85 rwpCR and rwEFS in Early-Stage Triple-Negative Breast Cancer Patients From Community Oncology Are Concordant With Trial End Point Estimates

Background

Clinical trials often exclude patients with comorbidities or advanced disease, limiting generalizability to real-world populations. Reliable real-world end points are essential for research and clinical practice. This study assessed concordance between real-world end points and clinical trial estimates in early-stage triple-negative breast cancer (eTNBC), using the KEYNOTE-522 control arm as a benchmark.

Methods

This retrospective, observational study used electronic health record data from patients treated in US community oncology clinics. Adults with stage II-III TNBC and no prior immunotherapy use were identified between January 1, 2020, and March 31, 2022, and followed through July 18, 2023. A trial emulation framework was used to define a real-world cohort that aligned with the trial control arm. The index was the date of neoadjuvant treatment initiation. Real-world end points definitions were aligned to the trial for event-free survival (EFS) and pathological complete responses (pCR). Adjustment for available baseline characteristics (age, race, ethnicity, clinical stage, and ECOG) was done through Matching Adjusted Indirect Comparison (MAIC). Kaplan-Meier plots and hazard ratios (HR) from Cox modeling were used to assess concordance between real-world EFS and trial estimates. A sensitivity analysis was conducted in which patients were censored at the initiation of subsequent immunotherapy. Risk ratios (RR) from generalized linear models were used to assess concordance between real-world pCR and trial estimates.

Results

A total of 311 real-world patients were identified as study-eligible. Median follow-up time was 24.6 months. Compared with the KEYNOTE-522 control group, real-world patients were older (median age 56 vs 48), had higher proportions of stage III disease (44% vs 25%) and ECOG 1 performance status of 1 or more (41% vs 13%). After MAIC adjustment, baseline characteristics were closely aligned between the 2 groups. The median EFS was not reached in the real-world cohort or the trial control arm. Strong concordance was observed for EFS (HR, 1.00; 95% CI, 0.72-1.39; P = .98; Figure). Real-world patients were 4% less likely to achieve pCR compared with the trial control arm (RR, 0.96; 95% CI, 0.80-1.14; P = .63) and this difference was not statistically significant. These results indicate the replicability of the trial end points with real-world data. The sensitivity analysis confirmed the robustness of survival estimates.

Conclusion

This real-world study demonstrates that early clinical real-world end points such as EFS and pCR are concordant with clinical trial estimates when applying a trial emulation framework. These findings support the reliability of real-world data in oncology research and their utility for clinical decision-making.