Real-World Progression-Free Survival as an Endpoint in Advanced Non-Small-Cell Lung Cancer: Replicating Atezolizumab and Docetaxel Arms of the OAK Trial Using Data Derived From Electronic Health Records

BackgroundEvaluating cancer treatments in real-world data (RWD) requires informative endpoints. Due to non-standardized data collection in RWD, it is unclear if and when common oncology endpoints are approximately equivalent to their clinical trial analogues. This study used RWD to replicate both the atezolizumab and docetaxel arms of the OAK trial. Outcomes using progression-free survival (PFS) derived from abstracted physicians notes in RWD (rwPFS) were then compared against PFS outcomes derived according to Response Evaluation Criteria in Solid Tumors (RECIST) from the clinical trial (ctPFS). MethodsAtezolizumab and docetaxel arms of the phase III OAK RCT (NCT02008227) were replicated in a US nationwide real-world database by applying selected OAK inclusion/exclusion criteria, followed by adjustment for baseline prognostic variables using propensity score-based methods. Multiple rwPFS definitions were characterized and a definition was chosen that was acceptable from both clinical and data analysis perspectives. Concordance of outcomes was assessed using Kaplan-Meier (KM) medians and hazard ratios (HRs). ResultsOverall, 133 patients receiving atezolizumab and 479 patients receiving docetaxel were selected for the RWD cohort. After adjustment, prognostic variables were balanced between RCT arms and corresponding RWD cohorts. Comparing rwPFS against ctPFS outcomes in terms of KM median and HR showed better concordance for docetaxel (2.99 vs 3.52 months; HR, 0.99, 95% CI, 0.85-1.15) than for atezolizumab (3.71 vs 2.76 months; HR, 0.8, 95% CI 0.61-1.02). The latter improved when events labelled "pseudo-progression" were excluded from the RWD (im-rwPFS) and immune-modified RECIST PFS (im-ctPFS) was used in the RCT Atezolizumab data (4.24 vs 4.14 months; HR, 0.95, 95% CI, 0.70-1.25). These findings were robust across several sensitivity analyses. ConclusionsWhile rwPFS and ctPFS were similar under docetaxel treatment, this was only the case for atezolizumab when immune-modified progression criteria were used, suggesting that similarity of RWD endpoints to their clinical trial analogues depends on drug category and possibly other factors. Replication of RCTs using RWD and comparison of outcomes can be used as a tool for characterizing RWD endpoints. Additional studies are needed to verify these findings and to better understand the conditions for approximate numerical equivalence of rwPFS and ctPFS endpoints.