A Senescence-Related Signature for Risk Stratification and Predicted Immunotherapy Response in Breast Cancer:...

Cellular senescence exerts context-dependent effects on breast cancer progression and therapeutic response through dynamic interactions with the tumor microenvironment; however, its pathological manifestations, cellular organization, and clinical significance remain incompletely understood. In this study, we integrated whole-slide pathology images, bulk transcriptomic profiles, single-cell ribonucleic acid (RNA) sequencing, spatial transcriptomics, and clinical data to systematically characterize senescence-associated features in breast cancer. Quantitative pathology and deep-learning features were extracted from hematoxylin and eosin-stained images, and a senescence-related signature (SRS) was constructed using machine-learning survival models. The SRS effectively stratified patients into distinct prognostic groups and showed consistent performance across the internally divided TCGA-BRCA subsets, with high-SRS patients exhibiting significantly poorer survival. Single-cell analyses revealed that senescence activity was predominantly enriched in epithelial and endothelial cells, whereas fibroblasts acted as major signal-sending populations and mediated enhanced collagen-related communication with tumor cells. Low-SRS tumors displayed greater immune-cell infiltration, higher immune-checkpoint expression, higher immunophenoscores, and more favorable Tumor Immune Dysfunction and Exclusion (TIDE)/immunophenoscore (IPS)-based predicted response profiles, while also exhibiting distinct drug-sensitivity profiles. Integrative pathological, transcriptomic, and spatial analyses further identified STARD10 as a senescence-associated candidate enriched in tumor regions and spatially concordant with senescence-marker expression at the spot level. Functional experiments showed that STARD10 knockdown reduced cellular metabolic activity in CCK-8 assays and suppressed migration and invasion, attenuated senescence-associated β-galactosidase staining and senescence-associated secretory phenotype (SASP)-related cytokine expression, and suppressed tumor growth in vivo. Collectively, these findings establish an integrated pathology- and multi-omics-derived senescence-related signature for prognostic stratification and characterization of TIDE/IPS-based predicted immunotherapy-response profiles in breast cancer, while identifying STARD10 as a candidate biomarker and a potential target for further mechanistic and therapeutic investigation.