Abstract Ovarian cancer (OC) has the highest fatality rate of all gynecologic cancers, with survival improving only 9% between 1975 and 2015, compared to a 21% increase across all cancers. This is largely driven by late-stage (III/IV) detection when five-year survival is 30%. Diagnosis is often confounded by vague abdominal symptoms (VAS) common in non-cancerous disorders that overlap with symptoms of early-stage OC. Despite diagnostic advancements, outcomes remain poor, highlighting the urgent unmet need for novel biomarker approaches. Mass spectrometry-based lipidomics has emerged as a powerful tool for novel biomarker identification, with the potential to transform early OC detection. We conducted multi-omic analysis of serum from two independent, clinically annotated discovery cohorts using UHPLC-MS untargeted lipidomics and a panel of protein biomarkers for OC by immunoassay (CA125, HE4, MUC1, FOLR1). Cohort 1 (N=433) specimens are from the University of Colorado Gynecologic Tissue and Fluid Bank and vendors. Samples included OC patients across subtypes and stages (N=60 early-stage, N=99 late-stage) and non-cancerous controls designed to mimic the symptomatic population, including healthy donors (N=75), benign gynecological disorders (N=155), and gastrointestinal disorders (N=44). Cohort 2 (N=399) specimens were collected from a prospectively enrolled symptomatic population through Manchester University NHS Foundation Trust and vendors. Samples included OC patients across subtypes and stages (N=52 EOC, N=56 LOC), benign gynecological disorders (N=85), and symptomatic normal individuals (N=206). Although all protein biomarkers show group-level differences, overlap between OC and controls limits clinical utility using standard thresholds. Lipidomic profiling showed unique signatures for OC compared to control groups, with separation between OC and controls by partial least squares discriminant analysis. Corresponding variable importance in projection scores highlight key lipid features driving separation, including phospholipids (PL), sphingomyelins (SM), and ceramides. Using z-score normalized sum values, we also observe class-level increases in SM, decreases in triglycerides (TG), and bi-directional alterations across various PLs. Across two independent cohorts with distinct patient populations, we identified similar alterations in lipid profiles for OC when compared to non-cancer groups. Alterations in SMs, TGs, and PLs have been implicated in cancer biology individually, but to the best of our knowledge this is the first study to report on their changes in individuals with VAS. As remodeling of lipid metabolism is associated with cancer biology, decreased TGs may reflect altered energy utilization. This remodeling also impacts lipid composition in classes such as PLs, SMs, and ceramides. Shifts in lipid profiles may indicate altered inflammation status, oxidative vulnerability, and membrane remodeling in the tumor microenvironment. Together, this highlights the utility of lipidomics for identifying novel tumor biomarkers for early-stage OC. Citation Format: Rachel Culp-Hill, Brendan M. Giles, Charles M. Nichols, Robert A. Law, Enkhtuya Radnaa, Kian Behbakht, Benjamin G. Bitler, Chloe E. Barr, Emma Crosbie, Vuna Fa, Abigail McElhinny. Serum-based multi-omic signatures of ovarian cancer in women with vague abdominal symptoms: analysis of two independent cohorts 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 (18Suppl): Abstract nr B019.
Culp‐Hill et al. (Fri,) studied this question.