Estimating the proportion of treatment effect explained by a surrogate marker

Area of Science:

• Biostatistics
• Clinical Trials
• Epidemiology

Background:

• Surrogate endpoints accelerate clinical trial evaluation.
• Quantifying the validity of surrogate endpoints is essential for reliable trial results.
• Existing methods for assessing surrogate endpoint validity can be complex.

Purpose of the Study:

• To develop and validate a statistical method for measuring the surrogate endpoint potential of a biological marker.
• To provide a quantitative measure of the proportional reduction in treatment effect attributable to a marker.
• To enable more efficient and accurate clinical trial analysis.

Main Methods:

• Utilizing Cox regression models to assess the impact of biological markers on clinical outcomes.
• Applying partial likelihood functions to estimate the proportion of treatment effect reduction.
• Developing asymptotically normal estimators with simple variance estimation for practical application.

Main Results:

• The proposed method provides a statistically sound measure of surrogate endpoint validity.
• The estimator is shown to be asymptotically normal with a straightforward variance estimator.
• Confidence intervals can be constructed using normal approximation or Fieller's theorem.

Conclusions:

• The developed statistical approach offers a practical and robust way to evaluate biological markers as surrogate endpoints.
• The methods are validated through extensive simulation studies, confirming their suitability for real-world applications.
• This work has direct implications for the design and interpretation of HIV/AIDS clinical trials.