Real-World Event-Free Survival and Real-World Pathological Complete Response in Early-Stage Triple-Negative Breast Cancer: Concordance with Estimates from the control arm of the KEYNOTE-522 Trial

Background : Clinical outcomes such as event-free survival (EFS) and pathological complete response (pCR) have the potential to support medical product development for early-stage breast cancer. Replicability of trial estimates for these endpoints in the real world is needed. Objective : To assess the concordance of real-world (RW) EFS and pCR for early-stage triple negative breast cancer (eTNBC) with estimates from the KEYNOTE-522 clinical trial control arm. Methods : This retrospective observational cohort study used electronic health record data from US community oncology practices within The US Oncology Network. The RW cohort was developed using eligibility criteria aligned with KEYNOTE-522, with eligibility determined at the date of neoadjuvant treatment initiation. Patients newly diagnosed with eTNBC between 1/1/2020 and 3/30/2022 were followed through 7/18/2023. To balance baseline characteristics between the RW and clinical trial populations, a matching-adjusted indirect comparison (MAIC) was implemented using inverse probability of treatment weighting (IPTW). Relative risks (RRs) for rwpCR and hazard ratios (HRs) for rwEFS were calculated, before and after MAIC-weighting. Results : A total of 311 patients in the RW cohort met eligibility criteria aligned with the KEYNOTE-522 clinical trial. The adjusted HR for rwEFS and RR for rwpCR were 1.00 (95% CI: 0.72, 1.38; p=0.980) and 0.96 (95% CI 0.80, 1.14; p=0.627), respectively. A sensitivity analysis for rwEFS that censored RW patients at the time of immunotherapy treatment gave consistent results. Conclusion : EFS and pCR were concordant between the RW cohort and the KEYNOTE-522 control arm. RW results can reflect clinical trial outcomes in a matched real-world population. These findings increase confidence in the replicability of early-stage trial estimates in RW settings.