Treatment Selection at Interim Analysis Based on a Correlated Early Endpoint in Adaptive Two-Stage Trials with a Primary Endpoint Based on Count Data

Adaptive two-stage design with treatment selection, also called seamless Phase II/III design, has been recognized to save sample size and time to enhance the development of effective treatments. These designs have been successfully implemented to support drug approval. When the primary endpoint takes a long time to observe, it is often desirable to base the treatment selection on early outcomes. However, existing literature predominantly focuses on such trial designs with continuous or binary endpoints, and there is limited research on those with recurrent event endpoints that have been used in various therapeutic areas. In this paper, we exploited a correlated early endpoint that matures faster and provides increased power for dose selection in trials with recurrent event endpoints. We addressed unique challenges that arise in practical scenarios, allowing for dose switching after treatment selection, and examined issues concerning the performance of adaptive two-stage designs, including type I error rate control in the final analysis and correlation between the recurrent event and the early endpoint. We compared performances of the inferential seamless design with p-value combination test, the inferential seamless design with Bonferroni adjustment, and the operational seamless design via simulation studies. Pros and cons of these design options were discussed.