A Bayesian phase I/II platform design with survival efficacy endpoint for dose optimization

Motivated by a real-world drug development program, we propose a Bayesian phase I/II platform design to co-develop therapies with time-to-event efficacy endpoint (BPCT). We jointly model the binary toxicity outcome and the time-to-event efficacy outcome, leveraging a Bayesian hierarchical framework to enable information sharing across indications. At each interim, we update the dose-toxicity and dose-efficacy estimates, as well as the utility for risk-benefit tradeoffs, based on observed data from all indications. This approach informs indication-specific decisions for dose escalation and de-escalation, and identifies the optimal biological dose for each indication. Simulation studies show that the proposed design has desirable operating characteristics, providing a highly flexible and efficient approach for dose optimization. The design has great potential to shorten the drug development timeline, save cost by reducing overlapping infrastructure, and expedite regulatory approval.