Single cell spatial profiling for identification of organ-specific immune features in the peritumoral microenvironment of triple-negative breast cancer with liver metastases.

1126 Background: Breast cancer liver metastasis (BCLM) is associated with poor prognosis. Prior studies have described liver metastases as immune-cold, yet the mechanisms underlying this phenotype remain incompletely understood. We hypothesized that the peritumoral microenvironment, in addition to the tumor itself, contributes to the distinct immunobiology of BCLM. Methods: We performed imaging mass cytometry (IMC) with a 42-antibody panel on TNBC tissues from primary breasts (PB) (n = 22) and liver metastases (LM) (n = 26). For each case, one region of interest (ROI) from tumor bed and one from peritumoral tissue were profiled. Cell segmentation was performed using DeepCell. Single-cell phenotyping and clustering were conducted using Scanpy with lineage markers. Functional marker expression was analyzed and differences between PB and LM were assessed using non-parametric tests with Benjamini–Hochberg false discovery rate (FDR) correction. Spatial analysis was performed using Squidpy with Wilcoxon testing and FDR correction across all tested interaction pairs. Results: IMC analysis resolved 407,160 cells into distinct immune, stromal, and tumor cell clusters across compartments. Within tumor beds, PB and LM exhibited largely similar functional marker expression and spatial organization, indicating minimal tumor-intrinsic divergence. In contrast, peritumoral tissues diverged substantially between PB and LM. Peritumoral tissues in PB demonstrated M2 macrophage dominance while LM exhibited M1 macrophage dominance. The comparative analysis of peritumoral tissues revealed that cancer cells in LM demonstrated significantly increased spatial proximity to CD8 T cells compared to PB with concurrent upregulation of PD-L1 on macrophages and NK cells, alongside elevated PD-1 expression on multiple immune cell types, indicating that CD8 T cells are potentially exhausted despite their physical presence. Additionally, CCR7, CD6, and CXCL13 were significantly elevated across all cell types in LM, while PB demonstrated higher expression of CD44, CXCR3 and CXCL12, indicating broad immunophenotypic differences in peritumoral tissues. Spatial analysis further revealed that CD4 T cells maintained closer proximity to cancer cells, M1 macrophage, and stromal populations in PB, indicating distinct spatial organization between sites. Conclusions: Single-cell spatial profiling reveals that organ-specific tumor and immune biology in TNBC BCLM is encoded primarily within the peritumoral tissue rather than the tumor bed itself. Our findings reveal a paradoxical phenotype in BCLM where CD8 T cells successfully infiltrate and localize near cancer cells but show features consistent with functional impairment. These findings support further investigation of immunotherapy in TNBC BCLM and biomarker validation.