Combining anti-cyclic citrullinated peptide with the American College of Rheumatology 1987 criteria failed to improve early rheumatoid arthritis diagnosis in the community-based very early arthritis cohort

Objectives. To test the performances of combining anti-CCP second generation (CCP2) with ACR 1987 classification criteria and to diagnose early RA in a community-based very early arthritis (VErA) patient cohort. Methods. The VErA cohort comprised 310 patients (median age 52 years; 68.1% women; median symptom duration 4.2 months; glucocorticoid- and DMARD naive) conservatively treated during the first 2 years. At 6 years of follow-up, a three-expert committee classified the patients into three groups: RA, other classified arthritis (OCA) or unclassified arthritis (UA). We calculated the performances of the different sets, including anti-CCP2 positivity, while retaining or deleting RF and rheumatoid nodule components with ACR 1987 criteria for early RA diagnosis. Models were subjected to receiver operating characteristics curve and logistic regression analyses to try to identify relevant sets able to classify very early RA. Results. At 6 years, 149 patients were diagnosed as RA and 119 as non-RA (95 OCA and 24 UA). The original ACR 1987 criteria had 77.9% sensitivity and 64.7% specificity for the RA diagnosis at 6 years. The modified set excluding rheumatoid nodules, including anti-CCP2 positivity and retaining RF performed significantly better than ACR 1987 criteria, with 79.9% sensitivity and 64.7% specificity and with a larger area under the curve. However, in the zone of interest, i.e. epsilon 4/7 criteria, the curves for these sets were superimposed. Conclusions. Adding anti-CCP2 positivity and deleting rheumatoid nodules failed to improve the performances of ACR 1987 classification criteria for the diagnosis of early RA.