MIDAS-2: an enhanced Bayesian platform design for immunotherapy combinations with subgroup efficacy exploration

Area of Science:

• Oncology
• Biostatistics
• Clinical Trial Design

Background:

• Immunotherapy combinations have transformed cancer treatment but rapid screening of optimal therapies and subgroup efficacy is challenging.
• Innovative and efficient clinical trial designs are crucial for advancing cancer therapy development.
• Existing platform designs often lack integrated subgroup analysis capabilities.

Purpose of the Study:

• To extend the MIDAS design to incorporate subgroup exploration for immunotherapy combinations.
• To propose an enhanced Bayesian information borrowing platform design, MIDAS-2.
• To enable efficient screening of promising combination strategies and their subgroup effects within a unified framework.

Main Methods:

• Utilized a regression model to characterize efficacy patterns within patient subgroups.
• Implemented Bayesian hierarchical modeling for information borrowing to enhance trial efficiency in subgroups.
• Employed time trend calibration to mitigate potential baseline drifts during the trial.

Main Results:

• MIDAS-2 demonstrated high probabilities of identifying effective drug combinations and promising subgroups.
• The design facilitates the selection of optimal treatments tailored to specific patient subgroups.
• The proposed design showed robustness against minor time trend drifts and controlled Type I error after calibration for significant drifts.

Conclusions:

• MIDAS-2 offers an adaptive framework for drug screening and subgroup exploration in immunotherapy.
• This design accelerates immunotherapy development through efficient, accurate, and integrated approaches.
• MIDAS-2 supports personalized medicine by optimizing treatment selection based on subgroup efficacy.