A Bayesian adaptive design in cancer phase I/II trials with drug combinations using escalation with overdose control (EWOC) and adaptive randomization
share
The use of drug combinations in clinical trials is increasingly common during the last years since a more favorable therapeutic response may be obtained by combining drugs that, for instance, target multiple pathways or inhibit resistance mechanisms. However, most of the existing methodology in phase I trials recommends a single maximum tolerated dose (MTD), which may result in a failed phase II since other MTDs may present higher treatment efficacy for the same level of toxicity. We are motivated by a phase I/II trial that combines cisplatin with cabazitaxel for patients with prostate cancer with visceral metastasis. We present a Bayesian adaptive phase I/II design with drug combinations where a binary dose limiting toxicity (DLT) is used for dose escalation in stage 1 and a time to event endpoint is used for dose allocation in stage 2. The overall goal is to estimate the dose combination region associated with the highest median time to progression (TTP) among doses along the MTD curve. Conditional escalation with overdose control (EWOC) is used in stage 1 to allocate dose combinations to subsequent cohorts of patients and estimate the MTD. Stage 2 starts by allocating a first cohort of patients to dose combinations equally spaced along the MTD curve, and then allocates subsequent cohorts of patients to dose combinations likely to have high posterior median TTP using adaptive randomization. We perform extensive simulation studies to evaluate the operating characteristics of our method.
READ FULL TEXT