Model-Free Approach to Evaluate a Censored Intermediate Outcome as a Surrogate for Overall Survival

Area of Science:

• Biostatistics
• Clinical Trial Design
• Survival Analysis

Background:

• Clinical trials often need long-term participant follow-up, increasing costs and time.
• Short-term surrogate outcomes are increasingly used to predict long-term primary outcomes.
• Existing statistical methods for evaluating surrogates with time-to-event data are limited.

Purpose of the Study:

• To propose a novel nonparametric approach for evaluating censored surrogate outcomes.
• To assess the treatment effect on a primary time-to-event outcome using a surrogate.
• To define and estimate the proportion of treatment effect explained (PTE) by a surrogate.

Main Methods:

• Developed a nonparametric method for censored time-to-event surrogate and primary outcomes.
• Defined and derived an optimal transformation to estimate the proportion of treatment effect explained (PTE).
• Utilized simulation studies and a real data application (metastatic colorectal cancer) for validation.

Main Results:

• The proposed method is model-free and has relaxed assumptions.
• The approach guarantees the true PTE is within (0, 1).
• Demonstrated effectiveness through simulations and a real-world cancer data analysis.

Conclusions:

• The new nonparametric approach effectively evaluates censored surrogate outcomes for time-to-event data.
• This method offers a flexible and assumption-relaxed alternative to existing techniques.
• Progression-free survival can be a valid surrogate for overall survival in metastatic colorectal cancer.