Introduction and Objective: Patients with T2D can be clustered into distinct etiologic subgroups, but it is not clear if the same is true for those with prediabetes (preDM). As metabolites mark the onset of T2D prior to clinical thresholds, we clustered individuals with preDM by preDM-related metabolites and examined subgroups for long-term T2D risk.Methods: Data were included from N=5567 participants, N=2003 of whom had preDM (100≤fasting glucose<126 mg/dL). Machine learning was used to select baseline metabolites that distinguished preDM from normoglycemia. Those with preDM were clustered into subgroups by their values of the selected compounds, and chi-squared tests quantified baseline subgroup differences. Cox-proportional hazards models, controlling for demographics, and fasting glucose and insulin, examined subgroup differences in T2D risk over 20 years follow-up.Results: N=31 metabolites discriminated preDM from normoglycemia (AUC=0.96;fig1A). Clustering individuals by these compounds identified 3 subgroups (fig1B). Fasting glucose and insulin did not differ between subgroups (p<.001;fig1C), but the likelihood of developing T2D did (P=.03;fig1C &1D).Conclusion: Individuals with preDM represent discrete metabolome-based subgroups, informative for T2D progression. Future analyses will include validation in additional DEFINE-T2D cohorts.
A. Wood: Research Support; Current; Beef Checkoff. Consultant; Ended; Lundquist Institute for Biomedical Innovation. M.R. Rooney: None. I.R. Konigsberg: None. Z. Chen: None. A. Manning: None. R.E. Gerszten: None. L. Szczerbinski: Research Support; Current; Novo Nordisk. Consultant; Current; Eli Lilly and Company. M. Mi: None. M. Sevilla-Gonzalez: Research Support; Current; Novo Nordisk. Q. Qi: None. Q. Pan: None. S.K. Das: None. J.I. Rotter: None.
Veterans Affairs, United States Department of Agriculture (USDA-3092-10700-068-006-N), and National Institutes of Health (1U01DK140757; HL181500; DK127073; U01DK140761; U01DK140738; U01DK140778)
Source link
Leave a Reply