Abstract MP14: Gestational Diabetes Mellitus, Circulating Metabolites, and Risk of Future Cardiovascular Disease in US Women

Introduction: Gestational diabetes mellitus (GDM) is associated with higher lifetime cardiovascular disease (CVD) risk in women, even without subsequent type 2 diabetes (T2D). Dysregulation of lipid and amino acid metabolism precedes CVD and has been associated with T2D in women with GDM history. However, the precise metabolic pathways contributing to the persistent CVD risk after GDM remain unknown. Hypothesis: A metabolomic signature of GDM history is associated with higher CVD risk. Methods: Using elastic net regression, we constructed a postpartum (24-72 hours after delivery) metabolomic signature of GDM using maternal plasma metabolomics data in the Boston Birth Cohort (BBC; 161 GDM; 1,178 non-GDM; 64% Black). The GDM signature was internally validated in the BBC using leave-one-out cross-validation and externally replicated in a subset of Nurses’ Health Study II (NHS2; 392 prior GDM; 2,326 no GDM; 98% White) with available postpartum (a median of 17.5 years after delivery) plasma metabolomics data. CVD was defined as a composite of coronary artery disease (CAD; coronary artery bypass grafting, myocardial infarction, and coronary death) and stroke. Cox proportional hazards model was used to examine associations between the GDM signature and incident CVD in both NHS2 and Nurses’ Health Study (NHS; no data on GDM history; 95% White). A genome-wide association study (GWAS) was conducted in NHS2 and NHS to identify genes implicated in the GDM signature. Results: The externally validated GDM signature consists of 44 metabolites (21 lipids, 10 amino acids, and others). Women with prior GDM had a higher signature score than those without in both BBC (p=2х10 -20 ) and NHS2 (p=0.004). In NHS2 (n=2,718), 97 CVD cases (17 cases in 392 women with prior GDM) were documented in 23 years of follow-up. The GDM signature was significantly associated with higher risk of CVD (HR = 1.31 [1.04-1.65] for per-SD increase in the signature score) and CAD (HR = 1.81 [1.30-2.52]) but not stroke (HR = 1.05 [0.76-1.45]) after adjusting for age, race, body mass index (BMI), blood pressure, baseline T2D, medication use, and lifestyle factors. The associations did not change after additionally adjusting for GDM history and incident T2D. The GDM signature mediated associations between GDM and CAD (5.7%; p=0.048) and CVD (4.4%; p=0.038). In NHS (n=4,469), 874 CVD cases were documented in 31 years of follow-up. The GDM signature was associated with incident CAD (adjusted HR = 1.18 [1.07-1.29]) but not CVD and stroke. In GWAS, we identified genetic loci harboring genes ( FADS1 , FADS2 , and SOX2-OT ) related to fatty acid metabolism, blood lipid levels, glycemic traits, BMI, and insomnia. Conclusions: We identified and validated a metabolomic signature associated with GDM history, which was associated an increased risk of future CVD, especially CAD. Ongoing work is underway to replicate these findings in other racial and ethnic groups.