|
Abstract:
|
Increasing scientific knowledge is creating substantial opportunities and challenges in oncology drug development. As diseases are sub-stratified into biomarker-based groups, usual paradigms for phase II and III trials may no longer apply. In some circumstances, a carefully conducted retrospective-prospective analysis may provide sufficient evidence of a predictive biomarker for clinical use. Prospectively planned, enrichment designs are appropriate when preliminary evidence suggests that patients with/without that marker profile do not benefit from treatments in question. An unselected design is optimal where preliminary evidence regarding treatment benefit and assay reproducibility is uncertain. The biomarker-based strategy design may be useful when there is a choice between many treatment options. Adaptive analysis designs allow for pre-specified data-driven marker-defined subgroup analyses from a RCT. Umbrella or basket trials enroll large groups of patients with subsequent assignment to either individual randomized trials or single arm investigations. We discuss features of these various novel design strategies in the context of multiple ongoing and planned real trials.
|
