LI can delay the onset of T2D. How inter-individual variability in response to LI relates to baseline phenotypic characteristics would help clinicians to advise on risk reduction. 328 subjects at high risk of T2D underwent a 12-weeks LI. Before and after LI, subjects were profiled for glucose and insulin measured through 75g-OGTT, peripheral (Matsuda (ISI) and HOMA) and hepatic (Hep-IR) insulin resistance indexes, insulinogenic index (IGI) , disposition index (DI=MI x dAUC0-180-Ins/glu) and target metabolomics. Hierarchical cluster analysis, based on indexes of insulin resistance (HOMA and ISI) and secretion (AUCins and IGI) , was performed to stratify the subjects at baseline. Pre vs. post values were compared in each cluster. Cluster analysis identified 4 groups with different degrees of glucose tolerance, insulin sensitivity and secretion (Figure A-D) , CL1 being the best. The 4 clusters showed different responses to LI: CL2-4 significantly improved insulin and reduced AUCins and α-ketoglutarate (p<0.05) , while no significant changes in CL1 (Figure E) .
Conclusions: Hierarchical clustering identified phenotypes that might predict metabolic responses to LI.
S.Sabatini: None. J.J.Nolan: Consultant; My Way Digital Health, Speaker's Bureau; Novo Nordisk A/S. D.O'gorman: None. A.Gastaldelli: Advisory Panel; Boehringer Ingelheim International GmbH, Novo Nordisk Global Business Services, Consultant; Boehringer Ingelheim International GmbH, Inventiva Pharma, Other Relationship; Eli Lilly and Company, Gilead Sciences, Inc., Pfizer Inc., Speaker's Bureau; Novo Nordisk.
