Model-based designs for phase I cancer clinical trials

Phase I clinical trials are conducted in all disease areas, but they are particularly important in cancer because of side effects associated with cytotoxic drugs for treating cancer patients. The main goal of phase I cancer clinical trials is to determine the highest dose of a new therapy associated with an acceptable level of toxicity for use in a subsequent phase II trial. More precisely, it is to estimate the maximum tolerated dose (MTD) such that the probability of the dose-limiting toxicity (DLT) given the MTD is equal to a target toxicity level. Here a "toxicity response" refers to a toxicity of grade three or higher as defined by the National Cancer Institute (NCI) Common Toxicity Criteria (CTC).; Two model-based designs for phase I clinical trials in cancer, the continual reassessment method (CRM, O'Quigley et al. [34]) and the escalation with overdose control method (EWOC, Babb et al. [3]), are considered here. In this research, we present and compare these two designs and propose a hybrid of these two so that we have faster convergence over EWOC and better safety protection over CRM for phase I cancer clinical trials. The performance of three designs (CRM, EWOC, and the hybrid designs) is also evaluated in terms of sensitivity to outliers.