An adaptive design for phase II non-oncology dose selection clinical trials.

A non-oncology dose selection phase II trial tests multiple active doses in a controlled fashion, and it not only needs to determine whether the treatment is effective but also to select the 'lowest efficacious' dose if the treatment is indeed efficacious. Few approaches exist in the literature for designing phase II non-oncology dose selection trials, and the standard design with a fixed sample size has been widely used. The objective of this study was to develop a more efficient design for phase II dose selection trials that terminates the trial early for futility and adjusts the sample size and number of doses at interim analyses when appropriate. One-sided statistical tests and confidence intervals were used to develop an adaptive design for non-oncology phase II dose selection trials. With several interim analyses built in, the adaptive design uses accumulated data to determine, at each interim analysis, whether the highest dose is efficacious and whether the low doses are as efficacious as the highest dose. Once a confident answer to either or both of these questions can be obtained, the trial may either be terminated early or some of the lower doses may be dropped to prevent assigning more patients to inferior doses and thus reduce the total sample size needed. Theoretical analyses and simulation studies show that the proposed adaptive design significantly outperforms the standard design with a fixed sample size. The proposed adaptive design should be preferred over the standard design especially in cases where enrolment is slow and efficacy can be measured after a relatively short period of time.