Background: The presence of autoantibodies is considered an important characteristic of rheumatoid arthritis (RA); therefore, both anticitrullinated protein antibodies (ACPA) and rheumatoid factor (RF) are included in the 2010 classification criteria for rheumatoid arthritis (RA). However, a considerable number of RA patients lack both these autoantibodies. Recently, several novel autoantibodies have been identified but their value for the classification of RA patients is unclear. Therefore, we studied the value of novel autoantibodies using the presence of anticarbamylated protein (anti-CarP) antibodies as an example for predicting RA development in patients with undifferentiated arthritis (UA). Methods: There were 1352 UA patients included in the Leiden Early Arthritis Clinic (EAC) cohort according to the 1987 criteria. When the 2010 criteria were used, there were 838 UA patients. Of these, we evaluated whether they fulfilled the 1987 or 2010 criteria after 1 year, respectively. Logistic regression analyses were performed with RA as outcome and ACPA, RF, and anti-CarP antibodies as predictors. Analyses were repeated after stratification for ACPA and RF. Results: Thirty-three percent of the 1987-UA patients and 6% of the 2010-UA patients progressed to RA during the first year of follow-up. For the 1987-UA patients, anti-CarP antibodies were associated with progression to RA, an association which remained when a correction was made for the presence of ACPA and RF (odds ratio (OR) 1.7, 95% confidence interval (CI) 1.2-2.4). After stratification for ACPA and RF, anti-CarP antibodies were associated with progression to RA only for ACPA- and RF-negative patients (OR 2.1, 95% CI 1.3-3.7). For the 2010-UA patients, anti-CarP antibodies were associated with progression to RA; however, they were not when a correction was made for the presence of ACPA and RF (OR 0.8, 95% CI 0.3-2.1). Conclusions: Our finding that anti-CarP antibodies have no additional value when RA is defined according to the 2010 criteria might be inherent to the composition of the 2010 criteria and therefore might also apply to other novel autoantibodies. Potentially it would be interesting to evaluate other, non-autoantibody biomarkers.
