Characterization of breast tumors and the tumor immune landscape in young, diverse participants of the Northern California Breast Cancer Family Registry

• Published in: Journal of clinical oncology Vol. 43; no. 16_suppl; p. e12551
• Main Authors: Mangiante, Lise, Simon, Brennan G., Weiss, Clemens, Ma, Zhicheng, Koo, Jocelyn, Ransohoff, Julia Dory, John, Esther M., Curtis, Christina, Caswell-Jin, Jennifer Lee, Kurian, Allison W.
• Format: Journal Article
• Language: English
• Published: American Society of Clinical Oncology 01.06.2025
• ISSN: 0732-183X; 1527-7755
• DOI: 10.1200/JCO.2025.43.16_suppl.e12551

e12551Background: Racial and ethnic disparities in breast cancer mortality persist across breast cancer subtypes, and variations in immune response may contribute. Single-cell spatial profiling enables detailed insights into the tumor microenvironment. Methods: We leveraged a cohort of young, racially and ethnically diverse women enrolled from 1995-2005 in the population-based Northern California Breast Cancer Family Registry (NC-BCFR). We determined feasibility of imaging-based spatial profiling on NC-BCFR TMAs with CosMx 6K SMI single-cell spatial transcriptomics assay, evaluating 140 samples from 88 women. Cells were stained with nuclear and membrane markers, imaged, and segmented with internal algorithms. Cell-level transcriptomic data of 6519 genes were normalized across the TMA and integrated with SCTransform. Manual cell typing was performed using publicly available annotations. Cell type composition for each sample for immune, fibroblast, cancer epithelium, and normal epithelium, normalized to 100 tumor cells/sample, was validated using predictions from matched H&E stained slides annotated with the HoverNet deep learning algorithm. Spatial transcriptomics data were used to predict HER2 and hormone status (HR). The immune microenvironment was examined across race, ethnicity and breast cancer subtype. Results: Race and ethnicity of 88 NC-BCFR participants were as follows: 16 Black, 22 Asian, 34 Hispanic, and 16 non-Hispanic (NH) White women. Profiles revealed 917,804 cells with mean transcript per cell of 838 (range: 517-1141). Following QC, normalization, and integration, 867,237 cells were typed across diverse phenotypes. With cell type normalized to 100 tumor cells/sample, there was a trend toward NH White samples having increased immune infiltrate vs. other racial groups, within both SMI (NH White 158, Black 116, Hispanic 89, Asian 55.3; ANOVA p = 0.263) and HoverNet (NH White 40.2, Black 24.9, Asian 29.1, Hispanic 18.9; ANOVA p = 0.117). Samples from NH White women had higher B-cell infiltrates (ANOVA p = 0.165) and macrophage content (ANOVA p = 0.000753) compared to cores from Asian, Black, and Hispanic women. Similar trends were observed in NK-cells (ANOVA p = 0.643) and T-cell infiltrates (ANOVA p = 0.318), and tertiary lymphoid structures were identified in one sample. Conclusions: Single-cell spatial transcriptomics can characterize the tumor immune microenvironment in TMAs from the 1990s, thus leveraging a highly diverse, population-based cohort study with long-term follow-up for novel, clinically relevant discoveries.