Dynamic enrichment of Bayesian small-sample, sequential, multiple assignment randomized trial design using natural...

Area of Science:

• Biostatistics
• Clinical Trial Design
• Rare Disease Research

Background:

• Recruiting patients for rare disease clinical trials, such as Duchenne muscular dystrophy (DMD), is difficult.
• Long-term placebo arms in trials raise ethical and retention concerns, challenging traditional drug development.
• Sequential trial designs face limitations in efficiency and data utilization.

Purpose of the Study:

• To propose a novel small-sample, sequential, multiple assignment, randomized trial (snSMART) design.
• To combine dose selection and confirmatory assessment within a single trial for rare diseases.
• To enhance the efficiency of treatment effect estimation in rare disease drug development.

Main Methods:

• The snSMART design re-randomizes patients to optimal dose levels based on early response.
• External control data (e.g., Duchenne Natural History Study) enriches the placebo arm.
• A robust meta-analytic combined (MAC) approach integrates data from multiple stages and external sources, addressing heterogeneity and bias.

Main Results:

• Reanalysis of a DMD trial using MAC-snSMART showed improved efficiency compared to the original trial.
• The robust MAC-snSMART method demonstrated more accurate estimators than traditional methods in most cases.
• The proposed design effectively utilizes data from all stages and external controls.

Conclusions:

• The snSMART design offers a more efficient approach to drug development for rare diseases like DMD.
• This methodology addresses ethical and retention concerns associated with traditional placebo-controlled trials.
• The MAC-snSMART approach provides a promising framework for optimizing rare disease clinical trials.