Abstract
In oncology trials, noninferiority concepts (e.g., shortening duration of chemotherapy) have sometimes been evaluated without preliminary knowledge of an experimental treatment being noninferior to the standard treatment, and thus, a prompt consideration for futility stop is vital. Because such concepts are usually examined in a patient population with a good prognosis, it is often the case that the number of events is quite small at earlier interim analyses. In the present study, we numerically assess the power reduction of when an aggressive inefficacy monitoring based on an earlier-available intermediate endpoint is used. We then propose a sample size determination method, and investigate the performance. An actual phase 3 oncology trial is used as a motivating example.
Acknowledgments
We would like to thank Dr. Yuichiro Ohe, Dr. Yasushi Goto, and Dr. Hidehito Horinouchi at National Cancer Center Hospital, Japan for permission to use design details of the SAVE study (JCOG1701) as an application example. We also thank the Editor, the Associate Editor, and two anonymous referees for their helpful comments.