Director of Biostatistics Shared Resource
Professor, Department of Computational Biomedicine
Samuel Oschin Cancer Center at Cedars-Sinai
Abstract: We present a two-stage Bayesian adaptive phase I/II design of a combination of cisplatin and cabazitaxel with continuous dose levels in the treatment of prostate cancer patients with visceral metastasis. The goal is to estimate dose combination regions that are tolerable and with the desired level of clinical benefit rate. In the first stage of the design, informative priors were used to calibrate the model parameters taking into account that the combination at 75 mg/m2 cisplatin and 15 mg/m2 cabazitaxel was tolerated without DLT observed in cycle 1 from a previous trial. In the second stage of the design, adaptive randomization is used to allocate patients to dose combinations along the MTD curve obtained from stage 1. The goal is to determine dose combination regions along the MTD curve with a clinical benefit rate that exceeds 0.15. An extension of the design where the MTD curve is updated during stage 2 is presented. Statistical power and Bayesian type I error are evaluated by presenting the operating characteristics under different scenarios for the true probability of treatment benefit rate.