A Novel Method for Inserting Dose Levels Mid-Trial in Early-Phase Oncology Combination Studies

Area of Science:

• Oncology
• Clinical Trials
• Biostatistics

Background:

• Combination treatments are increasingly used in early-phase oncology trials.
• Identifying the maximum tolerated dose (MTD) combination is a key objective.
• Challenges arise when the initial dose set lacks combinations near the target toxicity.

Purpose of the Study:

• To propose a novel method for inserting dose levels mid-trial in early-phase oncology studies.
• To address the challenge of finding MTD combinations when initial sets are suboptimal.
• To improve the precision of MTD identification in dose-finding trials.

Main Methods:

• Developed a method to search for a toxicity contour partitioning the dose space.
• The contour identifies combinations with toxicity above and below the target.
• The method triggers dose level insertion when no combination is close to the target toxicity.
• Evaluated the approach using simulations with the PIPE design and a 2D Bayesian logistic regression model (BLRM).

Main Results:

• The proposed insertion method increases the probability of selecting combinations close to the target toxicity.
• The method does not increase the probability of recommending subtherapeutic or toxic doses.
• Simulation results demonstrate the effectiveness of the novel approach.

Conclusions:

• The novel mid-trial dose insertion method offers a systematic solution for MTD finding in oncology trials.
• This approach enhances the ability to identify optimal dose combinations, improving trial success rates.
• The method is compatible with various model-based and model-assisted designs.