STATISTICAL DESIGNS FOR EARLY PHASES OF CANCER CLINICAL TRIALS

The main purpose of a Phase I trial of a new antitumor agent is to determine the appropriate dosing regimen and characterize the safety profile of a new molecular or monoclonal antibody. Phase II cancer clinical trials are conducted to assess the efficacy of a new anticancer therapy and to determine whether it has sufficient activity against a specific type of tumor to warrant further development. In this paper, commonly used statistical designs, based on either frequentist approaches or Bayesian methods, for Phase I and Phase II cancer clinical trials are reviewed and discussed. Future directions of designing more efficient trial are explored.