Using a surrogate with heterogeneous utility to test for a treatment effect

Area of Science:

• Biostatistics
• Clinical Trial Design
• Epidemiology

Background:

• Surrogate markers can shorten clinical trials but may have variable utility.
• Ignoring heterogeneity in surrogate marker utility can lead to inaccurate treatment effect conclusions.
• Existing methods may fail when patient populations differ from the marker evaluation study.

Purpose of the Study:

• To develop a novel statistical test for treatment effects that accounts for surrogate marker utility heterogeneity.
• To compare the proposed method against standard tests using primary outcomes and ignoring heterogeneity.
• To validate the new approach using theoretical properties and simulation studies.

Main Methods:

• Development of a new testing procedure for treatment effects incorporating surrogate marker heterogeneity.
• Derivation of asymptotic properties for the proposed estimator and variance estimates.
• Simulation studies to evaluate finite sample performance and compare with existing methods.

Main Results:

• The proposed testing procedure demonstrates validity and accounts for heterogeneity in surrogate marker utility.
• Simulation results show the new approach can outperform methods that ignore heterogeneity.
• The method was illustrated using an AIDS clinical trial with CD4 count as a surrogate marker for RNA.

Conclusions:

• The novel statistical test effectively addresses heterogeneity in surrogate marker utility.
• This approach offers a more accurate way to assess treatment effects in clinical trials with surrogate markers.
• The method provides a valuable tool for analyzing clinical trial data, exemplified by AIDS research.