Bayesian Power-Based Sample Size Determination for Single-Arm Clinical Trials With Time-to-Event Endpoints

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Cancer Therapy Research

Background:

• Exploratory clinical trials assess new treatment viability, often using single-arm designs with binary outcomes.
• Time-to-event endpoints are common in cancer therapy, typically analyzed with frequentist methods for sample size determination.
• Bayesian methods offer advantages by integrating prior information and enabling early trial termination.

Purpose of the Study:

• To propose a novel sample size determination method for clinical trials using Bayesian power.
• To leverage posterior and prior predictive probabilities of the hazard ratio for improved sample size calculations.
• To enhance clinical trial efficiency through the incorporation of early termination criteria.

Main Methods:

• Developed a Bayesian sample size determination method based on Bayesian power.
• Utilized posterior probability and prior predictive probability of the hazard ratio, assuming proportional hazards.
• Defined prior information using analysis priors and design priors to express parameter information and uncertainty.
• Extended the study design to include early termination criteria for efficient trial conduct.

Main Results:

• Informative analysis priors led to a decrease in the required sample size.
• Design priors accounting for variance resulted in a more conservative, larger sample size.
• The proposed method allows for sample size optimization while utilizing available prior information.

Conclusions:

• The Bayesian sample size determination method offers a flexible approach for clinical trial design.
• Incorporating prior information can significantly impact sample size, allowing for either reduction or a more conservative estimate.
• The inclusion of early termination criteria enhances the efficiency of clinical trial conduct.