A hierarchical Bayesian design for randomized Phase II clinical trials with multiple groups

Enhanced knowledge of the biological and genetic basis of cancer is re-defining the target population for new treatments. In oncology, potential targets for a new therapeutic agent often include various solid and hematologic malignancies that share common signaling pathways. New agents are often tested in multiple tumor types across which information can be borrowed. We propose a hierarchical Bayesian design (HBD) to simultaneously test a novel agent in multiple groups for randomized Phase II clinical trials with binary endpoints. Compared to parallel design for individual tumor groups, the HBD has greatly reduced sample size. Therefore, this improves efficiency and decreases the financial cost of conducting randomized Phase II clinical trials. An R package hbdct has been developed to implement the HBD and streamline the sample size calibration.