A flexible multi-metric Bayesian framework for decision-making in Phase II multi-arm multi-stage studies.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Multi-arm multi-stage phase II trials accelerate drug development but pose risks for early decisions due to small sample sizes and short follow-ups.
• Intermediate outcomes from biomarkers may not perfectly predict primary outcomes, and varying stakeholder risk tolerance necessitates comprehensive evidence summary beyond single hypothesis tests.

Purpose of the Study:

• To propose a flexible Bayesian framework supporting efficient interim decision-making in multi-arm multi-stage phase II clinical trials.
• To develop a multi-metric approach for ranking and comparing treatment arms against internal controls, considering point estimates, uncertainty, and evidence towards a Target Product Profile.

Main Methods:

• A Bayesian framework incorporating multiple metrics: point estimates, uncertainty quantification, and evidence towards desired thresholds (Target Product Profile).
• Application of the framework to a public-private partnership targeting novel tuberculosis (TB) arms.
• Simulation studies to evaluate framework performance with varying sample sizes and biomarker-outcome correlations.

Main Results:

• The multi-metric Bayesian framework provides sufficient confidence for interim decision-making with sample sizes as low as 30 patients per arm.
• The framework remains effective even when intermediate outcomes show only moderate correlation with the primary clinical outcome.
• Demonstrated utility in a TB drug development context.

Conclusions:

• The proposed flexible Bayesian framework offers a practical and efficient approach for assessing novel therapeutics in early-phase clinical trials.
• Reframing trial design and decision-making procedures enhances confidence and efficiency in drug development.
• The multi-metric approach accommodates diverse stakeholder needs and risk tolerances.