Bayesian adaptive randomization in the I-SPY2 sequential multiple assignment randomized trial

Area of Science:

• Oncology
• Clinical Trials
• Biostatistics

Background:

• The I-SPY2 trial is a Phase 2 platform evaluating neoadjuvant treatments for locally advanced breast cancer.
• It utilizes response-adaptive randomization (RAR) to assign patients to novel agents.
• Recently, I-SPY2 was reconfigured as a sequential multiple assignment randomized trial (SMART) with up to three therapy stages.

Purpose of the Study:

• To identify highly efficacious neoadjuvant treatment regimens for locally advanced breast cancer.
• To develop and implement a Bayesian RAR scheme for a SMART trial.
• To maximize the number of patients achieving pathological complete response (pCR).

Main Methods:

• The study employs a sequential multiple assignment randomized trial (SMART) design.
• A Bayesian response-adaptive randomization (RAR) approach was developed to update randomization probabilities at each stage.
• Randomization probabilities are updated based on the posterior probability that treatments are part of the optimal regimen.

Main Results:

• The Bayesian RAR approach resulted in more patients receiving treatment consistent with highly efficacious regimens.
• The method improved overall within-trial pathological complete response (pCR) rates.
• Optimal treatment regimes were identified post-trial at rates similar to or exceeding nonadaptive randomization.

Conclusions:

• The developed Bayesian RAR scheme effectively identifies optimal treatment regimens in the I-SPY2 SMART trial.
• This adaptive approach enhances patient treatment experience and improves trial-wide pCR rates.
• The SMART design with Bayesian RAR is a powerful strategy for accelerating the development of effective breast cancer therapies.