A multi-arm multi-stage clinical trial design for binary outcomes with application to tuberculosis.

Area of Science:

• Clinical trial methodology
• Pharmaceutical development
• Infectious disease research

Background:

• Randomised controlled trials (RCTs) face increasing cost and time demands.
• Multi-arm multi-stage (MAMS) designs were proposed to expedite treatment evaluation.
• MAMS designs discontinue poorly performing arms at interim analyses to improve efficiency.

Purpose of the Study:

• Adapt Royston's MAMS design for binary outcomes in tuberculosis (TB) trials.
• Demonstrate acceleration of new TB regimen evaluation using a seamless phase II/III design.
• Investigate bias in treatment effect estimation for arms continuing or ceasing randomization.

Main Methods:

• Adapted MAMS design for binary outcomes analyzed by absolute difference in proportions.
• Applied the design to tuberculosis drug development scenarios.
• Utilized simulations to validate methodology extensions and assess bias.

Main Results:

• Seamless phase II/III TB trial designs substantially reduce sample size and duration compared to separate trials.
• Bias in estimated treatment effects for the definitive outcome is minimal.
• Bias is particularly small when treatment selection uses intermediate outcomes or post-trial reanalysis.

Conclusions:

• The proposed seamless phase II/III designs are practical for various disease areas.
• These designs offer significant promise for accelerating new treatment evaluations, especially for TB.
• The methodology facilitates efficient testing of multiple new TB regimens.