mice, including the eye and genitalia, although it has been reported recently that infectious virus is present in the trigeminal ganglion up to 240 days after infection [16].
The possibility that the surgical trauma of ganglion transplantation or the site in which the transplant was placed (the ear) prevents viral reactivation from occurring and/or prevents infectious virus from leaving this site to infect the animal's nervous system must be considered. However, in the Acute and Reactivation Protocol mice, it was found that ganglia containing infectious virus, either in the acute stage of infection or following reactivation, placed into the ear pocket resulted in spread of the virus from this site to the nervous system of BALB/c scid mice, resulting in encephalitis and death. Additionally, injection of 10 viral particles into ganglion grafts resulted in viral infection and death of 50% of BALB/c scid mice, demonstrating the sensitivity of this in vivo system and confirming that the subcutaneous ear site is not a sequestered site that prevents the escape of infectious virus.
Conclusion
It is concluded that measurable infectious virus is not produced under the conditions of these experiments. Thus, the technical approach used here appears to be a valid and sensitive measure of the presence of infectious virus. The results of this study reveal that molecular reactivation, i.e., expression of HSV-1 genes and production of glycoproteins during latency, occurs in mice and extends this observation to establish that molecular reactivation does not necessarily lead to the production of infectious viral particles. The approach used in this investigation opens up new vistas for studying herpesvirus latency and reactivation.
Methods
Mice
Female BALB/cJ and BALB/c scidJ mice at 5 weeks of age (The Jackson Laboratory, Bar Harbor, ME) were used. Confirmation that the BALB/c scid mice were immune deficient was achieved by performing flow cytometric analysis of spleen cells for CD3+ T cells and membrane immunoglobulin-positive cells. No evidence of the presence of T or B lymphocytes in the BALB/c scid mice sacrificed throughout the course of this study was obtained (data not shown). Animals studies were approved by the Louisiana State University Health Sciences Center Institutional Animal Care and Use Committee (IACUC). All animals were provided with food and water ad libidum and were cared for according to the NIH Guidelines on the Care and Use of Animals in Research.
Virus
The McKrae strain of HSV-1, a strain which is known to spontaneously reactivate in rabbits, was propagated in and titered on Vero cells (American Type Culture Collection, Manassas, VA). At the time of infection, the virus stock was thawed and diluted so as to deliver 1 × 105 PFU in 4 μl of culture medium. BALB/c mice to be infected were anesthetized, their corneas were lightly scratched in a cross-hatched pattern, and 4 μl of the viral suspension was placed on the surface of each eye. In order to ensure survival, infected animals each received 0.1 ml of pooled human serum (Chemicon International, Temecula, CA) intraperitoneally at the time of infection. At 3 and 5 days after infection, the ocular surface of the animals was swabbed and tested for the presence of infectious virus by the viral plaque assay. Animals not giving evidence of infection were excluded from the study.
Analysis of the trigeminal ganglion for infectious virus during latency
Thirty-five BALB/c strain mice were infected with the McKrae strain of HSV-1 by the topical ocular route as described above. Five days after infection, the eyes of all animals were swabbed for the determination of infectious virus. Also on day 5, five animals were killed and their trigeminal ganglia were separately homogenized in 0.5 ml of tissue culture medium. The ganglion homogenates were tested for infectious virus on Vero cell monolayers in 24-well tissue culture plates. In this experiment, no attempt was made to quantitate infectious virus, but only to note the presence or absence of infectious virus in the ganglia. Eye swabs and trigeminal ganglion homogenates were similarly tested from five additional animals killed at each of the following time points: 10, 20, 30, 50, 70, and 100 days after infection.
Determination of the sensitivity of the assay for infectious virus in the ganglion
Groups of 10 uninfected BALB/c mice were sacrificed and their trigeminal ganglia removed intact and placed into tissue culture medium. The 20 ganglia from each group of mice were positioned under a stereoscopic microscope and each ganglion injected with a 5 μl volume of culture medium containing 100, 50, 10, 5, or 1 PFU of the McKrae strain of HSV-1. Ten ganglia from each group were immediately homogenized in 0.5 ml of culture medium, the homogenate clarified by centrifugation at 8000 × g in a microcentrifuge, and the supernatant plated on Vero cells in 24-well plates. Following viral attachment, the supernatant was removed and a 0.5% methylcellulose overlay was placed in each well. The plates were incubated for 2 days and plaques counted. The remaining 10 ganglia were transplanted into subcutaneous ear pockets in BALB/c scid mouse recipients as described below. The mice were observed daily for signs of viral pathogenesis and death.
Ganglion transplantation
Recipient mice were anesthetized with a mixture of ketamine and xylazine and positioned under a stereoscopic dissecting microscope such that the entire ear pinna could be seen at 10× magnification. The tip of the ear was gently grasped with sterile forceps and an incision made in the dorsal skin surface with a sterile lamellar blade (Wilson Ophthalmic, Mustang, OK). The lamellar blade was gently eased below the surface of the epithelium with a side-to-side and insertion-retraction motion, creating a pocket approximately 3 mm wide and 7 mm deep. Individual ganglia were gently inserted into the ear pockets so as to allow the open end of the ear pocket to close over the graft and self-seal, thus enclosing the ganglion in the pocket and avoiding the need for sutures. One application of neomycin and polymixin sulfate ointment externally was adequate to prevent bacterial infection. The recipient animals were returned to their cages for recovery from the anesthetic and the external condition of the graft site was observed daily to ensure the success of the transplant.
Three ganglion transplantation protocols were performed:
1) Acute Protocol: BALB/c mice and BALB/c scid mice received trigeminal ganglion grafts from BALB/c donors that had been infected 5 days previously with McKrae strain HSV-1. Recipient mice were observed for evidence of viral infection, development of tissue pathology at the site of the transplant, signs of morbidity, and death. At the time of sacrifice or the time of death, serum was collected for testing for antibodies to HSV-1.
2) Latent Protocol: Trigeminal ganglia from BALB/c mice that had been infected 35 days previously with McKrae strain HSV-1 were transplanted into BALB/c and BALB/c scid mice as described above. The recipient mice were observed as described above for evidence of viral infection, viral-induced tissue pathology at the transplant site, signs of morbidity, and death. In replicates of this experiment, recipient mice were anesthetized on days 5, 14, or 28 after transplantation and the ganglion transplants removed from the skin pockets and placed in tissue culture to test for the presence of latent, reactivatable herpesvirus in the transplanted ganglion. For in vitro incubation, ganglia from latent BALB/c mice and explanted ganglion transplants were incubated in separate wells of 12-well culture plates containing Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and an antibiotic/antimycotic mixture (GIBCO, Carlsbad, CA). The culture medium in each well was assayed for infectious virus on Vero cell monolayers at 1, 3, 5, 7, 10, 14, and 21 days of incubation.
3) Reactivation Protocol: Trigeminal ganglia latent with the McKrae strain of HSV-1 obtained from BALB/c mice were incubated in tissue culture for 3 days, and then transplanted into BALB/c and BALB/c scid recipients. The mice were observed for evidence of viral infection, virus-induced tissue pathology at the transplant site, signs of morbidity, and death. The recipients were tested for the development of antiviral antibody by ELISA.
ELISA for serum antibody
Serum was collected from the BALB/c and BALB/c scid animals at the time of sacrifice. ELISA plates were coated with a cell culture lysate from Vero cells infected 18 hours previously with McKrae strain HSV-1. This lysate contains a mixture of HSV-1 antigens and has been used in previous studies [30,31]. Equal numbers of ELISA plate wells were coated with a cell culture lysate of uninfected Vero cells. The plate wells were washed three times with Tris buffered saline (TBS) and 1:50 dilutions of each serum were tested in quadruplicate for reactivity with the infected and uninfected cell lysates. Binding of serum anti-HSV antibody was detected with a secondary rabbit anti-mouse IgG antibody coupled to horseradish peroxidase (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Following washing, the plate wells were incubated in tetramethylbenzidine substrate (DakoCytomation, Inc., Carpinteria, CA) for 15 minutes at room temperature and the reaction stopped with 1 M sulfuric acid. The optical densities were read at 450 nm in a plate reader and the optical density values for each serum sample tested on the infected cell lysate were corrected by subtraction of the optical density obtained with the serum on the uninfected cell lysate.
Footpad swelling assay for delayed type hypersensitivity
The same infected cell lysate used to coat the ELISA plates was treated with UV light for 10 minutes to inactivate infectious virus. Mice were anesthetized with a mixture of ketamine and xylazine. The left hind footpads were injected with 10 μl of the treated, infected lysate and the right hind footpad received 10 μl of the uninfected cell lysate. At 24 hours, the footpad swelling response was measured using a spring-loaded micrometer gauge (Starett, Inc., Athol, MA). Four measurements were made of each right and left footpad. Delayed-type hypersensitivity reactions were calculated as follows: specific footpad swelling = (24 hr measurement of left footpad - 0 hr measurement of left footpad) - (24 hr measurement of right footpad - 0 hr measurement of right footpad) × 103 mm. In each experiment, in addition to the test animals, mice not immune to HSV-1 and mice immunized with UV-inactivated virus were used as negative and positive controls.
Histology and immunohistochemical staining
Animals selected at random were sacrificed and the portion of the ear containing the ganglion transplant site was frozen and sectioned in a cryotome. The sections (10 μm) were placed on microscope slides and fixed in cold acetone for 5 minutes. Representative sections were stained with hematoxylin and eosin for histopathologic examination. Additional sections were stained for cells expressing HSV-1 antigens. A direct staining method employing a polyclonal, horseradish peroxidase-conjugated rabbit anti-HSV-1 antibody (DakoCytomation) was used. The antibody was diluted 1:200 in TBS containing 1% bovine serum albumin. Following incubation for 1 hour, the slides were washed in three washes of TBS for 5 minutes each and then incubated in the substrate consisting of diaminobenzidine and hydrogen peroxide (Pierce Biotechnology, Inc., Rockford, IL). Color development was stopped after 5 minutes and the sections were counterstained with methyl green. Control slides were incubated in an irrelevant peroxidase-labeled rabbit antibody followed by the substrate.
Cell viability in ganglion transplants
Transplanted ganglia were removed from ear pockets at 5, 14, 28, and 65 days after transplantation. Each ganglion was teased apart with forceps in 2 ml of calcium/magnesium-free Hank's balanced salt solution (GIBCO) containing 10 U DNase, 0.1 mg/ml dispase, 0.1 mg/ml collagenase, and 0.1 mg/ml trypsin. The tissue fragments were incubated with gentle stirring for 15 minutes at 37°C. The cells and tissue clumps were gently triturated and washed two times in DMEM/10% FBS. The cells were resuspended in 2 ml DMEM/10% FBS containing 1 μg/ml of Hoechst 33342 vital DNA stain (H342, Calbiochem, La Jolla, CA) for 15 minutes at 37°C. The cells were washed twice in DMEM/FBS and resuspended in 2 ml of medium. Living and dead cells were differentiated with the addition of 0.1% trypan blue, which quenches the H342 fluorescence of dead cells [32]. The cells were placed onto clean microscope slides, coverslipped, and immediately examined on a Nikon E600 fluorescence microscope with a UV-2E/C (330–380 excitation; 435–485 barrier) filter. The cells were categorized as either small (10–50 μm in diameter, non-neurons) or larger than 50 μm in diameter (neurons). For each suspension, 10 microscopic fields at 100× magnification were examined, and the total number of cells, the numbers of small and large cells, and the numbers of fluorescing (i.e., not quenched, viable) cells in each of the two size ranges were determined.
Statistical analysis
Analysis of numerical data and statistical analyses were performed with Microsoft Excel (Redmond, WA), Modstat (Modern Microcomputers, Mechanicsville, VA), and CoStat (Cohort Software, Monterey, CA). Fisher's exact test was used to compare the differences in survival frequencies between groups of mice. P values less than 0.01 were considered significant.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BMG carried out the ganglion transplantation experiments and immunoassays. WPH carried out the viral infections and plaque assays. BMG and WPH conceived of the study. BMG wrote the manuscript.
Acknowledgements
This work was supported in part by U.S. Public Health Service grants R01EY002672 (BMG), R01AI054104 (WPH), and P30EY002377 (LSU Eye Center Core grant) from the National Institutes of Health, Bethesda, Maryland, an unrestricted challenge grant (LSU Eye Center) from Research to Prevent Blindness, Inc., New York, New York, and a National Science Foundation EPSCoR grant (EPS0346458) to Montana State University (WPH).
==== Refs
Mester JC Rouse BT The mouse model and understanding immunity to herpes simplex virus Rev Infect Dis 1991 13 S935 S945 1664131
Hill TJ Harbour DA Blyth WA Isolation of herpes simplex virus from the skin of clinically normal mice during latent infection J Gen Virol 1980 47 205 207 6245170
Stevens JG Cook ML Maintenance of latent herpetic infection: an apparent role for anti-viral IgG J Immunol 1974 113 1685 1693 4372269
Tenser RB Edris WA Gaydos A Hay KA Secondary herpes simplex virus latent infection in transplanted ganglia J Virol 1994 68 7212 7220 7933103
Willey DE Trousdale MD Nesburn AB Reactivation of murine latent HSV infection by epinephrine iontophoresis Invest Ophthalmol Vis Sci 1984 25 945 950 6086546
Deatly AM Spivack JG Lavi E Fraser NW RNA from an immediate early region of the type 1 herpes simplex virus genome is present in the trigeminal ganglia of latently infected mice Proc Natl Acad Sci USA 1987 84 3204 3208 3033640
Stevens JG Wagner EK Devi-Rao GB Cook ML Feldman LT RNA complementary to a herpesvirus alpha gene mRNA is prominent in latently infected neurons Science 1987 235 1056 1059 2434993
Feldman LT Ellison AR Voytek CC Yang L Krause P Margolis TP Spontaneous molecular reactivation of herpes simplex virus type 1 latency in mice Proc Natl Acad Sci USA 2002 99 978 983 11773630 10.1073/pnas.022301899
Halford WP Gebhardt BM Carr DJJ Persistent cytokine expression in trigeminal ganglion latently infected with herpes simplex virus type 1 J Immunol 1996 157 3542 3549 8871654
Keadle TL Laycock KA Miller JK Hook KK Fenoglio ED Francotte M Slaoui M Stuart PM Pepose JS Efficacy of a recombinant glycoprotein D subunit vaccine on the development of primary and recurrent ocular infection with herpes simplex virus type 1 in mice J Infect Dis 1997 176 331 338 9237697
Liu T Khanna KM Carriere BN Hendricks RL Gamma interferon can prevent herpes simplex virus type 1 reactivation from latency in sensory neurons J Virol 2001 75 11178 11184 11602757 10.1128/JVI.75.22.11178-11184.2001
Liu T Khanna KM Chen X Fink DJ Hendricks RL CD8+ T cells can block herpes simplex virus type 1 (HSV-1) reactivation from latency in sensory neurons J Exp Med 2000 191 1459 1466 10790421 10.1084/jem.191.9.1459
Shimeld C Easty DL Hill TJ Reactivation of herpes simplex virus type 1 in the mouse trigeminal ganglion: an in vivo study of virus antigen and cytokines J Virol 1999 73 1767 1773 9971753
Shimeld C Whiteland JL Nicholls SM Grinfeld E Easty DL Gao H Hill TJ Immune cell infiltration and persistence in the mouse trigeminal ganglion after infection of the cornea with herpes simplex virus type 1 J Neuroimmunol 1995 61 7 16 7560014 10.1016/0165-5728(95)00068-D
Shimeld C Whiteland JL Williams NA Easty DL Hill TJ Reactivation of herpes simplex virus type 1 in the mouse trigeminal ganglion: an in vivo study of virus antigen and immune cell infiltration J Gen Virol 1996 77 2583 2590 8887494
Sawtell NM Quantitative analysis of herpes simplex virus reactivation in vivo demonstrates that reactivation in the nervous system is not inhibited at early times postinoculation J Virol 2003 77 4127 4138 12634371 10.1128/JVI.77.7.4127-4138.2003
Halford WP Maender JL Gebhardt BM Re-evaluating the role of natural killer cells in innate resistance to herpes simplex virus type 1 Virol J 2005 2 56 16022737 10.1186/1743-422X-2-56
Minagawa H Sakuma S Mohri S Mori R Watanabe T Herpes simplex virus type 1 infection in mice with severe combined immunodeficiency (SCID) Arch Virol 1988 103 73 82 2850780 10.1007/BF01319810
Bouley DM Kanangat S Rouse BT The role of the innate immune system in the reconstituted SCID mouse model of herpetic stromal keratitis Clin Immunol Immunopathol 1996 80 23 30 8674236 10.1006/clin.1996.0090
Minagawa H Yanagi Y Latent herpes simplex virus-1 infection in SCID mice transferred with immune CD4+ T cells: a new model for latency Arch Virol 2000 145 2259 2272 11205116 10.1007/s007050070019
Irie H Aita K Koyama AH Fukuda A Yoshida T Shiga J The role of donor CD4+ T cells in the reconstitution of oral immunity by herpes simplex virus type 1 in severe combined immunodeficiency mice J Infect Dis 2002 185 409 416 11865391 10.1086/338835
Geiger K Howes EL Sarvetnick N Ectopic expression of gamma interferon in the eye protects transgenic mice from intraocular herpes simplex virus type 1 infections J Virol 1994 68 5556 5567 8057437
Geiger KD Nash TC Sawyer S Krahl T Patstone G Reed JC Krajewski S Dalton D Buchmeier MJ Sarvetnick N Interferon-gamma protects against herpes simplex virus type 1-mediated neuronal death Virology 1997 238 189 197 9400592 10.1006/viro.1997.8841
Mikloska Z Cunningham AL Alpha and gamma interferons inhibit herpes simplex virus type 1 infection and spread in epidermal cells after axonal transmission J Virol 2001 75 11821 11826 11689663 10.1128/JVI.75.23.11821-11826.2001
Blatt AN Laycock KA Brady RH Traynor P Krogstad DJ Pepose JS Prophylactic acyclovir effectively reduces herpes simplex virus type 1 reactivation after exposure of latently infected mice to ultraviolet B Invest Ophthalmol Vis Sci 1993 34 3459 3465 8225880
Cook SD Paveloff MJ Doucet JJ Cottingham AJ Sedarati F Hill JM Ocular herpes simplex virus reactivation in mice latently infected with latency-associated transcript mutants Invest Ophthalmol Vis Sci 1991 32 1558 1561 1849874
Gebhardt BM Varnell ED Hill JM Kaufman HE Zak O, Sande MA Animal models of ocular herpes simplex virus infection (rabbits, primates, mice) Handbook of Animal Models of Infection: Experimental Models in Antimicrobial Chemotherapy 1999 San Diego, CA: Academic Press 919 926
Sawtell NM Thompson RL Rapid in vivo reactivation of herpes simplex virus in latently infected murine ganglionic neurons after transient hyperthermia J Virol 1992 66 2150 2156 1312625
Shimeld C Hill TJ Blyth WA Easty DL Reactivation of latent infection and induction of recurrent herpetic eye disease in mice J Gen Virol 1990 71 397 404 2155293
Gebhardt BM Hill JM T lymphocytes in the trigeminal ganglia of rabbits during corneal HSV infection Invest Ophthalmol Vis Sci 1988 29 1683 1691 3263346
Gebhardt BM Evidence for antigenic cross-reactivity between herpesvirus and the acetylcholine receptor J Neuroimmunol 2000 105 145 153 10742556 10.1016/S0165-5728(00)00204-6
Reynolds CP Black AT Woody JN Sensitive method for detecting viable cells seeded into bone marrow Cancer Res 1986 46 5878 5881 2428486
|
16109179
|
PMC1208961
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Aug 18; 2:67
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-67
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-711611532210.1186/1743-422X-2-71ResearchBaculovirus-mediated promoter assay and transcriptional analysis of white spot syndrome virus orf427 gene Lu Liqun [email protected] Hai [email protected] Ivanus [email protected] Li [email protected] Jimmy [email protected] Animal health biotechnology unit, Temasek life sciences laboratory, 1 Research Link, National University of Singapore, 117604, Singapore2005 23 8 2005 2 71 71 8 7 2005 23 8 2005 Copyright © 2005 Lu et al; licensee BioMed Central Ltd.2005Lu et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
White spot syndrome virus (WSSV) is an important pathogen of the penaeid shrimp with high mortalities. In previous reports, Orf427 of WSSV is characterized as one of the three major latency-associated genes of WSSV. Here, we were interested to analyze the promoter of orf427 and its expression during viral pathogenesis.
Results
in situ hybridization revealed that orf427 was transcribed in all the infected tissues during viral lytic infection and the translational product can be detected from the infected shrimp. A time-course RT-PCR analysis indicated that transcriptional products of orf427 could only be detected after 6 h post virus inoculation. Furthermore, a baculovirus-mediated promoter analysis indicated that the promoter of orf427 failed to express the EGFP reporter gene in both insect SF9 cells and primary shrimp cells.
Conclusion
Our data suggested that latency-related orf427 might not play an important role in activating virus replication from latent phase due to its late transcription during the lytic infection.
==== Body
Background
White spot syndrome virus (WSSV) was assigned to the genus Whispovirus belonging to new family Nimaviridae in the universal database of ICTV (International Committee of Taxonomy of Viruses, ). WSSV is probably the most important pathogen of the cultured penaeid shrimp resulting in high mortalities [1]. Even though WSSV represents one of the largest known animal viruses with a 305 kb double-stranded circular DNA genome, most of the putative 185 ORFs bear no homology to known genes in the GenBank [2,3]. The technical difficulty in characterization of the WSSV ORFs lies mainly in the lack of established shrimp cell lines for in vitro reproduction of the virus [4]. During viral lytic infection, just as other DNA viruses, the genes encoded by WSSV can be classified as immediately early, delayed early, late and very late genes. Most, if not all, immediate-early genes encode transcriptional regulation proteins. They are distinguished from other viral genes by the fact that their transcription is independent of prior viral gene product expressed in transient assays [5]. Although during the last decade, intensive efforts have been undertaken for characterization of the structural protein genes and a few non-structural protein genes that show homology to known sequences in the databases, little is known about the molecular mechanisms underlying the WSSV life cycle and mode of infection.
Recently, three viral transcripts (Orf427, Orf151 and Orf366) and their corresponding DNA sequence have been detected in both specific-pathogen-free (SPF) shrimps and WSSV-infected shrimps through a WSSV-specific DNA microarray study. From this study, Orf427, Orf151 and Orf366 were determined to be latency-associated genes of WSSV [6]. These data suggest that WSSV remains latent in healthy shrimps. In a similar global analysis, three immediately early (IE) genes (ie1, ie2, and ie3) of WSSV were identified in infected shrimps [7]. Identification of the IE genes and latency-associated genes can lead to better understanding of the life cycle of WSSV, shedding light on the molecular mechanisms in WSSV-induced mortality. In a previous study, we have found that latency-related ORF427 interacted with a shrimp protein phosphatase (PPs) [8]. To further characterize the orf427 gene, we were interested to analyze the promoter of orf427 and its expression during viral pathogenesis.
Results
To investigate whether promoter of orf427 is active without the existence of other viral proteins in the host cells, we tried to establish in vitro culture of fragments from lymphoid organ as reported previously [9]. However, the primary shrimp cells were very sensitive to standard liposome-based transfection reagents. Thus, for the promoter analysis, we employed a transduction method mediated by baculovirus [10]. Recombinant baculoviruses bearing EGFP-expressing cassettes were produced according to pFASTBac1 manufacturer instructions (Invitrogen) (Fig. 1A). Budded virus from insect cell culture medium was concentrated by ultrafiltration and infectious titers of both stock viruses were determined by plaque assay and adjusted to be 1010 plaque-forming units (PFU)/ml.
Figure 1 Baculovirus-mediated promoter analysis of orf427 compared with immediate-early gene ie1. A) Genomic organization of vAc-Proie1-EGFP and vAc-Pro427-EGFP. Pie1, promoter of ie1 gene; P427, promoter of orf427. Recombinant baculoviruses were constructed using the Bac-To-Bac system (Invitrogen). The EGFP-expressing cassettes were first cloned into the pFastBac1 shuttle vector at the indicated restriction sites and then integrated into the bacmid genome through site-specific transposition. B) Promoter activity of orf427 and ie1 gene in insect SF9 cells. Brightfield and EGFP fluorescence signals in SF9 cells infected with vAc-Proie1-EGFP and vAc-Pro427-EGFP at m.o.i of 10, respectively. C) Promoter activity of orf427 and ie1 gene in primary shrimp cells. Brightfield and EGFP fluorescence signals in primary shrimp cells transduced with vAc-Proie1-EGFP and vAc-Pro427-EGFP at m.o.i of 100, respectively. D) Western blot assay to confirm the expression of GFP in virus-infected or transducted cells. 1. Protein marker; 2. vAc-Proie1-EGFP infected SF9 cells; 3. vAc-Pro427-EGFP infected SF9 cells; 4. vAc-Proie1-EGFP transduced shrimp primary cells; 5. vAc-Pro427-EGFP transduced primary shrimp cells.
Infection of SF9 cells and transduction of shrimp primary cells with the recombinant baculovirses were carried out at a MOI of 10 and 100, respectively. As expected, the ie1 promoter drove the expression of the egfp reporter gene in both insect SF9 and the primary shrimp cells, as demonstrated by direct light and fluorescence microscopy; while the orf427 promoter didn't express egfp to a detectable level in either cell type (Fig. 1B and 1C). The expression of GFP could be confirmed in both cells through immunoblot assay using monoclonal anti-GFP antibody (Fig. 1D). We also noticed that the primary shrimp cells could only be transduced at a low percentage of about 5% (Fig. 1C).
In most cases, viruses establish latency in specific tissue(s). To test whether orf427 is transcribed only in specific latency sites or in all the tissues that support viral infection, in situ hybridization was performed on paraffin embedded tissue sections from shrimps at late infection (4 days after viral inoculation) using DIG-labeled antisense RNA probes specific for orf427. Results shown in fig. 2 indicated that in contrast to the control shrimp sections, orf427 was extensively transcribed in all the WSSV infected tissue sections including subcuticular epithelium cells (Fig. 2I), hemocytes lodged in the connective tissues (Fig. 2II), and stomach chamber lining cells (Fig. 2III). Also, we expressed and purified partial fragment of ORF427 in a GST-fusion form. Protein purity of the purified protein was more than 90% as judged by SDS-PAGE (figure not shown). Polyclonal antibody was developed by injection of the protein into Guinea pigs. ORF427 can be detected from homogenized infected shrimps through immunoblot assay using the anti-ORF427 antibody (Fig. 3).
Figure 2 Detection of orf427 mRNA in different tissue sections from WSSV-infected shrimp by in situ hybridization with specific orf427 antisense riboprobe. I: WSSV-infected shrimp; C: non-infected shrimp; the short bar is about 30 μm in length. 1) Section of subcuticle epithelium; 2) Section of hemocytes; 3) Section of stomach chamber lining cells.
Figure 3 Detection of ORF427 in infected shrimp through Western blot analysis. Western blotting analysis for detection of the endogenic ORF427 in infected shrimp cells. Polyclonal antibody toward Orf427 was raised using the bacterially expressed partial Orf427 as antigen from Guinea pigs. 1. Protein marker; 2, total shrimp cellular extracts sampled from normal shrimp; 3, total shrimp cellular extracts sampled from WSSV-infected shrimp.
In order to determine whether orf427 is transcribed in the early phase during viral lytic infection, we employed a RT-PCR approach to detect the transcriptional products of orf427. The sequences of the primers used are shown in Fig. 4A. P. monodon shrimps challenged through intramuscular injection with WSSV were sampled at different time points after viral inoculation, and total RNAs were extracted from the shrimp heads for RT-PCR analysis. As controls, fragments corresponding to the WSSV immediately early gene ie1 [7], delayed early gene dnapol [11], and late gene vp28 [12], were also amplified from the same RNA samples. A shrimp β-actin primer set was used as an internal control for RNA quality and amplification efficiency. Our results show that orf427 is only transcribed after 6 h post infection (Fig. 4B), which is at the late phase during viral lytic infection. As expected, ie1 can be detected from 3 h p.i., while dnapol and vp28 can be detected from 6 h p.i. (Fig. 4B).
Figure 4 Time course RT-PCR analysis of orf427 during viral pathogenesis. A) Gene specific primer sets used in the RT-PCR analysis as previously reported [6,7]. B) Agarose gel electrophoresis of RT-PCR products. Total RNA was sampled at the indicated time points post infection and RT-PCR was performed using primer sets specific for ie1, dnapol, vp28, orf427, and β-actin gene, individually. M: kb DNA ladder from Stratagene.
Discussion
Establishment and maintenance of latency in the host after primary infection have been investigated in some well-studied DNA viruses such as: herpes simplex virus (HSV) [13], human herpesvirus (HHV) [14], cytomegalovirus (CMV) [15], and Epstein-Barr virus [16]. However, the molecular mechanisms that control virus latency and reactivation remain to be elucidated. Because of problems associated with conducting molecular studies in animals, it has proven difficult for investigators to move beyond phenomenal description and identification of latency-associated transcripts (LATs). Most of the characterized LATs were expressed at low levels during lytic replication but were major transcripts during latent infection, and their functions were not understood. These include a set of latency-associated transcripts from the HHV-6 IE-A region [17], a set of genes controlled by the Qp promoter of Epstein-Barr virus [16], and latency-associated transcripts from both DNA strands in the ie1/ie2 region of CMV [15]. U94 gene of HHV-6 is one of the better-characterized LATs. U94 protein acts as a transactivator by binding to a transcription factor and enables the establishment and/or maintenance of latent infection at the primary infection site like monocytes and early bone marrow progenitor cells [18]. Our data indicate that orf427 is a very late gene during viral lytic infection, and this correlates with the finding that ORF427 is not a transcriptional regulator, but a protein phosphatase-interacting protein [8].
Most recently, nuclear protein phosphatase-1 was reported to regulate HIV-1 transcription both in vitro and in vivo [19]. Primary functional dissection of Orf427 suggests that orf427 most likely encodes a calcium-binding regulator of shrimp protein phosphatase, with the C terminus responsible for the binding of PPs (data not shown). This suggests that orf427 is not necessary for viral reactivation and only contributes to maintaining viral latency by affecting the function of shrimp protein phosphatase. Similarly, the LAT gene of HSV-1 has been shown to be dispensable for viral reactivation from latently infected mouse dorsal root ganglia cultured in vitro [20].
The development of a continuous shrimp cell line in vitro is urgently required for further characterization of WSSV infection at the molecular and cellular levels. In recent years, encouraging progress has been made in shrimp cell culture using conventional primary culture techniques. Several investigators have reported that WSSV infects the primary cultures of lymphoid organs from the black tiger shrimp, P. monodon; however, recent findings suggest that the replication of WSSV in lymphoid organ primary cell is low [4,9,21]. Besides this, the primary cell couldn't be transfected with common liposome methods. We thus took alternative approach to monitor the gene expression in the primary shrimp cells. Recently AcMNPV (Autographa californica multiple nucleopolyhedrovirus), containing an appropriate eukaryotic promoter, was used to efficiently transfer and express foreign genes in a variety of mammalian cells and several animal models [22]. Considering that shrimp is more phylogenically related to arthropods, the natural host of AcMNPV, we employed recombinant baculovirus-mediated transduction to introduce foreign genes into the primary shrimp cells. As expected, the primary shrimp cells were transduced in our experiments; and the low transduction efficiency might be due to the possible inhibition effect of L15 medium on the attachment of baculovirus to the cell membrane (for example, the pH value of medium for insect cells to amplify baculovirus is 6.8, while the pH value of L15 medium is above 7.0). The transduction efficiency might be significantly increased by using VSV-G-containing baculovirus as gene delivery vehicle [10]. The successful transduction of cultured shrimp cells with recombinant baculovirus may pave the way for the development of baculovirus-based vaccines for the shrimp farming industry.
Conclusion
The data presented here demonstrates that latency-associated Orf427 is only transcribed in the very late phase during viral lytic infection. In contrast to immediately early promoters, the promoter of orf427 couldn't drive the expression of an egfp reporter gene independently. Our data suggest that as a very late protein during viral lytic infection, ORF427 might only function in maintaining WSSV in the latent phase but is not required for virus reactivation.
Materials and methods
Virus, shrimp, and cells
WSSV used in this study was isolated from Penaeus monodon shrimps, which were imported from Indonesia. Purification of the virus was performed as previously described [6]. P. monodon shrimps challenged through intramuscular injection were sampled at different time points postinfection and immediately frozen and stored at -80°C. Adult P. monodon shrimps weighing approximately 30–100 g were used for primary cell culture. Monolayer cell cultures from minced fragments of lymphoid tissue were established as described by Chen [9]. Primary cells were maintained in 2 × L15 medium from Invitrogen. Insect SF9 cells (Invitrogen) were maintained and propagated in SF-900II serum-free medium (Invitrogen). Infection of SF9 cells and transduction of foreign genes into shrimp primary cells were performed as previously described [10].
Construction of recombinant baculoviruses
The ie1 basic promoter region from -1 to -512 was amplified using primer set of 5'-TCCCTACGTATCAATTTTATGTGGCTAATGGAGA-3' and 5'-ACGCGTCGA CCTTGAGTGGAGAGAGAGCTAGTTATAA-3' [7]. To make sure that the selected promoter region contained the full orf427 promoter, the upstream sequence of orf427, starting from -1 to -3500, was PCR-amplified from WSSV genome with primer set of 5'-TCCCTACGTATGGGTCAGAAAAGAACCC-3' and 5'-ACGCGTCGACATC TCAAAGGTTGCCAATGACCAACAT-3'. Both promoters were digested with SnaBI and SalI, and inserted into the corresponding sites of shuttle vector pFastBac1 (Invitrogen). The EGFP cDNA was first cut with SalI and NotI from the pEGFP-N1 vector (Clontech), followed by insertion into the pFASTBac1 vector bearing the promoter sequence of orf427 or ie1 gene. Recombinant baculoviruses bearing the EGFP-expression cassette were constructed according to the Bac-To-Bac protocol (Invitrogen). The infectious titers of the recombinant baculoviruses were determined by plaque assay with SF9 cells.
Development of polyclonal antibody and Western blot analysis
The C terminal partial fragment amplified from orf427 template using primer pair of 5'-CGGGATCCGTTAGAGCTTCAAAGGTGGA-3' and 5'-ACGCGTCGAC TTATTTTCCTTGATCTAGAG-3' was inserted into the pGEX4T-3 vector at BamH1 and Sal I site. The partial ORF427 was expressed and purified in E. coli as a glutathione S-transfererase (GST) fusion protein according to manufacturer's protol (Amersham Pharmacia). SPF Guinea pigs were immunized and specific antisera were prepared using standard procedures. Homogenized protein mixtures from infected shrimp or virus-infected cells were harvested and subjected to sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). Immunoblot analysis was performed according to standard protocol [23].
In situ hybridization
In situ hybridization was performed on paraffin embedded tissue sections using a DIG-labeled antisense RNA probes. Both WSSV-free shrimps and WSSV-infected shrimps were fixed in 4% (W/V) paraformaldehyde (PFA)-PBS, dehydrated, and embedded in paraffin. Sections of 6 μm in thickness were made and attached to 3-aminopropyltriethoxy-silane-coated slides. DIG-labeled antisense riboprobe specific for orf427 was synthesized by in vitro transcription using T7 RNA polymerase (Stratagene) and 10 × Dig labeling mix (Roche). The transcription template was PCR amplified from orf427 with a primer set of 5'-TAATACGACTCACTATAGGGCGCACCAGAAGAAAGGGTCT-3', and 5'-AAGGAAAC CATCGAGGCCTT-3'. The T7 promoter sequence was flanked at the 5' of the reverse primer. Hybridization was performed in 50% formamide and 5 × SSC in a humified chamber at 60°C for 14–16 h (the background is too high at 50°C in our hybridization system). The hybridization was visualized by using alkaline phosphatase-conjugated anti-digoxigenin antibody.
RT-PCR analysis
Total RNA was extracted from heads of the WSSV-infected shrimps using TRIzol-LS reagent (Life Technologies). For RT-PCR, an aliquot of total RNA (20 μg) was treated with 200 U of RNase-free DNase I (Gibco BRL) at 37°C for 30 min to remove residual DNA. First strand cDNA synthesis was performed using the oligo-dT primer, and 2 μl of the cDNA was subjected to PCR in a 50 μl reaction mixture.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Jimmy Kwang designed the study and critically reviewed the manuscript. Liqun Lu performed all the experiments and wrote the manuscript. Wang Hai helped perform in situ hybridization. Ivanus Manopo helped prepare shrimp primary cells and critically review the manuscript. Yu Li constructed and tested the plasmid containing WSSV ie1 promoter.
Acknowledgements
The authors would like to thank Dr. He Qigai for assistance in preparing the antibody and Dr. Beau James Fenner for reviewing the manuscript. This work was supported by Temasek Holdings Pte Ltd of Singapore.
==== Refs
Chou HY Huang CY Wang CH Chiang HC Lo CF Pathogenicity of a baculovirus infection causing white spot syndrome in cultured penaeid shrimp in Taiwan Dis Aquat Org 1995 23 165 173
van Hulten MCW Witteveldt J Peters S Kloosterboer N Tarchini R Fiers M Sandbrink H Klein Lankhorst R Vlak JM The white spot syndrome virus DNA genome sequence Virology 2001 286 7 22 11448154 10.1006/viro.2001.1002
Yang F He J Lin XH Li Q Pan D Zhang X Xu X Complete genome sequence of the shrimp white spot bacilliform virus J Virol 2001 75 11811 11820 11689662 10.1128/JVI.75.23.11811-11820.2001
Maeda M Saitoh H Mizuki E Itami T Ohba M Replication of white spot syndrome virus in ovarian primary cultures from the kuruma shrimp, Marsupenaeus japonicus J Virol Methods 2004 116 89 94 14715311 10.1016/j.jviromet.2003.10.013
Blissard GW Rohrmann GF Baculovirus diversity and molecular biology Annu Rev Entomol 1990 35 127 155 2154158 10.1146/annurev.en.35.010190.001015
Khadijah S Neo SY Hossain MS Miller LD Matharan S Kwang J Identification of white spot syndrome virus latency-related genes in specific-pathogen-free shrimps by use of a microarray J Virol 2003 77 10162 10167 12941929 10.1128/JVI.77.18.10162-10167.2003
Liu WJ Chang YS Wang CH Kou GH Lo CF Microarray and RT-PCR screening for the white spot syndrome virus immediately-early genes in cycloheximide-treated shrimp Virology 2005 334 327 341 15780883 10.1016/j.virol.2005.01.047
Lu L Kwang J Identification of a novel shrimp protein phosphatase and its association with latency-related ORF427 of white spot syndrome virus FEBS Lett 2004 577 141 146 15527775 10.1016/j.febslet.2004.08.087
Chen SN Wang CS Establishment of cell culture systems from penaeid shrimp and their susceptibility to white spot disease and yellow head viruses Methods cell sci 1999 21 199 206 10627672 10.1023/A:1009885929335
Leisy DJ Lewis TD Leong JC Rohrmann GF Transduction of cultured fish cells with recombinant baculoviruses J Gen Virol 2003 84 1173 1178 12692282 10.1099/vir.0.18861-0
Chen LL Wang HC Huang CJ Peng SE Chen YG Lin SJ Chen WY Dai CF Yu HT Wang CH Lo CF Kou GH Transcriptional analysis of the DNA polymerase gene of shrimp white spot syndrome virus Virology 2002 301 136 147 12359454 10.1006/viro.2002.1536
van Hulten MCW Westernberg M Goodall SD Vlak JM Identification of two major virion protein genes of white spot syndrome virus of shrimp Virology 2000 266 227 236 10639309 10.1006/viro.1999.0088
Halford WP Kemp CD Isler JA Davido DJ Schaffer PA ICP0, ICP4, or VP16 expressed from adenovirus vectors induces reactivation of latent Herpes Simplex virus type 1 in primary cultures of latently infected trigeminal ganglion cells J Virol 2001 75 6143 6153 11390616 10.1128/JVI.75.13.6143-6153.2001
Bolle LD Naesens L Clercq ED Update on human Herpesvirus 6 biology, clinical features, and therapy Clin Micro Rev 2005 18 217 245 15653828 10.1128/CMR.18.1.217-245.2005
Hummel M Abecassis MM A model for reactivation of CMV from latency J Clin Virol 2002 25 123 136 10.1016/S1386-6532(02)00088-4
Ruf IK Moghaddam A Wang F Sample J Mechanisms that regulate Epstein-Barr virus EBNA-1 gene transcription during restricted latency are conserved among lymphocryptoviruses of Old World primates J Virol 1999 73 1980 1989 9971778
Kondo K Shimada K Sashihara J Tanaka-Taya K Yamanishi K Identification of human herpesvirus 6 latency-associated transcripts J Virol 2002 76 4145 4151 11907257 10.1128/JVI.76.8.4145-4151.2002
Rotola A Ravaioli A Gonelli SD Cassai E Di Luca D U94 of human herpesvirus 6 is expressed in latently infected peripheral blood mononuclear cells and blocks viral gene expression in transformed lymphocytes in culture Proc Natl Acad Sci USA 1998 95 13911 13916 9811900 10.1073/pnas.95.23.13911
Ammosova T Jerebtsova M Beullens M Voloshin Y Ray PE Kumar A Bollen M Nekhai S Nuclear protein phosphatase-1 regulates HIV-1 transcription J Biol Chem 2003 278 32189 32194 12788939 10.1074/jbc.M300521200
Sedarati F Izumi KM Wagner EK Stevens JG Herpes simplex virus type 1latency-associated transcription plays no role in establishment or maintenance of a latent infection in murine sensory neurons J Virol 1989 63 4455 4458 2550678
Shi Z Wang H. Zhang J Xie Y Li L Chen X Edgerton BF Bonami JR Response of crayfish, procambarus clarkia, haemocytes infected by white spot syndrome virus J Fish Dis 2005 28 151 156 15752275 10.1111/j.1365-2761.2004.00607.x
Kost TA Condreay JP Recombinant baculoviruses as mammalian cell gene-delivery vectors Trends Biotech 2002 20 173 180 10.1016/S0167-7799(01)01911-4
Sambrook J Fritsch EF Maniatis T 2 Molecular cloning: a laboratory manual 1989 Cold Spring Harbor Laboratory, Cold Spring Harbor, NY
|
16115322
|
PMC1208962
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Aug 23; 2:71
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-71
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-721612022110.1186/1743-422X-2-72ResearchStrategically examining the full-genome of dengue virus type 3 in clinical isolates reveals its mutation spectra Chao Day-Yu [email protected] Chwan-Chuen [email protected] Wei-Kung [email protected] Wei-June [email protected] Hui-Lin [email protected] Gwong-Jen J [email protected] Institute of Epidemiology, College of Public Health, National Taiwan University (NTU), Taipei, Taiwan (100), Republic of China (R.O.C.)2 Institute of Microbiology, College of Medicine, NTU, Taipei, Taiwan (100), Republic of China (R.O.C.)3 Dept. of Parasitology, Chang Gung College of Medicine and Technology, Kwei-San, Tao-Yuan, Taiwan (100), Republic of China (R.O.C.)4 Hepatitis Research Center, NTU Hospital, Taipei, Taiwan (100), Republic of China (R.O.C.)5 Division of Vector-Borne Infectious Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention (CDC), Fort Collins, USA2005 24 8 2005 2 72 72 29 6 2005 24 8 2005 Copyright © 2005 Chao et al; licensee BioMed Central Ltd.2005Chao et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Previous studies presented the quasispecies spectrum of the envelope region of dengue virus type 3 (DENV-3) from either clinical specimens or field-caught mosquitoes. However, the extent of sequence variation among full genomic sequences of DENV within infected individuals remains largely unknown.
Results
Instead of arbitrarily choosing one genomic region in this study, the full genomic consensus sequences of six DENV-3 isolates were used to locate four genomic regions that had a higher potential of sequence heterogeneity at capsid-premembrane (C-prM), envelope (E), nonstructural protein 3 (NS3), and NS5. The extentof sequence heterogeneity revealed by clonal sequencing was genomic region-dependent, whereas the NS3 and NS5 had lower sequence heterogeneity than C-prM and E. Interestingly, the Phylogenetic Analysis by Maximum Likelihood program (PAML) analysis supported that the domain III of E region, the most heterogeneous region analyzed, was under the influence of positive selection.
Conclusion
This study confirmed previous reports that the most heterogeneous region of the dengue viral genome resided at the envelope region, of which the domain III was under positive selection pressure. Further studies will need to address the influence of these mutations on the overall fitness in different hosts (i.e., mosquito and human) during dengue viral transmission.
Quasispeciesmutation spectramicro-evolution of dengue virus serotype 3dengue hemorrhagic fever (DHF)sequence diversityTaiwan
==== Body
Background
Dengue viruses (DENV), which consisted of four antigenically distinct serotypes (DENV-1, 2, 3 and 4), are the most important arthropod-borne viruses affecting humans. After infection, it may result in dengue fever (DF), dengue haemorrhagic fever (DHF), dengue shock syndrome (DSS) or death [1,2]. It is estimated that close to 50–100 million cases of DF and 30,000 fatal cases of DHF/DSS occur annually in tropical and subtropical regions. With the increased numbers of dengue patients, it is indicated the global expansion of epidemic areas, and increased frequencies of severe DHF/DSS and case fatality [3]. Considerable efforts have been devoted to developing vaccines to prevent dengue, but the success of the vaccines will be dependent on the vaccine strain chosen to direct against the diversity and evolution of DENV genome.
DENV belongs to the genus Flavivirus, family Flaviviridae, possessing a positive-sense, single-stranded RNA genome, which is approximately 10,700 bases in length and contains a single open reading frame [4]. A single polyprotein translated from the viral RNA is cleaved into 3 structural proteins [capsid (C), premembrane (prM) and envelope (E) protein] and 7 nonstructural proteins (NS), with the gene order as 5'-C-prM/M-E-NS1-NS2A-NS2B-NS3-NS4A-NS4B-NS5-3'. Like many RNA viruses, the genomic sequence of a single DENV isolate exists in nature as a collection of highly similar but not identical variants known as quasispecies due to its high average mutation rate of 10-3 to 10-5substitution per nucleotide copied and per round of replication [5,6]. Previous studies using a clonal sequencing approach amplified viral RNA directly from DENV-3 infected patients' plasma and the extent of sequence heterogeneity in the envelope region with mean pairwise difference ranging from 0.21 to 1.67% have been observed [7]. There are obvious reasons for selecting the E gene region for this study, mainly due to its important biological functions such as receptor-mediated endocytosis, virus-induced cellular tropism and eliciting neutralization antibodies. However, one cannot exclude the biological significance of the sequence heterogeneity in other genomic regions including non-structural (NS) proteins, 5' and/or 3' non-coding regions (NCR). The well-studied example of hepatitis C virus (HCV) demonstrated that the quasispecies dynamics and composition of the NS5A region may play a role in disease prognosis and in response to interferon and ribavirin therapy [8]. Although the previous attempt to correlate the sequence heterogeneity of the capsid gene with NS protein 2B gene region of DENV-3 has observed very similar sequence heterogeneity with mean pairwise p-distance 0.12–1.2% [9], the extent of sequence variation among full genomic sequences of DENV within infected individuals remains largely unknown. Thus, it is important to address whether the evidence of different evolutionary processes, such as adaptive evolution, shape the population genetics of DENV at specific genomic regions other than the E region.
An outbreak of DHF, attributed to genotype 2 of DENV-3, resulted in 111 DF and 23 DHF cases in Tainan (southern Taiwan) from October 1998 to January of 1999 [10]. DENV-3 was the only serotype isolated during this outbreak, and the seroepidemological study clearly demonstrated that DHF cases were not associated with secondary DENV infection [10]. Here we report the selection of the most prominent variable regions identified by the full-genomic sequencing of DENV isolates from six clinical patients during this outbreak. The application of the clonal sequencing of those variable regions enabled us to study quasispecies structure of DENV isolates and to provide a better understanding of the changes in mutation spectrum at the clonal level and virus evolution.
Results
Heterogeneous regions identified at full genomic scale of DENV-3
To identify the potential heterogeneous regions of DENV-3 in the whole-genomic scale, acute-phase plasma samples were obtained from six dengue patients, including three DF (designated 1F, 2F and 3F) and three DHF patients (designated 1H, 2H and 3H). The sequencing strategy is depicted in Fig 1. These patients' plasma samples were used to infect the C6/36 mosquito cell line to obtain sufficient viral genomic RNA for full-genomic consensus sequencing for the identification of regions with sequence heterogeneity for follow-up clonal sequencing.
Figure 1 Strategy in clonal-sequencing the whole genome of genomic RNA of DENV-3.
The consensus sequence similarity of these six viruses was as high as 99.73%. The 2H and 3H virus each had two silent changes at nucleotide positions of 808 (G to A), 9979 (T to C), 4204 (C to T) and 8785 (T to C), respectively (Table 1). There were no consistent nucleotide changes that might correlate with disease severity among paired viruses using this consensus sequencing approach. However, the potential heterogeneous sequence regions were clearly observed and identified by close examination of the overlapping chromatogram files using the SeqMan program in the Lasergene software package (DNASTAR inc., Madison, WI). Special attention was paid to identify the regions which consistently presented mixed-chromatographic peaks in the respective trace files obtained from at least two independent sequencing primers. These potential heterogeneous regions, located at C/PrM, E, NS3 and NS5 genes (Table 1), were selected for the clonal sequencing analysis. Five genomic fragments were amplified directly from six patients' viremic plasma by five flanking primer pairs (Table 4) at nucleotide position of 1–764 (5'NCR/C/prM), 1259–2550 (E/NS1), 5443–6337 (NS3) and 8501–10316 (NS5/3'NCR) using Titan™ one tube RT-PCR System (Boehringer Mannheim). After excluding the primer sequences, the C/PrM region was 752 nucleotides in length with 225 amino acids in the coding region; the E/NS1 region was 1239 nucleotides in length covering 413 amino acids which included 40 amino acids at the N terminal end of NS1 protein, the NS3 region was 866 nucleotides in length covering 288 amino acids, and the NS5 region was 1791 nucleotides in length with 586 amino acids in viral coding sequences.
Table 1 Identification of the positions of potential heterogeneity nucleotide sequence by the full genome consensus sequence of DENV-3a viruses isolated during 1998–1999 dengue outbreak in Taiwan.
Virus ID Disease Statusc Nucleotide Changes at Positions Indicatedb
320–322 444–445 808 1693 1716 4204 5322 6045 6079 8785 9076 9979 10105 10128
1F DF RRR TG (CA) G G(C) C(T) C T A C T T T C C
1H DHF GGG TG G C C(T) C T A C T T T C C
2H DHF GGG TG A G C(T) C T A C T T(A) C T C
2F DF GGG TG G G C(T) C T A C T T T C C
3H DHF GGG TG G G(C) C(T) T T(C) A(C) C(T) C T T C C
3F DF GGG TG G G C(T) C T A C T T T C C(T)
aNucleotide in parentheses indicated "mix nucleotide sequence" based on mix chromatographic signals in the sequencing trace file and nucleotide position number referred to the reference strain H87 of DENV-3 (genebank accession number: M93130)
b DENV-3 viruses were isolated from the plasma of six dengue patients by one passage in the C6/36 mosquito cell culture.
c Disease status was classified based on WHO criteria [30]. DF: dengue fever; DHF: dengue hemorrhagic fever.
Table 4 The Oligonucleotide primers and conditions used for RT-PCR of full-length genome of DENV-3
PCR Primera Sequence (5' → 3') Genome Positionb Size(nt)c
P1A AGT TGT TAG TCT RCG TGG 1–18 1181
CP1181B TCC ARG CAC CTT CAG ATG 1181–1199
DC530A AAC AWR TGC ACC CTC 540–555 1164
CDC1694B TGC ATK GCT CCT TCT TGR 1694–1712
P1259A GGC AAG GGA AGC TTG GTG ACA TGC GC 1259–1285 1244
CDC2503B GGG AGT CTG CTT GGA ATT 2503–2521
DC2171A GCC ATT CTR GGW GAC ACC GCY TGG GA 2171–2197 1246
CDC3417B TCT CTT CTT TGT CMT TCA 3417–3435
D3-3142A CCA AAG AGT CTA GCT GGT CC 3142–3162 1535
D3-4677B CAT TGT GCG TCA ACA CTG CC 4677–4697
d3NS2B1A AGC TGG CCA CTG AAT GAG G 4124–4143 1562
D35686B CAA AAG TCT TCC TAC TAA GTT G 5686–5708
D3-5443A GCC GCA ATT TTC ATG ACA 5443–5461 2034
D3-7477B AAC AGC TAT CGT GGT GTT CC 7477–7497
d37246A AAG AAT CCA ACG GTG GAT GG 7246–7266 1454
d38750B TCC CTT GTG CAT AAT CTG GG 8750–8770
d38501A CAG GCT CAG CCT CCT CC 8501–8518 1654
d310316B GCT TCT TCC GTA CTG TGG C 10316–10335
d39991A CTT ACT GTC TGG AAC AGG G 9991–10010 648
d310688B GTT GAT TCA ACA GCA CCA TTC 10688–10709
a Primer names with A in the end indicate a viral-sense orientation; names with B in the end indicate a complementary sense orientation
b Genome positions are given according to the published sequence of strain H87 of dengue virus serotype 3
c nt indicated nucleotide
Clonal sequencing of the heterogeneous regions among dengue viral genomes
The pCRII-TOPO™ T/A cloning kit (Invitrogen, San Diego, CA) was used to clone PCR products representing heterogeneous sequence regions identified by the consensus sequencing as described previously [7]. At least 20 to 30 clones containing the PCR amplicons from four heterogeneous gene regions (C-PrM, E, NS3 and NS5) were sequenced, aligned, and analyzed using the program GCG and MEGA v3.0 [11]. In general, the transitional substitutions were higher than transversional substitutions for all samples analyzed. The transitional changes (A to G or T to C) constituted overall substitution rate of 72.8 ± 5.1%, 73.7 ± 11% at C/PrM, E of structural proteins, and the NS3 and NS5 regions had relatively lower such changes of 63.2 ± 7.5%, 44.7 ± 18%. The lowest nucleotide mutation frequencies were observed in NS3 region with a mean ± standard deviation (SD) of 0.6 ± 0.3 × 10-3 for all clones analyzed, followed by C/PrM (1.2 ± 0.15 × 10-3), NS5 (1.5 ± 0.4 × 10-3) and envelope region (1.8 ± 0.8 × 10-3), which was statistically significant (p < 0.01). Similarly, the substitution frequencies of amino acids were also variable among the viral genome regions with the lowest frequency observed in NS3 (1.3 ± 0.4 × 10-3), followed by C/PrM (2.3 ± 0.3 × 10-3), E (3.1 ± 1.8 × 10-3) and NS5 region (3.3 ± 0.8 × 10-3) (Table 2).
Table 2 Sequence diversity (mean p-distance) among different genomic regions of DENV-3
Nucleotide Amino acid
Virus ID No. Region No of Clones No of change/total Mutation frequencya (10-3) A→G or U→Cb (%) Mean p-distancec(10-3) Range (10-3) No of change/total Mutation frequencya(10-3) Mean p-distancec(10-3) Range (10-3)
1F C/PrM 23 24/17733 1.4 66.7 2.67 0–6.7 13/5175 2.5 5 0–8.9
1H C/PrM 29 29/22359 1.3 69 2.45 0–6.7 17/6525 2.6 5.12 0–22.2
2H C/PrM 21 16/16191 1.0 81.3 2.03 0–6.7 9/4725 1.9 3.81 0–13.3
2F C/PrM 22 18/16962 1.1 72.2 2.17 0–6.7 10/4950 2.0 4 0–13.3
3H C/PrM 13 11/10023 1.1 72.7 1.81 0–4 7/2925 2.4 4.7 0–17.8
3F C/PrM 24 24/18504 1.3 75 2.55 0–8 13/5400 2.4 4.8 0–17.8
Mean ± std C/PrM 22 1.2 ± 0.15 72.8 ± 5.1 2.28 ± 0.3 2.3 ± 0.3 4.6 ± 0.5
1F E/NS1 26 40/32240 1.2 72.5 2.23 0–6.5 21/10192 2.1 4.3 0–10.2
1H E/NS1 13 47/16250 2.9 61.7 3.82 0–6.4 17/5096 3.3 6.6 0–7.6
2H E/NS1 25 33/16250 2.0 93.9 3.88 0–5.6 34/9800 3.5 6.4 0–12.6
2F E/NS1 13 15/22103 0.7 66.7 3.18 0–2.8 4/5096 0.8 7.6 0–9
3H E/NS1 23 44/20240 2.2 72.7 3.83 0–4.8 55/9016 6.1 6.2 0–12.6
3F E/NS1 20 39/24960 1.6 74.4 2.56 0–6.8 20/7840 2.6 4.8 0–21.7
Mean ± std E/NS1 20 1.8 ± 0.8 73.7 ± 11 3.7 ± 0.7 3.1 ± 1.8 6 ± 1.2
1F NS3 19 5/17005 0.3 60 0.9 0–3.5 8/4978 1.6 3.3 0–11.4
1H NS3 27 16/24165 0.7 56.3 1.2 0–4.6 11/7074 1.6 3.1 0–11.4
2H NS3 26 9/23270 0.4 33.3 1.2 0–4.6 6/6812 0.9 3.2 0–11.5
2F NS3 25 16/22375 0.7 40 2.4 0–5.8 12/6550 1.8 5.7 0–15.3
3H NS3 23 25/20585 1.2 16 1.6 0–4.6 6/6026 1.0 2.0 0–11.4
3F NS3 18 8/16110 0.5 62.5 1.0 0–4.6 4/4716 0.8 1.6 0–11.4
Mean ± std 23 0.6 ± 0.3 44.7 ± 18 1.4 ± 0.6 1.3 ± 0.4 3 ± 1.4
1F NS5 13 29/23335 1.2 69 1.7 0–4.5 17/7748 2.2 4.1 0–8.4
1H NS5 17 32/30515 1.0 71.9 2.4 0–5.6 31/10132 3.1 3.5 0–8.4
2H NS5 18 51/32310 1.6 64.7 3.0 0–8.9 33/10728 3.1 6.1 0–18.5
2F NS5 25 60/44875 1.3 65 2.5 0–5.6 47/14900 3.2 3.6 0–8.4
3H NS5 16 64/28720 2.2 56.3 4.3 0–7.3 46/9536 4.8 6.8 0–15.2
3F NS5 26 69/46670 1.5 52.2 3.7 0–6.7 52/15496 3.4 6.5 0–11.8
Mean ± std 19 1.5 ± 0.4 63.2 ± 7.5 3.1 ± 0.9 3.3 ± 0.8 5.1 ± 1.5
aMutation frequency is defined as the proportion of mutations relative to the consensus nucleotide or amino acid sequence for each patient and calculated by dividing the number of mutations relative to consensus by the total number of nucleotides or amino acid sequenced in each sample.
b Percentage of A→G or U→C transitional mutations
c p-distance is calculated by pairwise comparison of nucleotide or amino acid sequences between clones by the program MEGA d Indicated the virus derived from mosquito inoculation by C6/36-passaged one virus of patient ID#1H.
The mean pairwise p-distance as described in the previous study [7] was employed to compare the extent of sequence variation among different viral genome regions. Consistently, NS3 had the lowest pairwise p-distance among NS5, C/PrM or E protein. The average mean p-distance in nucleotides and SD for NS3, C-PrM, NS5, and E were 1.4 ± 0.6 × 10-3, 2.3 ± 0.3 × 10-3, 3.1 ± 0.9 × 10-3, and 3.7 ± 0.7 × 10-3, respectively. At the amino acid level, NS3 also had the lowest mean p-distance (3 ± 1.4 × 10-3) and E proteins had the highest variability (6 ± 1.2 × 10-3) (Table 2). The difference of mean p-distance in nucleotides or amino acids among different genes was statistically significant (p < 0.01). No consistent correlation between any two different genes from the same human isolates with the extent of the nucleotide heterogeneity could be made. This would suggest that different genes are governed by different mutation rates, which resulted in different sequence (quasispecies) spaces/sizes in different gene regions.
Different selection pressures on different domains of E gene of DENV-3
Our previous analysis of the E gene of DENV-3 covered only 131 amino acids [7]. The PCR amplification by primer pair p1259A and cdc2503B in this study covered 1239 nucleotides encoding 413 amino acids, including portion of domain I and II, 3 hinge regions, and complete domain III to the end of the stem-anchor region [12]. However, genetic instability was observed when the PCR product was cloned into the T/A vector and propagated in E. coli. The genetic truncation occurred consistently at the location following amino acid position 412 of the envelope gene (E412). This truncation was observed in 29 clones (21.5%) out of 135 clones sequenced. In order to increase the sample size and to investigate the extent of amino acid substitution in the E protein, the deduced amino acid sequences of all 135 clonally obtained sequences from patients' viremic plasma were aligned and trimmed so that it contained 293 amino acids, ranging from E118 to E412, which include portions of domain I and II, the complete domain III and a portion of stem-anchor region for analysis (Table 3). Consistently, there was a higher mean amino acid p-distance in dengue hemorrhagic patients (1H: 0.008 ± 0.002, 2H: 0.012 ± 0.002, 3H: 0.009 ± 0.002) than in dengue fever patients (1F: 0.006 ± 0.001, 2F: 0.007 ± 0.002, 3F: 0.007 ± 0.001) with statistical significance (p < 0.05).
Table 3 Mean p-distance and ratio of dN to dS per site of amino acid among different domains of the E protein in DENV-3 infected patients
Virus ID No of sequences Envelope (293aa) Domain I (70aa) Domain II (106aa) Domain III (100aa)
Mean p-distance dN/dS Mean p-distance dN/dS Mean p-distance dN/dS Mean p-distance dN/dS
1F 26 0.006 ± 0.001 2.04 0.009 ± 0.003 1.82 0.005 ± 0.002 1.72 0.005 ± 0.002 2.8
1H 21 0.008 ± 0.002 1.11 0.008 ± 0.004 2.53 0.007 ± 0.003 0.59 0.007 ± 0.002 1.2
2F 18 0.007 ± 0.002 1.38 0.009 ± 0.004 2.52 0.006 ± 0.003 1.04 0.007 ± 0.003 2.57
2H 25 0.012 ± 0.002 1.06 0.013 ± 0.004 2.68 0.012 ± 0.004 1.08 0.011 ± 0.003 0.77
3F 23 0.007 ± 0.001 0.81 0.012 ± 0.004 0.96 0.005 ± 0.002 0.45 0.003 ± 0.002 1.76
3H 23 0.009 ± 0.002 1.46 0.014 ± 0.005 4.19 0.006 ± 0.003 0.62 0.005 ± 0.002 120.03
Average 22.7 0.008 ± 0.002 1.31 0.011 ± 0.002 2.45 0.007 ± 0.002 0.92 0.006 ± 0.002 21.52
Since the E protein is the major determinant of viral entry, cellular tropism and the target of both humoral and cellular immune selection [13,14], amino acid changes associated with particular site changes were further investigated using Phylogenetic Analysis by Maximum Likelihood program (PMAL) [15]. The non-synonymous (dN) to synonymous (dS) substitution ratio, referred to as parameter omega (ω) in the model, was calculated with the CODEML program from the PAML package, which analyzed and compared the ω ratios codon-by-codon using the maximum likelihood ratio test among three domains [16]. In this study, the M3 model of codon evolution was used since it often provides the best evidence for positive selection. Although some variations were observed, in general, domains III and I were influenced by positive selection as indicated by the dN/dS ratio larger than 1, but domain II was influenced by neutral selection, as the dN/dS ratio was smaller than 1. The average value of dN/dS was the highest in Domain III (21.52), followed by Domain I (2.45), then Domain II (0.92) (Table 3).
Phylogenetic analysis of DEN-3 Virus
To determine the evolutionary history of the DENV-3 viruses found in Taiwan in 1998, the nucleotide sequence of their partial E protein genes were compared with those from all previous published DENV-3 E gene sequence available in the GenBank. The phylogenetic tree analysis for 141 clonal sequences from six virus isolates of this study and 24 global DENV-3 sequences separated the viruses into five main subgroups, which had been previously defined as five different genotypes. As in previous studies of DENV-3 diversity, the 1963 Puerto Rico strain formed a distinct outlier, which served as the outgroup for the phylogenetic tree. The tree topology was very similar based on either neighbor-joining (NJ) or parsimony (PAR) method. Based on the phylogenetic tree, the virus isolates from Taiwan in 1998 formed a tight cluster with strong bootstrap support, which fell closer to the isolates from Thailand as they belong to DEN-3 genotype II, according to the classification of Lanciotti et al (20)(Fig 2). Most of the population from the clonal sequences formed a tightly cluster, which represented the highly homogeneous nucleotide sequences during the same epidemic. Interestingly, some clones from different individual isolates appeared to form the different subgroups under the Thailand genotype, with 50–100% bootstrap support. This indicated viral evolution did occur during the epidemic period, probably under selection pressure.
Figure 2 Phylegenetic tree showing the evolutionary relationships of the E gene among 54 sequences from 30 clonal sequences of 6 DEN-3 clinical isolates and 24 global isolates. Bootstrap support values presented as percentage are given for key nodes only and the genotype designations are given. The horizontal branch length of the trees was drawn to scale. GenBank accession numbers of the global DEN-3 strains used in this analysis are as follows: Fiji92 (L11422), India84 (L11424), Indonesia73 (L11425), Indonesia78 (L11426), Indonesia85 (L11428), Malaysia74 (L11429), Malaysia81 (L11427), Mozambique85 (L11430), H87 (L11423), Philippines83 (L11432), Puerto Rico77 (L11434), Puerto Rico63 (L11433), Samoa86 (L11435), SriLanka81 (L11431), Sri Lanka85 (L11436), Sri Lanka89 (L11437), Tahiti65 (L11439), Tahiti89 (L11619), Thailand62 (L11440), Thailand73 (L11620), Thailand87 (L11442).
Discussion and Conclusion
To the best of our knowledge, this was the first systematic attempt to understand the sequence spectrum of the entire genome of DENV-3. Previous studies, focused on certain genomic regions such as the envelope gene, the capsid gene or the NS2B gene, have revealed the presence of quasispecies structure as indicated by the simultaneous presence of multiple variant genomic sequences of the dengue virus isolates from either the clinical samples or field-caught mosquitoes [7,9,17-19]. Instead of arbitrarily choosing one genomic region in this study, the full genomic consensus sequences of six DENV-3 isolates were used to locate the four most prominent heterogeneous regions, the C/PrM and E in the structure, and NS3 and NS5 in the nonstructural regions.
Use of clonal sequencing to study the mutation spectrum needs to ensure that sequencing artifacts due to RT-PCR amplification are reduced to minimum. In this study, we used viral RNAs extracted from patients' viremic plasma directly. In addition, a thermostable polymerase with proof-reading function was incorporated in the RT-PCR, which has been shown to be a simple and valuable method for characterization of mutant spectra of virus quasispecies [20]. The nucleotide changes from four sequenced-viral genomic regions (range of 4.4–11.6 × 10-5 changes/nucleotide/cycle of PCR) were greater than those predicted based on reverse transcriptase (10-4) and proof-reading DNA polymerase (Pfu, 10-6 error/site/cycle) combined [21]. Based on the experimental data, Arias et al pointed out that the biological and molecular clones were statistically indistinguishable when defining the mutation spectrum with regard to the types and distributions of mutations, mutational hot-spots and mutation frequencies [20]. Similarly, we believe that the full-genomic characterization followed by clonal sequencing procedure employed in this study is a reasonable and justifiable approach for the characterization of mutation spectra (quasispecies dynamic) of DENV-3 viruses.
DENV, like other RNA viruses, exists as quasispecies with the sequence diversity of the envelope gene in the DENV-3 virus population from 6 clinical isolates, ranging from 0.22–0.39% of mean p-distances in this study. These values are within the range calculated by other studies (0.12 to 0.84%) for different portions of the E protein genes of DENV-3 viruses from either the clinical or field-caught mosquito isolates [7,9,17-19]. Our study confirmed use of the structural protein, especially the E gene with higher sequence heterogeneity to study the viral quasispecies, instead of NS protein and 5' and 3' NCR. The extent of sequence variation observed in this study was similar to or lower than what has been reported for acute infection of HIV-1 or HCV [8,22-25]. A study of sequence variation of HIV-1 after sexual transmission revealed that the nucleotide mean diversity of the E gene (gp120) was 0.24% and that of the gag gene (p17) was 0.5% [22]. Similar results by studying variants of hepatitis C virus (HCV) from a single infected blood donor and 13 viraemic recipients were traced to examine the sequence diversity in hypervariable region 1 with sequence p-distance ranged from 0.3% to 6.2% [23]. These data might support an important concept in the evolution of arthropod-borne RNA viruses (arboviruses) which evolve more slowly than RNA viruses transmitted by other routes due to intrinsic constraints associated with dual replication in mammalian and invertebrate hosts [26]. Consistent with this interpretation was that the lower sequence diversity was observed at the same E protein gene from the field-caught mosquito DENV-3 isolates [19] or after inoculation of clinical serum of DENV-3 into mosquitoes (data not shown).
The larger mutation spectra in structural proteins than non-structural proteins probably imply less genetic constraint on the structure proteins to maintain proper function than non-structural proteins. However, the mutations in the structural or non-structural proteins did not accumulate randomly during replication. The mutation rates vary in different functional/structural domains. Even within the envelope structure protein, where domain III, the proposed receptor-binding and neutralizing antibody-binding sites [13] had highest sequence heterogeneity than Domain I or II. The detail analysis in our study further indicated that the different selection pressure was exerted on different domain of the E gene of DENV-3. Domain III and domain I were under the influence of positive selection (dN/dS:21.52, 2.45) and domain II was under the influence of neutral selection (dN/dS:0.92). The particularly higher dN/dS ratio in domain III of viral isolates 3H was caused by the value of 0 of dS at the denominator. The positive selection on the domain III is not surprising since domain III contains the receptor-binding domain and major type-specific neutralization epitopes [12]. However, the complete E gene sequence may be required to clarify the evolutionary selection on domain I and II due to incomplete sequence obtained in this study.
In contrast to other studies which suggested the strong purifying selection in the E gene of dengue virus evolution, the consensus sequences used for analysis represented dengue viral gene conservation during long-term evolution [27]. The clonal sequences obtained from our study represented the selection pressure imposed on viral populations during the short term of evolution, which might explain the substantially different dN/dS value within hosts and among genotypes. The majority of the nonsynonymous mutations that arise within each host occurred as singletons with relatively low frequency in the population; thus are likely to be deleterious. Such heterogeneous gene pool may give rise to various viruses able to occupy new ecological niches or to adapt to sudden selection pressures on the cycle of replication. It is evident that certain nonsynonymous nucleotide mutations at specific sites repeatedly occurred among different virus isolates as well as after mosquito inoculation in our study (data not shown), which has been proposed as quasispecies memory in another study [28]. Further studies are needed to address the influence of these mutations on the overall fitness in different hosts (i.e., mosquito and human) during dengue viral transmission.
Materials and methods
Study subjects and virus isolation
Six dengue patients were identified by RT-PCR to be DENV-3 positive during the 1998 epidemic and their acute-phase viremic plasma samples were collected within seven days following the onset of fever. These plasma samples were used to infect C6/36 Aedes albopictus mosquito cell lines as described previously [29]. The study protocol was approved by the College of Public Health Research Ethics Review Committee at the National Taiwan University with the informed consent obtained from six dengue patients. Six adult dengue cases between 38 and 63 years of age, including one DF (F) and one DHF (H) cases, whose disease status were classified based on WHO criteria [30], were represented as 1F, 1H, 2F, 2H, 3F and 3H, respectively.
DENV-3 was confirmed by indirect immuno-fluorescent antibody (IFA) tests using serotype-specific monoclonal antibodies (DENV-1:H47, DENV-2:H46, DENV-3:H49, DENV-4:H48) [31]. The C6/36-passage one viral stock was used for full genomic consensus sequencing to identify regions with sequence heterogeneity for clonal sequencing as described later.
Preparation of viral RNA, RT-PCR amplification and consensus sequencing of PCR products
Viral RNA was extracted either from viremic plasma specimens or from the C6/36-passaged one cell culture fluids using QIAamp viral RNA mini kit (Qiagen, Germany) by following the manufacturer's protocol. The eluted RNA was used as the template and overlapping regions of DENV-3 genome amplified by Titan™ one tube RT-PCR System (Boehringer Mannheim, Germany) following the manufacturer's suggestions. The oligonucleotide primer pairs were designed based on published full-length DENV-3 sequence data for the strains of H87 and 80-2 (GenBank Accession number M93130 and AF317645) and the unpublished DENV-3 sequences (Chang, G-J. personal communication). Ten overlapping fragments were generated which spanned genomic regions of DENV-3 at the following nucleotide (nt) positions: 1 to 1199, 540 to 1712, 1259 to 2521, 2171 to 3435, 3142 to 4697, 4124 to 5708, 5443 to 7497, 7246 to 8770, 8501 to 10335, 9991 to 10709. Primer sequences used for PCR amplification were summarized in Table 4. The obtained PCR products were sequenced by using the Big Dye Terminator Sequencing kit (Perkin-Elmer, Applied Biosystems, Foster City, CA) and analyzed by the 3100 automate sequencer (Perkin-Elmer, Applied Biosystems) with a short capillary.
Preparation of plasmid templates for clonal sequencing
We used pCRII-TOPO™ T/A cloning kit (Invitrogen, San Diego, CA) to clone PCR products representing heterogeneity sequence regions identified by the consensus sequencing protocol at the previous section. The T/A vector ligated PCR product was used to transform Escherichia coli TOP10 competent cells (Invitrogen) and at least 30 white colonies were picked, to grow in 3 ml LB broth at 37°C overnight. Plasmid DNAs were extracted by the QIAprep Spin Miniprep kit (Qiagen), and each plasmid DNA with the desired inserts was completely sequenced using insert flanking primers, T7 and cSP6.
Nucleotide and Amino acid sequence analysis
Overlapping chromatogram files retrieved from the automate sequencer were analyzed and edited using the SeqMan program in the Lasergene software package (DNASTAR inc., Madison, WI). The derived consensus sequences after excluding the sequences of amplifying primers were aligned using GCG package (Genetic Computer Group, WI). For full-length genomic sequences we paid special attention to identify the regions which consistently presented mixed-chromatographic peaks in the trace file obtained from at least two independent sequencing primers. These regions were selected for the follow-up clonal sequence analysis. Pairwise comparisons of both nucleotide and amino acid sequences between isolates and clonal sequences were performed using the program MEGA v3.0 (Molecular Evolutionary Genetics Analysis, Pennsylvania State University, PA) to determine the numbers of transition and transversion changes, and the mean and proportion of difference, Hamming distance and p-distance, as described previously [8,32,33]. Synonymous (dS) and nonsynonymous (dN) distances relative to the consensus sequences were calculated within each isolate by maximum likelihood ratio method in the CODEML program from the PAML package [15]. Instead of assuming that all sites are under the same selection pressure with the same underlying dN/dS ratio, it allows variable selection intensity to vary among amino acid sites [34,35]. In this study, M3 model of codon evolution was applied for which often provides the best evidence for positive selection [16]. An excess of nonsynonymous substitutions over synonymous substitutions (ie. the ratio of dN/dS > 1) is an indicator of positive natural selection at the molecular level.
The results were expressed as the mean ± standard deviation (SD). T-tests were performed on two-sampled tests and a one-way ANOVA was performed to compare data from different genomic regions, family clusters or different domains in envelope region. In all tests, a p-value less than 0.05 was considered statistically significant.
Evolutionary analysis
The nucleotide sequences generated in this study were combined with those of all other DENV-3 E protein gene sequences available on GenBank, which resulted in a total data set of 154 sequences. Phylogenetic trees were estimated using parsimony method available in the Phylip v3.6 package [36]. Bootstrap resampling analysis of 500 replicates was generated with the SEQBOOT program to prove the stability of the trees. Phylogenetic trees were delineated using the TreeView (v.1.6.6) program by using Puerto Rico 1963 isolate as the outgroup. For better presentation of the phylogenetic tree, only 30 clonal sequences from six different clinical isolates and 24 global isolates were shown in Fig 2.
Nucleotide sequences accession numbers
The sequences from four heterogeneous regions of dengue viruses from the six patients studied here have all been submitted to GenBank, and their accession numbers are from DQ109039 to DQ109173 for the E region, from DQ109174 to DQ109305 for the capsid/prM region, from DQ109306 to DQ109405 for the NS5 region and from DQ109406 to DQ109524 for the NS3 region.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
DYC designed and performed all the experiments and helped drafted this manuscript. CCK helped with collecting field isolates and instructed the experiments, together with WKW and HLW. WJC helped for the mosquito injection experiments and GJC formulated the idea for this study and also provided critical comments regarding this manuscript.
Financial support
The study was supported by the grants from the National Health Research Institute (NHRI), Taipei, Taiwan (grant number: NHRI#DD01-861X-CR-501P and NHRI#CN-CL8903P) and International Society of Infectious Disease (ISID).
Acknowledgements
We sincerely thank Shih-Ting Ho at the Sin-Lau Christian Hospital, Chien-Ming Li at the Chi-Mei Foundation Medical Center and Shih-Chung Lin at the Kuo General Hospital for their enthusiasm for kindly providing the clinical samples. This study was supported by the grants from the National Health Research Institute (NHRI), Taipei, Taiwan (NHRI#DD01-861X-CR-501P and NHRI#CN-CL8903P) and the training grant to D.-Y. Chao from International Society of Infectious Disease (ISID).
==== Refs
Gubler DJ Dengue and dengue hemorrhagic fever Clin Microbiol Rev 1998 11 480 496 9665979
Halstead SB The XXth century dengue pandemic: need for surveillance and research Wld Hlth Statist Quart 1992 45 292 296
McBride WJH Bielefeldr-Ohmann H Dengue viral infections; pathogenesis and epidemiology Microbes Infect 2000 2 1041 1050 10967284 10.1016/S1286-4579(00)01258-2
Chambers TJ Hahn CS Galler R Rice CM Flavivirus genome: organization, expression and replication Annu Rev Microbiol 1990 44 649 688 2174669 10.1146/annurev.mi.44.100190.003245
Twiddy SS Holmes EC Rambaut A Inferring the rate and time-scale of dengue virus evolution Mol Biol Evol 2003 20 122 129 12519914 10.1093/molbev/msg010
Domingo E Quasispecies theory in virology J Virol 2002 76 463 465 10.1128/JVI.76.1.463-465.2002
Wang WK Lin SR Lee CM King CC Chang SC Dengue type 3 virus in plasma is a population of closely related genomes:quasispecies J Virol 2002 76 4662 4665 11932434 10.1128/JVI.76.9.4662-4665.2002
Farci P Shimoda A Coiana A Diaz G Peddis G Melpolder JC Strazzera A Chien DY Munoz SJ Balestrieri A Purcell RH Alter HJ The outcome of acute hepatitis C predicted by the evolution of the viral quasispecies Science 2000 288 339 344 10764648 10.1126/science.288.5464.339
Wang WK Sung TL Lee CN Lin TY King CC Sequence diversity of the capsid gene and the nonstructural gene NS2B of dengue-3 virus in vivo Virology 2002 303 181 191 12482670 10.1006/viro.2002.1635
Chao DY Lin TH Hwang KP Huang JH Liu CC King CC 1998 dengue hemorrhagic fever epidemic in Taiwan. Emerg Infect Dis 2004 10 552 554 15116715
Kumar S Tamura K Nei N MEGA3: Integrated software for Molecular Evolutionary Genetics Analysis and sequence alignment. Brief Bioinform 2004 5 150 163 15260895
Modis Y Ogata S Clements D Harrison SC Variable surface epitopes in the crystal structure of dengue virus type 3 envelope glycoprotein. J Virol 2005 79 1223 1231 15613349 10.1128/JVI.79.2.1223-1231.2005
Rey FA Heinz FX Mandl C Kunz C Harrison SC The envelope glycoprotein from tick-borne encephalitis virus at 2 A resolution Nature 1995 375 291 298 7753193 10.1038/375291a0
Modis Y Ogata S Clements D Harrison SC A ligand-binding pocket in the dengue virus envelope glycoprotein Proc Natl Acad Sci 2003 100 6986 6991 12759475 10.1073/pnas.0832193100
Yang Z PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 1997 13 555 556 9367129
Yang Z Nielsen R Goldman N Pedersen AMK Codon-substitution models for heterogeneous selection pressure at amino acid sites Genetics 2000 155 431 449 10790415
Craig S Thu HM Lowry K Wang X Holmes EC Aaskov J Diverse dengue type 2 virus populations contain recombinant and both parental viruses in a single mosquito host J Virol 2003 77 4463 4467 12634407 10.1128/JVI.77.7.4463-4467.2003
Wittke V Robb TE Thu HM Nisalak A Nimmannitya S Kalayanrooj S Vaughn DW Endy TP Holmes EC Aaskov J Extinction and rapid emergence of strains of dengue 3 virus during an interepidemic period Virology 2002 301 148 156 12359455 10.1006/viro.2002.1549
Lin SR Hsieh SC Yueh YY TH L Chao DY Chen WJ King CC Wang WK Study of sequence variation of dengue type 3 virus in naturally infected mosquitoes and human hosts: implications for transmission and evolution. J Virol 2004 78 12717 12721 15507664 10.1128/JVI.78.22.12717-12721.2004
Arias A Lazaro E Escarmis C Domingo E Molecular intermediates of fitness gain of an RNA virus: characterization of a mutant spectrum by biological and molecular cloning J Gen Virol 2001 82 1049 1060 11297679
Mullan B Sheehy P Shanahan F Fanning L Do Taq-generated RT-PCR products from RNA viruses accurately reflect viral genetic heterogeneity? J Viral Hepat 2004 11 108 114 14996344 10.1046/j.1365-2893.2003.00455.x
Zhu T Mo H N W D.S. N Cao Y Koup RA Ho DD Genotypic and phenotypic characterization of HIV-1 in patients with primary infection Science 1993 261 1173 1181
Love A Molnegren V Mansson AS Smaradottir A Thorsteinsson SB Widell A Evolution of hepatitis C virus variants following blood transfusion from one infected donor to several recipients: a long-term follow-up. J Gen Virol 2004 85 441 450 14769902 10.1099/vir.0.19439-0
Curran R Jameson CL Craggs JK Grabowska AM Thomson BJ robins A Irving WL Ball JK Evolutionary trends of the first hypervariable region of the hepatitis C virus E2 protein in individuals with differing liver disease severity J Gen Virol 2002 83 11 23 11752696
Delwart E Magierowska M Royz M Foley B Peddada L Smith R Heldebrant C Conrad A Busch M Homogeneous quasispecies in 16 out of 17 individuals during very early HIV-1 primary infection AIDS 2002 16 189 195 11807302 10.1097/00002030-200201250-00007
Domingo E Biebricher CK Eigen M Holland JJ Quasispecies and RNA virus evolutioon:principles and consequences Landes Bioscience, Texas, USA 2001 82 108
Holmes EC Patterns of intra- and interhost nonsynonymous variation reveal strong purifying selection in dengue virus. J Virol 2003 77 11296 11298 14512579 10.1128/JVI.77.20.11296-11298.2003
Domingo E Ruiz-Jarabo CM Sierra S Arias A Pariente N Baranowski E Escarmis C Emergence and selection of RNA virus variants: memory and extinction Virus Res 2002 82 39 44 11885948 10.1016/S0168-1702(01)00385-9
Gubler DJ Kuno G Sather GE Velez M Oliver A Mosquito cell cultures and specific monoclonal antibodies in surveillance for dengue viruses. Am J Trop Med Hyg 1984 33 158 165 6364855
World Health Organization Dengue haemorrhagic fever: diagnosis, treatment, prevention and control. 2nd edition. Geneva. Wld Hth Org 1997 12 23
Kao CL Wu MC Chiu YH Lin JL Wu YC Yueh YY Chen LK Shaio MF King CC Flow cytometry compared with indirect immunofluorescence for rapid detection of dengue virus type 1 after amplification in tissue culture. J Clin Microbiol 2001 39 3672 3677 11574589 10.1128/JCM.39.10.3672-3677.2001
Kumar S Tamura K Nei M MEGA: Molecular Evolutionary Genetics Analysis (Pennsylvania State Univ., University Park) 1993
Nei M Gojobori T Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol Biol Evol 1986 3 418 426 3444411
Yang Z Bielawski JP Statistical methods for detecting molecular adaptation. Trends Ecol Evol 2000 15 496 503 11114436 10.1016/S0169-5347(00)01994-7
Nielsen R Yang Z Likelihood models for detecting positively selected amino acid sites and applications to the HIV-1 envelope gene Genetics 1998 148 929 936 9539414
Felsenstein J PHYLIP: Phylogeny Inference Package (Univ. of Wahsington, Seattle), Version 3.5c 1993
|
16120221
|
PMC1208963
|
CC BY
|
2021-01-04 16:38:59
|
no
|
Virol J. 2005 Aug 24; 2:72
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-72
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-771613525610.1186/1743-422X-2-77Case ReportVaricella zoster virus acute retinal necrosis following eye contusion: case report Svozílková Petra [email protected]Říhová Eva [email protected]ík Pavel [email protected] Pavel [email protected]řík Zdeněk [email protected]á Bohdana [email protected] Department of Ophthalmology, 1st Faculty of Medicine, Charles University, Prague, Czech Republic2005 31 8 2005 2 77 77 19 6 2005 31 8 2005 Copyright © 2005 Svozílková et al; licensee BioMed Central Ltd.2005Svozílková et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Acute retinal necrosis is a sight-threatening disease caused by the group of herpesviruses. The aim of this paper is to report a case of acute retinal necrosis following ocular trauma in a patient initially treated with vaso-active drugs and corticosteroids for presumed ocular ischemic syndrome.
Case presentation
A 51-years-old otherwise healthy man, who suffered from sudden visual loss in the left eye following contusion, was commenced on vaso-active drugs and systemic corticosteroids for suspected ocular ischemic syndrome with extensive swelling of the optic disc and macular edema. Subsequently, vision in the initially uninvolved right eye decreased. Polymerase chain reaction of vitreous samples and retinal biopsy confirmed varicella zoster virus. Despite intensive treatment with intravenous antiviral medication, the patient became completely blind in both eyes.
Conclusion
Initial treatment of acute, unexplained visual decrease with systemic corticosteroids may lead to visual loss in patients with developing acute retinal necrosis. Ocular trauma could have induced and corticosteroid treatment promoted reactivation of a latent viral infection in our patient.
acute retinal necrosisacyclovircontusioncorticosteroidsvaricella zoster virus
==== Body
Acute retinal necrosis (ARN) is a sight-threatening clinical syndrome caused by the group of herpesviruses (herpes simplex virus; HSV-1 and HSV-2, varicella zoster virus; VZV, cytomegalovirus; CMV or Epstein-Barr virus; EBV). Rapidly progressing retinal inflammation leads to severe impairment of vision.
Case presentation
We present a case of a 51-year-old otherwise healthy man, who suffered from rapid visual loss in the left eye following contusion. Ocular trauma was caused during a football match by a ball, which hit an index finger located just in front of the bulbus. The patient attended our department on April 27, 2004, one week after the injury, when the vision in the left eye decreased to light perception with inaccurate light projection and hand movements in a lower part of visual field. The best-corrected visual acuity in the right eye was 1.0. Intraocular pressures were 18 mmHg in the right eye and 45 mmHg in the left eye. Examination of the anterior segment and fundus of the right eye revealed no pathology. The left eye showed discrete injection of the conjunctiva and keratic precipitates with mild anterior chamber flare and cells. There was iridodonesis, cleft syndrome and a relative afferent pupillary defect in the left eye. The fundus examination of the left eye revealed swelling of the optic disc, large ischemic macular edema, superficial retinal hemorrhages, narrowing of the arterioles and dilatation of the venules (Figure 1A). Fluorescein angiography of the left eye showed macular edema and vascular leakage in the venous phase (Figure 1B). Duplex Doppler ultrasonography and computed tomography scans of the brain and orbits were normal.
Figure 1 Red free fundus photographs and late phases of fluorescein angiography. (A, B) On initial examination, red free photograph and late phase of fluorescein angiography of the left eye demonstrated swelling of the optic disc, ischemic macular edema, narrowing of the arterioles, dilatation of the venules and superficial retinal hemorrhages. (C, D) Four weeks later, fundus examination of the initially uninvolved right eye revealed swelling of the optic disc with hemorrhages and ischemic macular edema.
Based on the clinical findings, the presumed diagnosis of ocular ischemic syndrome was made. The patient was initially treated with vaso-active drugs in addition to corticosteroids. Intravenous methylprednisolone (500 mg daily for 5 days) followed by 60 mg of oral prednisone daily was indicated due to swelling of the optic disc and macular edema. Despite intensive therapy, the fundus examination showed progression of ischemic lesions. Visual acuity in the left eye was light perception with inaccurate light projection. The finding on the right eye was without changes. The patient was discharged on oral prednisone 50 mg daily.
On May 24, 2004, four weeks after pulse intravenous corticosteroid treatment, vision in the initially uninvolved right eye decreased to 0.25. Fundus examination of the right eye disclosed swelling of the optic disc with hemorrhages, blurring of the optic disk margins and ischemic macular edema (Figure 1C, D). In the left eye, massive vitreous opacities made evaluation of the fundus impossible.
A differential diagnosis of antiphospholipide syndrome, masquerading syndrome, viral retinitis or specific inflammation was considered.
No neurological or other abnormalities were found on systemic examination. The findings from magnetic resonance imaging and magnetic resonance angiography of brain and orbits were within normal limits. The cerebrospinal fluid was negative for VZV DNA and enteroviruses RNA. Chest X ray and abdominal ultrasonography were normal. Leukocyte count, hematocrit and activated partial tromboplastin time (APTT) were normal, liver tests showed elevated levels of alaninaminotranspherase (ALT; 2.63 ukat/l). Anti-cardiolipin antibodies were negative. Serologic tests for syphilis and human immunodeficiency virus (HIV-1/-2) were negative. Serum was evaluated regarding evidence for herpesviruses by means of polymerase chain reaction (PCR). Low levels of VZV and EBV EBNA-1 IgG antibodies were detected in serum, whereas IgM antibodies were absent; as well antibodies of respiratory infections or neuroinfections were negative. Blood cultures were also negative. Immunofenotypization showed lower count of lymphocytes in peripheral blood, without plasma cell neoplasia.
On May 27, 2004, an aqueous tap of the left eye was performed and samples were submitted for cytological and virological analysis. PCR of aqueous humour was negative for herpesviruses family and cytology confirmed non-purulent intraocular inflammation. The patient was treated with corticosteroids. The best-corrected visual acuity in the right eye decreased to 0.02. Due to progressive impairment of the clinical status, the corticosteroid therapy was stopped. Fundus examination demonstrated several enlarging foci of necrotizing retinitis with extensive posterior pole involvement (Figure 2). Based on the clinical appearance, a diagnosis of presumed necrotizing herpetic retinopathy was made. The patient was commenced on high-dose intravenous acyclovir (4 × 500 mg per day for 2 weeks). Two days later, visual loss with acuity reduced to light perception with inaccurate light projection occurred in the right eye. In the left eye, there was progression of vitreous opacities and the vision was light perception with inaccurate light projection.
Figure 2 Red free fundus photograph of the right eye after intravenous corticosteroid treatment. Fundus examination disclosed several enlarging foci of necrotizing retinitis.
On June 4, 2004, a diagnostic pars plana vitrectomy and retinal biopsy were carried out in the left eye. The vitreous cavity was filled with 16% perfluoropropane (C3F8) gas. A part of retina and samples of diluted and undiluted vitreous were obtained. Due to failure of the antiviral treatment and ocular disease progression, the patient underwent a pars plana vitrectomy in the right eye on June 9, 2004.
PCR of retina of the left eye and undiluted vitreous of both eyes were positive for VZV. Undiluted vitreous was negative for HSV-1 and -2, CMV, EBV. PCR of diluted vitreous was negative for herpesviruses family. Mycobacterium tuberculosis was not detected using PCR in vitreous of both eyes. Cultivation of vitreous for bacteria and fungi was negative; Toxoplasma gondii antibodies were also negative. Histopathological examination confirmed non-purulent intraocular inflammation. Immunofenotypization of vitreous of both eyes showed no plasma cell neoplasia.
In two weeks, intravenous acyclovir was followed by oral acyclovir (5 × 400 mg daily). In the right eye, large foci of retinal atrophy with reduced inflammatory reaction were present. Owing the cataract induced by gas, fundus of the left eye was not visible.
The patient was discharged on acyclovir 4 × 400 mg daily. However, an exudative retinal detachment was seen in the right eye and vision decreased to 0. Vision in the left eye was light perception with inaccurate light projection. On examination 4 weeks later, B-scan ultrasonography of the left eye confirmed the exudative retinal detachment. Nevertheless, despite intensive treatment with intravenous antiviral medication, the patient became completely blind in both eyes.
Discussion
ARN is a serious ophthalmic manifestation of infection caused by the herpesviruses family. A rapid and accurate diagnosis of herpetic infection is crucial for prompt administration of specific antiviral therapy. Although the precise pathogenesis of ARN is not completely understood, Kezuka and coworkers [1] found out that a high proportion of patient with ARN associated with VZV displayed a transient loss of virus-specific delayed hypersensitivity, but their serum samples contained high titers of anti-VZV antibodies. The authors propose that idiopathic reactivation of VZV in one eye might promote suppression of delayed hypersensitivity, thereby eliminating the virus-specific CD4+ T cells that are required to prevent neural spread of the virus from the site of reactivation. ARN in the contralateral eye may be the inevitable consequence. On resolution of the intraocular inflammation, virus-specific delayed hypersensitivity recurred in most of these individuals. Patients with ARN syndrome should be followed because of possible recurrence.
We present a case of a 51-year-old otherwise healthy man with rapid visual loss initially treated with vaso-active drugs and systemic corticosteroids for presumed ocular ischemic syndrome with swelling of the optic disc and macular edema. The causative agent was diagnosed as VZV based on PCR analysis of vitreous and retinal samples. Possible mechanisms of VZV necrotizing retinopathy include reinfection by an exogenous virus or reactivation of a latent infection. In our opinion, ocular trauma probably induced reactivation of a latent virus in the presented patient. Absence of high VZV titers in the serum makes systemic reinfection unlikely. Thompson and coworkers [2] demonstrated three patients treated for ARN apparently caused by reactivation of latent HSV-2. Primary viral infection was probably congenital, with documented perinatal HSV-2 infection in two patients. In all these cases, periocular trauma preceded the development of retinitis by two to three weeks. To our knowledge, the possible reactivation of VZV by ocular trauma has never been reported.
A unique case of acute HSV encephalitis associated with bilateral ARN syndrome after craniotomy for resection of a suprasellar craniopharyngioma has been reported. The authors hypothesized reactivation of previously latent HSV in the area of the inferior frontal lobe and optic chiasm. Reactivated virus may have migrated to the retina by axonal transport, through the optic nerves, to induce the ARN syndrome [3]. The onset of bilateral necrotizing herpetic retinopathy three years after HSV encephalitis following pulse corticosteroid treatment has also been described. Based on the extremely rapid development of retinitis to involve the fellow eye after pulse corticosteroid therapy, the authors concluded that treatment with corticosteroids alone might increase the risk of reactivation of latent infection [4].
Ocular trauma could have induced, and systemic corticosteroid treatment probably promoted, reactivation of a latent virus in our patient. Initial treatment of acute, unexplained decrease of vision with systemic corticosteroids may lead to visual loss in patients with developing necrotizing herpetic retinopathy [5]. Since progression to profound and irreversible visual loss is rapid, early diagnostic vitreous biopsy must be performed before commencement of immunosuppressive drugs. PCR analysis of vitreous samples is a valuable tool in the early diagnosis and initiation of appropriate treatment.
Competing interests
The author(s) declare that they have no competing interests.
Acknowledgements
Written consent was obtained from the patient for publication of case including clinical photographs.
==== Refs
Kezuka T Sakai J Usui N Streilein JW Usui M Evidence for antigen-specific immune deviation in patients with acute retinal necrosis Arch Ophthalmol 2001 119 1044 1049 11448326
Thompson WS Culbertson WW Smiddy WE Robertson JE Rosenbaum JT Acute retinal necrosis caused by reactivation of herpes simplex virus type 2 Am J Ophthalmol 1994 118 205 211 8053466
Perry JD Girkin CA Miller NR Kerr DA Herpes simplex encephalitis and bilateral acute retinal necrosis syndrome after craniotomy Am J Ophthalmol 1998 126 456 460 9744385 10.1016/S0002-9394(98)00108-1
Verma L Venkatesh P Satpal G Rathore K Tewari HK Bilateral necrotizing herpetic retinopathy three years after herpes simplex encephalitis following pulse corticosteroid treatment Retina 1999 19 464 467 10546950
Benz MS Glaser JS Davis JL Progressive outer retinal necrosis in immunocompetent patients treated initially for optic neuropathy with systemic corticosteroids Am J Ophthalmol 2003 135 551 553 12654381 10.1016/S0002-9394(02)01978-5
|
16135256
|
PMC1208964
|
CC BY
|
2021-01-04 16:38:58
|
no
|
Virol J. 2005 Aug 31; 2:77
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-77
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-551611783710.1186/1477-7819-3-55Case ReportAdenocarcinoma in bladder diverticulum, metastatic from gastric cancer Matsuhashi Nobuhisa [email protected] Kazuya [email protected] Taiso [email protected] Kuniyasu [email protected] Yasuyuki [email protected] Yosuke [email protected] Department of Surgical Oncology, Gifu University Hospital, 1-1 Yanagido, Gifu City 501-1193, Japan2 Department of Pathology, Gifu University Hospital, 1-1 Yanagido, Gifu City 501-1193, Japan3 Department of Emergency & Disaster Medicine, Gifu University Hospital, 1-1 Yanagido, Gifu City 501-1193, Japan2005 24 8 2005 3 55 55 12 5 2005 24 8 2005 Copyright © 2005 Matsuhashi et al; licensee BioMed Central Ltd.2005Matsuhashi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Metastasis to the urinary bladder from gastric cancer is rare. Metastasis to a diverticulum of the bladder from gastric cancer is extremely rare. We report a case of isolated bladder metastasis from gastric cancer and invasion localized to the muscularis propria of the primary site (stomach).
Case presentation
A 90-year-old female presented with nausea and vomiting that was diagnosed as gastric cancer, the patient also had intermittent hematuria. Pelvic computed tomography identified an abnormally thickened area in the bladder wall that was diagnosed as a diverticulum of the bladder. A biopsy of the bladder wall revealed well differentiated tubular adenocarcinoma metastatic from gastric carcinoma.
Conclusion
Almost all cases of bladder metastasis from gastric cancer had peritoneal dissemination. This particular presentation of bladder metastasis from gastric cancer, to the best of our knowledge, has not been previously reported.
==== Body
Background
Metastasis to a diverticulum of the bladder from gastric cancer is extremely rare [1]. Gastric cancer has a tendency to metastasize widely, most commonly to the liver, lung, lymph nodes, bone and peritoneum [2]. The bladder may be involved in the late stages from metastasis and is usually associated with metastasis to other organs [3], but isolated bladder metastasis and invasion localized in the muscularis propria of the primary site (stomach) is extremely rare.
Case presentation
A-90-year-old female presented in November 2003 with a history of nausea, vomiting and dysphagia, with hematuria. On admission her abdomen was slightly distended, tympanic, and slightly tender in the upper abdominal regions, with normal bowel sounds and no palpable mass. Blood tests revealed a white blood cell count of 6,100/mm3, c reactive protein (CRP) 6.01 mg/dl, carcinoma antigen (CA) 19-9 50.6 mg/dl and carcinoembryonic antigen (CEA) of 2.9 mg/dl. Pelvic computed tomography (CT) identified an abnormal thickness of the bladder wall with enhance effect in a diverticulum and it's origin from bladder (Figure. 1). Another mass was seen in the antral portion of the stomach; however, pancreas and biliary tract were normal on computed tomography. There was not other lesion detected in other abdominal and pelvic organs.
Figure 1 Pelvic computed tomography revealed an abnormally thickened diverticulum of the bladder.
A gastroscopy was carried out which revealed type 3 pyloric stenosis. A biopsy of the stomach was taken which revealed well-differentiated tubular adenocarcinoma. Cystoscopy was performed which showed a lesion in the bladder diverticulum, a biopsy of the bladder wall revealed well-differentiated tubular adenocarcinoma metastasis from gastric carcinoma. At laparotomy, the pylorus segment of the stomach was viable with signs of edema, but no serosal invasion was identified. There were no sign of peritoneal dissemination in the intra-abdominal cavity. Peritoneal washings were negative for malignant cells. A palliative distal gastrectomy with gastrojejunostomy was performed to relieve pyloric obstruction.
However, cystectomy or diverticulectomy was not performed due to age of the patient and technical difficulties due to previous two surgeries performed for abnormal position of uterus and volvulus of intestine. The size of the macroscopic specimen was 3.0 × 2.5 cm (Figure. 2). Histology revealed well differentiated tubular adenocarcinoma invading to the muscularis propria (MP), 3type, Infiltrative growth pattern (inf) β, int, ly3, v0 (Figure. 3, 4a), Similar to that of bladder tumor (Figure 4b). The patient recovered with no further symptoms, and was discharged on the 19th postoperative day. However, patient later developed pyelonephritis, bilateral hydronephrosis, disseminated intravascular coagulation (DIC) and died three months after the surgery.
Figure 2 Macroscopic specimens identified a type-3 tumor at the pylorus of the stomach.
Figure 3 Microscopic specimens demonstrated localization in the muscularis propria on the invasion index (T2) (Hematoxylin and Eosin ×10).
Figure 4 Photomicrograph of a) stomach showing well differentiated tubular adenocarcinoma (left). (Hematoxylin and Eosin ×400) and b) bladder biopsy specimen showing well differentiated tubular adenocarcinoma (right) (Hematoxylin and Eosin ×400).
Discussion
Metastases to urinary bladder are rare, accounting for less than 2% of all bladder tumors, these are mostly found in advanced stages with peritoneal dissemination [2]. Information pertaining to bladder metastases is derived largely from autopsy studies, and known primary sites of origin in descending frequency are gastric cancer, malignant melanoma, breast and lung [2]. Potential mechanisms contributing to the appearance of secondary bladder tumors from adjacent organs are implantation of exfoliated cells from the bladder periphery or renal pelvis, and lymphogenous, hematogenous, or peritoneal dissemination from a distant primary source [3].
The relative infrequency of primary adenocarcinoma of the bladder causes the dilemma whether bladder adenocarcinoma represents a primary or secondary process [1]. Mostofi et al, have proposed several guidelines for such differentiation [2]. If the adjacent mucosa contains polypoid formation, Brunn's nests, or glandular or mucous metaplasia, a primary bladder lesion is likely. An additional feature favoring a bladder origin is the coexistence of transitional and squamous carcinoma. Mostofi et al, also stipulated that secondary bladder tumors rarely provoke urinary symptoms before the primary site is detected. In this case, histology indicated that neoplastic columnar cells formed small solid nests and /or small-sized glandular structures. In conclusion, it appears that the stomach was the preponderant site of the origin
More than 95% of the bladder tumors are transitional cell carcinoma and less than 1% is adenocarcinoma [4]. Almost all bladder adenocarcinoma originate from trigone of the bladder. Gastric cancer metastatic to the bladder may behave differently in the two sexes. Among 10 autopsied male patients with gastric cancer, Hermann found bladder metastases in only one; however, among 12 cases of Krukenberg's tumors (ovarian cancers arising from gastrointestinal origin), there were 6 cases with metastases to the bladder, uterus, and Fallopian tubes [3]. It was hypothesized that the ovary might somehow direct metastases to the pelvic organs, since bladder metastases are very rare in the absence of Krukenberg's tumor. Patients of metastatic linitis plastica described by Mizutani et al, [4], and Leddy et al, [5] were both females with metastatic tumors in at least one ovary as well as the bladder. Since the chief complaint of patients with bladder metastasis is intermittent hematuria, bladder metastases from gastric cancers have been reported mainly by urology surgeons, and thus cancer invasion of the stomach in almost all of the existing case reports was not analyzed [4-7]. Our results indicate that the finding of an abnormally thickened diverticulum of the bladder may provide prognostic value in computed tomography, and additionally in localized gastric cancer lesions with invasion limited to the muscularis propria too might metastasize by lymphogenous spread. To our knowledge this is the first reported case of isolated metastasis to a urinary bladder diverticulum.
Conclusion
Isolated metastasis to urinary bladder are rare, metastasis to a urinary bladder diverticulum is still rarer.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NM, KY, TT, YS and YA took part in the operation, performed the literature search and drafted the manuscript for submission.
KS performed histological examination and provided photomicrographs.
All authors read and approved the final manuscript.
Acknowledgements
The written permission was obtained from the patient for publication of this case report.
==== Refs
Mostofi FK Thompson RV Dean AL Mucous adenocarcinoma of the urinary bladder Cancer 1955 8 741 758 13240656
Ganem EJ Batal JT Secondary malignant tumors of the urinary bladder metastatic from primary foci in distant organs J Urol 1956 75 965 972 13320594
Hermann HB Metastatic tumors of the urinary bladder originating from the carcinoma of the gastrointestinal tract J Urol 1929 22 257
Mizutani Y Hashimura T Kitayama T Toshimitsu T Nonomura M A case of secondary tumor, the origin (gastric cancer) of which could not be identified before autopsy Hinyoukika Kiyo 1990 36 605 608
Leddy FF Peterson NE Ning TC Urogenital linitis plastica metastatic from stomach Urology 1992 39 464 467 1580041 10.1016/0090-4295(92)90249-V
Ota T Shinohara M Kinoshita K Sakoma T Kitamura M Maeda Y Two Cases of metastatic bladder cancers showing diffuse thickening of the bladder wall Jpn J Clin Oncol 1999 29 314 316 10418562 10.1093/jjco/29.6.314
Kim HC Kim SH Hwang SI Lee HJ Han JK Isolated bladder metastases from stomach cancer: CT demonstration Abd Imaging 2001 26 333 335 10.1007/s002610000163
|
16117837
|
PMC1208965
|
CC BY
|
2021-01-04 16:39:04
|
no
|
World J Surg Oncol. 2005 Aug 24; 3:55
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-55
|
oa_comm
|
==== Front
World J Surg OncolWorld Journal of Surgical Oncology1477-7819BioMed Central London 1477-7819-3-581613732810.1186/1477-7819-3-58Case ReportPancreatic metastasis of Merkel cell carcinoma and concomitant insulinoma: Case report and literature review Bachmann Jeannine [email protected] Jorg [email protected] Frank [email protected] Shailesh V [email protected] Wolfgang [email protected]üchler Markus W [email protected] Helmut [email protected] Department of General Surgery, University of Heidelberg, Heidelberg, Germany2 Department of Pathology, University of Heidelberg, Heidelberg, Germany3 Department of Dermatology, University of Heidelberg, Heidelberg, Germany2005 1 9 2005 3 58 58 19 5 2005 1 9 2005 Copyright © 2005 Bachmann et al; licensee BioMed Central Ltd.2005Bachmann et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Merkel cell carcinomas are rare neoplasm of neuroendocrine origin, usually observed in elderly people in areas with abundant sunlight, and predominantly located on the head and neck, extremities, and trunk. In many patients, a local recurrence after resection of the primary tumour and even distant metastases can be found.
Case presentation
We report an unusual occurrence of pancreatic metastases from a previously diagnosed Merkel cell carcinoma with the discovery of a concomitant insulinoma. An 82-year old lady suffered from recurrent attacks of hypoglycemia and presented with an abdominal mass, 2 years prior she had an excision done on her eyebrow that was reported as Merkel cell carcinoma. An extended distal pancreatectomy and splenectomy along with resection of the left flexure of the colon for her abdominal mass was carried out. Final histopathology of the mass was a poorly differentiated endocrine carcinoma in the pancreatic tail, in the peripancreatic tissue and in the surrounding soft tissue consistent with metastatic Merkel cell carcinoma in addition to an insulinoma of the pancreatic body.
Conclusion
This is the first documented case of a metastatic Merkel cell carcinoma and a concomitant insulinoma, suggesting either a mere coincidence or an unknown neuroendocrine tumor syndrome.
==== Body
Background
Merkel cell carcinoma is a rare cutaneous neoplasm of neuroendocrine origin, usually observed in elderly people in areas with abundant sunlight. Affected skin areas are predominantly the head and neck, but the extremities and trunk are other predisposed areas [1-4]. In a review of 1024 patients, the primary tumour was found on the head and neck in 40% of the patients, in 33% on the extremities and in 23% on the trunk [5]. In 98% of reported cases this cancer could be found in Caucasians, suggesting possible protection by darker skin pigmentation [5]. While the exact incidence remains unknown, Hodgson found an increase in the incidence of Merkel cell carcinoma from 0.15 cases per 100,000 in 1986 to around 0.44 cases per 100,000 in 2001; the highest incidence of 4.28 per 100,000 was found in patients over 85 years [6]. The literature review of 1024 patients showed a mean age of 69 years at diagnosis [5]. We report an unusual occurrence of pancreatic metastases from a previously diagnosed Merkel cell carcinoma with the discovery of a concomitant insulinoma.
Case presentation
A 82-year old lady presented to the Department of General Surgery at the University of Heidelberg, Germany with recurrent attacks of hypoglycemia and a large abdominal mass. While diagnostic tests repeatedly documented glucose levels below 40 mg/dl (normal levels 80 – 120 mg/dl), a computed tomography (CT) (Figure 1A, B) scan of the abdomen revealed a large lesion of around 5 to 6 cm in relation to the pancreatic body and tail. There were also large masses of about 3–5 cm in the retroperitoneum and in the area of the celiac trunk and around the mesenteric artery. Furthermore, in the pancreatic body there was a hypervascularized area (Figure 2), that was suspicious for an insulinoma. Clinically this lady, who was not thriving, reported a weight loss of 12 kilograms over the previous 4 months. A somatostatin receptor scintigraphy showed an enhanced uptake in the region of the pancreatic body/tail as well as in the right axilla (a palpable mass was also noted there) and excluded the possibility of other involved areas.
Figure 1 Computed tomography scan of the abdomen revealed a large lesion of 5 to 6 cm in relation to the pancreatic body and tail (venous phase A, B, arrows).
Figure 2 Arterial phase of the computed tomography scan of the abdomen shows a hypervascularized area (arrows) in the pancreatic body.
She gave a past history of an operation done on the right eyebrow 2 years prior for a 0.8 × 0.8 cm lesion (Figure 3) that was reported as a Merkel cell carcinoma. Histopathology showed rather uniform tumor cells in a trabecular growth pattern (Figure 4a) with monomorphous pale-stained nuclei and many mitoses (Figure 4b). There was invasion of dermal lymphatics and blood vessels (Figure 4a, c, insert). Immunohistochemistry revealed strong positivity for cytokeratin 20 (Figure 4c) and neurofilament (not shown) in the characteristic dot-like pattern and a weak expression of chromogranin A (Figure 4d). After excision, radiation therapy was also administered only at the site of the primary lesion, the draining lymphatic vessels and the first lymph node station. A year later, a large abdominal mass was noted of uncertain origin and an ultrasound guided biopsy showed an unspecified small cell cancer. In view of the large mass with additional suspicious areas being noted in the spleen, left adrenal gland and axilla, she had been subjected to palliative radiotherapy of 30 Gray over 2 months. However no definitive diagnosis of metastasis in these areas was established. With a working clinical diagnosis of symptomatic insulinoma not responding to medical measures, a decision for surgical resection of this large lesion was inevitable, the age of the patient and the previous history of palliative radiation just 6 months prior notwithstanding.
Figure 3 Merkel cell carcinoma on the right eyebrow.
Figure 4 Merkel cell carcinoma – primary tumor. A, B: Hematoxylin and eosin staining. A. Lower magnification showing intravasal tumor cells. B: Monomorphous tumor cells, pale staining nuclei, many mitotic figures. C: Strong cytokeratin 20 staining. Insert: Intravasal cytokeratin 20-positive tumor cells. D: Weak chromogranin A staining
Operative details
Surgical exploration revealed a large mass of about 5 cm in the tail of the pancreas, in close proximity to the spleen and the splenic flexure of the transverse colon. However there was no evidence of any metastatic disease to the liver, peritoneum and the adnexae. After a careful and meticulous mobilization, a distal pancreatectomy, splenectomy, and adrenalectomy along with resection of the splenic flexure of the colon were performed.
Pathological examination revealed a tumor with manifestations in the pancreatic tail, the adrenal gland, the peripancreatic tissue, and the surrounding soft tissue. Grossly, the mass displayed a whitish and glassy cut surface, containing extended areas of haemorrhage and necrosis. Histologically, the tumor displayed endocrine architecture with mostly solid formations of rather monomorphic cells. The tumor was mitotically highly active (mitotic count >10 per high power field) and contained abundant areas of necrosis. Immunohistochemically, the tumor cells were strongly positive for the endocrine marker synaptophysin and for cytokeratin 20 while there was no expression of insulin. The proliferative activity (MIB-1) reached approximately 80% (Figure 5).
Figure 5 Merkel cell carcinoma metastasis in the pancreas. A, B: Hematoxylin and Eosin staining. C: Synaptophysin staining. D: Cytokeratin 20 staining.
Furthermore, gross examination of the resected specimen revealed a well demarcated, brownish tumor of the pancreatic body, measuring 1.2 cm in diameter. This tumor microscopically displayed endocrine architecture with trabecular arrangements of uniform tumor cells, showing no mitotic activity. Immunohistochemistry revealed strong positivity for synaptophysin as well as focal positivity for insulin. The proliferative activity (MIB-1) was approximately 1% (Figure 6). The diagnosis of a poorly differentiated endocrine carcinoma (Merkel cell carcinoma) along with that of benign pancreatic insulinoma was thus made.
Figure 6 Insulinoma in the pancreatic body. A, B: Hematoxylin and Eosin staining. C: Insulin staining (note the focal positivity). D: Synaptophysin staining.
The patient had a smooth postoperative recovery, the bouts of hypoglycaemia completely disappeared, and she was discharged home within 3 weeks of surgery. She is presently asymptomatic and remains on regular follow up.
Discussion
Though a very rare tumor, Merkel cell carcinoma is inherently aggressive. The cells of origin are thought to be Merkel cells, which are neuroendocrine cells present in the basal layer of the epidermis. However the exact origin is controversial since often on immunohistological examination both epidermal and neuroendocrine features can be found. This tumor has a high tendency for local recurrence [2,7-9]. While in up to 50% of patients local recurrence after excision is observed, most are diagnosed within two years of the primary resection [1]. 50 – 75% of the patients develop lymph node metastases [5,9-12]. Distant metastases are described in many cases [1,5,7,8,13]. The survival depends on the treatment and the stage at diagnosis. In a study of 251 patients, the 5-year survival rate was 97% for those who had no positive lymph nodes on histological examination. In contrast the 5-year survival was 52% in those who presented with positive lymph nodes [1]. Shalhub et al, reported on a case of metastatic Merkel cell cancer. In their case the patient showed a lymph node in the axilla and in the groin. At the same time a CT scan showed multiple enlarged portocaval and gastrohepatic lymph nodes as well as a small lesions in the vicinity of the colon, which were metastatic Merkel cell carcinoma on histological examination [13].
Aggressive therapy of the primary tumor is crucial for improved outcomes. Surgery and adjuvant radiotherapy provide good local control of the primary tumor [5]. A retrospective study of 35 patients suffering from Merkel cell carcinoma showed that wide excision, lymph node dissection and adjuvant radiotherapy decreased the loco-regional recurrence, and patients had a better survival [14]. In a review of 1024 patients, the recurrence rate was 10.5% in those who received radiotherapy in comparison to 52.6% without radiation following resection [5].
Pancreatic Metastases from Merkel Cell Cancer
The majority of the neoplasms in the pancreas have their origin in the organ itself. In a study examining 973 surgical specimen, only 38 metastatic tumors to the pancreas were noted [15]. The origin of pancreatic metastases were lymphomas, carcinomas of stomach, kidney, lung and others, as well as a Merkel cell carcinoma [15]. A Medline search with the key words Merkel cell cancer, Merkel cell carcinoma, metastatic disease, pancreatic metastases and surgical treatment revealed two additional reports of pancreatic metastasis from a Merkel cell carcinoma: One patient had a 4 cm violaceous firm subcutaneous nodule of the left lower eyelid. The remaining findings on physical examination were unremarkable. A biopsy confirmed that as Merkel cell tumor. Later this patient presented with jaundice and a cystic lesion (approximately 5 cm in diameter) in the pancreas. Subsequent laparoscopy showed evidence of peritoneal carcinomatosis, which was histological confirmed as a metastasis of the primary Merkel carcinoma [7]. In another case report, a 64-year old man presented with a short history of jaundice. A CT scan of the abdomen showed a large 6 cm mass in the head of the pancreas. Histological examination revealed a metastasis of Merkel cell carcinoma approximately two years after resection of a Merkel cell carcinoma resected from his finger [8]. As regards the role of imaging in Merkel cell carcinoma, clinical information is still insufficient to fully appreciate the role of imaging in the management of this condition. A better imaging algorithm is expected with increased awareness and improved clinical understanding of this uncommon neoplasm. Currently, CT scan, along with magnetic resonance imaging, appears to be imaging modality of choice. Contrast enhanced computed tomography (CT) demonstrates high-attenuation adenopathy and soft-tissue nodules. Solid abdominal organs targeted by metastasis manifest as hypervascular lesions with ringlike enhancement [16,17].
Our case adds some more information to the existing pool of knowledge about this tumor. It is well documented that treatment principles of this rare tumor revolve around adequate surgical resection and the addition of radiotherapy [2,18]. In systemic disease, many combinations of chemotherapeutic agents have been used with poor results [19]; although in a recent study, Merkel cell carcinoma was chemosensitive but rarely chemocurable in patients with metastases or locally advanced tumors [20]. While this elderly lady was possibly subjected to palliative radiotherapy given the extent of her abdominal mass, her unrelenting symptoms and the fact that age over 60 years is a poor prognostic factor in Merkel cell carcinoma prompted us to consider a surgical resection in spite of her advanced age. However, the symptoms (severe hypoglycemia) were not related to the metastatic Merkel cell carcinoma, but to the concomitant insulinoma. While her symptoms completely resolved due to excision of the insulinoma, the R0 excision of the Merkel cell carcinoma could also be expected to provide survival benefit; however this remains to be seen on longer follow-up.
Interestingly, Fincher et al reported a Merkel cell carcinoma with a concomitant somatostatinoma. Our case is the second case of Merkel cell carcinoma with a concomitant endocrine tumor. As already speculated by Fincher et al, this "occurrence may represent a previously undescribed neuroendocrine tumor syndrome, and this possibility should be considered when either tumor is diagnosed" [21]. Modern molecular biological methods may help to elucidate whether common molecular alterations are present in these tumor types.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JB collated the information, searched the literature and wrote the manuscript.
FB and WH contributed to the pathological aspects of the manuscript and helped in preparing the manuscript.
JK and SVS assisted in literature search and writing of the manuscript.
MWB and HF managed the patient and helped in preparing the manuscript and edited the final version with JK.
All authors read and approved the final version of the manuscript.
Acknowledgements
Written consent was obtained from the patient for publication of this case.
==== Refs
Allen PJ Bowne WB Jaques DP Brennan MF Busam K Coit DG Merkel cell carcinoma: prognosis and treatment of patients from a single institution J Clin Oncol 2005 23 2300 2309 15800320 10.1200/JCO.2005.02.329
Eng TY Boersma MG Fuller CD Cavanaugh SX Valenzuela F Herman TS Treatment of merkel cell carcinoma Am J Clin Oncol 2004 27 510 515 15596922 10.1097/01.coc.0000135567.62750.f4
Ferringer T Rogers HC Metcalf JS Merkel cell carcinoma in situ J Cutan Pathol 2005 32 162 165 15606676 10.1111/j.0303-6987.2005.00270.x
Haag ML Glass LF Fenske NA Merkel cell carcinoma. Diagnosis and treatment Dermatol Surg 1995 21 669 683 7633811 10.1016/1076-0512(95)97513-6
Medina-Franco H Urist MM Fiveash J Heslin MJ Bland KI Beenken SW Multimodality treatment of Merkel cell carcinoma: case series and literature review of 1024 cases Ann Surg Oncol 2001 8 204 208 11314935
Hodgson NC Merkel cell carcinoma: changing incidence trends J Surg Oncol 2005 89 1 4 15611998 10.1002/jso.20167
Bachmeyer C Alovor G Chatelain D Khuoy L Turc Y Danon O Laurette F Cazier A N'Guyen V Cystic metastasis of the pancreas indicating relapse of Merkel cell carcinoma Pancreas 2002 24 103 105 11741190 10.1097/00006676-200201000-00014
Ouellette JR Woodyard L Toth L Termuhlen PM Merkel cell carcinoma metastatic to the head of the pancreas JOP 2004 5 92 96 15007190
Rufini V Perotti G Brunetti M Crescenzi A Fadda G Troncone L Unsuspected testicular metastases from Merkel cell carcinoma: a case report with therapeutic implications Am J Clin Oncol 2004 27 636 637 15577445 10.1097/01.coc.0000146413.25203.0d
Pitale M Sessions RB Husain S An analysis of prognostic factors in cutaneous neuroendocrine carcinoma Laryngoscope 1992 102 244 249 1545650
Queirolo P Gipponi M Peressini A Disomma CF Vecchio S Raposio E Guenzi M Sertoli MR Cafiero E Merkel cell carcinoma of the skin. Treatment of primary, recurrent, and metastatic disease: review of clinical cases Anticancer Res 1997 17 673 677 9066600
Westgate SJ Radiation therapy for skin tumors Otolaryngol Clin North Am 1993 26 295 309 8460044
Shalhub S Clarke L Morgan MB Metastatic Merkel cell carcinoma masquerading as colon cancer Gastrointest Endosc 2004 60 856 858 15557979 10.1016/S0016-5107(04)02173-X
Kokoska ER Kokoska MS Collins BT Stapleton DR Wade TP Early aggressive treatment for Merkel cell carcinoma improves outcome Am J Surg 1997 174 688 693 9409598 10.1016/S0002-9610(97)00193-1
Adsay NV Andea A Basturk O Kilinc N Nassar H Cheng JD Secondary tumors of the pancreas: an analysis of a surgical and autopsy database and review of the literature Virchows Arch 2004 444 527 535 15057558 10.1007/s00428-004-0987-3
Gollub MJ Gruen DR Dershaw DD Merkel cell carcinoma: CT findings in 12 patients AJR Am J Roentgenol 1996 167 617 620 8751663
Nguyen BD McCullough AE Imaging of Merkel cell carcinoma Radiographics 2002 22 367 376 11896226
Eng TY Naguib M Fuller CD Jones WEIII Herman TS Treatment of recurrent Merkel cell carcinoma: an analysis of 46 cases Am J Clin Oncol 2004 27 576 583 15577435 10.1097/01.coc.0000135926.93116.c7
Ratner D Nelson BR Brown MD Johnson TM Merkel cell carcinoma J Am Acad Dermatol 1993 29 143 156 8335732
Voog E Biron P Martin JP Blay JY Chemotherapy for patients with locally advanced or metastatic Merkel cell carcinoma Cancer 1999 85 2589 2595 10375107 10.1002/(SICI)1097-0142(19990615)85:12<2589::AID-CNCR15>3.0.CO;2-F
Fincher RK Christensen ED Tsuchida AM Ampullary somatostatinoma in a patient with Merkel cell carcinoma Am J Gastroenterol 1999 94 1955 1957 10406267
|
16137328
|
PMC1208966
|
CC BY
|
2021-01-04 16:39:04
|
no
|
World J Surg Oncol. 2005 Sep 1; 3:58
|
utf-8
|
World J Surg Oncol
| 2,005 |
10.1186/1477-7819-3-58
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1617041310.1371/journal.pcbi.001003705-PLCB-RA-0067R2plcb-01-04-02Research ArticleBioinformatics - Computational BiologyMicrobiologySystems BiologyEubacteriaTransition to Quorum Sensing in an Agrobacterium Population: A Stochastic Model Model of Quorum Sensing in
AgrobacteriumGoryachev Andrew B 1*Toh Da-Jun 1Wee Keng Boon 1Lee Travis 1Zhang Hai-Bao 2Zhang Lian-Hui 21 Systems Biology Group, Bioinformatics Institute, Singapore
2 Institute of Molecular and Cellular Biology, Singapore
Margalit Hanah EditorHebrew University of Jerusalem, Israel* To whom correspondence should be addressed. E-mail: [email protected] 2005 16 9 2005 1 4 e374 4 2005 8 8 2005 Copyright: © 2005 Goryachev et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Understanding of the intracellular molecular machinery that is responsible for the complex collective behavior of multicellular populations is an exigent problem of modern biology. Quorum sensing, which allows bacteria to activate genetic programs cooperatively, provides an instructive and tractable example illuminating the causal relationships between the molecular organization of gene networks and the complex phenotypes they control. In this work we—to our knowledge for the first time—present a detailed model of the population-wide transition to quorum sensing using the example of Agrobacterium tumefaciens. We construct a model describing the Ti plasmid quorum-sensing gene network and demonstrate that it behaves as an “on–off” gene expression switch that is robust to molecular noise and that activates the plasmid conjugation program in response to the increase in autoinducer concentration. This intracellular model is then incorporated into an agent-based stochastic population model that also describes bacterial motion, cell division, and chemical communication. Simulating the transition to quorum sensing in a liquid medium and biofilm, we explain the experimentally observed gradual manifestation of the quorum-sensing phenotype by showing that the transition of individual model cells into the “on” state is spread stochastically over a broad range of autoinducer concentrations. At the same time, the population-averaged values of critical autoinducer concentration and the threshold population density are shown to be robust to variability between individual cells, predictable and specific to particular growth conditions. Our modeling approach connects intracellular and population scales of the quorum-sensing phenomenon and provides plausible answers to the long-standing questions regarding the ecological and evolutionary significance of the phenomenon. Thus, we demonstrate that the transition to quorum sensing requires a much higher threshold cell density in liquid medium than in biofilm, and on this basis we hypothesize that in Agrobacterium quorum sensing serves as the detector of biofilm formation.
Synopsis
Understanding the interplay between the extracellular environment and intracellular decision circuitry of a cell is important but is an arduous goal to achieve since many interacting factors, difficult to measure and control in experiment, are involved. The authors address this problem by means of computational modeling using the example of a bacterial population that cooperatively switches on a common gene expression program if a certain critical population density is achieved. They developed a detailed model of the intracellular control network and demonstrated that it can operate as an “on–off” gene expression switch that is sensitive to environmental control and yet highly robust to intracellular molecular noise. The population-wide transition is further modeled using a novel method in which each bacterium is given a unique copy of an intracellular network. This approach, which allows monitoring of both the dynamics of individual cells and population behavior, provides an explanation for the gradual appearance of the transition to the “on” state that has been observed in experiments, and quantitatively predicts the critical value of the population density at which this transition occurs. Unexpectedly, a comparison of the cell densities required for the transition in different environmental conditions brought about a hypothesis regarding the previously elusive ecological and evolutionary function of this cooperative phenomenon.
Citation:Goryachev AB, Toh DJ, Wee KB, Lee T, Zhang HB, et al. (2005) Transition to quorum sensing in an Agrobacterium population: A stochastic model. PLoS Comput Biol 1(4): e37.
==== Body
Introduction
Molecular networks, which integrate signal transduction and gene expression into the unified decision circuitry, are ultimately responsible for the realization of all life activities of biological cells including internal developmental programs and responses to environmental factors. One of the main challenges of systems biology is to uncover and understand the relationships between the properties of these molecular circuits and the macroscopic cellular phenotypes that are controlled by them [1]. Particularly important are the phenotypes involving interaction and cooperative action of multiple cells. The mapping of networks onto phenotypes is still difficult to accomplish in multicellular eukaryotic organisms owing to their staggering complexity. Less complex and more experimentally accessible prokaryotic organisms became the systems of choice for “dissecting social behavior at the genetic level” [2]. The phenomenon of bacterial quorum sensing (QS) gives us a particularly unique opportunity to follow the causal relationships from molecular circuitry to cooperative population dynamics.
QS refers to the ability of bacterial populations to collectively activate certain gene expression programs, e.g., toxin release or antibiotic production, once some critical population density has been reached. QS is found in a vast variety of bacterial species and has been extensively studied experimentally [3–6]. In Gram-negative bacteria, the QS phenomenon is usually controlled by a small gene expression network that functions as an environmentally activated “on–off” gene expression switch [5,6] whose operation is analogous to that of radar. At the low cell density that normally corresponds to the “off” switch state, a key transcription factor required for the expression of proteins responsible for the phenotype is suppressed. At the same time, the cell steadily produces a small amount of the QS signaling molecule, termed the autoinducer, that can freely diffuse in and out of the cell. While the population density is low, most of the autoinducer molecules are washed out and dispersed in the environment by diffusion. As the cell density grows, more molecules of autoinducer enter the bacterium from outside. Once certain cell “quorum” is reached, the inbound autoinducer signal triggers the transition of the QS network to the “on” state, resulting in the production of the transcription factor and the expression of the target genes.
This transition on both intracellular and population-wide scales is the focus of our study. We investigate the phenomenon of QS in the soil-dwelling plant pathogen Agrobacterium tumefaciens, the causative agent of crown gall disease [7]. Bacteria of this species often harbor Ti (tumor-inducing) plasmids that endow their hosts with the unique ability to genetically modify susceptible plants through a cross-kingdom DNA transfer. Like many other soil bacteria, Agrobacterium is chemotactic to exudates released by plant wounds and is capable of catabolizing various nutrients that leave injured plant roots. Once bacteria form physical contact with the surface of the wound, Ti plasmids offer their hosts an extraordinary advantage over their plasmidless competitors. A fragment of the plasmid, termed the vir region, is injected into the plant cell in the form of a virion-like complex and is stably incorporated into the plant genome [7]. One of the imported genes is responsible for the synthesis of opines, a class of low-molecular-weight nitrogen-rich metabolites, that can be utilized as a nutrient only by the bacteria that harbor the Ti plasmid. Other transferred genes cause a vigorous proliferation of infected plant cells that eventually results in the formation of a characteristic gall tumor. Once productive infection is established, Ti plasmids attempt to propagate themselves into the plasmidless bacteria of the same or related species by means of genetic conjugation. It has been shown that the conjugal transfer of Ti plasmids requires the QS phenomenon [8].
Functional significance of QS for the control of Ti plasmid conjugation remains an ecological and evolutionary puzzle. It is widely believed [5,6] that QS controls processes, such as production of toxins and antibiotics, that are either inefficient or devoid of adaptive value if not performed on a population scale. Thus, the fact that the establishment of QS is upstream of the initiation of conjugation seems to imply that plasmids await the critical density of donors to collectively begin transfer to recipients. Since multiple donors cannot cooperate in DNA transfer, the necessity for collective action does not seem to be relevant in our case. Instead, to increase the probability of successful conjugation it would appear beneficial to exceed a certain number of recipients per donor. However the density of plasmidless recipients cannot be estimated using QS since they do not produce the autoinducer. This seemingly paradoxical situation may imply that our understanding of the biological function of QS is not yet complete. Indeed, an alternative function for QS as a sensor of the volume enclosing the bacteria has also been proposed [9]. To answer what bacteria really measure using QS in each particular situation, it is necessary to consider the ecologically relevant conditions of bacterial growth [2].
An experimental approach to this problem is often complicated by the technical difficulty of work in real ecosystems. On the other hand, mathematical modeling can significantly aid and complement experimental methods in answering biological questions that involve spatial and temporal scales of the QS phenomenon. Some aspects of either intracellular [10–13] or population [14–16] dynamics have been mathematically modeled to gain insight into the QS phenomenon in Pseudomonas aeruginosa and Vibrio fischeri. However, because of the lack of detailed molecular information, experimentally testable conclusions on the connections between intracellular and population dynamics have rarely been made. Here we develop a multi-level modeling approach that describes both the intracellular and the population-wide dynamics and allows us to follow the connections between them explicitly. Although much has been learned about the molecular details of the Agrobacterium QS network, it is not always clear what functions they perform. Here we construct a detailed model of the QS network in Agrobacterium and analyze it both quantitatively and qualitatively. We demonstrate that the network possesses properties of the on–off gene expression switch robust to molecular noise. We further develop a population-scale model that incorporates bacterial motion, cell division, and chemical communication while explicitly considering the individual intracellular dynamics of each cell. This allows us to describe the transition to QS on both cellular and population scales and quantitatively predict the values of critical autoinducer concentration and threshold cell density as functions of various intracellular and environmental parameters. Finally, comparing feasibility of the transition to QS in homogeneous medium and biofilm, we present a hypothesis explaining the ecological and evolutionary roles of QS in regulation of Ti plasmid conjugal transfer.
The QS Network and Model Assumptions
All genes that are thought to constitute the QS network are located on the Ti plasmid itself [7]. The entire QS network is controlled upstream by the availability of the plant-produced opines to ensure that energetically expensive conjugation machinery is activated only after the establishment of a successful plant wound infection. Based on the chemical nature of the encoded opines, Ti plasmids are divided into two major types [7], of which we consider only the octopine type. We reconstructed the layout of the QS network for the octopine-type Ti plasmids from the published experimental data (Figure 1). In this plasmid class, octopine molecules that are imported through the cell wall eventually cause activation of transcription from the operon occ [17]. In the model, we assume that octopine is constitutively available at the saturating concentration that results in the maximal rate of occ transcription. The last open reading frame of this operon codes for the QS transcription activator TraR. Binding of TraR to its cognate autoinducer is thought to occur only within a narrow window of time during traR mRNA translation when the newly formed protein chain tightly winds around a single molecule of Agrobacterium autoinducer (AAI) [18–20]. This total engulfment of AAI molecule makes formation of the TraR–AAI complex (TraR*) practically irreversible. Furthermore, TraR protein translated in the absence of AAI is misfolded, insoluble, and unable to bind AAI [18,20]. This has an important consequence in that the rate of production of TraR* depends on the concentrations of traR mRNA and AAI and does not depend on the accumulation of misfolded TraR protein, as explicitly shown in Figure 1. Once formed, TraR* quickly dimerizes to form a stable transcriptionally active TraR* dimer (TraRd) with a relatively short half-life of 35 min [18]. TraRd is capable of activating a number of operons that encode proteins necessary for conjugation. The first open reading frame of the trb operon codes for the acyl-homoserine lactone synthetase TraI, which utilizes two metabolites abundant in the bacterial cell to create AAI [21]. Since our model considers transition to QS in the mostly nutrient-rich, stress-free conditions of an optimized growth medium, we assume that the substrates of TraI are present in excess and their concentrations do not limit the rate of AAI production. Both traR and traI were shown to be expressed at some low constitutive rate even in the absence of octopine [7]. The TraR–TraI couple constitutes the classic QS positive feedback loop found in many Gram-negative bacteria. Additional feedback loops that also involve other components of the QS network are specific for Agrobacterium. Thus, negative control of QS is provided by the antiactivator traM, whose transcription is directly activated by TraRd [22]. TraM effectively sequesters TraRd through the formation of a very stable complex in which TraRd is unable to bind DNA [23,24]. Recently, a number of authors reported that, like TraR, TraM also forms a dimer [25–27]. The stoichiometry of the reaction between TraR and TraM, however, remains controversial [25–27]. In our model we follow the original hypothesis of Swiderska et al. [24], which assumes that the complex consists of one TraRd and one monomer of TraM. This hypothesis is partially supported by Chen et al. [26], who showed that the TraM dimer must dissociate to form a complex with TraR. Under these assumptions we disregard dimerization of TraM as not affecting the network behavior. An additional positive feedback loop arises because TraRd activates transcription of the msh operon, which is a suboperon of occ that contains traR itself.
Figure 1 Quorum-Sensing Network of an Octopine-Type Agrobacterium Ti Plasmid
Blue ellipses represent proteins, red rectangles indicate mRNA species, and green flattened circles denote metabolites. Open circle arrowheads represent enzymatic activation of a reaction or transport, and open triangle arrowheads denote translation. Essential bimolecular reactions are shown explicitly as open squares. S represents two substrates of the autoinducer synthetase TraI, and Ø denotes protein degradation.
Several lines of evidence suggest that active transporters facilitate traffic of the QS signaling molecules through the cell wall in a number of bacterial species including Agrobacterium [28–30]. In our model, we explore the hypothesis that AAI is imported from the environment by an active pump that is also under the transcriptional control of TraRd. Indeed, the msh operon contains five open reading frames (ophABCDE) that encode a putative ABC-type importer whose function is not completely understood but that has been hypothesized to be an active transporter of AAI into the cell [30]. Taking into consideration this uncertainty, the putative AAI importer in the model is denoted simply as Imp.
Results/Discussion
The QS Network Can Operate as a Gene Expression Switch
We first asked whether the intracellular model constructed by us purely from the individual molecular interactions indeed describes known biological properties of the QS network. Although numerical simulation of the full model can answer this question in principle, it does not bring qualitative insight into the system's behavior. Thus, we reduced the full model to only two nonlinear differential equations describing TraRd and intracellular AAI that are readily amenable to qualitative analysis (see Materials and Methods for details). Figure 2 demonstrates that the QS network indeed possesses the property of an environmentally activated gene expression switch. Depending on the value of Ae, the model possesses one or two stable stationary states. The only stationary state at a low extracellular concentration of AAI is characterized by a vanishing number of TraRd molecules. In fact, a copy number of TraRd significantly below one indicates that most of the time transcription factor is not present in the cell at all and the QS network is in the off state. As AAI accumulates in the environment, the position of the Ai nullcline elevates so that first the on state is born at some value of Ae and then the off state disappears at the critical extracellular concentration, thus triggering the transition to the on state. As a result of this transition, the copy number of TraRd dramatically rises from less than one to several hundreds, and the production of AAI increases nearly10-fold.
Figure 2 Nullclines of the Reduced Quorum-Sensing Model at Several Extracellular Concentrations of AAI
Nullclines represent lines on which the respective variable does not change with time (e.g., d[traRd]/dt = 0 on the TraRd nullcline), and their intersections correspond to the stationary states of the model. The TraRd nullcline is shown as a solid black line and the Ai nullcline at three values of Ae—(a) 35 nM, (b) 67 nM, and (c) 142 nM—is shown in color. Positions of stable stationary states are marked by filled circles for the off network state and by filled boxes for the on state. The open circle indicates the unstable steady state of a saddle type.
Sensitivity of the QS network to the changes in the extracellular concentration of AAI and consequently to the population density is defined by passive permeability of the cell wall to AAI and any active transport, if existent. Passive diffusion largely depends on the physical properties of the AAI molecule (e.g., length of the hydrocarbon chain) and cannot be controlled by the QS network. On the other hand, our model shows that an active importer, such as putative transporter Imp, can exert significant control over the transition threshold. Indeed, we found that the critical extracellular AAI concentration
depends linearly on the AAI–Imp dissociation constant
for a wide range of
values (Figure 3). One may also speculate that the availability of such a pump could give the Ti plasmid evolutionary flexibility in the changing environment since a large multi-subunit protein complex can quickly accumulate mutations that potentially affect its transporter properties and consequently alter the transition threshold.
Figure 3 Dependence of the Critical Extracellular Concentration of Autoinducer on the Autoinducer–Transporter Dissociation Coefficient = k5/k4 Predicted by the Intracellular Model
The solid line is fitted to the computed data points indicated by the filled squares. The value of
= 0.592 nM used throughout the simulations reported in this paper is shown as an empty square.
TraM Is Necessary for the Existence of the Off State
Our model clearly demonstrates that the negative feedback provided by traM antiactivator is essential for the very existence of the off state in the QS network. Historically, traM was identified as a gene whose loss of function resulted in constitutive conjugation among the bacteria [22,31]. Indeed, the removal of the TraRd–TraM reaction from the model destroys the bistability of the QS network and permits only the on stationary state at any extracellular concentration of AAI. The model explains the mechanism of TraM action through the mutual exclusion between TraM and TraRd. In the off state, an appreciable number of TraM molecules (over 100 copies per cell) ensures that TraRd does not accumulate. During the transition to QS, rapidly created TraR* dimers sequester the TraM pool and reduce it to less than one molecule per cell in the on state. Production rate of the traM mRNA remains high in the on state, but the substantial surplus of TraRd guarantees that all newly synthesized TraM molecules are quickly sequestered. If this mechanism is inactivated by a mutation, the QS network amplifies any small number of TraRd complexes to the high on level even in the absence of exogenous AAI. This model prediction has been verified by genetic analysis. Strain K588, which produces AAI constitutively, was found to contain a point mutation in traM that renders the TraM protein inactive. Complementation of K588 with an intact traM, which was administered on a separate plasmid, restored the wild-type phenotype completely (L. H. Z., unpublished data).
Sensitivity Analysis Reveals Critical Components of the QS Network
Additional information about the contribution of various components of the QS network to the control of the TraRd copy number can be gained from the analysis of sensitivity of the stationary concentration of TraRd to variation in the network parameters. One of the popular methods to assess sensitivity of molecular networks [32,33] is based on metabolic control analysis [34]. We computed the sensitivity of TraRd concentration to all network parameters in the off state at two values of Ae, far from the transition to QS (Ae = 0) and in the bistable region (Ae = 42 nM), as well as of the on state, deep into the region of its stability at Ae = 120 nM. The three values of sensitivity coefficient calculated for each parameter as described in the Materials and Methods are given in Table S1. As expected, sensitivity to most parameters peaks in the on–off transition region. Processes responsible for the production and degradation of traR mRNA exert the most powerful control over the copy number of TraRd. However, the contribution of the positive feedback loop due to the transcription of the msh operon becomes significant in comparison with the input from the octopine-induced transcription of occ only during and after the transition to the on state. Interestingly, the sensitivity analysis shows that various components of the QS network contribute very differently into the control of TraRd in the on and off states. Thus, TraRd degradation is unimportant in the off state, when the copy number of the transcription activator is controlled by TraM, and acquires prominence in the on state, when sequestration by TraM becomes irrelevant. Likewise the Imp plays no role in the off state but exerts control over TraRd during the transition to QS. In contrast, the subsystem responsible for the production of TraI looses significance in the on state, when a large pool of this protein accumulates in the cell.
The QS Network is Robust to Molecular Noise
We next set out to investigate whether the sharp switch-like behavior of the QS network predicted by the deterministic model is preserved when fluctuations in the number of molecules are considered. To answer this question, we simulated the full intracellular model stochastically (see Materials and Methods) and compared the results with the deterministic solution. Figure 4 demonstrates that there is remarkably good agreement between the deterministic model and the behavior of the stochastic intracellular model averaged over a long observation time. The same results were obtained by averaging the behavior over many independent realizations.
Figure 4 Transition to Quorum Sensing in the Stochastic Model of Intracellular Dynamics
TraRd concentration averaged over 6 × 106 s (filled squares) is plotted against extracellular concentration of AAI. The prediction of the deterministic model is shown as a solid line for comparison.
Undetectable in the average value, fluctuations in the copy number of TraRd deserve special attention as they can dramatically affect the network behavior. In particular, the off state predicted by our model is biologically meaningful only if the fluctuations of TraRd are controlled as tightly as its average concentration. Rare but significant departures from the off state in the absence of the extracellular AAI signal (Ae = 0) would result in spontaneous activation of genes encoding conjugation machinery under unfavorable conditions, bringing about selective penalty for the host bacteria. We performed a detailed analysis of the stability of the off state by investigating the TraRd probability density function at varying concentrations of extracellular AAI. The TraRd probability density function peaks at zero in the off state and decreases exponentially with the number of dimer molecules. At Ae = 0, where the average TraRd concentration is 0.034, fluctuations of TraRd are practically negligible. Thus, two copies per cell are found with probability 0.001, while three copies are found with probability of only 6.6 × 10−5. This demonstrates that in the absence of external autoinducer the QS network maintains robust control of the fluctuations in the TraRd copy number and effectively prevents spontaneous transitions to the on state. At the same time, other molecular species whose copy number is not controlled by the QS network, e.g., TraI, TraM, and intracellular AAI, exhibit large-amplitude fluctuations around their average levels, in agreement with earlier reports for other molecular networks [35]. As Ae increases, the fluctuations of TraRd also grow. At extracellular AAI concentrations in the range of 40–60 nM the network visits on and off states intermittently. For Ae ≥ 70 nM, the model is found solely in the on state.
Collective Robustness of the Population Transition to QS
We first investigated the transition to QS in the simplest case of a population growing exponentially in a homogeneous liquid medium. The stochastic population model was simulated to imitate actual population dynamics of motile bacteria in a small volume element (Ve = 10−5 ml) of a liquid medium bulk. During approximately 7 h of simulation time, the population grew more than 100 times to reach the maximal density N = 2.52 × 109 cells/ml. Figure 5 demonstrates the transition to QS in this system as detected by monitoring the intracellular dynamics of randomly selected bacterial cells. Observation of an individual bacterium over time shows that after the initial period of quiescent growth in the off state, TraRd exhibits sudden and fast switches to the on state (Figure 5A). These phases with variable duration and TraRd abundance alternate with the off phases in a random pattern until the cell finally settles in the on state. While the intracellular dynamics of individual bacteria appears to be totally erratic, the population average demonstrates an orderly, gradual transition to QS (Figure 5B). Moreover, ensemble-averaged behavior of the stochastic population model resembles that of the deterministic model (Figure 5C). For example, notice that in both cases prior to the transition to the on state TraM temporarily undergoes a maximum. To be precise, we can define the transition to QS to occur at the point where the lines for TraRd and TraM abundance intersect in Figure 5C. Then deterministic and stochastic models predict transition at the same critical value of extracellular autoinducer concentration,
= 580 ± 3 nM. Thus, the collective transition to QS in the spatially homogeneous bacterial population, as defined by the ensemble average of intracellular concentrations, is robust to the variability of individual bacteria and can be described by a simple deterministic model with a reasonable accuracy. Interestingly, the value of
predicted by our model is of the same order of magnitude as the experimentally found values (150 and 900 nM) for the two QS systems of Serratia liquefaciens [36].
Figure 5 Transition of a Model Bacterial Population to Quorum Sensing in a Homogeneous Liquid Medium
Intracellular concentrations of TraRd (thick red line) and TraM (thin blue line, filled circles) are plotted against the population density that exponentially grows with time.
(A) Dynamics of an individual cell in the stochastic population model.
(B) Dynamics of the stochastic population model averaged over ten bacteria.
(C) Behavior of the deterministic population model.
Transition to QS in the Growing Population is Dynamic
The value of the critical AAI concentration
for the exponentially growing spatially homogeneous population is more than ten times larger than the one that follows from the case in Figure 4. This discrepancy arises because in the computation of both stochastic and deterministic values presented in Figure 4 it is implicitly assumed that a bacterium remains in the medium with a given extracellular AAI concentration for a practically infinite amount of time. In the growing population, Ae rises exponentially together with the population density and the intracellular network does not have sufficient time to adjust to the extracellular environment. As a result, the transition to QS in a growing population always occurs in conditions far from stationary, and the values of the critical AAI concentration and the cell density threshold depend on the parameters of the population growth. The duration of the cell cycle that defines the population growth rate is one of the key parameters. The lower the population growth rate, the smaller the required
. In the unrealistic limit of an infinitely slowly growing population, the transition is guaranteed to occur at the value found for the static intracellular model (
≈ 48 nM). Figure 6 shows that the difference between the actual critical concentration
and its asymptotic value
first decreases dramatically with the increase in the cell cycle duration Tc and then slowly vanishes in accordance with an apparent power law
.
Figure 6 Critical Extracellular Concentration of Autoinducer Depends on the Growth Rate of a Bacterial Population
The difference between the actual critical concentration Aec computed using the deterministic population model and the value predicted by the intracellular model Aec,∞ is plotted against the duration of cell cycle Tc (filled circles). As shown by the linear fit (solid line), Aec approaches Aec,∞ according to the power law Aec − Aec,∞∝T − 0.61c.
The threshold population density depends on
and a number of other population parameters, e.g., the spatial distribution of bacteria in the habitat (see below). The value of the cell density Nc that has to be reached to achieve QS under specific environmental conditions can provide useful information on the possibility of achieving QS in an environmental niche. For example, experimental observation of the transition of an Agrobacterium population to QS in liquid medium can be problematic because of the large value of the predicted density threshold (
≈ 2.0 × 109 cells/ml by the stochastic model and
≈ 2.82 × 109 cells/ml by the deterministic approach). In response to depletion of the nutrients in the liquid medium, a typical Agrobacterium population begins the transition to the stationary growth phase at or even below the predicted N c values [37]. The ensuing general stress response activates transcription of lactonase attM, which efficiently destroys autoinducer molecules and abrogates the transition to QS [37–39]. Thus, even in the nutrient-rich conditions of the optimized liquid medium it is not unlikely that the transition to the stationary phase precedes and, therefore, precludes the transition to QS.
Spatial Heterogeneity Reduces Population Density Threshold
We interpreted the previous finding as an indication that the simplest experimental scenario of growth in a spatially homogeneous liquid medium is not ecologically relevant and the QS network is not “tuned on” to support the transition to QS in such conditions. Indeed, in nature, the transition to QS and subsequent bacterial conjugation take place in the thin but dense biofilm on the surface of a Ti-plasmid-induced plant tumor. In laboratory conditions, the experiments on detection and quantification of conjugation are normally performed in the quasi two-dimensional environment of polymer filters that provide bacteria with firm support for attachment and conjugation [40]. We therefore modified the stochastic population model to simulate an immotile bacterial population that grows exponentially on the surface of a filter that is placed at the bottom of a Petri dish and covered with 1 cm of an unstirred liquid medium. We also roughly estimated the transition to QS in the biofilm growing on the interface between the plant tumor and the soil using the same spatial layout but with the AAI diffusion coefficient reduced by a factor of ten. Although such simplistic approximation may not adequately represent complex conditions in the soil, it reveals a qualitative trend relevant for the goal of our study. In the absence of mechanical mixing and active bacterial motion, AAI excreted into the extracellular space freely diffuses into the medium and forms a sharp concentration gradient. As can be seen in the simulation results (Figure 7A), the AAI concentration drops exponentially from hundreds of nanomoles per liter in the plane of biofilm to barely detectable values on the medium boundary. Lower conductivity of the medium results in a steeper gradient, faster accumulation of AAI in the plane of biofilm, and reduced losses of AAI into the environment. Monitoring of the population-averaged intracellular variables shows that the transition to QS in the biofilm is not appreciably different from that in the bulk of the liquid medium (Figure 7B) and takes place at the similar critical extracellular AAI concentration
≈ 615 nM. In contrast, the threshold population densities, 4.43 × 108 cells/cm2 for the faster and 1.33 × 108 cells/cm2 for the slower diffusion, are considerably lower than the homogeneous bulk value. From analysis of electron microscopy images of Agrobacterium biofilms [41], we calculated a typical cell density to be between 0.5 and 5 cells/μm2 (0.5–5.0 × 108 cells/cm2). The values predicted by our model are thus well within the natural range and can be readily reached by a biofilm growing in the optimal nutrition conditions, e.g., on the feeding surface of a plant tumor.
Figure 7 Transition to Quorum Sensing in a Model Bacterial Population Growing as an Attached Biofilm
(A) Spatial gradient of autoinducer created by the biofilm after 7 h of growth at two different values of AAI diffusion coefficient: (a) 10−5 cm2/s and (b) 10−6 cm2/s. The x-axis is perpendicular to the plane of the biofilm, which is positioned at x = 0.
(B) Intracellular copy numbers of TraRd and TraM averaged over 20 randomly selected bacteria versus the population density for the slower diffusion of AAI.
Conclusions
Using the phenomenon of bacterial QS as an enlightening example, we investigated the relationship between the dynamics of an intracellular molecular network and the population-wide phenotype that is controlled by this network. We first reconstructed the QS network of Agrobacterium Ti plasmids from experimental data and demonstrated that the network actually possesses the properties required for a gene expression switch, such as high sensitivity to environmental control and robustness to molecular noise. We then developed a stochastic model of a bacterial population to explore the transition to QS in a number of simple experimental scenarios, namely, during exponential growth in homogeneous liquid bulk and in an attached biofilm.
One of the long-standing questions in this field is whether all cells in a population produce autoinducer at the same rate and experience transition to QS simultaneously [2]. Our results predict that even in spatially homogeneous medium there is dramatic variability between the cells in the level of transcription factor and therefore the rate of production of autoinducer. This variability presumably arises from the fact that in a single cell the transition to QS, essentially an autocatalytic process, vastly amplifies the natural stochasticity of gene expression inherent in bacteria [42–44]. As a result, individual cells experience transition to QS in a wide range of extracellular concentrations of autoinducer and at varied levels of population density. Such nongenetic variability in the behavior of individual bacterial cells has been demonstrated to be advantageous for population survival in some cases, e.g., in the phenomenon of bacterial persistence under antibiotic treatment [45,46]. In our system, this variability may improve chances of Ti plasmid propagation within the biofilm when the density of donor cells falls short of the critical value. Although no direct evidence for this has been reported, such variability can explain why the appearance of conjugation between donors and recipients is observed emerging gradually over an extended cell density range [47]. Additional factors responsible for the widening of the transition range may result from spatial heterogeneity within the biofilm. In the present stochastic model we disregarded inhomogeneity of cell distribution in the plane of the biofilm. In future research it would be interesting to explore the influence of spatial heterogeneity typical of natural biofilms on the transition to QS.
Despite large variability between individual cells, a single population-wide value can be meaningfully defined for both critical concentration of autoinducer and the threshold population density for the transition to QS if the intracellular dynamics of individual cells is averaged over the population. These values are robust to stochasticity of individual bacteria and can be predicted with sufficient accuracy by a deterministic population model provided that spatial heterogeneity is insignificant. Our model demonstrates that when driven by exponential growth, the population transition to QS does not occur under steady state conditions and therefore the critical values depend on the parameters of population growth. Thus, the critical concentration of autoinducer depends on the growth rate.
Importantly, bacteria grown in different experimental conditions require different population densities to reach QS. Transition to QS in the bulk of the liquid medium appears to be the least favorable and requires much higher population density than transition in a biofilm. This result suggests potential ecological and evolutionary significance of the QS phenomenon for Ti plasmid propagation. In natural conditions a bacterial population dwells in a heterogeneous habitat with both bulk (e.g., soil) and the attracting surface (the plant–soil interface). Given the difference in cell density thresholds, it is likely that the transition to QS occurs in the surface-attached biofilm but not in the bulk. Therefore, it is tempting to speculate that the Ti plasmid utilizes QS to detect whether its bacterial host is firmly attached to the biofilm in close proximity to the source of nutrition (octopine) and, therefore, is in favorable conditions to initiate the conjugal transfer. Contribution of QS to the maturation of biofilms has been suggested for a number of species (for review see [2]). Interestingly, in our case QS does not influence any morphogenetic process in the biofilm but rather appears to detect the condition when the biofilm is sufficiently advanced in its formation. The requirement of biofilm formation for conjugation might be explained by the fact that solid support was found to be essential for the success of Ti plasmid transfer between the bacterial cells [47]. In addition, location inside a biofilm implies a high density of neighbors and therefore a high probability of finding a conjugation partner in close proximity.
Thus, our systems-level model provides experimentally testable quantitative predictions regarding both the dynamics of the intracellular control network and the population-wide characteristics of the transition to QS. Experimental verification of these predictions can be achieved, for example, by using a combination of classical genetic techniques and modern fluorescent confocal microscopy. One of the less obvious model predictions amenable to this approach is the existence of an appreciable intracellular pool of TraM in the off state. Insertion of a fluorescent reporter, like the green fluorescent protein or one of its derivatives, into some or all operons controlled by TraRd would result in the development of a sensitive gage to directly observe the dynamics of the transition to QS in vivo. With such a reporter it should be possible to observe bistability in the transition region by simultaneously monitoring populations of cells in the off and on states, e.g., as has recently been done in a study of the lactose utilization network [48]. Ability to monitor in vivo the copy number of TraRd or, complementary to it, concentration of TraM in a statistically significant number of bacterial cells would also allow one to measure the critical concentration of AAI and the threshold population density. The dependence of the autoinducer critical concentration on the growth rate is also a testable prediction. Our model predicts that if the duration of the cell cycle increases from 1 h to 2 h,
should drop almost by a factor of two.
In addition to producing these predictions, our study suggests an answer to the long-standing question regarding the ecological and evolutionary role of the QS phenomenon in the genetic conjugation of Ti plasmids. Finally, our analysis demonstrates how computational modeling connects multiple scales of biological phenomena, from the level of molecular networks to that of multicellular populations.
Materials and Methods
Mathematical formulation of intracellular dynamics.
We formulated the chemical kinetics of the Agrobacterium QS network as a system of 16 mass-action rate law equations with 30 chemical constants and the extracellular concentration of AAI as a free parameter. Complex processes of transcription and translation are represented in our formulation by cumulative reactive mechanisms:
with three and one effective reaction constants, respectively. The action of the putative AAI importer is modeled according to the standard enzymatic mechanism:
where Ae and Ai are the extracellular and intracellular concentrations of AAI, respectively. Detailed formulation of each model equation is given in Table 1.
Table 1 Model Variables and Equations
Kinetic constants.
We extracted a number of crucial model parameters, e.g., lifetime of TraRd and reaction constant of TraRd and TraM, directly from the literature. Many parameters of a more general nature, such as velocities of transcription and translation, are not reported for our system and were estimated based on values obtained for other prokaryotic systems. We estimated the average volume of a bacterial cell Vb to be 1.4 × 10−13 cm3 (a cylinder with diameter 0.3 μm and length 2 μm) using high-quality electronic microscopy images of Agrobacterium colonies [41]. Based on this value we converted all concentrations from moles per liter (M) to molecules per cell (one molecule per cell corresponds to approximately 12 nM) and adjusted kinetic constants appropriately. A full set of the constants used in this work is presented in Table S1. To ensure validity of the population model predictions, it is essential to correctly evaluate the level of AAI production by a bacterial cell. We achieved this by fitting our model to the experimental data obtained for the conjugation constitutive mutant strain K588 [37] (Figure 8). This mutant produces copious amounts of AAI that can be readily detected in the medium throughout population growth (see Results and Discussion).
Figure 8 Extracellular Concentration of AAI in the Culture of the TraM-Defective Mutant Strain K588 versus Time in the Exponential Growth Phase
The model (solid line) is fitted to the experimental data (filled squares). All parameters are as in Table S1, except k
11, which is set to zero to model the inability of mutated TraM to sequester TraRd.
Sensitivity analysis.
We performed a local analysis of sensitivity of the stationary states of the full deterministic model to variation of the model parameters using the formalism of the metabolic control analysis [34] that recently has been extended for the analysis of signal transduction and gene expression networks [49,50]. In this framework, the sensitivity of the stationary concentrations of intracellular components Ci to variation of the model parameters kj is given by the concentration response coefficients:
These coefficients were computed using Jarnac, which was integrated into Systems Biology Workbench, a freely available software platform [51].
Reduction of the full model.
We used standard methods of chemical kinetics to reduce the dimensionality of the full QS network model to only two equations describing the dynamics of TraRd and intracellular AAI. We first eliminated variables describing vacant and occupied TraRd binding sites (see Table 1), assuming that the quasistationary approximation holds for these variables. This assumption results in the effective Michaelis–Menten approximation for all transcription events. We simultaneously introduced corresponding Michaelis–Menten constants, which in this case are the coefficients of dissociation for TraRd binding to respective cis-regulatory elements, e.g.,
. Thus, for example, the equation for traM mRNA becomes:
where D represents concentration of TraRd as in Table 1. In a similar fashion we eliminated the variable imp_A and introduced another Michaelis–Menten constant,
. We then applied quasistationary approximation to all three mRNA species as well as to the proteins Imp, TraI, and TraM and the monomeric complex TraR*. This allowed us to express all of the remaining variables through the concentrations of TraRd and AAI. For example, from the previous equation the quasistationary concentration of traM mRNA is:
Finally, we obtained the reduced model with two equations:
Population model.
Since the full model for the intracellular dynamics is expressed entirely in mass-action rate law equations, it can be simulated with both deterministic and stochastic methods. To model the transition to QS in the exponentially growing bacterial population deterministically, we complemented the intracellular model with two equations describing the dynamics of cell density N and extracellular autoinducer Ae:
where V′ is the ratio of the bacterial cell volume Vb to the full volume occupied by the population suspended in the medium Ve (1 ml) and the expression in square brackets represents the exchange of AAI between a bacterium and the environment (see Table 1 for details).
In the stochastic formulation, each bacterium is represented by a separate agent endowed with an independent stochastic realization of the QS network model. We assumed that bacteria can either randomly move in the medium (planktonic form) with effective diffusion coefficient Db = 10−6 cm2/s [52] or remain immotile (attached form). Both forms exchange AAI with the surrounding medium according to the model equations. In the extracellular environment AAI is represented by a spatially dependent continuous concentration field and its spatiotemporal evolution is modeled by the diffusion equation. While the diffusion coefficient of AAI is not known, from the size of the molecule we estimated that AAI diffuses in the liquid bacterial culture with DA = 10−5 cm2/s. Cells periodically divide, resulting in two replicas that are exactly identical at the moment of division and diverge thereafter. In both deterministic and stochastic representations average cell cycle is 1 h (α = 1.9 × 10−4 s−1) according to the estimate based on our experimental data.
Computational realization.
All deterministic simulations were performed with Matlab (MathWorks, Natick, Massachusetts, United States). The stochastic model of the intracellular network was simulated with the exact Gillespie algorithm [53] using public-domain cell-modeling software Cellware [54,55]. To model stochastic population dynamics we developed a dedicated parallel software platform [56] that manages dynamically growing bacterial populations and utilizes Cellware to compute the intracellular dynamics for each cell agent. The platform also simulates cell motion, exchange of AAI with the medium, and diffusion of AAI in the medium. The software was implemented in C++ and MPI to run on a commodity Linux cluster. All codes used in this work are freely available on request.
Supporting Information
Table S1 Model Parameters
(121 KB DOC)
Click here for additional data file.
Accession Numbers
The NCBI Entrez Protein (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=Protein) accession numbers for the proteins discussed in this paper are TraI (AAB95104), TraM (AAC28120), and TraR (AAC28121). The NCBI Entrez Gene (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene) accession numbers for the genes discussed in this paper are ophABCDE (1224221–1224225), traI (1224280), traM (1224219), and traR (1224220).
We acknowledge many people from the Bioinformatics Institute who contributed their help and advice. In particular we are grateful to the present and former members of the Cellware team: Pawan Dhar, Li Ye, Tan Chee Meng, Wu Song, and Sandeep Somani for their indispensable help with incorporation of their software into our computation platform. We thank Arun Krishnan, Francis Tang, and Stephen Wong for help with parallel programming and cluster computing. We also acknowledge inspirational and fruitful discussions with Guna Rajagopal and Santosh Mishra. We thank Vinay Sarathy and Jeffrey Pang for their contribution during the period of their SMA student internships. Many thanks go to Herbert Sauro and Brian Ingalls, who provided invaluable advice and software for the sensitivity analysis. ABG gratefully acknowledges the help of Baltazar Aguda, who read the manuscript and provided useful suggestions. This work was supported by the Agency for Science, Technology and Research of Singapore.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. ABG and LHZ conceived and designed the study. HBZ performed the experiments. DJT, KBW, and TL developed software and performed the numerical computations. ABG designed and analyzed the models and wrote the paper.
A previous version of this article appeared as an Early Online Release on August 8, 2005 (DOI: 10.1371/journal.pcbi.0010037.eor).
Abbreviations
AAI
Agrobacterium autoinducer
Imp
Agrobacterium autoinducer importer
QSquorum sensing
TraRdTraR–Agrobacterium autoinducer complex dimer
TraR*TraR–Agrobacterium autoinducer complex
==== Refs
References
Hartwell LH Hopfield JJ Leibler S Murray AW 1999 From molecular to modular cell biology Nature 402 C47 C52 10591225
Parsek MR Greenberg EP 2005 Sociomicrobiology: The connections between quorum sensing and biofilms Trends Microbiol 13 27 33 15639629
von Bodman SB Bauer WD Coplin DL 2003 Quorum sensing in plant-pathogenic bacteria Annu Rev Phytopathol 41 455 482 12730390
Henke JM Bassler BL 2004 Bacterial social engagements Trends Cell Biol 14 648 656 15519854
Fuqua C Greenberg EP 2002 Listening in on bacteria: Acyl-homoserine lactone signalling Nat Rev Mol Cell Biol 3 685 695 12209128
Fuqua C Parsek MR Greenberg EP 2001 Regulation of gene expression by cell-to-cell communication: Acyl-homoserine lactone quorum sensing Annu Rev Genet 35 439 468 11700290
Zhu J Oger PM Schrammeijer B Hooykaas PJ Farrand SK Winans SC 2000 The bases of crown gall tumorigenesis J Bacteriol 182 3885 3895 10869063
Zhang L Murphy PJ Kerr A Tate ME 1993
Agrobacterium conjugation and gene regulation by N-acyl-L-homoserine lactones Nature 362 446 448 8464475
Redfield RJ 2002 Is quorum sensing a side effect of diffusion sensing? Trends Microbiol 10 365 370 12160634
Dockery JD Keener JP 2001 A mathematical model for quorum sensing in Pseudomonas aeruginosa
Bull Math Biol 63 95 116 11146885
James S Nilsson P James G Kjelleberg S Fagerstrom T 2000 Luminescence control in the marine bacterium Vibrio fischeri: An analysis of the dynamics of lux regulation J Mol Biol 296 1127 1137 10686109
Viretta AU Fussenegger M 2004 Modeling the quorum sensing regulatory network of human-pathogenic Pseudomonas aeruginosa
Biotechnol Prog 20 670 678 15176867
Cox CD Peterson GD Allen MS Lancaster JM McCollum JM 2003 Analysis of noise in quorum sensing OMICS 7 317 334 14583119
Ward JP King JR Koerber AJ Williams P Croft JM 2001 Mathematical modelling of quorum sensing in bacteria IMA J Math Appl Med Biol 18 263 292 11817745
Chopp DL Kirisits MJ Moran B Parsek MR 2002 A mathematical model of quorum sensing in a growing bacterial biofilm J Ind Microbiol Biotechnol 29 339 346 12483476
Nilsson P Olofsson A Fagerlind M Fagerstrom T Rice S 2001 Kinetics of the AHL regulatory system in a model biofilm system: How many bacteria constitute a “quorum”? J Mol Biol 309 631 640 11397086
Wang L Helmann JD Winans SC 1992 The A. tumefaciens transcriptional activator OccR causes a bend at a target promoter, which is partially relaxed by a plant tumor metabolite Cell 69 659 667 1586946
Zhu J Winans SC 1999 Autoinducer binding by the quorum-sensing regulator TraR increases affinity for target promoters in vitro and decreases TraR turnover rates in whole cells Proc Natl Acad Sci U S A 96 4832 4837 10220379
Zhang RG Pappas T Brace JL Miller PC Oulmassov T 2002 Structure of a bacterial quorum-sensing transcription factor complexed with pheromone and DNA Nature 417 971 974 12087407
Zhu J Winans SC 2001 The quorum-sensing transcriptional regulator TraR requires its cognate signaling ligand for protein folding, protease resistance, and dimerization Proc Natl Acad Sci U S A 98 1507 1512 11171981
Hwang I Li PL Zhang L Piper KR Cook DM 1994 TraI, a LuxI homologue, is responsible for production of conjugation factor, the Ti plasmid N-acylhomoserine lactone autoinducer Proc Natl Acad Sci U S A 91 4639 4643 8197112
Hwang I Cook DM Farrand SK 1995 A new regulatory element modulates homoserine lactone-mediated autoinduction of Ti plasmid conjugal transfer J Bacteriol 177 449 458 7814335
Hwang I Smyth AJ Luo ZQ Farrand SK 1999 Modulating quorum sensing by antiactivation: TraM interacts with TraR to inhibit activation of Ti plasmid conjugal transfer genes Mol Microbiol 34 282 294 10564472
Swiderska A, Berndtson AK, Cha MR, Li L, Beaudoin GM 3rd, et al. 2001 Inhibition of the Agrobacterium tumefaciens TraR quorum-sensing regulator. Interactions with the TraM anti-activator J Biol Chem 276 49449 49458 11687576
Vannini A Volpari C Di Marco S 2004 Crystal structure of the quorum-sensing protein TraM and its interaction with the transcriptional regulator TraR J Biol Chem 279 24291 24296 15044488
Chen G Malenkos JW Cha MR Fuqua C Chen L 2004 Quorum-sensing antiactivator TraM forms a dimer that dissociates to inhibit TraR Mol Microbiol 52 1641 1651 15186414
Qin Y Smyth AJ Su S Farrand SK 2004 Dimerization properties of TraM, the antiactivator that modulates TraR-mediated quorum-dependent expression of the Ti plasmid tra genes Mol Microbiol 53 1471 1485 15387823
Welch M Todd DE Whitehead NA McGowan SJ Bycroft BW 2000 N-acyl homoserine lactone binding to the CarR receptor determines quorum-sensing specificity in Erwinia
EMBO J 19 631 641 10675332
Pearson JP van Delden C Iglewski BH 1999 Active efflux and diffusion are involved in transport of Pseudomonas aeruginosa cell-to-cell signals J Bacteriol 181 1203 1210 9973347
Fuqua C Winans SC 1996 Localization of OccR-activated and TraR-activated promoters that express two ABC-type permeases and the traR gene of Ti plasmid pTiR10 Mol Microbiol 20 1199 1210 8809772
Fuqua C Burbea M Winans SC 1995 Activity of the Agrobacterium Ti plasmid conjugal transfer regulator TraR is inhibited by the product of the traM gene J Bacteriol 177 1367 1373 7868612
Ortega F Ehrenberg M Acerenza L Westerhoff HV Mas F 2002 Sensitivity analysis of metabolic cascades catalyzed by bifunctional enzymes Mol Biol Rep 29 211 215 12241060
Ingalls BP Sauro HM 2003 Sensitivity analysis of stoichiometric networks: An extension of metabolic control analysis to non-steady state trajectories J Theor Biol 222 23 36 12699732
Kacser H Burns JA 1973 The control of flux Symp Soc Exp Biol 27 65 104 4148886
Morohashi M Winn AE Borisuk MT Bolouri H Doyle J 2002 Robustness as a measure of plausibility in models of biochemical networks J Theor Biol 216 19 30 12076125
Eberl L Winson MK Sternberg C Stewart GSAB Christiansen G 1996 Involvement of N-acyl-L-homoserine lactone autoinducers in controlling the multicellular behaviour of Serratia liquefaciens
Mol Microbiol 20 127 136 8861211
Zhang HB Wang C Zhang LH 2004 The quormone degradation system of Agrobacterium tumefaciens is regulated by starvation signal and stress alarmone (p)ppGpp Mol Microbiol 52 1389 1401 15165241
Dong YH Wang LH Xu JL Zhang HB Zhang XF 2001 Quenching quorum-sensing-dependent bacterial infection by an N-acyl homoserine lactonase Nature 411 813 817 11459062
Zhang HB Wang LH Zhang LH 2002 Genetic control of quorum-sensing signal turnover in Agrobacterium tumefaciens
Proc Natl Acad Sci U S A 99 4638 4643 11930013
Levin RA Farrand SK Gordon MP Nester EW 1976 Conjugation in Agrobacterium tumefaciens in the absence of plant tissue J Bacteriol 127 1331 1336 783141
Kaláb M 2004 Agrobacterium tumefaciens Lund (Sweden) SciMAT Available: http://anka.livstek.lth.se:2080/microscopy/Agrobact.htm . Accessed 15 August 2005.
Rosenfeld N Young JW Alon U Swain PS Elowitz MB 2005 Gene regulation at the single-cell level Science 307 1962 1965 15790856
Elowitz MB Levine AJ Siggia ED Swain PS 2002 Stochastic gene expression in a single cell Science 297 1183 1186 12183631
Ozbudak EM Thattai M Kurtser I Grossman AD van Oudenaarden A 2002 Regulation of noise in the expression of a single gene Nat Genet 31 69 73 11967532
Balaban NQ Merrin J Chait R Kowalik L Leibler S 2004 Bacterial persistence as a phenotypic switch Science 305 1622 1625 15308767
Kussell E Kishony R Balaban NQ Leibler S 2005 Bacterial persistence: A model of survival in changing environments Genetics 169 1807 1814 15687275
Piper KR Farrand SK 1999 Conjugal transfer but not quorum-dependent tra gene induction of pTiC58 requires a solid surface Appl Environ Microbiol 65 2798 2801 10347084
Ozbudak EM Thattai M Lim HN Shraiman BI van Oudenaarden A 2004 Multistability in the lactose utilization network of Escherichia coli
Nature 427 737 740 14973486
Kholodenko BN Kiyatkin A Bruggeman FJ Sontag E Westerhoff HV 2002 Untangling the wires: A strategy to trace functional interactions in signaling and gene networks Proc Natl Acad Sci U S A 99 12841 12846 12242336
Bruggeman FJ Westerhoff HV Hoek JB Kholodenko BN 2002 Modular response analysis of cellular regulatory networks J Theor Biol 218 507 520 12384053
Sauro HM Hucka M Finney A Wellock C Bolouri H 2003 Next generation simulation tools: The Systems Biology Workbench and BioSPICE integration OMICS 7 355 372 14683609
Berg HC 1993 Random walks in biology, expanded ed Princeton (New Jersey) Princeton University Press 152 p.
Gillespie DT 1977 Exact stochastic simulation of coupled chemical reactions J Phys Chem 81 2340 2361
Dhar P Meng TC Somani S Ye L Sakharkar K 2004 Grid Cellware: The first grid-enabled tool for modeling and simulating cellular processes Bioinformatics 21 1284 1287 15546936
Dhar P Meng TC Somani S Ye L Sairam A 2004 Cellware—A multi-algorithmic software for computational systems biology Bioinformatics 20 1319 1321 14871872
Toh DJ Tang F Lee T Sarda D Krishnan A 2004 Parallel computing platform for the agent-based modeling of multicellular biological systems Liew KM Shen H See S Parallel and distributed computing: Applications and technologies Heidelberg Springer 5 9
|
16170413
|
PMC1214540
|
CC BY
|
2021-01-05 09:19:23
|
no
|
PLoS Comput Biol. 2005 Sep 16; 1(4):e37
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010037
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1617263210.1371/journal.pcbi.001003905-PLCB-RA-0050R4plcb-01-04-03Research ArticleBioinformatics - Computational BiologyGenetics/Gene ExpressionSaccharomycesAnalysis of a Splice Array Experiment Elucidates Roles of Chromatin Elongation Factor Spt4–5 in Splicing Analysis of a Splice Array ExperimentXiao Yuanyuan 1Yang Yee H 2Burckin Todd A 3Shiue Lily 4Hartzog Grant A 3Segal Mark R 1*1 Department of Epidemiology and Biostatistics, Center for Bioinformatics and Molecular Biostatistics, University of California, San Francisco, California, United States of America
2 Department of Medicine, Center for Bioinformatics and Molecular Biostatistics, University of California, San Francisco, California, United States of America
3 Department of Molecular, Cell, and Developmental Biology, University of California, Santa Cruz, California, United States of America
4 Department of Molecular, Cell and Developmental Biology, University of California, Santa Cruz, California, United States of America
Bourne Philip E EditorUniversity of California, San Diego, United States of America.* To whom correspondence should be addressed. E-mail: [email protected] 2005 16 9 2005 1 4 e399 3 2005 8 8 2005 Copyright: © 2005 Xiao et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Splicing is an important process for regulation of gene expression in eukaryotes, and it has important functional links to other steps of gene expression. Two examples of these linkages include Ceg1, a component of the mRNA capping enzyme, and the chromatin elongation factors Spt4–5, both of which have recently been shown to play a role in the normal splicing of several genes in the yeast Saccharomyces cerevisiae. Using a genomic approach to characterize the roles of Spt4–5 in splicing, we used splicing-sensitive DNA microarrays to identify specific sets of genes that are mis-spliced in ceg1, spt4, and spt5 mutants. In the context of a complex, nested, experimental design featuring 22 dye-swap array hybridizations, comprising both biological and technical replicates, we applied five appropriate statistical models for assessing differential expression between wild-type and the mutants. To refine selection of differential expression genes, we then used a robust model-synthesizing approach, Differential Expression via Distance Synthesis, to integrate all five models. The resultant list of differentially expressed genes was then further analyzed with regard to select attributes: we found that highly transcribed genes with long introns were most sensitive to spt mutations. QPCR confirmation of differential expression was established for the limited number of genes evaluated. In this paper, we showcase splicing array technology, as well as powerful, yet general, statistical methodology for assessing differential expression, in the context of a real, complex experimental design. Our results suggest that the Spt4–Spt5 complex may help coordinate splicing with transcription under conditions that present kinetic challenges to spliceosome assembly or function.
Synopsis
Splicing is a key process for the regulation of gene expression in eukaryotes and is credited as being the main reason for the extraordinary complexity of the human proteome relative to the human genome. Accurate splicing is crucial for normal protein function; aberrant transcripts due to splicing mutations are known causes for 15% of genetic diseases. Therefore, elucidation of splicing mechanisms will not only help in understanding the complexity and diversity of higher organisms, but also potentially aid in new therapeutic strategies for treatments of splicing-related genetic disorders. It has been previously shown that splicing has important links to other steps involved with gene expression. In this study, the authors pursue a genome-wide approach, using yeast-based, splicing-sensitive, DNA microarrays in order to further characterize the roles of select splicing factors. They devise novel statistical and computational methods that enable identification of specific sets of genes that are mis-spliced in the chosen splicing factors. Follow-up investigation of known attributes of the genes so elicited indicates that these factors may help coordinate splicing and transcription in situations where additional energy is required to effect splicing.
Citation:Xiao Y, Yang YH, Burckin TA, Shiue L, Hartzog GA, et al. (2005) Analysis of a splice array experiment elucidates roles of chromatin elongation factor Spt4–5 in splicing. PLoS Comput Biol 1(4): e39.
==== Body
Introduction
Eukaryotic genes are fragmented into exons by intervening sequences (introns). After a gene is transcribed into pre-mRNA, the introns are removed from the transcript and the exons are joined by the spliceosome. This reaction, splicing, can also be used to create multiple transcripts from a single gene. For example, a particular exon may be included in one version of an mRNA, and skipped in another. This process of alternative splicing is subject to regulation in response to tissue, developmental, and environmental cues [1]. In humans, most genes are subject to splicing and more than half are likely subject to alternative splicing, which is credited as the most important source for the extraordinary enrichment in complexity of the human proteome relative to the genome [1]. Accurate splicing is crucial for normal protein function; aberrant transcripts due to splicing mutations are known causes for 15% of genetic diseases [1]. Therefore, elucidation of splicing mechanisms will not only help us understand the operating mechanisms underneath the functional complexity and diversity of higher eukaryotes, but also aid in new therapeutic strategies for treatments in splicing-related genetic disorders.
Although the different steps of gene expression are typically studied in isolation, it is clear that there are important functional links between them [2–4]. For example, the process of capping the 5′ end of pre-mRNAs is thought to influence both transcription and splicing [5,6]. Furthermore, the rate of transcription elongation appears to influence splicing and alternative splice site choice [7,8]. In addition, a number of pre-mRNA processing factors are recruited to transcripts via interaction with RNA polymerase II [2,3]. Thus, a comprehensive description of mRNA synthesis will require an understanding between these functional linkages of steps in gene expression.
Traditionally, gene expression is studied on an individual gene basis by ad hoc experiments. With the advent of eukaryotic genomic sequences, a global genomic view of mRNA production is achievable, and recently, several large-scale gene expression profiling experiments utilizing microarray technology have provided an unprecedented amount of information regarding the mechanisms underlying its regulation [4,9–11]. Saccharomyces cerevisiae, a simple yeast that has been used as a model to study eukaryotic gene expression, presents a convincing entry point to embark on this task. The yeast genome is completely sequenced and well annotated, and the splicing machinery of yeast is well conserved with that of humans. Among the more than ~5,800 genes in the yeast genome, only about 250 of them possess introns and only a handful have multiple introns or are alternatively spliced [12]. However, these 250 intronic genes give rise to 27% of the transcripts synthesized by the cell, an indication of the importance of splicing in yeast [13,14].
Clark et al. [15] designed a DNA microarray that allows the simultaneous analysis of splicing and mRNA levels in yeast. To discriminate between spliced and unspliced transcripts, oligonucleotide probes on these arrays were designed to detect the splice junctions (SJ), introns, and second exons of intron-containing genes (Figure 1A). SJ are found only in spliced transcripts, whereas introns exist only in unspliced transcripts and splicing intermediates. Second exons are present in both spliced and unspliced transcripts and are good indicators of total transcript level. To detect these different classes of transcripts, the arrays are competitively hybridized with probes derived from control and experimental yeast strains. For several splicing mutants, Clark et al. compared their whole genome splicing data to traditional molecular analyses of a small number of transcripts and found that reliance on only one or two genes as reporters may lead to misinterpretation of the role of a factor in splicing [15]. Thus, the whole genome approach provides a more reliable method for assessing the role of particular factors in splicing.
Figure 1 Splicing Array Probe and RT-PCR Primer Design
(A) Probe design of the splicing array. There are three oligonucleotide probes for each intron-containing gene: intron (red), splice-junction (blue), and exon (green). In addition, there are also about 800 probes for intronless genes (yellow). This figure is modified from Clark et al. [15].
(B) Primer design of RT-PCR. Primers are chosen to flank the intron–exon2 junction and the second exon or spliced mRNA.
Burckin et al. [4] recently used splicing-sensitive DNA microarrays to analyze 80 different yeast strains carrying mutations in genes encoding components of the gene expression machinery. Using clustering and machine learning techniques, they compared gene expression patterns in these mutants and discovered functional roles for specific factors at multiple steps in the gene expression pathway, further confirming the coordination and coupling of the machineries along the pathway.
Previously, Hartzog et al. [16] found evidence that the chromatin elongation factors Spt4 and Spt5 play a role in RNA processing in S. cerevisiae. Spt4 and Spt5 form a complex that regulates transcription elongation by RNA polymerase II. This complex is conserved across eukaryotes and has been proposed to both facilitate transcription by removing a nucleosomal barrier to transcript elongation and also suppress inappropriate transcription by reassembling nucleosomes behind transcribing polymerase [16]. The recent finding that Spt5 interacts physically and genetically with pre-mRNA capping factors suggests a role for Spt4–Spt5 in capping [17–20]. Because pre-mRNA capping is thought to increase the efficiency of splicing, Lindstrom et al. further analyzed splicing in spt4 and spt5 mutants and found that several genes were not spliced with normal efficiency [17]. In the splicing array study described above, Burckin et al. [4] found extensive but not universal splicing defects in spt4 and spt5 mutants. Interestingly, they also found that the capping enzyme appears to play an essential role in splicing. Thus, their genome-wide analysis of splicing provided particularly striking examples of linkages between steps in gene expression. However, the experimental design of that study precluded identification of specific genes dependent upon particular factors for their splicing.
Such identification is the purpose of our present study. While we also utilize splicing-specific DNA microarrays, we do so in the context of an experimental design that enables elicitation of specific intron-containing genes that are mis-spliced in spt4, spt5, or ceg1 mutants. In addition, we examine the aberrant splicing patterns caused by several phenotypically distinct spt5 mutations that had not been previously examined. Our primary data analytic task is therefore the determination of the set (possibly empty) of genes that have altered expression as reported by the SJ and intron probes. To do this, we assayed each mutant multiple times and then employed a recently devised statistical framework that robustly and efficiently identifies genes exhibiting differential expression (DE) in the mutants.
Many methods have been advanced for this task of identifying differentially expressed genes. Fold change has been extensively used to yield lists of genes that have altered expression beyond a prescribed threshold. Despite its methodological simplicity and intuitive appeal, fold change lacks a statistical framework (there is no accommodation of expression variation) and is biased toward selecting genes at low expression levels. Another class of frequently used methods treats the task of comparing expression levels in different biological states as a univariate testing problem, employing various corrections for test multiplicity [21]. Kerr et al. [22] propose using traditional analysis of variance (ANOVA) techniques, since these readily handle known sources of variation due to, for example, dye labeling and sample or array replicates. By removing these effects from the estimation of the error term, we achieve a reduction in this term and correspondingly sharper inferences. Wolfinger et al. [23] extend the ANOVA framework by treating some factors, for example, dyes and arrays, as random representatives of a large population (that is, as random effects) resulting in a mixed model. There are several Bayesian alternatives to the above approaches [24–27], as well as some intermediary approaches that yield regularized t statistics [28–30].
Our study employs a complex experiment design, featuring 22 dye-swap array hybridizations comprising both biological and technical replications (see Results). As elaborated in the next section, we initially analyzed these data with four ANOVA mixed models and the semiparametric hierarchical mixture model (SHMM) of Newton et al. [31]. Instead of arbitrating between these models and picking a single model on which to base DE declarations, we exploit the fact that all five models are estimating the same quantity and employ a novel synthesizing scheme [32], Differential Expression via Distance Synthesis (DEDS), to derive a list of differentially expressed genes in spt mutants. This method compares favorably with the best individual models, while enjoying improved robustness properties [32]. Further analysis of such genes, whose splicing is altered in spt mutants, reveals common biochemical characteristics and attributes, which may provide new insights into the mechanisms of RNA processing and its connections to transcription.
Results
Experimental Design and Data Pre-Processing
In yeast, SPT4 is a non-essential gene encoding a 102-amino-acid protein, and spt4Δ (null) mutants display mutant phenotypes and genetic interactions consistent with an elongation defect [16]. SPT5 encodes a large protein, and spt5 mutations typically display mutant phenotypes and genetic interactions similar to those observed for spt4 mutations, although they are often phenotypically more severe, consistent with the fact that SPT5 is essential for life [16]. In this work, we have analyzed an spt4 null mutation, and three partial loss-of-function mutations in SPT5. Two of these, spt5–4 and spt5–194, encode versions of Spt5 that are defective for binding Spt4 (GAH, J. Yamada, and T. Egelhofer, unpublished data). The third allele, spt5–242, causes a cold-sensitive growth defect [33], and displays splicing and other defects at all temperatures (GAH and TB, unpublished data; [17]). The Spt5–242 protein still binds Spt4, even at the non-permissive temperature (GAH, J. Yamada, and T. Egelhofer, unpublished data). In addition, we include analysis of ceg1–250, a temperature-sensitive mutation that causes rapid inactivation of the capping enzyme at the non-permissive temperature [6].
Two independent mRNA samples were prepared from each mutant, fluorescently labeled, and then hybridized to the splicing arrays competitively with a probe derived from wild-type cells. Experiments were performed using a replicated dye-swap study design (Figure 2A) [34]. Briefly, there were four arrays (A1–A4) for each mutant versus wild-type experiment. The first mRNA sample was hybridized to arrays A1 and A2 (Figure 2B), and the second was hybridized to A3 and A4. In A1 and A3, the mutant mRNA sample was labeled with Cy5 dye, and the wild-type sample was labeled with Cy3. The dye assignment was reversed for arrays A2 and A4. In addition to these 20 mutant arrays (four arrays × five mutants), there were two separate wild-type self-hybridization experiments, in which the wild-type was labeled with both Cy5 and Cy3. These self-hybridizations serve as technical replicates, that is, as controls for variation in labeling and hybridization.
Figure 2 Graphical Representation of Designs
(A) In this representation, vertices correspond to target mRNA samples and edges to hybridizations between two samples. By convention, we place the green-labeled sample at the tail and the red-labeled sample at the head of the arrow.
(B) Nested design of the experiment. The effect A is nested in S, and S is in turn nested in V. Note that there are two samples (S) for each mutant, but only one sample for the wild-type.
To provide a global view of splicing defects in the ceg and spt mutants, we plotted unnormalized log intensity values for signals from the two channels, mutant against wild-type, in Figure 3. Points that represent individual array features are color coded so that exon, SJ, intron, and intronless gene features can be visually differentiated. Genes lying on the diagonal have a ratio close to 1, and their expression in the mutants is therefore largely unaffected. For ceg1–250, shown in the lower right panel, introns (light blue points) deviate noticeably from the diagonal toward the ceg1–250 axis. This is a clear indication of intron accumulation in the ceg1 mutant. SJ (dark blue points) in ceg1–250, on the other hand, largely display ratios of less than 1, indicating a decrease in SJ formation. Taken together, an accumulation of introns and loss of SJ in ceg1–250 is indicative of a splicing defect. Compared with ceg1–250, the four spt mutants exhibit fewer alterations in splicing, with spt5–194 most severely affected, in agreement with its phenotypic characteristics. A control plot from the wild-type self-hybridization is depicted in the upper left panel. As expected, no separation is observed in introns and SJ, and all points conform closely to the diagonal.
Figure 3 Scatterplots of the Logarithm Intensities of Splicing-Related Probes
Points are color-coded as indicated.
Boxplots of normalized ratios of splicing-related probes stratified by mutants are shown in Figure 4. The general trend of the SJ probe ratios shows a shift from the horizontal zero line in the negative direction, signaling a decreased expression of SJ in the mutants. The ceg1–250 mutant showed the largest decrease, and spt5–194 was the most severely affected of the spt mutants. The boxplots of the exon probe ratios display a similar pattern of change—the expression of exon probes was also decreased in the mutants. This is consistent with the idea that the majority of the exon 2 probe signal for a transcript is derived from mRNA, which is stable and long-lived in comparison to pre-mRNA. It is of interest to investigate if the decrease of the SJ probe and exon probe ratios is correlated. Figure 5 displays the scatterplots between ratios of these probes. The upper panel shows evident correlation between SJ and exon ratios. In contrast to the exons and SJ, ratios of the intron probes do not show any shift from the horizontal zero line, but spread for the mutants is nonetheless increased. Furthermore, there is no obvious correlation between the intron and exon ratios. In both plots, however, the spread of the cloud of points is mutant dependent and related to the severity of splicing defects. From Figure 4, it is clear that several of the mutants tested, ceg1–250, spt5–194, and spt5–242, cause strong decreases in exon and SJ signals and, more idiosyncratic, gene-specific changes in intron signals. Do these changes reflect altered transcription, splicing, RNA decay, or a mixture of potential defects? To focus on alterations of splicing efficiency independent of changes in transcription, we used the intron accumulation (IA) and SJ indices of Clark et al. [15], which normalize ratios of intron and SJ signals to the ratios measured for the second exon. The SJ index is the change of the SJ probe signal normalized by the change of overall gene expression level as measured by the related exon probe signal: SJ = log(SpliceJunctionmut/SpliceJunctionwt)/(Exonmut/Exonwt). Similarly, the IA index is obtained as the normalized change of the intron probe signals: IA = (Intronmut/Intronwt)/(Exonmut/Exonwt). Relating changes in the SJ and intron signals to changes in the second exon takes into account changes in overall expression level that may occur as a result of alterations in other steps of gene expression.
Figure 4 Boxplots of Normalized Ratios of Splicing-Related Probes Stratified by Mutants
Splice-junction and exon probe ratios show a shift from the horizontal zero line in the negative direction, whereas intron probe ratios are centered at zero.
Figure 5 Scatterplots of Normalized Ratios of Splicing-Related Probes
Points are color-coded by their mutant identity. Gray horizontal and vertical reference lines indicate zero expression ratios.
DE Models
The experimental design of the splice mutant study motivated the use of four different mixed ANOVA models in addition to the SHMM (Table 1). These were separately applied to the two splicing indices. The four ANOVA models are distinguished by including wild-type self-hybridizations or not and allowing gene-specific heterogeneity or not. The experimental design (Figure 2B)—wherein array effect A is nested in sample effect S (S/A), and sample effect S is in turn nested in mutant effect V (V/S), argues for treating model terms involving S and A as random effects, with the remaining terms involving genes (G), mutants (V), and gene–mutant interactions (GV) being fixed effects. The four models thereby constitute nested, mixed-effect ANOVAs ; see Materials and Methods for fitting details.
Table 1 Summary of the Five Competing DE Models
To complement the ANOVA approaches described above, we also employed the SHMM advanced by Newton et al. [31]. This methodology was selected for the following reasons: (i) the SHMM is non-parametric where there is sufficient information (lots of genes) and parametric where there is limited information (observations per gene), and this synthesis makes for an appropriately balanced strategy; and (ii) as is standard, our ANOVA approaches treat gene (G), mutant (V), and gene–mutant interactions (GV) as fixed effects. Thus, there is no information sharing between genes. The SHMM achieves such sharing and does so in a more principled and flexible manner than some of the ad hoc approaches proposed that yield regularized t-statistics [28,30,35]. The SHMM also has limitations, the foremost of which perhaps is the adequacy of the parametric assumptions. The extent of such assumptions has been appreciably relaxed compared to the preceding fully parametric treatment of Kendziorski et al. [27]. Importantly, diagnostic tools are provided for assumption checking. Additionally, the present implementation supports only two group comparisons. Thus, there is some potential efficiency loss for the nested design employed in the splice study (Figure 2B). Details on the estimation methodology as well as extensive illustration of calibration, diagnostic, and performance aspects are provided in Newton et al. [31].
Model Synthesis and Selection of Differentially Expressed Genes
Models with heteroscedastic errors accommodate gene-specific variances, but typically, as here, replication is very limited and so the precision of the estimates is compromised. Models imposing homoscedastic errors yield precise estimates of the common error variance, and tests based on many degrees of freedom (df), since they permit combination over the large number of genes. However, the homoscedasticity assumption is both strong and difficult to evaluate. Differences in error df for the different models are presented in Table 2. Note that there are more than 5,000 df for error for the homoscedastic models and only about 20 df for the heteroscedastic models. A comparison of associated p-values from individual measures is provided in [32].
Table 2 Degrees of Freedom for the ANOVA Mixed Models
DEDS is a novel method combining statistics or summaries that measure the same phenomenon [32]. Rather than trying to arbitrate between models and pick a single model on which to base DE declarations, or informally distilling sets of genes that are differentially expressed under two or more models, we applied DEDS here as a robust means to refine selection of DE genes as furnished by the above five individual models. The simple underlying principle of DEDS is that genes that are highly ranked (as being differentially expressed) by all five models are more likely to be truly differentially expressed than genes that are high only for a single model. Further details concerning DEDS are provided in [32], while an algorithm outline is sketched in Materials and Methods.
The task of identifying differentially expressed genes consists of two components: (i) ranking genes in order of evidence for DE; and (ii) declaration of a set of DE genes by thresholding the ranked list. Here, we examine the robustness of DEDS with respect to both components. A comparison of ranking of DE genes by DEDS and individual measures, using an example based on the IA indices of spt5–242, is provided in Figure 6A. Ranks are logged so that correlations of DE genes (low ranks) are more clearly displayed. We see a similar level of concordance between ranks from DEDS and individual models. The numbers of genes identified as differentially expressed by DEDS under false discovery rates (FDR) 0.01 and 0.05 for SJ as well as IA indices are listed in Table 3. To examine the stability of the DE findings, we assessed the impact of scale choice of measures and the interrelated choice of distance metric. We investigated all combinations of (raw, logarithm) p-value scales * (Euclidean, city block) distances. Representative results are shown in Figure 6B. Genes are ordered according to their DE significance by the raw/Euclidean combination, so that the black points are monotone by definition. The dashed gray line marks the 0.05 q-value threshold. The corresponding numbers of declared DE genes for these four combinations range from 145 to 163, of which 139 are common to all four combinations. This demonstrates that, for the splice array experiment, DEDS-based selections of DE genes are largely insensitive to the different scales and distance metrics examined. This concordance, evident in Figure 6, pertains to the IA index and the ceg1–250 mutant. Analogous concordance was observed for the SJ index and the other mutants.
Figure 6 Analysis of DE Gene Ranking and Selection by DEDS
(A) A comparison of ranking of DE genes by DEDS and individual measures. Plotted are a scatterplot matrix of DE gene rankings by the five models and DEDS using spt5–242 IA indices. Ranks are logged so that correlations of DE genes (low ranks) are more clearly displayed.
(B) Sensitivity of DEDS declarations of DE to choice of scale and distance metric. Genes, ordered according to their DE significance by the raw/Euclidean combination (so that the black points are monotone by definition) are plotted against DEDS q-values for all four scale/distance combinations. The dashed gray line marks the 0.05 q-value threshold.
Table 3 Number of DE in SJ and IA Indices
The observation of greater numbers of genes identified as differentially expressed based on the IA index data than on the SJ data (Table 3) reinforced the finding in Clark et al. [15] that IA indices are a more sensitive indicator for splicing defects. The splicing defect in the yeast capping enzyme mutant ceg1–250 is catastrophic, whereas in the spt4 and spt5 mutants fewer genes exhibit a splicing defect. Overall, spt5–194 is the most severe splicing mutant among all spt mutants, with spt4Δ being the least impaired. The complete list of DE genes is provided in Table S1.
Validation of DE Genes
The identification of genes affected by spt4 and spt5 mutations using statistically robust methodology offers insight into the function of the Spt4–Spt5 complex, as well as the opportunity to better equate changes in IA with bona fide splicing defects. To validate our findings, we have used quantitative RT-PCR (QPCR) analysis to quantitatively examine five intron-containing genes, as well as two unspliced genes, in all five mutants. We previously performed a qualitative analysis of three of these genes, U3B, RPS25A, and RPL26A, and found that they were inefficiently spliced in spt4 and spt5 mutants [17]. By choosing primers that flank the intron–exon2 junction, we can specifically detect unspliced pre-mRNA (Figure 1B). We also picked primers to detect either the second exon, or spliced mRNA (Figure 1B). As with the microarrays, we can normalize changes in pre-mRNA levels to changes in spliced mRNA or total mRNA (that is, exon2).
As shown in Table 4, the results of the RT-PCR analysis generally agreed with the microarray analysis. Strikingly, in the four spt mutants, genes identified by DEDS showed an absolute increase in pre-mRNA levels, while in the ceg1 mutant none of the pre-mRNAs showed an absolute increase as compared to wild-type. After normalizing the pre-mRNA signals to the spliced mRNA or second exon signals to account for potential changes in transcription or transcript stability, ceg1 also showed a splicing defect as predicted by DEDS. Furthermore, the performance of DEDS was superior to the four ANOVA models and equivalent to the SHMM in terms of numbers of false positives and negatives over all five mutants (Table S2).
Table 4 QPCR Validation DE Microarray Data
Description and Analysis of DE Genes
There are likely multiple molecular mechanisms by which different genes were differentially expressed in the mutants discussed here. To account for some of these mechanisms, we subdivided the lists of DE genes with a q ≤ 0.05 (controlling FDR) before further analysis. First, we reasoned that positive and negative changes in IA likely occurred via different molecular mechanisms. Therefore, for each of the five mutants examined, the DE genes were divided into lists of genes with either positive or negative fold change (here, fold change refers to the IA index). Second, because ribosomal protein genes represent a large fraction of all spliced genes in yeast [36], and because they are subject to a common mode of regulation [37], we further subdivided our lists of DE genes into sublists of ribosomal (RP) and non-RP genes (Table 5). Finally, we focused upon the IA index, as it is more sensitive to alterations in splicing [15].
Table 5 Distribution of DE Genes
For the spt5 and ceg1 mutants, a large majority of the DE genes encoded RP proteins, whereas only ~40% of all intron-containing genes encode RP proteins (Table 5 and [36]). Furthermore, a number of translation and rRNA processing factors are among the non-RP genes found in our analysis, and it is possible that these genes are regulated by the same strategies as the RPs. Interestingly, for those DE genes with a negative fold change—that is, those that were apparently spliced more efficiently—we found no RP genes. This suggests that the genes with a negative or positive fold change in the IA index have distinct dependencies upon Spt4–Spt5 and Ceg1.
We next asked if the genes identified in this analysis shared any particular attributes. It has previously been noted that introns in yeast display a bimodal distribution of sizes and positions within genes [36]. RP protein genes have large introns that occur relatively early in a pre-mRNA, whereas non-RP genes typically have smaller introns that occur somewhat later in the mRNA. Furthermore, RP genes are highly transcribed, whereas non-RP genes tend to be less highly transcribed [14]. We therefore compared the transcription rates and size and positions of introns within the DE genes that displayed a positive fold change (Table 6). In the ceg1 mutant, the set of DE genes had no unusual properties other than the non-RP DE genes being transcribed somewhat more frequently than the average non-RP gene. In the spt mutants, intron position of the DE genes was not significantly different from the average for RP and non-RP genes (Table 6). In contrast, in the spt5–4 and spt5–194 mutants, the non-RP DE genes shared attributes of RP genes: they tended to have longer introns and be more highly expressed than the typical non-RP gene. The non-RP DE genes in the spt4Δ and spt5–242 mutants represent an intermediate case; their introns are not significantly longer than those of the typical non-RP intron-containing genes, but they are more highly transcribed.
Table 6 Properties of DE Genes with a Positive Fold Change (Average)
The DE genes with a negative fold change appear to represent a distinct class of genes. First, they encoded only non-RPs. Second, they resembled the typical non-RP intron-containing genes in that they had short introns; however, they were expressed at even lower levels than the typical non-RPs (Table 7), contrary to the DE genes with positive fold changes. Again, this is consistent with the idea that these genes were differentially expressed for reasons distinct from those leading to DE of genes with a positive fold change.
Table 7 Properties of DE Genes with a Negative Fold Change (Average)
Discussion
In this paper, we showcased splicing array technology and developed methodologies for its analysis in the context of a real, complex experimental design. We applied four ANOVA mixed models and a SHMM, and used DEDS [32] to derive a list of DE genes. The DEDS algorithm synthesizes statistics or methods that estimate the same quantity of interest. The underlying principle behind DEDS is that genes that are highly ranked by different methods are more likely to be truly differentially expressed than genes that rank highly on a single measure. In our previous work, we have evaluated DEDS on diverse datasets, featuring both one-channel Affymetrix oligonucleotide arrays and two-channel spotted arrays [32]. Using a set of spike-in (Affymetrix) datasets, where differentially expressed genes are known, we demonstrated that DEDS compares favorably with the best individual statistics while enjoying robustness properties lacked by the individual statistics [32].
Previous to this and other microarray studies, only four genes had been identified and confirmed for splicing defects in spt4 and spt5 mutants using traditional molecular techniques [17]. Recently, Burckin et al. have used splicing-sensitive DNA microarrays to compare patterns of splicing defects across a diverse set of mutations affecting gene expression [4], but this and the previous study lacked a statistical or quantitative framework for rigorous determination of specific genes that were differentially expressed. Here, we have used splicing-sensitive DNA microarrays combined with DEDS to analyze all known intron-containing genes in the yeast genome and to specifically identify those genes whose proper splicing is dependent upon SPT4, SPT5, or CEG1. Despite the differences in experimental goals and designs, the findings of these two studies are nonetheless consistent. In Burckin et al. [4], the analyses using hierarchical clustering and support vector machines showed that the overall impact of the loss of Ceg1 function in vivo is nearly identical to that of bona fide splicing factors, which is in line with the large number of DE genes we found whose splicing is abnormal in ceg1–250 (see Table 3). Also in our previous study, spt4Δ and spt5–194 mutants displayed a lesser splicing defect than the ceg1–250 mutant, which is again consistent with our current findings. Comparison of the lists of DE genes for the five mutants examined here revealed that most of the genes that were differentially expressed in the spt mutants were also differentially expressed in the ceg1 mutant (Figure 7 and Table S1). The spt5–242 mutant differed from the other spt5 mutants in that it did not preferentially affect the splicing of non-RP genes with long introns. We do not understand the mechanistic basis for this observation, although it is consistent with our previous observations that this spt5 mutation is phenotypically distinct from other spt5 alleles and therefore may cause a distinct biochemical defect [33,38]. Our data further suggest that Spt4′s contribution to splicing is modest, as only a handful of genes were differentially expressed in the spt4 mutant. This is consistent with the observation that, in contrast to SPT5, SPT4 is not essential for life. Furthermore, this observation suggests that the defects caused by the spt5–4 and spt5–194 mutations extend beyond the Spt4 binding defect we have observed for the Spt5–4 and Spt5–194 proteins. Since there is currently no evidence that Spt4 functions independently of Spt5 [39], these observations suggest that Spt4 assists in, but is not essential for, the functions of Spt4–Spt5 in splicing.
Figure 7 Venn Diagram of DE Genes from Different Mutants
(A) compares DE genes among the three spt5 mutants (spt5–194, spt5–4, and spt5–242). Statistical test shows that the common 43 genes are highly significant, with a p-value < 0.001. In (B), spt5 refers to the 43 common genes among all spt5 mutants. The overlaps between spt5 and ceg1–250 (40, p < 0.001), spt5, and spt4 (8, p< 0.001), spt4, spt5, and ceg1–250 (7, p< 0.001) are all significant.
The smaller number of DE genes in the spt mutants compared to ceg1–250 may indicate a lesser effect on splicing rather than an effect on a distinct subset of intron-containing genes. It is interesting to note, however, that highly transcribed genes with long introns—that is, RP genes and a subset of non-RP genes with long introns—were most sensitive to the spt mutations. These data suggest that the Spt4–Spt5 complex may play a particular role in coordinating splicing with transcription under conditions that present kinetic challenges to the spliceosome or its assembly, that is, when splice sites are widely separated, increasing the separation in time and space between the synthesis of the 5′ and 3′ splice sites, or when a gene is highly transcribed, creating the need for rapid and repeated assembly of spliceosomes over one site on a gene. In addition, these data are consistent with recent evidence demonstrating an effect of RNA polymerase II elongation rates on alternative splicing in higher eukaryotes [40]. In contrast, the non-RP genes spliced more efficiently in the spt mutants tend to be transcribed less frequently than the average non-RP gene (Table 7). Thus, as is the case for transcription, the Spt4–Spt5 complex may have both positive and negative effects on splicing [16]. Furthermore, this is consistent with previous observations that altered transcription elongation may lead to increased splicing, presumably due to increased opportunities for recognition of suboptimal splice sites [7,8]. Whether the effects we have measured here are due to altered elongation rates, or they indicate a more direct role of Spt4–Spt5 in splicing is currently under investigation.
Materials and Methods
Sample preparation and array hybridization.
All yeast strains (Table 8) used were isogenic to S288C and Gal+ [41]. Yeast were grown overnight in rich medium (YPD) at 30 °C to early log phase (> 1 × 107 cells/ml), spun down, and resuspended in pre-warmed 39 °C media, and allowed to grow at 39 °C for 45 min after shift to restrictive temperature. Cells were collected by centrifugation at room temperature for 4 min, washed once with sterile water, flash frozen in liquid nitrogen, and stored at −80 °C. Total RNA was isolated by a hot phenol method [42] and quantitated by UV absorbance. Fluorescently labeled probe preparation, hybridization, and data acquisition were performed as previously described [15] using 15 μg of total RNA/sample. For each mutant, RNA was prepared from two independently grown cultures. Each RNA sample was used to probe two arrays, and was labeled with Cy3 for the first array and Cy5 for the second.
Table 8 Yeast Strains
Data normalization and pre-processing.
To effectively and properly normalize the data, we used non-linear loess normalization [43] based on the subset of intronless genes. After normalization, for each array the four replicates of each SJ, intron, and exon probes were summarized using averages. This was followed by the calculation of SJ and IA indices.
ANOVA mixed models.
We applied four different ANOVA mixed models corresponding to all combinations of wild-type versus wild-type (in/out) by gene-specific variance heterogeneity (yes/no). DE is examined by the two-sample t test, when including the two wild-type versus wild-type samples, whereas a one-sample t test is applied when excluding these two samples. Not allowing for gene-specific variance imposes the assumption that all genes exhibit a similar degree of variability, so they can be jointly analyzed using a common estimate of error variance [22]. Conversely, allowing different variances for different genes [23] mandates fitting gene by gene.
Model specifics.
Model I—one-sample/homoscedastic errors: Let Ygvsa be the splicing related index, SJ or IA, from gene g (g = 1, 2,…, 254 for SJ and 1, 2,…, 263 for IA), mutant v (v = 1, 2,…, 5), sample s (s = 1,2), and array a (a = 1,2; corresponding to the dye-swap pair). The first model can be represented as
Table 9 Derivation of Variance Components for Model I
Effects (V/S)s, (V/S/A)a, (GV/S)gvs, and ɛgvsa are assumed to be normally distributed normal variables with zero means and variance components
,
and σ2, respectively. The derivation of the variance components is shown in Table 9. The remaining effects in the model are fixed effects. The parameter of interest in this model is μgv = μ + Gg + Vv + GVgv, which measures the mean of the SJ/IA indices of gene g in mutant v. The following null hypothesis therefore defines the absence of DE in mutant v and gene g:
The variance of the treatment mean μ̂gv can be computed by the following equation:
where nS = 2 and nA = 2.
Model II—one-sample/heteroscedastic errors: Model II is different from Model I by assuming that each gene has its own error distribution, so the model is fitted gene by gene. It can be represented by the following equation:
The parameter of interest in this model is μgv = μg + Vv, which measures the mean of the SJ/IA indices of gene g in mutant v. The following null hypothesis defines the absence of DE in mutant v and gene g:
The variance of the treatment mean μ̂gv can be computed by the following equation:
Model III—two-sample/homoscedastic errors: Model III differs from Model I by including the indices derived from the two wild-type self-hybridizations. Because of this inclusion, the study design is rendered unbalanced. To be more specific, the arrays in the two wild-type self-hybridizations came from the same sample, whereas the samples of four slides related to a mutant were from two distinct samples (see Figure 2B). The model can be represented by the following equation:
The parameter of interest in this model is μgv = μ + Gg + Vv + GVgv, which measures the mean of the SJ/IA indices of gene g in mutant v. The following null hypothesis defines the absence of DE in mutant vm and gene g compared to the wild-type:
The variance of the treatment mean μ̂gv can be computed by the following equation:
, where nS = 2 for mutants and nS = 1 for the wild-type.
Model IV—two-sample/heteroscedastic errors: Model IV differs from Model II by including the indices derived from the two wild-type self-hybridizations. The model can be represented by the following equation:
The parameter of interest in this model is μgv = μg + Vv, which measures the mean of the SJ/IA indices of gene g in mutant v. The following null hypothesis defines the absence of DE in mutant vm and gene g compared to the wild-type:
The variance of the treatment mean μ̂gv can be computed by the following equation:
, where nS = 2 for mutants and nS = 1 for the wild-type.
SHMM model.
Implementation of the SHMM model uses the R package EBarrays, available from ftp://ftp.biostat.wisc.edu/pub/newton/Arrays/tr1074/Rcode/. The output posterior probabilities for (directional) DE from the package have dual utilities: (i) ranking (genes), and (ii) calibration (providing FDR). We utilized the former for DEDS synthesis.
DEDS procedures.
Fit the five DE models, and assume the resulting p values for gene i and model j are pij (i = 1, 2, …, n, j = 1, 2, …, 5) in data matrix P.
Locate the most extreme point E as a vector of zeros of length five.
Calculate distance di of all genes to E and order d(1) ≤ d(2) ≤ … ≤ d(n).
Generate B sets of reference distribution by:
Center the columns of P at mean 0.
Compute the singular value decomposition P = UDVT.
Calculate P* = PV.
Create Z* by drawing uniform distribution over the range of the columns of P*.
Back transform Z
* by Z = Z*VT to obtain the reference data Z.
For each reference dataset b, di values are calculated and ordered in the way of
For a typical gene i, compute the median number of falsely called genes by computing the median number of values among each of the B sets of
that are smaller than d
(i); and the q-value (controlling FDR) of gene i is computed as the median of the number of falsely called genes divided by the number of genes called significant. Illustration of determination of the extreme point E when using statistics (instead of p-values), in known null and non-null situations, is provided in Figure S1.
Analysis of DE genes.
Gene annotations were obtained from the Ares lab intron database (http://www.cse.ucsc.edu/research/compbio/yeast_introns.html), and transcription frequency data was obtained from the Young lab (http://web.wi.mit.edu/young/expression/transcriptome.html).
The collection of all intron-containing genes was divided into sets of RP and non-RP genes, and averages and standard deviations were calculated for their transcription frequencies, intron lengths, and intron start sites. Several genes were omitted from these analyses because there was no good data concerning their transcription frequency or intron position or size. In addition, Mtr2, which has multiple, overlapping introns, was considered to have a single intron for this analysis (see Table S1). To determine if the properties of DE genes in a mutant were significantly different from those of all RP or non-RP intron-containing genes, we used a non-parametric resampling method. Briefly, a referent null distribution was generated by first taking 10,000 random samples of size N from the sets of all intron-containing RP or non-RP genes (N is the number of DE RP or non-RP genes for a particular mutant), and then calculating the averages of each sample. The p-value was derived as the percentage within the referent distribution that is more extreme than the observed property.
QPCR analysis.
cDNA synthesis for QPCR was performed as described for fluorescently labeled target synthesis, except that equal concentrations of all four deoxyribonucleotides and no Cy dyes were used. Reactions lacking reverse transcriptase were performed to control for genomic DNA contamination. Amplifications were conducted in a Bio-Rad iCycler using iQ SYBR Green Supermix (Bio-Rad, Hercules, California, United States) and 200 μM primer according to the manufacturer's instructions, using the oligonucleotide primers found in Table S3. Representative transcripts were assayed in triplicate. To compare the QPCR with array values, we normalized QPCR values to the OSH3 mRNA. OSH3 was chosen as a suitable reference gene, since the array data indicated that its expression was unchanged in the five mutants used in the comparison.
Supporting Information
Table S1 A Complete List of Differentially Expressed Genes for the Five Mutants in SJ and IA Indices
(346 KB XLS)
Click here for additional data file.
Table S2 Comparison of the Five Models and DEDS in Terms of Numbers of False Positives/Negatives in the QPCR Tested Five Genes
(23 KB DOC)
Click here for additional data file.
Table S3 Oligo Sequences Used in the QPCR Validation of the Microarray Analysis
(14 KB XLS)
Click here for additional data file.
Figure S1 Illustration of DEDS Extreme Point Determination
Application of DEDS in (a) the Affymetrix spike-in data ([45]); (b) the Affymetrix spike-in data with the top 100 DE genes removed to generate a dataset with only null genes. This dataset consists of 12,626 probe sets, 14 of which are spiked in at varying concentrations, and the rest are null. DEDS was applied synthesizing t statistics, fold change, and SAM (Significance Analysis for Microarrays) measures. The diagonal and upper triangle display Q-Q plots and scatterplots of the respective measures, while the lower triangle gives corresponding correlation coefficients. Red spots are differentially expressed by DEDS. E of (t statistic, fold change and SAM) was found to be (171.9, 7.2, 82.34) in panel (a) and at (7.5, 0.5, 3.1) in panel (b).
(39 KB DOC)
Click here for additional data file.
Accession Numbers
The Saccharomyces Genome Database (SGD; http://www.yeastgenome.org/) accession numbers for the genes and gene products discussed in this paper are ceg1 (S000003098), RPL26A (S000004336), RPS25A (S000003259), SNR17B (U3B) (S000007441), Spt4 (S000003295), Spt5 (S000004470), U3B (S000007441), YDR064W (RPS13) (S000002471), YGR027C (RPS25A) (S000003259), YLR344W (RPL26A) (S000004336), and YOL127W (RPL25) (S000005487).
This work was supported by grants to GAH from the National Institutes of Health (GM60479) and the University of California Cancer Research Coordinating Committee. LS was supported by a grant from the Packard Foundation. We thank Manny Ares for many stimulating discussions related to this work.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. GAH and MRS conceived and designed the experiments. TAB performed the experiments. YX, YHY, and MRS analyzed the data. YHY, LS, and GAH contributed reagents/materials/analysis tools. YX, YHY, TAB, GAH, and MRS wrote the paper.
Abbreviations
ANOVAanalysis of variance
DEdifferential expression
DEDSDifferential Expression via Distance Synthesis
dfdegrees of freedom
IAintron accumulation
QPCRquantitative RT-PCR
RPribosomal
SHMMsemiparametric hierarchical mixture model
SJsplice junction
==== Refs
References
Maniatis T Tasic B 2002 Alternative pre-mRNA splicing and proteome expansion in metazoans Nature 418 236 243 12110900
Maniatis T Reed R 2002 An extensive network of coupling among gene expression machines Nature 416 499 506 11932736
Proudfoot NJ Furger A Dye MJ 2002 Integrating mRNA processing with transcription Cell 108 501 512 11909521
Burckin TA Nagel R Mandel-Gutfreund Y Shiue L Clark TA 2005 Exploring functional relationships between components of the transcription, splicing and mRNA export machinery by gene expression machinery Nat Struct Mol Biol 12 175 182 15702072
Schwer B Shuman S 1996 Conditional inactivation of mRNA capping enzyme affects yeast pre-mRNA splicing in vivo RNA 2 574 583 8718686
Fresco LD Buratowski S 1996 Conditional mutants of the yeast mRNA capping enzyme show that the cap enhances, but is not required for, mRNA splicing RNA 2 584 596 8718687
Howe KJ Kane CM Ares M Jr. 2003 Perturbation of transcription elongation influences the fidelity of internal exon inclusion in Saccharomyces cerevisiae
RNA 9 993 1006 12869710
de la Mata M Alonso CR Kadener S Fededa JP Blaustein M 2003 A slow RNA ploymerase II affects alternative splicing in vivo Mol Cell 12 525 532 14536091
Fehlbaum P Guihal C Bracco L Cochet O 2005 A microarray configuration to quantify expression levels and relative abundance of splice variants Nucleic Acids Res 33 e47 15760843
Johnson JM Castle J Garrett-Engele P Kan Z Loerch PM 2003 Genome-wide survey of human alternative pre-mRNA splicing with exon junction microarrays Science 302 2141 2144 14684825
Pan Q Shai O Misquitta C Zhang W Saltzman AL 2004 Revealing global regulatory features of mammalian alternative splicing using a quantitative microarray platform Mol Cell 16 929 941 15610736
Barrass JD Beggs JD 2003 Splicing goes global Trends Genet 19 295 298 12801718
Lopez PJ Seraphin B 1999 Genomic-scale quantitative analysis of yeast pre-mRNA splicing: Implications for splice-site recognition RNA 5 1135 1137 10496213
Ares M Jr. Grate L Pauling MH 1999 A handful of intron-containing genes produces the lion's share of yeast mRNA RNA 5 1138 1139 10496214
Clark TA Sugnet CW Ares M Jr. 2002 Genomewide analysis of mRNA processing in yeast using splicing-specific microarrays Science 296 907 910 11988574
Hartzog GA Speer JL Lindstrom DL 2002 Transcript elongation on a nucleoprotein template Biochim Biophys Acta 1577 276 286 12213658
Lindstrom DL Squazzo SL Muster N Burckin TA Wachter KC 2003 Dual roles for Spt5 in pre-mRNA processing and transcription elongation revealed by identification of Spt5-associated proteins Mol Cell Biol 23 1368 1378 12556496
Pei Y Shuman S 2002 Interactions between fission yeast mRNA capping enzymes and elongation factor Spt5 J Biol Chem 277 19639 19648 11893740
Wen Y Shatkin AJ 1999 Transcription elongation factor hSPT5 stimulates mRNA capping Genes Dev 13 1774 1779 10421630
Mandal SS Chu C Wada T Handa H Shatkin AJ 2004 Functional interactions of RNA-capping enzyme with factors that positively and negatively regulate promoter escape by RNA polymerase II Proc Natl Acad Sci U S A 101 7572 7577 15136722
Dudoit S Yang YH Callow MJ Speed TP 2002 Statistical methods for indentifying differentially expressed genes in replicated cDNA microarray experiments Statistica Sinica 12 111 139
Kerr MK Martin M Churchill GA 2000 Analysis of variance for gene expression microarray data J Comput Biol 7 819 837 11382364
Wolfinger RD Gibson G Wolfinger ED Bennett L Hamadeh H 2001 Assessing gene significance from cDNA microarray expression data via mixed models J Comput Biol 8 625 637 11747616
Efron E Tibshirani R Storey J Tusher VG 2001 Empirical bayes analysis of a microarray experiment J Am Stat Assoc 96 1151 1160
Lee ML Lu W Whitmore GA Beier D 2002 Models for microarray gene expression data J Biopharm Stat 12 1 19 12146717
Newton MA Kendziorski CM Richmond CS Blattner FR Tsui KW 2001 On differential variability of expression ratios: Improving statistical inference about gene expression changes from microarray data J Comput Biol 8 37 52 11339905
Kendziorski CM Newton MA Lan H Gould MN 2003 On parametric empirical Bayes methods for comparing multiple groups using replicated gene expression profiles Stat Med 22 3899 3914 14673946
Lönnstedt I Speed TP 2001 Replicated microarray data Statistica Sinica 12 31 46
Smyth GK 2004 Linear models and empirical Bayes methods for assessing differential expression in microarray experiments Stat Appl Genet Mol Biol 3: Article 3
Baldi P Long AD 2001 A Bayesian framework for the analysis of microarray expression data: Regularized t-test and statistical inferences of gene changes Bioinformatics 17 509 519 11395427
Newton MA Noueiry A Sarkar D Ahlquist P 2004 Detecting differential gene expression with a semiparametric hierarchical mixture method Biostatistics 5 155 176 15054023
Yang YH Xiao Y Segal MR 2005 Identifying differentially expressed genes from microarray experiments via statistic synthesis Bioinformatics 21 1084 1093 15513985
Hartzog GA Wada T Handa H Winston F 1998 Evidence that Spt4, Spt5, and Spt6 control transcription elongation by RNA polymerase II in Saccharomyces cerevisiae
Genes Dev 12 357 369 9450930
Chen Y Dougherty ER Bittner ML 1997 Ratio-based decisions and the quantitative analysis of cDNA microarray images J Biomed Opt 2 364 374 23014960
Tusher VG Tibshirani R Chu G 2001 Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci U S A 98 5116 5121 11309499
Spingola M Grate L Haussler D Ares M Jr. 1999 Genome-wide bioinformatic and molecular analysis of introns in Saccharomyces cerevisiae
RNA 5 221 234 10024174
Wade JT Hall DB Struhl K 2004 The transcription factor Ifh1 is a key regulator of yeast ribosomal protein genes Nature 432 1054 1058 15616568
Lindstrom DL Hartzog GA 2001 Genetic interactions of Spt4-Spt5 and TFIIS with the RNA polymerase II CTD and CTD modifying enzymes in Saccharomyces cerevisiae
Genetics 159 487 497 11606527
Kim DK InuKai N Yamada T Furuya A Sato H 2003 Structure-function analysis of human Spt4: Evidence that hSpt4 and hSpt5 exert their roles in transcriptional elongation as parts of the DSIF complex Genes Cells 8 371 378 12653964
Nogues G Kadener S Cramer P de la Mata M Fededa JP 2003 Control of alternative pre-mRNA splicing by RNA Pol II elongation: Faster is not always better IUBMB Life 55 235 241 12880204
Winston F Dollard C Ricupero-Hovasse SL 1995 Construction of a set of convenient Saccharomyces cerevisiae strains that are isogenic to S288C Yeast 11 53 55 7762301
Zavanelli MI Ares M Jr. 1991 Efficient association of U2 snRNPs with pre-mRNA requires an essential U2 RNA structural element Genes Dev 5 2521 2533 1752442
Yang YH Dudoit S Luu P Lin DM Peng V 2002 Normalization for cDNA microarray data: A robust composite method addressing single and multiple slide systematic variation Nucleic Acids Res 30 e15 11842121
Holstege FC Jennings EG Wyrick JJ Lee TI Hengartner CJ 1998 Dissecting the regulatory circuitry of a eukaryotic genome Cell 95 717 728 9845373
Irizarry RA Bolstad BM Collin F Cope LM Hobbs B 2003 Summaries of Affymetrix GeneChip probe level data Nucleic Acids Res 31 e15 12582260
|
16172632
|
PMC1214541
|
CC BY
|
2021-01-05 09:18:22
|
no
|
PLoS Comput Biol. 2005 Sep 16; 1(4):e39
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010039
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1618418910.1371/journal.pcbi.001004105-PLCB-RA-0131R1plcb-01-04-06Research ArticleBioinformatics - Computational BiologyDevelopmentGenetics/Gene FunctionGenetics/Genome ProjectsGenetics/Functional GenomicsGenetics/Gene ExpressionNeuroscienceMus (Mouse)A Digital Atlas to Characterize the Mouse Brain Transcriptome A Digital Atlas for the Mouse Brain TranscriptomeCarson James P 12*Ju Tao 3Lu Hui-Chen 4¤Thaller Christina 2Xu Mei 5Pallas Sarah L 5Crair Michael C 4Warren Joe 3Chiu Wah 12Eichele Gregor 61 Program in Structural and Computational Biology and Molecular Biophysics, National Center for Macromolecular Imaging, Baylor College of Medicine, Houston, Texas, United States of America
2 Verna and Marrs McLean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, Texas, United States of America
3 Department of Computer Science, Rice University, Houston, Texas, United States of America
4 Division of Neuroscience and Program in Developmental Biology, Baylor College of Medicine, Houston, Texas, United States of America
5 Department of Biology, Georgia State University, Atlanta, Georgia, United States of America
6 Max Planck Institute of Experimental Endocrinology, Hanover, Germany
Eisen Michael B EditorLawrence Berkeley National Laboratory, United States of America* To whom correspondence should be addressed. E-mail: [email protected]¤ Current address: The Cain Foundation Laboratories, Department of Pediatrics, Division of Neuroscience and Program in Developmental Biology, Baylor College of Medicine, Houston, Texas, United States of America
9 2005 23 9 2005 1 4 e419 6 2005 16 8 2005 Copyright: © 2005 Carson et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Massive amounts of data are being generated in an effort to represent for the brain the expression of all genes at cellular resolution. Critical to exploiting this effort is the ability to place these data into a common frame of reference. Here we have developed a computational method for annotating gene expression patterns in the context of a digital atlas to facilitate custom user queries and comparisons of this type of data. This procedure has been applied to 200 genes in the postnatal mouse brain. As an illustration of utility, we identify candidate genes that may be related to Parkinson disease by using the expression of a dopamine transporter in the substantia nigra as a search query pattern. In addition, we discover that transcription factor Rorb is down-regulated in the barrelless mutant relative to control mice by quantitative comparison of expression patterns in layer IV somatosensory cortex. The semi-automated annotation method developed here is applicable to a broad spectrum of complex tissues and data modalities.
Synopsis
The mammalian brain is a complex organ with hundreds of functional parts. Describing when and where genes are expressed in the brain is thus a potentially powerful method for understanding the function of gene products. In recent years, several mammalian genomes including those of human and mouse have been characterized. There are now efforts around the world that aim to determine the expression patterns for all genes in the mouse brain. To search these expression data readily, they must be placed into an atlas. The authors propose a new method for bringing such genetic data into a common spatial framework so that one can perform spatial searches and comparisons of gene expression patterns. To create this atlas, the authors developed a series of maps of the brain using a graphical modeling method called subdivision. These maps were deformed to match the shape of tissue sections, and genetic activity information was associated with the appropriate coordinates on the map. After placing 200 genes into the context of this atlas, the authors illustrate its application in discovering genes potentially involved in diseases and brain development.
Citation:Carson JP, Ju T, Lu HC, Thaller C, Xu M, et al. (2005) A digital atlas to characterize the mouse brain transcriptome. PLoS Comput Biol 1(4): e41.
==== Body
Introduction
High-resolution maps of gene expression provide important information about how genes regulate biological processes at cellular and molecular levels. Therefore, a multitude of efforts are in progress to depict gene expression at single cell resolution in specimens ranging from organs to embryos (http://mamep.molgen.mpg.de [1]; http://genepaint.org/ [2]; http://brainatlas.org/ [3]; http://mahoney.chip.org/mahoney/ [4]; http://www.ncbi.nlm.nih.gov/projects/gensat/ [5]). Common to these genome-scale projects is that they generate vast numbers of images of expression patterns that reveal the presence of transcripts or proteins in a particular cell or group of cells within a natural context. However, large collections of images are of limited usefulness per se without efficient means to mine these images and to characterize and compare gene or protein expression patterns. In analogy to the requirements for mining genomic sequence information, meaningful retrieval of expression patterns requires suitable annotation. By annotation, we mean associating sites and strengths of expression with a digital representation of the anatomy of a specimen.
The annotation approach taken by the Gene Expression Database [6] is to hand-curate published gene expression patterns using an extensive dictionary of anatomical terms. This annotation is facilitated by the Edinburgh Mouse Atlas Project (EMAP), which provides anatomical ontology relationships using a hierarchical tree [7]. Visualization is achieved by associating these terms with locations in a volumetric model [8]. The Edinburgh Mouse Atlas Project also provides tools to map in situ hybridization (ISH) images directly into a three-dimensional (3D) atlas [7]. Although hand curation is an effective method for annotation, it is not an efficient means for handling the large-scale datasets systematically collected by robotic ISH [9]. In addition, if future changes are made to anatomical designations, updating the annotation may require a laborious review of previously annotated data.
Here we present a completely novel approach that uses a geometric modeling technique to create a digital atlas of the postnatal day 7 (P7) mouse brain. This deformable atlas can then be adjusted to match the major anatomical structures present in P7 mouse brain tissue sections, accurately define the boundaries between structures, and provide a smooth multi-resolution coordinate representation of small structures. When combining this technique with a method for detecting strength of gene expression, one can efficiently and automatically annotate a large number of gene expression patterns in a way that subsequently allows queries and comparisons of expression patterns in user-defined regions of interest.
P7 mouse brain was selected as the specimen because at this developmental stage, many complex brain functions begin to be established yet the existing information on underlying molecular mechanisms is still relatively limited. We describe here the creation of a prototype 200-gene dataset generated using robotic ISH, and the application of our deformable atlas-based annotation method to this dataset. We then demonstrate the utility of the approach with two examples: searching for genes expressed in the substantia nigra, and identifying genes potentially involved in functional regionalization of the cortex.
Results
Construction of an Atlas Using a Subdivision Mesh Technique
In building the atlas of the P7 mouse brain, we first selected a set of 11 cresyl-violet-stained standard sagittal brain sections that approximate the 11 sagittal sections in Valverde's atlas of the postnatal mouse brain [10]. These standard sections exhibit the hemi-brain by spanning from lateral (section 1) to paramedial (section 11). The boundaries of 15 major brain structures (amygdala, basal forebrain, cerebellum, cortex, globus pallidus, hippocampus, hypothalamus, medulla, midbrain, olfactory bulb, pons, septum, striatum, thalamus, and ventral striatum) were then delineated on each of the 11 standard sections. (The boundaries for standard section 4 are shown in Figure 1A.) For each of these major structures in a standard section, we created a representation using a coarse quadrilateral mesh. Figure 1B shows an example of creating a coarse mesh for the thalamus in standard section 4. The subdivision algorithm applies an iterative refinement of this coarse mesh, resulting in a fine mesh that both smoothly overlays internal regions of structures and explicitly defines their boundaries (Figure 1C). The complete mesh across an entire section is an accurate map representing all major anatomical structures (Figure 1D). Performing the described process for all 11 standard sections resulted in 11 maps, which together constitute an atlas of the P7 mouse brain (Figure S1).
Figure 1 P7 Mouse Brain Atlas Construction and Application
(A) Standard Nissl-stained P7 sagittal standard section number 4 with major anatomical boundaries drawn in red: amygdala (am), basal forebrain (bf), cerebellum (cb), cortex (ctx), globus pallidus (gp), hippocampus (hi), medulla (med), midbrain (mb), olfactory bulb (ob), pons (p), striatum (st), thalamus (th), and ventral striatum (vst).
(B) The coarse mesh, shown here for the thalamus, is constructed by defining vertices of quadrilaterals.
(C) Iterative application of subdivision generates smooth boundary curves and a smooth internal representation of smaller quadrilaterals. Fixed vertices (large squares) allow crease angles to be added to the otherwise smooth boundary curve.
(D) The atlas for standard section number 4. Each coarse quadrilateral is associated with a particular anatomical structure, an association inherited during subdivision.
(E) Expression pattern of Cannabinoid receptor 1 in a section similar to standard map 4.
(F) The atlas (D) is deformed by moving vertices so that the anatomical boundaries match those in the Cannabinoid receptor 1 section (E).
(G) Quadrilaterals overlying the DG (insert) were marked in 59 fitted maps using a mesh generated after two rounds of subdivision. In every section, the same four quadrilaterals were found to overlap the bulk of the DG.
Each of the 11 maps is deformable and hence can be precisely fitted to an anatomically similar experimental sagittal brain section (e.g., Figure 1E). The shape of the fine mesh is controlled by repositioning the vertices of the coarse mesh (indicated by black squares in Figure 1B–1D). An automated global fit of the map can be used for an initial approximation of the related map to the experimental section [11]. A manual adjustment by dragging vertices into new positions then allows the map to fit the boundaries of the anatomical structures in the experimental section accurately (Figure 1F).
Anatomical substructures in the mouse brain maintain a consistent spatial relationship with neighboring structures when specimen age and strain do not change. Thus, the location of any given substructure should be consistently represented by a set of quadrilaterals in the fitted map. This important property was examined by fitting standard map 6 to 59 different experimental sections and then determining which quadrilaterals contained the dentate gyrus (DG), a substructure of the hippocampus. Although the shape of the DG and its relative position within the hippocampus varied to some extent (e.g., because of tissue compression/stretching in the sectioning process), the same four quadrilaterals always contained most of the DG, with adjacent quadrilaterals sometimes containing the edge of the DG (Figure 1G). This suggests that the subdivision mesh-based atlas not only explicitly delineates the boundaries between major structures, but can also be used to define the location of internal substructures such as the DG.
Establishment of Annotated Gene Expression Patterns
Nonradioactive ISH data.
We have assessed the subdivision atlas with a comprehensive test dataset of ~5,000 images of entire sagittal sections from P7 mouse brain produced using robotic ISH for 200 different genes (Table S1). Each gene expression image set spans the left half of the brain and consists of at least 24 sections spaced a maximum of 200 μm apart. Digital images were captured in a bright field microscope at 1.6 μm per pixel resolution. This resolution is sufficient to view individual cell bodies and estimate the strength of expression as reflected by the amount of precipitate in each cell using a previously reported quantification algorithm, Celldetekt [12]. Figure 2 illustrates the types of data—cellular resolution images with diverse expression patterns (insets of Figure 2C and 2D)—that were subjected to annotation by subdivision mesh fittings.
Figure 2 Examples of Gene Expression Patterns
Shown are gene expression patterns revealed by nonradioactive robotic ISH on sagittal sections. Digoxigenin-tagged RNA probes hybridized to cellular mRNA are visualized via a serial amplification that produces a blue-purple signal.
(A) Purkinje cell protein 4 in section 4.
(B) Ly6/neurotoxin 1 in section 6.
(C) 4931408A02Rik in section 9 with inset showing localized expression in midbrain neurons.
(D) A230109K2Rik in section 9 with inset showing localized expression in hypothalamus.
(E) RAS protein-specific guanine nucleotide-releasing factor 1 in section 11.
(F) Gastrin releasing peptide in section 2.
(G) Nephroblastoma overexpressed gene in section 4.
(H) Somatostatin in section 6.
Linking expression levels to locations in the atlas.
From the ~24 sections for each gene, we identified the sections that best matched the anatomy represented by standard maps 2, 4, 6, 9, and 11, which collectively are sufficient to characterize all 15 different major anatomical structures in the atlas. The standard maps were deformed to fit appropriate tissue sections (e.g., Figure 1D–1F). We applied Celldetekt to classify the expression levels for cells in the tissue sections, and associated the local levels of expression with the overlying quadrilaterals in the finely subdivided mesh (e.g., Figure 1F). This created a digital dataset of cellular expression levels at all locations across 1,000 mesh-fitted experimental sections representing 200 different genes.
Knowledge Discovery Using the P7 Mouse Brain Gene Expression Patterns
Homologous pattern query.
Within the context of an anatomical atlas, comparison of expression patterns in a region of interest provides a mechanism for identifying candidate genes involved in regionalized biological or pathological processes. Idiopathic Parkinson disease (IPD) is a progressive neurodegenerative disorder characterized in part by the loss of dopaminergic neurons in the substantia nigra, resulting in decreased dopamine release in the striatum and severe impairment of motor function. To search for genes potentially involved in IPD, we performed a homologous pattern query for genes in the dataset that best match the expression pattern of dopamine transporter 1 (Slc6a3), a marker for dopaminergic neurons, in the substantia nigra (Figure 3A). Genes in the dataset were ranked by their similarity to this query pattern, calculated as the weighted sum of differences in detected cellular expression strengths across all selected quadrilaterals. The top 12 ranked genes are shown in Figure 3B. Query patterns are not limited to genes already in the dataset; as shown in the next example, queries can be performed using user-created query patterns.
Figure 3 A Search for Genes that Are Expressed in the Substantia Nigra
(A) The pattern of Slc6a3 gene expression in and around the substantia nigra at standard section number 6 is set as the query pattern for a search of all 200 expression patterns in the current dataset. Note the color-coded shading of the query pattern, with red indicating the strong expression of Slc6a3 in the substantia nigra, and grey indicating no expression in the tissue surrounding the substantia nigra.
(B) The expression patterns in the substantia nigra pars compacta of the 12 genes found to match the search criterion best are shown: dopamine receptor 2 (Drd2); vesicular monoamine transporter 2 (Slc18a2); tyrosine hydroxylase (Th); alpha synuclein (Snca); a gene encoding a nuclear orphan receptor (Nr4a2); limb expression 1 homolog (Lix1); a gene encoding an aldehyde dehydrogenase (Aldh1a1); protein tyrosine phosphatase, receptor type L (Ptprl); chaperonin subunit 8 (Cct8); synaptic vesicle glycoprotein 2c (Sv2c); transmembrane protein 1 (Tmem1); and LIM homeobox transcription factor 1 beta (Lmx1b).
Expression difference detection.
ISH can reveal changes in gene expression that result from experimental or genetic modification. The present dataset offers the opportunity to obtain a list of genes expressed in a structure that is presumed altered because of such modification. barrelless (brl) was chosen to demonstrate this type of analysis. This mutant lacks “barrels,” the discrete cylindrical structures in layer IV of the primary somatosensory cortex that receive sensory input from facial whiskers [13,14]. The phenotype associated with brl results from a loss-of-function mutation in calcium/calmodulin-stimulated adenylate cyclase 1 [15], a cAMP-synthesizing enzyme.
A search of our dataset for genes expressed more strongly in layer IV of the barrel field than in layers I and II/III (Figure 4A) returned the transcription factor retinoid-related orphan receptor beta (Rorb) (Figure 4B) and the metabotropic glutamate receptor type 2 (Grm2) (Figure 4C). We then sought to determine whether there were significant changes in the strength of Rorb and Grm2 expression in brl mice. Three pairs of P7 brains from brl mutants and their heterozygous littermate controls (possessing intact barrel maps) were subjected simultaneously to robotic ISH using Rorb and Grm2 riboprobes. Cellular expression strengths were determined using Celldetekt [12]. Subdivision mesh atlases were fitted to five adjacent 25-μm-thick tissue sections located between standard sections 2 and 3, and the identical 12 quadrilaterals were selected in each mesh to define a common region of comparison in the barrel field. Although the percentage of cells expressing Rorb was similar in control and brl tissue (brl, 96% ± 17% of control), we found in the brl brains a significant (p = 0.02) decrease in the relative percentage of cells expressing Rorb strongly (brl, 51% ± 7% of control) (Figure 4D). By contrast, an identical analysis of Grm2 expression did not reveal differences in either total expression (brl, 113% ± 14% of control) or strong expression (brl, 111% ± 35% of control).
Figure 4 Quantitative Analysis of Rorb and Grm2 Expression in Control and brl P7 Brains
(A) The dataset of 200 genes was searched using a query pattern defined as strong expression in layer IV of the somatosensory cortex (SsCx) (red) and no expression in layers I and II/III somatosensory cortex (grey) for standard section 2. Rorb and Grm2 were two of the top matches returned.
(B) The strong expression Rorb in control somatosensory cortex layer IV coincides with the anatomical shape of the barrels that are absent in the brl mouse. For both genotypes, cellular expression was detected and color-coded by signal strength using the Celldetekt software, followed by fitting of the appropriate subdivision mesh to the shape of the cortex. A row of 12 quadrilaterals in the subdivision mesh defines the area of comparison in the somatosensory cortex layer IV. Note the greater prevalence of strongly expressing cells (red) in the control tissue. Moderately expressing cells and weakly expressing cells are indicated by blue and yellow, respectively.
(C) Quantification of Grm2 expression in somatosensory cortex layer IV as described for Rorb showed no difference in expression strength distribution between control and brl.
(D) Statistical comparisons between control and brl revealed no significant changes in the percentage of somatosensory cortex layer IV cells expressing either Rorb (p = 0.8) or Grm2 (p = 0.5). However, a significant decrease in the percentage of strongly expressing cells was found for Rorb in brl (p = 0.02), but not for Grm2 (p = 0.8).
(E) The somatosensory cortex containing the barrel region was dissected as indicated (highlighted and boxed) and used for quantitative real-time PCR analysis.
(F) Consistent with the ISH data, a statistically significant decrease in Rorb expression was found in brl by quantitative real-time PCR (p = 0.008).
To validate the results of our method for difference detection, we performed quantitative real-time PCR. Mice homozygous for brl were paired with their heterozygous littermates for Rorb expression analysis of the somatosensory cortex containing the barrel region (Figure 4E). We found that Rorb expression in the brl mice was consistently and significantly lower (p < 0.01) than that in the control mice (brl, 79% ± 4% of control) (Figure 4F).
Resources Available
To facilitate distribution and application of the methods of this project, we have made the atlas, dataset, and demonstration queries publicly available online at http://www.geneatlas.org/. The atlas resources consist of the 11 Nissl-stained standard sagittal section images with the major anatomical regions labeled, and the corresponding 11 standard subdivision mesh maps. We also provide an interactive demonstration that allows visitors to deform a map onto an experimental section (as in Figure 1D–1F). All 1,000 reduced-resolution images produced by Celldetekt for this project are also available at this Web site. In addition, most of the ~5,000 images of raw ISH data are available and viewable at http://www.genepaint.org/. The 1,000 images of gene expression patterns can be queried using a graphical search tool that allows users to duplicate the searches in Figure 3 and Figure 4A, as well as to specify different regions of interest and query patterns for their own customized queries.
Discussion
In this study, we have constructed and applied a subdivision mesh-based atlas to sagittal mouse brain sections revealing the localization of transcripts visualized by ISH. Expression patterns revealed with bacterial artificial chromosome vectors [16], radioactive ISH [17], or immunohistochemistry can readily be subjected to subdivision mesh fitting and thus be represented in the atlas shown here. In addition, it may be possible to capture the architecture of fiber tract connectivity [10], micro MRI data [18], and “tissue voxel-based” microarray-based expression profiles [19] in our subdivision maps. Such multimodality will greatly enhance the discovery power of such an atlas. The subdivision mesh-based atlas can also be used to create tables with sites, levels, and patterns of expression and thus can emulate a text-based annotation procedure [20].
Generating unbiased portraits of gene expression patterns and placing these into a common spatial framework greatly facilitates the discovery of biologically important information. In the case of the brl mice, we first searched our 200-gene dataset for genes that are expressed in the developing barrel field region (see Figure 4A). The subsequent detection of down-regulation of the transcription factor gene Rorb in brl cortex (Figure 4D) raises the possibility that activity-controlled signaling, mediated by adenylate cyclase 1 in cortical map formation, converges on gene transcription. This discovery also establishes that our annotation can both identify cortical-layer-specific marker genes and estimate quantitative differences in the level of gene expression. Differences in expression levels were more dramatic when using the histology-based method, which accurately delineated the region of interest, than when using quantitative real-time PCR on RNA isolated from a block of cortical tissue (Figure 4D–4F).
The ability to align multiple known expression patterns is the strength of the method described here. We exploited this by searching for genes expressed in a pattern similar to that of Slc6a3, which encodes a dopamine transporter and is transcribed in the substantia nigra (see Figure 3). Twelve genes were identified with our homologous pattern search. Seven of these have been previously connected to IPD. alpha synuclein and a nuclear orphan receptor (Nr4a2) are causative genes in some forms of familial IPD [21,22]. dopamine receptor 2 and tyrosine hydroxylase have been implicated in IPD on the basis of polymorphisms [23,24]. LIM homeobox transcription factor 1 beta regulates domamineric neurogenesis [25]. Expression of an aldehyde dehydrogenase (Aldh1a1) has been shown recently to be decreased markedly in individuals with IPD [26]. One gene, vesicular monoamine transporter 2, is similar to the gene used as the query pattern in that both are involved in monoamine transport.
The five other identified genes have not been previously connected to IPD. synaptic vesicle glycoprotein 2c regulates synaptic vesicle exocytosis and has a particularly restricted expression pattern in comparison to other genes in its family, suggesting a potential relationship to the substantia nigra and IPD [27]. The product of chaperonin subunit 8 is involved in protein folding and assembly [28]. This biochemical property may be a link to IPD because one aspect of this disorder is protein aggregation, mostly of alpha synuclein in Lewy bodies. protein tyrosine phosphatase, receptor type L encodes a transmembrane receptor with tyrosine phosphatase activity that has been implicated in cell–cell contact [29]. limb expression 1 homolog and transmembrane protein 1 are genes with completely unknown functions. limb expression 1 homolog is initially expressed in the precursor cells of the substantia nigra and later in its pars compacta [30]. These results suggest that it is worth considering synaptic vesicle glycoprotein 2c, chaperonin subunit 8, protein tyrosine phosphatase, receptor type L, transmembrane protein 1, and limb expression 1 homolog as candidates for further investigation into their relationship with IPD. This prototype dataset demonstrates the usefulness of this approach even with a dataset of only 200 genes. By extending this dataset to thousands of genes, our approach would yield a more comprehensive set of candidate genes involved in brain functions and disease mechanisms.
Although the atlas can reliably detect expression in substructures such as the substantia nigra (see Figure 3), cortical layers (see Figure 4A and 4B), and the DG (see Figure 1G), there are limitations in how small a structure the subdivision mesh can consistently locate. This can be addressed by increasing the complexity of the mesh through additional control points. The disadvantage of increased complexity is that fitting the mesh to experimental sections will become more time-consuming. This can be alleviated by focusing on specific anatomical substructures (e.g., just the thalamus), for which new specialized maps could be created.
One of the greatest strengths of the subdivision-based atlas is the ability to fit the maps efficiently and accurately to tissue sections, despite the varying section-to-section deformations introduced by tissue fixation, sectioning, and transfer of sections to slides. By applying this mesh-fitting process, an individual can easily map the expression patterns of 10–20 genes per day. For application of the method to the entire transcriptome, future development efforts should focus on reducing the time involved in the mesh fitting process, e.g., automated fitting based on associating anatomical landmarks with each mesh vertex [31]. In addition, the subdivision method can be extended to create a 3D volumetric subdivision atlas. When coupled with a robust method to stack tissue sections into a 3D volume of gene expression patterns, a 3D subdivision atlas may allow more efficient alignments of expression patterns than a set of two-dimensional maps.
Materials and Methods
Non-radioactive ISH.
Tissue preparation, riboprobe preparation, automated ISH, and digitization were performed as previously described [9,32–34] and as described online at http://www.genepaint.org/RNA.htm. Briefly, brains were embedded in OCT and fresh frozen in a chamber that allows stereotaxic alignment of the specimen. Serial sagittal sections at 25 μm thickness were cut with a cryostat through the left half of the brain to just past the midline. Sections from a single specimen were alternately distributed into eight different sets, resulting in a spacing of 200 μm between sections within a set. Each set consisted of approximately 24 sections (four per slide, six slides). Slides were assembled into a flow-through hybridization chamber and placed into position in a Tecan (Mannedorf, Switzerland) Genesis liquid-handling robot, which performs ISH on 192 slides in less than 24 h. Digoxigenin-tagged riboprobes were produced by in vitro transcription from PCR-generated DNA templates using bacteriophage RNA polymerases. Probes were detected by a dual amplification procedure [35].
Microscopy.
After ISH, slides were cover-slipped and digitally scanned at 1.6 μm/pixel using a custom-made automated Leica (Wetzlar, Germany) microscope [9]. Images were cropped and stored in TIFF format with LWZ lossless compression.
Atlas creation.
Each standard cross-section was modeled using a Catmull-Clark subdivision mesh [36] partitioned by a network of crease curves. Our subdivision method [37] consisted of two simple transformations: bilinear subdivision that splits each quadrilateral into four subquadrilaterals followed by centroid averaging to reposition vertices (Figure S2). Each quadrilateral in the coarsest mesh was associated with the appropriate anatomical structure. This association is maintained during subdivision.
Atlas fitting.
ISH sections most similar to the selected maps were visually selected. This was a rapid step requiring less than 1 min for each gene. Standard atlas meshes were then deformed to fit ISH sections using a semi-automated process of computing an affine fit using principal component analysis, performing a local fit using iterated least squares, and verifying visually [11]. Because of the intuitive flexibility of the subdivision meshes, any necessary manual corrections of the mesh fitting were simple and could be performed in 2–5 min per ISH section.
Pattern query scoring.
As part of an expression pattern similarity query, a total difference score in relation to the query pattern is calculated for each pattern in the dataset. This score, S, is the sum of the individual differences, d, for each quadrilateral pair within the region of the search, j:
. Each d is calculated as a weighted L
1 norm between the vector of the number of cells at different Celldetekt-calculated expression strength levels, c = [strong, moderate, weak, none] for the query pattern quadrilateral, q, and the current dataset pattern quadrilateral, p. Specifically,
, with weights w = [9, 4, 1, 0].
Rorb and Grm2 analysis.
Each brl and littermate control mouse brain pair was subjected to ISH simultaneously. Prior to Celldetekt analysis, image intensity level adjustment was performed on pairs so that the percentage of strongly expressing cells was approximately equivalent from pair to pair. All p-values were calculated using two-tailed paired t-tests that compared brl brain section sets in relation to their control pairs.
RNA extraction and cDNA generation.
The somatosensory cortex was isolated from brl mice (n = 16) and heterozygous littermate control mice (n = 22) in a total of six group pairs as previously described [38]. Total RNA was extracted, cleaned with DNase I, and then reverse transcribed. Conventional PCR for Rorb was performed in samples from heterozygous control and homozygous brl animals. The PCR amplicons were sequenced to confirm their identity across control and brl samples. The resulting sequences were used for the design of TaqMan (Roche Molecular Systems, Alameda, California, United States) primers and probes for quantitative real-time PCR.
Quantitative real-time PCR.
The TaqMan probe and primer pair for Rorb were as follows: probe, 5′-FAM TCAGAAGAACCACCTGGATGATGAGACCC TAMRA-3′; forward primer, GATTTATTTTGCACTGCAACATGTG; and reverse primer, ACTGCCGTGATAGTTGGTATCTTG. Relative quantification of Rorb expression was performed with 18S rRNA as an endogenous control. Each sample was run in triplicate to reduce pipetting error and increase consistency of the results. PCR was carried out at 50 °C for 2 min, 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The expected size of the PCR products was confirmed by gel electrophoresis. In addition, a conventional PCR omitting the hybridization probe was run on a thermocycler to verify PCR specificity. Equal amplification efficiency of Rorb to 18S rRNA was achieved, validating the relative quantification.
Animals.
C57BL/6 wild type mice from Jackson Laboratory (Bar Harbor, Maine, United States) were the source of the line of mice used for all 200 genes. The discovery of brl mice resulted from a spontaneous mutation in a line from ICR stock at Université de Lausanne [39]. brl mice used in our experiments were from the eighth backcross generation of the incipient C57BL/6J-brl congenic inbred strain. Genotypes were determined by genomic PCR as described [38]. Data analysis was performed blind to genotype. All animals were treated in compliance with the guidelines of both the U.S. Department of Health and Human Services and Baylor College of Medicine's Animal Care and Use Committee.
Supporting Information
Figure S1 The Subdivision-Based Anatomical Atlas of the Postnatal Mouse Brain
Eleven sagittal maps compose this subdivision-based postnatal mouse brain atlas. The 15 major anatomical structures are color-coded as indicated.
(7.6 MB TIF)
Click here for additional data file.
Figure S2 Subdivision Mesh
(A) An initial coarse mesh (left). The two transformations of subdivision: bilinear subdivision (middle) and then centroid averaging (right).
(B) The mesh subdivided twice (left), thrice (middle), and four times (right).
(453 KB TIF)
Click here for additional data file.
Table S1 List of 200 Riboprobes Used
(242 KB DOC)
Click here for additional data file.
Accession Numbers
The Entrez (http://www.ncbi.nlm.nih.gov/gquery/) accession numbers for the genes discussed in this paper are 4931408A02Rik (XM_354970), A230109K2Rik (AK020723), Aldh1a1 (NM_013467), alpha synuclein (NM_009221), cannabinoid receptor 1 (NM_007726), chaperonin subunit 8 (NM_009840), dopamine receptor 2 (NM_010077), gastrin releasing peptide (NM_175012), Grm2 (M92075), LIM homeobox transcription factor 1 beta (NM_010725), limb expression 1 homolog (NM_025681), Ly6/neurotoxin 1 (NM_011838), nephroblastoma overexpressed gene (NM_010930), Nr4a2 (NM_013613), protein tyrosine phosphatase, receptor type L (NM_011214), Purkinje cell protein 4 (NM_008791), RAS protein-specific guanine nucleotide-releasing factor 1 (NM_011245), Rorb (NM_146095), Slc6a3 (NM_010020), somatostatin (NM_009215), Synaptic vesicle glycoprotein 2c (XM_127490), transmembrane protein 1 (XM_125775), tyrosine hydroxylase (NM_009377), and vesicular monoamine transporter 2 (NM_013031).
We thank G. Alvarez-Bolado, M. Chen, A. Liang, A. Visel, and M. Yaylaoglu for technical assistance, and D. Armstrong, M. Bello, I. Kakadiaris, and J. Maunsell for advice and discussions. This research was supported by a fellowship (National Library of Medicine grant number 5T15LM07093) from the Keck Center for Computational and Structural Biology of the Gulf Coast Consortia, the Burroughs Welcome Fund, National Institutes of Health (grants P41RR02250, EY-12696, R01 MH62639, and F32 NS11034), and the National Science Foundation (grants IBN-0078110 and EIA-0325004).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. JPC, TJ, HCL, JW, WC, and GE conceived and designed the experiments. JPC, HCL, CT, and MX performed the experiments. JPC, HCL, and GE analyzed the data. JPC, TJ, HCL, CT, MX, SLP, MCC, JW, WC, and GE contributed reagents/materials/analysis tools. JPC, TJ, HCL, CT, MX, SLP, MCC, JW, WC, and GE wrote the paper.
A previous version of this article appeared as an Early Online Release on August 16, 2005 (DOI: 10.1371/journal.pcbi.0010041.eor).
Abbreviations
3Dthree-dimensional
brl,
barrelless
DGdentate gyrus
Grm2,
metabotropic glutamate receptor type 2
IPDidiopathic Parkinson disease
ISHin situ hybridization
P7postnatal day 7
Rorb,
retinoid-related orphan receptor beta;
Slc6a3,
dopamine transporter 1
==== Refs
References
Neidhardt L Gasca S Wertz K Obermayr F Worpenberg S 2000 Large-scale screen for genes controlling mammalian embryogenesis, using high-throughput gene expression analysis in mouse embryos Mech Dev 98 77 94 11044609
Visel A Thaller C Eichele G 2004 GenePaint.org: An atlas of gene expression patterns in the mouse embryo Nucleic Acids Res 32 D552 D556 14681479
Boguski MS Jones AR 2004 Neurogenomics: At the intersection of neurobiology and genome sciences Nat Neurosci 7 429 433 15114353
Gray PA Fu H Luo P Zhao Q Yu J 2004 Mouse brain organization revealed through direct genome-scale TF expression analysis Science 306 2255 2257 15618518
Gong S Zheng C Doughty ML Losos K Didkovsky N 2003 A gene expression atlas of the central nervous system based on bacterial artificial chromosomes Nature 425 917 925 14586460
Hill DP Begley DA Finger JH Hayamizu TF McCright IJ 2004 The mouse Gene Expression Database (GXD): Updates and enhancements Nucleic Acids Res 32 D568 D571 14681482
Baldock RA Bard JB Burger A Burton N Christiansen J 2003 EMAP and EMAGE: A framework for understanding spatially organized data Neuroinformatics 1 309 325 15043218
Brune RM Bard JB Dubreuil C Guest E Hill W 1999 A three-dimensional model of the mouse at embryonic day 9 Dev Biol 216 457 468 10642785
Carson JP Thaller C Eichele G 2002 A transcriptome atlas of the mouse brain at cellular resolution Curr Opin Neurobiol 12 562 565 12367636
Valverde F 1998 Golgi atlas of the postnatal mouse brain New York Springer-Verlag 146 p.
Ju T Warren J Eichele G Thaller C Chiu W 2003 A geometric database for gene expression data Kobbelt L Schröder P Hoppe H Geometry processing 2003 Aachen (Germany) Eurographics Association 166 176
Carson JP Eichele G Chiu W 2005 A method for automated detection of gene expression required for the establishment of a digital transcriptome-wide gene expression atlas J Microsc 217 275 281 15725131
Fukuchi-Shimogori T Grove EA 2001 Neocortex patterning by the secreted signaling molecule FGF8 Science 294 1071 1074 11567107
Woolsey TA Van der Loos H 1970 The structural organization of layer IV in the somatosensory region of mouse cerebral cortex: The description of a cortical field composed of discrete cytoarchitectonic units Brain Res 17 205 242 4904874
Abdel-Majid RM Leong WL Schalkwyk LC Smallman DS Wong ST 1998 Loss of adenylyl cyclase I activity disrupts patterning of mouse somatosensory cortex Nat Genet 19 289 291 9662407
Heintz N 2004 Gene expression nervous system atlas (GENSAT) Nat Neurosci 7 483 15114362
Lein ES Zhao X Gage FH 2004 Defining a molecular atlas of the hippocampus using DNA microarrays and high-throughput in situ hybridization J Neurosci 24 3879 3889 15084669
Sadowski M Tang CY Aguinaldo JG Carp R Meeker HC 2003 In vivo micro magnetic resonance imaging signal changes in scrapie infected mice Neurosci Lett 345 1 4 12809974
Barlow C Lockhart DJ 2002 DNA arrays and neurobiology—What's new and what's next? Curr Opin Neurobiol 12 554 561 12367635
Carson JP Ju T Thaller C Warren J Bello M 2004 Automated characterization of gene expression patterns with an atlas of the mouse brain Proceedings of the 26th Annual International Conference of the IEEE EMBS 2004 September 1–5; San Francisco, California. Madison (Wisconsin) Omnipress 2917 2920
Le WD Xu P Jankovic J Jiang H Appel SH 2003 Mutations in NR4A2 associated with familial Parkinson disease Nat Genet 33 85 89 12496759
Polymeropoulos MH Higgins JJ Golbe LI Johnson WG Ide SE 1996 Mapping of a gene for Parkinson's disease to chromosome 4q21-q23 Science 274 1197 1199 8895469
Plante-Bordeneuve V Taussig D Thomas F Said G Wood NW 1997 Evaluation of four candidate genes encoding proteins of the dopamine pathway in familial and sporadic Parkinson's disease: Evidence for association of a DRD2 allele Neurology 48 1589 1593 9191771
Haavik J Toska K 1998 Tyrosine hydroxylase and Parkinson's disease Mol Neurobiol 16 285 309 9626667
Smidt MP Asbreuk CH Cox JJ Chen H Johnson RL 2000 A second independent pathway for development of mesencephalic dopaminergic neurons requires Lmx1b Nat Neurosci 3 337 341 10725922
Galter D Buervenich S Carmine A Anvret M Olson L 2003 ALDH1 mRNA: Presence in human dopamine neurons and decreases in substantia nigra in Parkinson's disease and in the ventral tegmental area in schizophrenia Neurobiol Dis 14 637 647 14678778
Janz R Sudhof TC 1999 SV2C is a synaptic vesicle protein with an unusually restricted localization: Anatomy of a synaptic vesicle protein family Neuroscience 94 1279 1290 10625067
Stoldt V Rademacher F Kehren V Ernst JF Pearce DA 1996 Review: The Cct eukaryotic chaperonin subunits of Saccharomyces cerevisiae and other yeasts Yeast 12 523 529 8771707
Yan HX He YQ Dong H Zhang P Zeng JZ 2002 Physical and functional interaction between receptor-like protein tyrosine phosphatase PCP-2 and beta-catenin Biochemistry 41 15854 15860 12501215
Moeller C Yaylaoglu MB Alvarez-Bolado G Thaller C Eichele G 2002 Murine Lix1, a novel marker for substantia nigra, cortical layer 5, and hindbrain structures Brain Res Gene Expr Patterns 1 199 203 12638132
Kakadiaris I Bello M Arunachalam S Kang W Ju T 2004 Landmark-driven, atlas-based segmentation of mouse brain tissue slices containing gene expression data Barillot C Haynor D Hellier P Medical image computing and computer-assisted intervention—MICCAI 2004 7th International Conference, Proceedings Part I; 26–29 September 2004, Rennes/Saint-Malo, France. pp. 192–199.
Visel A Ahdidan J Eichele G 2002 A gene expression map of the mouse brain: Genepaint.org—A database of gene expression patterns Kotter R Neuroscience databases: A practical guide Boston Kluwer Academic Publishers 19 36
Herzig U Cadenas C Sieckmann F Sierralta W Thaller C 2001 Development of high-throughput tools to unravel the complexity of gene expression patterns in the mammalian brain Bock G Goode J Novartis Foundation Symposium: Complexity in biological information processing Chicester John Wiley and Sons 129 149
Yaylaoglu MB Titmus A Visel A Alvarez-Bolado G Thaller C 2005 Comprehensive expression atlas of fibroblast growth factors and their receptors generated by a novel robotic in situ hybridization platform Dev Dyn E-pub ahead of print.
Reymond A Marigo V Yaylaoglu MB Leoni A Ucla C 2002 Human chromosome 21 gene expression atlas in the mouse Nature 420 582 586 12466854
Catmull E Clark J 1978 Recursively generated B-spline surfaces on arbitrary topological meshes Comput Aided Des 16 350 355
Warren J Weimer H 2002 Subdivision methods for geometric design: A constructive approach San Francisco Morgan Kaufmann 299 p.
Lu HC She WC Plas DT Neumann PE Janz R 2003 Adenylyl cyclase I regulates AMPA receptor trafficking during mouse cortical ‘barrel' map development Nat Neurosci 6 939 947 12897788
Van der Loos H Welker E Dorfl J Rumo G 1986 Selective breeding for variations in patterns of mystacial vibrissae of mice. Bilaterally symmetrical strains derived from ICR stock J Hered 77 66 82 3711643
|
16184189
|
PMC1215388
|
CC BY
|
2021-01-05 09:18:22
|
no
|
PLoS Comput Biol. 2005 Sep 23; 1(4):e41
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010041
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1615930810.1371/journal.pmed.0020303Policy ForumBioethicsEpidemiology/Public HealthHealth EconomicsHealth PolicyHIV/AIDSResource allocation and rationingHealth PolicyHIV Infection/AIDSHealth EconomicsMedicine in Developing CountriesRationing Antiretroviral Therapy for HIV/AIDS in Africa: Choices and Consequences Policy ForumRosen Sydney *Sanne Ian Collier Alizanne Simon Jonathon L Sydney Rosen, Alizanne Collier, and Jonathon L. Simon are at the Center for International Health and Development, Boston University, Boston, Massachusetts, United States of America. Ian Sanne is at the Clinical HIV Research Unit, University of the Witwatersrand, Johannesburg, South Africa.
Competing Interests: Ian Sanne is the Chief Executive Officer of Right to Care, a not-for-profit organization in South Africa that provides treatment to HIV/AIDS patients.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 20 9 2005 2 11 e303Copyright: © 2005 Rosen et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.The question facing African governments and societies, say Rosen and colleagues, is not whether to ration such therapy, but how to do so in a way that maximizes social welfare.
==== Body
In the past three years, expanding access to antiretroviral therapy (ART) for HIV/AIDS has become a global objective and a national priority for many countries in sub-Saharan Africa. Large-scale treatment programs have been launched in countries spanning the continent from Lesotho to Ghana, paid for by domestic funds mobilized by African governments and by international donor contributions.
While these funds, which reach into the billions of dollars, will pay for ART for many thousands of HIV-positive Africans, there is almost no chance that African countries will have the human, infrastructural, or financial resources to treat everyone who is in need. National plans for treatment rollout typically call for a specific number of patients to initiate therapy within the first one or two years of the program. Though the target patient numbers are extremely ambitious—often requiring a 10-fold expansion of services over a two-year period—they still represent a minority of those who are eligible for antiretrovirals on even the most conservative medical grounds. Table 1 indicates the demand for and supply of ART in several African countries and globally, based on starting ART at a CD4 count of 200 cells/μl or an AIDS-defining illness.
Table 1 Targets for Treatment Coverage in Selected African Countries and Globally
Data taken from [40] unless otherwise indicated.
a This is the midpoint of the high and low estimates made by the World Health Organization [40]. For most countries, it appears to include estimates of patients in the public sector, nongovernmental sector, and private sector.
b Calculated as the sum of the estimated number eligible for therapy in 2004 [40] plus 6% of the adult HIV-positive population at the end of 2003 as estimated by UNAIDS [46], which is the proportion expected to become eligible over the course of 2005 due to normal disease progression [5].
c End 2005 target as multiple of number on therapy, December 2004.
d Target for July 2005.
The message of Table 1 is clear: rationing of ART is already occurring and will persist for many years to come. The question facing African governments and societies is not whether to ration ART, but how to do so in a way that maximizes social welfare, now and in the future.
Inevitably, the social and economic consequences of rationing a scarce and valuable resource—treatment for a life-threatening illness—will vary widely depending on the rationing system chosen. In a previous article [1], we argued that the chances of achieving a socially desirable outcome from the global intervention now being launched will be higher if an open public-policy debate is conducted and policies are selected that make transparent the trade-offs inherent in any rationing system. We also identified a number of possible rationing systems and proposed several criteria that could be used to select among them. In this paper, we examine these issues in more detail and use an expanded set of criteria to evaluate several rationing systems that already exist in sub-Saharan Africa.
Systems for Rationing
In economic terms, any policy or practice that restricts consumption of a good is a rationing system [2]. A rationing system restricts demand for a scarce resource so that it matches supply [3]. In the marketplace, price is the basis for rationing: those who can and are willing to pay the market price obtain the resource, while those who cannot or will not pay go without. Nonmarket goods, such as access to free or subsidized medical care, are rationed in a variety of other ways [4]. As used by economists, rationing is a morally neutral concept. It does not imply an intent to deprive some people of a good, but rather describes the allocation of a resource of which there is not enough to go around. Non-price rationing of health care has a long history and is widespread and accepted in many parts of the world, reflecting the widely held view that access to health care should be based on some notion of need, and not determined solely by ability to pay [4]. At the same time, non-price rationing is inherently political. It can be, and often is, used to channel resources toward or away from particular groups for reasons unrelated to their absolute or relative need for the resource.
Different approaches to rationing HIV drugs will have different social and economic consequences for African populations
(Illustration: Margaret Shear)
In this paper, we define an ART rationing system as any allocation of public resources that prioritizes access to HIV/AIDS treatment on the basis of any geographic, social, economic, cultural, or other nonmedical factor. This is important, as virtually all programs will set a medical threshold for access to treatment, in most cases having a CD4 count lower than 200 cells/μl or an AIDS-defining illness. A less conservative medical eligibility threshold, such as that of the United States Department of Health and Human Services, which recommends that ART be started at a CD4 count of 350 cells/μl, would dramatically increase the number of eligible patients and intensify the need for rationing [5]. Even with the more conservative eligibility threshold now being applied, however, the figures in Table 1 indicate that demand for treatment will exceed supply. In the remainder of this paper, we will focus our attention on the nonmedical bases for rationing.
Explicit Rationing Systems
In many cases, governments will set explicit criteria for which types of patients should be eligible for ART first or at lowest cost. The criteria can target selected subpopulations directly, or they can set eligibility requirements that intentionally give some patients better access than others. Possible subpopulations for direct targeting of treatment include:
Mothers of new infants
Rather than face an ever-increasing burden of orphan support, many countries are making ART preferentially available to HIV-positive mothers through testing and treatment at antenatal clinics. The “MTCT-Plus” initiative (“MTCT” is mother-to-child transmission, and “plus” refers to an essential HIV care package for the mother in addition to strategies to reduce MTCT), which has been implemented in many African countries, is the main example of this strategy [6].
Skilled workers
African countries face the loss of vast numbers of educated or trained workers, whose skills are vital to maintaining social welfare, sustaining output, and generating economic growth. Human capital can be conserved by giving treatment priority to nurses, teachers, engineers, judges, police officers, and other skilled workers whose contributions are important to economic development or social stability. Botswana [7], Zambia [8], and Uganda [9] have recently announced plans to target soldiers, university faculty and students, and civil servants, respectively.
Poor people
The social justice agenda pursued by some governments and many nongovernmental organizations argues that the poorest members of society, who are least likely to be able to afford private medical care, should have preferential access to publicly funded treatment programs [10]. Means-testing, which can be applied at the level of the household or the community and calibrated to achieve the desired number of patients, is a common way to ration social benefits [11].
High-risk populations
The extent to which ART can curb HIV transmission is a subject of current debate in the literature [12,13]. If treatment reduces the probability of transmission by suppressing viral load, then a public health argument can be made for giving preferential access to high-risk populations, such as commercial sex workers, truck drivers, or intravenous drug users.
Governments can also intentionally create eligibility requirements that result in rationing, without specifying particular target populations. Rationing systems of this type include:
Residents of designated geographic areas
One obvious way to limit access to treatment is to offer it only to those who reside in specified geographic catchment areas [14]. These areas can be distributed around the country, centered in regions of high HIV prevalence, or concentrated in urban centers or politically important regions. Excluding patients who do not live within the designated areas may not be feasible, but most patients will not be able to afford the cost of regular transport or permanent relocation.
Ability to co-pay
If patients are required to contribute even a small share of the cost of treatment, the number who can access therapy is likely to fall dramatically. Governments could in principle match supply and demand by setting and adjusting the level of co-payment required. The obvious outcome is a rationing system that favors the upper socioeconomic tiers of patients, who likely include the majority of skilled workers. In some societies men will also have preferential access when a cash payment is required [15]. A drawback of requiring co-payment is that poorer patients may stop therapy because they run out of funds. This is the reason for stopping cited by nearly half of all non-adherent patients in a recent study in Botswana [16].
Commitment to adherence to therapy
Adherence to treatment regimens has been found to be the most important determinant of the success of ART at the individual patient level [17]. One way to improve the success of a large-scale treatment program, while at the same time limiting access, could therefore be to restrict therapy to patients who are judged to have the ability and willingness to adhere or who demonstrate high adherence after initiating therapy. Results of pilot projects suggest that requiring attendance at pre-treatment counseling sessions helps to screen for adherence commitment, for example [18].
Implicit Rationing Systems
The alternative to specifying explicitly who will have priority access to resources is to allow implicit rationing systems to arise. These can be thought of as the default conditions that will prevail in the absence of explicit choices.
Access to HIV testing
Voluntary counseling and HIV testing (VCT) is typically the entry point into an HIV/AIDS treatment program. If some subpopulations, such as youth or particular occupational groups, are targeted for HIV education and VCT services or promotion campaigns, they will have an advantage over others in seeking treatment, as will those who simply live closer to VCT facilities [19].
Patient costs
Most countries will scale up their treatment programs incrementally, at first offering services at only a few facilities before gradually adding more. Ghana started with four public treatment sites in 2004, for example, but is aiming to have 16 in operation by the end of 2006 [20]. For most patients, bus or taxi fare will be required for regular trips to the clinic, and each trip will take up a good deal of time. Previous research has found that indirect costs due to travel time and transport play an important role in limiting access to medical care [21–24]. Unless transport is subsidized, limiting the number of service sites will effectively ration treatment to those who live nearby and to better-off households that have the resources to travel.
First come, first served
In the absence of any other requirements, most facilities are likely to treat everyone who is medically eligible, until the supply of drugs, diagnostics, or expertise runs out. Patients who arrive after that happens may be put on a waiting list, sent to another facility, or simply sent away. This approach, which reflects an absolute shortage of treatment “slots,” is likely to favor three groups of patients: those who are already paying privately for antiretroviral drugs and shift over to publicly funded treatment once it is available; those who develop AIDS-related symptoms first, in most cases because they were infected earliest; and the few HIV-positive individuals who do not yet have AIDS but have taken the initiative to go for a test and know their own status.
Queuing
One of the most common ways to ration scarce resources is the time-honored, time-consuming tradition of queuing. While it is possible to create a waiting list that keeps track of individuals' places in line, in many African countries the queue is a literal line outside the clinic door. Such queuing will favor patients whose opportunity cost of time is low [23]. This group is likely to be dominated by unemployed men and by women who can bring their small children with them. It may penalize employed persons and farming households that face a high seasonal demand for labor.
No matter what system is used, informal and/or illicit arrangements can often be made that give preferential access to treatment to those with social, economic, or political influence. In all of the implicit systems, and in some of the explicit ones, there will very often be a high degree of queue jumping. Elites capture a disproportionate share of resources in all countries; in developing countries, where enforcement of rules tends to be weak and informal arrangements common, it is safe to assume that members of the elite who are medically eligible for therapy will find a way to get it. De facto rationing on the basis of social or economic position will thus occur. It is the phenomenon of queue jumping that turns what appear to be equitable, if inefficient, rationing systems, such as first-come, first-served, into an inequitable and inefficient approach.
Many other potential criteria for rationing ART have been proposed or are in use [11,14,25,26]. Treatment access could be targeted, for example, to young people (because they respond best to the therapy and have their most productive years ahead of them); families of current patients (to promote adherence); those with debts (so that the loan default rate does not increase); patients with tuberculosis (to suppress transmission of tuberculosis); or children (who are least able to protect themselves).
Evaluating the Systems
The different approaches to rationing ART described above will inevitably have very different social and economic consequences for African populations. In this section, we assess the rationing systems' probable outcomes using criteria that capture most of the principles that governments use to evaluate policies and social investments. They are by no means the sole criteria of interest, nor should they necessarily be given equal weight. We propose them only as a starting point for thinking about the consequences of alternative approaches.
Effectiveness
Does the rationing system produce a high rate of successfully treated patients? “Successful treatment” could be defined as a fully suppressed viral load or high CD4 count over a sustained period of time. It might also incorporate some measure of viral resistance to the drugs. We assume that early diagnosis and high adherence improve effectiveness and that patient motivation improves adherence, but level of education and socioeconomic status do not [27].
Cost savings
Is the cost per patient treated low, compared to other approaches? Cost is characterized in this way to maintain internal consistency: a rating of “high” in this domain is desirable, as it is for all other criteria. Cost is defined broadly, to incorporate costs incurred by patients, providers, insurers, and the public health system, including the identification of medically eligible patients and management of opportunistic infections, side effects, and treatment failure.
Feasibility
Are the human and infrastructural resources needed for implementation available? We define an approach as feasible if there are no obstacles to carrying it out that appear to be insurmountable under typical conditions in sub-Saharan Africa.
Economic efficiency
To what extent does the system mitigate the long-term impacts of the HIV epidemic on economic development? AIDS has the potential to affect economic development in many ways [28]. We focus on human capital accumulation, where human capital is defined as the accumulated skill, knowledge, and expertise of workers [29].
Social equity
Do all medically eligible patients, including those from poor or disadvantaged subpopulations, have equal access to treatment? We define “equity” as equitable access for all at the current time, not redistribution of resources to redress past injustices, and we assume that a system's susceptibility to queue-jumping reduces its equity.
Rationing potential
Will the chosen system sufficiently reduce the number of patients? The purpose of any rationing system is to match demand to the available supply.
Impact on HIV transmission
To what extent does treatment reduce HIV incidence? Preferentially treating those who are likely to transmit the virus could reduce HIV incidence more than treating those who are not likely transmitters [13,30].
Sustainability
Can the system be sustained over time? This criterion pertains to the durability of the source of funding. We assume that donor support will hold out for some time but will ultimately ebb, leaving national governments responsible for an increasing share of the costs of treatment [31].
Effect on the health care system
How does the system for allocating ART affect the country's health care system as a whole? The choice of rationing strategies could influence whether expanding treatment access will strengthen general health services for poor communities or drain resources from non-HIV health care to meet the demand for ART, further crippling general health services.
To demonstrate how one might evaluate alternative rationing strategies, Table 2 uses the first six of these criteria to compare each of the rationing systems described above. The table omits the impact on HIV transmission and the impact on resistance, because there is little agreement on how the treatment of different populations is likely to affect these outcomes. It also omits the two long-term “systems” criteria—sustainability and effect on the health care system—because most treatment programs are so new that even informed speculation is difficult to offer.
Table 2 Comparison of Rationing Systems
There are several limitations to the analysis presented in Table 2. First, we do not “know” the outcomes of the strategies described above, because most of them have either not yet been tried or, if tried, have not been evaluated. Prior experience in delivering health care in sub-Saharan Africa suggests that some of our assumptions and ratings are correct. We are confident, for example, that an implicit rationing system based on queuing will result in queue-jumping, and therefore be inequitable. We are also confident that targeting skilled workers will improve labor productivity and therefore promote economic efficiency. Our ratings of some of the other systems, in contrast, are largely speculative. Table 2 also cannot capture the possibility that outcomes will vary by country, setting, or context.
A second limitation involves the criteria we selected for evaluation. We applied six criteria that we believe capture the key considerations in designing an HIV/AIDS treatment program, and we identified but did not apply three others that could alter the overall outcomes. Undoubtedly, there are other criteria that also matter. We also could not account for potential interactions among the criteria. Cost and feasibility are clearly related, for example; at some level of cost, any system could be considered feasible. Many would argue that social equity is essential to sustainable economic development, and that efficiency and equity cannot therefore be separated. While we recognize that these relationships exist, neither data nor experience allows us to address them here.
Conclusions
During the initial months of existing ART programs in sub-Saharan Africa, limited access to health care services and widespread reluctance to be tested for HIV or enroll in treatment programs have greatly limited patient numbers [32,33]. This phenomenon, for as long as it persists, may prevent demand from exceeding supply, and no rationing will be necessary. There are already some situations, however, in which patients demanding access to care have overwhelmed available resources [34]. Even under the most optimistic scenarios for reaching universal coverage, there will be a period of at least several years when treatment is scarce.
Rationing of medical care is not a new phenomenon, nor is it by any means limited to developing countries. Waiting lists, whether for specific procedures, organs for transplant, or experimental treatments, are common in North America and Europe. Many state governments in the US are explicitly limiting access to more expensive AIDS drugs [35]. The HIV/AIDS crisis in Africa is simply bringing the need for rationing into stark relief.
There is no single rationing system, or combination of systems, that will be optimal for all countries at all times. Table 2 highlights the trade-off between economic efficiency and social equity: rationing systems that rate high in terms of efficiency generally rate low in terms of equity. African societies will place different weights on the values inherent in goals such as equity and efficiency, and decisions about rationing will be made at multiple levels of the health care system. International funding agencies have already begun to express their priorities through the amounts and conditions of their grants. Ministries of health will set policies that reflect national priorities, followed by district and local departments of health. Even individual health care workers, such as nurses at clinics where antiretroviral drugs are available but scarce, will be forced to ration access to patients who meet the clinic's or their own criteria [36].
Because access to antiretroviral drugs is a matter of life or death for patients with AIDS, the choice of rationing systems matters deeply. African governments can take one of two courses: ration deliberately, on the basis of explicit criteria, or allow implicit rationing to prevail. Implicit rationing is not likely to maximize social welfare, nor does it allow for transparency and accountability in policy making. We believe that the magnitude of the intervention now underway and the importance of the resource allocation decisions to be made call for public participation, policy analysis, and political debate in the countries affected. Several proposals have been made for how such processes could be carried out ([11,26,38,39]; A. Acharya, unpublished data). In the absence of such processes, decisions about access to treatment will be made arbitrarily and will, most likely, result in inequity and inefficiency—the worst of both worlds. Governments that make deliberate choices, in contrast, are more likely to achieve a socially desirable return from the large investments now being made than are those that allow queuing and queue-jumping to dominate. Countries that promote an open policy debate have the opportunity to ration ART in a manner that sustains both economic development and social cohesion—in the age of AIDS, the best of both worlds.
Funding for the research presented in this paper was provided by the South Africa Mission of the US Agency for International Development through the Child Health Research Project, G/PHN/HN/CS, Global Bureau, the US Agency for International Development (USAID), under the terms of Cooperative Agreement No. HRN-A-00-96-90010-00, the Applied Research on Child Health Project, and by the South Africa Mission of USAID through Cooperative Agreement No. 674-A-00-02-00018 to Right to Care. The opinions expressed herein are those of the authors and do not necessarily reflect the views of USAID. The funding agency did not influence the conduct or outcomes of the analysis or exercise any editorial control over this paper.
Citation: Rosen S, Sanne I, Collier A, Simon JL (2005) Rationing antiretroviral therapy for HIV/AIDS in Africa: Choices and consequences. PLoS Med 2(11): e303.
Abbreviations
ARTantiretroviral therapy
USAIDthe US Agency for International Development
VCTvoluntary counseling and HIV testing
==== Refs
References
Rosen S Sanne I Collier A Simon JL Hard choices: Rationing antiretroviral therapy for HIV/AIDS in Africa Lancet 2005 365 354 356 15664233
Stiglitz JE Economics of the public sector 2004 New York: Norton Second Edition 754
European Observatory on Health Systems and Policies Glossary 2005 Available: http://www.euro.who.int/observatory/glossary/toppage . Accessed 22 February 2005
Maynard A Rationing health care: An exploration Health Policy 1999 49 5 11 10827287
Auvert B Males S Puren A Taljaard D Carael M Can highly active antiretroviral therapy reduce the spread of HIV?: A study in a township of South Africa J Acquir Immune Defic Syndr 2004 36 613 621 15097305
Mitka M MTCT-Plus program has two goals: End maternal HIV transmission + treat mothers JAMA 2002 288 153 154 12095369
[Anonymous] Botswana: Anti-AIDS drugs for armed forces 2005 Available: http://www.plusnews.org/AIDSReport.ASP?ReportID=4575 . Accessed 15 March 2005
[Anonymous] Zambia: University launches free ARV treatment on campus 2005 Available: http://www.plusnews.org/AIDSreport.asp?ReportID=4446 . Accessed 4 February 2005
[Anonymous] Uganda: Free ARVs for HIV-positive civil servants 2005 Available: http://www.plusnews.org/AIDSreport.asp?ReportID=4680 . Accessed 8 April 2005
[Anonymous] Zambia: MSF says rural poor lack access to AIDS drugs 2004 Available: http://www.plusnews.org/AIDSReport.ASP?ReportID=2971 . Accessed 28 January 2004
Macklin R Ethics and equity in access to HIV treatment: 3 by 5 initiative 2004 Available: http://www.who.int/ethics/en/background-macklin.pdf . Accessed 22 February 2005
Blower SM Bodine E Kahn J McFarland W The antiretroviral rollout and drug-resistant HIV in Africa: Insights from empirical data and theoretical models AIDS 2005 19 1 14 15627028
Salomon JA Hogan DR Stover J Stanecki KA Walker N Integrating HIV prevention and treatment: From slogans to impact PLoS Med 2005 2 e16 10.1371/journal.pmed.0020016 15647780
[Anonymous] Africa: Treatment criteria—Deciding who gets to live 2004 Available: http://www.plusnews.org/webspecials/ARV/afrdec.asp . Accessed 17 December 2004
[Anonymous] Zambia: Second-class women left behind in access queue 2004 Available: http://www.plusnews.org/AIDSreport.asp?ReportID=4295 . Accessed 18 December 2004
Weiser S Wolfe W Bangsberg D Thior I Gilbert P Barriers to antiretroviral adherence for patients living with HIV infection and AIDS in Botswana J Acquir Immune Defic Syndr 2003 34 281 288 14600572
Garcia DO Knobel H Carmona A Guelar A Lopez-Colomes JL Impact of adherence and highly active antiretroviral therapy on survival in HIV-infected patients J Acquir Immune Defic Syndr 2002 30 105 110 12048370
Hosseinipour MC Kazembe PN Sanne IM van der Horst CM Challenges in delivering antiretroviral treatment in resource poor countries AIDS 2002 16 Suppl 4 S177 S187 12699015
Khumalo-Sakutukwa G Routh J Fiamma A Lane T Fritz K Facilitating acceptance of HIV testing: Mobile HIV voluntary counseling and testing in Sub-Saharan Africa [abstract] 2004 In: XVth International AIDS Conference; Bangkok, Thailand, July 11–16. Abstract nr WeOrE1268
[Anonymous] Ghana: Government makes ARV drugs more widely available 2005 Available: http://www.plusnews.org/AIDSreport.asp?ReportID=4356 . Accessed 13 January 2005
Dor A Gertler P Van Der GJ Non-price rationing and the choice of medical care providers in rural Cote d'Ivoire J Health Econ 1987 6 291 304 10285439
Melese M Alemayehu W Friedlander E Courtright P Indirect costs associated with accessing eye care services as a barrier to service use in Ethiopia Trop Med Int Health 2004 9 426 431 14996373
Ensor T Cooper S Overcoming barriers to health service access: Influencing the demand side Health Policy Plan 2004 19 69 79 14982885
Scott VE Chopra M Conrad L Ntuli A How equitable is the scaling up of HIV service provision in South Africa? S Afr Med J 2005 95 109 113 15751205
Wilson DP Blower SM Designing equitable antiretroviral allocation strategies in resource-constrained countries PLoS Med 2005 2 e50 10.1371/journal.pmed.0020050 15737005
Committee on Examining the Probable Consequences of Alternative Patterns of Widespread Antiretroviral Drug Use in Resource-Constrained Settings Scaling up treatment for the global AIDS pandemic: Challenges and opportunities 2004 Washington (District of Columbia) National Academies Press 325
Orrell C Bangsberg DR Badri M Wood R Adherence is not a barrier to successful antiretroviral therapy in South Africa AIDS 2003 17 1369 1375 12799558
Haacker M The macroeconomics of HIV/AIDS 2004 Available: http://www.imf.org/external/pubs/ft/AIDS/eng/ . Accessed 7 August 2005
[Anonymous] Biz/ed economics glossary 2004 Available: http://bized.ac.uk/glossary/econglos.htm . Accessed 9 February 2004
Porco TC Martin JN Page-Shafer KA Cheng A Charlebois E Decline in HIV infectivity following the introduction of highly active ART AIDS 2004 18 81 88 15090833
Attawell K Mundy J Provision of ART in resource-limited settings: A review of experience up to August 2003 2003 Available: http://www.who.int/3by5/publications/documents/en/ARTpaper_DFID_WHO.pdf . Accessed 11 August 2005
Kilewo C Massawe A Lyamuya E Semali I Kalokola F HIV counseling and testing of pregnant women in sub-Saharan Africa: Experiences from a study on prevention of mother-to-child HIV-1 transmission in Dar es Salaam, Tanzania J Acquir Immune Defic Syndr 2001 28 458 462 11744835
Maman S Mbwambo J Hogan NM Kilonzo GP Sweat M Women's barriers to HIV-1 testing and disclosure: Challenges for HIV-1 voluntary counselling and testing AIDS Care 2001 13 595 603 11571006
[Anonymous] Africa: Focus on Mozambique—ART in Africa 2004 Available: http://www.plusnews.org/webspecials/ARV/ovrmoz.asp . Accessed 16 December 2004
[Anonymous] Increased number of low-income HIV-positive people, high price of Fuzeon creating care ‘rationing’ 2004 Available: http://www.kaisernetwork.org/daily_reports/rep_index.cfm?hint=1&DR_ID=21639 . Accessed 13 January 2004
Loewenson R McCoy D Access to antiretroviral treatment in Africa BMJ 2004 328 241 242 14751875
Cheek R Playing God with HIV: Rationing HIV treatment in southern Africa 2001 Available: http://www.iss.org.za/PUBS/ASR/10No4/Cheek.html . Accessed 22 February 2005
Daniels N Fair process in patient selection for antiretroviral treatment in WHO's goal of 3 by 5 Lancet 2005 366 169 171 16005341
McGough LJ Reynolds SJ Quinn TC Zenilman JM Which patients first? Setting priorities for antiretroviral therapy where resources are limited Am J Public Health 95 1173 1180
World Health Organization “3 × 5” progress report 2004 Geneva World Health Organization Available: http://www.who.int/3by5/progressreport05/en/ . Accessed 8 August 2005
[Anonymous] The treatment era: ART in Africa: Treatment map 2004 Available: http://www.plusnews.org/aids/treatment.asp . Accessed 12 January 2004
[Anonymous] Malawi hopes to treble AIDS drug rollout 2005 Available: http://www.iol.co.za/index.php?set_id=1&click_id=68&art_id=qw110535084416M443 . Accessed 15 January 2005
[Anonymous] Nigeria: Authorities predict 250,000 people on ARVs by mid-2006 2005 Available: http://www.plusnews.org/AIDSreport.asp?ReportID=4537 . Accessed 1 March 2005
South Africa Department of Health Operational plan for comprehensive HIV and AIDS care, management, and treatment for South Africa 2003 Available: http://www.info.gov.za/otherdocs/2003/aidsplan.pdf . Accessed 22 February 2005
Wines M Zambia's president vows to extend AIDS treatment by 2005 2004 Available: http://query.nytimes.com/gst/health/article-page.html?res=9504E6D8163DF936A15753C1A9629C8B63 . Accessed 7 August 2005
Joint United Nations Programme on HIV/AIDS 2004 Report on the global AIDS epidemic 2004 Geneva Joint United Nations Programme on HIV/AIDS 231
Desvarieux M Landman R Liautaud B Girard PM Antiretroviral therapy in resource-poor countries: Illusions and realities Am J Public Health 2005 95 1117 1122 15933242
Ekouevi DK Rouet F Becquet R Inwoley A Viho I Immune status and uptake of antiretroviral interventions to prevent mother-to-child transmission of HIV-1 in Africa J Acquir Immune Defic Syndr 2004 36 755 757 15167296
Gill CJ Hamer D Simon JL Thea DM Sabin L No room for complacency about adherence to antiretroviral therapy in sub-Saharan Africa AIDS 2005 19 1243 1249 16052078
|
16159308
|
PMC1215468
|
CC BY
|
2021-01-05 10:40:26
|
no
|
PLoS Med. 2005 Nov 20; 2(11):e303
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020303
|
oa_comm
|
==== Front
PLoS MedPLoS MedpmedplosmedPLoS Medicine1549-12771549-1676Public Library of Science San Francisco, USA 1615930710.1371/journal.pmed.0020317Policy ForumHealth PolicyMedical EthicsMedical LawMental HealthPsychiatryPsychiatryHealth PolicyHuman rightsEthicsPsychiatric Health Laws in Pakistan: From Lunacy to Mental Health Policy ForumGilani Ahmed Ijaz *Gilani Umer Ijaz Kasi Pashtoon Murtaza Khan Murad Musa Ahmed Ijaz Gilani is in the Department of Basic Health Sciences, Shifa College of Medicine, Islamabad, Pakistan. Umer Ijaz Gilani is an intern at Orr Dignam and Company Advocates, Islamabad, Pakistan. Pashtoon Murtaza Kasi is a medical student at Aga Khan University, Karachi, Pakistan. Murad Musa Khan is in the Department of Psychiatry, Aga Khan University, Karachi, Pakistan.
Competing Interests: The authors declare that no competing interests exist.
*To whom correspondence should be addressed. E-mail: [email protected] 2005 20 9 2005 2 11 e317Copyright: © 2005 Gilani et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Recent changes in mental health services and in the laws governing treatment if mentally ill individuals are encouraging, but further improvement is needed.
==== Body
The South Asian country of Pakistan is the sixth most populous nation in the world. Although modern-day Pakistan came into being only 58 years ago, it is heir to a rich historical heritage spanning thousands of years, first records of which date back to the 4,000-year-old Indus valley civilization. The sociodemographic and socioeconomic characteristics of Pakistan today are shown in Table 1; the major features are a high fertility rate, a correspondingly large, young, and predominantly rural population, and a poor economy.
Table 1 Socioeconomic and Demographic Variables and Psychiatric Facilities in Pakistan
Please see http://www.pak.gov.pk for more information. GDP, gross domestic product.
One of the major health-care problems of the country is mental illness. A systematic review of risk factors, prevalence, and treatment of anxiety and depressive disorders in Pakistan found that the overall mean prevalence of these disorders in the community was 34% (range is 29%–66% for women and 10%–33% for men) [1]. Factors positively associated with these disorders were female sex, middle age, low level of education, financial difficulty, being a housewife, and relationship problems—suggesting that social factors play an important part in the aetiology of anxiety and depression in Pakistan. Other major mental health problems are developmental disorders, psychosis, and drug abuse, although credible estimates for these are lacking.
Islam plays a major role in determining the value system of Pakistani society. On the one hand, society is generally contemptuous of and biased against individuals who are mentally ill [2,3]. On the other hand, good treatment of individuals who are mentally ill is deemed greatly desirable under the society's strong religious and ethical values.
In this article, we examine the infrastructure of mental health services in Pakistan and the Pakistani laws that govern treatment of individuals who are mentally ill. We look critically at how these laws have changed over time. Changes in these laws reflect deeper changes in Pakistani society's attitude towards individuals who are mentally disordered. We stress the need for further improvement in these laws and suggest that Pakistani mental health laws should meet international standards for the treatment of mentally ill people.
Mental Health Infrastructure
There are many players and factors involved in the access, provision, delivery, functioning, and uptake of mental health services in Pakistan (Figure 1; Table 1). Awareness about mental illness is still poor in Pakistan. Such illness is generally attributed to supernatural causes—it is considered to be a curse, a spell, or a test from God.
Figure 1 The Players and Factors Involved in the Access, Provision, Delivery, Functioning, and Uptake of Mental Health Services in Pakistan
(Image: Aslam Bashir, Aga Khan University)
Those who experience mental illness often turn first to religious healers, rather than mental health professionals, since patients and their families tend to have great faith in these healers. Religious healers use verses from the Koran to treat patients. Next, patients turn to traditional and alternative healers, who are also popular in Pakistani society.
Help from the mainstream health-care system is usually sought late in the course of the illness. In Pakistan's health-care system, a pyramidal model is followed, starting with primary health care at the bottom (Figure 2). However, the referral system is inefficient and, particularly in the case of individuals who are mentally ill, patients are usually taken by their families directly to tertiary or specialist hospitals, rather than to primary-care practitioners. It is, however, important to note that many mental illnesses can be treated and managed by primary-care practitioners. The private sector also plays a major role in providing psychiatric care. For those who can afford it, private psychiatric care is an option frequently used.
Figure 2 The Usual Model of Seeking Help for Mental Health Illnesses in Pakistan
(Image: Aslam Bashir, Aga Khan University)
Recent Improvements in Provision
Pakistan has come a long way since it gained its independence in 1947, when there were only three psychiatric hospitals in the country. Today around 20 medical colleges support psychiatric wards. At the moment, there are some 4,100 beds in the public and private sector and about 342 practicing psychiatrists, mostly located in major cities (Table 1) [4]. Behavioural sciences and psychiatric training form an essential part of undergraduate medical training.
The National Mental Health Programme, developed in 1986, aims at achieving universal provision of mental health and substance-abuse services by incorporating them into primary health care. Via this programme, primary-care physicians are being trained, and training manuals are being developed for lady health visitors (a type of health worker who provides a variety of services to urban and rural communities, including basic nursing care, maternal and child health services, and training of community workers [5]). In addition, junior psychiatrists are being trained in community mental health. The importance of including spiritual healers in the mainstream health-care and referral system has also been recognised by the National Mental Health Programme, as they are frequently the ones having first contact with individuals who are mentally ill.
Pakistani Law and Mental Illness
The law has important implications for the lives of all citizens, including those who are mentally ill. The laws governing the treatment of mentally ill people give a clear indication of a country's attitude towards such people. The relationship between a society's attitude and the law is a dynamic one, and a two-way affair.
In Pakistan, until 2001, the major source of laws relating to individuals who are mentally ill was the Lunacy Act of 1912 [6] (Box 1), enacted by the colonial government, at the time, for the whole of British India. After the partition, Pakistani law continued to be based on the relics of its colonial past, although sporadic changes were brought about in the light of drastically changed conditions and the requirements of an Islamic republic. The Lunacy Act of 1912, however, like most other laws, remained in effect, despite occasional protests by the medical profession and society at large.
Box 1. The Lunacy Act of 1912
The Lunacy Act [6] was enacted in 1912 for British India. Until recently, it was the most important piece of psychiatric legislation in Pakistan. The statute is divided into four major parts dealing with definitions of crucial terms, rules pertaining to reception, care, and treatment of individuals who are mentally ill, and procedural rules for establishing whether or not an individual is mentally ill. Even a cursory glance at the statute reveals it as woefully inadequate and obsolete for the needs of a modern state. In 2001, the act was replaced by the Pakistan Mental Health Ordinance [7].
On February 20, 2001, the Pakistan Mental Health Ordinance came into effect [7]. The Lunacy Act of 1912 consequently stood repealed. The 2001 ordinance has brought about significant changes in the law “relating to mentally disordered persons with respect to their care and treatment and management of their property and other related matters” [7], as the preamble of the ordinance boldly proclaims.
The ubiquitous presence of ordinances is a peculiar feature of Pakistani law. Ordinances are presidential orders, tantamount to valid law, passed in emergency situations or in the absence of a sitting Parliament. Theoretically, they lapse after three months if opposed by an act of Parliament. In reality, ordinances form an important part of Pakistani law, and like the Pakistan Mental Health Ordinance, permanently hold as much legal value as any act passed by Parliament.
How Did the 2001 Ordinance Change the Law?
Before 2001, one of the most striking features of the Pakistani law regarding mentally disordered people was the sustained usage of archaic, imprecise, and often undefined terms. The term “lunacy” is a classic example because, despite being shunned by the psychiatric profession for being inhumane and imprecise, it still formed a part of the outdated Lunacy Act's title and was used, along with its derivatives, throughout the act. Lunatic was defined as “an idiot or person of unsound mind” (section 4), a definition that, on account of its vagueness, has also given rise to much case law [8,9].
The 2001 law has discarded such outmoded and imprecise terms as “lunatic”, “criminal lunatic” (an oxymoron, given that “lunatics” cannot be held responsible for their acts), and “asylum”, and has provided its own more comprehensive set of definitions. The ordinance uses the term “mental health” as a part of its title and defines the converse—mental disorder—as “mental illness, including mental impairment, severe personality disorder, severe mental impairment, and any other disorder or disabling of mind…” (section 2(1)(m) of [7]). For all of these categories of mental disorder, comprehensive definitions have also been provided. For example, severe personality disorder is described in the 2001 law as persistent disorder or disability of mind (whether or not it includes significant impairment of intelligence), which results in abnormally aggressive or seriously irresponsible conduct on the part of the person concerned.
The term “mentally disordered prisoner” replaces “criminal lunatic”; similarly, instead of “asylum”, “health facility” and “psychiatric facility” are used wherever suitable. A very pragmatic step forward has been the introduction of a definition of “informed consent” for treatment. In light of the new definition, consent would only be considered valid when it is informed, that is, when the patient (or guardian or nearest relative, in case of a minor) has been adequately informed of the purpose, nature, likely effects, and risks of the treatment, including the likelihood of its success, any alternative, and the costs to be incurred. By addressing this pertinent issue, an important gap in the law has been bridged.
Terminology deeply reflects the mood of those who use it. The adoption, by Pakistani society and law, of more scientifically appropriate, precise, and humane terms indicates that a more empathetic attitude has replaced an earlier attitude of summary dismissal and a lack of understanding towards individuals who are mentally disordered. This is, indeed, a positive development.
An important development brought about by the new law is the establishment of the Federal Mental Health Authority, comprising seven “eminent psychiatrists of at least 10 years standing” (section 3(3)(v) of [7]) and seven other members, largely bureaucrats. This body has been given the responsibility of overseeing the state of mental health provision in the country, setting up national standards of care and treatment, and performing a host of other tasks.
But instead of making immediate changes needed by the mental health system of the country, this authority makes only promises. Its objectives are long term and to an extent vaguely defined. The effect that this institution will have on the lives of individuals who are mentally disordered is hard to judge from what the law says about the institution; the effect depends, rather, on how the institution performs in the future.
Fewer Days of Forced Detention
Under the Lunacy Act of 1912, the combined effect of sections 13–16 was to allow the detention of people alleged to be “lunatics” for a period of ten days, extendable by the magistrate's permission to a maximum of 30 days, before an actual inquiry was held to establish the detainee's mental status. Given the rampant corruption and widespread abuse of power endemic in the Pakistani legal system, this provision was bound to be exploited. The recent trend in legal reform in Pakistan has been towards restricting police powers of arrest and detention. This trend reflects the realisation that the police have a reputation for abusing their powers.
In the new 2001 ordinance, section 19(2) clearly limits the period of forced detention under the above-mentioned circumstances to a maximum of 72 hours. During this time period, examination by a psychiatrist or the psychiatrist's nominated medical officer has to be ensured, and necessary arrangements must be made for starting care and treatment. It is hoped that this will prevent the widespread abuses of the law that often occurred before 2001.
A patient who is mentally ill on leave from a psychiatric facility may be ordered to be brought back by the magistrate in the area on the advice of the treating psychiatrist.
Criminal and Civil Liability
One of the most important legal issues that arise with regard to people who are mentally disordered is the issue of the extent of criminal and civil liability. In British law, for instance, individuals who are mentally disordered qualify for the “defence of insanity and automatism”, which leads to a legally stipulated reduction in their liability for criminal offences (see Book 4, chapter 2 in [10]). Similarly, in Pakistani civil law, individuals who are mentally disordered do not have the capacity to enter into valid contracts and, hence, cannot be held liable for breach of contract (section 3(3)(v) of [7]).
The new ordinance, just like its precursor, leaves the crucial question of the extent of criminal and civil liability of people who are mentally ill unanswered, which means that the law on this matter has to be derived from other sources of criminal and civil law. The new ordinance, therefore, has had no bearing whatsoever on one of the most crucial questions of law pertinent to individuals who are mentally disordered: how will it be determined that someone is or is not mentally disordered?
Human Rights
The latter half of the 20th century has seen the rise of the doctrine of human rights. Chapter VII of the new ordinance concerns the protection of human rights of persons who are mentally disordered, and in this manner, it is a great advance. Besides stressing the requirement of informed consent, it grants patients the right to confidentiality, stipulating, “No patient shall be publicised nor his identity disclosed to the public through press or media unless such person chooses to publicise his own condition.” Moreover, suicide, per se, is not to be considered a sign of mental illness, though persons attempting suicide must subsequently be assessed by an approved psychiatrist to ascertain their mental status.
A list of four distinct offences has been created by section 52 (chapter VIII) of the ordinance. These crimes are (1) wilfully making false statements so as to discredit someone as mentally disordered, (2) negligence of a manager of the estate of a person who is mentally disordered or such person's refusal to submit accounts, (3) ill treatment of a patient by the staff of a psychiatric facility, and (4) ill treatment or exploitation, including the traditional practice of induced microcephaly (Box 2), of any person who is mentally ill by members of the public at large. Punishments have been prescribed for these criminal offences, which include fines and imprisonment ranging from six months to five years.
Box 2. Induced Microcephaly: The Making of “Rat Children”
According to a legend, infertile women are blessed with children when they pray at the shrine of Shah Dola, a saint buried in Gujrat, Pakistan. But the first-born child in such situations, says the legend, is always microcephalic and must, therefore, be handed over to the custodians of the shrine. The microcephalic children (the “rat children” of Shah Dola) are severely handicapped both mentally and physically, but are considered close to God and, thus, are given charity. It is alleged that this phenomenon is actually the work of criminal gangs, who use iron rings to induce microcephaly in otherwise healthy infants to exploit them as beggars [13].
Doctors Are Not above the Law
The section of the new ordinance most relevant to the psychiatric profession, perhaps, is section 56 (chapter XI), which deals with specialised psychiatric treatments. It stipulates that all electroconvulsive treatments shall preferably be administered under general anaesthesia and advised by a psychiatrist, not a medical officer or anyone else, and that the reasons for not using other available methods must be recorded.
Also, contrary to popular clinical practice, the ordinance states that “administration of long acting anti-psychotic depot injections shall only be carried out upon the advice of a psychiatrist for a period specified in the prescription and such cases shall be reviewed periodically”. Such stringent measures can help prevent excesses being committed by the profession, though, at times, these restrictions might be found to be overly rigid. Finally, the most stringent controls have been placed on the practice of psychosurgery, obviating any possibility of it being performed, except when found to be absolutely crucial by a comprehensive, stipulated panel of doctors.
Together, the introduction of these measures shows the attention that has been devoted towards streamlining the provision of medical help for individuals who are mentally disordered and towards bridging the gaps in the law. The flip side may be that such detailed legislation might open the floodgates to those who wish to sue psychiatric professionals, though it is likely that such litigation will be rare. There is a lack of professional expertise regarding cases of personal injury caused by medical negligence. Another factor that makes litigation rare is the strong traditional belief in predestination—the belief that ill fate or death is fixed, and thus someone's negligence cannot effect it to a large extent.
The Road Ahead
The recent changes in the law do signify its dynamic nature. The law's response to changed social and professional attitudes, in the form of the new ordinance, though much belated, is a fitting one. But, as we have indicated in this article, many gaps in the law still remain. Also, steps such as the establishment of the Federal Mental Health Authority can only be judged by performance over the course of the coming years.
We firmly believe that the road ahead in the mental health laws of Pakistan also lies in seeking to comply with international standards. Despite the progress made in comparison with the old law, the current law still falls short of standards in relevant international conventions.
The Universal Declaration of Human Rights [11] and its extension with regard to individuals who are mentally disabled—the Declaration of the Rights of the Mentally Retarded [12]—can be useful guides to action. Briefly put, the latter declaration makes seven guarantees to individuals who are mentally disordered: (1) equal rights to the maximum degree of feasibility; (2) proper education, care, and treatment for self-development; (3) the right to economic security and a decent standard of living; (4) the right to live with one's own family or the closest possible alternative; (5) the right to a qualified guardian, if necessary; (6) protection from exploitation, abuse, and degrading treatment, and restricted civil and criminal liability; and (7) the right that any restriction of rights must be legally monitored, must not be arbitrary, and must be subject to appeal and periodic review.
Although these conventions cannot be incorporated into our domestic law, per se, they do provide standards that we must strive to meet. The recent ordinance has brought us one step closer to such compliance. But many problems still remain, so it has become ever more important that the law be subjected to periodic review by a team of experts who measure its performance with reference to the above-mentioned standards, and suggest the necessary reforms.
The authors are indebted to Unaiza Niaz, Umama Naeem, and Abdul Wahab Khan for their inspirational teaching of psychiatry. Thanks are also due to Ajmal Rizvi (Multimedia Designer, Aga Khan University) for his final help on the figures.
Citation: Gilani AI, Gilani UI, Kasi PM, Khan MM (2005) Psychiatric health laws in Pakistan: From lunacy to mental health. PLoS Med 2(11): e317.
==== Refs
References
Mirza I Jenkins R Risk factors, prevalence, and treatment of anxiety and depressive disorders in Pakistan: Systematic review BMJ 2004 328 794 15070634
Qidwai W Azam SI Psychiatric morbidity and perceptions on psychiatric illness among patients presenting to family physicians, in April 2001 at a teaching hospital in Karachi, Pakistan Asia Pacific Fam Med 2002 1 79 82
Naeem F Ayub M Izhar N Javed Z Irfan M Stigma and knowledge of depression: A survey comparing medical and non-medical students and staff in Lahore, Pakistan Pak J Med Sci 2005 21 155 158
Karim S Saeed K Rana MH Mubbashar MH Jenkins R Pakistan mental health country profile Int Rev Psychiatry 2004 16 83 92 15276941
Upvall MJ Sochael S Gonsalves A Behind the mud walls: The role and practice of lady health visitors in Pakistan Health Care Women Int 2002 23 432 441 12171694
Government of Pakistan Lunacy Act 1912 1912
Government of Pakistan Mental Health Ordinance, 2001 2001 Available: http://www.emro.who.int/MNH/WHD/Pakistan-Ordinance.pdf . Accessed 11 August 2005
Lahore High Court Case No. 001L219 (2001 CLC 97) 2001 Lahore (Pakistan) Lahore High Court
Lahore High Court Case No. 984L2533 (1984 MLD 438) 1984 Lahore (Pakistan) Lahore High Court
Blackstone W Commentaries on the laws of England 1769 Chicago University of Chicago Press 496
United Nations General Assembly Universal declaration of human rights 1948 New York United Nations General Assembly Official Records General Assembly Resolution 217A (III), UN Doc A/810
Office of the United Nations High Commissioner for Human Rights Declaration on the rights of mentally retarded persons 1971 General Assembly Resolution 2856 (XXVI). Available: http://www.ohchr.org/english/law/res2856.htm . Accessed 11 August 2005
Galpin R The rat children of Pakistan 1998 June 29 London BBC News Available: http://news.bbc.co.uk/1/hi/world/south_asia/122670.stm . Accessed 11 August 2005
|
16159307
|
PMC1215469
|
CC BY
|
2021-01-05 10:39:17
|
no
|
PLoS Med. 2005 Nov 20; 2(11):e317
|
utf-8
|
PLoS Med
| 2,005 |
10.1371/journal.pmed.0020317
|
oa_comm
|
==== Front
AIDS Res TherAIDS Research and Therapy1742-6405BioMed Central London 1742-6405-2-71615689110.1186/1742-6405-2-7MethodologyAssessment of HIV-1 entry inhibitors by MLV/HIV-1 pseudotyped vectors Siegert Sandra [email protected] Sonja [email protected] Ralf [email protected] Barbara S [email protected] Georg-Speyer-Haus, Institute for Biomedical Research, Paul-Ehrlich-Strasse 42-44, D-60596 Frankfurt am Main, Germany2 Paul-Ehrlich Institute, Abt. 2/01, Paul-Ehrlich Strasse 51-59, D-63225 Langen, Germany3 Department of Medicine III, Johannes Gutenberg University, 55101 Mainz, Germany4 Institute of Medical Microbiology and Hygiene, University of Regensburg, Franz-Josef-Strauss Allee 11, 93053 Regensburg, Germany5 Friedrich Miescher Institute, Maulbeerstrasse 66, CH-4058 Basel, Switzerland2005 12 9 2005 2 7 7 25 7 2005 12 9 2005 Copyright © 2005 Siegert et al; licensee BioMed Central Ltd.2005Siegert et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Murine leukemia virus (MLV) vector particles can be pseudotyped with a truncated variant of the human immunodeficiency virus type 1 (HIV-1) envelope protein (Env) and selectively target gene transfer to human cells expressing both CD4 and an appropriate co-receptor. Vector transduction mimics the HIV-1 entry process and is therefore a safe tool to study HIV-1 entry.
Results
Using FLY cells, which express the MLV gag and pol genes, we generated stable producer cell lines that express the HIV-1 envelope gene and a retroviral vector genome encoding the green fluorescent protein (GFP). The BH10 or 89.6 P HIV-1 Env was expressed from a bicistronic vector which allowed the rapid selection of stable cell lines. A codon-usage-optimized synthetic env gene permitted high, Rev-independent Env expression. Vectors generated by these producer cells displayed different sensitivity to entry inhibitors.
Conclusion
These data illustrate that MLV/HIV-1 vectors are a valuable screening system for entry inhibitors or neutralizing antisera generated by vaccines.
==== Body
Background
The acquired immunodeficiency syndrome (AIDS) was first described about 20 years ago. Since then almost 20 million people have died from human immunodeficiency virus (HIV-1) infection and 42 million are infected with. New drugs and an effective vaccine are urgently needed. In particular, new drugs that block the HIV type 1 (HIV-1) entry into host cell have clear advantages over the currently used drugs. They should abrogate the establishment of a productive infection and consequently could diminish the chances of HIV-1 developing resistance. Furthermore, a vaccine that prevents AIDS should elicit broadly cross-reactive neutralizing antibodies to prevent infection. A safe and simple assay for measuring neutralizing activities against different HIV-1 strains is critical for the development of such a vaccine or entry inhibiting drugs.
We previously generated a retroviral vector which specifically transfers genes into human CD4+ cells [1,2]. This vector was derived by pseudotyping murine leukemia virus (MLV) capsid particles with a variant of the HIV-1 envelope protein (Env) containing the surface glycoprotein gp120-SU and a carboxyl-terminally truncated transmembrane (TM) protein with only 7 cytoplasmic amino acids. HIV-1 Env facilitates vector attachment to target cells and membrane fusion, which is initiated by the interaction of HIV-1 Env with the CD4 receptor molecule on the surface of the target cell. CD4 binding induces a conformational change in the envelope glycoprotein and allows the binding of a co-receptor of the chemokine receptor family [3]. The co-receptor usage is virus strain dependent: R5 viruses, which infect monocytes and macrophages, use CCR5 and X4 viruses, which infect T cell lines, use CXCR4. X4R5 strains can use CXCR4 as well as CCR5 for entry. The transfer of a marker gene by MLV/HIV-1 vectors is therefore a safe and simple method to assay entry mediated by HIV-1 Env and can be used to evaluate HIV-1 entry inhibitors, such as small molecules or neutralizing antibodies in sera of vaccinated animals or patients.
Here, we optimized production of the MLV/HIV-1 vector to allow analysis of different HIV-1 Envs and demonstrate that the responsiveness to viral entry inhibitors was dependent on the HIV-1 strain the Env was derived from. This illustrates that MLV/HIV-1 pseudotyped vectors are useful tools for analyzing HIV-1 entry.
Results and discussion
Generation of a stable producer cell line encoding the 89.6 P HIV-1 Env
We and others have previously reported that MLV capsids can be pseudotyped with cytoplasmatically truncated variants of the HIV-1 or HIV-2 envelope glycoproteins possessing only 7 cytoplasmic amino acids. These MLV/HIV pseudotyped vectors have the HIV host range [4,1,5]. We used a X4 HIV-1 Env variant (BH10) for pseudotyping, which restricted vector entry to CD4 and CXCR4 receptor-positive cells [2]. HIV-1 Env was expressed from an expression construct that also encoded Rev, which is required to transport the rev responsive element (RRE)-containing env mRNA from the nucleus to the cytoplasm. In the present study we evaluated a codon-usage-optimized HIV-1 env gene that encodes the truncated Env variant of the X4R5 89.6 P HIV-1 isolate and lacks rev sequences. Western blot analysis of transfected 293T cells showed that the change from lentiviral to mammalian codon usage allowed high HIV-1 Env protein expression in the absence of Rev (Figure 1). The 89.6 P Env showed different migration in polyacrylamide gels from the BH10 isolate, which might be caused by different glycosylation patterns of the protein and reflects strain-specific differences in Env.
Figure 1 Expression of the 89.6 P HIV-1 Env. 293T cells were transfected with 3 μg plasmid DNA and cell lysates were analyzed after two days for HIV-1 Env expression by Western blot analysis. The two forms of Env are indicated as gp140 (C-terminally truncated precursor) and gp120 (SU).
We previously constructed a MLV/HIV-1 producer cell line based on FLY cells [6], which expresses the HIV-1 Env of the X4 HIV-1 BH10 strain and a retroviral vector encoding the green fluorescent protein (GFP) [2]. These cells are further referred to as FLY-HIV-87-GFP cells. HIV-1 Env protein expression is driven by the strong human elongation factor 1α promoter and stable clones were selected via the puromycin resistance gene (pac) encoded on the bicistronic messenger RNA.
We cloned a codon-usage-optimized 89.6 P HIV-1 Env into this vector and stable cell clones were rapidly isolated by puromycin selection. Protein expression of HIV-1 Env was ensured by expansion of the cells in the continued presence of puromycin. A single clone was further transduced with a GFP-encoding retroviral vector to obtain the producer cell line FLY-syn-GFP.
Characterization of vectors particles derived from FLY-HIV-87-GFP or FLY-syn-GFP
The characterization of the two producer cell lines was started by determining the amount of infectious retroviral vector particles released from the producer cells. Infectious particles can be easily detected by the transfer of the gfp gene. NIH3T3 CD4/CXCR4 cells that express the CD4 and CXCR4 receptors were transduced with supernatants derived from FLY-HIV-87-GFP or FLY-syn-GFP cells and analyzed after two days by flow cytometry. Figure 2A gives a typical FACS analysis of NIH3T3-CD4/X4 cells transduced with serially diluted vector supernatants. Titers are given in Figure 2B in infectious units per ml and represent the average values of five experiments. The titers produced by FLY-HIV-87-GFP cells were reproducibly higher than those obtained from FLY-syn-GFP cells. However, the X4R5 phenotype of the 89.6 P HIV-1 Env was retained by the pseudotypes. Only vector particles derived from FLY-syn-GFP cells were able to transduce NIH3T3 cells expressing CD4 and CCR5 (Figure 3).
Figure 2 Titer of MLV/HIV-1 pseudotyped vector particles released from producer cells. A: NIH3T3-CD4/X4 cells (1 × 105) were incubated with 1, 0.1 or 0.01 ml supernatant from producer cells in a total volume of 1 ml. For titer determination, the number of GFP+ cells (upper left) was determined. B: Titers were calculated by measuring the percentage of GFP-positive cells after transduction. Values are the average of 5 experiments.
Figure 3 Co-receptor usage of MLV/HIV-1 vectors. NIH3T3 cells expressing the CD4 receptor and either CXCR4 or CCR5 were transduced with vector particles derived from FLY cell lines, and the titers were calculated by measuring the percentage of GFP-positive cells. The X4R5 89.6 P Env in vector particles derived from FLY-syn-GFP cells allowed the transduction of both target cell lines.
The changed codon usage of the 89.6 P Env resulted in high expression in FLY cells; however, vector titers were always lower than those of vectors containing BH10 Env. To analyze the amount of Env incorporated into particles, the supernatants of the producer cells were collected and centrifuged through a 30% sucrose cushion. The Western blot analysis of the pellets revealed that high levels of the 89.6 P Env was incorporated into vector particles (Figure 4). Equal loading was verified by detection of the MLV p30 Gag protein. These data imply that the amount of Env in viral vector particles does not correlate with their titer. Instead, it seems likely that the fusion ability of Env, which displays strain-specific differences, affects vector infectivity.
Figure 4 Incorporation of HIV-1 Env into vector particles. Incorporation of Env into MLV/HIV-1 particles was analyzed by Western blot analysis of particles concentrated from FLY cell supernatants by ultracentrifugation. Equal loading was confirmed after stripping the blot and incubating with an anti-MLV-Gag (p30) antibody.
Evaluation of HIV-1 entry inhibitors
Attachment and entry of HIV-1 into CD4 cells involve a series of conformational changes in Env which allow co-receptor binding and finally fusion of viral and cell membranes. AMD-3100 is a small molecule inhibitor of gp120 attachment to the CXCR4 receptor, and T-20 is a synthetic peptide corresponding to a helical region of HIV-1 gp41 that blocks fusion of the cellular and the viral membrane. While AMD-3100 is only active against X4 and X4R5 HIV-1 strains, T20 inhibits fusion of most HIV-1 strains.
As shown in Figure 5A, the sensitivity of the BH10 and 89.6 P Env-containing vectors to AMD-3100 was slightly different. BH10 was less sensitive, indicating a higher affinity for CXCR4. However, the responsiveness to T20 was not significantly different between both Env proteins (Figure 5B). The inhibitor concentrations used in these studies were higher than those previously published because C-terminally truncated Envs have a fast fusion kinetic and are thus less sensitive to entry inhibitors [7]. It has been shown that the cytoplasmic tail slows the folding of HIV-1 Env from a late prebundle configuration into the six-helix bundle, and thereby also slows down the fusion process [8]. Inhibition of HIV-1 Env mediated transduction was specific, since amphotropic MLV could not be inhibited by AMD3100 or T20 (data not shown).
Figure 5 Inhibition of MLV/HIV-1 mediated gene transfer by HIV-1 entry inhibitors. Indicated amounts of either (A) AMD-3100 or (B) T20 were added to vector supernatants during the transduction of NIH3T3 cells expressing CD4 and CXCR4. Transfer of the gfp gene was measured and titers are given as percentage of those obtained from untreated supernatants. Inhibition of MLV/HIV-1 vectors was done 2–3 times and had the same outcome. Inhibition of transduction of vectors derived from FLY-HIV-87-GFP cells by AMD3100 was done only once.
We present here two producer cell lines that release MLV/HIV-1 pseudotyped retroviral vectors particles. Transfer of the gfp gene can be used as an indication of an infection process mediated by the HIV-1 envelope glycoprotein. This assay is robust and simple to perform and GFP expression can be rapidly monitored by flow cytometry without further staining of the target cells. Expression of the HIV-1 Env from a bicistronic vector allowed fast establishment of stable producer cell lines. Optimization of HIV-1 Env codon usage led to high expression without the need for Rev co-expression and will, in combination with the bicistronic vector, facilitate the easy exchange of Env sequences. The system can also be applied to transient vector production, and synthetic genes will permit fast testing of diverse HIV-1 Envs, including those from drug resistant strains.
Conclusion
MLV/HIV-1 vectors are a valuable screening system for entry inhibitors or neutralizing antisera generated by vaccines.
Methods
Plasmids
The truncated variant of the envelope glycoprotein HIV-1 Env Tr712 [1] was derived from the plasmid pLßAc/env-Tr712-neo as a 3.1 kb SalI/XhoI fragment and was cloned into the XhoI site of the bicistronic vector pEF-IRES-P [9]. The sequence of the 89.6 P HIV-1 Env isolate (aa 1 – 712) was chemically synthesized and the codons were modified to high GC content without changing the coding sequence. The Env signal peptide sequence was exchanged with that of the CD5 receptor. Env was excised as an EcoRI/XhoI fragment, blunt-ended and cloned into the XhoI site of the vector pEF-IRES-P, resulting in the clone pEF-IRES-P-89.6 P.
Cells
NIH 3T3 derivatives [10], 293T (ATCC #CRL-11268) and FLY [11] cells were grown in Dulbecco's modified Eagle's medium (GIBCO BRL, Eggenstein) supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin and 1% L-glutamine. FLY cells are based on human HT1080 cells and express the MLV Gag/Pol gene product [11]. Stably transfected FLY-HIV-87-GFP or FLY-syn-GFP cells were grown in the medium described above supplemented with 2.5 μg/ml puromycin (Sigma, Deisenhofen).
Stable transfection of cells
FLY cells (106) were seeded in a 10 cm tissue culture plate. The following day, cells were transfected with 10 μg of the bicistronic construct pEF-IRES-P-89.6 P. Transfection was performed with SuperFect™ Transfection Reagent (Qiagen, Hilden). Forty-eight hours after transfection, 2.5 μg/ml puromycin was added to the medium. After the selection and isolation of single cell clones, cells were tested for HIV-1 envelope glycoprotein expression by Western blot analysis and the best expressing clone was transduced with VSV-G pseudotyped retroviral vectors encoding GFP (pMX-EGFP) [12].
Retroviral transduction and titer determination
Serial dilutions of vector supernatants from packaging cells were passed through 0.45-μm filters and incubated with 1 × 105 NIH 3T3-CD4/CXCR4 or NIH 3T3-CD4/CCR5 cells for 6 h and longer incubation (e.g. over night) did not change the titer. Supernatants were mixed with different amounts of T20 (generous gift of Prof. von Laer, Georg-Speyer-Haus, Frankfurt) or AMD-3100 (NIH AIDS Research and Reference Reagent Program) for inhibition assays. The numbers of GFP-expressing cells were detected by FACS analysis 48 – 72 hours after transduction. The titers are given in infectious units per ml (IU/ml) and were determined by calculating the percentage of GFP-positive cells. GFP expression was monitored by a shift to green fluorescence (FL-1). FACS analysis was performed with a FACScan (Becton Dickinson, Heidelberg) using the Cellquest software.
Preparation of viral proteins and immunoblots
For the analysis of incorporation of Env proteins into the vector particles, supernatant of producer cells was filtered through a 0.45-μm filter (Greiner, Frickenhausen, Germany) and centrifuged through a 30% sucrose cushion for 90 min at 26,000 rpm in an SW28 rotor. The pellet was resuspended in 50 μl SDS loading buffer [13] and 25 μl were analyzed. Cell lysates were prepared as described before [1] and 40 μg protein were analyzed. Proteins were separated by electrophoresis through a 7.5% polyacrylamide gel and electroblotted onto a nitrocellulose transfer membrane (Schleicher & Schuell, Dassel, Germany). Immunostaining was performed in Tris base saline (pH 7.4) with 5% milk powder and 0.1% Tween 20. Proteins were detected by incubation with goat antiserum against gp120 (Dunn Labtech, Ansbach, Germany) or a polyclonal antiserum directed against MLV p30 [1], followed by an incubation with a horseradish peroxidase-labeled anti-goat antibody or protein A. Protein bands were finally visualized by enhanced chemiluminescence detection using the ECL kit (Amersham, Braunschweig, Germany) according to the manufacturer's instructions.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Sandra Siegert and Sonja Thaler performed the experiments, Ralf Wagner and Barbara Schnierle participated in the design of experiments, and Barbara Schnierle oversaw the conduction of the experiments and the interpretation of the results.
==== Refs
Schnierle BS Stitz J Bosch V Nocken F Merget-Millitzer H Engelstadter M Kurth R Groner B Cichutek K Pseudotyping of murine leukemia virus with the envelope glycoproteins of HIV generates a retroviral vector with specificity of infection for CD4-expressing cells Proc Natl Acad Sci U S A 1997 94 8640 5 9238030 10.1073/pnas.94.16.8640
Thaler S Schnierle BS A packaging cell line generating CD4-specific retroviral vectors for efficient gene transfer into primary human T-helper lymphocytes Mol Ther 2001 4 273 9 11545619 10.1006/mthe.2001.0445
Chen B Vogan EM Gong H Skehel JJ Wiley DC Harrison SC Structure of an unliganded simian immunodeficiency virus gp120 core Nature 2005 433 834 841 15729334 10.1038/nature03327
Mammano F Salvatori F Indraccolo S De Rossi A Chieco-Bianchi L Gottlinger HG Truncation of the human immunodeficiency virus type 1 envelope glycoprotein allows efficient pseudotyping of Moloney murine leukemia virus particles and gene transfer into CD4+ cells J Virol 1997 71 3341 5 9060707
Hohne M Thaler S Dudda JC Groner B Schnierle BS Truncation of the human immunodeficiency virus-type-2 envelope glycoprotein allows efficient pseudotyping of murine leukemia virus retroviral vector particles Virology 1999 261 70 8 10441556 10.1006/viro.1999.9847
Cosset FL Morling FJ Takeuchi Y Weiss RA Collins MK Russell SJ Retroviral retargeting by envelopes expressing an N-terminal binding domain J Virol 1995 69 6314 22 7666532
Wilk T Pfeiffer T Bosch V Retained in vitro infectivity and cytopathogenicity of HIV-1 despite truncation of the C-terminal tail of the env gene product Virology 1992 189 167 77 1604808 10.1016/0042-6822(92)90692-I
Abrahamyan LG Mkrtchyan SR Binley J Lu M Melikyan GB Cohen FS The cytoplasmic tail slows the folding of human immunodeficiency virus type 1 Env from a late prebundle configuration into the six-helix bundle J Virol 2005 79 106 115 15596806 10.1128/JVI.79.1.106-115.2005
Hobbs S Jitrapakdee S Wallace JC Development of a bicistronic vector driven by the human polypeptide chain elongation factor 1alpha promoter for creation of stable mammalian cell lines that express very high levels of recombinant proteins Biochem Biophys Res Commun 1998 252 368 72 9826536 10.1006/bbrc.1998.9646
Deng HK Unutmaz D KewalRamani VN Littman DR Expression cloning of new receptors used by simian and human immunodeficiency viruses Nature 1997 388 296 300 9230441 10.1038/40894
Cosset FL Takeuchi Y Battini JL Weiss RA Collins MK High-titer packaging cells producing recombinant retroviruses resistant to human serum J Virol 1995 69 7430 6 7494248
Onishi M Nosaka T Misawa K Mui AL Gorman D McMahon M Miyajima A Kitamura T Identification and characterization of a constitutively active STAT5 mutant that promotes cell proliferation Mol Cell Biol 1998 18 3871 9 9632771
Laemli UK Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970 227 680 685 5432063 10.1038/227680a0
|
16156891
|
PMC1215470
|
CC BY
|
2021-01-04 16:38:33
|
no
|
AIDS Res Ther. 2005 Sep 12; 2:7
|
utf-8
|
AIDS Res Ther
| 2,005 |
10.1186/1742-6405-2-7
|
oa_comm
|
==== Front
Aust New Zealand Health PolicyAustralia and New Zealand Health Policy1743-8462BioMed Central London 1743-8462-2-181611149610.1186/1743-8462-2-18ResearchStrengthening Medicare: Will increasing the bulk-billing rate and supply of general practitioners increase access to Medicare-funded general practitioner services and does rurality matter? Day Susan E [email protected] Katrina [email protected] David [email protected] Stuart [email protected] Lyle [email protected] Don [email protected] Centre for Health and Society, School of Population Health, University of Melbourne, Australia2 Program Evaluation Unit, School of Population Health, University of Melbourne, Australia3 Cancer Control Research Program, British Columbia Cancer Research Centre, Canada4 Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Population Health, University of Melbourne, Australia5 Department of Public Health Sciences, University of Alberta, Canada2005 20 8 2005 2 18 18 21 2 2005 20 8 2005 Copyright © 2005 Day et al; licensee BioMed Central Ltd.2005Day et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent increases in the bulk-billing rate have been taken as an indication that the Federal government's Strengthening Medicare initiative, and particularly the bulk-billing incentives, are 'working'. Given the enduring geographic differences in the supply of general practitioners (GPs) it is timely to reconsider the impact that this increase in the provision of 'free care' will have on access to Medicare-funded GP services in rural and urban areas of Australia. Utilisation has been modelled as two different stochastic processes: the decision to consult and the frequency of consultation.
Results
In the decision to consult model the supply of FFS GPs is a more important predictor of utilisation than the bulk-billing rate. Paradoxically the modelling predicts that ceteris paribus increases in either GP supply or the bulk-billing rate appear to have perverse effects in some areas by decreasing utilisation. In the frequency of consultation model, GP density is not a predictor and increasing the bulk-billing rate will unambiguously increase the frequency of consultation across all areas. In both models, the positive impacts associated with changes in supply and cost are constrained outside the inner metropolitan area by reduced geographic accessibility to Medicare-funded GP services. The modelling also shows that people are more likely to consult a GP in areas of high socioeconomic disadvantage, although socioeconomic status is not a predictor of frequency of consultation.
Conclusion
Bulk-billing rates and the supply of FFS GPs are important features of the Australian health care system that are, potentially, amenable to policy manipulation. The implications of this research are that government policies designed to achieve similarity in these characteristics across geographic areas will not result in equity of access because they fail to address problems caused by geographic inaccessibility in rural and remote areas. Attempting to increase bulk-billing rates in some of these areas may, in fact, reduce access to FFS GP services.
==== Body
Background
In April 2003 the Federal government announced a policy initiative called A Fairer Medicare. This initiative was heavily criticised and enabling legislation failed to pass the Senate. A revised version of the policy, called Medicare Plus, was announced in November 2003. Further revisions ensued before Medicare Plus was passed by the Federal parliament in March 2004 and its name was subsequently changed to Strengthening Medicare [1]. Strengthening Medicare aimed to improve access (including affordability) to out-of-hospital medical services and contained twenty-seven separate measures including incentive payments for general practitioners (GPs) to encourage bulk-billing of concession card holders and children under sixteen years (particularly in regional, rural and remote areas) and a series of measures to attract GPs to work in areas of undersupply such as outer metropolitan, regional and remote areas [1,2].
Although Strengthening Medicare and its precursors were not specifically aimed at increasing the bulk-billing rate, there was considerable debate about the impact that the initiatives would have on the bulk-billing rate and access to GP services (see for example, the first and second reports of the Senate Select Committee on Medicare) [3,4]. Also, one of the effectiveness indicators developed by the Department of Health and Ageing to measure progress towards Departmental outcomes in relation to access to Medicare is 'the percentage of Medicare services that are direct billed with no gap charged' [2]. Recent increases in the overall bulk-billing rate for GP services have been seen as an indication that the Strengthening Medicare package, and bulk-billing incentives in particular, are 'working' [1,5]. Given the recent increases in the level of bulk-billing (i.e. 'free care') provided by fee-for-service (FFS) GPs and enduring geographic differences in the supply of GPs, it is pertinent to question what impact increases in the bulk-billing rate will have on access to Medicare-funded GP services in different parts of Australia.
The most widely accepted study examining the relationship between price and utilisation of doctors' services is the Rand Health Insurance Experiment (HIE) in the United States. HIE results show that utilisation of outpatient services decreases when patients are required to pay a co-payment and that there are no differences in the nature of the response across geographic areas [6]. Richardson applied the HIE results to Australian Medicare data and concludes that 78% of the change in GP service utilisation in the 1976–1986 period and 94% of the change in the 1984/85–1989/90 period were not related to decreases in consumer co-payments – other factors were at work [7].
A review of Australian cross sectional studies examining factors associated with GP consultation rates indicates that higher frequencies are associated with lower cost, [8-10] patients having a health care card, [11] patients being in poorer health, [9,11] and with areas of higher proportions older age residents [10]. Lower frequencies are to be found in areas with higher levels of education [8] and among female patients with higher levels of education, [9] in less geographically accessible (eg rural) areas [10,11] and among female patients with an internal Health Locus of Control [9]. One study with highly aggregated levels of data showed an increase in the frequency of consultation in low socioeconomic status areas [12]. Another indicated that the relationship was not so straightforward. Low socioeconomic status was associated with an increase in the consultation rate in highly accessible areas and low consultation rates in inaccessible areas, [13] suggesting supply side factors are also important.
Cross sectional results for the impact of GP supply generally indicate an increase in supply being associated with an increase in frequency of consultation [10,14,15]. One Australian study by Doessel [16] indicated there was no association between GP supply and population based frequency of utilisation rates. However, Richardson and Peacock have indicated that these results should be interpreted carefully [15].
Observed associations between increases in doctor supply and increases in frequency of utilisation have given rise to the notion of supplier induced demand (SID). SID suggests that doctors are imperfect agents and can induce demand for health care, which directly conflicts with the full information and consumer sovereignty assumptions of the orthodox model of demand and supply [17]. Volumes of empirical evidence addressing the possibility of SID have been presented from a range of different health systems, and have been reviewed elsewhere [18-20]. Many of the studies reviewed have used cross-sectional data sets to examine the effect of doctor supply on the frequency of utilization of health care services [14,21-24].
However, studies examining only the frequency of utilisation do not take into account the nature of the decision making process in the demand for non-emergency care which is thought to reflect sequential decisions involving both the individual and the GP [25]. Initially an individual decides whether to seek out health care or advice. This decision to consult is derived from the demand for health [26] and reflects an individual's beliefs about the severity of their condition, the availability of health care and their expectation of any potential benefits. Once the contact decision has been made and acted upon, the individual and the doctor decide on the type and amount of health services to be provided and the frequency of future consultations. Although the decision about the type, amount and frequency of future consultations is made primarily by the GP [6,27,28] the patient also has a role in this decision. The extent to which the patient participates in the decision depends on the nature of the doctor-patient agency relationship [29].
A small number of studies have attempted to examine the effect of the supply of doctors upon the patient-initiated and doctor-initiated components of utilisation. The results of these studies are inconclusive. Rossiter and Wilensky found that increases in doctor supply had no effect on patient-initiated visits but did have a small effect on doctor-initiated visits and took this to be evidence of SID [30,31]. On the other hand, Escarce found that increased availability increased initial contacts but had little effect on the intensity of subsequent visits, the doctor-initiated component [32]. There is only one Australian study examining the patient-initiated component of the utilisation process. This study found a higher proportion of the resident population consulting a GP in areas with a greater supply of GPs, areas which have higher proportions of younger and older age residents, and areas of lower cost (higher bulk-billing rates). Lower proportions were associated with rural areas [10].
One of the shortcomings of much of the Australian work on the relationship between cost, supply and utilisation of GP services is the inability of the studies to account for border-crossing by patients. There is ample evidence that patients do not necessarily visit the nearest GP when they decide to see a doctor [33,34]. The studies also indicate significant interactions between the variables associated with utilisation of GP services yet these interactions do not appear to have been systematically explored. And the studies are not always inclusive: some target specific demographic groups, others omit certain age groups or geographic areas from their analyses. Finally, despite their widespread acceptance, [35] the generalisability of the HIE results has been questioned [36,37] and their relevance to the Australian situation is not clear.
This research uses publicly accessible data to investigate the likely impact of an increase in the bulk-billing rates and GP supply on access to Medicare-funded GP services in rural and urban areas. The methodology is based on an analysis of aggregated data from various sources and has been developed to be inclusive of all geographic areas and age groups, to take into account border crossing by patients, to systematically explore the interactions among the variables and to accommodate the two-part decision-making process thought to underlie the utilisation of GP services.
Methods
Unit of analysis
Divisions of General Practice consist of 'geographically co-located group(s) of general practitioners who have formed an organization to work together ... to improve health outcomes at the local level' and provide GPs with a 'corporate identity' as well as a 'method of influencing the organization of health care delivery' [38]. Divisions can, therefore, be regarded as geographically defined service delivery systems and are appropriate aggregation units for modelling the relationship between price, supply and utilisation of health care services. One hundred and twenty-one Divisions are included in the analysis.
Data
Two utilisation models have been developed. The first reflects the decision to consult an FFS GP and the second reflects the frequency of consultation once the decision to consult has been made. These contact and frequency decisions are modelled as two different stochastic processes [28]. Failure to do this can lead to inconsistent parameter estimates and misinterpretation of results [25].
Decision to Consult Model
A patient to population ratio (decision to consult index) was considered an appropriate outcome variable for the contact model and, to overcome the problems of border crossing, the index has been defined as the number of Whole Patient Equivalents (WPEs) per head of resident population. (See the notes to Table 1 for the definition of WPE.)
Table 1 Variables and sources of data for the regression models
Variables Data Sources
Dependent: decision to consult index
Whole Patient Equivalents (WPEs) per head of population* WPEs: Table S2 in the Statistical Appendix to the report The General Practice Workforce in Australia†
Population: HealthWIZ v6.2‡ based on 1996 census
Dependent: frequency of consultation
Group A1 and A2 consults per Standardized Whole Patient Equivalent (SWPE)¶ Group A1 and A2 Consultations: available at accessed Sept 7 2003
SPWEs: Table S2 in the Statistical Appendix to the report The General Practice Workforce in Australia†
Predictor variables
Geographic accessibility: Population weighted ARIA values for each Division % Population in postcodes for each Division: available at accessed Jul 11 2003
Population in each postcode: HealthWIZ v6.2‡ based on 1996 census
ARIA values for postal areas originally accessed at accessed 26 Feb 2002. This url is no longer available but information about the data can be found at .
Bulk billing rate for general practice consultations HealthWIZ v6.2‡
Dr Density: number of GPs and Other primary medical care doctors per 1,000 head of population Vocationally Registered GPs and Other primary care practitioners: Table S2 in the Statistical Appendix to the report The General Practice Workforce in Australia†, derived from the 1998 Medical labour force survey undertaken by the Australian Institute of Health and Welfare (AIHW), Canberra
Population: HealthWIZ v6.2‡ based on 1996 census
Index of Disadvantage Population at each level of disadvantage within a Division: HealthWIZ v6.2‡
% population born in NESB country; % female, av age HealthWIZ v6.2‡ based on 1996 census
Notes: * WPEs are derived by the GP Branch of the Commonwealth Department of Health and Ageing as an indicator of patient load for each practice. If a patient consulted at only one general practice during a financial year, the patient is counted as one WPE for the practice. If a patients visits more than one general practice, the patient is counted as a fraction of a WPE based on the schedule fee value for each general practice consulted.
† Australian Medical Workforce Advisory Committee: The General Practice Workforce in Australia. AMWAC Report 2000.2.
‡ Commonwealth Department of Health and Aged Care, National Social Health Statistical Database, developed by Prometheus Information Pty Ltd, commonly known as HealthWIZ. See Other Products at
¶ SWPEs are standardised WPEs. 'The standardisation process is based on National Medical and Department of Veterans' Affairs claims figures for each of 16 age/gender categories. The standardisation is achieved by allocating each patient to one of sixteen categories and multiplying the WPE value for each patient by the appropriate weight.
Frequency of Consultation Model
Group A1 (General Practitioner) and Group A2 (Other un-referred) professional attendances are the basic building blocks of the Medicare system. In the 1998–99 financial year these services accounted for 95% of Medicare all services and 92% of the Medicare dollar benefits paid (excluding oral maxillofacial services) [39]. Frequency of consultation has been defined as the number of Group A1 and Group A2 services per Standardized Whole Patient Equivalent (SWPE). (See the notes to Table 1 for the definition of SWPE.)
Based on the review of the literature, the aim of the research and an exploration of the available data, the predictor variables initially entered into both models were: a Divisional measure of geographic accessibility, the bulk-billing rate, GP density, level of socioeconomic disadvantage of the resident population, and the proportion of the resident population born in a non-English speaking background (NESB) country. The proportion of the resident population that is female, and the average age of the resident population were also entered into the decision to consult model but not in the frequency of consultation model as age and gender are controlled for in the outcome variable. Sociodemographic information is based on the 1996 census and all other data relates to the 1998–99 financial year. The relevant data sources for the variables are shown in Table 1.
Assuming that Australian Bureau of Statistics postal areas are an adequate approximation of postcode areas, a population weighted geographic accessibility value was calculated for each Division by multiplying the Accessibility/Remoteness Index of Australia (ARIA) [40] for each postal area by the proportion of the Divisional population living in the matching postcode and then summing the results across all Divisional postcodes. Values range from zero to twelve and low values indicate greater geographic accessibility. The Divisional index of disadvantage has been derived from the 1996 Socio-Economic Indexes for Areas (SEIFA) Index of Disadvantage [41] statistics for each Division. Values of this index range from one to eleven and the original variable has been recoded so that higher values indicate higher levels of socioeconomic disadvantage. GP density is the number of vocationally registered plus other GPs per 1,000 head of resident population. Summary statistics for the variables used in the modelling are shown in Table 2.
Table 2 Summary of the dependent and predictor variables
Highly Accessible* n = 64 Accessible* n = 37 Moderately Accessible* n = 7 Remote* n = 6 Very Remote* n = 5
Dependent Variables
WPEs/head of popn Mean 0.95 0.83 0.76 0.70 0.54
SD 0.09 0.06 0.10 0.05 0.12
Consults/SWPE Mean 6.52 4.98 4.8 5.08 4.61
SD 0.91 0.53 0.42 0.44 0.62
Predictor Variables
ARIA value Mean 0.22 2.39 4.77 6.87 10.12
SD 0.28 0.80 0.65 0.56 0.82
Bulk-billing rate Mean 82.8% 55.7% 52.1% 65.5% 71.4%
SD 11.2% 13.4% 15.2% 7.0% 20.6%
GP density Mean 1.21 0.97 0.87 0.77 0.82
SD 0.33 0.16 0.27 0.17 0.16
Index of disadvantage Mean 6.80 5.00 4.50 3.61 3.93
SD 2.31 1.38 0.84 0.22 0.52
% born NESB country Mean 16.0% 3.9% 4.9% 4.8% 3.9%
SD 10.4% 1.6% 2.6% 1.5% 1.9%
% female Mean 50.4% 50.3% 49.5% 48.5% 47.6%
SD 1.3% 1.0% 1.8% 2.7% 3.1%
Average age (yrs) Mean 35.22 35.62 33.86 34.09 32.08
SD 2.45 1.76 2.20 2.32 2.15
Data Sources: See Table 1
Notes: * Categories based on the following ARIA values: highly accessible = 0 – 1.84, accessible = >1.84 – 3.51, moderately accessible = >3.51 – 5.80, remote = >5,80 – 9.08, very remote = <9.02 – 12. Source: Commonwealth Department of Health and Ageing, Accessibility/Remoteness Index of Australia (ARIA), available at Accessed 8th January 2000.
Analysis
Stepwise ordinary least squares (OLS) regression analyses have been undertaken for both models. A forward selection algorithm with an exclusion threshold of p > 0.2 has been used to enter each of the predictor variables. The main predictor variables which were not excluded using this threshold formed the 'stem' and their two-way and three-way interactions were entered sequentially using the same algorithm and exclusion criteria.
Results
Decision to consult model
The main predictor variables for the decision to consult model are GP density, bulk-billing rate, socioeconomic disadvantage of the resident population and geographic accessibility (ARIA). Age, gender and NESB status of the resident population are not predictors in this model. The signs of the predictor variables indicate that:
• Higher levels of socioeconomic disadvantage in the resident population are associated with higher levels of the decision to consult index.
• Lower ARIA scores (higher geographic accessibility) are associated with higher levels in the decision to consult index.
• Although the signs for GP density and bulk-billing rates are positive, the three-way interaction term in the model indicates that the relationships between the decision to consult an FFS GP and the supply of GPs and bulk-billing rate are not straightforward. (Table 3)
Table 3 Regression model for decision to consult
Model Unstandardized coefficients Standardized coefficients t-value Significance
B Std Error Beta
Constant 0.497 0.039 12.897 0.000
GP density 0.234 0.020 0.559 11.717 0.000
Index of disadvantage 0.018 0.003 0.299 6.458 0.000
Bulk-billing rate 0.127 0.032 0.177 3.977 0.000
ARIA -0.010 0.006 -0.205 -1.812 0.073
ARIA * GP density * Bulk-billing rate -0.037 0.008 -0.444 -4.308 0.000
Notes: ARIA = population weighted Accessibility/Remoteness Index of Australia
Higher index of disadvantage = higher level of socioeconomic disadvantage
Dependent variable = number of Whole Patient Equivalents (WPEs) per head of population (see Table 1)
Rewriting the regression equation in the following way clarifies the relationship between the decision to consult and the bulk-billing rate.
Decision to consult = 0.497 + 0.234*GP density + 0.018*Index of disadvantage + Bulk-billing rate*(0.127 - 0.037*ARIA*GP density) - 0.010*ARIA
This equation shows that the positive effect associated with any given bulk-billing rate will be greatest in the most highly accessible Divisions (i.e. where ARIA = 0), and the 'turning point' occurs where the bracketed term equals zero (i.e. when. ARIA*GP density = 3.4 and 0.037*ARIA*GP density = 0.127). In the data set used for this analysis, there are twenty-one Divisions where ARIA*GP density is greater than 3.4. The modelling predicts that increasing the bulk-billing rate ceteris paribus in these twenty-one Divisions will lead to a decrease in the number of people consulting an FFS GP within the Division. As can be seen in Table 4, not all these Divisions are characterised by low bulk-billing rates and/or low decision to consult indices.
Table 4 Divisions in which an increase in the bulk-billing rate will decrease the number of people consulting an FFS GP
Geographic Accessibility Bulk-billing rate Index of disadvantage† GP density‡ WPEs per head of population¶
Accessible*
1. 230 59% 6.8 0.96 0.82
2. 412 73% 4.7 1.19 0.93
3. 413 81% 4.5 1.06 0.85
4. 507 69% 8.0 1.37 0.90
5. 609 47% 5.3 0.89 0.80
Moderately Accessible*
6. 231 56% 7.2 0.90 0.78
7. 411 51% 5.6 0.96 0.79
8. 509 21% 7.7 1.26 0.84
9. 511 69% 7.1 0.84 0.86
10. 801 62% 5.8 1.07 0.67
Remote*
11. 241 76% 9.1 0.69 0.76
12. 416 64% 7.3 0.54 0.65
13. 417 64% 7.9 0.73 0.71
14. 512 72% 7.8 0.90 0.74
15. 611 59% 5.3 1.01 0.71
16. 612 58% 7.0 0.78 0.63
Very Remote*
17. 233 93% 8.9 0.71 0.70
18. 415 57% 6.8 0.65 0.50
19. 610 93% 8.0 1.06 0.38
20. 614 49% 4.8 0.79 0.61
21. 802 65% 6.8 0.89 0.51
Notes: * Categories based on the following ARIA values: highly accessible = 0 – 1.84, accessible = >1.84 – 3.51, moderately accessible = >3.51 – 5.80, remote = >5,80 – 9.08, very remote = <9.02 – 12. Source: Commonwealth Department of Health and Ageing, Accessibility/Remoteness Index of Australia (ARIA), available at Accessed 8th January 2002.
† Higher values indicate higher levels of socioeconomic disadvantage
‡ Vocationally registered and other GPs per 1,000 head of population
¶ WPE = whole patient equivalent (see Table 1)
The standardized beta coefficients in Table 3 indicate that GP density is the most important predictor in the decision to consult model. The regression equation can also be rewritten to emphasise the relationship between utilisation and GP density.
Decision to consult = 0.497 + 0.018*Index of disadvantage + 0.127*Bulk-billing rate + GP density*(0.234 - 0.037*ARIA*Bulk-billing rate) - 0.010*ARIA
The positive effect associated with GP density will be greatest in highly accessible Divisions (ARIA = 0) and the 'turning point' will occur when ARIA*Bulk-billing rate equals 6.3. In the current data set this occurs in the three very remote Divisions: 233, 610 and 802. The modelling predicts that increasing GP density ceteris paribus in these Divisions would lead to a decrease in the number of people consulting an FFS GP – an apparently perverse result.
Frequency of consultation model
GP density and socioeconomic disadvantage of the resident population do not enter as predictor variables in the frequency of consultation model. As shown in Table 5, the main predictors are the bulk-billing rate squared and the proportion of the resident population born in a non-English speaking background (NESB) country. There are two interaction terms: bulk-billing rate squared*ARIA; and bulk-billing rate squared*Proportion of the resident population born in an NESB country.
Table 5 Regression model for the frequency of consultation
Model Unstandardized coefficients Standardized coefficients t-value Significance
B Std Error Beta
Constant 4.296 0.120 35.777 0.000
Bulk-billing rate squared 2.767 0.204 0.619 13.539 0.000
Bulk-billing rate squared * ARIA -0.192 0.025 -0.248 -7.621 0.000
Propn popn born NESB country -4.154 1.841 -0.369 -2.256 0.026
Propn popn born NESB country * Bulk-billing rate squared 8.108 2.091 0.675 3.878 0.000
Notes: ARIA = population weighted Accessibility/Remoteness Index of Australia
NESB = Non-English Speaking Background
Dependent variable = number of consults per Standardised Whole Patient Equivalent (SWPE – see Table 1)
To emphasise the relationship between the bulk-billing rate and frequency of consultation the regression equation can be written as:
Frequency of consultation = 4.296 + Bulk-billing rate sq*(2.767 + 8.108*Propn population born NESB country - 0.192*ARIA) - 4.154*Propn population born NESB country
The positive impact associated with the bulk-billing rate will be highest in the most geographically accessible Divisions (i.e. where ARIA = 0) which have a high proportion of the resident population born in an NESB country. Assuming the proportion of the population born in an NESB country is zero, then the 'turning point' in the relationship between bulk-billing and frequency of consultation will occur when 0.192*ARIA equals 2.767 (i.e. ARIA = 14.4). Since the maximum ARIA value is twelve the 'turning point' will not be reached and the modelling predicts that a ceteris paribus increase in the bulk-billing rate will unambiguously increase the frequency of consultation across all Divisions.
Performance of the models
In both models the residuals are normally distributed and both have good explanatory power. The linear R2 value for the decision to consult is 0.845 and for the frequency of consultation it is 0.902. (Figure 1)
Figure 1 Performance of the models. Actual and predicted values of the dependent variables in the regression models (a) Decision to consult, (b) Frequency of consultation.
Discussion
Using publicly accessible data, aggregated to the level of the Division of General Practice, this research explores the associations between the bulk-billing rate, the supply of FFS GPs and utilisation of Medicare-funded GP services in rural and urban areas of Australia. Although the use of aggregated data has been described as a useful 'first-cut method' for getting insights into the 'interplay of macroeconomic variables for the analysis of economic policy', [42] such data has been criticised because it conceals individual differences in economic behaviour [43]. Because the primary aim of this research is to look at the likely impact of the Federal government's Strengthening Medicare policy on access to Medicare-funded GP services it is considered an appropriate methodology although there are some caveats that need to be noted.
No inferences can be drawn about the modelled relationships at the intra-Divisional level (i.e. at the individual or postcode level). It also needs to be borne in mind that the data used in the modelling related to a period some six years prior to the introduction of Strengthening Medicare. Policy changes in the intervening six years may have changed the nature of the modelled relationships. And a major assumption underlying the analysis is that the allocation of a single patient to different geographic areas is the result of border crossing by the patient. To the extent that the geographic allocation also results from border crossing by GPs (i.e. patients seen by a GP in one Division are linked to provider numbers in a different area) then the extent of border crossing by patients will be overstated. It is not clear from the data to what extent the results are confounded by GPs' border crossing.
Bearing these caveats in mind, the results indicate that the relationship between bulk-billing and utilisation is complex. This complexity arises, in part, because utilisation is a dynamic process involving both the decision to consult a doctor and the frequency of consultation once contact has been established. Each stage in this process involves a unique set of interacting predictor variables. In addition, each Division represents a unique combination of the predictor variables which means that the impact of changes in bulk-billing will differ from Division to Division. Despite this complexity it is possible to draw a number of inferences from the modelling that have important policy implications.
The level of socioeconomic disadvantage in the resident population has no impact on frequency of consultation but higher levels of socioeconomic disadvantage in the resident population are associated with higher decision to consult ratios. These results suggest that, at the Divisional level, the delivery of FFS GP services in the 1998–99 period was responsive to the needs of socioeconomically disadvantaged groups.
In many, but not all, Divisions an increase in the bulk-billing rate will unambiguously increase utilisation of Medicare-funded GP consults within a given period: it will increase both the number of people consulting an FFS GP and the number of visits per patient. However, in a number of Divisions, an increase in the bulk-billing rate will simultaneously increase the number of times existing patients consult and decrease the number of people consulting in a given period. The most straightforward interpretation of this phenomenon is that new patients are 'crowded out' by existing patients' increased use of services.
Increasing the supply of FFS GPs in a Division will, in all but three very remote Divisions, increase the number of people consulting an FFS GP in a given period without impacting on the number of times patients consult. This would mean an increase the total number of FFS GP consults in a Division in a given period and, assuming a stable population, an increase in the number of consults per head of resident population. The positive relationship between supply and the number of people contacting a GP is consistent with both the SID hypothesis and orthodox economic theory, but the modelling also implies that SID is not occurring at the point in the decision-making process where it might be most expected to occur – in the frequency of consultation. These results are consistent with Escarce's findings in relation to surgeons [32]. The reasons for the apparently perverse results in the three very remote Divisions cannot be clarified in this modelling exercise and require a more in-depth analysis to understand the dynamics underlying the result.
Conclusion
Bulk-billing rates and the supply of FFS GPs are two important supply-side elements in the Australian health care system that are potentially amenable to policy manipulation. To the extent that the Strengthening Medicare policy has increased bulk-billing rates, and bearing in mind the qualifications expressed in relation to the data, the modelling indicates that an increase in bulk-billing rates will not necessarily increase access across all Divisions. Unlike the American HIE results which showed no difference in response across geographic areas, in Australia, rurality is clearly important in understanding the relationships between cost, supply and access to medical care and needs to be considered when modelling the impact that policy initiatives on access to FFS GP services.
Strengthening Medicare was specifically designed to increase access for financially disadvantaged patients and those less than 16 years of age. The modelling suggests that, in the late 1990s, FFS GPs were responsive to the needs of socioeconomically disadvantaged patients and age was not an important predictor in the utilization of Medicare-funded GP services. If this situation still existed prior to the introduction of the new policy, then it is likely that the impact of the initiative on access to FFS GP services for these two groups of patients would be somewhat restrained.
Finally, geographic differences in bulk-billing rates and the supply of FFS GPs are often taken to be evidence of inequities in the provision of heath services. According to Hancock, social equity requires that the 'provision of health services ... should not discriminate on the grounds of ... rurality or geographical location' [44]. However, policy strivings for equivalence in performance indicators such as bulk-billing rates across geographic areas will not necessarily result in geographic equity of access. If geographic accessibility is not addressed, ensuring equity of access to FFS GP services in Australia would seem to require positive discrimination in favour of rural areas such that the supply of FFS GPs and bulk-billing rates are higher in rural than metropolitan areas.
Competing interests
This work is funded by an Australian Research Council grant and the North East Victorian Division of General Practice is an industry partner in the research.
Authors' contributions
SED: participated in the design, implemented the study drafted and revised the manuscript
KA: conceived the study, participated in its design and helped draft and revise the manuscript
DD: participated in the design of the study and revised the manuscript
SP: provided technical advice in the design of the study and revised the manuscript
LG: participated in the design and implementation of the data analysis and revised the manuscript
DV: conceived the study, participated in its design and revised the manuscript
Acknowledgements
The authors acknowledge the input of Dr Mark Robinson and Mr David Dart of the North East Victorian Division of General Practice in the initial development of this work.
==== Refs
Pratt A Health Insurance Amendment (100% Medicare Rebate and Other Measures) Bill 2004 Department of Parliamentary Services, Parliament of Australia Accessed 10th May, 2005
Department of Health and Ageing Annual Report 2003–04 2004 Canberra: Australian Government
Senate Select Committee on Medicare Medicare – healthcare or welfare? Senate Select committee on Medicare Secretariat Accessed 9th May, 2005
Senate Select Committee on Medicare Medicare Plus: the future for Medicare? Senate Select Committee on Medicare Secretariat Accessed 9th May, 2005.
Abbott T Ministerial media releases. ABB178/04. Bulk-billing rates continue to rise Department of Health and Ageing, Australian Government Accessed 12th May, 2005
Manning WG Health Insurance and the Demand for Medical Care: Evidence from a Randomized Experiment American Economic Review 1987 77 251 277 10284091
Richardson J The Effects of Consumer Co-payments in Medical Care 1991 National Health Strategy Background paper No. 5. Canberra: Department of Community Services and Health
Richardson J Does bulk billing cause abuse of Medicare? Community Health Stud 1987 11 98 107 3621895
Young AF Dobson AJ Byles JE Determinants of general practitioner use among women in Australia Soc Sci Med 2001 53 1641 51 11762890 10.1016/S0277-9536(00)00449-4
Rosenman SJ Mackinnon A General practitioner services under Medicare Aust N Z J Public Health 1992 16 419 26
Knox SA Britt H The contribution of demographic and morbidity factors to self-reported visit frequency of patients: a cross-sectional study of general practice patients in Australia BMC Family Practice 2004 5 7 published online 15084223
Furler JS The inverse care law revisited: impact of disadvantaged location on accessing longer GP consultation times Med J Aust 2002 177 80 3 12098344
Turrell G Oldenburg BF Harris E Jolley DJ Kimman ML Socioeconomic disadvantage and use of general practitioners in rural and remote Australia Med J Aust 2003 179 325 6 12964920
Richardson J The Inducement Hypothesis: That Doctors General Demand for Their Own Services Health, economics, and health economics: proceedings of the World Congress on Health Economics 1981 Leiden, The Netherlands: North-Holland Pub Co
Richardson J Peacock S Supplier Induced Demand Reconsidered Centre for Health Program Evaluation Accessed 26th May, 2005
Doessel DP Is an increased medical workforce a "problem" in the health sector? Theory and evidence the Nineteenth Australian Conference of Health Economists 1998 Canberra: Centre for Health Program Evaluation
Dranove D Demand inducement and the physician/patient relationship Economic Enquiry 1988 26 281 298
Folland S Goodman AC Stano M The Economics of Health and Health Care 2001 Upper Saddle River, NJ: Prentice Hall
Rice TH Labelle RJ Do Physicians Induce Demand for Medical Service? Journal of Health Politics, Policy and Law 1989 14 587 600 2794423
Donaldson C Gerrard K Paying general practitioners: shedding light on the review of health services Journal of the Royal College of General Practitioners 1989 39 114 117 2555489
Fuchs V The supply of surgeons and the demand for operations Journal of Human Resources 1978 13 35 56 722069
Fuchs VR Kramer MJ Determinants of Expenditures for Physicians' Services in the United Stated, 1948–1968 1972 New York: National Bureau of Economic Research
Cromwell J Mitchell JB Physician-induced demand for surgery Journal of Health Economics 1986 5 293 313 10317903 10.1016/0167-6296(86)90006-8
Phelps CE Induced Demand – Can We Ever Know Its Extent? Journal of Health Economics 1986 5 355 365 10282335 10.1016/0167-6296(86)90010-X
Scott A Culyer AJ, Newhouse JP Economics of General Practice Handbook of Health Economics 2000 Amsterdam: Elsevier 1176 1200
Grossman M Culyer AJ, Newhouse JP The Human Capital Model Handbook of Health Economics 2000 Amsterdam: Elsevier 347 408 10.1016/S1574-0064(00)80166-3
Duan N Manning W Morris CN Newhouse JP A comparison of alternative models of the demand for medical care Journal of Business and Economic Statistics 1983 1 115 126
Pohlmeier W Ulrich V An Econometric Model of the Two-Part Decisionmaking Process in the Demand for Health Care Journal of Human Resources 1995 30 339 361
McGuire TG Culyer AJ, Newhouse JP Physician Agency Handbook of Health Economics 2000 Amsterdam: Elsevier 461 536 10.1016/S1574-0064(00)80168-7
Rossiter LF Wilensky GR Identification of Physician-Induced Demand Journal of Human Resources 1984 19 231 244
Rossiter LF Wilensky GR A Reexamination of the Use of Physician Services: The Role of Physician-Initiated Demand Inquiry 1983 20 162 172 6222986
Escarce JJ Explaining the Association between Surgeon Supply and Utilisation Inquiry 1992 29 403 415 1473864
Humphreys JS Weinand HW Health status and health care in rural Australia: a case study Community Health Stud 1989 13 258 275 2605898
Strong K Trickett P Titulaer I Bhatia K Health in Rural and Remote Australia 1998 AIHW Cat No PHE 6. Canberra: Australian Institute of Health and Welfare
Cutler DM Zeckhauser RJ Culyer AJ, Newhouse JP The Anatomy of Health Insurance Handbook of Health Economics 2000 1A Amsterdam: Elsevier
Rice T Morrison KR Patient cost sharing for medical services: a review of the literature and the implications for health care reform Medical Care Review 1994 51 235 287 10138049
Zweifel P Manning WM Culyer AJ, Newhouse JP Moral Hazard and Consumer Incentives in Health Care Handbook of Health Economics 2000 1A Amsterdam: Elsevier
Review Panel The Future Role of the Divisions Network Commonwealth of Australia Accessed 25th July, 2004
Australian Government Medicare Benefits Schedule (MBS) Group Statistics Report Health Insurance Commission Accessed 14th August, 2004
GISCA ARIA vs RRMA National Key Centre for Social Applications of Geographical Information Systems (GISCA), University of Adelaide Accessed 6th October, 2002
Australian Bureau of Statistics Census of Population and Housing: Socio-economic Indexes for Areas (SEIFA), Australia Australian Bureau of Statistics Accessed 22nd September, 2003
Stoker TM Empirical Approaches to the Problem of Aggregation Over Individuals Journal of Econometric Literature 1993 XXXI 1827 1874
Schwartz S The Fallacy of the Ecological Fallacy: The Potential Misuse of a Concept and the Consequences Am J Public Health 1994 84 819 824 8179055
Hancock L Hancock L Rights and markets: What makes sustainable health policy? Health Policy in the Market State 1999 St Leonards NSW: Allen & Unwin 1 15
|
16111496
|
PMC1215471
|
CC BY
|
2021-01-04 16:38:28
|
no
|
Aust New Zealand Health Policy. 2005 Aug 20; 2:18
|
utf-8
|
Aust New Zealand Health Policy
| 2,005 |
10.1186/1743-8462-2-18
|
oa_comm
|
==== Front
Behav Brain FunctBehavioral and brain functions : BBF1744-9081BioMed Central London 1744-9081-1-151613140410.1186/1744-9081-1-15ResearchAssociation study of polymorphisms in synaptic vesicle-associated genes, SYN2 and CPLX2, with schizophrenia Lee Hee Jae [email protected] Ji Young [email protected] Jong Woo [email protected] Sheng-Yu [email protected] Mi Suk [email protected] Jin Kyoung [email protected] Joo-Ho [email protected] Hiroki [email protected] Yasuyuki [email protected] Medical Science Institute, Kangwon National University, Chunchon, Republic of Korea2 Department of Neuropsychiatry, College of Medicine, Kyung Hee University, Seoul, Republic of Korea3 Kohwang Medical Research Institute, College of Medicine, Kyung Hee University, Seoul, Republic of Korea4 Division of Disease Genes, Research Center for Genetic Information, Medical Institute of Bioregulation, Kyushu University, Fukuoka, Japan2005 31 8 2005 1 15 15 28 6 2005 31 8 2005 Copyright © 2005 Lee et al; licensee BioMed Central Ltd.2005Lee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The occurrence of aberrant functional connectivity in the neuronal circuit is one of the integrative theories of the etiology of schizophrenia. Previous studies have reported that the protein and mRNA levels of the synapsin 2 (SYN2) and complexin 2 (CPLX2) genes were decreased in patients with schizophrenia. Synapsin 2 and complexin 2 are involved in synaptogenesis and the modulation of neurotransmitter release. This report presents a study of the association of polymorphisms of SYN2 and CPLX2 with schizophrenia in the Korean population.
Methods
Six single nucleotide polymorphisms (SNPs) and one 5-bp insertion/deletion in SYN2 and five SNPs in CPLX2 were genotyped in 154 Korean patients with schizophrenia and 133 control patients using direct sequencing or restriction fragment length polymorphism analysis. An intermarker linkage disequilibrium map was constructed for each gene.
Results
Although there was no significant difference in the genotypic distributions and allelic frequencies of either SYN2 or CPLX2 polymorphisms between the schizophrenia and control groups, the two-way haplotype analyses revealed significant associations with the disease (P < 0.05 after Bonferroni correction). The three-way haplotype analyses also revealed a significant association of SYN2 with schizophrenia (P < 0.001 after Bonferroni correction).
Conclusion
These results suggest that both SYN2 and CPLX2 may confer susceptibility to schizophrenia in the Korean population.
==== Body
Background
Schizophrenia is a severe, chronic mental illness affecting 0.5–1.5% of the general population worldwide [1]. The contribution of genetic factors to the vulnerability to schizophrenia has been well established by family, twin, and adoption studies that have suggested a significant heritability of approximately 50–70% [2]. Many studies have attempted to identify the allelic variants that confer susceptibility to the illness, but no single genes have been identified that produce a major effect on the vulnerability [3].
Recently the synaptic hypothesis of schizophrenia has gained attention by attributing the fundamental pathology of schizophrenia to the dysfunction of synaptic transmission involving various molecules [4]. Synapsins, a family of synaptic vesicle-associated phosphoproteins, play a crucial role in the regulation of neurotransmission, synaptogenesis, and neuronal plasticity [5]. Three human synapsin genes have been identified (SYN1, 2, and 3; OMIM 313440, 600755, and 602705) [6]. Complexin 1 and complexin 2, which are encoded by CPLX1 (OMIM 605032) and CPLX2 (OMIM 605033), respectively, and are also called synaphins, are pre-synaptic membrane proteins that preferentially bind to syntaxin within the SNARE (soluble N-ethylmaleimide-sensitive fusion attachment protein receptors) complex. These proteins are important regulators of transmitter release immediately preceding vesicle fusion [7]. Previous studies have demonstrated that the concentrations of synapsins and complexins are reduced in the brains of schizophrenics [8,9]. The expression levels of both synapsins were significantly decreased in the hippocampal tissue of schizophrenic patients [10]. The levels of synapsin 2 and complexin 2 mRNA were also significantly reduced in the prefrontal cortex, cerebellum, and hippocampus of schizophrenics [11-14].
SYN2 was mapped to chromosome 3p25 [15], and CPLX2 is located on chromosome 5q35.3 (OMIM 605033). These loci were identified as potential regions conferring susceptibility to schizophrenia in diverse populations [16-18]. Based on their localization, well-established neurobiological roles, and expression patterns in schizophrenic patients, we selected SYN2 and CPLX2 as candidate genes for conferring susceptibility to schizophrenia. In this report, we present an association study of SYN2 and CPLX2 with schizophrenia using 12 polymorphisms in the Korean population.
Results
SYN2 polymorphisms in the schizophrenia and control groups
Of the seven polymorphisms in SYN2, rs2623873 (SYN2-1) is located in the promoter region, whereas the others are all located in the intronic regions (SYN2-2–7) (Fig. 1, Table 1). The genotypic distributions and allelic frequencies of polymorphisms in SYN2 were determined in 113 schizophrenic patients and 114 normal healthy controls by direct sequencing or DdeI RFLP. The genotypic distributions and allelic frequencies of polymorphisms in SYN2 are shown in Table 2. The average allelic frequency of the SNPs was 0.312. Given the equivalent frequency for the susceptible allele, the expected detection power for SYN2 was 0.9538 to 0.9929 under the multiplicative model with a genotype relative risk = 1.8 to 2.0 [22]. None of the SNPs showed any significant deviation from Hardy-Weinberg equilibrium (P > 0.05). We observed no significant difference in the genotypic distributions and allelic frequencies between the schizophrenics and control groups (Table 2).
Figure 1 Genomic organization of SYN2 and CPLX2 and locations of SNPs. a; SYN2 spans over 140 kb and is composed of 14 exons. Seven markers are indicated with the dbSNP reference ID . b; CPLX2 spans over 83 kb and is composed 3 exons. Five markers are indicated with the dbSNP reference ID .
Table 1 Twelve polymorphisms genotyped in this study.
Gene Name dbSNP rs# Region Allele Methods
SYN2 SYN2-1 rs2623873 Promoter G/T Direct Sequencing
SYN2 SYN2-2 rs2308169 Intron ATGCT/- Direct Sequencing
SYN2 SYN2-3 rs308961 Intron T/G Direct Sequencing
SYN2 SYN2-4 rs308963 Intron C/G Direct Sequencing
SYN2 SYN2-5 rs308952 Introna A/G DdeI RFLP
SYN2 SYN2-6 rs2279750 Introna A/C Direct Sequencing
SYN2 SYN2-7 rs310762 Intron C/T Direct Sequencing
CPLX2 CPLX2-1 rs2247916 Promoter G/T MaeIII RFLP
CPLX2 CPLX2-2 rs2243404 5'UTR C/T Cac8I RFLP
CPLX2 CPLX2-3 rs890736 Intron C/T AvaII RFLP
CPLX2 CPLX2-4 rs890737 Intron C/T Direct Sequencing
CPLX2 CPLX2-5 rs4390706 Intron G/A Direct Sequencing
a Tissue inhibitor of metalloproteinase 4 (Timp4) gene is nested within the intron of SYN2 in reverse orientation.
Table 2 Genotype distributions and allele frequencies of each polymorphism of the SYN2 and CPLX2 in the schizophrenia and control groups.
Polymorphism Subjects Genotype distribution (frequency) Allele frequency
11 12 22 Pa 1 2 Pb
SYN2-1 Cases (n = 113) 47 (0.416) 50 (0.442) 16 (0.142) 0.527 0.637 0.363 0.679
Controls (n = 114) 41 (0.360) 59 (0.518) 14 (0.122) 0.618 0.382
SYN2-2 Cases (n = 113) 40 (0.354) 55 (0.487) 18 (0.159) 0.758 0.597 0.403 0.438
Controls (n = 114) 36 (0.316) 56 (0.491) 22 (0.193) 0.561 0.439
SYN2-3 Cases (n = 113) 83 (0.735) 30 (0.265) 0 (0.000) 0.762 0.867 0.133 0.975
Controls (n = 114) 85 (0.746) 28 (0.246) 1 (0.009) 0.868 0.132
SYN2-4 Cases (n = 113) 49 (0.434) 45 (0.398) 19 (0.168) 0.722 0.633 0.367 0.612
Controls (n = 114) 44 (0.386) 51 (0.447) 19 (0.167) 0.610 0.390
SYN2-5 Cases (n = 113) 72 (0.637) 35 (0.310) 6 (0.053) 0.973 0.792 0.208 0.869
Controls (n = 114) 74 (0.649) 34 (0.298) 6 (0.053) 0.798 0.202
SYN2-6 Cases (n = 113) 67 (0.593) 39 (0.345) 7 (0.062) 0.786 0.765 0.235 0.622
Controls (n = 114) 63 (0.553) 44 (0.386) 7 (0.061) 0.746 0.254
SYN2-7 Cases (n = 113) 39 (0.345) 54 (0.478) 20 (0.177) 0.847 0.584 0.416 0.576
Controls (n = 114) 42 (0.368) 55 (0.482) 17 (0.149) 0.610 0.390
CPLX2-1 Cases (n = 154) 132 (0.857) 22 (0.143) 0 (0.000) 0.441 0.929 0.071 0.612
Controls (n = 133) 113 (0.850) 18 (0.135) 2 (0.015) 0.917 0.083
CPLX2-2 Cases (n = 154) 132 (0.857) 22 (0.143) 0 (0.000) 0.254 0.929 0.071 0.198
Controls (n = 133) 108 (0.812) 23 (0.173) 2 (0.015) 0.898 0.102
CPLX2-3 Cases (n = 154) 115 (0.747) 34 (0.221) 5 (0.032) 0.129 0.857 0.143 0.072
Controls (n = 133) 85 (0.639) 43 (0.323) 5 (0.038) 0.801 0.199
CPLX2-4 Cases (n = 154) 110 (0.714) 40 (0.260) 4 (0.026) 1.000 0.844 0.156 0.849
Controls (n = 133) 94 (0.707) 35 (0.263) 4 (0.030) 0.838 0.162
CPLX2-5 Cases (n = 154) 124 (0.805) 30 (0.195) 0 (0.000) 0.312 0.903 0.097 0.541
Controls (n = 133) 112 (0.842) 20 (0.150) 1 (0.008) 0.917 0.083
a Fisher's exact probability tests, case vs controls (2 × 3 genotype-based analysis)
b Fisher's exact probability tests, case vs controls (2 × 2 allele-based analysis)
We compared the LD for all possible two-way comparisons of the SNPs in the controls (Table 3). The pairwise D' values for the seven SNPs were consistently high, except in one instance (SYN2-2 vs. SYN2-6; D' = 0.300, r2 = 0.200). Out of the 21 possible pairs of SNPs, significant haplotype associations with schizophrenia were observed for 4 pairs: SYN2-1 – SYN2-2 (χ2 = 27.58, df = 3, P = 4.45 × 10-6), SYN2-2 – SYN2-4 (χ2 = 16.46, df = 3, P = 9.12 × 10-4), SYN2-2 – SYN2-7 (χ2 = 8.08, df = 3, P = 0.044), and SYN2-3 – SYN2-4 (χ2 = 10.66, df = 3, P = 0.014) (Table 3). Even after the Bonferroni correction (number of haplotypes, n = 21), the associations of the SYN2-1 – SYN2-2 and SYN2-2 – SYN2-4 haplotypes with schizophrenia remained significant (Pcorr = 9.35 × 10-5 and Pcorr = 0.019) (Table 3). The T allele-the deletion allele haplotype for the SYN2-1 – SYN2-2 combination and the deletion allele-the G allele haplotype for the SYN2-2 – SYN2-4 combination were observed more frequently in schizophrenia than the controls (Table 4).
Table 3 Pairwise linkage disequilibrium and haplotype association of SNPs in SYN2.
SYN2-1 SYN2-2 SYN2-3 SYN2-4 SYN2-5 SYN2-6 SYN2-7
SYN2-1 0.532 0.776 0.758 0.865 0.516 0.642
0.222 0.056 0.553 0.306 0.482 0.397
SYN2-2 4.45 × 10-6 a 0.613 0.453 0.734 0.300 0.429
0.044 0.250 0.174 0.200 0.225
SYN2-3 0.246 0.632 1.000 0.997 0.480 1.000
0.097 0.038 0.433 0.097
SYN2-4 0.845 9.12 × 10-4 a 0.014 1.000 0.480 0.806
0.395 0.433 0.650
SYN2-5 0.896 0.655 0.135 0.133 0.938 0.865
0.651 0.295
SYN2-6 0.116 0.892 0.602 0.602 0.147 0.722
0.278
SYN2-7 0.646 0.044 0.155 0.392 0.404 0.0.72
Upper diagonal top: D', bottom: r2 in controls; Lower diagonal: P value by χ2 test (df = 3)
a P < 0.05 after Bonferroni correction
Table 4 Estimated haplotype frequencies of the SYN2-1 – SYN2-2 and SYN2-2 – SYN2-4 combination on the SYN2
Haplotypes Estimated frequency
SYN2-1 SYN2-2 Cases Controls
T Deletion 0.583 0.461
T ATGCT 0.054 0.158
G Deletion 0.014 0.100
G ATGCT 0.349 0.281
SYN2-2 SYN2-4 Cases Controls
Deletion G 0.569 0.463
Deletion C 0.028 0.098
ATGCT G 0.064 0.146
ATGCT C 0.339 0.293
We also investigated the association of three-way haplotypes formed by SYN2-1, SYN2-2, and SYN2-4 with schizophrenia. A significant difference in the haplotype frequencies between the schizophrenia and control groups was observed (χ2 = 35.0, df = 7, Pcorr = 1.1 × 10-5). For the combination of SYN2-1, SYN2-2, and SYN2-4, the estimated frequencies of the T-deletion-G haplotype differed between the schizophrenia (0.570) and controls groups (0.440).
CPLX2 polymorphisms in schizophrenia and control groups
Of the five SNPs in CPLX2, rs2247916 (CPLX2-1) is located in the promoter region, rs2243404 (CPLX2-2) is located in the 5'UTR, and the others are located in the intronic regions (Fig. 1, Table 1). We determined the genotypic distributions and allelic frequencies of the SNPs in 154 schizophrenic patients and 133 normal healthy controls by direct sequencing or RFLP analysis. The genotypic distributions and allelic frequencies for CPLX2 SNPs are shown in Table 2. The average allelic frequency of the SNPs was 0.126. Given the equivalent frequency for the susceptible allele, the expected detection power for CPLX2 was 0.7445 to 0.8802 based on the multiplicative model with the genotype relative risk = 1.8 to 2.0 [22]. None of the five SNPs showed any significant deviations from Hardy-Weinberg equilibrium. We observed no significant differences in genotypic distributions or allelic frequencies between the schizophrenia and control groups (Table 2).
We compared LD for all possible two-way comparisons of the SNPs in controls (Table 5). The pairwise D' values for the five SNPs were consistently low, except in one instance (CPLX2-2 vs. CPLX2-4; D' = 0.715, r2 = 0.011). Only one pair of SNPs (CPLX2-1 vs. CPLX2-2) showed a significant haplotype association with schizophrenia (χ2 = 16.28, df = 3, P = 0.0009), even after the Bonferroni correction (n = 10, Pcorr = 0.009, Table 5). For the combination of CPLX2-1 – CPLX2-2, the G allele-the C allele haplotype was observed more frequently in the schizophrenia group than the control group (Table 6).
Table 5 Pairwise linkage disequilibrium and haplotype association of SNPs in CPLX2.
CPLX2-1 CPLX2-2 CPLX2-3 CPLX2-4 CPLX2-5
CPLX2-1 0.412 0.325 0.058 0.008
0.136 0.300 0.001 0.008
CPLX2-2 9.0 × 10-4 a 0.309 0.715 0.301
0.042 0.011 0.001
CPLX2-3 0.494 0.056 0.027 0.019
0.015 0.016
CPLX2-4 0.994 0.830 0.564 0.143
0.000
CPLX2-5 1.000 0.564 0.650 0.992
Upper diagonal top: D', bottom: r2 in controls; Lower diagonal: P value by χ2 test (df = 3)
a P < 0.01 after Bonferroni correction
Table 6 Estimated haplotype frequencies of the CPLX2-1 – CPLX2-2 combination on the CPLX2
Haplotypes Estimated frequency
CPLX2-1 CPLX2-2 Cases Controls
T C 0.010 0.036
T T 0.061 0.047
G C 0.919 0.862
G T 0.010 0.055
Discussion
In this study, we observed significant, pairwise haplotype associations with schizophrenia for two pairs of SNPs in SYN2 (SYN2-1 – SYN2-2 and SYN2-2 – SYN2-4; Pcorr = 9.35 × 10-5 and Pcorr = 0.019, respectively) and one pair of SNPs in CPLX2 (CPLX2-1 – CPLX2-2, Pcorr = 0.009) (Table 3, 5). The three-way haplotype (SYN2-1, SYN2-2, and SYN2-4) also showed a significant association with schizophrenia (Pcorr = 1.1 × 10-5). The SYN2-1 and CPLX2-1 SNPs are located in the respective promoter regions, -98 and -156. SYN2-1 was located within the GC box motif and CPLX2-1 within the C/EBP motif in a database search . The positive haplotype associations seem to be based on an increase of LD in the schizophrenia group compared to the control group because the D' values of the schizophrenia group were higher than those of the controls [(SYN2-1 – SYN2-2, 0.935 vs. 0.531 (schizophrenics vs. controls)), (SYN2-2 – SYN2-4, 0.750 vs. 0.453)] (Table 3). A similar situation was also observed with the positive association of the haplotype in CPLX2 with schizophrenia [CPLX2-1 – CPLX2-2, 0.852 vs. 0.412 (schizophrenics vs. controls)] (Table 5).
Chen et al. [23] recently reported an association study of four SNPs in SYN2 using Han Chinese samples. They found significant associations of SNP rs795009 and a haplotype constructed by the four SNPs with schizophrenia. Chen et al. [23] and our study examined two SNPs (rs2308169 and rs308963) in common, and their genotypic and allelic frequencies were similar in both studies. Although Chen et al. [23] did not mention the pairwise haplotype association study that we performed, they did report a significant difference in the overall four-way haplotype frequencies between schizophrenics and controls. Since two independent studies have reported a significant haplotype association of SYN2 with schizophrenia, this gene is probably involved in the pathogenesis of schizophrenia.
Several studies have suggested that the decreased expression of synaptic genes is characteristic of schizophrenia. In the hippocampus of schizophrenic patients, several studies have shown a consistent pattern of decreases in presynaptic proteins and their encoding mRNAs, such as synapsin 2, synaptophysin, and synaptosomal-associated protein-25 (SNAP-25) [8-10,24]. Furthermore, a reduction in the synapsin 2 mRNA levels was observed in the prefrontal cortex of schizophrenic patients [14], but controvertible results have also been reported [25]. The altered expression levels of other presynaptic proteins, complexin 1 and complexin 2, have been reported in schizophrenic patients [11-13]. Interestingly, complexin 1 is enriched in axosomatic regions, inhibitor neurons, and their synapses, while complexin 2 is enriched in the axodendritic terminals [9,26]. The differential expression of complexins 1 and 2 implies their involvement in the excitatory synapse in the hippocampus of schizophrenic patients [11]. These observations suggest that abnormal expression of SYN2 and CPLX2 may cause the vulnerability to schizophrenia by altering neurotransmitter release and neuroplasticity.
Conclusion
We found significant differences in the haplotype frequencies in both SYN2 and CPLX2 polymorphisms between schizophrenia and control groups. In addition, the haplotype constructed from three polymorphisms (SYN2-1, SYN2-2, and SYN2-4) showed a significant association with schizophrenia. Our results suggest that both SYN2 and CPLX2 polymorphisms may contribute susceptibility to schizophrenia in the Korean population.
Methods
Subjects
A total of 154 unrelated Korean schizophrenia patients (80 male and 74 female with a mean ± SD age of 43.8 ± 11.4 yr) and 133 unrelated Korean controls (64 male and 69 female; age 50.6 ± 11.7 yr) were recruited. For the SYN2 analysis, 113 unrelated Korean schizophrenia patients (60 male and 53 female with a mean ± SD age of 42.2 ± 11.3 yr) and 114 unrelated Korean controls (60 male and 54 female; age 51.7 ± 10.9 yr) were participated. The schizophrenia patients were diagnosed using the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV criteria. The control subjects were recruited after they had been designated as mentally healthy in a general health check-up program. The average age of the controls exceeded 50 years because we tried to avoid misincorporation of patients with late onset schizophrenia in the control group, while it may produce statistical bias potentially. Written informed consent was obtained from all subjects. This study was approved by the Ethics Committee of Kyung Hee University, Faculty of Medicine. Genomic DNA was extracted from whole blood cells using a NucleoSpin® Blood kit (Macherey-Nagel, Easton, PA).
SNP Selection and PCR-based Genotyping
Since the genomic sizes of SYN2 and CPLX2 are about 187 and 89 kb, respectively, we initially intended to select common polymorphisms at intervals of approximately 20–50 kb from the dbSNP . After validating the frequency of each polymorphism in 24 healthy Korean individuals using direct sequencing, we selected seven common polymorphisms from SYN2 and five from CPLX2 for further analyses (Fig. 1, Table 1). We amplified the fragments containing polymorphisms individually and genotyped DNA samples for each SNP with either PCR-based restriction fragment length polymorphism (RFLP) assays or direct sequencing performed with an ABI PRISM® Dye Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA) on an ABI PRISM® 3100 DNA sequencer (Applied Biosystems) (Table 1). In case of unclear sequence data, we repeated direct sequencing under various conditions until the genotype was determined correctly.
Statistics
The deviation of the genotypic frequencies from Hardy-Weinberg equilibrium was examined using the chi-square test (df = 1). Statistical differences in the genotypic distributions and allelic frequencies between the schizophrenia and control groups were examined using the Fisher's exact probability test. We calculated D' and r2 to evaluate the magnitude of linkage disequilibrium (LD) [19]. We estimated haplotype frequencies using the EH program, version 1.14 [20]. The statistical analysis of haplotype association was done as previously described [21]. We applied the Bonferroni correction to multiple testing based on the number of haplotypes. The significance level for all the statistical tests was 0.05.
Authors' contributions
HJ Lee conceived of the study, carried out sequencing, participated in the interpretation of the data and drafted manuscript, JY Song, JW Kim and JK Park recruited the samples of schizophrenia patients, SY Jin, MS Hong, and J-H Chung recruited the samples of normal control, H Shibata and Y Fukumaki participated in its design, carried out the statistical analyses, and participated in the interpretation of data. All authors read and approved the final manuscript.
Acknowledgements
This work was supported by JPSS and KOSEF, and in part by a grant of the Korean Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea. (0405-NS01-0704-0001).
==== Refs
Lewis DA Lieberman JA Catching up on schizophrenia: natural history and neurobiology Neuron 2002 28 325 334 11144342 10.1016/S0896-6273(00)00111-2
Owen MJ O'Donovan MC Gottesman II McGuffin P, Owen MJ, Gottesman II Schizophrenia Psychiatric Genetics & Genomics 2002 Oxford 247 266
Owen MJ Williams NM O'Donovan MC The molecular genetics of schizophrenia: new findings promise new insights Mol Psychiatry 2004 9 14 27 14581932 10.1038/sj.mp.4001444
Frankle WG Lerma J Laruelle M The synaptic hypothesis of schizophrenia Neuron 2003 39 205 215 12873379 10.1016/S0896-6273(03)00423-9
Hilfiker S Pieribone VA Czernik AJ Kao H-T Augustine GJ Greengard P Synapsins as regulators of neurotransmitter release Phil Trans R Soc Lond B 1999 354 269 279 10212475 10.1098/rstb.1999.0378
Kao H-T Porton B Hilfiker S Stefani G Pieribone VA DeSalle R Molecular evolution of the synapsin gene family J Exp Zool (Mol Dev Evo) 1999 285 360 377 10.1002/(SICI)1097-010X(19991215)285:4<360::AID-JEZ4>3.0.CO;2-3
Tokumaru H Umayahara K Pellegrini LL Ishizuka T Saisu H Betz H SNARE complex oligomerization by synaphin/complexin is essential for synaptic vesicle exocytosis Cell 2001 104 421 432 11239399 10.1016/S0092-8674(01)00229-X
Browning MD Dudek EM Rapier JL Leonard S Freedman R Significant reductions in synapsin but not synaptophysin specific activity in the brains of some schizophrenics Biol Psychiatry 1993 34 529 535 8274580 10.1016/0006-3223(93)90195-J
Harrison PJ Eastwood SL Preferential involvement of excitatory neurons in medial temporal lobe in schizophrenia Lancet 1998 352 1669 1673 9853440 10.1016/S0140-6736(98)03341-8
Vawter MP Thatcher L Usen N Hyde TM Kleinman JE Freed WJ Reduction of synapsin in the hippocampus of patients with bipolar disorder and schizophrenia Mol Psychiatry 2002 7 571 578 12140780 10.1038/sj.mp.4001158
Eastwood SL Harrison PJ Hippocampal synaptic pathology in schizophrenia, bipolar disorder and major depression: a study of complexin mRNAs Mol Psychiatry 2000 5 425 432 10889554 10.1038/sj.mp.4000741
Eastwood SL Cotter D Harrison PJ Cerebellar synaptic protein expression in schizophrenia Neuroscience 2001 105 219 229 11483314 10.1016/S0306-4522(01)00141-5
Knable MB Barci BM Webster MJ Meador-Woodruff L Torrey EF Molecular abnormalities of hippocampus in severe psychiatry illness: postmortem findings from the Stanley Neuropathology Consortium Mol Psychiatry 2004 9 609 620 14708030 10.1038/sj.mp.4001471
Mirnics K Middleton FA Marquez A Lewis DA Levitt P Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex Neuron 2000 28 53 67 11086983 10.1016/S0896-6273(00)00085-4
Karlin S Chen C Gentles AJ Cleary M Associations between human disease genes and overlapping gene groups and multiple amino acid runs Proc Nat Acad Sci 2002 99 17008 17013 12473749 10.1073/pnas.262658799
Lewis CM Levinson DF Wise LH DeLisi LE Straub RE Hovatta I Genome scan meta-analysis of schizophrenia and bipolar disorder, Part II: Schizophrenia Am J Hum Genet 2003 73 34 48 12802786 10.1086/376549
Pulver AE Lasseter K Kasch L Wolyniec P Nestadt G Blouin JL Schizophrenia: a genome scan targets chromosomes 3p and 8p as potential sites of susceptibility genes Am J Med Genet 1995 60 252 260 7573181 10.1002/ajmg.1320600316
Sklar P Pato MT Kirby A Kirby A Petryshen TL Medeiros H Carvalho C Genome-wide scan in Portuguese island families identifies 5q31-5q35 as a susceptibility locus for schizophrenia Mol Psychiatry 2004 9 213 218 14699422 10.1038/sj.mp.4001418
Taillon-Miller P Bauer-Sardina I Saccone NL Putzel J Laitinen T Cao A Kere J Pilia G Rice JP Kwok PY Juxtaposed regions of extensive and minimal linkage disequilibrium in human Xq25 and Xq28 Nat Genet 2000 25 324 328 10888883 10.1038/77100
Xie X Ott J Testing linkage disequilibrium between a disease gene and marker loci Am J Hum Genet 1993 53 1170
Sham P Sham P The analysis of allelic association Statistics in Human Genetics 1997 London: Arnold 145 185
Ohashi J Tokunaga K The power of genome-wide association studies of complex disease genes: statistical limitations of indirect approaches using SNP markers J Hum Genet 2001 46 478 482 11501946 10.1007/s100380170048
Chen Q He G Wang XY Chen QY Liu XM Gu ZZ Positive association between synapsin II and schizophrenia Biol Psychiatry 2004 56 177 181 15271586 10.1016/j.biopsych.2004.05.010
Thompson PM Egbufoama S Vawter MP SNAP-25 reduction in the hippocampus of patients with schizophrenia Prog Neuropsychopharmacol Biol Psychiatry 2003 27 411 417 12691775 10.1016/S0278-5846(03)00027-7
Imai C Sugai T Iritani S Niizato K Nakamura R Makifuchi T A quantitative study on the expression of synapsin II and N-ethylmaleimide-sensitive fusion protein in schizophrenic patients NeuroSci Lett 2001 305 185 188 11403936 10.1016/S0304-3940(01)01844-4
Takahashi S Yamamoto H Matsuda Z Ogawa M Yagyu K Taniguchi T Identification of two highly homologous presynaptic proteins distinctly localized at the dendritic and somatic synapses FEBS Lett 1995 368 455 460 7635198 10.1016/0014-5793(95)00713-J
|
16131404
|
PMC1215472
|
CC BY
|
2021-01-04 16:39:21
|
no
|
Behav Brain Funct. 2005 Aug 31; 1:15
|
utf-8
|
Behav Brain Funct
| 2,005 |
10.1186/1744-9081-1-15
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-1531596702410.1186/1471-2105-6-153Research ArticleEffective ambiguity checking in biosequence analysis Reeder Janina [email protected] Peter [email protected] Robert [email protected] InternationaI NRW Graduate School of Bioinformatics and Genome Research, Center of Biotechnology (CeBiTec), Bielefeld University, Postfach 10 01 31, 33501 Bielefeld, Germany2 Practical Computer Science, Faculty of Technology, Bielefeld University, Postfach 10 01 31, 33501 Bielefeld, Germany2005 20 6 2005 6 153 153 22 12 2004 20 6 2005 Copyright © 2005 Reeder et al; licensee BioMed Central Ltd.2005Reeder et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Ambiguity is a problem in biosequence analysis that arises in various analysis tasks solved via dynamic programming, and in particular, in the modeling of families of RNA secondary structures with stochastic context free grammars. Several types of analysis are invalidated by the presence of ambiguity. As this problem inherits undecidability (as we show here) from the namely problem for context free languages, there is no complete algorithmic solution to the problem of ambiguity checking.
Results
We explain frequently observed sources of ambiguity, and show how to avoid them. We suggest four testing procedures that may help to detect ambiguity when present, including a just-in-time test that permits to work safely with a potentially ambiguous grammar. We introduce, for the special case of stochastic context free grammars and RNA structure modeling, an automated partial procedure for proving non-ambiguity. It is used to demonstrate non-ambiguity for several relevant grammars.
Conclusion
Our mechanical proof procedure and our testing methods provide a powerful arsenal of methods to ensure non-ambiguity.
==== Body
Background
The ambiguity problem in biosequence analysis
Biosequence analysis problems are typically optimization problems – we seek the best alignment of two protein sequences under a similarity score, or the most stable secondary structure of an RNA molecule under a thermodynamic model. In such a problem, there is a "good" and a "bad" type of ambiguity. The good one is that there are many solutions to choose from. The bad one is that our algorithm may find the same solution several times, or even worse, it may study seemingly different solutions, which in fact represent the same object of interest. The cause of all these phenomenona has been called ambiguity, because it is closely related to the ambiguity problem of formal languages. It is not quite the same problem, however. In striving for avoidance of ambiguity, we want to get rid of the bad type and retain the good.
Ambiguity is not a problem with a dynamic programming (DP) algorithm that returns a single, optimal score, together with a solution that achieves this score, and does not make assertions about other solutions in the search space. Then, it does not matter whether this solution is analyzed several times, or that there are other solutions achieving the optimal score. In other cases, ambiguity can cause a DP algorithm to return an "optimal" answer which is plainly wrong. In the presence of ambiguity, the Viterbi algorithm cannot report the most likely structure [1], a folding program cannot produce a complete and non-redundant set of suboptimal structures [2], and statistics like counts, sum over all scores (by an Inside-type algorithm), or expected number of feasible or canonical structures [3] cannot be computed.
Previous work
The phenomenon of ambiguity has been formalized and studied in [3] in a quite general framework of dynamic programming over sequence data. There, it is shown that for a proof of non-ambiguity, a canonical model of the studied domain is required. The canonical model plays an essential role. It is the mathematical formalization of the real-world domain we want to study, and "canonical" means one-to-one correspondence. Any formal proof can only deal with the formalization of the real-world domain, and when the one-to-one correspondence does not hold, all proofs of (non-) ambiguity would be meaningless for the real world. In general, it may be quite difficult to find a canonical model for some real-world domains. Our case, however, is easy. When RNA secondary structure is our domain of study, base pair sets or the familiar dot-bracket strings can serve as a canonical model, as they uniquely represent secondary structures. To ensure non-ambiguity, there must exist an injective (i.e. one-to-one) mapping from derivation trees (according to the grammar underlying the DP algorithm) to the canonical model. While such a mapping may be easy to specify, the proof of its injectivity remains a problem.
Recently, Dowell and Eddy have re-addressed this problem [1] in the framework of stochastic context free grammars (SCFGs). In a probabilistic framework, ambiguity matters when a best, i.e. most likely solution is computed. This solution is wrong if several "different" solutions represent the same real-world object. Dowell and Eddy experimented with two ambiguous SCFGs, and showed that the quality of results may range from just slightly wrong to totally useless. After having shown that one cannot get by with ignoring ambiguity, they provide four non-ambiguous SCFGs for RNA structure analysis; however, a proof of their non-ambiguity is not given. Instead, they suggest a testing approach to check for the presence of ambiguity, which, of course, cannot prove its absence.
In this contribution, we first review the ambiguity problem in the framework of SCFG modeling, explain some of its sources, prove its algorithmic undecidability, and suggest three ways to deal with it: ambiguity avoidance, testing for ambiguity, and, best of all when successful, a mechanical proof of absence.
Formalization of ambiguity
We formalize the problem at hand in two steps, going from context free grammars (CFGs) to stochastic context free grammars, and then differentiating between syntactic and semantic ambiguity.
Formal grammars
A formal language is a subset of the set of all strings over a finite alphabet. Formal languages are typically described by formal grammars. In general, a formal grammar consists of an alphabet, a set of nonterminal symbols, and a set of production rules. There exist various grammar types, differing in the laws for construction of these production rules. The expressive power of a grammar type depends on these laws. In 1956, Noam Chomsky introduced a hierarchy of formal grammars that ranks grammar types by their expressive power, the Chomsky hierarchy [4]. It consists of four levels: regular grammars, context-free grammars, context-sensitive grammars, and unrestricted grammars. Here, we only address context-free grammars. These are suitable to describe the pseudoknot-free secondary structure of RNA. When considering pseudoknots, context-sensitive grammars are needed.
Context free grammars
A context free language is described by a context free grammar G, given by a set of terminal symbols (the alphabet), a set of nonterminal symbols, including a designated axiom symbol, and a set of production rules of the form X → α, where X is a nonterminal symbol, and a is a string of terminal and nonterminal symbols, α may be the empty string, denoted ε. Starting with the axiom symbol, by successive replacement of nonterminal symbols by right-hand sides of corresponding productions, we can derive a set of terminal strings. They constitute the language of the grammar, denoted L(G) Without loss of generality, derivations are canonized by replacing, in each step, the leftmost nonterminal symbol in the string obtained so far. Each such derivation can uniquely be represented as a derivation tree, and if the same terminal string has two different derivation trees, the grammar is called ambiguous.
Our first example is Dowell and Eddy's grammar G1 [1] to describe RNA secondary structures:
G1: S → aSu | uSa | cSg | gSc | gSu | uSg
S → aS | cS | gS | uS
S → Sa | Sc | Sg | Su
S → SS
S → ε
In the following, we shall use a shorthand notation, where a stands for any base A,C,G,U, while α and occurring in the same rule stand for either one of the base pairs (A,U), (U,A), (C,G), (G,C), (G,U), or (U,G).
G1: S → aS| aS | Sa | SS | ε
Four different derivation trees of the grammar G1 are shown in Figure 1. As they all emerge from the same terminal string acaggaaacuguacggugcaaccg, this grammar is ambiguous.
Figure 1 Four derivation trees. Four derivation trees for RNA sequence "acaggaaacuguacggugcaaccg", two (left) representing the annotation sequence ((((....)))).((((...)))) and two (right) the annotation sequence .(((....)))((...)).......
Stochastic context free grammars
Stochastic context free grammars associate a (nonzero) probability with each production, such that the probabilities for all alternative productions emerging from the same nonterminal symbol add up to 1. As a string is derived, probabilities of the involved rules multiply.
We extend the CFG G1 to a SCFG by the following example probabilities:
PS→aS = 0.2
PS→aS = 0.2
PS→Sa = 0.2
PS→SS = 0.2
PS→ε = 0.2
For simplicity, we chose probabilities independent of certain bases. In SCFG design, often also non-canonical base pairings are allowed with a low probability.
For grammar G1, the derivations shown in Figure 1 have probabilities of 5.24·10-14, 2.1·10-15, 4.19·10-16 and 4.19·10-16 (from left to right).
All derivations for a string can be constructed by a CYK-type parser [5]. The parser may compute the overall probability of a given string, summing up probabilities over all its derivations, in which case it is called the Inside algorithm. Or, the parser can return the most likely derivation of the input string, in which case it is known as the Viterbi algorithm. For grammar G1, the corresponding CYK-based Viterbi algorithm is shown here:
Input: Sequence x = x1 ... xn
Initialization: for 1 ≤ i ≤ n
S(i, i) = PS→ε
Iteration: for 1 ≤ i ≤ j ≤ n
Syntactic versus semantic ambiguity
Above, we introduced the formal language-theoretic notion of ambiguity: if the same symbol sequence has two or more different derivation trees, the grammar is called ambiguous. For clarity, we will refer to it as fl-ambiguity. In this sense, grammar G1 (and every other grammar in this manuscript) is in any case fl-ambiguous. This is demonstrated by the fact that the four derivation trees of Figure 1 all belong to the same symbol sequence. We now need to refine this notion of ambiguity.
In modeling with SCFGs, derivations do not merely produce strings, but they represent objects of interest themselves. With RNA, a derivation of an RNA sequence represents a possible secondary structure of this sequence. A more compact representation of a secondary structure is the widely used dot-bracket notation, as shown at the bottom of Figure 1. In the following, we will use the term annotation sequence for the dot-bracket string representing one secondary structure of the underlying RNA sequence. The one-to-one correspondence between (molecular) structures and (in silico) annotation sequences qualifies the latter as a canonical model of the grammar.
By the term syntactic ambiguity we denote the fact that typically an RNA sequence has many secondary structures, i.e. annotation sequences, hence many derivations. Figure 1 shows two example annotation sequences of the same RNA sequence.
Semantic ambiguity exists when there are, for some sequence, several derivations that represent the same annotation sequence, and hence, the same secondary structure. This is our point of study. In this case, the probability of a certain annotation sequence is split up into the probabilities of its multiple derivations. In Figure 1, this is exemplified by the two derivations on the left that both represent the annotation sequence ((((....)))).((((...)))), and the two derivations on the right, that both represent the annotation sequence .(((....)))((...))....... Thus, grammar G1 is syntactically as well as semantically ambiguous.
Semantic ambiguity is the "bad", syntactic ambiguity the "good" type of ambiguity in SCFG modeling and dynamic programming that was mentioned above. On the pure formal language level, they cannot be distinguished – both are manifest as fl-ambiguity. The bad ambiguity hides with the good, which is why its presence is sometimes overlooked.
Semantic ambiguity is not a problem with the Inside algorithm, as a probability sum over all derivations is computed anyway. With the Viterbi algorithm, we can certainly obtain the most likely derivation, but we do not know whether it represents the most likely annotation sequence. Some other annotation sequence may be more likely, but as its probability is the sum of many different derivations, none of these derivations may come out optimal. And even if the most likely annotation sequence is returned by the Viterbi algorithm, its computed probability is too small when there are further derivations of this annotation sequence.
As Dowell and Eddy have shown, this happens in practice and the effects are severe. For correct modeling with SCFGs, we need grammars that are syntactically, but not semantically ambiguous.
Semantic ambiguity in dynamic programming
Our treatment here extends to all dynamic programming algorithms that fall into the class known as algebraic dynamic programming (ADP) [6]. However, some definitions must be refined, as the ADP approach uses so-called yield grammars rather than (S)CFGs. We will not introduce the ADP formalism here, but remain within the SCFG terminology. Still, we shall refer to some DP algorithms that are not based on SCFGs, where our treatment also applies.
SCFGs for RNA secondary structure analysis
We will further exemplify the above using the grammars G1 to G6 studied by Dowell and Eddy:
Dowell and Eddy showed that grammars G1 and G2 are semantically ambiguous, while G3 to G6 passed a partial test for non-ambiguity.
Results and discussion
In this section, we first review some sources of ambiguity and suggest three ways to deal with it: ambiguity avoidance, testing for ambiguity, and, best of all when successful, a mechanical proof of absence.
Sources of ambiguity, and how to avoid them
We first study some standard patterns that give rise to ambiguity in our grammars. Thereafter, we make some observations with respect to the potential of testing procedures.
Three simple cases
Ambiguity does not sneak into our grammars by chance and non-awareness. There are two competing goals in grammar design, and both may foster ambiguity.
Small grammars have the advantage that they require fewer parameters and can be trained more quickly. Larger grammars allow a more sophisticated distinction of cases, hence providing a more fine-tuned model. However, if the underlying "distinct" cases lead to the same annotation sequence, then the grammar is ambiguous. This case is witnessed by grammar G2, where along with the introduction of base pair specific rules, another source of ambiguity is introduced.
Often, non-ambiguous grammars require more space in their implementation via a CYK parser. For example, the non-ambiguous Wuchty algorithm (RNAsubopt, [2]) requires four tables for storing intermediate results, while the ambiguous Zuker-Stiegler recurrences (Mfold, [7]) require only two. Two other cases in point are (a) and (b) below, while (c) shows that the non-ambiguous grammar can also be smaller.
Ambiguity can have many sources. Here, we present three common situations that lead us to write ambiguous rules, but can be easily avoided.
(a) Lists of adjacent elements of the same type, {Sn}:
Consider S → SS|U versus L → LS|S, S → U. The left-hand rule generates the language {Sn} in an ambiguous way. For example, S3 has the two derivations S → SS → SSS and S → SS → SSS, where the generating nonterminal symbol is written in bold face. By contrast, with the right-hand rules there is only the derivation L → LS → LSS → SSS. The price for non-ambiguity is the new nonterminal symbol L, more parameters in the training set, and possibly another DP table in the implementation.
(b) Embedded elements, {amTan}:
Consider R → aR|Ra|T versus R → aR|V, V → Va|T.
For a given string amTan, the first two alternatives of the left-hand rule produce the initial string am and the terminal an in arbitrary order, while the right-hand rules produce amcompletely before an, allowing for only one derivation. An analog case is the embedding {amTbn}. As above, an extra nonterminal symbol is required to achieve non-ambiguity.
(c) ε-rules, L → ε:
Sometimes it is tempting to add a special case by using ε. Consider L → LS|S|ε, which generates {Sn |n ≥ 0} by adding an ε-rule to the non-ambiguous rules in (a). Now, each string of length > 0 has two derivations, e.g. L → LS → S and L → S. The solution here is to drop the middle alternative, L → S.
The general case of ε-rules may be more tricky to handle. In general, all context free languages can be described without ε-rules, except possibly one for the axiom symbol. However, if ε-rules were used relentlessly, eliminating them without affecting the language may require a major redesign of the grammar.
Degree of ambiguity and consequences for testing
Dowell and Eddy showed that semantic ambiguity produces sometimes mildly, sometimes drastically false results. In one experiment, they showed that the CYK algorithm for the semantically ambiguous grammar G1 does not give the optimal secondary structure for about 20% of a sample set of 2455 sequences. The same experiment for grammar G2 even gave a rate of 98% false results. The explanation of the difference in effect lies with the degree of ambiguity. The degree of ambiguity of a given annotation sequence is the number of its derivations, i.e. a degree of 1 means that this annotation sequence is not ambiguous. Depending on the involved productions, a particular string can have a constant, polynomial, or exponential number of derivations. The latter is the rule rather than the exception. It is easy to calculate for the left production rule of case (b) above that the sequence {amTan} has derivations starting from S. Moreover, if derivations emerging from T are also ambiguous, the degrees of ambiguity multiply.
Studying sources of ambiguity helps to better understand the nature of the error. Depending on the grammar, certain types of RNA structures may have their probability split up over a large number of derivations, while others are unaffected. This makes it difficult to judge the amount of testing required, and the confidence achieved with the approaches presented in the next section.
Testing for ambiguity
Performing a test for semantic ambiguity allows us to obtain more confidence in the grammar, although testing cannot prove non-ambiguity, but only ambiguity.
Algorithmic arsenal for ambiguity testing
First, we create several variants of the Inside and Viterbi algorithms, which are our algorithmic arsenal for testing. Gl serves as the expository example here; for any other grammar, recurrences can be given in an analogous way:
Input: Sequence x = x1 ... xn
Initialization: for 1 ≤ i ≤ n
S (i, i) = PS→ε
Iteration: for 1 ≤ i ≤ j ≤ n
Scoring schemes:
= 1
PV→α = 0 for all other rules
By different interpretations of the operations H, o and P, different scoring schemes can be plugged in. The recurrences may also be "conditioned" by annotating the symbol sequence x with a given annotation sequence s [1]. In that case, the rule S → aS is only allowed when the bases involved are anno-tated to form a base pair in s. This version of the recurrences will be denoted by Gs.
Using the first two scoring schemes, we obtain the Viterbi and the Inside algorithm. Using the other two, we obtain an algorithm for counting the number of derivations for the input string, and an algorithm for base pair maximization. Base pair maximization will not be used in the sequel, it is included only to indicate the swiftness of transition from SCFG modeling to other DP-based analyses. These algorithms are available at the accompanying website [8], where readers are welcome to practice their insight on ambiguity matters.
In the following, we write G (σ, x) for running the CYK parser based on grammar G with scoring scheme σ on input x.
We recalled above that the formal treatment of semantic ambiguity requires a canonical representation of the objects under study. For RNA secondary structures, there is an obvious choice, our annotation sequences in the widely used dot-bracket notation (cf. Figure 1). Each secondary structure (excluding pseudoknots) is uniquely represented by such a string. The scoring scheme Dotbracket makes the CYK algorithm report all the structures it has analyzed for a given input sequence by producing their annotation sequences.
Here, the objective function H merely collects lists of dot-bracket strings, each PV→α is a function adding dots or brackets to strings. is string concatenation. PS→SS is also string concatenation, but has the unusual type String → (String → String), in order to fit into our recurrences smoothly. Here, o expects a dot-bracket string as its left argument, a function as its right argument, and applies the latter to the former. For example, the function calls PS→SS (PS→gSu (PS→ε)) (PS→aS (PS→ε)) generate the annotation sequence ". ()" for the symbol sequence "agu". The reader may verify (using the aforementioned website) that G1 (Dotbracket, "agu") = ["(.)","(.)",".()","...","...", etc.], where the duplicate entries result from the ambiguity of G1.For example, the annotation sequence "..." is found 48 times.
Using these algorithms in concert for some RNA sequence x, we obtain from G(Viterbi, x) the probability of the most likely derivation for x, from G(Counting, x) the number of possible derivations, and from G(Dotbracket, x) the complete list of the annotation sequences associated with these derivations – possibly containing duplicates in the case of semantic ambiguity.
Testing procedures
Brute force testing
Checking for duplicates in G(Dotbracket, x). We can simply enumerate the dot-bracket representation of all structures exhaustively for a given input string and check for any repeats. There are some duplicates in G(Dotbracket, x) if and only if x can fold into an ambiguous annotation sequence (which may be precluded by its nucleotide content). Performing this test on a large number of inputs x should give a good hint whether ambiguity is present. Of course, enumerating the annotation sequences for all possible derivation trees creates voluminous output, and the automated check for duplicates requires some careful programming. Hence, this test is practical only for short sequences.
Sampling structures from sample sequences
Test G(Viterbi, x) = Gs(Inside, x)? Dowell and Eddy suggested a testing procedure that relies on a comparison of the results from the Viterbi and the Inside algorithms, where the latter is conditioned on the most likely annotation sequence s returned by the Viterbi run. Gs(Inside, x) sums up probabilities over all derivations representing annotation sequence s. The tested equation therefore holds if and only if the annotation sequence s has exactly one derivation tree. If there are more than one, the Inside algorithm will return a higher probability than the Viterbi run, which indicates ambiguity of s (and hence G). Similarly, Gs(Counting, x) directly computes the number of derivations for s, where a result larger than 1 signals ambiguity.
Dowell and Eddy suggest to run the test also for a sample of suboptimal annotation sequences for x. As a variant, we can do the same test based on a minimizing Viterbi run (setting H = min). Since the minimizing Viterbi run gives us the least probable derivation tree, we may have a higher chance to find an ambiguous one (if present) than in the maximizing run.
In any case, this test works with samples of sub-optimal annotation sequences for a test set of sequences, and it is difficult to give general guidelines how much testing is required. The four grammars G3 - G6 passed the Dowell-Eddy test in [1], and in the next section we shall prove their non-ambiguity. In this sense, we can state that this test has already worked quite well in practice. However, the eternal dilemma of testing persists – only if we confirmed the above equation for all x, semantic non-ambiguity would be assured.
Structure counting for sample sequences
Test G(Counting, x) = R(Counting, x)? An even stronger test is possible when we have a reference grammar R available that generates the same language and is known to be semantically non-ambiguous. Grammar G will produce counts that are larger than those of R if and only if G allows ambiguous derivations for x. Still, if this test succeeds, this does not imply non-ambiguity of G. But this test is much more thorough than our previous one, as the entire structure space of each tested x is analyzed. For example, a sequence of length 30 has an expected number of 175550 feasible structures [3]. Thus, one run of this test has the testing power of 175550 runs of the previous one. Several non-ambiguous reference grammars for RNA are known – the critical part here is to assure that our grammar G to be tested describes the same language as R. Both grammars must impose the same restrictions on loop sizes, lonely base pairs, etc. This may be obvious in many cases, but in general, language equivalence is an undecidable problem in formal language theory.
Just-in-time testing
Test G(Counting, x) = R(Counting, x)? While testing cannot guarantee the non-ambiguity of the grammar, we can convert the previous idea to a test that ensures for each application run that the results are not affected by ambiguity. Prior to running G(Viterbi, x) for a given x, we test whether the property G (Couniing, x) = R(Counting, x)holds. This costs a constant factor in runtime, but solves the problem in the sense that when we get a positive test, we know the Viterbi result is correct for this input. If the grammar is ambiguous, this will be detected with the first application where it occurs.
Proving non-ambiguity
Proving the absence of ambiguity in a grammar is of course better than any test procedure.
Semantic ambiguity in dynamic programming is unde-cidable
Ambiguity of context free grammars is well-known to be algorithmically undecidable [9]. There exists no program that can determine for an arbitrary grammar G whether or not G is fl-ambiguous. Here, the problem is to decide whether a given SCFG is se-mantically ambiguous. It is not surprising that this problem is not easier:
Theorem 1 Semantic ambiguity in dynamic programming is formally undecidable
Proof
We show that for a given CFG G there exists a DP problem and an associated canonical model such that the DP algorithm is semantically am biguous if and only if the grammar is fl-ambiguous. Given an algorithm to decide ambiguity for DP problems, we could hence decide ambiguity for context free grammars, which is impossible. Details are given in the appendix. □
While this result rules out an automated proof procedure for arbitrary grammars used in SCFG modeling, there might still be the possibility to design such a procedure for a restricted class of grammars, say all grammars which describe RNA secondary structures. However, no such method is currently known.
Hand-made proof of non-ambiguity
A hand-made de-novo proof of the non-ambiguity of a new grammar G requires an inductive argument on the number of parses corresponding to the same annotation sequence. We constructed one such proof for the grammar published in [3]. It is not mathematically deep, but rather a tedious exercise, and the likelihood to produce errors or oversights is high. An easier approach is the use of a known, non-ambiguous "reference" grammar R, such that L(G) = L(R). By showing that a one-to-one mapping between parse trees of G and R exists, it is possible to prove the non-ambiguity of G. Such a proof remains manageable if the grammars are rather similar and the correspondence between derivations is easy to maintain. For grammars that are rather distinct, the proof is as messy as the de-novo proof.
Mechanical proof of non-ambiguity
We now present a mechanical technique that is a partial proof procedure for the case of modeling RNA structure with SCFGs: If it succeeds, it proofs non-ambiguity, if it fails, we do not know. We shall show that the method succeeds on several relevant grammars.
The technique described in the following comprises two steps. First, we remove the syntactic ambiguity of the grammar and reduce a possibly existent semantic ambiguity to fl-ambiguity. Then we use a parser generator to check the transformed grammar for fl-ambiguity. This test can prove non-ambiguity of a large number of grammars.
Ambiguity reduction
Paired bases can always also be unpaired – this creates the syntactic (good) ambiguity. For example, grammar Gl has four rules of the form S → aS, one for each base A, C, G, U, and six rules of the form S → aS for the six valid base pairs. Used in concert, they create the "good" ambiguity that allows us to parse "CAAAG" either as "(...)" or as ".....".
Remember that the dot-bracket notation is a canonical representation for RNA secondary structure. For any G, we denote by G* the transformed grammar that arises when we replace base pairs a, by "(" and ")", and other base symbols by ".". Take for example
G5 : S → aS | aSS | ε (11 productions), which is transformed to
G5* : S → '.'S | '('S')' S | ε (3 productions).
This transformation removes the syntactic ambiguity of G5 by differentiating between paired and unpaired bases and reduces the semantic ambiguity -if present – to fl-ambiguity of G5*. Note that the transformation from G to G* works for any grammar for RNA structure, as long as we can identify the corresponding bases of a base pair.
Theorem 2 Let G* be derived from G according to the above rules. Then, G* is fl-ambiguous if and only if G is semantically ambiguous
Proof
Every dot-bracket string describes exactly one possible secondary structure. If G* is fl-ambiguous, there exist different derivations in G* for the same dot-bracket string z. Then, for an RNA sequence x compatible with z, using the corresponding productions there are different derivations in G which represent the same secondary structure z. This is equivalent to semantic ambiguity of G. If G* is non-ambiguous, only a single derivation exists for every z in L(G*). A single derivation exists in G for a compatible RNA sequence x, and hence, G is semantically non-ambiguous. □
Non-ambiguity proof
By the transformation described above, the task of proving semantic non-ambiguity of G is transformed to the task of proving fl-non-ambiguity of G*. As stated above, this question is undecidable in general. However, compiler technology provides a partial proof procedure: If a deterministic parser can be generated for a grammar, then it is non-ambiguous [5]. We shall apply a parser generator to G*.
Simply speaking, a parser generator takes a file with a context free grammar as input, and generates a program which implements the parser for this grammar. This parser must be deterministic, and, in contrast to our CYK parsers, it only exists for non-ambiguous grammars. There are many such generators available; we will focus on the class of LR(k) grammars [10] and their parser generators. A context free grammar is called LR(k) if a deterministic shift reduce parser exists that uses k symbols of lookahead. By definition, an LR(k) grammar is non-ambiguous, and for a given k it is decidable whether a grammar is LR(k). This decision can be assigned to a parser generator. Given the grammar and the lookahead k, a parser generator tries to construct a parser that uses k symbols of lookahead. When successful, the non-ambiguity of the grammar is proved. When the grammar is not LR(k), the generator will not be able to create a deterministic parser and reports this situations in form of "shift-reduce" and "reduce-reduce"-conflicts to the user. In this case, we do not know whether the parser generator might be successful for a larger k, and the question of ambiguity remains undecided.
Applications
For our experiments, we used the MSTA parser generator of the COCOM compiler construction toolkit [11]. MSTA is capable of generating LR(k) parsers for arbitrary k. Note that compiler writers prefer other parser generators like yacc [12] and bison [13], which for efficiency reasons only implement LR(1) parsers. We, however, are not planning to run the parser at all. Its successful construction is the proof of non-ambiguity; for applying our SCFG, we need the original grammar and its CYK parser.
MSTA accepts input files in the widely used yacc format. The following shows the input file for grammar G5:
Feeding this file into MSTA with k = 1 yields a deterministic shift-reduce parser for grammar G5. This proves that G5 is LR(1), has a deterministic LR(1) parser, and is therefore non-ambiguous.
Table 1 summarizes the results for grammars G1 to G6. For G1 and G2, the results only show that both grammars are not LR(1), LR(2) or LR(3). Although no real proof, the magnitude and growth of the number of conflicts with increasing k gives a strong hint at the ambiguity of these grammars.
Table 1 Results of mechanical proof procedure. Number of shift-reduce (SR) and reduce-reduce (RR) conflicts when feeding example grammars G1 to G8 into parser generator MSTA. A 0/0 entry indicates a successful proof of non-ambiguity. Note that for increasing k, the number of conflicts may remain constant or even grow before it goes down to 0/0.
Grammar k SR/RR conflicts
G1 1 24/12
G1 2 70/36
G1 3 195/99
G2 1 25/13
G2 2 59/37
G2 3 165/98
G3 1–3 4/0
G3 4 16/0
G3 5 0/0
G4 1 0/0
G5 1 0/0
G6 1 0/0
G7 1–6 5/0
G7 7 0/0
G8 1 0/0
Grammar G3 is LR(5) and G4 to G6 are LR(1). Therefore, we have proved mechanically that the four "good" grammars studied by Dowell and Eddy are definitely non-ambiguous. The two additional grammars G7 and G8 from [1], not reproduced here, were also included in the study and proved to be non- ambiguous.
In Table 1 we also report on the number of conflicts found by the parser generator for increasing values of k. While the nature of these conflicts is not relevant for us, the table shows that various behaviors are possible. Their numbers may grow (G3) or may remain constant (G7) before they go to zero for some k.
Experience from a larger example
The parser generator test works quite well for the small grammars we presented so far. However, there exist cases where, due to the finite lookahead of the generated parser, the parser generator reports conflicts while the grammar is in fact non-ambiguous. In the following, we report on one such case, and show how to deal with this situation.
In his thesis [14], Björn Voss introduced a new grammar that promises to handle dangling bases of multiloop components in a non-ambiguous way. With 28 nonterminal symbols and 79 rules, the grammar is quite large. In such a case, mechanical assistance is strongly required. Our first approach with the parser generator succeeded, except for one small part of the grammar for which it reports a con-flict. Figure 2 shows two example derivations where this conflict occurs.
Figure 2 Two example derivations. Two example derivations of a grammar taken from [14]. The left side is part of a multiloop derivation, the right side part of a left bulge.
The central nonterminal of the grammar is CL, which splits up into closed structures like hairpin loops, bulges, and multiloops. Due to the necessity to handle dangling bases in a non-ambiguous way, the rules for multiloops are the most complicated of this grammar. Altogether, 11 nonterminals and 35 rules are used exclusively for this purpose. The construction of these rules guarantees, that every derivation of a multiloop must lead to at least two closed substructures. One of these derivations is shown on the left side of Figure 2. Therefore, a derivation of a multiloop can by no means conflict with a derivation of a left bulge, which must include a single closed substructure. However, the parser generator runs into a conflict here. Consider the following annotation sequence:
Here, the string "((...((" appears two times in the annotation sequence. The first appearance denotes a left bulge, the second the beginning of a multiloop. The decision which of these two is given can only be made after the first closed substructure is completely processed. Since the generated parser can only read a limited number of input characters ahead (k), the parser generator is not able to construct a deterministic parser for this situation and reports a conflict.
However, we can circumvent this problem by extending the alphabet of the annotation sequence by an additional character (say, ':') for unpaired bases in left bulges1:
Since a multiloop's derivation can not conflict with that of a bulge, this modification does not alter the ambiguity or non-ambiguity of the grammar. The important difference is that positional information is turned into symbolic information.
After this modification, the parser generator runs smoothly through the grammar, which proves its non-ambiguity.
Conclusion
In this work, we have presented testing methods and a partial proof procedure to analyze the semantic ambiguity of SCFGs. We have shown that the problem is not decidable for dynamic programming over sequence data in general, and that hence there is no standard solution that works for all cases. It remains open whether specifically for the class of grammars that describe RNA secondary structure, this problem is decidable. We have proposed several tests, and a partial, mechanical proof procedure. We mechanically proved that the six grammars that passed Dowell and Eddy's test for non-ambiguity are actually non-ambiguous. We also reported on a proof of the non-ambiguity of a new and large grammar for RNA secondary structures, whose sophistication makes it inadvisable to rely solely on human reasoning.
We want to point out that the non-ambiguity proofs for the grammars studied here do not solve the problem of ambiguity for modeling of RNA secondary structures once and for all. New scientific interests and research questions will always demand new grammars. An example is a grammar that is restricted to a special class of structures of an RNA family. This allows us to define a thermodynamic matcher, which uses the minimum free energy as a scoring scheme and focuses only on a specific realm of secondary structures. Here, for every new RNA family, a new grammar must be devised. This demonstrates a continuous need for new, specialized grammars. Every time we develop a new grammar, the dragon of ambiguity raises its head, but with the weapons presented here, we can be confident to defeat it.
Methods
Our way to describe various tests by combining a grammar with varying scoring schemes is derived from the algebraic dynamic programming method, described in detail in [6]. The theoretical framework of this method also underlies the proof of Theorem 1. The parser generator MSTA used as a partial proof method is available at [11].
Authors' contributions
RG suggested the topic and contributed the undecid-ability proof. JR worked out the testing procedures and PS the mechanical proof procedure. All authors cooperated closely in writing the manuscript.
Appendix: Ambiguity in DP is undecidable
Dynamic programming is a very general programming technique, and its scope is not precisely circumscribed. We prove our undecidability result for the well defined class of algebraic dynamic programming [6] problems, which of course implies undecidability in general. Simply speaking, a DP problem is given by a grammar G and a scoring scheme σ (not necessarily stochastic), as was exemplified in Section Testing for ambiguity.
Theorem 1 Semantic ambiguity in dynamic programming is formally undecidable
Proof
For an arbitrary context free grammar G, we can construct a DP problem where L(G) serves as the canonical model, and show that the context free grammar G is ambiguous if and only if the DP problem is semantically ambiguous.
Let G be a context free grammar. Without loss of generality, we can assume that each production is either of the form A → t, generating a terminal symbol, or of A0 → A1...An, n ≥ 0, generating a series of nonterminal symbols. We construct a scoring scheme σ for grammar G such that G(σ, x) computes all derivation trees for x. Similar to the scoring scheme Dotbracket, we set H = collect and xof = f(x). For each production π we use a unique tree label Tπ. We define ...An (an)...(a1) = ...An (a1,...,an), and PA→t = t.
By construction, G(σ, x) constructs the list of all derivation trees for x. The canonical mapping ν (from derivation trees to their derived strings) is simply given by ν(...An (a1,...,an)) = ν(a1)...ν(an) and ν(t) = t. By construction, the domain of v are the derivation trees of G, its range is L(G). Hence, v is injective if and only if G is non-ambiguous. Could we formally decide the semantic ambiguity of an arbitrary DP problem, we could do so for the problem given by G and σ, and hence, ambiguity of context free languages would be decidable. □
Note
1For the same reason, this modification is also necessary in the rules for internal loops.
Acknowledgements
We are grateful to Jens Reeder for discussion and the anonymous referees for very detailed and helpful comments. Work of JR is funded by DFG Graduate College 635 Bioinformatik.
==== Refs
Dowell R Eddy SR Evaluation of several lightweight stochastic context-free grammars for RNA secondary structure prediction BMC Bioinformatics 2004 5 71 15180907 10.1186/1471-2105-5-71
Wuchty S Fontana W Hofacker I Schuster P Complete Suboptimal Folding of RNA and the Stability of Secondary Structures Biopolymers 1999 49 145 165 10070264 10.1002/(SICI)1097-0282(199902)49:2<145::AID-BIP4>3.0.CO;2-G
Giegerich R Explaining and Controlling Ambiguity in Dynamic Programming Proc Combinatorial Pattern Matching, Springer LNCS 1848 2000 46 59
Chomsky N Three Models for the Description of Language IRE Transactions on information theory 1956 2 113 124 10.1109/TIT.1956.1056813
Aho A Ullman J The Theory of Parsing, Translation and Compiling 1973 Englewood Cliffs, NJ: Prentice-Hall [I and II]
Giegerich R Meyer C Steffen P A Discipline of Dy namic Programming over Sequence Data Science of Computer Programming 2004 51 215 263 10.1016/j.scico.2003.12.005
Zuker M Stiegler P Optimal Computer Folding of Large RNA Sequences using Thermodynamics and Auxiliary Information Nucleic Acids Research 1981 9 133 148 6163133
Effective Ambiguity Checking in Biosequence Analysis
Chomsky N Schützenberger MP Braffort P, Hirschberg D The algebraic theory of context-free languages Computer Programming and Formal Systems 1963 Amsterdam: North-Holland 118 161
Knuth D On the Translation of Languages from Left to Right Information and Control 1965 8 607 639 10.1016/S0019-9958(65)90426-2
COCOM tool set
Johnson SC YACC: Yet Another Compiler Compiler Computing Science TR 1975 32
Bison parser generator
Voss B Advanced Tools for RNA Secondary Structure Analysis PhD thesis 2004 University of Bielefeld
|
15967024
|
PMC1215473
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Jun 20; 6:153
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-153
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2101612238610.1186/1471-2105-6-210Research ArticleVariation in structural location and amino acid conservation of functional sites in protein domain families Pils Birgit [email protected] Richard R [email protected] Jörg [email protected] Department of Bioinformatics, University of Würzburg, Biocenter, Am Hubland, 97074 Würzburg, Germany2 Wellcome Trust Centre for Human Genetics, University of Oxford, Headington, OX3 7BN Oxford, UK2005 25 8 2005 6 210 210 24 5 2005 25 8 2005 Copyright © 2005 Pils et al; licensee BioMed Central Ltd.2005Pils et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The functional sites of a protein present important information for determining its cellular function and are fundamental in drug design. Accordingly, accurate methods for the prediction of functional sites are of immense value. Most available methods are based on a set of homologous sequences and structural or evolutionary information, and assume that functional sites are more conserved than the average. In the analysis presented here, we have investigated the conservation of location and type of amino acids at functional sites, and compared the behaviour of functional sites between different protein domains.
Results
Functional sites were extracted from experimentally determined structural complexes from the Protein Data Bank harbouring a conserved protein domain from the SMART database. In general, functional (i.e. interacting) sites whose location is more highly conserved are also more conserved in their type of amino acid. However, even highly conserved functional sites can present a wide spectrum of amino acids. The degree of conservation strongly depends on the function of the protein domain and ranges from highly conserved in location and amino acid to very variable. Differentiation by binding partner shows that ion binding sites tend to be more conserved than functional sites binding peptides or nucleotides.
Conclusion
The results gained by this analysis will help improve the accuracy of functional site prediction and facilitate the characterization of unknown protein sequences.
==== Body
Background
Protein function is determined by the spatial configuration and type of amino acids at functional sites. Knowledge of functional sites provides valuable information for the assignment of molecular function, potential physiological binding partners and hence drug design. Tasks performed by functional sites range from the binding of small molecules like ions, cofactors, metabolic substrates or high molecular weight compounds such as nucleic acids and peptide chains, to catalysing chemical reactions in the active centre of enzymes.
The exponentially growing number of uncharacterised protein sequences in the public databases has turned the development of automatic identification of functional sites into an important research field and many computational methods focusing on this area have been described in recent years (for review see [1-3]). In contrast to structural approaches that search for ligand binding pockets on the protein surface using molecular modelling [4,5], network analysis [6], or compare the protein surface to structures with known interacting sites [7,8], many methods are based on a set of homologous sequences combined with evolutionary or structural information. The evolutionary trace (ET) method [9,10], for example, searches for a structural cluster of conserved residues. Beginning with a sequence identity tree from a set of homologous proteins, the tree is scanned for subgroup-specific residues, which are invariant within the subgroup but vary between subgroups. These residues, called evolutionary trace residues, and the residues that are invariant in all sequences are then mapped onto a representative 3D structure and clusters of high ranking residues, corresponding to the inner nodes of the tree, are searched. These clusters usually coincide with the functional center of the protein. Improvements of the ET method use sequence weights based on their similarity (weighted evolutionary tracing) and an amino acid substitution matrix to account for biochemically similar amino acids in the identification of the trace residues [11], they consider the evolutionary distance between proteins due to the phylogenetically biased databases [12] or allow different rates of amino acid substitutions at protein sites [13]. A similar approach is focusing more on structural information and calculates a conservation score at each position under consideration of the behaviour of spatial neighbours [14].
Most of the above mentioned methods assume that functional sites are under high selective pressure and conserved within the protein, so that functional sites can easily be detected by lower rates of amino acid substitutions. However, functional sites can vary in subfamilies and homologous protein sequences can perform different functions using a different set of functional residues. Accordingly, interaction interfaces can vary in their location in distant homologues and this has to be considered if interaction interfaces are inferred from homologous proteins [15,16].
Prior to their prediction, it is necessary to understand the arrangement and properties of functional sites, as well as how protein families and single sequences differ in their use. Effort towards this direction has been made by several groups, who studied physicochemical properties of protein-protein interaction interfaces found in homodimer, heterodimer or intra-chain domain complexes [17-21], as well as protein-DNA interaction interfaces [22].
Here, we perform a large-scale analysis of functional sites extracted from experimentally determined protein-ligand complexes stored in the protein data bank (PDB) [23], grouped by the presence of conserved protein domains described in the SMART database [24]. The classification into protein families enables us to find differences in amino acid conservation and use of specific locations for functional sites between the families. The analysis shows that domains vary strongly in the conservation of interacting sites and provides useful information for the prediction of interacting sites based on homologous protein sequences.
Results and discussion
Our analysis of interacting sites is based on conserved protein domains found in the structures of the protein data bank (PDB). Family sequence alignments of protein domains were retrieved from the SMART database and used to scan the protein sequences of the PDB with domain-specific Hidden Markov Models (HMMs). Wherever a domain was identified, all interactions between an amino acid belonging to this domain and any ligand were extracted and the position of the interacting amino acid was transferred onto the HMM consensus.
Table 1 lists the number of sequences with ligand interactions extracted from PDB and the number of interacting sites found in these sequences [see Additional file 1]. The HMM consensus was used as a reference sequence to be able to compare the location of interactions among different sequences of a domain family. The positions in the consensus sequence correspond to positions shared by most sequences of the domain family (match states) and most domains interact only in regions for which an HMM match state exists. In contrast, if the interacting position corresponds to an insert state in the alignment to the HMM consensus, the information of interaction could not be transferred to the domain consensus. There are several domains, which have a comparatively large number of interacting sites located in loop regions. These amino acid sites are only present in subfamilies of the domain and are lacking an HMM match state, so that mapping the information of interaction onto the HMM was not possible. An overview of the number of interactions at HMM insert states for all investigated protein domains is given in table 1 [see Additional file 1]. In domains of the immunological system, e.g. the immunoglobulin (IG) and immunoglobulin V type (IGv) domains, up to 50% of the interacting sites are located in these regions. Strikingly, other extra-cellular domains (fibronectin type 3 (FN3), C-type lectin (CLECT), leucine rich repeat C-terminal domain (LRRCT)) also present a large number of interacting sites in loop regions. These domains all have a common involvement in highly specific recognition processes, where any restrictions in the choice of amino acid or structural constraints would be disadvantageous for the function of the domain. The loop regions without any match states exactly fulfil this condition and therefore biochemical properties of the domain can be fine-tuned to complement the appearance of the ligand. The increased use of variable loop regions for functional sites seems to be a common characteristic of extracellular highly ligand-specific domains.
Conservation of location of interacting sites
In order to compare the use of specific interacting positions within a domain family we introduced a score (ConsInt) to measure the conservation of interaction at a given site in the family alignment. The score reflects the importance of an amino acid site in domain interactions and ranges between 0 and 1. Scores greater than 0 mean that at least two non-identical sequences interact at this position, and a score of 1 arises if all sequences interact at this position. For the comparison of the interaction scores and amino acid conservation, we only calculated the score for sites, where the interaction is achieved by atoms of the amino acid side chain. These scores are generally smaller than those calculated for side chain and backbone interactions, because side chain interactions are only part of the total interactions and many sites are very flexible in the contribution of atoms to the interaction interface. Corresponding positions in homologous sequences can interact with side chain atoms in one sequence and exclusively with backbone atoms in the next sequence.
The distribution of interaction scores is shown in figure 1. It is remarkable that there are only a few positions in all of the investigated domains with the maximum interaction score. Smaller interaction scores can emerge from the use of ligands with distinct functions in the PDB data, for example, if a domain is in complex with its native substrate or another time with a regulatory protein that binds to a remote part of the domain. Absence of the substrate in the latter case leads to lower interaction scores for important functional sites: even medium interaction scores can indicate significant interactions. A large proportion of amino acid sites have very low interaction scores, presenting sites that are only occasionally involved in interactions or that are specific to a subgroup. Subgroups tend to use the same functional positions, while the functional sites can vary between subgroups. This behaviour is also reflected in the distance trees of the domains. The carbohydrate binding RICIN domain in figure 2 exemplifies the divergence of functional sites within a domain family. By inspecting the arrangement of functional sites in the different RICIN sequences, the RICIN family can be divided into three main subgroups of distinct functions. While one subgroup possesses two carbohydrate binding sites and a peptide binding region (group II in figure 2), one subgroup is limited to the N-terminal carbohydrate binding region (group I) and one to the C-terminal carbohydrate binding region (group III). Absence of a carbohydrate binding site in the first or third group is unlikely to be an artefact of the crystallization process, since the domains were crystallized in complex with an adjacent sugar bound domain. The classification observed by the interaction profile is also reflected in the cladogram given to the left of the interaction profile. Interestingly, the subgroups that preserved only one of the carbohydrate binding sites originate from proteins with tandem RICIN domains, which arose from gene duplications [25], so that the proteins are again provided with two carbohydrate binding sites. A single RICIN domain can be divided into three subunits of approximately 40 amino acids in lengths that have evolved from an ancient galactose binding peptide [26]. These subunits represent the differently specialized binding sites in the domain family. In proteins carrying two RICIN domains, only the first subunit of the first RICIN domain and the last subunit of the second RICIN domain preserved their ability to bind carbohydrates, while one subunit has specialized on binding peptides (subgroup II in figure 2). The RICIN domain is a good example how homologous sequences belonging to different subfamilies of a domain specialized on binding different ligands and thus, on various functions. The functional sites are no longer conserved throughout the whole family and they could not be inferred from homologous sequences. However, functional sites can be predicted, if the information is taken from the most closely related sequences.
Figure 1 Distribution of interaction scores. The interaction score reflects the importance of a functional sites in establishing an interaction. Surprisingly, only few interacting sites are absolutely conserved in their location within the whole protein family and characterized by high interaction scores. The majority of interacting sites feature small interaction scores. This shows that these sites are only used by a few sequences of the domain family for establishing an interaction, which can also be caused by the different nature of ligands.
Figure 2 Interaction profile of the RICIN domain. Alignment of positions corresponding to an HMM match state. Sites interacting with saccharides are indicated in blue, peptide interactions in orange, and sites interacting with both ligands, saccharides and peptides, are indicated in purple. Light colours represent backbone interactions, darker colours involve side chain atoms. The amino acid conservation is visualized by green bars below the alignment. Sugar binding sites described in the literature are indicated by red arrows above the alignment [41]. Several positions (1, 3, 4, 22, 42, 58, 88, 90, 122) are located in the vicinity of a glycosylation site, but do not specifically interact with saccharides. The unrooted tree reflects the classification into three main subgroups with different interaction sites. Group II harbours two sugar-binding sites, group I and III originate from tandem RICIN domains, in which group I preserved the N-terminal sugar-binding site and group III the carboxy-terminal binding site. PDB identifiers from top to bottom: 1PC8 (B: 5–131), 1TFM (B: 5–131), 2MLL (B: 5–131), 1CE7 (B: 5–131), 1ONK (B: 9–135), 1PUM (B: 9–135), 1M2T (B: 257–383), 1OQL (B:13–139), 1ABR (B: 13–139), 2AAI (B: 8–134), 1HWO (B: 10–135), 1HWP (B: 10–135), 1HWN (B:10–135), 1HWM (B:3–266), 1V6U (A: 312–436), 1ISW (A:312–436), 1ISV (A:312–436), 1ITO (A:312–436), 1V6W (A: 312–436), 1V6X (A: 312–436), 1XYF (A:312–436), 1ISY (A: 312–436), 1ISZ (A:312–436), 1V6V (A:312–436), 1ISX (A:312–436), 1KNM (A:7–131), 1KNL (A:9–133), 1BFM1MC9(A:9–133), 1QXM (A: 29–157), 1PUM (B: 140–262), 1M2T (B: 390–510), 1ONK (B: 140–262), 1OQL (B: 140–262), 1PC8 (B: 136–254), 1TFM (B: 136–254), 2MLL (B: 136–254), 1CE7 (B:136–254), 2AAI (B: 138–261), 1ABR (B: 143–266), 1HWO (B: 138–262), 1HWP (B: 138–262), 1HWM (B: 138–262), 1HWN (B: 139–263), 1FWU (A: 3–123), 1DQG (A: 4–124), 1DQO (A: 4–124), 1FWV (A: 3–123)
Alternating locations of functional sites are especially prominent in DNA binding domains, like the C2H2-zinc finder and homeodomains [22]. Another example is the high mobility group (HMG) domain, which is shown in complex with its target DNA in figure 3. Here, an amino acid side chain pointing into the DNA helix recognizes the DNA bases of the target sequence. This key position is located in the loop connecting the first and second alpha helix of the domain. The contact is carried out by a serine residue in figure 3a (pdb id: 1hry[27]). In contrast, a phenylalanine establishes the contact in figure 3b (pdb id: 1ckt[28]) and the serine residue corresponding to the structure shown in 3a is pointing away from the DNA helix. In the sequence alignment, these two key positions are located adjacent to each other. Many other domains show this variability in the location of interacting amino acid residues and profit from the flexibility to fine-tune substrate specific binding sites based on the same structurally conserved protein fold.
Figure 3 Variable location of interacting amino acid residues in the HMG domain. Sequence specific interaction by the high mobility group (SMART: HMG) domain (green) is achieved by an amino acid side chain (pink) pointing into the DNA double helix (blue). The interaction is achieved by a phenylalanine in figure 3a [28] or by a serine residue in 3b [27]. The sequence alignment (figure 3c) reveals that these two interacting residues are not located at corresponding position.
Amino acid conservation of interacting sites
Having observed great differences in the location of interacting sites within conserved domains, the question arises how these sites behave with regard to their amino acid conservation. It is generally believed that interacting sites are more highly conserved than non-interacting solvent-assessable sites. However, about one third of all interactions in our analysis were achieved by backbone atoms only, so that the kind of amino acid has no direct effect on the interaction. For the remaining two thirds, which account for side chain interactions, we calculated a score for the conservation of amino acid similar to the interaction score ConsInt. We next compared the relative frequency distribution of the amino acid conservation score of interacting sites with the one of non-interacting sites, which are composed of amino acids located in the core of the domain as well as non-interacting surface residues. The distribution of interacting sites is slightly shifted to higher amino acid conservation scores (figure 4). This shift is clearly visible and is statistically significant (Kolmogorov-Smirnoff test: p-value < 2, 2 ×10-16), although the data have not been restricted to surface residues. Our finding that interacting sites are more highly conserved, on average, is in accordance with the results of other groups who reported that domain-domain interfaces are better conserved than the rest of the surface residues [29].
Figure 4 Amino acid conservation of interacting and non-interacting sites. Non-interacting sites (yellow) are slightly more highly conserved than interacting sites (red) as shown by the shift to higher amino acid conservation of interacting sites.
Strikingly, there are many interacting sites with very low amino acid conservation scores. A possible explanation could be that very few residues of the interaction interface are important for a stable interaction and conserved in their amino acid. It has been shown that only a subset of interface residues contribute a crucial part to the binding energy, while residues around these so-called hot spots are less conserved [30,31]. Another explanation could be the increased specificity for various ligands. The data set contains orthologous sequences, which might be conserved in function and substrate specificity and paralogous sequences, which might have accumulated mutations throughout evolution and adopted new substrate specificity. The zinc finger domain, for example, interacts mainly through an aromatic residue with the nucleic acid, but specificity is provided by a nearby position, which can vary in its amino acid and is also interacting with the bases of the nucleic acid (figure 5). Hence, the variety of amino acids at functional sites could be advantageous to recognize numerous different ligands by the same domain family.
Figure 5 Substrate specific interaction by varying the type of amino acid. Substrate specificity in the zinc finger domain (SMART: ZnF_C2H2) is ensured by various amino acids that interact with the bases of the DNA. The protein domain is highlighted in green, the DNA chain in orange and the zinc atom in red.
Correlation of interaction and amino acid conservation
In order to test whether often-used interacting sites coincide with highly conserved sites we plotted the interaction score against the amino acid conservation score (figure 6). Since our dataset includes various ligands, which might have different preferences in terms of the amino acid conservation or interaction, we divided the data by the type of ligand. Groups analyzed correspond to peptides, nucleotides, ions and all ligands, including those, which could not be classified into one of the prior groups. Statistical analysis detected a significant positive correlation between the interaction score and its amino acid conservation in all observed groups (see figure 6 for details). The trend to higher interaction scores with increasing amino acid conservation is clearly visible if the data are divided into groups and the median calculated for each group. The median values of ion binding sites are shifted to higher amino acid conservation scores compared to the other three ligand groups, indicating that ion binding sites are more highly conserved than sites binding peptides, nucleotides or other small molecules. The highest median interaction scores are found in the group of nucleotide ligands, consequently nucleotide binding sites preferentially use the equivalent positions in homologous sequences. Interestingly, sites interacting with peptides are not less conserved in their amino acids but are more flexible in the location of interaction compared to nucleotides. An unexpected finding is the great variance of amino acid conservation scores for high interaction scores, especially in the nucleotide and peptide group. This indicates that these interacting sites are very flexible in the type of amino acid and specialized to complement the ligand and to increase specificity of the protein-ligand interaction.
Figure 6 Correlation of interaction scores and amino acid conservation. For better visualization of the correlation, the data was divided into five groups corresponding to the 0–20% quantile, 20–40% quantile, etc. of the amino acid conservation scores and then the median interaction score and median amino acid conservation score was calculated for each group and plotted with red dots. The correlation coefficient, p-value and population are indicated for each ligand above the graphs. The correlation coefficient was calculated according to Pearson's method under the null-hypothesis of no correlation (c = 0).
Conclusion
Our analysis reveals that functional sites can be highly variable in their amino acid conservation and very flexible in using various locations in the protein domain. The properties of functional sites are dependent on the protein family and can vary from highly conserved, as observed in enzymes involved in DNA replication, to protein families that are highly variable with various amino acids at various locations, as for example immunoglobulins or carbohydrate-binding domains. Similar results were obtained by other groups. Pachenko and co-workers analyzed 86 domains from the CDD database and report that functional sites of homologous sequences can greatly differ in their physicochemical properties and their location in the three-dimensional structure [32]. Variability in functional sites was also described by Devos and Valencia. By comparing the conservation of binding sites in structural alignments, they found high conservation in diverged sequences contrasting highly similar sequences with different interacting sites [33]. Our findings present valuable information for the improvement of methods to predict functional sites. In most of these methods, the prediction is based on a set of homologous sequences. This approach results only in reliable predictions if the investigated protein family is conserved in most of the functional sites. Approaches to improve the accuracy of prediction have been made recently by using orthologous proteins with presumably the same ligand specificity [34] or by sub-typing protein families [35]. With the growing number of sequences within protein families, trends that consider variability of functional sites and use subgrouping aided by experimental information might become widely accepted in the future and promise to be successful in the prediction of functional sites.
Methods
Data set
The analysis is based on the October version of PDB (27969 structures). All protein sequences extracted from PDB files were scanned against all SMART domains (667) using hmmsearch from the HMMER package (version 2.3.2) [36]. Profile HMMs were retrieved from the SMART family alignments and score thresholds were used as assigned by SMART for each individual domain [37]. The search resulted in 8747 protein structures containing at least one of 480 SMART domains. For those structures containing a protein-ligand interaction, we calculated the distance for each atom of all protein compounds to each atom of all ligands. We considered amino acids as interacting if the distance of any atom of the amino acid to any atom of the ligand was smaller than 4 Angstrom, which is a very conservative threshold and is in the range of the two oxygen atoms in a hydrogen bond. Interactions to water molecules were neglected, as well as interactions between identical chains, because homodimers can present artefacts arising from the crystallization process. We are aware of losing information about naturally occurring homodimers. If more than one model of the protein structure exists, for example in the case of NMR data, all models were taken into account and a position was treated as interacting if more than 50% of all models harbour an interaction at this position. For each domain family, a multiple sequence alignment was generated from the sequences identified in the PDB scan for SMART domains. The alignments were created with hmmalign (HMMER package), according to the SMART profile HMMs. Distance trees were created with PROTDIST and FITCH, both from the Phylip package [38].
A problem with using the protein data bank is the overrepresentation of some proteins, while others are completely absent. To deal with the biased nature of the database, we used sequence weights, correlated to the evolutionary distance between the sequences. The distance is small for similar sequences, so that these proteins are weighted with a negligible small factor. Although many proteins are redundant in our dataset it is advantageous to consider all proteins because they can be bound to different ligands or the complex crystallized under different conditions. The most important interacting sites should clearly stand out in the large scale analysis.
Ligands interacting with protein domains were, wherever possible, classified into groups of peptides, nucleotides or ions. The group of ions was restricted to small ions and excluded typical buffer anions. The groups of peptides and nucleotides also included modified molecules that are functionally alike. We also analysed all ligands together, including carbohydrates, buffer ions and other small molecules.
Calculation of scores
For each alignment position consistent with the HMM consensus sequence, we calculated a position-specific interaction conservation score (ConsInt) to describe how well a position is conserved in ligand interactions within the domain family.
where N is the number of sequences in the alignment, Dist(seqa, seqb) is the phylogenetic distance between sequence a and b obtained from the phylogenetic tree and Inti takes a value of 1 if sequence a and b both interact at position i, and 0 otherwise. The score ranges from 0 to 1 and is 1 if all sequences interact at the position of interest. To obtain a score greater than 0 for a certain position at least two non-identical sequences have to interact at this position. In this way, interactions to non-physiological ligands like artificially synthesized peptides should be lost, and important interacting positions should have noticeably higher scores. The score takes into account the phylogenetic distances between the sequences so that highly similar sequences are weighted more weakly, in contrast to interacting sites in divergent sequences, which are weighted more strongly.
Similar to the interaction score, a position specific amino acid conservation score (ConsAA) was calculated.
Here, Substi(a, b) measures the similarity between the amino acids at position i in sequences a and b. It is based on the VTML 160 substitution matrix [39]. The conservation score was normalized between 0 (low aa conservation) and 1 (high aa conservation) ranging from the lowest to the highest score of the amino acid substitution matrix. It is important to note that our amino acid conservation scores do not describe protein families as found in the SMART database, but only the data set used in our analysis, so that the amino acid conservation score can be compared with the interaction conservation score for each position in the family alignment.
The software suite R was used for the statistical analyses [40]. The concentration of data points on vertical lines found for higher amino acid conservation scores in the scatter plots of figure 6 are an effect of the discrete values of the amino acid substitution matrix and the few substitutions at conserved sites.
Authors' contributions
BP performed the analysis and prepared a draft of the manuscript. RRC contributed to the generation of the data set and JS supervised the project. All authors were involved in the design of the project and the refinement of the manuscript.
Supplementary Material
Additional file 1
Table 1 – Interacting sites in protein domains For each domain, the number of domains with ligand interactions extracted from PDB (# PDB), the number of interacting sites corresponding to HMM match states (int sites), the number of interacting sites corresponding to HMM insert states (loop sites), the percentage of insert state interacting sites from the total number of interacting sites (% loop) and the number of interacting sites in the domain consensus (domain int sites) are given. The last column considers conserved positions in the domain consensus sequence, while all other columns count sites in sequences belonging to the domain family. The number of conserved interacting sites (domain int sites) can be 0 despite plenty of interacting sites in family sequences if the site-specific interaction score does not yield a positive value due to identical sequences. Abbreviations of domain names are according to SMART [24].
Click here for file
Acknowledgements
We would like to thank Ivica Letunic for scanning PDB for SMART domains and Wolfgang Huber for his assistance on the statistical analysis. This work was supported by the DFG grant BO-1099/5-2 (BP).
==== Refs
Lichtarge O Sowa ME Evolutionary predictions of binding surfaces and interactions Curr Opin Struct Biol 2002 12 21 27 11839485 10.1016/S0959-440X(02)00284-1
Campbell SJ Gold ND Jackson RM Westhead DR Ligand binding: functional site location, similarity and docking Curr Opin Struct Biol 2003 13 389 395 12831892 10.1016/S0959-440X(03)00075-7
Jones S Thornton JM Searching for functional sites in protein structures Curr Opin Chem Biol 2004 8 3 7 15036149 10.1016/j.cbpa.2003.11.001
Bhinge A Chakrabarti P Uthanumallian K Bajaj K Chakraborty K Varadarajan R Accurate detection of protein:ligand binding sites using molecular dynamics simulations Structure (Camb) 2004 12 1989 1999 15530363 10.1016/j.str.2004.09.005
Laurie AT Jackson RM Q-SiteFinder: an energy-based method for the prediction of protein-ligand binding sites Bioinformatics 2005 21 1908 1916 15701681 10.1093/bioinformatics/bti315
Amitai G Shemesh A Sitbon E Shklar M Netanely D Venger I Pietrokovski S Network analysis of protein structures identifies functional residues J Mol Biol 2004 344 1135 1146 15544817 10.1016/j.jmb.2004.10.055
Stark A Shkumatov A Russell RB Finding functional sites in structural genomics proteins Structure (Camb) 2004 12 1405 1412 15296734 10.1016/j.str.2004.05.012
Kinoshita K Nakamura H Identification of the ligand binding sites on the molecular surface of proteins Protein Sci 2005 14 711 718 15689509 10.1110/ps.041080105
Lichtarge O Bourne HR Cohen FE An evolutionary trace method defines binding surfaces common to protein families J Mol Biol 1996 257 342 358 8609628 10.1006/jmbi.1996.0167
Madabushi S Yao H Marsh M Kristensen DM Philippi A Sowa ME Lichtarge O Structural clusters of evolutionary trace residues are statistically significant and common in proteins J Mol Biol 2002 316 139 154 11829509 10.1006/jmbi.2001.5327
Landgraf R Fischer D Eisenberg D Analysis of heregulin symmetry by weighted evolutionary tracing Protein Eng 1999 12 943 951 10585499 10.1093/protein/12.11.943
Armon A Graur D Ben-Tal N ConSurf: an algorithmic tool for the identification of functional regions in proteins by surface mapping of phylogenetic information J Mol Biol 2001 307 447 463 11243830 10.1006/jmbi.2000.4474
Pupko T Bell RE Mayrose I Glaser F Ben-Tal N Rate4Site: an algorithmic tool for the identification of functional regions in proteins by surface mapping of evolutionary determinants within their homologues Bioinformatics 2002 18 Suppl 1 S71 7 12169533
Landgraf R Xenarios I Eisenberg D Three-dimensional cluster analysis identifies interfaces and functional residue clusters in proteins J Mol Biol 2001 307 1487 1502 11292355 10.1006/jmbi.2001.4540
Aloy P Ceulemans H Stark A Russell RB The relationship between sequence and interaction divergence in proteins J Mol Biol 2003 332 989 998 14499603 10.1016/j.jmb.2003.07.006
Rekha N Machado SM Narayanan C Krupa A Srinivasan N Interaction interfaces of protein domains are not topologically equivalent across families within superfamilies: Implications for metabolic and signaling pathways Proteins 2005 58 339 353 15562516 10.1002/prot.20319
Jones S Thornton JM Analysis of protein-protein interaction sites using surface patches J Mol Biol 1997 272 121 132 9299342 10.1006/jmbi.1997.1234
Lijnzaad P Argos P Hydrophobic patches on protein subunit interfaces: characteristics and prediction Proteins 1997 28 333 343 9223180 10.1002/(SICI)1097-0134(199707)28:3<333::AID-PROT4>3.0.CO;2-D
Larsen TA Olson AJ Goodsell DS Morphology of protein-protein interfaces Structure 1998 6 421 427 9562553 10.1016/S0969-2126(98)00044-6
Lo Conte L Chothia C Janin J The atomic structure of protein-protein recognition sites J Mol Biol 1999 285 2177 2198 9925793 10.1006/jmbi.1998.2439
Valdar WS Thornton JM Protein-protein interfaces: analysis of amino acid conservation in homodimers Proteins 2001 42 108 124 11093265 10.1002/1097-0134(20010101)42:1<108::AID-PROT110>3.0.CO;2-O
Luscombe NM Thornton JM Protein-DNA interactions: amino acid conservation and the effects of mutations on binding specificity J Mol Biol 2002 320 991 1009 12126620 10.1016/S0022-2836(02)00571-5
Deshpande N Addess KJ Bluhm WF Merino-Ott JC Townsend-Merino W Zhang Q Knezevich C Xie L Chen L Feng Z Green RK Flippen-Anderson JL Westbrook J Berman HM Bourne PE The RCSB Protein Data Bank: a redesigned query system and relational database based on the mmCIF schema Nucleic Acids Res 2005 33 Database Issue D233 7 15608185
Letunic I Copley RR Schmidt S Ciccarelli FD Doerks T Schultz J Ponting CP Bork P SMART 4.0: towards genomic data integration Nucleic Acids Res 2004 32 D142 D144 14681379 10.1093/nar/gkh088
Villafranca JE Robertus JD Ricin B chain is a product of gene duplication J Biol Chem 1981 256 554 556 7451458
Rutenber E Ready M Robertus JD Structure and evolution of ricin B chain Nature 1987 326 624 626 3561502 10.1038/326624a0
Werner MH Huth JR Gronenborn AM Clore GM Molecular basis of human 46X,Y sex reversal revealed from the three-dimensional solution structure of the human SRY-DNA complex Cell 1995 81 705 714 7774012 10.1016/0092-8674(95)90532-4
Ohndorf UM Rould MA He Q Pabo CO Lippard SJ Basis for recognition of cisplatin-modified DNA by high-mobility-group proteins Nature 1999 399 708 712 10385126 10.1038/21460
Littler SJ Hubbard SJ Conservation of orientation and sequence in protein domain--domain interactions J Mol Biol 2005 345 1265 1279 15644220 10.1016/j.jmb.2004.11.011
Clackson T Wells JA A hot spot of binding energy in a hormone-receptor interface Science 1995 267 383 386 7529940
Keskin O Ma B Nussinov R Hot regions in protein--protein interactions: the organization and contribution of structurally conserved hot spot residues J Mol Biol 2005 345 1281 1294 15644221 10.1016/j.jmb.2004.10.077
Panchenko AR Kondrashov F Bryant S Prediction of functional sites by analysis of sequence and structure conservation Protein Sci 2004 13 884 892 15010543 10.1110/ps.03465504
Devos D Valencia A Practical limits of function prediction Proteins 2000 41 98 107 10944397 10.1002/1097-0134(20001001)41:1<98::AID-PROT120>3.0.CO;2-S
Mirny LA Gelfand MS Using orthologous and paralogous proteins to identify specificity-determining residues in bacterial transcription factors J Mol Biol 2002 321 7 20 12139929 10.1016/S0022-2836(02)00587-9
Hannenhalli SS Russell RB Analysis and prediction of functional sub-types from protein sequence alignments J Mol Biol 2000 303 61 76 11021970 10.1006/jmbi.2000.4036
Eddy S HMMER: sequence analysis using profile hidden Markov models [http://hmmerwustledu]
Schultz J Milpetz F Bork P Ponting CP SMART, a simple modular architecture research tool: identification of signaling domains Proc Natl Acad Sci U S A 1998 95 5857 5864 9600884 10.1073/pnas.95.11.5857
Felsenstein J PHYLIP - Phylogeny Inference Package (Version 3.2) Cladistics 1989 5 164 166
Muller T Spang R Vingron M Estimating amino acid substitution models: a comparison of Dayhoff's estimator, the resolvent approach and a maximum likelihood method Mol Biol Evol 2002 19 8 13 11752185
R Development Core Team R: A language and environment for statistical computing 2004 Vienna, Austria, R Foundation for Statistical Computing
Pascal JM Day PJ Monzingo AF Ernst SR Robertus JD Iglesias R Perez Y Ferreras JM Citores L Girbes T 2.8-A crystal structure of a nontoxic type-II ribosome-inactivating protein, ebulin l Proteins 2001 43 319 326 11288182 10.1002/prot.1043
|
16122386
|
PMC1215474
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Aug 25; 6:210
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-210
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2181613732110.1186/1471-2105-6-218DatabaseTmaDB: a repository for tissue microarray data Sharma-Oates Archana [email protected] Philip [email protected] David R [email protected] Academic Unit of Pathology, University of Leeds, Leeds, LS1 3EX, UK2 School of Biochemistry & Microbiology, University of Leeds, Leeds, LS2 9JT, UK2005 1 9 2005 6 218 218 19 4 2005 1 9 2005 Copyright © 2005 Sharma-Oates et al; licensee BioMed Central Ltd.2005Sharma-Oates et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Tissue microarray (TMA) technology has been developed to facilitate large, genome-scale molecular pathology studies. This technique provides a high-throughput method for analyzing a large cohort of clinical specimens in a single experiment thereby permitting the parallel analysis of molecular alterations (at the DNA, RNA, or protein level) in thousands of tissue specimens. As a vast quantity of data can be generated in a single TMA experiment a systematic approach is required for the storage and analysis of such data.
Description
To analyse TMA output a relational database (known as TmaDB) has been developed to collate all aspects of information relating to TMAs. These data include the TMA construction protocol, experimental protocol and results from the various immunocytological and histochemical staining experiments including the scanned images for each of the TMA cores. Furthermore the database contains pathological information associated with each of the specimens on the TMA slide, the location of the various TMAs and the individual specimen blocks (from which cores were taken) in the laboratory and their current status i.e. if they can be sectioned into further slides or if they are exhausted. TmaDB has been designed to incorporate and extend many of the published common data elements and the XML format for TMA experiments and is therefore compatible with the TMA data exchange specifications developed by the Association for Pathology Informatics community. Finally the design of the database is made flexible such that TMA experiments from several types of cancer can be stored in a single database, which incorporates the national minimum data set required for pathology reports supported by the Royal College of Pathologists (UK).
Conclusion
TmaDB will provide a comprehensive repository for TMA data such that a large number of results from the numerous immunostaining experiments can be efficiently compared for each of the TMA cores. This will allow a systematic, large-scale comparison of tumour samples to facilitate the identification of gene products of clinical importance such as therapeutic or prognostic markers. In addition this work will contribute to the establishment of a standard for reporting TMA data analogous to MIAME in the description of microarray data.
==== Body
Background
Tissue microarray technology provides a high-throughput method for large-scale immunohistochemical analyses under uniform experimental conditions. The TMA technology was developed to assist genome-scale molecular pathology studies and facilitate the analysis of molecular alterations in thousands of tissue specimens in parallel at the DNA, RNA, or protein level [1]. TMAs are constructed by drilling 0.6 mm cylindrical cores at the site of interest from various paraffin-embedded specimen blocks, referred to as donor blocks, and re-embedding them into an empty paraffin block, referred to as recipient block (Figure 1).
Figure 1 Schematic representation of the construction of a TMA block. TMAs are assembled by extracting cylindrical cores (using a 0.6 mm needle) from specific locations in the paraffin-embedded tissue blocks "donor" and re-embedded in an empty paraffin block "recipient" [2]. The TMA block is then sectioned off onto slides.
On a single TMA block up to 1000 specimens can be arrayed representing a number of different pathologies and tissue types both, normal and tumour. The recipient block is then sectioned off into a number of slides (from 50 – 400) [3]. These slides can then be screened at the DNA, RNA and protein level by fluorescence in situ hybridisation (FISH), in situ hybridisation (ISH), and immunohistochemistry (IHC), respectively.
There is a large quantity of data associated with a TMA experiment. These include images, quantified experimental results, patient demographics and pathology reports of each specimen [4]. The vast quantity of data generated by TMA experiments and the different types of information associated with each experiment has led to the need for a central data repository. Thus far there have only been three papers published on approaches for the archiving and analysis of TMA data and all three have limitations [5,6]. The relational database management system of Manley et al. [5] associates pathology data with immunohistochemical staining results but is restricted to just one type of cancer in this case being prostate cancer and cannot be adapted to include pathology data from different types of cancer. Liu et al. [6] on the other hand focus on the archiving and the analysis of TMA staining results but offer no means to relate the pathology information. Finally, the TAD database developed by Coombes et al. [7] is a tool focusing on enabling pathologists to solely score cores on TMA slides generated from a single TMA block. It is therefore considered too restrictive as a database for archiving TMA data.
Currently there are no central databases that facilitate the storage and analysis of TMA data that can also accommodate pathological information from different types of cancer. This has led to the design and implementation of TmaDB a relational database that archives all aspects of TMA data into a single database that can be accessed from remote locations to both update and retrieve data. The database design is flexible, thus incorporating pathological information from different types of cancer into one central database. The database design is compliant with recently recommended data exchange specifications [4,8].
Construction and contents
The relational database is implemented in MySQL server version 14.5. The database schema was designed by Entity-Relationship (ER) modelling (Figure 2) and can be found on the TmaDB Web site.
Figure 2 Entity relation (ER) diagram of TmaDB.
The "block" table is the key table in the database to which all other tables are related (directly or indirectly). This table contains the specimen identifier and other data relating to a particular specimen. The "path_report" table is another important table that stores some generic information about a patients' medical history regardless of the specific cancer diagnosis. The specific information associated with a particular type of cancer is stored in the disease-specific tables. The disease specific information is obtained from the national minimum data set required for the reporting of cancer pathological investigations. This minimum data set is a standard supported by the Royal College of Pathologists. The "tma" table stores the information about the design and the construction of the TMA block (also referred as the recipient block), which can then be sectioned onto several slides (Figure 1). Each of these slides is stained with a specific antibody or a histochemical stain. The staining protocol, the slide image and other details of the experiment are stored in the "tma_experiment" table. Each of the cores on the TMA slide has a stain result and a scanned image associated with it that is stored in the "core_expt_results" table. Other information regarding the quality of the core and the tissue diagnosis is stored in the "core_info" table. The "whole_tissue_section" table stores the staining protocol as well as the results and the images of the stain generated from the traditional whole tissue section (if available) for the clinical specimens used in the TMA.
Content
The data submitted to the database comes from a number of different sources. One of the data sources is the pathology report for each clinical specimen, which is sent with the specimen either directly from the hospital, or provided as part of the clinical trials. These reports are usually provided as paper copies in the form of a letter. If sent as a paper copy, they are digitised by an individual who then extracts the information that would be required for the national minimum data set for each of the specific types of cancer (Figure 3). The national minimum data set for reporting specific types of cancer is supported by the Royal College of Pathologists and the forms can be downloaded from the Royal College of Pathologists website . This information can be entered directly on the Web to be assimilated into the database or saved as a tab delimited text file that can be uploaded from the Web site. Most of the information obtained is stored in the "path_report", "patient" and "block" tables (Figure 2). It should be noted that patients' names and addresses are not stored in the database.
Figure 3 TmaDB homepage.
Information on the storage location in the laboratory of each block and whether the block is still in use is stored in the block table along with when this information was last updated. This enables a block to be found easily in the event it is required for future experiments. This information is uploaded from the tab delimited text file containing details of TMA construct design. The information recorded in this file is shown on the Web site, and can be downloaded and used as a template for recording data alongside experimental work.
The pathological details of each of the cores together with additional information such as the location on the original block from which the core was obtained and the quality of the core are uploaded from a tab delimited text file and stored in the core location table. The TMA slide immunohistological staining protocol and the results are also recorded in a tab delimited text file, an example of which is shown as a template on the Web site.
Each TMA slide is scanned using software that provides a very high-resolution image of the TMA slide enabling users to "zoom in" and "zoom out" on the image over the web. TMA slide images are saved on an image server. They can be divided into a grid such that each core image can be saved as a separate image. The TmaDB has a link to the image server for each of the cores and results from the analysis of the images are stored in the "core_expt_result" table (Figure 2). Only the filenames of the images are stored by TmaDB and the actual images are stored on the image server.
Currently there are 18 TMA experiments stored in the database although this number is set to rise. The number of people whose clinical specimens have been analysed on TMA is 1095 (April 2005). Most of this data has been generated as part of clinical trials.
Utility and discussion
The MySQL relational database is interfaced with the World Wide Web making the database user-friendly as well as enabling access from remote locations for multiple users at any time. The web site also enables users to submit and retrieve data from the database. The homepage provides a brief background on TMA technology and outlines the advantages and disadvantages of the technique, it also provides a menu for easy navigation of the web site. A help page provides detailed explanations for each of the menu links. A brief description is given below:
The search page allows users to retrieve data from the database by searching with a specific keyword. Specific searches can be performed for cases where the "specimen_id" is already known by the user (Figure 4).
Figure 4 An example query of the database.
The "Display all records" link connects to a page containing all the tables in the database. Clicking on individual table names fetches all the data contained in the particular table. The "Database design" link, displays the ER diagram (Figure 2). The "MySQL search" link is an html form that enables the user to retrieve data from the database with MySQL "select" statements. For security reasons, the user only has permission to search the database but not write to the database. However badly constructed MySQL "select" statements with incorrect syntax will not be executed and the MySQL error message will be displayed on screen. In addition, the provision of SQL select facilities could leave the database vulnerable to badly constructed queries that would be damaging to performance but not data, although this has not been a problem to date. The other four links on the menu are for inputting data into the database. The "Submit TMA design" link allows the user to upload a tab delimited text file which is then saved into a directory on the server side and an email is sent automatically to the curator who then checks the file to make sure there are no errors and then the information is loaded into the database automatically. The data is loaded sequentially into the database beginning with the TMA design ("Submit TMA design") then the specific details of each of the cores on the TMA ("Submit core path data") followed by experimental protocol and results ("Submit experimental protocol and results") and finally the pathological data ("Submit block path report"). The file formats for each of the files to be uploaded are shown on the Web site. The final link on the menu connects to the image server where all images for all TMAs are archived. All files can also be uploaded in XML file format (the schema for this is provided on the Web site). In addition the output from all queries to the database can be displayed or downloaded as an XML file. The XML file format is compatible with the data exchange specification developed by Berman et al. [4] in that we have adopted their extensibility mechanism and therefore our instance documents are still valid under their exchange specifications. The basic XML structure developed by Berman et al. [4] has been extended with the aim of establishing a community standard that is similar to MAGE-ML for microarray databases [9].
The purpose of a central cancer TMA database is to provide a single repository for laboratory experimentalists to store TMA experiments. Storage of all TMA experiments will enable researchers to make correlations between the intensity of staining of a specific antibody and the stage of cancer, or between different types of cancers. The curation of a database will also encourage researchers to report results in a standard format thereby enabling comparisons between experiments performed by different individuals. An additional advantage of the database is to be able to determine where a specimen block is stored or the person/institution in possession of a specimen block of interest for further experiments. Furthermore it is anticipated that the database will also be used as a learning tool for pathologists.
Further developments of the database include the addition of other information from different experiments (such as cDNA microarray) on the same clinical specimen used in the TMA experiment and allowing users to carry out complex queries (see below).
An example of a complex query to the database:
Find all cases where specimens are stained with antibody "p53" AND type of cancer is "colon" AND pT stage is "3".
Conclusion
The TmaDB provides a central repository for archiving all aspects of TMA data. The relational database includes the vast majority of the published Common Data Elements for a TMA experiment. This will therefore enable efficient data exchange as well as contributing to the establishment of a standard for reporting TMA data analogous to MIAME in the description of microarray data [10]. The database design is adaptable such that pathological data from several different types of cancer can be included in one database.
It is anticipated that in addition to providing a resource for archiving and querying TMA data TmaDB will also enable large-scale analyses of TMA data.
Availability and requirements
Tmadb web site: .
The database and the software to upload files and query the database are available under the GPL as a package for installation on a local server [see Additional files].
Authors' contributions
ASO participated in the design, implementation of the study and drafted the manuscript. PQ conceived of the study. DRW participated in the design and coordination of the study. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
This compressed (gz) file contains two directories tmadb_bmc_html and tmadb_bmc and two files, create_tmadb.txt and a README file which can be extracted using gunzip software. The create_tmadb.txt file contains all the MySQL create commands for creating tables contained in the database. The README file provides instructions to help the user install the software. The tmadb_bmc_html directory contains html, xml and text files required for interfacing with the cgi programs. The tmadb_bmc directory contains ten files, nine files with the extension cgi and a file named config.pl. config.pl Contains variables that require modification during installation. colo_form_input.cgi Program to upload colorectal pathology information from the Web form. colo_path_input.cgi Program to upload colorectal pathology information from the Web. core_path.cgi Program to upload specific information relating to each core from the Web. keysearch.cgi Program to query the database using a keyword search or a specific specimen identifier. mysql_search.cgi Program to query the database using MySQL statements. table_contents.cgi Program to display the contents of each table in the database. tma_construct.cgi Program to upload TMA design construct information from the Web. tma_result_input.cgi Program to upload TMA experiment protocol and results from the Web. unknown_path.cgi Program to upload pathology information from the Web for specimens where the diagnosis is unknown.
Click here for file
Acknowledgements
This work is supported by Yorkshire Cancer Research.
==== Refs
Kallioniemi OP Wagner U Kononen J Sauter G Tissue microarray technology for high-throughput molecular profiling of cancer Hum Mol Genet 2001 10 657 662 11257096 10.1093/hmg/10.7.657
Kononen J Bubendorf L Kallioniemi A Barlund M Schraml P Leighton S Torhorst J Mihatsch MJ Sauter G Kallioniemi OP Tissue microarrays for high-throughput molecular profiling of tumor specimens Nat Med 1998 4 844 847 9662379 10.1038/nm0798-844
Henshall S Tissue microarrays J Mammary Gland Biol Neoplasia 2003 3 347 358 14973378 10.1023/B:JOMG.0000010034.43145.86
Berman JJ Edgerton ME Friedman BA The tissue microarray data exchange specification: a community-based, open source tool for sharing tissue microarray data BMC Med Inform Decis Mak 2003 3 5 12769826 10.1186/1472-6947-3-5
Manley S Mucci NR De Marzo AM Rubin MA Relational database structure to manage high-density tissue microarray data and images for pathology studies focusing on clinical outcome: the prostate specialized program of research excellence model Am J Pathol 2001 159 837 843 11549576
Liu CL Prapong W Natkunam Y Alizadeh A Montgomery K Gilks CB van de Rijn M Software tools for high-throughput analysis and archiving of immunohistochemistry staining data obtained with tissue microarrays Am J Pathol 2002 161 1557 1565 12414504
Coombes KR Zhang L Bueso-Ramos Brisbay S Logothetis C Roth J Keating MJ McDonnell TJ TAD: a web interface and database for tissue microarrays Appl Bioinformatics 2002 1 155 158 15130842
Berman JJ Datta M Kajdacsy-Balla A Melamed J Orenstein J Dobbin K Patel A Dhir R Becich MJ The tissue microarray data exchange specification: implementation by the Cooperative Prostate Cancer Tissue Resource BMC Bioinformatics 2004 5 19 15040818 10.1186/1471-2105-5-19
Spellman PT Miller M Stewart J Troup C Sarkans U Chervitz S Bernhart D Sherlock G Ball C Lepage M Swiatek M Marks WL Goncalves J Markel S Iordan D Shojatalab M Pizarro A White J Hubley R Deutsch E Senger M Aronow BJ Robinson A Bassett D Stoeckert CJ JrBrazma A Design and implementation of microarray gene expression markup language (MAGE-ML) Genome Biol 2002 3 RESEARCH0046 12225585 10.1186/gb-2002-3-9-research0046
Brazma A Hingamp P Quackenbush J Sherlock G Spellman P Stoeckert C Aach J Ansorge W Ball CA Causton HC Gaasterland T Glenisson P Holstege FC Kim IF Markowitz V Matese JC Parkinson H Robinson A Sarkans U Schulze-Kremer S Stewart J Taylor R Vilo J Vingron M Minimum information about a microarray experiment (MIAME)-toward standards for microarray data Nat Genet 2001 29 365 71 11726920 10.1038/ng1201-365
|
16137321
|
PMC1215475
|
CC BY
|
2021-01-04 16:27:26
|
no
|
BMC Bioinformatics. 2005 Sep 1; 6:218
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-218
|
oa_comm
|
==== Front
BMC Cardiovasc DisordBMC Cardiovascular Disorders1471-2261BioMed Central London 1471-2261-5-251612238110.1186/1471-2261-5-25Research ArticleRegulation and splicing of scavenger receptor class B type I in human macrophages and atherosclerotic plaques Svensson Per-Arne [email protected] Mikael CO [email protected]äckestrand Magnus SC [email protected]ägg Daniel A [email protected] Bertil G [email protected] Veronika [email protected] Lillemor [email protected] Dag S [email protected] Björn [email protected] Olov [email protected] Lena MS [email protected] Björn [email protected] Research Centre for Endocrinology & Metabolism, Department of Internal Medicine, The Sahlgrenska Academy, Göteborg University, S-413 45 Göteborg, Sweden2 The Wallenberg Laboratory for Cardiovascular Research, The Sahlgrenska Academy, Göteborg University, S-413 45 Göteborg, Sweden3 Cardiovascular Research Unit, Center for Molecular Medicine, Karolinska Institute, Stockholm, Sweden4 Department of Medicine, Cardiovascular Institute, The Sahlgrenska Academy, Göteborg University, Göteborg, Sweden5 Department of Body Composition and Metabolism, The Sahlgrenska Academy, Göteborg University, Göteborg, Sweden2005 25 8 2005 5 25 25 28 1 2005 25 8 2005 Copyright © 2005 Svensson et al; licensee BioMed Central Ltd.2005Svensson et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The protective role of high-density lipoprotein (HDL) in the cardiovascular system is related to its role in the reverse transport of cholesterol from the arterial wall to the liver for subsequent excretion via the bile. Scavenger receptor class B type I (SR-BI) binds HDL and mediates selective uptake of cholesterol ester and cellular efflux of cholesterol to HDL. The role of SR-BI in atherosclerosis has been well established in murine models but it remains unclear whether SR-BI plays an equally important role in atherosclerosis in humans. The aim of this study was to investigate the expression of SR-BI and its isoforms in human macrophages and atherosclerotic plaques.
Methods
The effect of hypoxia and minimally modified low-density lipoprotein (mmLDL), two proatherogenic stimuli, on SR-BI expression was studied in human monocyte-derived macrophages from healthy subjects using real-time PCR. In addition, SR-BI expression was determined in macrophages obtained from subjects with atherosclerosis (n = 15) and healthy controls (n = 15). Expression of SR-BI isoforms was characterized in human atherosclerotic plaques and macrophages using RT-PCR and DNA sequencing.
Results
SR-BI expression was decreased in macrophages after hypoxia (p < 0.005). In contrast, SR-BI expression was increased by exposure to mmLDL (p < 0.05). There was no difference in SR-BI expression in macrophages from patients with atherosclerosis compared to controls. In both groups, SR-BI expression was increased by exposure to mmLDL (p < 0.05). Transcripts corresponding to SR-BI and SR-BII were detected in macrophages. In addition, a third isoform, referred to as SR-BIII, was discovered. All three isoforms were also expressed in human atherosclerotic plaque. Compared to the other isoforms, the novel SR-BIII isoform was predicted to have a unique intracellular C-terminal domain containing 53 amino acids.
Conclusion
We conclude that SR-BI is regulated by proatherogenic stimuli in humans. However, we found no differences between subjects with atherosclerosis and healthy controls. This indicates that altered SR-BI expression is not a common cause of atherosclerosis. In addition, we identified SR-BIII as a novel isoform expressed in human macrophages and in human atherosclerotic plaques.
==== Body
Background
The presence of lipid-loaded macrophages (foam cells) in the vascular wall is one of the hallmarks of atherosclerosis. High levels of low-density lipoprotein (LDL) cholesterol increase the risk of atherosclerosis and the lipids accumulated in the macrophages are derived mainly from modified forms of LDL. Oxidation of LDL is believed to be the major modification of LDL cholesterol in vivo [1]. As the atherosclerotic plaque increases in size due to accumulation of lipid and fibrous elements, local supply of oxygen and nutrients may become insufficient. In advanced atherosclerotic plaques, zones of hypoxia have been identified [2,3].
The level of high-density lipoprotein (HDL) cholesterol is inversely related to the risk of developing atherosclerosis. A proposed mechanism for the athero-protective role of HDL is the transfer of cholesterol from the arterial wall via the liver to the bile, a process known as reverse cholesterol transport [4]. Scavenger receptor class B type I (SR-BI; also known as CLA-1) has been shown to mediate selective uptake of cholesterol esters from HDL particles into cells [5]. Moreover, SR-BI mediates cholesterol efflux in vitro [6,7] and has been suggested to contribute to cellular cholesterol efflux also in vivo. Therefore, SR-BI may be of importance also for the first step in reverse cholesterol transport, i.e. the removal of cholesterol from the arterial wall [4].
The role of SR-BI in HDL metabolism and atherosclerosis has been well established in murine models. Experiments in rodents have shown that SR-BI is expressed in tissues with selective cholesterol uptake, e.g. liver, adrenals, ovaries and testes [5,8]. Hepatic overexpression of SR-BI in transgenic mice decreases the development of atherosclerosis in cholesterol-fed LDL receptor knockout (KO) mice [9]. Recent studies have demonstrated that lack of SR-BI expression in macrophages increases atherosclerotic lesion formation in mice [10]. The link between human SR-BI and atherosclerosis in man is less well established. Human SR-BI mediates selective cholesterol uptake into cells and displays a tissue distribution similar to that in rodents [11,12]. SR-BI expression has been detected in human atherosclerotic plaques [13] and variants of the human SR-BI gene have been associated to lipid abnormalities [14,15]. Recently, a variant of the human SR-BI promoter, that displays decreased expression, has been associated to increased plasma HDL cholesterol levels [16]. Therefore, decreased macrophage expression of human SR-BI may be a mechanism that promotes the development of atherosclerosis.
The aim of this study was to investigate the expression of SR-BI in human macrophages and atherosclerotic plaques.
Methods
Subjects and samples
Patients diagnosed with myocardial infarction or unstable angina pectoris were identified via the INTERGENE study, which is a population study recruiting subjects from Västra Götaland, Sweden. For details on the sampling and procedures in the INTERGENE study, see Berg et al [17]. First-degree relative (siblings, children or parents) of the patients were screened by ultra-sound examination (Sequoia ultra sound scanner) of the intima thickness of the carotid and femoral arteries to determine the extent of atherosclerosis. Fifteen subjects with sub-clinical atherosclerosis were recruited to this study (the macrophage INTERGENE study). The inclusion criteria for the atherosclerotic group were at least one atherosclerotic plaque in the carotid or femoral arteries and a verified family history of myocardial infarction or unstable angina pectoris, as documented in the INTERGENE study. Fifteen healthy control subjects matched by sex and age, which had no such family history, were also recruited from the INTERGENE population study. The exclusion criteria for both groups were, age <40 years, diabetes mellitus, clinical cardiovascular disease, smoking during the last 10 years, severe hypercholesterolemia, current infection (C-reactive protein >5 mg/L), and severe chronic disease. Additional exclusion criteria for the control group were hypertension.
Specimens from atherosclerotic carotid arteries were obtained from two females and two males undergoing carotid endarterectomy (n = 4).
The lipid analyses were performed at the Wallenberg Laboratory and the remaining biochemical analyses at the central laboratory at Sahlgrenska University Hospital. Triglyceride and cholesterol levels were analysed by fully enzymatic techniques [18,19]. LDL cholesterol was calculated as described by Friedewald et al. [20]. HDL was determined after precipitation of apolipoprotein (apo) B-containing lipoproteins with manganese chloride and dextran sulfate. The clinical and biochemical characteristics of the study groups (macrophage INTERGENE subjects) are presented in Table 1. The atherosclerotic subjects had higher total cholesterol levels and higher systolic blood pressure compared to the healthy control subjects.
Table 1 Clinical and biochemical characteristics of the study groups (The macrophage INTERGENE subjects).
Characteristic Healthy control subjects (n = 15) Atherosclerotic subjects (n = 15)
Age1 (years) 57 (47–73) 58 (43–74)
Weight (Kg) 79 ± 13 80 ± 13
Systolic blood pressure (mmHg) 126 ± 14 144 ± 16*
Diastolic blood pressure (mmHg) 74 ± 8 78 ± 11
Serum Na (mmol/L) 140 ± 2 140 ± 2
Serum K (mmol/L) 4.10 ± 0.26 4.11 ± 0.27
Serum Ca (mmol/L) 2.36 ± 0.07 2.37 ± 0.07
Blood Hb (g/L) 142 ± 6 143 ± 9
Blood glucose (mmol/L) 5.1 ± 0.8 5.0 ± 0.7
Total cholesterol (mmol/L) 5.33 ± 0.64 5.99 ± 0.69*
LDL cholesterol (mmol/L) 3.23 ± 0.72 3.78 ± 0.75
HDL cholesterol (mmol/L) 1.56 ± 0.50 1.40 ± 0.49
Serum triglycerides (mmol/l) 1.37 ± 0.46 1.78 ± 1.07
Data are presented as means ± SD. 1mean and range. *P < 0.05 (Student's t-test).
The different studies in the project were approved by the ethics committees of Göteborg University and Karolinska Institute.
Macrophage preparation and culture
Human mononuclear cells were isolated by Ficoll-Paque (Amersham Biosciences, Uppsala, Sweden) using buffy coats from the atherosclerotic subjects or healthy control subjects in the macrophage INTERGENE study or healthy volunteers. The mononuclear cell preparations were washed five times with PBS, pH 7.2, without calcium and magnesium, but containing 10 mM EDTA. The cell preparation was performed at room temperature. Mononuclear cells were resuspended and seeded at a density of 107 cells per each 100-mm plastic dish in a serum-free medium (Macrophage-SFM, GIBCO BRL; Grand Island, NY), supplemented with penicillin 100 U/ml and streptomycin 100 μg/ml. The mononuclear cells were allowed to adhere for 1 h. Non-adhered cells were eliminated with three washes with PBS. Adherent monocytes were cultured in Macrophage-SFM medium with antibiotics and supplemented with human granulocyte macrophage colony stimulating factor (GM-CSF), 70 U/ml (R&D Systems Europe Ltd., Abingdon, UK). The medium was discarded after 3d and the cells were washed once with PBS. The cells were then allowed to grow for another 3d in the same medium, but without GM-CSF and were then defined as monocyte-derived macrophages, and referred to as "macrophages" in the following text.
Preparation of human LDL and cell treatment
Preparation of LDL was performed as previously described [21]. Minimally modified LDL (mmLDL) was generated by storing sterile native LDL in the dark at 4°C for three months (thiobarbituric acid-reactive substances (TBARS) = 5 nmol MDA/mg protein). Macrophages from healthy volunteers or from subjects in the macrophage INTERGENE study were incubated without (control) or with mmLDL (50 μg protein/ml) in Macrophage-SFM medium for 24 h. Macrophages from healthy volunteers were also incubated under normoxic (21% O2) or hypoxic (0% O2) conditions for 24 h as previously described [22]. Total RNA was isolated from the macrophages using RNeasy kit (Qiagen, Hilden, Germany).
Real-time PCR analysis of SR-BI gene expression
Oligonucleotide primers and probes were designed with Primer Express 1.5 software (Applied Biosystems, Foster City, CA). Primers (forward: 5'-CCG CAC CTT CCA GTT CCA-3', reverse: 5'- ATG TTG GGC ATG ACG ATG TAG TC-3') and probe (5'-TCC AAG TCC CAC GGC TCG GAG A-3') were purchased from Applied Biosystems. These primers detects both SR-BI and the SR-BII isoform mRNA. The probes consisted of oligonucleotides that were labeled at the 5' end with the reporter dye FAM and at the 3' end with the quencher TAMRA. Reagents (TaqMan® Reverse Transcriptase reagents and TaqMan® Universal PCR Master mix, Applied Biosystems) and conditions were used according to the manufacturer's protocol. Briefly, cDNA was synthesized using reverse transcriptase (RT) from RNA samples. Each amplification reaction consisted of diluted cDNA (corresponding to 20 ng RNA), 300 nM of each primer, 200 nM TaqMan probe and TaqMan® Universal PCR reaction mix. Amplification and detection of specific products were performed with the ABI Prism 7700 sequence detection system (Applied Biosystems) using default cycle parameters. Pre-developed assay reagents for human RPLP0 (large ribosomal protein) was obtained from Applied Biosystems and used as reference to normalize the expression levels between the samples. All standards and samples were analyzed in triplicates.
Analysis of SR-BI isoforms by reverse transcriptase -PCR
As an isoform of SR-BI, SR-BII, is expressed in mouse tissue and human cell lines, primers able to detect both isoforms were used (Figure 3A). Total RNA was isolated from macrophages from healthy volunteers and from atherosclerotic plaque tissue using the RNeasy kit (Qiagen, Hilden, Germany). RNA was reverse transcribed using random hexamers and 20 units AMV-reverse transcriptase (Promega Corp., Madison, WI). PCR was performed in Taq buffer (Roche Diagnostics, Mannheim, Germany) with 2 units of Taq polymerase (Roche Diagnostics), 1 μM of primers hCLA-1 z5 (5'-GGG AAG ATC GAG CCA GTA-3'; Genset, Paris, France) and hCLA-1 z3 Not I (5'-GCG CGG CCG CGG GGA CAG TGT GAC ATC T-3'; Genset), 400 ng cDNA and dNTPs (0.2 mM each) using GeneAmp PCR system 9600 (Perkin-Elmer, Foster City. CA). The amplification is predicted to generate a 476 bp fragment for the SR-BI transcript and a 347 bp fragment for the SR-BII transcript. PCR products and Nco I-digested PCR products were separated on a 3% agarose gel containing ethidium bromide. The PCR products were cloned into the pCRII vector (Invitrogen, San Diego, CA) and verified by DNA sequencing.
Figure 3 Analysis of SR-BI expression in human macrophages and human carotid atherosclerotic plaque by RT-PCR. Schematic illustration of the human SR-BI cDNA and the location of the primers used for PCR (A). Analysis of expression of SR-BI and SR-BI isoforms in RNA extracted from macrophages by RT-PCR (B). PCR products from four donors (D1–D4) and PCR products from donor 4 digested with Nco I (D4+Nco I), non-template control (ntc), 1 kb DNA-molecular marker (m). Analysis of expression of SR-BI and SR-BI isoforms in RNA extracted from atherosclerotic plaque tissue by RT-PCR (C). PCR products from atherosclerotic plaque tissue cDNA from four patients (P1–P4), PCR product from patient 4 digested with Nco I (P4+Nco I). PCR product corresponding to undigested SR-BI (476 bp), SR-BII (347 bp) and SR-BIII (541 bp) are indicated.
Statistical analysis
Statistical analyses were performed using Student's t-test or by ANOVA followed by Student's t-test.
Results
Effect of hypoxia on macrophage SR-BI expression
As the thickness of the atherosclerotic plaque increases, diffusion of oxygen is decreased and zones of low oxygen tension can be detected within atherosclerotic plaques. Several studies have shown that hypoxia alters a large number of macrophage functions [2]. We therefore investigated if macrophage SR-BI expression is regulated by hypoxia in vitro. Human macrophages from 13 healthy volunteers were exposed to hypoxia for 24 h. The expression of macrophage SR-BI was decreased from 1.26 ± 0.17 arbitrary units to 0.73 ± 0.10 arbitrary units (mean ± SEM; p < 0.005) after hypoxia exposure as determined by real-time RT-PCR analysis.
Effect of mmLDL on macrophage SR-BI expression
The uptake of lipids and cholesterol by macrophages is a key event implicated in the development of atherosclerosis. Therefore, scavenger receptor expression is thought to be a critical determinant of lipid accumulation in macrophages. Human macrophages from 4 healthy volunteers were treated with mmLDL (50 μg protein/ml) for 24 h. The expression of SR-BI was significantly up regulated in macrophages after mmLDL treatment compared to control macrophages (p < 0.05, Figure 1) as determined by real-time RT-PCR analysis.
Figure 1 Effect of mmLDL on the expression of SR-BI in macrophages. Macrophages from 4 different healthy voluntary donors were exposed to mmLDL (50 μg protein/ml) for 24 h and SR-BI gene expression was analyzed by real-time RT-PCR using RNA from mmLDL treated and untreated control macrophages. The primers for real-time RT-PCR analysis used in this experiment were designed to detect SR-BI and the SR-BI isoforms. SR-BI expression was normalized to the reference gene RPLP0. The results are presented as mean +/- SEM. *, p < 0.005.
Expression of SR-BI in macrophages derived from subjects with atherosclerosis
As SR-BI expression in macrophages may modulate the development of atherosclerosis we investigated if macrophages derived from subjects with atherosclerosis displayed an altered expression pattern compared to macrophages derived from matched healthy controls. SR-BI gene expression in macrophages and mmLDL-treated macrophages from 15 atherosclerotic subjects and 15 controls (the macrophage INTERGENE subjects) were analyzed by real-time RT-PCR (Figure 2). No difference in gene expression was detected between the subjects with atherosclerosis and the matched healthy controls (Figure 2). However, the analysis confirms the previous findings that SR-BI expression is increased in response to 24 h mmLDL treatment of macrophages from both the subjects with atherosclerosis and the control subjects (p < 0.05 and p < 0.05, respectively). Macrophage SR-BI expression levels were not correlated to serum HDL cholesterol levels in the macrophage INTERGENE subjects (data not shown).
Figure 2 SR-BI expression in macrophages from atherosclerotic subjects. Macrophages from 15 atherosclerotic subjects and 15 healthy matched control subjects were treated with mmLDL (50 μg protein/ml) for 24 h. SR-BI expression was analyzed in both groups both before and after mmLDL treatment by real-time RT-PCR. SR-BI expression was normalized to the reference gene RPLP0. The primers for real-time RT-PCR analysis used in this experiment detect SR-BI and the SR-BI isoforms. The results are presented as mean +/- SEM. *, p < 0.05.
Analysis of SR-BI isoforms in macrophages and atherosclerotic plaques
As different isoforms of SR-BI may modulate receptor function [23], the expression of SR-BI isoforms was investigated in human macrophages by RT-PCR using primers designed to detect both SR-BI and SR-BII transcripts. SR-BI (476 bp) and SR-BII (347 bp) transcripts were detected in all the samples examined (Figure 3B). The identities of the PCR products were verified by restriction enzyme mapping. Digestion with Nco I resulted in a 53-bp decrease in fragment size of both SR-BI (423 bp) and SR-BII (294 bp) as predicted from the mRNA sequence. Sequence analysis of the 347 bp PCR product (clones from 2 subjects) confirmed that this isoform is the human equivalent to the SR-BII isoform originally cloned in mouse. An additional PCR product that decreased in size with Nco I digestion was also detected. The PCR product (541 bp) was sequenced (clones from 2 subjects) and identified as a novel isoform of the SR-BI receptor and was therefore designated SR-BIII (GenBank accession number AF254409). This novel isoform appears to be generated by the use of an alternative splice acceptor site in intron 11, resulting in a frame shift after aa 467, and can be predicted to generate a unique intracellular C-terminal domain containing 53 aa (Figure 4A and 4B).
Figure 4 Alternative splicing generating SR-BI, SR-BII, and SR-BIII transcripts. Schematic illustration of the predicted alternative splicing generating the SR-BI, SR-BII, and SR-BIII transcripts (A). White boxes indicate exons, arrows indicate stop codons, black boxes indicate the part coding for the C-terminal transmembrane domain and the gray box indicates the alternative reading frame used in the SR-BIII isoform. Nucleotide and amino acid sequence of the human SR-BI, SR-BII, and SR-BII isoforms (B). Putative stop codons are underlined.
To analyze whether the SR-BI isoforms were also expressed in human atherosclerotic plaque tissue, the experiment was repeated using RNA samples from plaques obtained from four patients that underwent carotid endarterectomy. PCR products corresponding to SR-BI (476 bp), SR-BII (347 bp) and SR-BIII (541 bp) were detected in atherosclerotic plaques from all four patients analyzed (Figure 3C).
Analysis of the intracellular C-terminal domain of the SR-BI isoforms
SR-BI and SR-BI isoforms are predicted to contain two transmembrane domains and two intracellular domains at the amino (N)- and carboxy (C)-termini, respectively. The N-terminal intracellular domain of SR-BI is predicted to contain 8 amino acids (aa) and the C-terminal intracellular domain of SR-BI is predicted to consist of 47 aa. A peroxisomal targeting sequence (PTS1) is present in the C-terminal domain of SR-BI, which is not present in the SR-BII and SR-BIII isoforms [8,24]. The intracellular C-terminal domain of human SR-BII contains 44 aa and has a high content of proline aa (12 of 44 aa). The predicted amino acid sequences of SR-BII from other species also have high proline content in the intracellular C-terminal domain (Figure 5B). The prolines in human SR-BII are in some places interspersed with two other amino acids generating PXXP motifs. The PXXP motif is the minimal consensus binding-motif for src homology 3 (SH3) domain-containing proteins [25]. One of these PXXP motifs (aa 488–491 of SR-BII) is conserved between species (Figure 5B). The novel SR-BIII transcript is predicted to encode an intracellular C-terminal domain that is longer (53 aa) than the intracellular C-terminal domains of both SR-BI and SR-BII (Figure 5A). The C-terminal domain of SR-BIII is less proline-rich than SR-BII and one SH3 binding site in SR-BIII is conserved between species (Figure 5B).
Figure 5 Intracellular domains of SR-BI isoforms. Schematic illustration of the topology of the three isoforms (A). The PTS1 domain of SR-BI, the proline-rich domain of SR-BII and the lengths of the intracellular domains are indicated. Species alignment of the intracellular domain of SR-BI, SR-BII and SR-BIII (B). Bovine, pig, hamster, rat and mouse SR-BIII sequences were deduced from SR-BI mRNAs. Sequence identities (dark gray boxes) are indicated. The isoforms differ in amino acid sequence downstream of Gln-467 (arrow). The PTS1 motif of SR-BI (white bold characters), and proline residues of the Src Homology (SH3) binding motif (black bold) are indicated. Potential sites for protein kinases A and C dependent phosphorylation, SH2 and SH3 binding motifs are indicated by bars.
Discussion
The key role of SR-BI in the development of atherosclerosis has been shown in genetically modified mice models. In particular, hepatic overexpression of SR-BI reduces atherosclerotic lesion formation [9], which is attributed to the role of SR-BI in the uptake of HDL-cholesterol in the liver. This is supported by the finding that SR-BI disruption increases plaque formation [26]. Genetically modified mice models have also been used to investigate the role of macrophage SR-BI expression in the development of atherosclerosis. Bone marrow transplantation studies have shown that SR-BI disruption in macrophages augment the development of advanced atherosclerotic plaques [10,27]. This effect is attributed to SR-BI role in cholesterol efflux. However, a recent study by Van Eck et al shows that the development of early atherosclerotic plaque (fatty streaks) is reduced in macrophages lacking SR-BI expression [28]. This effect could be related to the scavenger receptor function of SR-BI by enhancing uptake of oxidized LDL or VLDL [28]. This indicates that macrophage SR-BI probably plays multiple roles in the development of atherosclerosis.
As the thickness of the atherosclerotic plaque increases, diffusion of oxygen is decreased and zones of low oxygen tension within atherosclerotic plaques have been detected [3]. In this study we show for the first time that SR-BI gene expression is down regulated by hypoxia. Reduced macrophage SR-BI gene expression during hypoxic condition may lead to reduced cholesterol efflux from macrophages in hypoxic zones of the atherosclerotic plaque and may increase foam cell formation in these areas.
Previous studies have shown disparate effect of highly oxidized LDL (oxLDL) on SR-BI expression in human monocyte-derived macrophages. Hirano et al reported an increased expression of SR-BI during macrophage differentiation and that oxLDL induced SR-BI mRNA and protein expression after 24 h of treatment [29]. In contrast, Han et al reported that 10 days differentiated macrophages displayed a decreased SR-BI expression in response to 16 h oxLDL treatment [30]. However, less differentiated macrophages (3 days differentiation) responded to 16 h of oxLDL treatment with increased SR-BI expression [30]. This indicates that the differentiation status of the macrophages and the time of oxLDL treatment influences SR-BI expression. In this study we show that SR-BI expression in macrophages is increased in response to 24 h of mmLDL treatment. Differences in chemical properties of mmLDL compared to oxLDL can affect macrophage SR-BI expression. It is unclear which of these in vitro foam cell formation models that best resembles foam cell formation in vivo.
It has been suggested that genetic variants involved in the development of complex disorders such as atherosclerosis, are often located in regulatory regions of the genome and affect the transcription of the susceptibility gene. As SR-BI gene expression in macrophages may modulate the development of atherosclerosis we investigated if macrophages derived from subjects with atherosclerosis displayed an altered expression pattern compared to macrophages derived from matched healthy controls. The results of this experiment indicate that altered expression of SR-BI in macrophages is not a common component in the development on atherosclerosis in the Swedish population. Recently, a variant of the human SR-BI promoter has been identified in a Taiwanese Chinese population. This promoter variant displays decreased gene expression and has been associated to increased plasma HDL cholesterol levels [16]. However, the allele frequency of this functional variant is rather low. SR-BI and members of the ATP-binding cassette transporters have been shown to facilitate cellular cholesterol efflux. SR-BI mainly binds phospholipid-rich HDL particles, whereas ABCA1 preferably uses lipid-poor HDL particles as cholesterol acceptors [31]. However, the precise role and interactions between these proteins is not completely understood.
The presence of multiple SR-BI isoforms in macrophages and atherosclerotic plaques makes the interpretation of the role of SR-BI in human atherosclerosis complex. We have previously identified a PTS1 motif in the C-terminus of SR-BI [8,24], which is not present in SR-BII and SR-BIII. The PTS1 motif is recognized by the peroxisomal targeting import receptor Pex5p [32], which mediates uptake of proteins in the peroxisome and may therefore be of importance for SR-BI action as the peroxisome is an important subcellular site for cholesterol and bile acid metabolism. A protein from rat liver membrane extracts named CLAMP has been shown to interact with the C-terminal end of SR-BI [33]. The human SR-BII isoform was also detected in macrophages and atherosclerotic plaques. The mouse SR-BII isoform binds HDL and mediates both cholesterol uptake and efflux but seems to be less efficient than SR-BI [23]. Lower surface expression or reduced stability of SR-BII compared to SR-BI may contribute to the differences between the two molecules [24]. It has recently been shown that SH3 domain containing proteins may interact with the SR-BII C-terminal domain [34]. This indicates that the different SR-BI isoforms may have distinct functional and signaling properties.
We have identified a novel isoform of SR-BI, SR-BIII, which is expressed in macrophages and atherosclerotic plaques. The SR-BIII isoform is probably generated by the use of an alternative splice acceptor site in intron 11 and encodes a unique intracellular C-terminal of the receptor. Further studies are needed to determine if SR-BIII also functions as a HDL receptor and what physiological function it may have.
Conclusion
We conclude that SR-BI is regulated by proatherogenic stimuli such as hypoxia and mmLDL in human macrophages. However, no differences in SR-BI expression were detected between macrophages from subjects with atherosclerosis and healthy controls. This indicates that altered SR-BI expression is not a common feature of atherosclerosis. In addition, we identified SR-BIII as a novel SR-BI isoform expressed in human macrophages and in human atherosclerotic plaques.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
All the authors have contributed to the design of the study, the data analysis and the writing of the manuscript. P-A.S. performed RT-PCR analysis and bioinformatical analysis. M.C.O.E. and B.O. performed the macrophage culture and mmLDL treatment. M.S.C.S carried out the sequencing and sequence analysis. D.H: performed the real-time RT-PCR analysis. V.S. collected the atherosclerosis plaque samples and extracted RNA. B.F. and D.T. coordinated and phenotyped the macrophage INTERGENE subjects, L.M.H. and O.W. contributed with the macrophage hypoxia experiments. L.M.S.C and B.C. designed the study, and participated in its coordination and data interpretation. The final version of the manuscript has been read and approved by all the authors.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by grants from AFA, Swegene, IngaBritt and Arne Lundberg Forskningsstiftelse, Emmeles fond, Fredrik and Ingrid Thurlings fond, Wilhelm och Martina Lundgrens Vetenskapsfond, Kungliga och Hvitfeldtska Stipendiestiftelsen, Stiftelsen Fonden för studerande av läkarvetenskap vid Sahlgrenska sjukhuset, Göteborg Medical Association and the Swedish Medical Research Council (Grants 6816, 11285, 11502 and 13141).
==== Refs
Steinberg D Low density lipoprotein oxidation and its pathobiological significance J Biol Chem 1997 272 20963 20966 9261091 10.1074/jbc.272.34.20963
Lewis JS Lee JA Underwood JC Harris AL Lewis CE Macrophage responses to hypoxia: relevance to disease mechanisms J Leukoc Biol 1999 66 889 900 10614769
Bjornheden T Levin M Evaldsson M Wiklund O Evidence of hypoxic areas within the arterial wall in vivo Arterioscler Thromb Vasc Biol 1999 19 870 876 10195911
Tall AR Plasma high density lipoproteins. Metabolism and relationship to atherogenesis J Clin Invest 1990 86 379 384 2200802
Acton S Rigotti A Landschulz KT Xu S Hobbs HH Krieger M Identification of scavenger receptor SR-BI as a high density lipoprotein receptor Science 1996 271 518 520 8560269
Ji Y Jian B Wang N Sun Y Moya ML Phillips MC Rothblat GH Swaney JB Tall AR Scavenger receptor BI promotes high density lipoprotein-mediated cellular cholesterol efflux J Biol Chem 1997 272 20982 20985 9261096 10.1074/jbc.272.34.20982
Jian B de la Llera-Moya M Ji Y Wang N Phillips MC Swaney JB Tall AR Rothblat GH Scavenger receptor class B type I as a mediator of cellular cholesterol efflux to lipoproteins and phospholipid acceptors J Biol Chem 1998 273 5599 5606 9488688 10.1074/jbc.273.10.5599
Johnson MS Svensson PA Helou K Billig H Levan G Carlsson LM Carlsson B Characterization and chromosomal localization of rat scavenger receptor class B type I, a high density lipoprotein receptor with a putative leucine zipper domain and peroxisomal targeting sequence Endocrinology 1998 139 72 80 9421400 10.1210/en.139.1.72
Arai T Wang N Bezouevski M Welch C Tall AR Decreased atherosclerosis in heterozygous low density lipoprotein receptor-deficient mice expressing the scavenger receptor BI transgene J Biol Chem 1999 274 2366 2371 9891004 10.1074/jbc.274.4.2366
Zhang W Yancey PG Su YR Babaev VR Zhang Y Fazio S Linton MF Inactivation of macrophage scavenger receptor class B type I promotes atherosclerotic lesion development in apolipoprotein E-deficient mice Circulation 2003 108 2258 2263 14581413 10.1161/01.CIR.0000093189.97429.9D
Cao G Garcia CK Wyne KL Schultz RA Parker KL Hobbs HH Structure and localization of the human gene encoding SR-BI/CLA-1. Evidence for transcriptional control by steroidogenic factor 1 J Biol Chem 1997 272 33068 33076 9407090 10.1074/jbc.272.52.33068
Murao K Terpstra V Green SR Kondratenko N Steinberg D Quehenberger O Characterization of CLA-1, a human homologue of rodent scavenger receptor BI, as a receptor for high density lipoprotein and apoptotic thymocytes J Biol Chem 1997 272 17551 17557 9211901 10.1074/jbc.272.28.17551
Chinetti G Gbaguidi FG Griglio S Mallat Z Antonucci M Poulain P Chapman J Fruchart JC Tedgui A Najib-Fruchart J Staels B CLA-1/SR-BI is expressed in atherosclerotic lesion macrophages and regulated by activators of peroxisome proliferator-activated receptors Circulation 2000 101 2411 2417 10821819
Acton S Osgood D Donoghue M Corella D Pocovi M Cenarro A Mozas P Keilty J Squazzo S Woolf EA Ordovas JM Association of polymorphisms at the SR-BI gene locus with plasma lipid levels and body mass index in a white population Arterioscler Thromb Vasc Biol 1999 19 1734 1743 10397692
Hong SH Kim YR Yoon YM Min WK Chun SI Kim JQ Association between HaeIII polymorphism of scavenger receptor class B type I gene and plasma HDL-cholesterol concentration Ann Clin Biochem 2002 39 478 481 12227853 10.1258/000456302320314485
Hsu LA Ko YL Wu S Teng MS Peng TY Chen CF Lee YS Association between a novel 11-base pair deletion mutation in the promoter region of the scavenger receptor class B type I gene and plasma HDL cholesterol levels in Taiwanese Chinese Arterioscler Thromb Vasc Biol 2003 23 1869 1874 12816880 10.1161/01.ATV.0000082525.84814.A9
Berg CM Lissner L Aires N Lappas G Toren K Wilhelmsen L Rosengren A Thelle DS Trends in blood lipid levels, blood pressure, alcohol and smoking habits from 1985 to 2002: results from INTERGENE and GOT-MONICA Eur J Cardiovasc Prev Rehabil 2005 12 115 125 15785296 10.1097/00149831-200504000-00005
Bokemark L Wikstrand J Wedel H Fagerberg B Insulin, insulin propeptides and intima-media thickness in the carotid artery in 58-year-old clinically healthy men. The Atherosclerosis and Insulin Resistance study (AIR) Diabet Med 2002 19 144 151 11874431 10.1046/j.1464-5491.2002.00664.x
Borner K Klose S [Enzymatic determination of total cholesterol with the Greiner Selective Analyzer (GSA-II) (author's transl)] J Clin Chem Clin Biochem 1977 15 121 130 870610
Friedewald WT Levy RI Fredrickson DS Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge Clin Chem 1972 18 499 502 4337382
Ohlsson BG Englund MC Karlsson AL Knutsen E Erixon C Skribeck H Liu Y Bondjers G Wiklund O Oxidized low density lipoprotein inhibits lipopolysaccharide-induced binding of nuclear factor-kappaB to DNA and the subsequent expression of tumor necrosis factor-alpha and interleukin-1beta in macrophages J Clin Invest 1996 98 78 89 8690807
Rydberg EK Krettek A Ullstrom C Ekstrom K Svensson PA Carlsson LM Jonsson-Rylander AC Hansson GI McPheat W Wiklund O Ohlsson BG Hulten LM Hypoxia increases LDL oxidation and expression of 15-lipoxygenase-2 in human macrophages Arterioscler Thromb Vasc Biol 2004 24 2040 2045 15358603 10.1161/01.ATV.0000144951.08072.0b
Webb NR Connell PM Graf GA Smart EJ de Villiers WJ de Beer FC van der Westhuyzen DR SR-BII, an isoform of the scavenger receptor BI containing an alternate cytoplasmic tail, mediates lipid transfer between high density lipoprotein and cells J Biol Chem 1998 273 15241 15248 9614139 10.1074/jbc.273.24.15241
Johnson MS Johansson JM Svensson PA Aberg MA Eriksson PS Carlsson LM Carlsson B Interaction of scavenger receptor class B type I with peroxisomal targeting receptor Pex5p Biochem Biophys Res Commun 2003 312 1325 1334 14652019 10.1016/j.bbrc.2003.11.076
Sudol M From Src Homology domains to other signaling modules: proposal of the 'protein recognition code' Oncogene 1998 17 1469 1474 9779993 10.1038/sj.onc.1202182
Trigatti B Rayburn H Vinals M Braun A Miettinen H Penman M Hertz M Schrenzel M Amigo L Rigotti A Krieger M Influence of the high density lipoprotein receptor SR-BI on reproductive and cardiovascular pathophysiology Proc Natl Acad Sci U S A 1999 96 9322 9327 10430941 10.1073/pnas.96.16.9322
Covey SD Krieger M Wang W Penman M Trigatti BL Scavenger receptor class B type I-mediated protection against atherosclerosis in LDL receptor-negative mice involves its expression in bone marrow-derived cells Arterioscler Thromb Vasc Biol 2003 23 1589 1594 12829524 10.1161/01.ATV.0000083343.19940.A0
Van Eck M Bos IS Hildebrand RB Van Rij BT Van Berkel TJ Dual role for scavenger receptor class B, type I on bone marrow-derived cells in atherosclerotic lesion development Am J Pathol 2004 165 785 794 15331403
Hirano K Yamashita S Nakagawa Y Ohya T Matsuura F Tsukamoto K Okamoto Y Matsuyama A Matsumoto K Miyagawa J Matsuzawa Y Expression of human scavenger receptor class B type I in cultured human monocyte-derived macrophages and atherosclerotic lesions Circ Res 1999 85 108 116 10400916
Han J Nicholson AC Zhou X Feng J Gotto AMJ Hajjar DP Oxidized low density lipoprotein decreases macrophage expression of scavenger receptor B-I J Biol Chem 2001 276 16567 16572 11278882 10.1074/jbc.M011302200
Yancey PG Bortnick AE Kellner-Weibel G de la Llera-Moya M Phillips MC Rothblat GH Importance of different pathways of cellular cholesterol efflux Arterioscler Thromb Vasc Biol 2003 23 712 719 12615688 10.1161/01.ATV.0000057572.97137.DD
Fransen M Brees C Baumgart E Vanhooren JC Baes M Mannaerts GP Van Veldhoven PP Identification and characterization of the putative human peroxisomal C-terminal targeting signal import receptor J Biol Chem 1995 270 7731 7736 7706321 10.1074/jbc.270.13.7731
Ikemoto M Arai H Feng D Tanaka K Aoki J Dohmae N Takio K Adachi H Tsujimoto M Inoue K Identification of a PDZ-domain-containing protein that interacts with the scavenger receptor class B type I Proc Natl Acad Sci U S A 2000 97 6538 6543 10829064 10.1073/pnas.100114397
Mulcahy JV Riddell DR Owen JS Human scavenger receptor class B type II (SR-BII) and cellular cholesterol efflux Biochem J 2004 377 741 747 14570588
|
16122381
|
PMC1215476
|
CC BY
|
2021-01-04 16:30:06
|
no
|
BMC Cardiovasc Disord. 2005 Aug 25; 5:25
|
utf-8
|
BMC Cardiovasc Disord
| 2,005 |
10.1186/1471-2261-5-25
|
oa_comm
|
==== Front
BMC Cell BiolBMC Cell Biology1471-2121BioMed Central London 1471-2121-6-321612022010.1186/1471-2121-6-32Research ArticleCrosstalk between the actin cytoskeleton and Ran-mediated nuclear transport Minakhina Svetlana [email protected] Ron [email protected] Marina [email protected] Ruth [email protected] Waksman Institute, Department of Molecular Biology and Biochemistry, NJ Cancer Center, Rutgers University, 190 Frelinghuysen Road, Piscataway, NJ 08854-8020, USA2005 24 8 2005 6 32 32 16 12 2004 24 8 2005 Copyright © 2005 Minakhina et al; licensee BioMed Central Ltd.2005Minakhina et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Transport of macromolecules into and out of the nucleus is a highly regulated process. The RanGTP/RanGDP gradient controls the trafficking of molecules exceeding the diffusion limit of the nuclear pore across the nuclear envelope.
Results
We found genetic interaction between genes establishing the Ran gradient, nuclear transport factor 2 (ntf-2), Ran GTPase activating protein (Sd), and the gene encoding Drosophila Profilin, chickadee (chic). The severe eye phenotype caused by reduction of NTF2 is suppressed by loss of function mutations in chic and gain of function mutations in Sd (RanGAP). We show that in chic mutants, as in Sd-RanGAP, nuclear export is impaired.
Conclusion
Our data suggest that Profilin and the organization of the actin cytoskeleton play an important role in nuclear trafficking.
==== Body
Background
In eukaryotic cells the nuclear envelope serves as a barrier between the nucleus and cytoplasm. The transport of molecules between the nucleus and cytoplasm occurs through nuclear pore complexes (NPCs). Although some small molecules (<40 kDa) diffuse through the pore, most proteins and RNAs require facilitated transport and special receptors called importins and exportins. This facilitated transport further depends on the small Ras family GTPase Ran (for review see [1,2]).
Similar to other GTPases, Ran is regulated by conformational changes driven by GTP hydrolysis and nucleotide exchange. GDP to GTP exchange happens in the nucleus and is catalyzed by the chromatin-associated protein RCC1 [3]. The GTPase activating protein RanGAP controls RanGTP hydrolysis in the cytoplasm. Spatial partition of these two processes generates a RanGTP concentration gradient across the nuclear pore. This gradient is thought to guide the directionality of nuclear transport.
Protein cargo containing nuclear localization signals (NLS) is recognized by importins and translocates through nuclear pores into the nuclei, where binding to RanGTP causes release of the cargo. In the nuclei, proteins containing nuclear export signals (NESs) form export complexes together with RanGTP. These complexes are then exported, and upon RanGTP hydrolysis in the cytoplasm, the cargo is released. RanGDP nuclear reentry is mediated by the nuclear transport factor 2 (NTF2, [4,5]).
NTF2 was originally identified by its ability to stimulate protein import into nuclei in permeabilized mammalian cells [6]. NTF2 was further shown to have a critical role in actively replenishing the nuclear stock of Ran [4,7,8]. NTF2 catalyzes RanGDP nuclear import, and the concentration of nuclear RanGTP is ultimately increased because of the activity of the RCC1 exchange factor.
Although the RanGTP gradient is required for both import and export, components of the pathway that regulate the gradient have an effect on directionality of cargo nuclear transport. For example, decrease in NTF2 has a primary effect on nuclear import. Conditional alleles of yeast ntf-2 show defects in protein nuclear import [9]. Also, depletion of NTF2 using antibodies inhibits nuclear import of NLS-containing proteins in HeLa cells [10]. In Drosophila, partial loss of function of ntf-2 affects nuclear import of Rel proteins in immune response and some loss of function alleles show a strong eye phenotype [11,12]. It has been shown in vertebrates that both low levels as well as increased levels of NTF2 impair nuclear import [13-15].
In contrast to NTF2, the intracellular localization of RanGAP appears to be critical for nuclear export [16]. In interphase cells RanGAP is localized to the cytoplasm and a large fraction of the cytoplasmic protein is modified by the ubiquitin related protein SUMO and localized to fibers of the NPCs [17,18]. This strategic position of RanGAP is thought to control the steep concentration gradient of RanGTP across the nuclear envelope.
Mislocalization of RanGAP can reduce nuclear RanGTP levels and lead to reduction of NES-mediated nuclear export, as happens in Segregation distortion (Sd) mutants in Drosophila [16]. In Sd mutants Sd-RanGAP, an enzymatically active protein refractory to SUMO modification, is expressed in addition to wild-type RanGAP. In these mutants RanGAP is found in higher levels in the cytoplasm and is also detected in nuclei. Sd-RanGAP presumably catalyzes hydrolysis of RanGTP in the nuclei thereby interfering with cargo export. A similar effect can be caused by over-expression of wild-type RanGAP [19].
Ran, in addition to functioning in nuclear-cytoplasmic transport, controls mitotic spindle formation. RanGTP stimulates polymerization of microtubules and RanGAP is found to associate with mitotic spindles [20,21]. Ran also functions in nuclear envelope and nuclear pore assembly [22,23]. These functions require the activities of RanGAP and RCC1, but the contribution of NTF2 is so far unclear.
In a genetic screen we identified mutants in Drosophila Profilin as modifiers of the partial loss of function eye phenotype of ntf-2. We find that Profilin is essential for normal nuclear export. This is surprising because the main function of Profilin is to control actin polymerization. RanGAP controls nuclear export and we find that gain of function mutants in Sd-RanGAP also suppress the ntf-2 eye phenotype. Our studies suggest a close connection between the organization of the actin cytoskeleton and nuclear transport.
Results
NTF2 and eye development
ntf-2 is an X-linked essential gene. Depending on the allele, animals die between the 2nd larval instar and the pupal stage. Some alleles have an adult survival rate of 8–15% of expected (Table 1), and all survivors show a small eye phenotype, strongly reduced numbers of ommatidia [11,12,24]. The eye phenotype varies from 30% of normal size to a more severe phenotype displaying one or two small patches of 10–40 ommatidia (Fig. 1A).
Table 1 Genetic interactions between ntf-2, chic, and Sd. The genetic crosses performed (top of the table). #s in parenthesis indicate columns in the table. The progeny resulting from the cross of ntf-2/FM7 females to males carrying the suppressing chromosome were counted. Four alleles of ntf-2 are listed in column 1. The genotype of the suppressing chromosomes are presented in the column 2. The numbers of trans-heterozygous females (ntf-2/+; Su/+; column 3), of non-mutant males (FM7/Y; Su/+, column 4), and of potentially suppressed mutant males (ntf-2/Y; Su/+) are presented. The ntf-2/+; Su/+ males were divided into two columns (5 and 6) depending on their eye phenotype and the viability of these males is indicated (100% X ntf-2 males/ntf-2/+ females). The number of progeny of the cross shown on the top and marked by * are not shown.
Parents
ntf-2 (1)/FM7c; +/+ × +/Y; Su(2)/Balancer
Progeny
ntf-2/+; Su/+(3) ntf-2/Y; Su/+ (5, 6)
FM7c/+; Su/+ * FM7c/Y; Su/+ (4)
ntf-2/+; Balancer/+* ntf-2/Y; Balancer/+*
FM7c/+; Balancer/+* FM7c/Y; Balancer/+
ntf-2 allele (1) Su (2) chromosome Cytology (2) ntf-2/ +; Su/+ (3) FM7/Y;Su/+ (4) ntf-2/ Y; Su/+ small eyes (5) ntf-2/Y; Su/+ normal eyes (6) ntf-2/Y; viability Suppression of ntf-2 eye
P49 + or Balancer 508 328 72 0 14% -
P7 + or Balancer 521 325 62 0 12% -
G0086 + or Balancer 503 154 42 0 8% -
G0337 + or Balancer 452 180 39 0 9% -
P49 Df(2R)Px2 60B; 60D 50 13 4 19 46% +
P7 Df(2R)Px2 60B; 60D 104 28 10 1 11% +
P49 In(2LR)Px[4] 60C-60D,21-22A 51 14 4 15 39% +
P7 In(2LR)Px[4] 60C-60D,21-22A 155 53 6 1 5% +
P7 P{lacW}l(2)04111 [k13009] 22A 68 24 0 4 6% +
P49 Df(2L)cl-h2 25D; 25F 9 5 0 10 100% +
P7 Df(2L)cl-h2 25D; 25F 23 13 0 13 56% +
P49 Df(2L)GpdhA 25D; 26A 66 14 2 2 6% +
P7 Df(2L)GpdhA 25D; 26A 49 27 1 2 6% +
P49 chicK13321 26A 53 25 3 1 8% +
P7 chicK13321 26A 67 40 1 11 18% +
P49 chic221 26A 30 12 0 10 33% +
P7 chic221 26A 65 48 8 3 17% +
G0086 chic221 26A 43 22 0 6 14% +
G0337 chic221 26A 69 44 2 4 9% +
P49 chic01320 26A 46 35 0 4 9% +
P49 chic2 26A 74 68 0 3 4% +
P7 chic2 26A 43 33 4 1 12% +
P49 In(2R)SD72, In(2R)NS, Sd[72] 35 30 0 12 34% +
G0086 In(2R)SD72, In(2R)NS, Sd[72] 40 7 0 2 5% +
P49 UAS-RanGAP12A-6, hsp70-GAL4 16 6 0 5 31% +
P7 UAS-RanGAP12A-6, hsp70-GAL4 26 11 0 7 27% +
P7 UAS-RanGAP12A-6, arm-GAL4 28 12 1 4 18% +
Figure 1 The ntf-2 eye phenotype is rescued by mutants in Profilin (chic). (A) Wild-type eye, a representative ntf-2 eye and the phenotype of a ntf-2 eye suppressed by chic/+. Note that the antennae (arrow) are normal in mutant animals. (B) Wild-type and ntf-2 eye-antennal discs. The antennal discs (ant) are normal in wild-type and mutant, while the ntf-2 eye disc (eye) shows abnormal growth and patterning. Size bar represents 10 μm.
The mutant eye-imaginal discs are smaller than wild-type and are often abnormally shaped (Fig. 1B). Overall, the structure of the mutant eye discs is perturbed and the organization of the actin cytoskeleton is strongly altered (compare Fig. 2A and 2B,C). Only few disorganized, irregularly spaced rabdomere-like structures are apparent in the posterior compartment of the eye disc (arrow in Fig. 2A–C).
Figure 2 The ntf-2 eye dics are disorganized. Wild-type eye disc (A, D; arrowhead indicates morphogenetic furrow, arrow indicates rabdomeres). In ntf-2 mutants (B, C, E, F) the furrow fails to move and fewer rabdomeres are formed; the organization of the actin cytoskeleton (green) and distribution of RanGAP (red) look abnormal. Squares are magnified in panels D, E, F. In all Figures DNA is shown in blue and the size bar represents 10 μm.
A deficiency screen to identify dominant suppressors of ntf-2
We took advantage of the partial loss of function eye phenotype of ntf-2 alleles to identify genes functioning with ntf-2, and performed a dominant suppressor screen of the eye phenotype. Males from 2nd and 3rd chromosomal deficiency stocks (deficiency/balancer) uncovering 70% to 80% of the two autosomes, or about 60% of the Drosophila genome, were crossed with ntf-2P7/FM7 females (Table 1 top). In the next generation the number of surviving ntf-2 males also carrying a deletion was counted and the survivors monitored for their eye phenotype. For our screen we set up 136 individual crosses, many of them repeatedly in order to obtain at least 150 adult progeny to screen for the eye phenotype. We only identified deletions and rearrangements in four regions of the second chromosome that showed suppression (Table 1). The suppression was confirmed using a second ntf-2 (P49) allele.
DNA rearrangements affecting regions 22A and 60B-D showed different results with the two ntf-2 alleles tested and were not pursued. Df(2l)cl-h2 (25D-F) appeared to rescue both viability and the eye phenotype, but the gene responsible for the suppression could not be identified. Df(2L)GpdhA (25D-26A) rescued the eye phenotype, but not viability. To identify the gene(s) responsible for the suppression of the eye phenotype we tested mutations in several genes that are uncovered by Df(2L)GpdhA and are available from the Drosophila stock center.
Mutants in one gene, chickadee (chic), encoding Drosophila Profilin [25], uncovered by Df(2L)GpdhA, showed suppression of the ntf-2 eye phenotype. We tested several loss-of-function alleles of chic, including a complete lethal null allele (chic221) and other partially viable alleles, that are either female, or male and female sterile. All chic alleles were crossed with at least 2 ntf-2 alleles, except chic221 that was tested with 4 different ntf-2 alleles. The suppression of the eye phenotype was observed in all crosses and the majority of surviving trans-heterozygous males showed suppression of the ntf-2 eye phenotype, restoration of wild-type eyes (Fig. 1A). The percent of males with wild-type eyes varied in different allele combinations. Surprisingly, the eye phenotype was usually either small or wild-type and virtually no eyes of intermediate size were observed.
Mutations in chic (Profilin) affect nuclear export
To investigate the cause underlying the suppression of the ntf-2 phenotype and possible function of Profilin in nuclear transport, we used a reporter gene approach. We assayed nuclear transport using UAS-NLS-NES reporter constructs C-terminally tagged with GFP in different mutant backgrounds. One construct contains a wild-type NLS and NES (UAS-NLS-NES-GFP), the other a wild-type NLS but a mutant NES that is not recognized by the nuclear export machinery (UAS-NLS-NESP12-GFP; [16,26]). Expression of the transgenes was driven by a heatshock-GAL4 driver, and the distribution of GFP was analyzed in salivary glands. As previously shown, the activity of the wild-type NES is stronger then that of the NLS [26]. Hence, in wild-type the NLS-NES-GFP is usually localized in the cytoplasm (Fig. 3B). In contrast, NLS-NESP12-GFP has impaired nuclear export and strongly accumulates in nuclei (Fig. 3A). In homozygous chic01320 and the hetero-allelic combination chic2/chic221, the distribution of the GFP reporter is altered. In contrast to the cytoplasmic distribution of NLS-NES-GFP in wild-type, in the chic mutant salivary glands the GFP reporter is found predominantly in the nucleus (Fig. 3D,F). The localization of NLS-NESP12-GFP is similar in chic and wild-type (Fig. 3C,E), indicating that NLS-mediated import is not affected.
Figure 3 Chic impairs nuclear export. Localization of GFP reporter appended with wild-type NLS and NES (B, D, F) or mutant NESP12 (A, C, E) in salivary gland cells. In wild-type (A) and chic mutants (C, E) NLS-NESP12-GFP is predominantly localized to the nucleus. In wild-type salivary glands NLS-NES+-GFP is mostly found in the cytoplasm (B). In contrast, in chic mutants, chic2/chic221 and chic01320 homozygotes, NLS-NES+-GFP is predominantly nuclear (D, F).
Sd (RanGAP) suppresses the ntf-2 phenotype
RanGAP functions in nuclear export of cargo and in Sd-RanGAP mutants the NLS-NES-GFP is found in the nucleus and NLS-NESP12-GFP is distributed the same as in wild-type [16,19]. This failure of exporting NLS-NES-GFP in Sd-RanGAP mutants is reminiscent of what we observe in chic alleles (Fig. 3).
Given the similarity in nuclear export phenotypes in Sd and chic mutants, we tested if Sd would also suppress the eye phenotype of ntf-2 alleles. We crossed the Sd (Sd72, [27]) chromosome with two ntf-2 alleles and found that the eye phenotype was suppressed in both of them. To confirm that the SD-RanGAP mutation, and not other genes on the Sd chromosome, is responsible for the suppression, we expressed a mutated Sd-RanGAP transgene (UAS-Sd-RanGAP12A-6; [16]) driven by hsp70-GAL4 or arm-GAL4 in ntf-2P7 and ntf-2P49 males and observed similar levels of suppression as seen with Sd72 (Table 1).
The genetic interaction between Sd-RanGAP and ntf-2 is not altogether surprising because both RanGAP and NTF2 are known to function in the formation of the RanGTP-GDP gradient. To investigate if RanGAP is affected in ntf-2 mutants we studied the distribution of RanGAP in eye discs.
In wild-type cells Ran-Gap is present in low levels in the cytoplasm and forms a clearly visible punctuated circle around the nucleus (Fig. 2A,D). The punctuate pattern of RanGAP is due to its association with nuclear pores [17,18]. This distribution is different in ntf-2 discs. Patches of cells are observed in which RanGAP aggregates in small or large clumps near the nuclei (Fig 2E,F), but in other cells the distribution of the protein looks relatively normal. This observation suggests, that the clumping of RanGAP is an effect of the abnormal organization of the cells within the ntf-2 disc. The cells with clumped RanGAP are usually in close proximity to cells with high levels of F-actin.
Lack of Profilin also affects RanGAP distribution
To investigate a connection between Profilin, RanGAP, and actin, we next asked whether the function of Profilin or actin polymerization might have an effect on RanGAP localization. We generated clones in eye discs of null alleles of the two genes chic (chic221) and, as a control, act up (acuE636). Acu participates in actin de-polymerization, the opposite function of Profilin [28].
In chic clones RanGAP protein is increased around the nuclear envelope and its distribution is uneven and patchy on the nuclear envelope surface (Fig. 4D arrows). In wild-type even, punctuated circles are observed (arrowheads). This abnormal distribution was found in 100% of examined clones (more then 50). In chic clones the level of F-actin was reduced as previously shown [28]. In the acu control clones high levels of F-actin are detected as expected (results not shown, [28]), but the distribution of RanGAP is not significantly changed (Fig. 4E).
Figure 4 Localization of RanGAP on chic nuclear envelopes is irregular. chic221 eye disc clones show only minor changes in distribution of (A) Lamin (red) and (B) Nups (red). But RanGAP localization (C, D red) is strongly altered appearing more clustered and patchy (arrows) than in wild-type cells, where RanGAP forms uniform dotted rings around nuclei (arrowheads). act up loss of function does not affect RanGAP (red) in acuE636 eye disc clones (E). Mutant clones are marked by the absence of green. The borders of some clones are highlighted with dashed lines.
To test whether this patchy protein distribution of RanGAP on nuclear pores of chic22 cells is caused by problems in nuclear envelope assembly, we analyzed the distribution of Lamin and nuclear pore proteins (Nups) in chic221 clones (Fig. 4A,B). The distribution of both Lamin and Nups is affected in about 30% of clones. This is likely due to the mislocalization of RanGAP. It has been shown previously that RanGTPase functions in nuclear pore and envelope formation [22,23].
The staining experiments show higher levels of RanGAP around nuclei in chic eye disc clones. We investigated if this is due to overall higher levels of RanGAP in mutant cells. The chic alleles used in the clonal analysis are homozygous lethal therefore we prepared extracts from wild-type and mutant 1st instar larvae. In western blots from extracts of chic221 (lethal at first and early second larval instar) and chic01320 (viable and female sterile) larvae, the amount of RanGAP present in mutants is not dramatically changed compared to wild-type (results not shown). This may be because RanGAP and Profilin are maternally contributed and therefore at these early stages a difference in levels is not detected. We then dissected eye-antennal discs from normal larvae and larvae with chic clones (see experiments shown in Figure 4). The dissected tissues also contained some brain material because eye-antennal discs are next to the brain hemispheres and are difficult to separate. In two separate experiments we see an increase of 30–50% in the intensity of the RanGAP band in extracts from discs carrying chic221 somatic clones compared to normal eye discs from chic221/+ larvae. The intensity of the RanGAP bands were normalized to that of the control Bic-D band and equals 2.6 for discs with clones and 1.8 for wild-type discs (Fig. 5).
Figure 5 RanGAP levels in chic mutants. Western blot of extracts from eye-antennal discs and some brain lobes. Compare the amount of RanGAP in chic221/+ tissue without somatic clones and chic221/+ with somatic clones; chic null tissue, represents 10–30% of eye discs cells and 2–5% of antennae and brain cells. The extracts were loaded at two different concentrations to allow comparison of levels of protein.
Discussion
NTF2 regulates nuclear import in every cell of the organism. Some ntf-2 alleles can produce male sterile but female fertile adults, that all have a striking eye phenotype. This phenotype appears to be caused by lower levels of NTF2 and not an altered protein since alleles showing the eye phenotype have a P-element insertion in the 5' UTR [11,12,24]. The difference in response of tissues to the lower levels of NTF2 is surprising. For instance, the wild-type and mutant antennal discs and antennae appear normal (Fig. 1A arrows), while the eye disc of the mutant has a strongly modified appearance.
We used partial loss of function ntf-2 alleles to screen for dominant suppressors of the eye phenotype. Using deletions uncovering more then half of the genome, we identified four regions that can function in the control of the RanGTP-RanGDP balance. While unable to identify specific genes responsible for the dominant interaction in three of these regions, we did find one suppressor, chic, encoding Profilin.
We tested several alleles of chic with several alleles of ntf-2 and observed suppression of the eye phenotype in all combinations. Reducing Profilin suppresses the effects of lower than normal amounts of NTF2. In all cases we observe either small eyes or completely restored eyes similar to wild-type. We detected no intermediate phenotypes, suggesting that a threshold level exists for each protein.
The ntf-2 eye phenotype can also be suppressed by a gain of function mutation in Sd (RanGAP). RanGAP regulates the Ran-GTP-to-Ran-GDP balance and is involved in nuclear export. NTF2 controls nuclear import of RanGDP and thus the nuclear trafficking of cargo. Because of the low viability of ntf-2 mutants we cannot obtain ntf-2 flies expressing a driver and NLS-NES-GFP reporter genes. Hence, the nuclear transport phenotype of ntf-2 alleles cannot be determined as we did for chic mutants, and was done for Sd mutants [16]. Previous investigations in several organisms indicate that in ntf-2 mutants, the Ran gradient or its formation is changed affecting cargo import [13-15].
Why lowering the level of Profilin that functions in actin polymerization suppresses the ntf-2 phenotype is not immediately apparent, but there are several possible explanations. Lower levels of Profilin may result in reduction of the abnormal actin polymerization in ntf-2 mutant eye discs (see Fig. 2). But our finding that the ntf-2 eye phenotype is suppressed by the over-expression of RanGAP suggests that the disorganized appearance of F-actin is an indirect result of abnormal nuclear trafficking. Therefore lowering Profilin seems to also affect the abnormal nuclear trafficking inherent to ntf-2 eye discs. This supposition is bolstered by our finding that Profilin is essential for normal nuclear export. Our results are consistent with F-actin being regulated by nuclear transport, and in turn, Profilin and Actin controlling aspects of nuclear trafficking.
Unpolymerized actin is found on NPC-attached nucleoplasmic filaments. It has been shown to function in the nuclear export of proteins and RNA [29]. Unpolymerized actin also associates with Profilin and is exported from the nuclei in a Ran-dependant manner [30]. We do not think that these processes have a primary role in our mutant phenotypes because staining of ntf-2 eye discs and chic clones with anti-actin antibody display no obvious difference in the distribution of non-polymerized actin (results not shown). Nevertheless, these processes have to be considered as part of the crosstalk between the actin cytockeleton and Ran-mediated nuclear trafficking.
That Profilin controls the localization of RanGAP is evident from the abnormal distribution of the protein in chic clones. The uneven distribution of RanGAP at the nuclear envelope is not due simply to higher levels of protein. In Sd transgenic lines that express wild-type or mutant RanGAP, higher levels of protein are found uniformly distributed in the cytoplasm and nucleus [19]. In chic mutant cells, the RanGAP level is about doubled, but the protein distribution is different than that observed in the over-expressing lines.
Methods
Drosophila stocks and suppressor screen
The w118 stock was used as wild-type stock (WT) in all experiments. All fly stocks were obtained from the Bloomington Stock Center [24], except UAS-RanGAP12A-6, UAS-NLS-NESP12-GFP and UAS-NLS-NES-GFP transgenic flies that were sent by Barry Ganetzky and Edwin Chan [16,31].
The following ntf-2 stocks were used for suppression experiments: w ntf-2P7/FM7i, P{w [+mC]=ActGFP}JMR3, y[1] f[1], w ntf-2P49/FM7c, w ntf-2G0086/FM7c, ntf-2G0337/FM7c. All alleles carry different P-element insertions in the 5' UTR of the gene, 102–112 bases upstream from the start of transcription that cause reduced viability and the eye phenotype (see Table 1, [11,12,24]).
Suppressor screen
To identify suppressors of ntf-2, ntf-2P7/FM7 and ntf-2P49/FM7 virgins were crossed with deficiency males. In the next generation the ntf-2 males not carrying a balancer chromosomes were analyzed. We tested 66 deficiencies on the second chromosome and about 70 deficiencies on the third chromosome, and about 25 existing mutations and P-element insertions mapping to the 22–30 cytological region of chromosome two. All were obtained from the Bloomington Stock center.
Alleles of chickadee, chicK1332, chic221, chic01320, chic2, In(2R)SD72, In(2R)NS, Sd[72], GAL4 drivers and Balancer chromosomes were obtained from Bloomington Stock Center. UAS-RanGAP12A-6, UAS-NLS-NESP12-GFP and UAS-NLS-NES-GFP transgenic flies were kindly sent to us by Barry Ganetzky and Edwin Chan [16].
Nuclear transport assay
To identify mutant animals at larval stages we used the GFP marked balancers FM7i P{w [+mC]=ActGFP}JMR3 and CyO, P{w [+mW.hs]=Ubi-GFP.S65T}PAD1. Mutant 2nd instar larvae were distinguished from their heterozygous siblings by the absence of green fluorescent protein. The expression of GFP-reporter genes were induced by 1 h incubation at 36°C. To diminish the possible effect of heating on nuclear trafficking we waited for 24 h before dissecting and fixing salivary glands. Localization of GFP was observed and fluorescence images obtained using a Zeizz Axioplan 2 (Zeiss) microscope and Image Pro Plus software.
Eye disc clones
FRT40 chic221 and FRT40 acuE636 chromosomes were obtained from Jessica Treisman. Clones in eye discs were generated as described [28].
Antibodies
Rabbit polyclonal antibody against Drosophila RanGAP was provided by Sinthia Stabber and Barry Ganetzki [16], mab414 recognizing nucleoporins with FXF repeats [32,33] was obtained from BAbCO (Richmond, CA), and anti-Lamin antibody was a gift from Paul Fischer, Stonybrook. Western analysis was performed as previously described [16,17].
Immunostaining
For antibody staining 3rd instar larvae were inverted in phosphate-buffered saline (PBS) and immediately fixed in 4% paraformaldehyde in PBS with 2% DMSO for 40 min and washed several times in PBT (PBS, 0.1% Triton X-100). Then tissues were blocked for 2 hours in PBS containing 1% bovine serum albumin (BSA) and 1% Triton X-100. Antibody incubations were done in PBT with 1% BSA overnight at 4°C. Anti-RanGAP rabbit serum was used at a 1:1000 or 1:800 dilution, and anti-RanGAP-1 monoclonal antibody was used at a 1:400 dilution, mab414 was used at a 1:300 dilution, anti-Lamin – 1:30. Secondary antibodies were used at a dilution of 1:500. Cy-3 conjugated anti-mouse and anti-rabbit IgG were purchased from Jackson Immuno Research Laboratories Inc. (West Grove, PA). F-actin was visualized by incubation with Alexa488 Phalloidin at 1:80 for 2 hours. DNA was stained using Hoechst 33258 (Molecular Probes, Eugene, OR). Samples were mounted in Vectashield (Vector Laboratories) and examined with a Leica DM IRBE (Leica) laser scanning confocal microscope. The images were analyzed with Leica Microsystems software and further processed using Adobe PhotoShop.
Authors' contributions
SM participated in the design of the study, and was involved in all experiments presented. RM carried out the suppressor screen and characterized chic as a suppressor of ntf-2. MD carried out the immunoassays. RS conceived the study, worked on the genetic screens. SM and RS wrote the paper. All authors read and approved the final manuscript.
Acknowledgements
We thank Sinthia Stabber, Edwin Chan, Barry Ganetzki, Jessica Treisman, and Paul Fischer for transgenic flies and antibodies. We are grateful to Ursula Stochaj, Frank Solomon, and Chris Rongo for helpful suggestion on the manuscript. We also thank Le Nguyen, Ananya Bhattacharya, Sarika Patel and Sini Abraham for initial work and technical help. This work was supported by a grant from the NIH and by the W. Horace Goldsmith Foundation.
==== Refs
Weis K Regulating access to the genome: nucleocytoplasmic transport throughout the cell cycle Cell 2003 112 441 451 12600309 10.1016/S0092-8674(03)00082-5
Gorlich D Kutay U Transport between the cell nucleus and the cytoplasm Annu Rev Cell Dev Biol 1999 15 607 660 10611974 10.1146/annurev.cellbio.15.1.607
Bischoff FR Ponstingl H Mitotic regulator protein RCC1 is complexed with a nuclear ras-related polypeptide Proc Natl Acad Sci U S A 1991 88 10830 10834 1961752
Ribbeck K Lipowsky G Kent HM Stewart M Gorlich D NTF2 mediates nuclear import of Ran Embo J 1998 17 6587 6598 9822603 10.1093/emboj/17.22.6587
Paschal BM Gerace L Identification of NTF2, a cytosolic factor for nuclear import that interacts with nuclear pore complex protein p62 J Cell Biol 1995 129 925 937 7744965 10.1083/jcb.129.4.925
Moore MS Blobel G Purification of a Ran-interacting protein that is required for protein import into the nucleus Proc Natl Acad Sci U S A 1994 91 10212 10216 7937864
Smith A Brownawell A Macara IG Nuclear import of Ran is mediated by the transport factor NTF2 Curr Biol 1998 8 1403 1406 9889103 10.1016/S0960-9822(98)00023-2
Gorlich D Seewald MJ Ribbeck K Characterization of Ran-driven cargo transport and the RanGTPase system by kinetic measurements and computer simulation Embo J 2003 22 1088 1100 12606574 10.1093/emboj/cdg113
Corbett AH Silver PA The NTF2 gene encodes an essential, highly conserved protein that functions in nuclear transport in vivo J Biol Chem 1996 271 18477 18484 8702493 10.1074/jbc.271.31.18477
Steggerda SM Black BE Paschal BM Monoclonal antibodies to NTF2 inhibit nuclear protein import by preventing nuclear translocation of the GTPase Ran Mol Biol Cell 2000 11 703 719 10679025
Bhattacharya A Steward R The Drosophila homolog of NTF-2, the nuclear transport factor-2, is essential for immune response EMBO Rep 2002 3 378 383 11943764 10.1093/embo-reports/kvf072
Peter A Schottler P Werner M Beinert N Dowe G Burkert P Mourkioti F Dentzer L He Y Deak P Benos PV Gatt MK Murphy L Harris D Barrell B Ferraz C Vidal S Brun C Demaille J Cadieu E Dreano S Gloux S Lelaure V Mottier S Galibert F Borkova D Minana B Kafatos FC Bolshakov S Siden-Kiamos I Papagiannakis G Spanos L Louis C Madueno E de Pablos B Modolell J Bucheton A Callister D Campbell L Henderson NS McMillan PJ Salles C Tait E Valenti P Saunders RD Billaud A Pachter L Klapper R Janning W Glover DM Ashburner M Bellen HJ Jackle H Schafer U Mapping and identification of essential gene functions on the X chromosome of Drosophila EMBO Rep 2002 3 34 38 11751581 10.1093/embo-reports/kvf012
Feldherr C Akin D Moore MS The nuclear import factor p10 regulates the functional size of the nuclear pore complex during oogenesis J Cell Sci 1998 111 ( Pt 13) 1889 1896 9625751
Hu W Jans DA Efficiency of importin alpha/beta-mediated nuclear localization sequence recognition and nuclear import. Differential role of NTF2 J Biol Chem 1999 274 15820 15827 10336485 10.1074/jbc.274.22.15820
Tachibana T Hieda M Sekimoto T Yoneda Y Exogenously injected nuclear import factor p10/NTF2 inhibits signal-mediated nuclear import and export of proteins in living cells FEBS Lett 1996 397 177 182 8955342 10.1016/S0014-5793(96)01180-5
Kusano A Staber C Ganetzky B Nuclear mislocalization of enzymatically active RanGAP causes segregation distortion in Drosophila Dev Cell 2001 1 351 361 11702947 10.1016/S1534-5807(01)00042-9
Mahajan R Delphin C Guan T Gerace L Melchior F A small ubiquitin-related polypeptide involved in targeting RanGAP1 to nuclear pore complex protein RanBP2 Cell 1997 88 97 107 9019411 10.1016/S0092-8674(00)81862-0
Matunis MJ Coutavas E Blobel G A novel ubiquitin-like modification modulates the partitioning of the Ran-GTPase-activating protein RanGAP1 between the cytosol and the nuclear pore complex J Cell Biol 1996 135 1457 1470 8978815 10.1083/jcb.135.6.1457
Kusano A Staber C Ganetzky B Segregation distortion induced by wild-type RanGAP in Drosophila Proc Natl Acad Sci U S A 2002 99 6866 6870 11997467 10.1073/pnas.102165099
Ohba T Nakamura M Nishitani H Nishimoto T Self-organization of microtubule asters induced in Xenopus egg extracts by GTP-bound Ran Science 1999 284 1356 1358 10334990 10.1126/science.284.5418.1356
Wilde A Lizarraga SB Zhang L Wiese C Gliksman NR Walczak CE Zheng Y Ran stimulates spindle assembly by altering microtubule dynamics and the balance of motor activities Nat Cell Biol 2001 3 221 227 11231570 10.1038/35060000
Walther TC Askjaer P Gentzel M Habermann A Griffiths G Wilm M Mattaj IW Hetzer M RanGTP mediates nuclear pore complex assembly Nature 2003 424 689 694 12894213 10.1038/nature01898
Zhang C Goldberg MW Moore WJ Allen TD Clarke PR Concentration of Ran on chromatin induces decondensation, nuclear envelope formation and nuclear pore complex assembly Eur J Cell Biol 2002 81 623 633 12494999 10.1078/0171-9335-00288
The FlyBase C The FlyBase database of the Drosophila genome projects and community literature Nucleic Acids Research 2003 31 172 175 http://flybase.org/ 12519974 10.1093/nar/gkg094
Cooley L Verheyen E Ayers K chickadee encodes a profilin required for intercellular cytoplasm transport during Drosophila oogenesis Cell 1992 69 173 184 1339308 10.1016/0092-8674(92)90128-Y
Stade K Ford CS Guthrie C Weis K Exportin 1 (Crm1p) is an essential nuclear export factor Cell 1997 90 1041 1050 9323132 10.1016/S0092-8674(00)80370-0
Powers PA Ganetzky B On the components of segregation distortion in Drosophila melanogaster. V. Molecular analysis of the Sd locus Genetics 1991 129 133 144 1936954
Benlali A Draskovic I Hazelett DJ Treisman JE act up controls actin polymerization to alter cell shape and restrict Hedgehog signaling in the Drosophila eye disc Cell 2000 101 271 281 10847682 10.1016/S0092-8674(00)80837-5
Hofmann W Reichart B Ewald A Muller E Schmitt I Stauber RH Lottspeich F Jockusch BM Scheer U Hauber J Dabauvalle MC Cofactor requirements for nuclear export of Rev response element (RRE)- and constitutive transport element (CTE)-containing retroviral RNAs. An unexpected role for actin J Cell Biol 2001 152 895 910 11238447 10.1083/jcb.152.5.895
Stuven T Hartmann E Gorlich D Exportin 6: a novel nuclear export receptor that is specific for profilin.actin complexes Embo J 2003 22 5928 5940 14592989 10.1093/emboj/cdg565
Kusano A Staber C Chan HY Ganetzky B Closing the (Ran)GAP on segregation distortion in Drosophila Bioessays 2003 25 108 115 12539236 10.1002/bies.10222
Davis LI Blobel G Identification and characterization of a nuclear pore complex protein Cell 1986 45 699 709 3518946 10.1016/0092-8674(86)90784-1
Aris JP Blobel G Yeast nuclear envelope proteins cross react with an antibody against mammalian pore complex proteins J Cell Biol 1989 108 2059 2067 2661560 10.1083/jcb.108.6.2059
|
16120220
|
PMC1215477
|
CC BY
|
2021-01-04 16:31:30
|
no
|
BMC Cell Biol. 2005 Aug 24; 6:32
|
utf-8
|
BMC Cell Biol
| 2,005 |
10.1186/1471-2121-6-32
|
oa_comm
|
==== Front
BMC Clin PatholBMC Clinical Pathology1472-6890BioMed Central London 1472-6890-5-81614657610.1186/1472-6890-5-8Case ReportPersistence of Natural Killer (NK) cell lymphocytosis with hyposplenism without development of leukaemia Khan Sujoy [email protected] K [email protected] Department of Haematology, Prince Charles Hospital, Merthyr Tydfil, Wales, CF47 9DT UK2 Department of Immunopathology, St. Bartholomew's Hospital, 51-53 Bartholomew's Close, West Smithfield, London, EC1A 7BE UK2005 7 9 2005 5 8 8 15 8 2004 7 9 2005 Copyright © 2005 Khan and Myers; licensee BioMed Central Ltd.2005Khan and Myers; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Natural killer (NK) cell lymphocytosis usually has an indolent course and can progress into massive lymphocytosis with development of cytopenias and neoplastic diseases. NK-cells usually express one or more "NK-associated" antigens (CD16, CD56, CD57). Reactive expansions are seen in autoimmune diseases, viral infections, solid tumours and non-Hodgkin's lymphoma.
Case presentation
We report a lady with a benign clinical course over 10 years and persistent CD8+/CD3-/CD57+/CD16+ LGL proliferation with presence of Howell-Jolly bodies (functional hyposplenism), an association not previously described.
Conclusion
We discuss the possible causes of clonal expansion and conclude that this may be part of the spectrum of immune dysregulation associated with NK-cell lymphocytosis.
==== Body
Background
The lymphoproliferative disease of granular lymphocytes (LDGL) results from the chronic proliferation of large granular lymphocytes (LGL) that may result from antigenic stimulation1,2. Natural Killer (NK) cells constitute approximately 15% of the peripheral blood mononuclear cell fraction. NK cells lack both CD3 and T-cell receptor expression, majority express CD56 and/or CD16 (Fcγ receptor). Granular lymphocytosis greater than 2,000/μL lasting for more than 6 months is regarded as the criteria to define the disease [1,2]. Patients with chronic, indolent NK lymphocytosis may develop cytopenias, splenomegaly, vasculitic skin lesions, and peripheral neuropathy [3]. We discuss a unique case of chronic, indolent NK lymphocytosis who presented with severe hyposplenism and has not developed leukaemia over a decade.
Case presentation
A 46-year-old lady was referred to the haematology clinic for evaluation of lymphocytosis in May 1993. She had severe lethargy and intermittent right upper abdominal discomfort without any significant loss in weight. Her past medical history included essential hypertension controlled on atenolol 100 mg once daily and was also on frusemide 40 mg once daily. She had no significant surgical history other than having undergone cholecystectomy in 1972. She had never smoked nor consumed alcohol. Physical examination showed no evidence of lymphadenopathy. Complete blood count showed normal haemoglobin concentration 14.8 g/dl, macrocytosis (MCV 100.1), raised white cell count at 13.4 × 109/L, lymphocytosis (absolute number 6.3 × 109/L), and normal neutrophil count (absolute number 5.6 × 109/L). Peripheral blood showed numerous Howell-Jolly bodies within erythrocytes. Thyroid function tests, protein electrophoresis, C-reactive protein, immunoglobulin levels and autoimmune screening were normal. Ultrasonography and computed tomography scan of the abdomen and pelvis did not reveal retroperitoneal or mediastinal lymphadenopathy, but the spleen was noted to be very atrophic. Gastroscopy showed multiple gastric erosions and the initial impression was of celiac disease complicated by lymphoma and lymphocytosis.
Duodenal biopsy showed well-formed villi and no increase in intraepithelial lymphocytes thereby excluding celiac disease. Colonoscopy and barium studies were normal. She was clinically diagnosed to have functional hyposplenism, considering the presence of Howell-Jolly bodies, and was given hemophilus influenzae (HiB) vaccine, pneumovax vaccine and counselled for long term oral Penicillin V. In November 1994, another complete blood count showed a white cell count of 15.3 × 109/L with absolute number of lymphocytes of 6.6 × 109/L. Bone marrow biopsy showed 25% infiltration with large granular lymphocytes. Peripheral blood smear revealed that the majority of lymphocytes had atypical morphology with large atypical nuclei and abundant cytoplasm containing fine azurophilic granules. Immunophenotyping showed CD16+/CD3- cells which were mainly reactive and another clonal cells which were CD57+/CD8+ suggesting NK-LGL (Natural Killer-Large Granular Lymphocyte) activity. 80% CD3+ cells were α/β and 20% γ/δ and a clonal population of CD16+/CD56+ cells.
She presented in March 1995 with lethargy and diffuse enlargement of her thyroid gland. Lymphocytosis persisted (WBC 15.7 × 109/L, absolute lymphocyte count 8.2 × 109/L), antimicrosomal antibody was positive at a titre of 1:6400, anti thyroglobulin antibody was positive at 1:1280 but thyroid function tests (TSH 1.69, fT4 13.0) and isotope thyroid scan I123 were normal. Immunoglobulins, C3, C4 and C-reactive protein levels were normal. There was spontaneous regression of her thyroid swelling by December 1996 at which time antimicrosomal antibody was negative. Follow-up for 10 years showed the persistence of lymphocytosis (February 2004-WBC 12.5 × 109/L, absolute lymphocyte count 5.4 × 109/L) without development of autoimmune disease, autoantibodies, neutropenia or infections.
Conclusion
This is the first report of the unusual occurrence of Howell-Jolly bodies in a patient with persistent NK-LGL lymphocytosis. She did not develop autoimmune disease, neutropenic infections or vasculitic syndromes but the lymphocytosis persisted.
Cytogenetic, molecular analyses and monoclonal antibodies (MoAbs) both against the V-gene regions of the T-cell receptor (TCR) and the molecules of the p58 family on NK cells [4] are used to characterize the expanded cells. Polyclonal proliferation may represent a preneoplastic condition, since there is some evidence that suggests that it is probably a multistep process [5]. In such cases, the clonal expansion might evolve from the abnormal immunoregulation of a response to inciting antigens, for example, viruses [5].
The Yorkshire Leukaemia Group investigated 870 adults with 'persistent' LGL/NKa+ (Large Granular Lymphocyte/Natural Killer associated) expansions suggest that clonal expansions may be more frequent than reported and found high proportion with persistent neutropenia and all patients with CD8+NKa+ abnormalities had rearranged TCR genes [6]. Splenomegaly has been reported in 19 % in a case series of 68 patients with clonal T-LGL proliferations [7], but uniquely, in our case, the lymphocytosis has remained stable and Howell-Jolly bodies were observed. The presence of Howell-Jolly bodies and liver-spleen scintigraphy showing absence of spleen is considered diagnostic of functional hyposplenia [8]. An autoimmune mechanism seems probable but not certain [9].
Clonality does not imply malignancy. It has been detected during autoimmune processes, including rheumatoid arthritis, and in bone marrow transplantation recipients [9]. Clonal populations in such cases most likely represent epiphenomenon of an immunoregulatory disorder. There is evidence of persistent CD8+ clonal expansions in normal elderly individuals [10] and clonal V-alpha 12.1+ T-cell expansions in uncomplicated rheumatoid arthritis [11] that could mean that these proliferations are benign.
In summary, our lady with persistent NK cell lymphocytosis and hyposplenism had a benign clinical course over a decade. We report this case as presence of Howell-Jolly bodies has not been previously reported and postulate that the hyposplenism is probably a part of the spectrum of immune dysregulation associated with NK-LGL.
Abbreviations
LGL-large granular lymphocytes; NK-Natural Killer; TCR-T-cell receptor.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SK drafted the initial manuscript. KM made changes to the manuscript and as corresponding author had full access to all data and was responsible for the decision to submit for publication.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Epling-Burnette PK Bai F Wei S Chaurasia P Painter JS Olashaw N Hamilton A Sebti S Djeu JY Loughran TP ERK couples chronic survival of NK cells to constitutively activated Ras in lymphoproliferative disease of granular lymphocytes (LDGL) Oncogene 2004 23 9220 9 15516985
Lamy T Loughran TP Jr Clinical features of large granular lymphocyte leukemia Semin Hematol 2003 40 185 95 12876667 10.1016/S0037-1963(03)00133-1
Tefferi A Li CY Witzig TE Dhodapkar MV Okuno SH Phyliky RL Chronic natural killer cell lymphocytosis: a descriptive clinical study Blood 1994 84 2721 5 7919384
Moretta L Ciccone E Mingari MC Biassoni R Moretta A Human natural killer cells: Origin, clonality, specificity, and receptors Adv Immunol 1994 55 341 80 7508176
Zambello R Loughran TP JrTrentin L Pontisso P Battistella L Raimondi R Facco M Sancetta R Agostini C Pizzolo G Serological and molecular evidence for a possible pathogenetic role of viral infection in CD3-negative NK-type lymphoproliferative disease of granular lymphocytes Leukemia 1995 9 1207 11 7630196
Scott CS Richards SJ Sivakumaran M Short M Child JA Hunt KM McEvoy M Steed AJ Balfour IC Parapia LA Transient and persistent expansions of large granular lymohocytes (LGL) and NK-associated (NKa) cells: the Yorkshire Leukaemia Group Study Br J Haematol 1993 83 505 515 8485057
Dhodapkar MV Li CY Lust JA Tefferi A Phyliky RL Clinical Spectrum Of Clonal Proliferations of T-Large Granular Lymphocytes: A T-Cell Clonopathy of Undetermined Significance? Blood 1994 84 1620 7 8068951
Christian Görg Miriam Eichkorn Gerhard Zugmaier The small spleen: Sonographic patterns of functional hyposplenia or asplenia Journal of Clinical Ultrasound 2003 31 152 155 12594800 10.1002/jcu.10148
Wardrop CA Dagg JH Lee FD Singh H Dyet JF Moffat A Immunological abnormalities in splenic atrophy Lancet 1975 2 4 7 1975 Jul 5 49638 10.1016/S0140-6736(75)92949-9
Posnett DN Sinha R Kabak S Russo C Clonal populations of T cells in normal ederly humans: The T cell equivalent to "benign monoclonal gammopathy." J Exp Med 1994 179 609 (published erratum appears in J Exp Med 179:1077, 1994) 8294871 10.1084/jem.179.2.609
DerSimonian H Sugita M Glass DN Maier AL Weinblatt ME Reme T Brenner MB Clonal V-alpha 12.1+ T-cell expansions in peripheral blood of rheumatoid arthritis patients J Exp Med 1993 177 1623 8496681 10.1084/jem.177.6.1623
|
16146576
|
PMC1215478
|
CC BY
|
2021-01-04 16:27:55
|
no
|
BMC Clin Pathol. 2005 Sep 7; 5:8
|
utf-8
|
BMC Clin Pathol
| 2,005 |
10.1186/1472-6890-5-8
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.