Abstract 13814: DMGV is a Novel Marker of Liver Fat and Predicts Future Development of Type 2 Diabetes

Background: Unbiased “non-targeted” metabolite profiling techniques hold considerable promise for biomarker and pathway discovery, though there have been few successful applications to human disease. We aimed to uncover novel biomarkers of non-alcoholic fatty liver disease (NAFLD) and cardiometabolic disease using non-targeted metabolomic profiling in well-phenotyped human cohorts. Methods: Non-targeted metabolite profiling was performed using hydrophilic interaction chromatography on a UHPLC Q-Exactive hybrid orbitrap LC/MS system. We searched HMDB, Metlin, Chemspider, and Pubchem databases using a 5-ppm mass tolerance for potential identifications of unknown metabolites. Following unambiguous metabolite identification, a high-throughput targeted metabolite assay was generated using a triple quadrupole MS. Stepwise logistic regression was used to determine the association of metabolites with clinical traits and outcomes. Results: The non-targeted platform was applied first to a discovery cohort of 1066 Framingham Heart Study (FHS) Gen 3 participants, of whom 470 underwent a CT scan as part of a comprehensive assessment of adipose depots, including liver fat. An unknown metabolite with a mass/charge ratio of 202.1185 had the strongest association with CT-defined liver fat, which was quantified using the liver-to-phantom ratio (P=1.16E-10, adjusting for age, sex, smoking, alcohol consumption, HDL and triglyceride concentration, HOMA-IR, hypertension, and BMI). Following structural elucidation by LC-MS and the synthesis of a chemical standard, we confirmed the peak identity as α-keto-δ-(N G N G -dimethylguanidino)-valeric acid (DMGV). DMGV was significantly associated with non-alcoholic steatohepatitis (NASH), in a cohort of 36 patients with biopsy-proven NASH and 36 age, sex, and BMI-matched controls (OR=1.96, P=0.011, conditionally adjusted for age, sex, and BMI). Further, DMGV was predictive of future type 2 diabetes mellitus (T2D) in 196 cases of incident diabetes and 126 controls in the Malmo Diet and Cancer Study (OR=1.56, P=8.6E-4, adjusting for age, sex, fasting glucose, and BMI; mean follow-up 12.8 years). Conclusions: DMGV is a novel marker of NAFLD and NASH, and is a predictor of future T2D.