Type 2 Diabetes Sub-Phenotypes and Their Association with Cardiovascular Disease Risk: A Multi-Center Study

Aims/HypothesisType 2 diabetes mellitus (T2D) is a heterogeneous condition influenced by lipid metabolism, inflammation, and genetic predisposition, all of which contribute to variable cardiovascular disease (CVD) risk. Identifying robust T2D sub-phenotypes and understanding their interactions with genetic predisposition is critical for personalized CVD risk assessment and care. This study aims to derive clinically relevant T2D sub-phenotypes and assess their association with CVD risk by employing robust methodology and replication across cohorts. MethodsWe analyzed data from the Rotterdam Study (n=1,250), applying Gaussian mixture clustering to derive T2D sub-phenotypes based on nine metabolic risk factors: age at diabetes diagnosis, sex, body mass index (BMI), fasting blood glucose, HOMA-IR, cholesterol levels (total, HDL, LDL), and C-reactive protein (CRP). Cox proportional hazard models adjusted for confounders were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for associations between T2D sub-phenotypes and a composite CVD outcome (coronary heart disease and stroke). Kaplan-Meier (KM) survival curves were created to study the risk of incident CVD across T2D sub-phenotypes, with the lowest-risk sub-phenotype as the reference group. Polygenic risk scores (PRS) for T2D, divided into tertiles, were included to explore the interaction of genetic predisposition with diabetes sub-phenotypes. Clustering was replicated in the KORA (n=243) and PROSPER (n=179) cohorts, with association analyses validated in the KORA cohort. We considered effect size and confidence intervals, not just p-values, for comprehensive result interpretation. ResultsThree distinct T2D sub-phenotypes emerged: (1) an "unspecified" sub-phenotype (53.4%) with lower levels of metabolic risk factors, (2) an "insulin-resistant" sub-phenotype (23.8%) characterized by higher BMI, HOMA-IR, and CRP, and (3) a "dyslipidemic" sub-phenotype (22.3%) with elevated total and LDL-cholesterol. Compared to the dyslipidemic sub-phenotype (reference group based on KM analyses), the adjusted HR for incident CVD was 1.04 (95% CI: 0.76, 1.42) for the unspecified sub-phenotype and 1.20 (95% CI: 0.84, 1.72) for the insulin-resistant sub-phenotype, indicating a slightly elevated risk of CVD for the insulin-resistant sub-phenotype. Among individuals with high T2D PRS, the insulin-resistant sub-phenotype exhibited the highest CVD risk (HR 2.28, 95% CI 1.13, 4.60) compared to low and medium PRS from T2D. The robustness of the sub-phenotypes and their associations with CVD risk was confirmed in independent KORA and PROSPER cohorts. Conclusions/InterpretationThese findings emphasize the importance of understanding metabolic and clinical diversity within T2D to better guide personalized management strategies. Further research through longitudinal studies, diverse populations, and advanced molecular profiling is essential to refine sub-phenotypic classifications and uncover underlying mechanisms to enhance patient outcomes