Remission in early rheumatoid arthritis: predicting treatment response.

Abstract

Optimizing therapeutic strategies to induce remission requires an understanding of the initial features predicting remission. Currently no suitable model exists. We aim to develop a remission score using predictors of remission in early rheumatoid arthritis (RA). We used a dataset from a UK randomized controlled trial that evaluated intensive treatment with conventional combination therapy, to develop a predictive model for 24-month remission. We studied 378 patients in the trial who received 24 months' treatment. Our model was validated using data from a UK observational cohort (Early RA Network, ERAN). A group of 194 patients was followed for 24 months. Remission was defined as 28-joint Disease Activity Score < 2.6. Logistic regression models were used to estimate the associations between remission and potential baseline predictors. Multivariate logistic regression analyses showed age, sex, and tender joint count (TJC) were independently associated with 24-month remission. The multivariate remission score developed using the trial data correctly classified 80% of patients. These findings were replicated using ERAN. The remission score has high specificity (98%) but low sensitivity (13%). Combining data from the trial and ERAN, we also developed a simplified remission score that showed that younger men with a TJC of 5 or lower were most likely to achieve 24-month remission. Remission was least likely in older women with high TJC. Rheumatoid factor, rheumatoid nodules, and radiographic damage did not predict remission. Remission can be predicted using a score based on age, sex, and TJC. The score is relevant in clinical trial and routine practice settings.