Quantitative coupling of clonal CNV evolution and spatially restricted malignant states identifies MFGE8 as a candidate late-state-associated target in breast cancer

BACKGROUND: Breast cancer progression reflects heterogeneous malignant states shaped by clonal evolution and microenvironmental selection, yet how genomic instability quantitatively couples to transcriptional state transitions and spatial niche organization remains unclear. METHODS: We integrated single-cell RNA sequencing and spatial transcriptomics across ER-positive, HER2-positive, and triple-negative tumors, combined with CNV inference, branched trajectory analysis, transcription factor regulon modeling, cell-cell communication inference, and spatial niche mapping. Trajectory-derived evolutionary gene programs were translated to bulk-tumor stratification and multicohort prognostic modeling. MFGE8 was further evaluated as a therapeutic target using structure-based virtual screening with docking and molecular dynamics refinement. Functional validation of MFGE8 was performed using knockdown, overexpression, rescue, and in vitro phenotypic assays. RESULTS: Malignant epithelial cells formed a branched three-state trajectory that was strongly coupled to CNV-defined clonal architecture, with the tightest coupling in triple-negative tumors. Regulon analyses revealed coordinated transcription factor rewiring along progression, converging on a late State 3 program enriched for extracellular matrix and adhesion signaling. Spatial transcriptomics localized State 3 to a discrete invasive-front niche characterized by stromal-dominant ligand-receptor interactions. MFGE8 emerged as a State 3-biased output associated with LHX2 activity, and in silico MFGE8 perturbation induced transcriptional rollback from the late-state branch. Structure-guided screening prioritized candidate MFGE8 binders with stable interaction dynamics. Evolutionary gene programs enabled robust bulk stratification and supported a multicohort-validated 10-gene prognostic model associated with genomic instability patterns and immunotherapy-related signatures. CONCLUSIONS: CNV-linked clonal evolution and spatial niche restriction are associated with a late malignant program and prioritize MFGE8 as a candidate late-state-associated therapeutic target in breast cancer.