Increasing the efficiency of oncology basket trials using a Bayesian approach

Area of Science:

• Oncology
• Biostatistics
• Clinical Trial Design

Background:

• Targeted and immune-oncology therapies necessitate advanced statistical designs for early-phase oncology trials.
• Basket trials evaluate drugs across multiple tumor types with shared biological targets, but often analyze indications independently, missing potential similarities.

Purpose of the Study:

• To develop and evaluate a novel statistical methodology for enhancing oncology basket trials.
• To improve trial efficiency and power by leveraging biological similarities among different indications.

Main Methods:

• Proposes an interim analysis to assess response rate homogeneity across indications.
• Employs Bayesian hierarchical modeling in the second stage for indications with similar response rates.
• Utilizes simulation studies to quantify efficiency gains compared to conventional parallel designs.

Main Results:

• The proposed method increases study power by enabling information sharing between indications with similar response rates.
• Efficiency gains are substantial when response rates are comparable across most indications.
• The method demonstrates improved performance over conventional approaches even when response rates vary considerably.

Conclusions:

• The Bayesian hierarchical modeling approach enhances oncology basket trials by pooling information from similar indications.
• This methodology offers a more flexible and efficient design for evaluating novel cancer therapies.
• The approach provides a statistically robust framework for adaptive trial designs in precision oncology.