A two-stage algorithm for designing phase I cancer clinical trials for two new molecular entities

Area of Science:

• Oncology
• Biostatistics
• Clinical Trial Design

Background:

• Phase I cancer clinical trials traditionally focus on single new molecular entities.
• Simultaneous targeting of multiple pathways with two molecular entities is a growing therapeutic strategy.
• Existing Bayesian methods like the continual reassessment method (CRM) are established for single-agent trials.

Purpose of the Study:

• To introduce a novel two-stage algorithm for designing Phase I cancer clinical trials involving two new molecular entities.
• To adapt Bayesian and likelihood-based approaches for the complexities of dual-agent therapy evaluation.
• To improve the efficiency and safety of dose escalation for combination cancer treatments.

Main Methods:

• A two-stage algorithm combining a modified continual reassessment method (CRM) with likelihood-based estimation.
• Stage 1 utilizes a modified CRM for initial dose-finding with two agents.
• Stage 2 employs accumulated data from Stage 1 to refine model parameter estimates for subsequent dose selection.

Main Results:

• The proposed algorithm provides a framework for designing Phase I trials of two new molecular entities.
• The two-stage approach allows for data-driven adaptation and improved estimation of toxicity.
• This method facilitates the efficient exploration of dose-combinations for novel cancer therapies.

Conclusions:

• The developed two-stage Bayesian and likelihood-based algorithm is suitable for Phase I clinical trials of two new molecular entities.
• This approach offers a robust statistical framework for dose-finding in combination cancer therapy development.
• The methodology enhances the design of early-phase trials for dual-agent cancer treatments.