Validation of the RSClin risk calculator using the National Cancer Database (NCDB).

Abstract

549 Background: Clinical practice guidelines recommend the use of genomic assays to aid decision making regarding the use of adjuvant chemotherapy (CT) for hormone receptor-positive, HER2-negative (HR+/HER2-) early breast cancer (EBC). Recently, the RSClin clinical tool, which integrates the 21-gene recurrence score (RS) and clinicopathologic features, was developed using data from the TAILORx trial. By integrating clinical and genomic risk, RSClin demonstrated greater precision in guiding adjuvant CT use in HR+/HER2- EBC than the 21-gene RS alone. As outcomes differed by race and ethnicity in TAILORx, further validation in real world datasets of diverse populations is needed. Methods: This study includes patients (pts) from the NCDB who were diagnosed with HR+/HER2- EBC from 2010-2017 and received adjuvant endocrine therapy (ET) with or without CT. RSClin provides a predicted absolute reduction in distant recurrence at 10 years with CT, while the NCDB database only provides overall survival (OS) metrics. While OS underestimates distant recurrence, we correlated RSClin predictions with survival differences between pts receiving ET versus those who received ET plus CT. Inverse probability of treatment weighting was used to correct for differences in age, comorbidity index, insurance, and race/ethnicity. OS benefit of CT was assessed in pts with low (< 2%), intermediate (2-5%), and high (>5%) absolute CT benefit (ACB) per RSClin using the log-rank test. Results: 150,268 pts with EBC were included. Average (avg) age was 59 yrs; 84% were white, 8% black, 4% Hispanic, and 4% Asian/Pacific Islander. Avg tumor size was 1.7 cm, avg RS 17, and 82% had lymph node (LN) negative disease. A significant OS benefit with CT was seen in the intermediate (HR 0.70) and high (HR 0.66) RSClin predicted ACB in the LN negative pts. When analyzed by racial / ethnic subgroup, all pts with a high RSClin ACB had improvement in survival with CT; in pts with an intermediate RSClin ACB, only white pts demonstrated significantly higher OS with CT (HR 0.72). Conclusions: Although developed to predict distant recurrence rates, the RSClin tool also correlates with OS in pts receiving ET with or without CT, which can aid in clinical decision making in pts with HR+/HER2- EBC. Predictive accuracy of RSClin differs by race/ethnicity. Accurate risk stratification in diverse populations is essential in ensuring equitable treatment and mitigating disparities in breast cancer outcomes in different racial/ethnic groups.