Resilience Measures for the Surrogate Paradox

2025-06-13

AI Summary

In clinical trials, surrogate endpoints may induce the “surrogate paradox”—where an intervention improves the surrogate marker yet harms the true primary outcome. To address this, we propose the first robustness assessment framework that does not rely on the cross-study transportability assumption of mean outcome functions. We formally define *resilience probability*, *resilience bound*, and *resilience set* to quantify an intervention’s resistance to adverse effects on unobserved primary outcomes. Our method integrates functional-class perturbation modeling, potential outcomes inference, distributional estimation, and sensitivity analysis. Evaluated via simulations and two HIV clinical trials, it demonstrably enhances the reliability and safety of surrogate endpoint use. The framework yields interpretable, bias-controlled robust criteria for surrogate endpoint validation and decision-making. (136 words)

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness
• Reasoning under Uncertainty: Probabilistic Programming

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements
• Responsible Web: Measurement, analysis, and circumvention of Web censorship

Abstract

Surrogate markers are often used in clinical trials to evaluate treatment effects when primary outcomes are costly, invasive, or take a long time to observe. However, reliance on surrogates can lead to the surrogate paradox, where a treatment appears beneficial based on the surrogate but is actually harmful with respect to the primary outcome. In this paper, we propose formal measures to assess resilience against the surrogate paradox. Our setting assumes an existing study in which the surrogate marker and primary outcome have been measured (Study A) and a new study (Study B) in which only the surrogate is measured. Rather than assuming transportability of the conditional mean functions across studies, we consider a class of functions for Study B that deviate from those in Study A. Using these, we estimate the distribution of potential treatment effects on the unmeasured primary outcome and define resilience measures including a resilience probability, resilience bound, and resilience set. Our approach complements traditional surrogate validation methods by quantifying the plausibility of the surrogate paradox under controlled deviations from what is known from Study A. We investigate the performance of our proposed measures via a simulation study and application to two distinct HIV clinical trials.

Problem

• Assessing resilience against surrogate paradox in clinical trials
• Estimating treatment effects on unmeasured primary outcomes
• Quantifying plausibility of surrogate paradox under deviations

Innovation

• Formal resilience measures against surrogate paradox
• Estimates potential treatment effects distribution
• Quantifies plausibility under controlled deviations