A Preplanned Multi-Stage Platform Trial for Discovering Multiple Superior Treatments With Control of FWER and Power

Area of Science:

• Clinical Trials
• Biostatistics
• Statistical Methodology

Background:

• Platform trials offer flexibility in adding/removing treatment arms and early stopping.
• Controlling error rates, particularly family-wise error rate (FWER), is crucial in complex trial designs.
• Existing methods may not adequately address sample size and error control when new arms are added sequentially.

Purpose of the Study:

• Introduce a pre-planned multi-stage design for platform trials allowing sequential addition of treatment arms.
• Develop methods to maintain family-wise error rate control throughout the trial.
• Determine sample size requirements for achieving desired statistical power, even after a superior treatment is identified.

Main Methods:

• Propose a multi-stage statistical design for platform trials.
• Derive calculations for expected sample size.
• Evaluate the design using a motivating clinical trial example and compare configurations.

Main Results:

• The proposed design enables sequential addition of treatment arms while preserving family-wise error rate control.
• Sample size calculations are provided for various scenarios, including continued testing after superiority is found.
• Comparisons suggest platform trials may not always yield sample size benefits over multiple separate trials when FWER control is required.

Conclusions:

• The developed multi-stage design offers a structured approach to flexible platform trials with robust error control.
• Understanding sample size implications is key for efficient platform trial implementation.
• The findings highlight the need for careful consideration of platform trial advantages versus traditional designs based on specific statistical goals like FWER control.