Phase II clinical trials in oncology: Strengths and limitations of two-stage designs

Two-stage designs are used widely in Phase II oncology clinical trials to reduce the number of patients placed on ineffective experimental therapies. They provide clear-cut rules for stopping early in the event that treatment is not succeeding as hoped and are relatively simple to implement. Such designs, however, can lead to situations in which patient accrual is continued in the face of a clearly inferior treatment. In situations where patients' response can be determined for many or most subjects before additional patients are enrolled, analyses using Bayesian methodology can lead to earlier termination of studies of ineffective treatments and better align the statistical assessment of treatment effect with the therapeutic objectives of study. These points are discussed in context of the role of Phase II clinical trials in the development of new treatments for cancer.