Global lipidomics analysis of early- and late-stage ovarian cancer serum.

e17587 Background: Ovarian cancer (OC) is the fifth leading cause of cancer-related deaths among women. For most patients, detection occurs at late stages (III/IV) when five-year survival is <30%. Currently, there is an unmet medical need for effective diagnostic tools to detect OC at earlier stages (I/II) when survival rates are greatly improved. The use of LCMS-based lipidomics has become an invaluable technology for the characterization of the serum lipid profile from normal to early- and late-stage disease. Understanding these alterations can offer a unique insight into molecular mechanisms of disease progression, tumor biology, and metabolic pathways with the potential for diagnostic and therapeutic targeting. Here, we have defined the serum lipid profile of patients with OC, spanning all stages and multiple subtypes. Methods: We conducted untargeted lipidomics using UHPLC-MS (Exploris 240, Thermo Scientific) on a cohort from the University of Colorado Gynecologic Tissue and Fluid Bank and commercial vendors (N=301). Samples included pre- and post-menopausal patients with OC across subtypes and stages (N=219 total: 80 early-stage, 139 late-stage) and normal donors with representative demographics (N=82). Feature assignments and relative quantitation were performed using internal standards, characteristic fragment moieties, and comparison against MS/MS spectral databases. Results: 38 lipid classes were detected, comprising 1570 unique lipid compounds with significant aberrations in both early- and late-stage OC when compared against normal serum. Significantly decreased classes across all OC stages included cardiolipins, triglycerides, and phospholipid classes such as LPA, LPC, LPE, PE, and PI. Conversely, significantly increased classes across all OC stages included bile acids, carnitines, free fatty acids, gangliosides, sphingomyelins, and sterols. Additionally, several classes contained species that were both upregulated and downregulated, such as MG, DG, PA, PC, PS, LPG, and LPS. All OC stages, regardless of subtype, exhibit clear differences as compared to normal serum. Conclusions: In summary, we have identified an altered lipid profile in both early- and late-stage OC serum. Similarity between early- and late-stage disease suggests early-stage OC serum is distinguishable from normal serum. Consistent with this, preliminary modeling yielded an AUC of 0.94 (0.91-0.97) for all stages and subtypes of OC vs. normal. Excitingly, early stage (I/II) vs. normal AUC was 0.93 (0.89-0.97), and early-stage high grade serous (HGS) OC vs normal was 0.94 (0.88-0.99.) This novel class of lipid biomarkers offers a promising diagnostic avenue with greater sensitivity than currently available options. Future studies will interrogate the mechanism underlying this unique lipid profile and build a robust statistical model that effectively differentiates non-cancer from OC with high sensitivity and specificity.