Predictive ability, validity, and responsiveness of the multi-biomarker disease activity score in patients with rheumatoid arthritis initiating methotrexate

Abstract

Background/ObjectiveWe assessed the predictive value, validity, and responsiveness of the multi-biomarker disease activity (MBDA) score in rheumatoid arthritis (RA) patients initiating methotrexate. MethodsWe examined data from a 16-week, open-label study of methotrexate in RA. Disease activity was assessed and the MBDA score was calculated using serum that was collected and banked from baseline and week 16. Multivariable logistic regression models assessed whether MBDA scores predicted treatment response. Pearson correlations assessed the convergent validity and external responsiveness of the MBDA score with other measures of RA disease activity. Internal responsiveness was assessed by calculating standardized response means (SRMs). ResultsA total of 130 patients initiated the study, with follow-up MBDA scores available on 95 patients. Baseline MBDA scores did not predict ACR response or achieving low disease activity. Higher baseline DAS28-ESR scores were significantly associated with an ACR20 response (odds ratio 1.89 per unit, 95% CI 1.20–2.96) but not ACR50, ACR70, or low disease activity. The MBDA score moderate-to-weakly correlated with the DAS28-ESR and ESR at baseline and week 16, with weak-to-very weak correlations with patient global and function. Change in MBDA scores moderately correlated with changes in DAS28-ESR and ESR, while weakly correlating with changes in patient global and function. The DAS28-ESR (SRM 1.31) demonstrated greater responsiveness following methotrexate treatment than the MBDA score (SRM 0.71). ConclusionsMBDA scores did not predict treatment response to methotrexate. The MBDA score weak-to-moderately correlated with baseline and post-treatment disease activity measures and was less responsive to methotrexate-related improvement than the DAS28-ESR.