J Clin Endocrinol Metab. 2024 Nov 20:dgae812. doi: 10.1210/clinem/dgae812. Online ahead of print.
ABSTRACT
OBJECTIVE: This study aimed to investigate serum metabolomic biomarkers associated with incident type 2 diabetes mellitus (T2DM) and evaluate their performance in improving T2DM risk prediction.
METHODS: Untargeted proton nuclear magnetic resonance (1H NMR) spectroscopy-based metabolomics analyses were conducted in the Multi-Ethnic Study of Atherosclerosis (MESA; n=3460; discovery cohort) and Rotterdam Study (RS; n=1556; replication cohort). Multivariable cause-specific hazards models were used to analyze the associations between 23,571 serum metabolomic spectral variables and incident T2DM. Replicated metabolites required an FDR-adjusted P<0.01 in MESA, P<0.05 in RS, and consistent direction of association. Pathway and network analyses were conducted to elucidate biological mechanisms underlying T2DM development. Utility of the replicated metabolites in improving T2DM risk prediction was assessed based on the Framingham Diabetes Risk Score. A 2-sample Mendelian randomization was conducted to assess causal associations.
RESULTS: Nineteen metabolites were significantly associated with incident T2DM. Pathway analyses revealed disturbances in aminoacyl-tRNA biosynthesis, metabolism of branched-chain amino acids (BCAAs), glycolysis/gluconeogenesis, and glycerolipid metabolism. Network analyses identified interactions with upstream regulators including p38 MAPK, c-JNK, and mTOR signaling pathways. Adding replicated metabolites to the Framingham Diabetes Risk Score showed modest to moderate improvements in prediction performance in MESA and RS, with Δ c-statistic of 0.05 (95% CI, 0.04-0.07) in MESA and 0.03 (95% CI, 0.01-0.05) in RS. Genetically increased BCAAs and mannose were associated with T2DM.
CONCLUSIONS: 1H NMR measured metabolites involved in aminoacyl-tRNA biosynthesis, BCAA metabolism, glycolysis/gluconeogenesis, and glycerolipid metabolism were significantly associated with incident T2DM and provided modest to moderate predictive utility beyond traditional risk factors.
PMID:39566105 | DOI:10.1210/clinem/dgae812
