Keywords: oncology, group sequential, clinical relevance, assessment of schedule, PFS
Two issues that should not be overlooked in the design of group sequential oncology trials are the clinical relevance of the assumed treatment effect and the frequency and timing of assessments of progression. Especially for lower-risk disease these studies may take several years. Since the expected efficacy for progression-free survival (PFS) or overall survival (OS) is likely uncertain at the time of design, group sequential evaluations of efficacy can substantially reduce the possible duration relative to conservative assumptions for which a trial is powered. Early stopping bounds should consider the clinical relevance of the minimum treatment effect at the interim or final analysis for which a trial is called positive. The assessment of schedule is usually determined by common clinical practice including clinical factors like radiation exposure for patients as well as the trial budget. While these are factors that would suggest less frequent scanning, they must be balanced against the fact that more frequent scanning would improve the accuracy of estimation of the time-to-event distribution and, specifically, median PFS. A case study considering different designs driven by clinical relevance and simulation results to assess the bias of less frequent assessment of schedule for imaging scans will be presented. Recommendations and remedies by design and analysis will be provided.