Feasibility and predictive value of proposed early endpoints for randomized controlled trials in knee osteoarthritis

The designation of osteoarthritis (OA) as a serious disease allows for accelerated approval of a treatment based on an intermediate clinical endpoint. We evaluated proposed endpoints and assessed their feasibility for use in a randomized controlled trial (RCT). We examined associations between endpoints and subsequent total knee replacement (TKR), a proxy for long-term clinical benefit.

We selected knees from the Osteoarthritis Initiative meeting typical RCT inclusion criteria, including pain and radiographic OA. Endpoints included TKR, end-stage knee OA (esKOA), Composite Knee Osteoarthritis Symptom Outcome (CKOASO), the FNIH OA Biomarkers Consortium endpoint, and several combinations of pain and/or functional limitations. We determined the cumulative incidence (CI) over four years and the associated sample size required for an RCT to detect a hazard ratio (HR) of 0.67 with 80% power (to assess feasibility). We assessed the association between reaching each endpoint and TKR over the subsequent 5 years.

1350 knees (one per participant) were included. 4-year CI of TKR was 4.3%. CIs were highest for FNIH (13.8%), esKOA (34.4%), and CKOASO (63.6%). The required sample size per arm ranged from 229 (CKOASO) to 3028 (TKR). The relative risk (RR) of subsequent TKR was highest for the esKOA endpoint (RR 5.7), followed by CKOASO (4.3) and FNIH (2.6).

Based on feasibility (sample size) and clinical relevance (association with subsequent TKR), we propose that the esKOA, CKOASO, and FNIH endpoints be prioritized for further consideration. Given feasibility limitations, we would not recommend TKR alone as a DMOAD trial endpoint.