Abstract PS2-08-13: Spatially resolved tumor subclones in HER2-positive breast cancer

Abstract Background: HER2-positive breast cancer displays significant tumor and tumor microenvironment (TME) heterogeneity, which directly impacts treatment response and survival outcomes. Treatment resistance can arise due to the presence of somatic tumor subclones and is further driven by interactions with TME. Spatial transcriptomics (ST) enables mapping of these subclones in the tissue context. Methods: We performed spatial transcriptomics (10X Genomics Visium) on frozen HER2-positive breast cancer surgical samples from untreated primary tumors (n = 25). We then utilized H 0.05, log2(fold change) 1) and subjected them to gene set over-representation analysis. Subsequently, we queried subclone markers against LINCS L1000 drug perturbation profiles (Enrichr) to nominate compounds predicted to modulate subclone-specific transcriptional programs. Finally, we reconstructed evolutionary relationships among subclones with MEDICC2. Results: We identified a total of 71 spatial tumor subclones in our cohort. Our results demonstrate substantial multi-clonality of HER2-positive breast cancer (median of 3 subclones per sample), with only 5 samples comprising a single subclone. Over-representation analysis results suggest that multi-clonal samples contain subclones with diverse gene expression profiles, exhibiting varying associations (FDR 0.05) to pathways related to immune-rich (e.g., interferon alpha/gamma response, IL2 STAT5 signaling), stromal (e.g., epithelial mesenchymal transition), or luminal phenotypes (e.g., estrogen response early/late). We further demonstrate that subclones with similar phenotypes within the same sample often share a recent common ancestor, indicating that they diverged after acquiring shared copy number alterations that likely enable specific, conserved transcriptional programs. Conversely, subclones with different phenotypes emerge on distinct evolutionary branches, likely due to the absence of these key copy number changes. Additionally, we observe statistically significant differences (Kruskal-Wallis, p-value 0.05) in ERBB2 expression between subclones within the sample in the majority of our cohort (18 out of 25 samples). We hypothesized that the presence of low-ERBB2 subclones may hinder the response to targeted anti-HER2 treatment. Additionally, enrichment tests on drug perturbations gene sets reveal that subclones within the same sample show varying treatment responses to the commonly used chemotherapy regimen. We present a case of a patient-derived sample that recurred after adjuvant carboplatin-docetaxel therapy. Enrichment analysis of spatial subclones revealed striking heterogeneity in carboplatin sensitivity, with only one of three subclones showing a significant response (FDR = 0.006). This intra-sample variation in treatment response underscores the importance of profiling spatial subclones. Conclusions: Our findings underscore the intra-tumor somatic heterogeneity of HER2-positive breast cancer depicted by ST and its potential clinical impact. Further analyses focused on the interactions of existing subclones with the TME could refine our understanding of the mechanisms driving treatment resistance in this disease. Citation Format: L. Cervenkova, M. Rediti, M. Serra, F. Lifrange, C. Tommasi, N. Occelli, X. Wang, D. Vincent, G. Rouas, L. Buisseret, D. Larsimont, F. Libert, A. Lefort, N. Van Renne, D. Venet, F. Rothé, C. Sotirou. Spatially resolved tumor subclones in HER2-positive 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 PS2-08-13.