Some ethical issues in phase II trials in acute leukemia

This paper addresses several scientific and ethical issues that arise in the design and conduct of phase II clinical trials of experimental therapies. Although we discuss chemotherapy trials in acute leukemia, the issues pertain to a much larger class of earlyphase clinical trials. Our focus is on the manner in which numerical values of standard and targeted response rates of a statistical design are specified, the number of interim tests that are applied, and the effects of these design features on early stopping probabilities and the treatments that patients actually receive. These points are illustrated by numerical comparisons of alternative designs for a particular phase II trial of a new drug for relapsed or refractory acute myelogenous leukemia. We show that statistical designs that target inappropriately low response rates or that apply early stopping rules too infrequently are at odds with good statistical and medical practice and that such designs often provide less benefit to the patients in the trial than would be obtained by simply treating all patients with standard therapy. The general conclusions are that statistical designs have both scientific and ethical implications, and that science, statistics, and ethics cannot be treated as separate issues.