Determination of the most appropriate ACR response definition for contemporary drug approval trials in rheumatoid arthritis

Objectives: To evaluate which American College of Rheumatology (ACR) response definition (ACR20, 50 or 70) should primarily be used for efficacy claims in future drug approval trials of rheumatoid arthritis (RA).

Methods: We systematically searched EMBASE, Medline and the Cochrane Library for randomised controlled RA drug approval trials of biological and targeted synthetic disease-modifying antirheumatic drugs (DMARDs). We included full-text articles reporting ACR response rates for multiple time points over a 24-week placebo-controlled period and visualised normalised response trajectories over time in different patient populations. Using mixed-effect logistic regression, we calculated the proportion of ACR responders per outcome and time point, and compared the discriminant validity of these metrics at multiple time points.

Results: We screened 12 680 records and included 45 in the final analysis. Discriminative capacity of the ACR20 was high across all time points, whereas ACR50 and ACR70 showed highest discrimination towards the end of the placebo-controlled periods. This effect could be observed in all patient populations and compound groups. Faster response to treatment was observed in DMARD naïve patient populations when compared with DMARD insufficient responders.

Conclusion: ACR20 remains the most powerful discriminator between active treatment and placebo, especially when early discrimination is of primary interest. At the same time, our results support the selection of more stringent thresholds if later time points shall be evaluated, given their comparable discriminant but higher clinical face validity.