Data-driven subtypes of type 2 diabetes and risk of dementia, stroke, and brain structural changes in the UK Biobank

Abstract

AIMS: Type 2 diabetes (T2D) is associated with increased risks of dementia and stroke with heterogeneity. We aimed to identify data-driven T2D subtypes and examine associations with dementia, stroke, mortality, and brain structure. METHODS: In 14,353 UK Biobank participants with T2D (mean age 59.8 years), k-means clustering was applied using age at diagnosis, body mass index, glycated hemoglobin, insulin resistance, systolic blood pressure, and C-reactive protein. Cox models estimated hazard ratios (HRs) and 95% confidence intervals (CIs). Brain structural measures were analyzed in a subset. RESULTS: Three subtypes emerged: severe obesity-related inflammatory diabetes (SOID), mild metabolic diabetes (MMD, reference), and mild age-related hypertension-predominant diabetes (MARD-H). Versus MMD, SOID showed higher risks of all-cause dementia (HR 1.24, 95% CI 1.01-1.52), vascular dementia (HR 1.42, 95% CI 1.01-1.99), stroke (HR 1.38, 95% CI 1.14-1.68), ischemic stroke (HR 1.48, 95% CI 1.21-1.82), all-cause mortality (HR 1.59, 95% CI 1.45-1.74), and cardiovascular death (HR 1.88, 95% CI 1.59-2.23). MARD-H showed modest relative but highest absolute risks. SOID showed lower gray matter and greater white matter hyperintensity. CONCLUSIONS: T2D subtypes show distinct vascular and neurological risks. The obesity-inflammation subtype carries elevated risks of dementia, stroke, mortality, and adverse brain changes.