Biomarker informed add-arm design for unimodal response

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Dose-finding studies are crucial for identifying optimal drug dosages.
• Existing designs may not fully leverage biomarker information or unimodal dose-response relationships.
• Biomarkers can predict treatment efficacy and guide study design.

Purpose of the Study:

• To propose a novel biomarker-informed add-arm design for dose-finding.
• To optimize dose selection when a biomarker for the primary endpoint exists and a unimodal dose-response relationship is expected.
• To evaluate the statistical performance, including Type I error control and power, of the proposed design.

Main Methods:

• Development of a statistical framework for a biomarker-informed add-arm design.
• Consideration of designs with up to seven active treatment arms.
• Extensive simulation studies to assess power and Type I error control.

Main Results:

• The proposed biomarker-informed add-arm design effectively controls Type I error.
• The design demonstrates superior power performance compared to a biomarker-informed two-stage winner design.
• Simulations confirm the practical utility and efficiency of the new design.

Conclusions:

• The proposed add-arm design is a statistically sound and powerful approach for dose-finding.
• Integrating biomarker information optimizes dose selection in unimodal response settings.
• This design offers an advancement in clinical trial methodology for drug development.