Value of anti–modified citrullinated vimentin and third-generation anti–cyclic citrullinated peptide compared with second-generation anti–cyclic citrullinated peptide and rheumatoid factor in predicting disease outcome in undifferentiated arthritis and rheumatoid arthritis
Autoantibodies such as rheumatoid factor (RF) and anti–citrullinated protein autoantibodies (ACPAs) determined by testing with second-generation anti–cyclic citrullinated peptide (anti–CCP-2) are frequently measured in clinical practice because of their association with disease outcome in undifferentiated arthritis (UA) and rheumatoid arthritis (RA). Recently, 2 new ACPA tests were developed: third-generation anti-CCP (anti–CCP-3) and anti–modified citrullinated vimentin (anti-MCV) autoantibody tests. To facilitate the decision on which autoantibody to test in daily practice, this study evaluated the capability of these autoantibodies and combinations of them to predict 3 outcome measures: progression from UA to RA, the rate of joint destruction in RA, and the chance of achieving sustained disease-modifying antirheumatic drug (DMARD)–free remission in RA.
Patients with UA (n = 625) were studied for whether UA progressed to RA after 1 year. Patients with RA (n = 687) were studied for whether sustained DMARD-free remission was achieved and for the rate of joint destruction during a median followup of 5 years. Positive predictive values (PPVs) for RA development and for associations with the disease course in RA were compared between single tests (anti–CCP-2, anti–CCP-3, anti-MCV, and RF) and between combinations of these tests.
Among the single tests performed in patients with UA, anti–CCP-2 tended to have the highest PPV for RA development (67.1%), but the 95% confidence intervals of the other tests overlapped. Among the single tests in patients with RA, all 4 tests showed comparable associations with the rate of joint destruction and with the achievement of remission. In both ACPA-positive and ACPA-negative RA, the presence of RF was not associated with more joint destruction. For all outcome measures, performing combinations of 2 or 3 autoantibody tests did not increase the predictive accuracy compared with performing a single test.
For clinical practice, a single autoantibody test is sufficient for risk estimation in UA and RA.
Rheumatoid arthritis (RA) is considered to have an autoimmune origin because of the presence of self-reactive autoantibodies. In addition to rheumatoid factor (RF), which is, to date, the only serologic measure included in the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) 1987 criteria for RA (1), several other autoantibodies have been described in recent years (2). The discovery of anti–citrullinated protein autoantibodies (ACPAs) has led to the development of various new tests for autoantibodies in RA. The anti–cyclic citrullinated peptide (anti-CCP) test, which is directed against a synthetic citrullinated peptide, revealed a higher specificity for RA when the first-generation anti-CCP (anti–CCP-1) test was used in comparison with the RF test (91–96% versus 74–91%) (3–7). Subsequently, a commercially available second-generation anti-CCP test (anti–CCP-2) was developed, showing an even better specificity for RA (90–97%) (4–9). RF and anti–CCP-2 autoantibodies can also be present in the preclinical phase and are associated with future RA development (10, 11). Consequently, tests for anti–CCP-2 and RF are now widely used as diagnostic tools in clinical practice.
Recently, 2 other serologic tests have emerged, the third-generation anti-CCP (anti–CCP-3) and the anti–modified citrullinated vimentin (anti-MCV) autoantibody tests. The anti–CCP-3 test has been reported to have sensitivities and specificities comparable with those of anti–CCP-2 (69–83% and 93–95%, respectively) (9, 12). The other novel autoantibody test, anti-MCV, targets modified citrullinated vimentin and has its origin in the older anti-Sa autoantibody test that has been shown to target citrullinated vimentin (13). Compared with anti–CCP-2, studies have demonstrated somewhat lower specificities and higher sensitivities for RA when anti-MCV is used (79–92% and 70–84%, respectively) (4–6, 14).
The aforementioned data were obtained in case–control studies comparing RA patients with non-RA patients or healthy individuals, and the resulting test characteristics quantify the proportion of patients who are identified as positive for RA by the specific autoantibody test (sensitivity) or the proportion of healthy individuals who are identified as negative for RA (specificity). As such, these measures, as well as the likelihood ratio of a test, provide information on the quality of the test. In clinical practice, the value of determining the presence of ACPAs or RF is related to the ability of these tests to predict the disease course. The chance for an individual patient to have a certain disease course is expressed by the positive predictive value (PPV) and negative predictive value (NPV).
A clinical state in which knowledge of the presence of RF and ACPAs can be particularly helpful is undifferentiated arthritis (UA). In the subgroup of patients with early arthritis, no diagnosis can be established according to existing classification criteria, and the presence of RF or anti–CCP-2 indicates an increased risk of RA development (15, 16). Thus far, the PPVs and NPVs for the risk of developing RA in patients with UA have not been studied using anti–CCP-3 or anti-MCV, and the 4 autoantibodies have not been subjected to a head-to-head comparison. Furthermore, the additive value of testing several combinations of autoantibodies for the prediction of RA development in individual patients with UA has not been addressed. Therefore, the first aim of this study was to compare the accuracy of anti–CCP-2, anti–CCP-3, anti-MCV, and RF in predicting RA development in patients with UA, and to explore whether testing combinations of these autoantibodies increases the predictive accuracy.
RF and anti–CCP-2 not only are important predictors of RA development but also are some of the most potent predictors of the outcome in RA, as measured by the rate of radiographic joint destruction (17–19). Thus far, only 1 study has compared the ability of anti-MCV with that of anti–CCP-2 for predicting radiographic progression, and the results in 273 RA patients provided evidence that anti-MCV is a better predictor than anti–CCP-2 (20). The effect of testing combinations of all 4 autoantibodies, however, was not studied. Thus, the second aim of the present study was to compare anti–CCP-2, anti–CCP-3, anti-MCV, and RF for their ability to predict the rate of joint destruction, and to explore whether combinations of these autoantibodies can increase the predictive ability, taking advantage of a longitudinal cohort of 687 RA patients with a median followup of 5 years.
A second disease outcome in RA is the achievement of remission, which, with the introduction of new aggressive treatment strategies, has increasingly become an attainable goal. We chose a strict definition and defined sustained disease-modifying antirheumatic drug (DMARD)–free remission as the persistent absence of synovitis for at least 1 year after cessation of DMARD therapy (21). Since the predictive value of the 4 autoantibodies in relation to the chance of remission has been scarcely explored, the present study compared the 4 tests for their ability to predict sustained DMARD-free remission in RA patients treated with conventional DMARDs.
In summary, to support the choice on which autoantibody to test in daily practice, this study uses a large, longitudinal cohort to assess the value of determining the presence of anti–CCP-2, anti–CCP-3, anti-MCV, and RF for predicting 3 outcome measures: progression from UA to RA, the rate of joint destruction in RA, and the chance of achieving sustained DMARD-free remission of RA. In addition, the predictive value of combining several tests is investigated.
PATIENTS AND METHODS
All patients included in this study were selected from the Leiden Early Arthritis Clinic (EAC) cohort that was started in 1993 (22). Patients were referred by general practitioners when arthritis was the suspected diagnosis. For inclusion, patients were required to have arthritis confirmed at the time of physical examination and to have a symptom duration of <2 years. Written informed consent was obtained from all participants. The study was approved by the local medical ethics committee.
At inclusion, patients were asked to report their joint symptoms and underwent a physical examination. Blood samples were obtained for routine diagnostic laboratory screening (including testing for IgM-RF), and the samples were stored to determine the presence of other autoantibodies at a later time. Followup visits were performed on a yearly basis and included radiography of the hands and feet.
Since the start of the Leiden EAC, treatment strategies for RA have changed, and 4 different strategies were applied depending on the inclusion period. RA patients included in the Leiden EAC between 1993 and 1995 were treated initially with analgesics and subsequently with chloroquine or sulfasalazine (delayed treatment) if they had persistent active disease (23). From 1996 to 1998, RA patients were treated promptly with either chloroquine or sulfasalazine (early treatment) (22, 23). From 1998 to 2002, RA patients were treated promptly with either sulfasalazine or methotrexate (early treatment). Finally, patients included in 2002 and thereafter were treated promptly with either sulfasalazine or methotrexate combined with treatment adjustments based on the disease activity (early treatment and disease activity–based treatment). Treatment of UA patients was not subject to a standard protocol.
UA was defined as a clinical presentation not fulfilling any of the existing classification criteria for a rheumatic disease diagnosis 2 weeks after the first presentation (24). Thus, patients with other rheumatic diseases, such as Sjögren's syndrome, psoriatic arthritis, or spondylarthritis, whose diagnosis was established at baseline were excluded from the UA group (Figure 1). Thus, 625 patients with UA who were consecutively included between 1993 and 2006 were studied. After 1 year of followup, 201 patients with UA (32.2%) had disease that progressed to RA, the diagnosis of which fulfilled the ACR 1987 criteria for RA (1).
In addition, 687 patients with RA who were included between 1993 and 2006 were studied for their disease outcome. Of these, 486 RA patients had already fulfilled the ACR criteria for RA at the time of inclusion, and 201 patients were initially diagnosed as having UA and developed RA within the first year of followup (Figure 1).
Determination of autoantibodies.
IgM-RF was determined by enzyme-linked immunosorbent assay (ELISA). Anti–CCP-2 autoantibodies (total IgG) were measured by ELISA (Immunoscan RA Mark 2; Euro-Diagnostica, Arnhem, The Netherlands). The cutoff level for anti–CCP-2 autoantibody positivity was set at 25 arbitrary units, according to the manufacturer's instructions. Anti–CCP-3 autoantibodies (IgA and IgG subforms) and anti-MCV autoantibodies were also measured by ELISAs (Quanta Lite CCP version 3.1 for IgG/IgA from Inova Diagnostics, San Diego, CA as well as ELISAs from Orgentec Diagnostika, Mainz, Germany). The cutoff level for both the anti–CCP-3 test and the anti-MCV test was 20 arbitrary units, according to the manufacturers' instructions,
The numbers of patients in whom testing for RF, anti–CCP-2, anti–CCP-3, and anti-MCV was performed showed little variation; testing for RF, anti–CCP-2, anti–CCP-3, and anti-MCV was performed in 665, 653, 629, and 629 of 687 RA patients, respectively, and in 623, 624, 597, and 597 of 625 UA patients, respectively. The slight variation in sample sizes was due to the difference in the timing of the performance of the tests. RF was determined routinely at inclusion, whereas the ACPAs were determined using serum samples that had been stored. The recently introduced anti–CCP-3 and anti-MCV assays were performed on all available sera in 2008, whereas the anti–CCP-2 assay was performed earlier.
Radiographs of the hands and feet were obtained on consecutive years starting at baseline and were scored for radiographic progression according to the Sharp/van der Heijde method (25). Among all of the patients with RA, 579 had data on the presence of autoantibodies as well as radiographic data. To encompass a reliable sample size during followup, radiographic data were restricted to a maximum of 7 years of followup. The number of available radiographs varied per time point and declined from 552 at baseline to 478 after 1 year of followup, to 426 after 2 years, to 358 after 3 years, to 299 after 4 years, to 270 after 5 years, to 207 after 6 years, and to 156 after 7 years. Due to the study design (an inception cohort), not all patients had a similar duration of followup (median 5 years).
All radiographs were scored by an experienced scorer (MPMvL) who was blinded with respect to the patient's autoantibody status, treatment, and other clinical data. Scoring was performed with time order of the radiographs, which is more sensitive to change compared with scoring of radiographs with unknown time sequence (26). Among the total number of scored radiographs, 499 radiographs were rescored by the same reader, consisting of 149 radiographs obtained at baseline and 350 radiographs obtained during the followup period, belonging to 60 randomly selected RA patients. The intraobserver intraclass correlation coefficients were 0.91 for all scored radiographs, 0.84 for baseline radiographs, and 0.97 for the radiographic progression rate.
Identification of sustained DMARD-free remission of RA.
Remission of RA was defined in its most stringent form as the persistent absence of synovitis for at least 1 year after cessation of DMARD therapy and the identification of remission by the patient's rheumatologist (27). The remission status could be reliably ascertained in 635 RA patients. Most patients who achieved remission were followed up longer than the minimum requirement of 1 year; the median duration of observation after discontinuation of DMARDs in the absence of swollen joints was 2.5 years. Patients who had a recurrence of their arthritis after discharge could easily return to the Leiden University Medical Center, the only referral center for rheumatology in a health care region of ∼400,000 inhabitants. The frequency of relapse of RA was recorded, and patients who experienced a relapse were included in the nonremission group (n = 6).
The PPV (proportion of UA patients who had a positive test result for RA and whose disease progressed to RA) and the NPV (proportion of UA patients whose disease did not develop into RA) were determined. Because radiographic data are not normally distributed, nonparametric Mann-Whitney tests were used to compare the Sharp/van der Heijde scores at individual time points for patients with and those without autoantibodies or for those with and those without combinations of autoantibodies. In addition, to take advantage of the prospective character of the data consisting of repeated measurements, and to avoid multiple testing by performing statistical tests for each time point, a linear mixed model with an autoregressive correlation structure with heterogeneous variances was used. This model estimates the linear progression rate of radiographic joint destruction using normalized, log-transformed Sharp/van der Heijde scores, taking missing observations into account. This means that the model compares the progression rates for the different patient groups. In the mixed model analyses, corrections were applied for age, sex, and inclusion period/treatment strategy. Correction for treatment strategy was performed by including, as a variable, the study inclusion period in the linear mixed model. This was done because treatment strategies improved over time, and an influence of the treatment strategy (reflected by the inclusion period) on the progression of radiographic joint damage has been observed previously as well as in the present study (data not shown).
In addition, the available followup duration differed between patients, and therefore the number of radiographs per time point declined during followup. Since the patients with the longest followup were included in the earliest inclusion period and, thus, would have been treated with the least aggressive treatment strategy, correction for inclusion period was performed. In order to prevent overfitting of the data, no corrections were applied for other variables.
Analysis of sustained DMARD-free remission of RA was performed by Cox regression analysis, to take into account the differences in followup times among patients. For patients who achieved remission, the dependent variable was the time-to-event, indicating the time until reaching remission. For patients not achieving remission, the time to last followup was used, with a maximum followup of 10 years.
For all statistical analyses, SPSS version 16.0 (SPSS, Chicago, IL) was used. P values less than 0.05 were considered significant. All reported P values are 2-sided.
Predicting progression from UA to RA.
Among the total Leiden EAC cohort, consisting of more than 2,000 patients with early arthritis, 625 patients with UA were selected and studied for disease progression to RA after 1 year of followup. At inclusion, the mean ± SD age was 50.9 ± 17.0 years, 371 patients (59.1%) were female, and the self-reported symptom duration was a mean ± SD of 5.5 ± 8.5 months. Anti–CCP-2, anti–CCP-3, anti-MCV, and RF were present in 149 of 624 UA patients (23.9%), 172 of 597 UA patients (28.8%), 199 of 597 UA patients (33.3%), and 155 of 623 UA patients (24.9%), respectively. The presence of autoantibodies overlapped; e.g., UA patients who tested positive for anti–CCP-2 were frequently also positive for the other autoantibodies (Figure 2).
The PPV for predicting progression from UA to RA was compared between the 4 autoantibody tests (Table 1). Anti–CCP-2 had the highest PPV, at 67.1%, compared with PPVs of 64.0%, 56.3%, and 61.7% for anti–CCP-3, anti-MCV, and RF, respectively. The NPVs of all 4 tests were comparable (∼80% for each). Thus, when only a single autoantibody test was performed, a positive result on the anti–CCP-2 test tended to correlate with the highest risk of RA development, but overlapping 95% confidence intervals (95% CIs) from the other tests hamper a definite differentiation.
Table 1. Comparison of the anti–CCP-2, anti–CCP-3, anti-MCV, and RF tests for predicting progression from UA to RA*
We then determined whether performing 2 autoantibody tests would result in a better estimation of the risk of RA than would performing only a single test. Since anti–CCP-2 and anti–CCP-3 are related tests, and since 90% of anti–CCP-2–positive patients were also positive for anti–CCP-3, the possible combinations of anti-MCV, anti–CCP-2, and RF were assessed first. The proportions of patients who developed RA (the PPVs) among those who had 2 positive test results were as follows: anti–CCP-2+/anti-MCV+ 69.9% (95% CI 62.1–77.7%), anti–CCP-2+/RF+ 74.1% (95% CI 65.8–82.3%), and anti-MCV+/RF+ 70.6% (95% CI 62.1–79.2%). Additional analyses using anti–CCP-3 instead of anti–CCP-2 yielded comparable results (data not shown). Taken together, these results show that when 2 tests are performed, none of these combinations is clearly superior to the other. Furthermore, no additive value of performing 2 autoantibody tests instead of 1 autoantibody test could be observed.
When a single autoantibody test is performed, no information about the presence or absence of other autoantibodies is obtained. However, the eventual coexisting presence of other, unmeasured autoantibodies can affect the risk of RA. To determine the risk of RA as conferred by the individual autoantibodies or by the number of different autoantibodies, the PPVs for progression to RA were determined in the group of 596 UA patients for whom information on all 3 autoantibodies (anti–CCP-2, anti-MCV, and RF) was available. The difference in comparison with the above-mentioned analyses was that the presence of 2 of the autoantibodies would indicate that the third was absent, whereas in the above-mentioned analyses, only 2 tests were performed, and the third autoantibody test could yield either a positive result or a negative result. The PPV for RA development in patients without any of these 3 autoantibodies was 18.8% (95% CI 14.7–22.9%). In the presence of 1 autoantibody, the PPV was 26.5% (95% CI 17.9–35.0%), and the PPV increased significantly, to 59.6% (95% CI 45.5–73.6%), in the presence of 2 autoantibodies (additional information available from the corresponding author upon request). The PPV for RA development was highest in the presence of all 3 autoantibodies (PPV 73.3%, 95% CI 64.6–81.9%).
Thus, when a single autoantibody test is used for predicting progression to RA in clinical practice, none of the 4 autoantibody tests is clearly superior. Although the presence of 2 autoantibodies significantly increased the risk of RA compared with the presence of 1 autoantibody, the performance of 2 autoantibody tests, at least in clinical use, does not significantly increase the predictive performance compared with the performance of 1 test. This finding is likely explained by the presence of other, unmeasured autoantibodies that affect the risk of RA.
Predicting joint destruction in RA.
Among the 579 RA patients with available radiographic data, the mean ± SD age at baseline was 56.2 ± 15.5 years, and 405 patients (69.9%) were female. Anti–CCP-2, anti–CCP-3, anti-MCV, and RF were present in 313 of 565 RA patients (55.4%), 322 of 544 RA patients (59.2%), 331 of 544 RA patients (60.8%), and 334 of 572 RA patients (58.4%), respectively. Anti–CCP-2–positive patients also tested positive for anti-MCV and/or RF (Figure 2).
The association between autoantibody positivity and the rate of joint destruction was first assessed by performing only 1 autoantibody test. For all autoantibody tests (anti–CCP-2, anti–CCP-3, anti-MCV, and RF), a positive test result, as compared with a negative finding, was associated with a higher Sharp/van der Heijde score at all time points except baseline (P < 0.001 by Mann-Whitney test) and was also associated with a higher rate of joint destruction over a period of 7 years (P < 0.001 by mixed model). As shown in Figure 3A, there was no difference among the 4 tests with regard to their ability to predict joint destruction.
Next, we investigated whether the addition of a second autoantibody test could increase the predictive value for the rate of joint destruction. As depicted in Figure 3B, no differences were seen between testing positive for anti–CCP-2, anti-MCV, or RF alone and testing positive for combinations of these autoantibodies. These results indicate that for clinical use, a positive test result for 1 of these autoantibodies predicts a severe disease course, and a second or third autoantibody test does not increase the predictive accuracy for the rate of joint destruction. Testing for anti–CCP-3 instead of anti–CCP-2 yielded comparable results (data not shown).
To identify the contribution of the individual autoantibodies to the association with the rate of joint destruction, the effect of the number of positive autoantibodies (with the known absence of the other autoantibodies) was investigated using data on the presence of RF, anti–CCP-2, and anti-MCV. The presence of either 2 or 3 of these autoantibodies was associated with a higher rate of joint destruction compared with the presence of none or 1 of these autoantibodies (P < 0.001 for 2 or 3 autoantibodies compared with none or 1 autoantibody, by mixed model) (Figure 4A). No significant difference between the presence of 2 autoantibodies and the presence of 3 autoantibodies or between no autoantibodies and the presence of 1 autoantibody was observed. It should be noted that the group with 1 autoantibody consisted almost exclusively of patients who were anti-MCV positive (n = 33) or RF positive (n = 39); only 2 patients were positive for anti–CCP-2 and negative for anti-MCV and RF.
To more specifically investigate the role of RF in relation to the ACPAs (anti–CCP-2, anti–CCP-3, or anti-MCV) and the rate of joint destruction, the additional effect of RF in the presence or absence of ACPAs was determined (Figure 4B). This revealed that not only in the presence of ACPAs but also in the absence of ACPAs, RF did not significantly contribute to the rate of joint destruction.
Predicting sustained DMARD-free remission of RA.
Among 635 RA patients, 78 patients achieved sustained DMARD-free remission after a median followup of 39.5 months. These 78 patients had a mean ± SD age of 59.4 ± 15.7 years, and 57 (73.1%) were female. Anti–CCP-2 autoantibodies were present in 11.8% of the patients who achieved remission, and anti–CCP-3 autoantibodies, anti-MCV autoantibodies, and RF were present in 21.9%, 28.8%, and 25.0% of these patients, respectively.
The 4 autoantibody tests were compared for their association with the achievement of remission (Figure 5A). The hazard ratio (HR) for each of the 4 tests, which is an indication of the test's ability to predict the likelihood of not achieving sustained DMARD-free remission, was 11.6 (95% CI 5.8–23.4) for anti–CCP-2, 6.0 (95% CI 3.4–10.4) for anti–CCP-3, 4.9 (95% CI 3.0–8.2) for anti-MCV, and 4.7 (95% CI 2.8–8.0) for RF.
Subsequently, the additive value of performing 2 autoantibody tests compared with 1 test was investigated. The HRs for not achieving sustained DMARD-free remission of RA were 15.6 (95% CI 6.7–36.4), 14.0 (95% CI 6.4–31.0), and 11.5 (95% CI 5.4–24.5) for the combinations of anti–CCP-2 and RF, anti–CCP-2 and anti-MCV, and anti-MCV and RF, respectively. These data indicate that to predict the chance of remission, performing 2 tests has no additional value compared with performing the anti–CCP-2 test alone.
To investigate whether the number of autoantibodies present could affect the chance of achieving sustained DMARD-free remission, the HRs were determined for the presence of 1, 2, or 3 autoantibodies (with the other autoantibodies known to be absent), using data on anti–CCP-2, anti-MCV, and RF (Figure 5B). The HRs for not achieving sustained DMARD-free remission were as follows: 3.7 (95% CI 1.1–12.3), 15.5 (95% CI 5.9–41.2), and 17.1 (95% CI 6.8–43.3) for the presence of 1, 2, and 3 autoantibodies, respectively, compared with no autoantibodies. Although the 95% CIs were overlapping, the results suggest that the more autoantibodies that are present, the lower the chance is of achieving sustained DMARD-free remission.
Among the autoantibodies tested in RA, only RF and ACPAs are considered clinically useful. In clinical practice, a physician or a patient is interested in the chance that the arthritis will progress to RA as compared with the chance of RA not developing, given a positive test result and a negative test result, respectively; these risks are reflected by the PPV and NPV, respectively. In addition, in early arthritis, the predictive accuracy of these autoantibody tests is valued the most in patients in whom, at presentation, no definite diagnosis can be established (i.e., patients with UA), since the disease in only one-third of these patients progresses to RA after 1 year (15). Moreover, in patients with RA, the autoantibody tests form one of the most potent predictors to obtain an indication of the severity of the future disease course. As such, information on the results of the autoantibody tests can influence treatment decisions in individual patients with UA and RA (24). Nevertheless, it is thus far unknown which test, or which combination of tests, is most powerful in predicting the progression from UA to RA and the disease progression in RA, and therefore the present study was undertaken.
To evaluate the prediction of progression from UA to RA using a single test, the PPVs of the 4 tests were compared. This revealed that a positive result on the anti–CCP-2 test tended to have the highest predictive value (PPV of 67%, compared with PPVs of 62% and 56% for RF and anti-MCV, respectively), but due to overlapping 95% CIs, a definite differentiation could not be made. In addition, performing a second test did show a tendency to increase the predictive accuracy in evaluating the chance of RA development, when both of the tests yielded positive results; the highest risk of RA was found in patients who were both anti–CCP-2 positive and RF positive, with a PPV of 74%, but again, the 95% CIs overlapped. To formally conclude whether the PPVs of 67% and 74% are statistically significantly different compared with the other 2 tests or compared with performing a single test, more than 1,800 UA patients would be required (using a P value of 5% and a power of 90%). Based on the present data, it can be concluded that the addition of a second test does not result in increased predictive accuracy. Notably, UA patients who tested positive for anti–CCP-2 were also frequently positive for other autoantibodies (Figure 2). Thus, the finding of comparable prognostic performances between 1 test and 2 tests is likely due to the coexisting presence of autoantibodies that have not been measured by a single test but that do affect the RA risk estimation.
The development of RA was assessed after 1 year of followup. This time point was chosen in order to have a similar duration of followup for all studied patients. However, this may have introduced misclassification, since patients with UA may have experienced progression to RA after more than 1 year of followup. With all available followup data, 25 patients (4.4%) showed progression to RA later than 1 year, indicating that the current PPVs may be marginally underestimated. Since this misclassification was present in the total group of UA patients, this does not hamper a comparison of tests.
Two measures were studied for the severity of the disease course in RA: achieving sustained DMARD-free remission and the level of radiographic joint destruction during a median followup of 5 years. Although, as shown in Figures 3A and 5A, the results may lead to the impression that anti–CCP-2–positive patients have a higher rate of joint destruction and achieve sustained DMARD-free remission less frequently compared with patients positive for the other autoantibodies, these differences were not statistically significant. Similar to the data on RA development, performing a second test appeared not to result in a more accurate prediction of the disease outcome in RA. This observation, which is in contrast to the findings in an earlier report (28), can be explained by the presence of unmeasured autoantibodies that are associated with a progressive course of RA.
To obtain a more detailed comprehension of the current results, the contribution of the individual autoantibodies to disease progression (RA development, joint destruction, and achievement of remission) was investigated. This showed that the presence of 2 autoantibodies indicated a significantly increased risk compared with the presence of 1 autoantibody. However, the group with 1 autoantibody consisted mostly of anti-MCV–positive or RF-positive patients. Patients who were positive for only anti–CCP-2 were very rare. Therefore, the possibility cannot be excluded that the increased risk of RA in the presence of 2 autoantibodies compared with 1 autoantibody is due to the effect of anti–CCP-2 rather than to the effect of an additional autoantibody. Nevertheless, in general, it was observed that the presence of a higher number of autoantibodies resulted in a higher risk of RA or a higher risk of progressive disease. This is consistent with recently published data showing that a broader autoantibody response was associated with disease progression (29).
Since the association of RF with the presence of RA is primarily explained by its interaction with ACPAs (30), we investigated whether there was a similar effect on the progression of joint destruction in RA. Intriguingly, the rate of joint destruction both in the ACPA-positive and in the ACPA-negative groups was not affected by the presence or absence of RF. This finding further supports the notion that RF does not, by itself, contribute to disease progression.
To observe the effect of anti-MCV on the rate of joint destruction, RA patients who were positive for anti-MCV only (n = 33) were studied. This revealed that the rate of joint destruction was comparable with that in patients with no autoantibodies (data not shown), indicating that anti-MCV alone does not strongly affect the level of joint damage in RA. The results of the present study thus indicate that for risk estimation of the disease course in clinical practice, performing a single autoantibody test is sufficient, both in UA and in RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. van der Linden had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Van der Linden, Huizinga, Toes, van der Helm-van Mil.
Acquisition of data. Van der Linden, Levarht, Stoeken-Rijsbergen, Huizinga, Toes, van der Helm-van Mil.
Analysis and interpretation of data. Van der Linden, van der Woude, Ioan-Facsinay, Huizinga, Toes, van der Helm-van Mil.
We thank Cypress Bioscience for providing anti–CCP-3 and anti-MCV ELISAs.