Replication of Oncology Randomized Trial Results using Swedish Registry Real World-Data: A Feasibility Study.

Abstract

Nationwide healthcare registries are potential important real-world data (RWD) sources for assessing drug effectiveness in oncology. However, it is unclear whether registry-derived RWD are suitable for clinical development. In this study, we replicate results from the comparator arm of two previously published oncology randomized controlled trials (RCTs) using RWD from Swedish nationwide healthcare registries. For replication 1, the RCT included 553 patients and the RWD included 283 patients treated with sorafenib for advanced hepatocellular cancer. The median overall survival (OS) was 11.2 (95% confidence interval (CI): 10.1-13.2) months in the RCT and 8.2 (95% CI: 7.0-9.9) months in the RWD, unadjusted hazard ratio (HR) 0.75 (95% CI: 0.63-0.88). For time-to-treatment discontinuation (TTD), the HR was 1.00 (95% CI: 0.87-1.16). For replication 2, the RCT included 154 patients and the RWD included 704 patients treated with melphalan, prednisone, and thalidomide for untreated multiple myeloma. The median OS was 52.6 (95% CI: 40-not applicable) months in the RCT and 36.9 (95% CI: 33.8-40.5) months in the RWD, unadjusted HR 0.67 (95% CI: 0.51-0.87). For TTD, the HR was 0.89 (95% CI: 0.74-1.06). The results were similar when applying various propensity-based confounding adjustments. In conclusion, OS was shorter in the RWD, whereas TTD was similar. Importantly, the data necessary (ex: eligibility criteria and baseline confounders) for replicating RCTs was mostly not available and these results further underscore the importance of developing frameworks for capturing relevant patient-level RWD for clinical and regulatory decision making in oncology.