Predicting responses in patients with rheumatoid arthritis to disease-modifying agents using baseline clinical data

Abstract

Objectives: The optimal treatment for active rheumatoid arthritis (RA) is unresolved, particularly in early RA. We used data from an observational cohort to develop the simple predictor algorithm and evaluated its application in two completed clinical trials in early and established RA. We assessed whether using a simple algorithm can identify patients who have persisting active disease despite treatment with disease-modifying drugs (DMARDs). We also examined if patients who have lower likelihoods of persisting active RA are likely to benefit from intensive treatment. Methods: We developed a simple predictive score for persisting disease activity using conventional clinical assessments in an observational cohort of patients with early RA (ERAN). It was tested in two trials in early (CARDERA) and established (TACIT) RA. Persistent disease activity was defined as disease activity score for 28 joints (DAS28) >3.2 at both 6 and 12 months. Results: Regression modelling identified three main predictors of persisting active disease in ERAN; tender joint counts, health assessment questionnaire (HAQ) scores and ESR. We dichotomised these predictors (≥6 tender joint counts, ≥1.0 HAQ ≥20 mm/h ESR) in a four-point prediction score. This simple prediction score predicted persisting active disease in the ERAN cohort and both CARDERA and TACIT trials. Patients with high scores were more likely to have persistently active disease at 6 and 12 months. The relationship was weaker in TACIT because no patients were without any predictive factors. Conclusions: Combining tender joint counts, ESR and HAQ in a simple predictive score prospectively identifies patients with higher risks of persistent disease activity over the next 12 months. More patients with all three risk factors had persistent active disease than those with none or one risk factor.