A Bayesian design for dual-agent dose optimization with targeted therapies

Area of Science:

• Clinical trial design
• Pharmacology
• Biostatistics

Background:

• Combination therapies, such as MEK and PIK3CA inhibitors, present challenges in dose selection.
• Higher doses do not always correlate with improved efficacy in targeted therapies.
• Optimizing the risk-benefit profile is crucial for effective treatment strategies.

Purpose of the Study:

• To propose a novel two-stage phase I-II adaptive clinical trial design for molecularly targeted therapy combinations.
• To identify optimal dose combinations that achieve a desirable risk-benefit trade-off.
• To enhance the safety and efficiency of dose-finding studies in oncology.

Main Methods:

• A two-stage design incorporating escalation with overdose control (EWOC) in Stage I.
• Adaptive randomization in Stage II based on continuously updated model parameters.
• Utilizing a flexible cubic spline model to represent efficacy response distributions.

Main Results:

• The proposed design demonstrated superior safety and efficiency in simulations compared to existing algorithms.
• The design effectively identifies optimal dose combinations for targeted therapy regimens.
• Performance was evaluated across various scenarios, including sample size variations and model misspecification.

Conclusions:

• The novel adaptive design offers a safer and more efficient approach for determining optimal dose combinations in targeted therapy trials.
• This methodology facilitates better risk-benefit assessment in complex drug development settings.
• The design is robust and adaptable to different clinical trial parameters and modeling assumptions.