Bayesian optimal phase II clinical trial design with time-to-event endpoint

Area of Science:

• Clinical Trial Design
• Biostatistics
• Bayesian Inference

Background:

• Phase II clinical trials are crucial for evaluating treatment efficacy.
• Traditional designs may lack flexibility and power for complex endpoints.
• Time-to-event endpoints, such as progression-free survival (PFS), are common in oncology.

Purpose of the Study:

• To propose a novel Bayesian Optimal Phase II (BOP2) design for clinical trials.
• To accommodate time-to-event endpoints or co-primary endpoints including time-to-event and categorical data.
• To enhance treatment selection by maximizing power while controlling Type I error.

Main Methods:

• Utilizes an exponential-inverse gamma model for time-to-event data.
• Employs adaptive probability cutoffs for go/no-go decisions at interim analyses.
• The BOP2 design is flexible regarding the number of interim looks and trial arms (single or two-arm).

Main Results:

• Simulation studies demonstrate favorable operating characteristics for the BOP2 design.
• The BOP2 design exhibits higher power compared to some existing Bayesian Phase II designs.
• It shows a reduced risk of premature trial termination due to incorrect decisions.

Conclusions:

• The BOP2 design offers a robust and flexible approach for Phase II clinical trials.
• It effectively bridges Bayesian and frequentist design principles.
• The design is practical, easy to implement, and supported by freely available software.