Objective. Early recognition and treatment of RA is associated with an improved outcome. The 2010 ACR/EULAR criteria for RA identify RA patients earlier than the 1987 ACR criteria. Nevertheless, we recently observed that 24% of the 2010 unclassified arthritis (UA) patients develop RA during follow-up. Here we studied this frequency in other cohorts and evaluated the prognostic accuracy of ACPA and the Leiden prediction rule in 2010 UA patients.
Methods. The 2010 UA patients from three Early Arthritis Clinics were studied: 776 from Leiden, 121 from Birmingham and 322 from Amsterdam. Fulfilment of the 1987 ACR criteria during follow-up was studied as the primary outcome. DMARD prescription during the year and having a persistent course of arthritis over 7 years were studied as secondary outcomes in one cohort. The presence of ACPA and the prediction score at baseline were evaluated in relation to these outcomes.
Results. In the three cohorts, 24%, 26% and 12%, respectively, of the 2010 UA patients fulfilled the 1987 criteria after 1 year. However, some of these patients already fulfilled the 1987 criteria at baseline. In 1987 and 2010 UA patients, 15%, 21% and 9%, respectively, developed RA (1987) at 1 year. In these patients, 0–6% of the patients were ACPA positive and 0–1% had high prediction scores. Consequently a large majority of the UA patients with an unfavourable outcome was not recognized by these prognostic tools.
Conclusion. A proportion of 2010 UA patients progress to RA. ACPA and the Leiden prediction rule are not useful in identifying these patients. These results imply that other predictive markers should be developed for 2010 UA patients.