Abstract 4735: In silico reconstruction of primary and metastatic estrogen receptor positive breast tumor architecture using geographic information science-augmented spatial transcriptomics

Abstract The tumor microenvironment consists of different cell populations, whose interactions contribute to tumor heterogeneity and plasticity and pose challenges to therapy efficacy and prognosis prediction. Breast cancer is one of the highly heterogeneous diseases, and estrogen receptor (ER) positive subtype, the most common one, shows worse prognosis once metastasized. Spatial transcriptomics (ST) is emerging as a powerful tool that offers valuable insights into spatial complexity of tumor microenvironment as well as the cell interactions underpinning tumor heterogeneity. In this study, we aimed to examine tumor heterogeneity of ER positive metastatic breast cancer by applying our geographic information science (GIS)-augmented ST analysis pipeline. Visium datasets were prepared from four metastatic liver tumors from xenograft mouse model and twelve primary and metastatic bone tumors from human patients, including three matching pairs. We integrated geospatial tools, such as multivariate spatial association analysis, with spatial omics data structures to check degree of auto-correlations and visualize localizations of single gene or gene sets. We further compensated for multi-cellular resolution of our spatial data via cell type mapping and deconvolution of cell type proportion (STdeconvolve). We identified hotspot regions displaying extensive co-localization between estrogen response gene set and lipid metabolism gene sets in the fulvestrant-treated metastatic liver tumors from xenograft mouse model. These hotspots were not concentrated in the same area within the tumor and differed by gene sets, suggesting intra-tumor metabolic heterogeneity of the endocrine responsive cells. Our primary samples from human patients showed similar pattern of co-occurrence between estrogen response and metabolic programs as well as cell-cycle related gene sets. On the other hand, mutual exclusivity was observed for estrogen response with immune-related gene sets and metastasis-related ones (epithelial mesenchymal transition, angiogenesis). Compared to primary tumors, metastatic bone tumors showed inter-tumor heterogeneity as such consistency in spatial pattern was obscured. Deconvoluting cell type proportions mapped a higher proportion of mature luminal cells in endocrine responsive hotspots, whereas cancer associated fibroblasts and pericytes were prevalent in the hotspots of metastasis-related gene sets. Our study fully utilized the potential power of ST data by delineating the tumor microenvironment in relation to intra-tumoral heterogeneity as well as cell type proportions. Findings from our study support the development of strategies to target different endocrine resistant cell populations with a combination of metabolic and metastasis-related signaling inhibitors. Citation Format: Jin Young Yoo, Sabrina Akter, Qianying Zuo, Mahima Goel, Audrey Lam, Debapriya Dutta, Betsy Barnick, Maria Grosse-Perdekamp, Aiman Soliman, Zeynep Madak-Erdogan. In silico reconstruction of primary and metastatic estrogen receptor positive breast tumor architecture using geographic information science-augmented spatial transcriptomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 4735.