SCI: A Bayesian adaptive phase I/II dose-finding design accounting for semi-competing risks outcomes for...

Area of Science:

• Clinical Trials
• Biostatistics
• Immunotherapy Research

Background:

• Phase I/II immunotherapy trials commonly use progression-free survival as an efficacy endpoint.
• Disease progression in these trials can ethically terminate toxicity events, creating semi-competing risks.
• Late-onset outcomes complicate the analysis of progression-free survival in immunotherapy trials.

Purpose of the Study:

• To propose a novel Bayesian adaptive phase I/II design for immunotherapy trials that accounts for semi-competing risks outcomes.
• To address the challenges posed by late-onset outcomes in the context of semi-competing risks.
• To develop an efficient and adaptive dose-finding algorithm for immunotherapy clinical trials.

Main Methods:

• A novel Bayesian adaptive phase I/II design, termed the dose-finding design accounting for semi-competing risks outcomes (SCI) design, is proposed.
• The likelihood function is reconstructed based on individual patient follow-up times to handle semi-competing risks.
• A data augmentation method is employed for efficient posterior sampling from Beta-binomial distributions.
• A curve-free dose-finding algorithm is developed for adaptive optimal biological dose identification without parametric assumptions.

Main Results:

• The proposed SCI design demonstrates good operating characteristics for dose selection.
• The design shows effectiveness in patient allocation within immunotherapy trials.
• Numerical studies indicate favorable trial duration performance with the SCI design.

Conclusions:

• The novel SCI design effectively manages semi-competing risks and late-onset outcomes in phase I/II immunotherapy trials.
• The Bayesian adaptive approach facilitates accurate optimal biological dose identification.
• The SCI design offers a robust framework for improving immunotherapy clinical trial efficiency and outcomes.