Abstract 6351: Characterizing the spatial heterogeneity of pregnancy-associated triple negative breast cancer

Abstract Background: Pregnancy-associated triple negative breast cancer (PA-TNBC) is one of the highest-risk breast cancers, marked by an aggressive phenotype that lacks targeted treatment options. Although pregnancy is thought to decrease long-term breast cancer risk, studies have shown that breast cancers diagnosed immediately following post-lactational involution have a 2-3x greater risk of metastasis and death than other breast cancers. The molecular basis for this is not fully understood. Methods: Spatial transcriptomics using the GeoMx Digital Spatial Profiler was performed on 46 PA-TNBC tissues from 33 patients, collected from Louisiana State University. 10 patients were diagnosed with TNBC pre-involution (PRE) and 23 patients were diagnosed post-involution (POST, <3 years after delivery). Tissues were H&E stained, and Region of Interest (ROI) selection was performed by a pathologist. ROIs were segmented into PanCK+ (epithelial) and PanCK- (tumor microenvironment - TME) regions and individually sequenced. After quality control (QC), lactation-associated gene expression agnostically validated the tissue classification as PRE vs POST. Linear mixed models identified differentially expressed genes (FDR < 0.05) between involution groups and epithelial types. The cell types in the TME were deconvoluted, and differentially co-occurring cell types were inferred between the groups. Finally, soft fuzzy clustering of the POST samples was performed to identify gene modules that exhibit similar expression patterns as a function of time of diagnosis (years after delivery). Results: After QC there were 1562 segments (471 PRE and 1091 POST) from 883 ROIs. We find that the main difference between PRE vs POST lies in the pre-cancerous tissue and is likely influenced by the TME. Differential analysis shows that POST non-invasive epithelia have an inflammatory phenotype and are actively remodeling, going through complement-mediated cell state changes. Further, POST-TME shows increased immune, inflammatory, and angiogenesis signaling, especially in the vicinity of the non-invasive epithelia. Cell type deconvolution confirmed increased immune presence and co-localization of immune cell types in POST. Finally, we find that transcriptomic signals from these processes peak in patients diagnosed 2 years post-delivery. Conclusions: This study uses spatial transcriptomics to identify 1) markers of aggressive precancerous biology in POST non-invasive epithelia, 2) TME influence on the normal→tumor transformation, and 3) a pseudo temporal gradation of molecular severity to understand how long after delivery are PA-TNBC patients at risk for aggressive cancer. Our study is the first to apply spatial transcriptomics to PA-TNBC tissue and our findings will inform future treatment possibilities for this understudied patient group. Citation Format: Darya Veraksa, Kavitha Mukund, Jerneja Tomsic, David Frankhouser, Augusto Ochoa, Victoria Seewaldt, Shankar Subramaniam. Characterizing the spatial heterogeneity of pregnancy-associated triple negative breast cancer [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 6351.