Sample Size Reassessment and Hypothesis Testing in Adaptive Survival Trials

Area of Science:

• Biostatistics
• Clinical Trial Design
• Survival Analysis

Background:

• Mid-study design modifications are increasingly accepted in clinical trials.
• Standard statistical theories struggle with time-to-event endpoints during interim analyses.
• Ensuring controlled error rates is crucial for trial validity.

Purpose of the Study:

• To analyze existing methods for mid-study design modifications using full interim data for time-to-event endpoints.
• To develop and evaluate an alternative statistical test that incorporates all observed event times.
• To assess the power of the proposed method in a cancer therapy clinical trial context.

Main Methods:

• Analysis of current statistical approaches for adaptive clinical trials with time-to-event outcomes.
• Development of a novel test statistic that utilizes all available event time data.
• Examination of type I error control through conservative assumptions.
• Power analysis using a simulated cancer clinical trial comparing two therapies.

Main Results:

• Current methods may exclude significant portions of observed event times when design modifications are made.
• An alternative test incorporating all event times is proposed.
• This alternative test requires a conservative assumption to maintain type I error control.
• The power of the new approach is evaluated in a relevant clinical trial scenario.

Conclusions:

• Addressing mid-study design modifications for time-to-event endpoints requires specialized statistical methods.
• The proposed alternative test offers a way to include all event data while controlling statistical error.
• Further investigation and validation are needed for widespread adoption in clinical research.
• This work contributes to the robust design of adaptive clinical trials for critical endpoints.