Analyses of drug combinations using missing data shortens trial periods in phase I/II oncology trials

Area of Science:

• Oncology
• Clinical Trial Design
• Biostatistics

Background:

• Traditional phase I/II oncology trials risk prolonged durations due to reliance on complete best overall response and toxicity data.
• Existing methods for dose combination selection in early-phase oncology trials can lead to unfeasibly long study periods.

Purpose of the Study:

• To propose and evaluate a novel approach for shortening the duration of phase I/II oncology drug combination trials.
• To enable earlier patient access to potentially effective cancer therapies and reduce clinical development costs.

Main Methods:

• A novel method is proposed where the next cohort's dose combination is determined before all current patient data is finalized.
• Efficacy in patients with undetermined best overall response is treated as missing data, modeled using nonparametric prior processes.
• Data augmentation is applied to estimate efficacy and toxicity probabilities, guiding the selection of the next cohort's dose combination.

Main Results:

• Simulation studies indicate the proposed approach significantly shortens oncology trial durations compared to existing methods.
• The method maintains trial design performance without compromising results.
• Reduced trial duration allows for earlier patient treatment and decreased costs for sponsors.

Conclusions:

• The proposed method offers an efficient strategy to accelerate oncology drug development.
• Shortening trial durations benefits both patients seeking timely treatment and sponsors aiming for cost-effective drug development.
• This innovative approach addresses a critical bottleneck in early-phase oncology clinical trials.