Breast cancer is a heterogeneous disease, with prognosis and treatment shaped by subtype and stage. In this study we integrated 31 single-cell RNA-seq datasets, totaling 1.2 million cells from 376 samples. Our analysis revealed distinct subtype- and stage-specific TME states, and in both TNBC and Luminal A, we uncovered connections between tumor-intrinsic programs and the microenvironment, directly linked to patient survival and therapeutic response. In TNBC, tumor programs differed by stage, with primaries linked to immune-modulatory states and metastases to poor survival. These programs showed elevated TROP2 expression, suggesting sensitivity to TROP2-targeted therapies. Stage-specific checkpoints linked tumor programs to PD-1/LAG3 in primaries and CTLA-4 in metastases. In Luminal A, ER⁺ tumors showed proliferative signaling linked to favorable endocrine therapy response but also immune exclusion, whereas ER⁻ tumors exhibited MYC-driven programs with poor prognosis, yet higher immune infiltration. Our breast cancer atlas outlines evolving therapeutic vulnerabilities, providing a framework for precision medicine and clinical trials design.
Schächter et al. (2025) studied this question.