Novel framework for the validation of intermediate clinical endpoints in oncology real-world data

This is the authors' abstract. We don't add key points for this paper.

INTRODUCTION: The role of real-world data in oncology has been substantially increasing, helping to answer clinically unmet needs. No valid framework for real-world data is currently available for the evaluation of surrogate endpoints, with no intermediate clinical endpoint (ICE) being validated as a surrogate for overall survival (OS) across most cancer types. METHODS: A real-world retrospective cohort study was conducted using the nationwide US Flatiron Health database. Four ICEs were evaluated real-world progression-free survival (rwPFS), and three exposure-derived endpoints: Time to Next line of Therapy (TTNT), Time to Last Treatment Administration (TTLA) of first line therapy, and Time to End of First-Line Therapy (TTEFL). Two conditions were evaluated to establish surrogacy: correlation between ICE and OS (condition 1), and correlation of group effects (condition 2). RESULTS: A total of 19,164 patients with metastatic breast cancer (MBC) were included. Median OS was 36.3 (95%CI, 35.6-37.2) months across the study population. For condition 1, a high correlation was found between each of rwPFS, TTNT and TTLA, and OS (Kendall's tau [τ] ≥ 0.5; R2 ≥ 0.7 between 36-month OS and 12-month rates for each ICE across subgroups of interest). For condition 2, group effects were found to be strongly correlated only between rwPFS and OS (R2 ≥ 0.7 across each subgroup comparison). CONCLUSION: Among patients with MBC, rwPFS is a strong surrogate for OS. This study establishes a novel framework applicable to real-world data for the validation of intermediate clinical endpoints. This study can serve as a blueprint for future efforts of real-world data analysis.