Does anti-mutated citrullinated vimentin have additional value as a serological marker in the diagnostic and prognostic investigation of patients with rheumatoid arthritis? A systematic review
- Department of Rheumatology, Erasmus Medical Center, University Medical Center Rotterdam, Rotterdam, The Netherlands
- Correspondence to J J Luime, Erasmus Medical Center, Department of Rheumatology, PO Box 2040, 3000 CA, Rotterdam, The Netherlands;
- Accepted 3 March 2009
- Published Online First 15 March 2009
Objective: To review the diagnostic and prognostic value of anti-mutated citrullinated vimentin (MCV) in rheumatoid arthritis, taking into account the already available serology.
Methods: Medline was searched via PubMed (1966 to May 2008) for anti-MCV and related terms, arthritis and arthropathies. Studies with anti-MCV, arthritis/arthropathy, and primary data on diagnosis and/or prognosis were included. Their methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) instrument for diagnostic studies and the modified Hayden list for prognostic studies.
Results: Of 14 eligible studies, 11 included diagnostic data and 3 included prognostic data. No study evaluated anti-MCV as an added diagnostic test to the already available anti-cyclic citrullinated peptide (CCP) and rheumatoid factor serology. One study included the optimal patient spectrum resulting in a sensitivity of 0.59 and specificity of 0.98. A total of 10 diagnostic case-control studies using the same anti-MCV kit showed a sensitivity of 0.64–0.84 and a specificity of 0.79–0.96. This almost equalled the performance of anti-CCP in the same studies. The prognostic evaluation of anti-MCV was limited by differences in study methodology, outcome and statistical modelling. Individual studies showed moderate associations for anti-MCV and radiological progression with the strength of the association comparable to that of anti-CCP.
Conclusions: Study heterogeneity, choice of study population and methodological limitations limited overall conclusions about the true diagnostic and prognostic test performance of anti-MCV. Evidence from the diagnostic case-control studies suggests that anti-MCV may be used as an alternative for anti-CCP.
Early intervention in rheumatoid arthritis (RA) improves the prognosis substantially,1 2 urging the need for its early detection. Serological markers play an important role in the detection of rheumatoid arthritis. Rheumatoid factor (RF) and anti-cyclic citrullinated peptide (anti-CCP) antibodies are currently key markers in the diagnostic investigation of patients with arthritis.3 A positive anti-CCP result increases the likelihood of having RA 6-fold to 20-fold in patients with early arthritis.4 5 6 7 A positive RF result increases the likelihood by 3-fold to 10-fold.4 5 6 7 8 9 False-positive rates for anti-CCP are less than 5%, while for RF it varies between 10% to 30% depending on the underlying study population.3 RF and anti-CCP both contribute to the prediction of RA in patients with early arthritis.10 11
A test for mutated citrullinated vimentin antibody (anti-MCV) was recently developed by Bang et al.12 Vimentin is an intermediate filament that is widely expressed in mesenchymal cells and macrophages.13 It is usually not in a citrullinated state, but deimination of this protein occurs in macrophages undergoing apoptosis. Citrullinated vimentin (CV) may emerge as a consequence of inadequate clearance of apoptotic material.14
CV is present in RA pannus and synovial fluid.12 15 CV was first described in 1989 as Sa,16 named after the first patient with RA it was discovered in.15 Previous studies suggested that CV is one of the first antibodies to occur in the disease process and could be more sensitive than the synthetic CCP antibodies.14 17 However, the sensitivity of the anti-Sa (anti-citrullinated vimentin) antibody for detection of rheumatoid arthritis has been shown to be limited in established RA15 18 19 20 21 22 and early arthritis.8 23 24
Bang et al identified vimentin isoforms in which arginine residues were replaced by glycine, which they named mutated citrullinated vimentin.12 Developing an antibody against this isoform (anti-MCV) improved the diagnostic accuracy substantially compared to anti-Sa.12 In this study we scrutinised the literature about anti-MCV and present diagnostic and prognostic data of anti-MCV use. Our primary study aim was to evaluate whether anti-MCV should play a role in the diagnostic investigation of patients with arthritis at risk for RA and the prognostic investigation of RA taken into account the clinical availability of RF and anti-CCP.
This review is based on the principles of Cochrane systematic reviews,25 and the recommendations for reviewing the accuracy of diagnostic tests26 and prognostic studies.27 Studies were identified in the Medline database through PubMed (1966 to May 2008). To retrieve all relevant publications related to anti-MCV we searched for MCV, anti-MCV, anti-Sa, anti-citrullinated peptide antibodies and anti-citrullinated protein/peptide antibody (ACPA) combined with the text words arthritis and arthropathy. One reviewer (JJL) screened abstracts of the retrieved citations on anti-MCV, and related terms and diagnostic terms: sensitivity, specificity, receiver operating characteristic (ROC) curves or prognostic evaluation. Relevant articles were obtained from the library and their reference lists were screened to find additional studies.
Full articles were screened for the following inclusion criteria: anti-MCV, arthritis/arthropathy and primary data on either diagnosis or prognosis. For each diagnostic study, details were extracted on study population (setting, sampling, age, sex and diagnosis), clinical tests, reference tests and outcome (sensitivity and specificity). When raw data were available, likelihood ratios with 95% CIs28 were calculated for individual findings. In addition, the added value of MCV over the known serological markers, RF and anti-CCP, was extracted if presented in the primary studies. For each prognostic study details were extracted on duration of follow-up, other important prognostic factors and outcomes of the prognostic modelling.
The patient spectrum is an important item in the external validity of diagnostic studies. The test accuracy varies across patient spectra.29 Therefore diagnostic articles were categorised based on their patient spectrum in phase I to IV studies according to the suggestions of Sackett and Haynes30; see table 1 for explanation of the phases. Patients with arthritis or inflammatory arthropathy, not diagnosed as crystal or pseudocrystal arthropathy, were regarded as the representative sample of patients who would receive the test in practice (phase III diagnostic study).
The methodological quality of the studies was evaluated by two reviewers (EMC, JJL) using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)31 tool for diagnostic studies and the modified quality list of Hayden for prognostic studies.32 A third reviewer (EL) evaluated the methods used for the autoimmune serology. QUADAS includes 14 questions about the spectrum of patients studied, selection criteria, test verification, test description, blinding, uninterpretable results and study withdrawals. These questions could be scored as positive if the item was fulfilled, negative if the item was not fulfilled, or unclear if the item was not described. The prognostic quality instrument consists of six categories: study population, study attrition, prognostic factor measurement, outcome measurement, confounding measurement and analysis. This list was modified for the specific use of biomarker evaluation based on the REMARK (for “REporting recommendations for tumor MARKer prognostic studies”) recommendations for prognostic biomarker studies33 and can be found in the Supplementary material. If the reviewers were discordant on an item, the item was discussed, and if discrepancies still existed these were resolved by the third reviewer (EL). The limitations of each study were described. The studies were not allocated into arbitrary categories of low, medium or high quality.
Our primary database search retrieved 420 abstracts. About 13 articles contained information about anti-MCV, of which 10 fulfilled the inclusion criteria.12 34 35 36 37 38 39 40 41 42 Reasons for exclusion were no primary data provided on anti-MCV,43 44 and patients selected on anti-MCV status or anti-CCP status.45 Search of bibliographies did add one additional study.46 All included studies12 34 35 36 37 38 39 40 41 42 46 provided data on sensitivity and specificity of anti-MCV, while three studies34 36 37 reported information on the prognostic value of anti-MCV. Baseline characteristics of these studies can be found in the Supplementary material.
Diagnostic value of anti-MCV
The performance of anti-MCV as a diagnostic test was examined in patients with arthritis suspected as being RA in one study.36 The remaining studies compared cohorts of patients with RA to control patients with other inflammatory joint disorders and/or healthy controls. The ACR criteria for RA47 were used as reference test in all studies. Sample sizes varied from 98 subjects to 1642 subjects, with a median of 301 subjects per study.
Most studies did not describe details concerning blinding of the operators of the anti-MCV test12 34 35 36 37 38 42 and blinding of the rheumatologist who diagnosed the inflammatory disorders.12 34 35 38 39 40 41 42 Uninterpretable or intermediate results of the anti-MCV test or diagnostic process, which are likely to occur in serology tests and the diagnostic process of RA, were seldom reported.40 Withdrawals were not reported, possibly due to the retrospective nature of most studies.
Anti-MCV as a single test
The diagnostic accuracy data of anti-MCV are given in table 2, together with the diagnostic value of anti-CCP and RF if presented in the primary papers. Table 2 is stratified according to the phases of diagnostic studies as explained in table 1. Among patients with aspecific arthritis suspected for rheumatoid arthritis a sensitivity of 0.59 and a specificity of 0.9236 was reported In the 10 case-control studies evaluated anti-MCV among patients with established RA, other inflammatory joint disorders and/or healthy controls, sensitivity ranged between 0.64–0.84 and specificity between 0.79–0.12 34 35 37 38 39 40 41 42 46
Additional diagnostic value of anti-MCV
None of the studies evaluated the value of anti-MCV in the diagnostic investigation with the already available RF and anti-CCP outcomes. In the case-control studies that presented data on anti-MCV and anti-CCP as single tests the diagnostic accuracy varied slightly between the two tests.34 35 37 38 39 40 42 46 The specificity of anti-MCV was slightly lower compared to the specificity of anti-CCP. The sensitivity varied more in two studies in favour of anti-MCV,12 37 and in two studies in favour of anti-CCP,34 38 while in four studies the antibodies were equal.39 40 41 46 Some studies presented details on the number of cases identified by anti-MCV alone varying from 1% to 20%12 22 36 37 38 40 and by anti-CCP alone varying between 2% to 41%.22 36 37 38
Prognostic value of anti-MCV
Three studies evaluated the prognostic value of anti-MCV.34 36 37 Their baseline characteristics are provided in the Supplementary material. The methodological quality varied. None of the studies included a rational for the sample size and treatment was not randomised or protocol based.34 36 37 The patient cohort was partly defined in one study,34 two studies36 37 included less than 75% of known prognostic factors (treatment, anti-CCP, RF, disease activity, erosions at baseline, extra-articular manifestations, smoking), while univariate analysis for the different prognostic factors was lacking or only partly presented in two studies.34 37 Limited information on statistical modelling and/or conceptual framework was a limitation in one study.37
Data on the prognostic value of anti-MCV is presented in table 3. The relation between anti-MCV and disease activity was evaluated in two studies.34 36 Radiological progression was studied using the Larssen score34 37 and the Sharp–van der Heijde (SvH) score.36
Based on the outcome of individual studies, anti-MCV seems to have moderate associations with radiological progression.34 36 37 The strength of this association was similar to anti-CCP in these studies. One study reported longitudinal data on disease activity and anti-MCV levels, showing a very weak longitudinal relationship,36 while one other study showed less reduction of the disease activity score (DAS) in patients positive for anti-MCV.34
Study heterogeneity and choice of study population limited overall conclusions about the diagnostic value of anti-MCV in patients with arthritis who would be likely to undergo serology tests in practice. Although information about the true test performance of anti-MCV is lacking, the diagnostic case-control studies provided information about anti-MCV compared to anti-CCP. In the 10 case control studies anti-MCV was as sensitive as anti-CCP, but slightly less specific in detecting rheumatoid arthritis.12 34 35 37 38 39 40 41 42 These results suggest that anti-MCV may be used as an alternative for anti-CCP. This was confirmed in one cohort study with a more representative patient population.36 No studies presented data on the added diagnostic value of anti-MCV in the diagnostic investigation to the already available anti-CCP and RF outcomes.
Cases identified by anti-MCV and anti-CCP were not always identical, meaning that patients negative for anti-CCP could be anti-MCV positive and vice versa. This was previously also shown for anti-citrullinated fibrinogen.48 As early detection of rheumatoid arthritis is increasingly important and the presence of autoantibodies tends to guide more intensive treatment,1 it may be relevant to test patients who were negative anti-CCP for the presence of other autoantibodies such as anti-MCV and/or anti-citrullinated fibrinogen. The question here however is, who should be tested to prevent unnecessary costly testing of all patients with arthritis. One possibility could be to test the cases negative for anti-CCP and positive for rheumatoid factor. Alternatively, one could narrow the arthritis patient spectrum using a prediction rule such as presented by Visser et al,10 to identify patients who are seronegative with higher risk.
With regard to the use of anti-MCV as prognostic biomarker in patients with RA or persistent arthritis, three studies were available. The studies were heterogenic in their design, statistical modelling and outcome measures. Most studies lacked important prognostic factors such as baseline disease activity, erosions, extra-articular manifestations, rheumatoid noduli and cigarette smoking.1 None of the studies protocolised or randomised their treatment to prevent confounding by severity.49 Treatment is the most important inhibitor for radiological progression and persistent high disease activity and is therefore intertwined in studying the longitudinal relationship between anti-MCV and radiological progression. There are, however, reasons to believe that in the cohorts described here, treatment for RA was still at its infancy in reducing radiological progression as most data was collected in the mid-1990s to 2000.24 36 37 Individual prognostic studies showed moderate association between anti-MCV and radiological progression equal to that of anti-CCP in these studies. This suggests that there is little evidence of added value of anti-MCV in the prognostic investigation of patients with RA compared to the already existing serology.
The resembling agreement between anti-MCV and anti-CCP in the diagnostic and prognostic investigation questions the relevance of measuring biologically available autoantibodies against proteins such as vimentin and fibrinogen. It seems that the specific presence of the MCV antibody in itself is not relevant as anti-CCP as synthetic peptide seem to be as effective in detection of patients with RA. However, the different citrullinated proteins might be involved in separate pathophysiological entities of RA and therefore could be of help to predict and stratify RA subgroups.50
Reviewing the literature on anti-MCV we came across a few challenges, from which we would like to address three important issues: the reporting of primary data, test reliability and study population. Reporting of biomarkers studies could be improved by using guidelines such as STARD (“STAndards for the Reporting of Diagnostic accuracy studies”)51 and REMARK.33 Studies on the diagnostic accuracy in this review often did not reported the study design, how diagnosis were obtained, whether test results were interpreted without knowledge of the reference standard, training of the test operators, reliability of the test and whether intermediate or uninterpretable results were found in either the biomarker test or the reference test. The main challenges in the prognostic studies were the lack of sample size calculation, no measures taken to prevent confounding by severity, not all known prognostic factors included and lack of information on univariate outcome of statistical modelling.
Test reliability is an issue in ACPA assessment, making comparison between different assays difficult, but only few primary papers included in this review reported reliability figures. Coenen et al39 and Bizzaro et al41 evaluated the performance of commercial available ACPA methods in series of serum samples of patients with established RA. Both showed substantial variation in interassay and intra-assay reliability of the assays. We underscribe their notion that standardisation of anti-CCP assays is needed and suggest that authors report information about interassay and intra-assay reliability of the assays in diagnostic and prognostic biomarker studies.
Differences in study population are always challenging when performing systematic reviews, but we would like to focus on this, as the outcome in diagnostic and prognostic research strongly depends on the characteristics of patients included in the study.27 30 52 The diagnostic case-control designs were very helpful to see whether anti-MCV could distinguish diseased from healthy and RA from other inflammatory diseases, but do not represent true test performance among patients in whom it is clinically sensible to suspect RA. Studies with a case-control design tend to report inflated sensitivity and specificity numbers.52 This is due to the inclusion of patients with established RA, which will generate less false-negative cases, while the inclusion of healthy subjects will reduce the number of false-positive cases. Readers should be made aware of this.
This review has certain limitations. One Korean paper was not evaluated in the review for its methodological quality because of language difficulties.46 Data was extracted from the English abstract and tables presented in the article. Publication bias could not be excluded in this review as this is an substantial issue in observational studies including studies of biomarkers.27 33 Regarding the results of the included studies, showing limited additional value of anti-MCV to the currently used biomarkers, we expect that there still may be underreporting of non-significant results. Adding non-significant results, however, will not change the conclusion of this review.
In summary, study heterogeneity and choice of study population limited overall conclusions about the diagnostic value and prognostic value of anti-MCV. Evidence from diagnostic case-control studies suggests that anti-MCV may be used as an alternative for anti-CCP. Conclusions about the prognostic value are limited to the observation that individual studies showed equal moderate performance of anti-MCV and anti-CCP.
▸ Additional data (supplementary tables 1 and 2 and supplementary appendix) are published online only at http://ard.bmj.com/content/vol69/issue2
Competing interests None declared.
Provenance and Peer review Not commissioned; externally peer reviewed.