Abstract PR018: Spatially resolved epithelial and immune heterogeneity reveals distinct endometriosis-related precursors to ovarian cancer

Abstract Evidence suggests that certain histotypes of ovarian carcinoma -mainly clear cell (CCOC) and endometrioid (EOC) carcinoma-may arise from endometriosis (EAOC), although the biological mechanisms underlying this phenomenon remain unclear. Previously, we identified that approximately one-third of simple endometriosis (SE) lesions exhibit a PD-1/PD-L1-positive immune signature like that of EAOCs. This hypothesis generating study aims to elucidate the spatial transcriptomic profile of epithelial and immune components in different endometriosis-related diseases. GeoMx™ whole-transcriptome digital spatial profiling (∼19000 probes) quantified RNA in 165 regions of interest (72 PanCK+, 55 CD3+, 38 CD68+) across 14 FFPE oophorectomy specimens: SE (n = 2), atypical endometriosis (AE, n = 2), PD-L1-positive “cancer-like” endometriosis (CLE, n = 4), CCOC (n = 3) and EOC (n = 3). After TMM/RUV4 normalization of 11029 genes, we applied multinomial modeling, differential expression, cell-type deconvolution, ligand–receptor inference, spatial graph statistics and pseudotime reconstruction. Unsupervised embedding revealed that epithelial transcriptional programs principally dictated regional clustering. SE maintained a ciliated, progesterone-responsive signature; in AE hormonal responsiveness decreased and switched on WNT-dependent progenitor and cell-cycle networks. CLE displayed different profiles both at the epithelial and the immune infiltrate (upregulation of IL-17, cell chemotaxis, ECM degradations pathways) when compared to SE and AE. Here, ferroptotic-stress genes (HMOX1, FTH1) and neutrophil-chemotactic chemokines (CXCL1/5/8) rose along with S100 alarmins, in a loop of chronic oxidative injury coupled to IL-17-driven inflammation. Pseudotime reconstruction suggested a possible trajectory from SE to CLE, toward EAOCs. Intriguingly, epithelium directly adjacent to CCOC/EOC already expressed LEFTY1 and DKK4, respectively, implying molecular field-cancerization ahead of histological change. Immune deconvolution added some spatial context. In CLE, CD4+cells and CD15+ neutrophils formed a dominant milieu, with a phenotypic drift toward LYVE1+ pro-angiogenic macrophages (p < 0.05). Graph-based spatial statistics showed these immune cells progressively relocating from stroma into the epithelial border, a prelude to immune exhaustion in carcinomas. Finally, CLE showed an extracellular-matrix/complement enrichment, centered on CD44–SPP1 interactions and collagen and laminin isoforms. This “window-of-implantation-like” matrix provided a fertile substrate for malignant conversion. Spatial transcriptomics revealed distinct epithelial–immune landscapes within lesions that appeared identical under the microscope. A chronic CLE niche—defined by ferroptotic stress, neutrophil-dominated inflammation and extracellular-matrix remodeling—emerged as a possible precursor to malignancy. HMOX1, MUC5B and CD15+ neutrophils worths evaluation as biomarkers for early risk stratification, while ECM/IL-17 signaling may represent pathways to target therapy development. Citation Format: Valentina Iacobelli, Luca Mastrantoni, Stefano Alivernini, Giulia Scglione, Floriana Camarda, Rita Trozzi, Francesca Sillano, Alessia Piermattei, Luciano Giacò, Giulia Mantini, Gianfranco Zannoni, Anna Fagotti, Camilla Nero. Spatially resolved epithelial and immune heterogeneity reveals distinct endometriosis-related precursors to ovarian cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Ovarian Cancer Research; 2025 Sep 19-21; Denver, CO. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl):Abstract nr PR018.