A design for phase II testing of anticancer agents within a phase III clinical trial

A design for testing new anticancer agents is proposed such that the initial testing of new agents (phase II trials) is included within the framework of a comparative clinical trial (phase III). Randomization between phase II trials and the treatment groups of the phase III trial enforces consistency of patient selection and evaluation of response criteria. Patients who progress on the phase II trials of the new agents are randomized to one of the treatments of the phase III trial. Design issues, such as sample size and power, and analysis of the proposed design, are discussed. Advantages and disadvantages of the design are illustrated by sample size calculations for a current clinical trial in advanced breast cancer.