Abstract 5389: Uncovering spatial relationships of the tumor microenvironment in the Carolina Breast Cancer Study

Breast cancer is the second leading cause of cancer-related deaths in women in the United States. Although breast cancer mortality rates have dropped over the last few decades, the mortality rate for Black women is still 40% higher than White women. Even among the breast cancer subtype with the best prognosis (ER+/HER2-), Black women still have a higher risk of recurrence (ROR) by PAM50. Some work has been done to uncover differences in the biology of breast cancer between White and Black women, but there has been little study of how immune responses differ by race. To identify spatial relationships between tumor and specific immune subpopulations in the microenvironment of breast cancer, we used the GeoMx (NanoString) platform to perform digital spatial profiling (DSP) of tumor whole sections and tumor microarrays (TMA) from the Carolina Breast Cancer Study (CBCS), a large population-based study that oversampled for Black and younger women. DSP identifies relationships among immune subpopulations, analyzing approximately 40 immune markers in areas selected for CD45, CD68, or pan-cytokeratin content. Thus regions of interest (ROIs) can be selected to evaluate immune ‘hot spots' and the tumor microenvironment, separately. In whole tumor slides, we found that immune hot spots (high CD45 infiltrated ROIs) had elevated expression of many immune markers, suggesting that there was a widespread, robust immune response that included Macrophages, Dendritic cells, and B-cells in addition to T-cells. However, we also observed that immune markers were detectable in tumor-enriched areas, and that tumor-enriched areas had distinct immune signals depending upon subtype, with Regulatory T cell (Treg) markers being higher in Basal-like breast cancers and Dendritic cell markers being higher in Luminal A breast cancers. Extending these findings to tumor-enriched ROIs on TMAs, we evaluated 76 patients and once again found subtype-specific immune responses, including enrichment of Treg markers in Basal-like breast cancers. Receiver operating characteristic (ROC) analysis of Treg marker expression as a predictor for Basal-like versus Luminal A breast cancer had an area under the curve (AUC) score of 0.754, pointing to Treg markers as an effective predictor of breast cancer subtype. With DSP analysis, immune marker expression can elucidate spatial differences in the tumor microenvironment and predict breast cancer subtype and other clinicopathologic and prognostic features. Our