A Novel CAF-Related Signature for Precise Prediction of Clinical Outcomes and Immunotherapy Response for Breast Cancer Patients: Based on Multiomics Analyses and Experimental Validation.

OBJECTIVE

Breast cancer (BC) remains a leading cause of cancer-related mortality worldwide. Cancer-associated fibroblasts (CAFs) is a central stromal component of the tumor microenvironment (TME), critically influence BC progression and therapeutic resistance. However, the association between CAF heterogeneity and patient prognosis or response to immunotherapy remains poorly characterized. Here, we aimed to develop a CAF-associated gene signature by integrating single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq data to predict clinical outcomes and immunotherapeutic response in BC.

METHODS

Gene expression profiles and clinical data from BC patients were sourced from TCGA and GEO databases. scRNA-seq data preprocessed (quality control, PCA, UMAP using Seurat) identified CAF-related genes. Prognostic genes were identified via univariate Cox, lasso, and multivariate Cox regression. Single-cell Gene Set Enrichment Analysis (scGSEA) assessed the signature's link to immune infiltration and immunotherapy genes. R tools evaluated signature characteristics and real-world applications. GO enrichment analysis was used to explore signaling pathways. CAF factor expression and CD8+ T-cell correlation in clinical BC samples were validated using qPCR, immunohistochemical (IHC), multiplex immunofluorescence (mIF), and Western blot.

RESULTS

scRNA-seq analysis identified multiple CAF-specific marker genes that formed the core of our signature. Eight genes (ANXA5, APOD, CXCL14, GSN, IGFBP4, PPIB, TCF7L2, and TMEM98) were associated with favorable prognosis (low-risk), whereas three genes (SDC1, EMP1, and FAM114A1) conferred higher risk. A risk score model based on these 11 genes independently predicted overall survival (OS) across diverse BC pathological subtypes, demonstrating robust prognostic accuracy. Immune infiltration analysis revealed significantly reduced immune cell abundance in the high-risk group compared to the low-risk group, suggesting diminished responsiveness to immunotherapy. In tumor tissues relative to adjacent nontumor tissues, mRNA and protein levels of the high-risk genes (SDC1, EMP1, and FAM114A1) were consistently elevated (all p < 0.05). Moreover, both Western blotting and mIF showed significantly higher CAF abundance in high-risk samples (p < 0.01), concomitant with markedly lower CD8+ tumor-infiltrating lymphocyte counts (p < 0.05). GO enrichment analyses indicated that CAFs promote BC evolution and progression through complex signaling networks. Key pathways included extracellular matrix (ECM) remodeling, cell adhesion, tumor associated inflammation, and oncogenic cascades such as KRAS, WNT, IL-6/STAT3, and TNFα/NF-κB highlighting CAFs as pivotal regulators and potential therapeutic targets in BC.

CONCLUSION

We present a novel CAF associated gene signature that robustly predicts prognosis and immunotherapy response in BC. As an independent prognostic indicator strongly correlated with immune infiltration, this model holds promise for guiding personalized therapeutic strategies. Future validation in large, multicenter cohorts and extension to other malignancies are warranted to facilitate clinical translation of CAF targeted biomarkers.