How to add baskets to an ongoing basket trial with information borrowing

Area of Science:

• Clinical Trial Design
• Biostatistics
• Pharmacology

Background:

• Basket trials evaluate a single treatment across multiple patient groups under one protocol.
• Adaptive designs allowing new basket incorporation are desirable but face challenges with small sample sizes and power.
• Existing Bayesian methods for small sample sizes have not addressed the impact of adding new baskets.

Purpose of the Study:

• To investigate methods for incorporating new baskets into ongoing adaptive basket trials using information borrowing.
• To develop and evaluate a novel calibration approach for efficacy criteria to ensure robust error control.
• To assess the performance of these methods regarding power and type I error rates.

Main Methods:

• Exploration of information borrowing techniques for adding baskets to ongoing trials.
• Development of a novel calibration strategy for efficacy criteria, considering alternative scenarios beyond the null.
• Simulation studies to evaluate type I error control and power for new basket incorporation.

Main Results:

• Adding new baskets can significantly improve statistical power for detecting treatment effects.
• Potential inflation of error rates was observed with standard approaches when incorporating new baskets.
• The proposed calibration procedure demonstrated a reduction in error rate inflation.

Conclusions:

• The developed methods offer improved power when adding baskets to adaptive trials.
• A novel calibration approach enhances the reliability of efficacy assessments and controls decision-making errors.
• This research provides a more robust framework for adaptive basket trial design and analysis.