[Research design for patient selection in clinical trials phase II].

Phase II trials play a key role in the development of a treatment, because they are crucial in deciding whether or not to proceed to phase III. In the event of the evaluation of a new treatment given to only one group of subjects, multi-stage designs were developed to include subjects in phase II trials. Gehan's two stage designs stop inclusion of subjects if no success is observed among the first n1 subjects. They are easy to use, but have the disadvantage of exposing many subjects to an inefficient treatment at the second stage. Fleming's multi-stage designs and Simon's two stage designs permit the early termination of a trial when the intermediate results are extreme, either in favor of the efficiency or in favor of the inefficiency of the treatment. Simon's plans minimize either the average or the maximum number of subjects exposed to ineffective treatment. Ensign's three stage designs combine the first stage of Gehan with the two stages of Simon. These plans are interesting because they can reject a treatment quickly after a long run of failures which minimizes the average number of subjects exposed to an ineffective treatment.