Associations of predictive biomarkers, MHC-I and MHC-II, with clinical and molecular features in a diverse breast cancer cohort.

517 Background: The addition of immunotherapy (IO) to chemotherapy in triple negative breast cancer (TNBC) has improved survival outcomes and recent studies have also shown improvements in pCR rates in early high-risk HR+/HER2- breast cancer. However, these improvements are not without toxicities and thus there is a critical need for biomarkers to select patients that are most likely to benefit from immunotherapy. MHC-I and tumor-specific MHC-II (tsMHC-II) expression are candidate biomarkers for predicting PD-(L)1 checkpoint inhibition benefit and have previously shown potential for predicting treatment response in multiple tumor types, including breast cancer. However, the existing data were generated from clinical trials with limited racial/ethnic diversity. Here we report associations between MHC-I and MHC-II expression and clinicopathological features in a large, racially and ethnically diverse breast cancer cohort. Methods: We performed multiplexed immunofluorescent tissue-based analysis of participants in the Carolina Breast Cancer Study (CBCS). Intrinsic subtype and P53 mutant-like status were identified using RNA-based multigene assays. To evaluate MHC-I we ranked all participants based on tumoral MHC-I intensity, with the top 33% categorized as "MHC-I High "; MHC-II+ expression on tumor cells was defined as ≥5% of tumor cells. Association between MHC-I/II and clinicopathological features by race was tested using Chi-square test or fisher-exact test and relative frequency difference (RFD) was calculated. Results: This study included 1630 participants (48% Black and 52% non-Black) with Stage I-III breast cancer. Black participants had a higher incidence of TNBC (25% vs. 12.5%, p=<0.001) and Basal-like (29% vs. 14%, p=<0.001) tumors. Basal-like and TP53 mutant tumors were more frequently observed among Black participants with HR+/HER2- tumors (11% vs. 5.5%, p=0.006 for Basal-like; 26% vs. 17%, p=0.003 for p53 mutant-like). While there was a similar distribution of tsMHC-II+ tumors between Black and non-Black participants, the relative frequency of tsMHC-II+ tumors was significantly more common in Black participants with HR+/HER2- breast cancer compared to non-Black participants (7.9% vs. 5%, p=0.04). Moreover, Black participants also were significantly more likely to have MHC-I high tumors (38.9% vs. 28.4%, p=0.01), which, after IHC subtype stratification, was only significant among participants with HR+/HER2- breast cancer (28.2% vs. 22.3%, p=0.02). Conclusions: In this diverse breast cancer cohort, MHC-I and MHC-II expression, candidate predictive biomarkers of IO benefit, were more prevalent in tumors from Black women with HR+/HER2- breast cancer. These results underscore the importance of diverse and equitable clinical trial enrollment in future IO trials, especially in HR+/HER2- breast cancer.