Dose-Finding Methods for Two-Agent Combination Phase I Trials

Abstract

In this chapter, we discuss the toxicity-based dose-finding methods for two-agent combinations in phase I oncology trials. The model-based approaches, such as the continual reassessment method (CRM), have been gradually applied to single-agent trials to determine the maximum tolerated dose (MTD). By contrast, the rule-based approaches have commonly been applied to two-agent combination trials, probably due to the absence of well-understood model-based methods for two-agent combination trials. In developing a dose-finding method for two-agent combination trials, we require a reasonable model that can adequately capture joint toxicity probabilities for two agents, taking into consideration of possible interactions of the two agents on toxicity probability (such as synergistic or antagonistic effects). We provide an overview of two useful dose-finding approaches based on Bayesian copula-type models and partial orderings across dose levels for two-agent combination trials. We also supply examples of successful software implementations and discuss the operating characteristics of these approaches.