|
Abstract:
|
Phase I clinical trials are increasingly designed to examine dual-agent therapies, but existing Phase I clinical trial designs identify the MTD of a single agent, and few published designs for Phase I trials of two agents exist. We propose a hierarchical Bayesian design for Phase I clinical trials designed specifically to examine the combined toxicity of two previously studied agents. We directly model the combined effects of both agents on the probability of toxicity and constrain the model so that the probability of toxicity increases with dose increases of either agent. We use adaptive Bayesian methods to determine the combination assigned to each subject. Through simulations, we demonstrate excellent operating characteristics for our design, in terms of identifying the optimal combination, as well as assigning a sufficient proportion of study subjects to the optimal combination.
|
