The key message of this study is that anti-CCP antibody status remains stable up to 5 years after disease onset in patients with IP. Patients switching from anti-CCP positive to negative retained the disease characteristics of persistently anti-CCP-positive patients. In contrast, 5-year radiographic outcomes were significantly worse in individuals converting from anti-CCP negative to positive, compared with those who remained negative; however, only 1.5% of patients negative for both anti-CCP and RF at baseline had converted to anti-CCP positive at 5 years. Repeating anti-CCP antibody measurement at 5 years added no additional predictive value in detection of radiographic damage or functional disability over baseline anti-CCP status alone.
Consistent with previous studies, RF status was more likely to change over time, with a lower proportion of subjects testing positive for RF at 5 years than at presentation [15
]. However, baseline anti-CCP status was a strong predictor of change in RF status, and after accounting for the effect of anti-CCP, change in RF status in either direction was not significantly associated with clinical outcomes. This is in keeping with our previous findings in this cohort that anti-CCP antibodies predict more-severe disease irrespective of RF status [8
Consistent with the known specificity of a positive anti-CCP status for RA and with its strong association with disease severity, patients with IP who will not satisfy the 1987 criteria for RA at any time during the first 5 years of follow-up were much less likely to be anti-CCP positive and displayed a much milder disease course. However, it is now impossible for practitioners to predict accurately at baseline which patient will prospectively go on to satisfy cumulatively the 1987 ACR criteria for RA within the 5 first years of follow-up. This should be, however, improved with the use of the 2010 ACR/EULAR criteria.
Baseline anti-CCP titer predicted both the presence and severity of erosive joint damage at 5 years in this cohort. A dose-response relation was observed, and individuals with a baseline anti-CCP concentration of more than 4 times the upper limit of normal were 10 times more likely to have a Larsen score above 15 (highest tertile) at 5 years as were those with a baseline anti-CCP titer of less than 2 (Table ). Consistent with this, higher anti-CCP antibody levels have been found to associate with increased disease activity and radiographic progression in patients with RA [5
]. However, a previous study investigating this relation in subjects with undifferentiated IP found no significant association between baseline anti-CCP concentration and DAS28, HAQ, or radiographic progression at 2 years [16
]. The disparity with our results may relate to the short follow-up time in that study, as the strength of correlation between baseline anti-CCP titer and clinical indices has been shown to increase gradually in the first 5 years of disease in patients with early RA [17
]. In support of our findings, higher anti-CCP titers predicted development of persistent arthritis in patients with IP [27
]. Lowering the threshold for anti-CCP positivity marginally improved the predictive value for erosive disease in our cohort, but at the cost of loss of specificity, which is similar to the findings of Mjaavatten et al.
] for prediction of persistent arthritis. Insufficient evidence exists to recommend altering the threshold for anti-CCP positivity; however, these findings highlight the importance of considering the absolute anti-CCP titer when evaluating the likely prognosis in patients with early IP [13
In keeping with previous studies, we found no significant associations between change in anti-CCP titers and 5-year clinical outcomes [16
]. It is not clear whether anti-CCP titers are affected by DMARD treatment, and study findings have been inconsistent [17
]. We included an adjustment for ever having received a DMARD, which did not materially alter our findings; however, we cannot exclude that DMARD treatment may have influenced associations between change in anti-CCP concentrations and clinical outcomes. Nevertheless, available data suggest that any associations between variation in anti-CCP titer and disease severity are relatively weak and are unlikely to add significant prognostic information over that of baseline anti-CCP titer alone.
The proportion of patients positive for anti-CCP (30%) or RF (26%) at 5 years was much lower than is typically seen in cohorts with established RA, even though 73% of patients fulfilled ACR criteria for RA by 5 years. These percentages are similar to those observed in other early arthritis cohorts [14
]. Our findings indicate that this discrepancy is unlikely to be due to a large proportion of subjects converting to autoantibody positivity in later disease. An alternative explanation may be that seronegative individuals are more likely to have milder disease and therefore are underrepresented in hospital-based cohorts with established RA.
Strengths of this study include the large sample size tested in the analysis, the availability of serum samples at baseline and 5 years, and the repeated testing of borderline samples to ensure the most accurate results. A potential weakness is the fact that no information was available regarding RF or anti-CCP status in the intervening years, so it is possible that "ever positive" status is a better predictor of outcome. However, given the stability observed for anti-CCP status over a 5-year period, it is unlikely to vary to a greater extent over the intervening years.