Validation of real-world event-free survival (rwEFS) in early-stage triple-negative breast cancer.

Abstract

11166 Background: Real-world (RW) evidence has been used to demonstrate effectiveness of early-stage cancer treatments. Early-stage clinical endpoints such as event-free survival (EFS) have the potential to support clinical decision-making as an indicator of real-world effectiveness for novel therapies. However, less is known regarding real-world replicability of rwEFS in early triple negative breast cancer (eTNBC). We therefore applied selected target trial emulation methods to examine the concordance between rwEFS and KEYNOTE-522 (KN-522) trial ‘chemotherapy only’ arm EFS estimates. Methods: Electronic health records from the US Oncology Network were used to identify stage II-III triple negative breast cancer patients who initiated neoadjuvant chemotherapy only from 1/1/20 to 3/31/22. Patients were followed through 07/18/23. Patients who received immunotherapy at any time during this observation period were excluded. KN-522 trial eligibility criteria and endpoint definitions were applied to develop the rwEFS endpoint. Matching-Adjusted Indirect Comparison (MAIC) was used to adjust the real-world cohort based on available baseline demographic and clinical characteristics (age, stage, and ECOG performance status) to approximate the KN-522 population. Kaplan-Meier curves and unadjusted and adjusted hazard ratios with 95% CIs were used to compare EFS between the RW cohort and the control arm of KN-522. Results: Real-world patients (n=199) were older (median age 59 vs 48), in more advanced stages (61% vs 25% stage III), and more often ECOG>0 (40% vs 13%) than participants in the control arm of KN-522. Median EFS was not reached in either the RW cohort or control arm of KN-522. At 36 months, estimated EFS was 76% for patients in the RW cohort and 77% in the KN-522 control arm. There was no statistically significant difference in EFS between the RW and KN-522 cohorts in unadjusted analysis (HR: 0.99, 95% CI: 0.68 - 1.46), indicating concordance between the estimates. After MAIC weighting, baseline values of age, stage, and ECOG performance were balanced between the two cohorts. The difference between RW and trial estimates was also not statistically significant in the adjusted analysis (HR: 0.76, 95% CI: 0.50 - 1.14). Conclusions: EFS is a valuable endpoint for evaluating the effectiveness of neoadjuvant therapy in eTNBC patients. With the application of real-world definitions to align key study design elements with the trial and leveraging curated real-world data, it is possible to reproduce the trial EFS estimate in a real-world cohort. The development of a robust, replicable rwEFS endpoint may support clinical decision-making and guide the choice of effective treatments for eTNBC patients.