Associations of Insulin Sensitivity with Risk of All-Cause and Cause-Specific Dementia: A Prospective Cohort Study

Impaired insulin sensitivity has been implicated in neurodegeneration, but whether clinically accessible indices of insulin sensitivity predict dementia risk in the general population remains unclear. We examined associations of the estimated glucose disposal rate (eGDR), a validated marker of insulin sensitivity, with incident all-cause dementia (ACD), Alzheimer's disease (AD), and vascular dementia (VaD), and assessed whether eGDR modifies genetic susceptibility to AD. We included 290,898 dementia-free participants from the UK Biobank. eGDR was calculated using waist circumference, hypertension, and glycated hemoglobin. AD polygenic risk score (AD-PRS) was categorized into tertiles. Longitudinal associations were assessed using Kaplan-Meier (KM) survival analysis and Cox proportional hazards models, with restricted cubic splines employed to examine potential non-linear relationships. Joint effects and interactions between eGDR and AD-PRS were evaluated. The robustness of findings was tested through extensive subgroup and sensitivity analyses. Over a mean follow-up of 13.23 years, 4794 participants developed ACD, 2138 developed AD, and 1070 developed VaD. KM curve revealed that dementia incidence differed significantly across eGDR quartiles (all log-rank P < 0.001). RCS showed significant non-linear inverse associations between eGDR and dementia risk (all P for non-linearity < 0.001). In fully adjusted Cox models, each 1-unit higher eGDR was associated with lower risk of ACD (HR 0.92, 95% CI 0.90-0.94), AD (0.94, 0.92-0.97), and VaD (0.82, 0.78-0.85). Compared with eGDR-Q1, Q4 had lower risks of ACD (HR 0.67), AD (HR 0.70), and VaD (HR 0.42) (all P < 0.001). eGDR modified AD genetic risk (P for interaction < 0.001); participants with low AD-PRS and high eGDR had the lowest AD risk (HR 0.17, 95% CI 0.12-0.23 vs. high AD-PRS and low eGDR group). Findings were consistent across subgroup and sensitivity analyses. Higher eGDR, indicative of greater insulin sensitivity, was associated with reduced risk of incident all-cause and cause-specific dementia, with a significant non-linear inverse dose-response relationship observed. Individuals with both high eGDR low high AD-PRS exhibited the lowest risk of developing AD. These findings suggest that insulin sensitivity may represent a modifiable target for dementia prevention, particularly among individuals with elevated genetic susceptibility.