Abstract
<title>Abstract</title> <p>Background: Ovarian cancer remains the most lethal gynecologic cancer, with limited improvements in patient survival despite targeted therapies and a high recurrence rate (~80%). Current standard-of-care for frontline treatment involves platinum-based chemotherapy, but the emergence of resistant clones limits long-term efficacy. Existing models often overlook critical interactions between cancer cells and their microenvironment. Therefore, we investigated the ovarian cancer microenvironment to identify cell populations and markers driving treatment resistance. Methods: We employed a multi-modal systems biology approach, integrating multiplex immunohistochemistry, bulk, and single-cell RNA sequencing to characterize the ovarian cancer microenvironment. Cell type composition was quantified using ImageJ (mIF) and computational deconvolution tools (CIBERSORTx, singleR) for benign (n=6-13) and cancer (n=7-20) samples. Platinum-sensitivity was determined by mapping single-cell data to a clinically annotated reference. Differential expression analysis and pathway enrichment were performed to identify key biological processes between benign vs cancer and sensitive vs resistant phenotypes. Additionally, a combinatorial marker identification tool (COMET) was used to determine a resistant signature in the sc-RNAseq dataset, which was validated using pseudotime in sc-RNAseq and the TCGA-OV bulk-RNAseq cohort. Results: Across modalities, results showed an increase in macrophage and T cell marker expression with an upregulation of inflammatory and immune pathways, alongside decreased fibroblast abundance, in cancer compared to benign tissues. Resistant samples also showed high expression of macrophage and fibroblast markers paired with an enrichment of the epithelial-to-mesenchymal transition pathway while sensitive samples showed high expression of T and NK cell markers and the upregulation of immune pathways. COMET identified two distinct resistant programs: an EMT-associated fibroblast signature characterized by INHBA, TIMP3 and NNMT; and a canonical epithelial ovarian cancer signature characterized by SLPI, MMP7, and WFDC2. Resistant signature scoring of bulk data from the TCGA-OV cohort predicted shorter treatment-free intervals for patients with higher signature scores and longer treatment-free intervals for patients with lower scores. Conclusions: These findings highlight the importance of tumor microenvironment components, particularly macrophages and fibroblasts, as key contributors to resistance in ovarian cancer and establish a potential resistant signature for biomarker discovery.</p>