Optimal single-arm two-stage designs with consideration of dependency on efficacy and safety.

Abstract

In phase II clinical trials on cancer, it is of great interest to establish the efficacy and safety of a new treatment simultaneously. Existing hypotheses may not achieve this goal effectively. We introduce two new sets of hypotheses that consider the association between the two factors, then construct the optimal two-stage designs for the hypotheses. The proposed designs strictly control the maximum type I error rate at the given nominal level , maintain the minimum power at least the given , and have the smallest expected total sample size under the null hypothesis. Furthermore, an algorithm is provided to compute these designs. R-codes are given in the Supplemental Material.