Abstract Introduction. Despite advances in breast cancer (BC) diagnosis and treatment, there is still a lack of robust early prognostic tools to accurately identify patients at high risk of metastatic progression. This gap limits timely access to personalized therapies, leading to overtreatment in low-risk cases and delayed intervention in those who develop metastases. Current prognostic models rely mainly on tumor-intrinsic features, such as size, histological grade, or hormone receptor status, without integrating key components of the tumor microenvironment (TME). Among these, cancer-associated fibroblasts (CAF) are non-neoplastic cells that, in response to signals like TGF-β, PDGF, and IL-6, acquire a myofibroblast phenotype with high extracellular matrix (ECM) remodeling capacity, promoting stromal desmoplasia and facilitating tumor invasion. CAF also drive EMT, angiogenesis, and local immunosuppression, contributing to tumor aggressiveness and therapy resistance. Given their role in tumor progression, CAF are a promising source of prognostic biomarkers. We hypothesized that CAF from primary tumors harbor a transcriptomic signature with independent prognostic value for distant metastasis. Methods. A total of 182 FFPE breast tumor biopsies from women with invasive carcinoma were analyzed. Patients had ≥5 years of clinical follow-up, represented all five molecular subtypes, and had not received neoadjuvant chemotherapy, radiotherapy, or systemic treatment. Reactive stroma was quantified using H 95% CI: 1.98-7.09; p0.01). The 10-gene CAF signature identified in metastatic tumors was involved in biological processes such as ECM organization, cell adhesion, and migration. This signature demonstrated strong predictive performance in the training cohort (AUC=0.938), and in independent datasets (AUC=0.956; specificity 100%), outperforming models based solely on conventional clinicopathological parameters. Patients expressing the signature showed a 19.95-fold increased risk of metastasis (p0.001), independently of hormone receptor status, histological grade, or tumor size. Notably, the signature retained prognostic value across molecular subtypes and sequencing platforms, supporting its clinical robustness. Conclusion. This study highlights the prognostic relevance of the stromal compartment, particularly CAF, in BC metastasis. The validated 10-gene signature provides a novel, TME-based biomarker that enables early risk stratification at the time of diagnosis. By incorporating this signature into clinical workflows, oncologists may identify high-risk patients earlier and guide personalized therapeutic decisions, ultimately improving outcomes and reducing mortality in BC Acknowledgments. CORFO 22CVID-206707; ANID VIU24P0138. Citation Format: D. P. Barrera, J. Contreras-Riquelme, V. Toledo, J. Sapunar, B. Chahuán, L. Moyano, B. Prieto, J. Cerda-Infante. Beyond the tumor: a CAF-Driven Gene Signature for Early Clinical Stratification of Metastasis Risk in Breast Cancer abstract. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS2-09-11.
Barrera et al. (Tue,) studied this question.