Reactive Stroma as a Transversal Prognostic Biomarker for Metastasis in Breast Cancer: Integration of Digital Histopathology and Transcriptomic Profiling

Distant metastasis is the main cause of breast cancer (BC) mortality, yet current prognostic models remain largely tumor-centric and underutilize stromal biology. In this study, we quantified reactive stroma, a collagen-rich and fibrotic fraction of the stromal compartment, as a subtype-independent biomarker of metastatic risk. A retrospective cohort of 182 FFPE primary BC biopsies (2006-2020) was analyzed. Total stroma was quantified on H&E-stained sections and reactive stroma on Masson's trichrome using QuPath with pathologist validation. Cutoffs were defined using maximally selected rank statistics, and overall survival (OS) and metastasis-free survival (MFS) were evaluated by Kaplan-Meier analysis and multivariable Cox regression. RNA sequencing was performed in a subset of cases to characterize associated transcriptomic programs. While total stromal content showed univariate associations with OS and MFS, it was not independently prognostic after adjustment. In contrast, high reactive stroma (cutoff 53.2%) independently predicted shorter MFS (HR = 3.76; <i>p</i> < 0.001), irrespective of molecular subtype and clinicopathological variables. Tumors with high reactive stroma exhibited upregulation of extracellular matrix and profibrotic genes (including FN1, OLR1, and EDN2), enrichment of collagen remodeling and TGF-β signaling pathways, and reduced T-cell activation signatures. These findings demonstrate that quantitative assessment of reactive stroma from standard histological stains is a reproducible, subtype-independent biomarker of metastatic risk in BC and can be readily integrated into routine pathology workflows to improve risk stratification.