159P Identification of 4 molecular HR+/HER2- breast cancer subtypes with distinct biological and clinical behaviors using spatial transcriptomics
(2025) Annals of Oncology
Files
KARAKOSE-2025-AnnOfOnc.pdf
Open Access - Adobe PDF
- 204.51 KB
Details
- Authors
- Abstract
- Background HR-positive, HER2-negative (HR+/HER2-) breast cancers (BC) show notable heterogeneity; some patients relapse despite being classified as low-risk by gene expression assays. We used spatial transcriptomics (ST) to explore tumor microenvironment (TME) heterogeneity and identify clinically distinct subtypes. Methods ST (10x Visium) was applied to 86 frozen HR+/HER2- tumors of no special type. H&E slides were morphologically annotated and ST spots were deconvoluted using external scRNA-seq data. Unsupervised clustering identified spatial transcriptome profiles and WGCNA uncovered gene modules. Multikernel learning integrated spatial clusters, morphology, cell-type infiltration, and significant module’s scores to define spatial subtypes. These were projected onto the METABRIC HR+/HER2- cohort (n=1041) for validation using gene signatures. Results We identified four spatial subtypes with distinct microenvironments, biology, and clinical outcomes. The Endocrine-Responsive subtype (n=23) had strong estrogen signaling, low proliferation, low-risk scores (by computing Oncotype and MammaPrint), and best outcomes (95.7% 10-years relapse free survival (10y RFS)). The Stromal-Enriched subtype (n=14) showed ECM remodeling, fibroblast enrichment, and high relapse (64.3% 10y RFS) despite low-risk scores (both Oncotype and MammaPrint). The Proliferative-Inflammatory subtype (n=33) was marked by high proliferation, strong interferon signaling, an inflamed yet immunosuppressive TME, with the worst prognosis (57.6% 10y RFS). The Luminal-Mixed Adaptive subtype (n=16) showed partial HER2 activation, high-risk scores, and high relapse (68.7% 10y RFS). Subtype assignment was independent of tumor size, age, and nodal status (p>0.05). All subtypes were reproduced in METABRIC, confirming biological and clinical relevance. In METABRIC, the Endocrine-Responsive subtype also exhibited better RFS compared with all other subtypes (p=0.0019). Conclusions ST uncovers prognostically meaningful subtypes of HR+/HER2- BC, defined by immune-stromal features often missed by standard assays. These findings could inform personalized treatment strategies pending further validation.
- Affiliations
Institut Jules Bordet Breast Cancer Translational Research Dept CHU de Liège - Sart Tilman Site Department of Pathology The AIRC Institute of Molecular Oncology Metabolic Reprogramming in Solid Tumors Lab, IFOM ETS Institute Jules Bordet Pathological Department Institute Jules Bordet Department of Pathology
Citations
Karakose, B., Serra, M., van Renne, N., Lifrange, F., Venet, D., Vincent, D., Rouas, G., Rediti, M., Craciun, L., Larsimont, D., Vikkula, M., Duhoux, F. P., Rothé, F., & Sotiriou, C. (2025). 159P Identification of 4 molecular HR+/HER2- breast cancer subtypes with distinct biological and clinical behaviors using spatial transcriptomics. Annals of Oncology, 36(s2), S258. https://doi.org/10.1016/j.annonc.2025.08.593 (Original work published 2025)