Correlation of Breast Cancer Index (BCI) prognostic and predictive results to Ki67 and tumor grade in early stage HR+ breast cancer.

Abstract

e12075 Background: Markers of tumor proliferative status, e.g., Ki67 and grade (G), are prognostic for early distant recurrence (DR) in HR+ breast cancer and integrated in adjuvant chemotherapy decisions; however, their impact on late DR and extended endocrine therapy (EET) decisions is less clear. BCI is a genomic assay that provides 2 results: BCI Predictive, based on HoxB13/IL17BR (H/I ratio), reports a prediction of likelihood of benefit from EET; BCI Prognostic, based on the algorithmic combination of H/I and proliferation genes, reports risk of late DR. In this study, BCI results were examined within clinicopathologic risk categories based on Ki67 and G. Methods: The BCI Clinical Database for Correlative Studies is an IRB-approved de-identified database that contains > 50 clinicopathologic and molecular variables from cases submitted for BCI testing in clinical practice (N = 14,463). Clinicopathologic variables were abstracted from pathology reports, and were available for a subset of cases. High Ki67 was defined as ≥14%. Chi-squared tests were used to compare BCI results between Ki67 and G subgroups. Results: Analyses included 3395 LN- pts (median age 59.1y; 73% ≥ 50y). BCI Prognostic showed a wide distribution of individual risk assessments in pts with high or low Ki67. A greater proportion of pts with high Ki67 was classified by BCI as high risk of late DR compared to low Ki67 (68.2% vs 26.5%; P < 0.0001); however, substantial proportions of high Ki67 pts were classified as BCI low risk (31.8%) and pts with low Ki67 classified as BCI high risk (26.5%). Overall, there was a moderate correlation between individual BCI Scores and individual Ki67 scores (Correlation = 0.519). 45.5% of pts with high Ki67 were classified as High BCI Predictive (H/I) compared to 35.5% of pts with low Ki67 (p < 0.001). Both BCI Prognostic and BCI Predictive (H/I) classified increasing proportions of pts as high risk or high benefit with increasing G (P < 0.0001). Conclusions: These findings help characterize differential stratification based on tumor biology vs Ki67/G for pts considering EET. While moderately correlated, both BCI Prognostic and BCI Predictive (H/I) identified distinct populations compared to Ki67 and G.