Metrics to find a surrogate endpoint of OS in metastatic oncology trials: a simulation study

Surrogate endpoint (SE) for overall survival (OS) in cancer patients is essential to improving the efficiency of oncology drug development. In practice, we may discover a new patient level association with OS in a discovery cohort, and then measure the trial level association across studies in a meta-analysis to validate the SE. In this work, we simulated pairs of metrics to quantify the surrogacy at the patient level and the trial level and evaluated their association, and to understand how well various patient level metrics from the initial discovery would indicate the eventual utility as a SE. Across all the simulation scenarios, we found tight correlation among all the patient level metrics, including C index, integrated brier score and log hazard ratio between SE values and OS; and similar correlation between any of them and the trial level association metric. Despite the continual increase in the true biological link between SE and OS, both patient and trial level metrics often plateaued coincidentally in many scenarios; their association always decreased quickly. Under the SE development framework and data generation models considered here, all patient level metrics are similar in ranking a candidate SE according to its eventual trial level association; incorporating additional biological factors into a SE are likely to have diminished return in improving both patient level and trial level association.