Pathophysiology-based subtypes of individuals at high-risk of type 2 diabetes: multi-omics profiles and lifestyle differences across subtypes.
Previous studies identified subtypes among individuals at elevated risk of type 2 diabetes mellitus (T2DM), yet the molecular signatures distinguishing these subtypes remain poorly characterized. We aimed to identify and characterize T2DM risk subtypes using routine and non-routine clinical variable lists, and to compare their metabolomic, proteomic, and lifestyle profiles.
In the Netherlands Epidemiology of Obesity study (median age 56 years; median BMI 29 kg/m2), we applied partitioning around medoids (PAM) clustering using two variable lists (routine, N = 5235; non-routine, N = 1510), both including variables derived from a liquid mixed meal challenge. Cox proportional hazards models estimated associations between subtypes and T2DM incidence. Random forest models identified discriminative metabolites and proteins across subtypes.
Each variable list yielded four subtypes ranging from an insulin-sensitive and lean profile (subtype 1) to an obese profile with ectopic fat accumulation and insulin resistance (subtype 4), with a graded increase in T2DM risk (hazard ratios ranging from 1.9 [95% confidence interval (CI): 0.4-9.8] to 19.5 [95% CI: 9.1-41.6]). Both subtyping schemes captured metabolic heterogeneity beyond conventional weight-by-glycemia categories, e.g., redistributing overweight or obese but normoglycemic individuals across subtypes with divergent metabolic profiles and T2DM risks. While multi-omics profiling revealed shared metabolic and proteomic markers across higher-risk subtypes (e.g., glycoprotein acetyls, glucose, hepatocyte growth factor), subtype assignment was predominantly driven by fasting glucose and lipoprotein levels (e.g., very-low-density lipoprotein [VLDL]), with additional subtype-specific molecular signatures (e.g., branched-chain amino acids). Individuals in the higher-risk subtypes also exhibited less healthy lifestyle characteristics, including poorer dietary quality.
Four metabolic subtypes with graded T2DM risk were identified in a predominantly overweight or obese, middle-aged population, revealing metabolic heterogeneity among individuals who appear homogeneous under conventional weight-by-glycemia categories. Although subtype differentiation was largely driven by fasting glucose and lipoproteins, multi-omics profiling uncovered additional molecular signatures, suggesting that data-driven subtyping may complement conventional risk markers and inform targeted prevention.
In the Netherlands Epidemiology of Obesity study (median age 56 years; median BMI 29 kg/m2), we applied partitioning around medoids (PAM) clustering using two variable lists (routine, N = 5235; non-routine, N = 1510), both including variables derived from a liquid mixed meal challenge. Cox proportional hazards models estimated associations between subtypes and T2DM incidence. Random forest models identified discriminative metabolites and proteins across subtypes.
Each variable list yielded four subtypes ranging from an insulin-sensitive and lean profile (subtype 1) to an obese profile with ectopic fat accumulation and insulin resistance (subtype 4), with a graded increase in T2DM risk (hazard ratios ranging from 1.9 [95% confidence interval (CI): 0.4-9.8] to 19.5 [95% CI: 9.1-41.6]). Both subtyping schemes captured metabolic heterogeneity beyond conventional weight-by-glycemia categories, e.g., redistributing overweight or obese but normoglycemic individuals across subtypes with divergent metabolic profiles and T2DM risks. While multi-omics profiling revealed shared metabolic and proteomic markers across higher-risk subtypes (e.g., glycoprotein acetyls, glucose, hepatocyte growth factor), subtype assignment was predominantly driven by fasting glucose and lipoprotein levels (e.g., very-low-density lipoprotein [VLDL]), with additional subtype-specific molecular signatures (e.g., branched-chain amino acids). Individuals in the higher-risk subtypes also exhibited less healthy lifestyle characteristics, including poorer dietary quality.
Four metabolic subtypes with graded T2DM risk were identified in a predominantly overweight or obese, middle-aged population, revealing metabolic heterogeneity among individuals who appear homogeneous under conventional weight-by-glycemia categories. Although subtype differentiation was largely driven by fasting glucose and lipoproteins, multi-omics profiling uncovered additional molecular signatures, suggesting that data-driven subtyping may complement conventional risk markers and inform targeted prevention.
Authors
Deng Deng, Hameete Hameete, Schrauwen Schrauwen, Wagner Wagner, van Hylckama Vlieg van Hylckama Vlieg, Rosendaal Rosendaal, Mook-Kanamori Mook-Kanamori, le Cessie le Cessie, van Dijk van Dijk, de Mutsert de Mutsert, Li-Gao Li-Gao
View on Pubmed