Breast cancer is a heterogeneous disease composed of distinct molecular subtypes that influence prognosis and treatment response, with subtype discordance between primary and metastatic tumors contributing to therapeutic failure. Using GeoMx spatial transcriptomics, we profiled non-metastatic primary tumors, metastatic primary tumors, and lymph node (LN) metastases to characterize transcriptional and immune-spatial changes during disease progression. Comparative analysis revealed strong tumor microenvironment reprogramming across clinical states. In stromal regions of metastatic tumors, we observed enrichment of extracellular matrix remodeling, tumor-associated macrophage activity, and epithelial-to-mesenchymal transition-related pathways. In epithelial compartments, non-metastatic and LN tumors showed a secretory-proliferative program, whereas primary metastatic tumors were enriched for antigen presentation and interferon signaling. Immune deconvolution demonstrated distinct immune landscapes, with non-metastatic tumors enriched for innate immune cells, primary metastatic tumors showing increased T- and B-cell infiltration, and LN metastases exhibiting an immunosuppressive microenvironment dominated by M2-like macrophages and memory B cells (p < 0.05). Subtype analysis revealed frequent triple-negative breast cancer (TNBC) subtype switching during LN metastasis, with 70% of cases transitioning from non-basal subtypes to unspecified or immunomodulatory subtypes, both associated with poorer prognosis (HR = 1.8, 95% CI 1.2-2.9). Spatial immune profiling further identified "excluded" and "ignored" tumors lacking T-cell infiltration, while "inflamed" tumors retained active cytotoxic immune responses. Integrating molecular subtype, immune infiltration, and spatial architecture, we developed a novel framework termed Immuno-Spatial Molecular Subtypes (ISMS), which captures TNBC heterogeneity and immune-spatial plasticity across disease progression.
Huang et al. (2026) studied this question.