Efficacy of many novel therapies in oncology might be restricted to a subset of patient population. Predictive biomarker(s) have been co-developed with novel therapies to achieve precision medicine. Adaptive seamless design with subpopulation selection has been recently proposed to conduct interim analysis for selection and/or enrichment of patient population and combined phase II and phase III data for final analysis of efficacy. However, the application of adaptive seamless design in phase II/III oncology trials must face a few challenges: choice of interim endpoint, decision rule for subpopulation selection, performance criteria for design evaluation and so on.
We will demonstrate a case study of adaptive seamless design with subpopulation selection in a phase II/III oncology trial through a retrospective real example, and present accompanying simulation studies to evaluate its operating characteristics through working through those challenges.