Bayesian optimal Phase II survival trial design with event-driven approach

Bayesian design incorporates prior knowledge and external information, making it an attractive option during the early phase of a clinical trial. A number of Bayesian optimal designs have been proposed to make go/no-go decisions based on posterior probabilities while also having desired frequentist operating characteristics. However, existing Bayesian designs either are not appropriate for time-to-event endpoints or rely on an exponential distribution assumption on the data. In this paper, we propose a Bayesian optimal Phase II event-driven design (BOP2e) that allows for futility and/or superiority stopping for single-arm trials with a time-to-event endpoint. The proposed BOP2e design is optimal in minimizing the expected sample size under null hypothesis while also controlling the frequentist Type I error. Simulation studies are performed to explore the operating characteristics of the proposed BOP2e designs. A user-friendly Shiny application is available to help implement the proposed designs.