EP495 - ECE_1400 - Mendelian randomization identifies no causal relationship of insulin resistance on hippocampal volume

Abstract

Insulin resistance is a systemic condition associated with altered brain metabolism and increased risk of cognitive and neurodegenerative disorders. Observational and neuroimaging studies suggest links between peripheral insulin resistance and structural brain changes, but these findings are limited by confounding. Mendelian randomization (MR) provides a genetic approach to determine whether insulin resistance causally influences brain structure, with important implications for endocrine mechanisms and metabolic therapies. To assess the causal relationship between genetically predicted systemic insulin resistance and hippocampal volume (HV) using two-sample MR. A two-sample Mendelian randomization study was performed using publicly available data to assess the causal effect of fasting insulin (FI)/glycated hemoglobin (HbA1c) on brain structure. Independent genome-wide significant variants (P < 5 × 10−⁸) selected from large-scale FI/HbA1c GWAS meta-analyses in individuals of European ancestry were used as instrumental variables. HV based on magnetic resonance imaging (MRI) GWAS data were available from UK Biobank Causal estimates of FI/HbA1c on HV were derived using inverse variance weighted (IVW) analysis and compared to those from weighted median and MR-Egger methods for consistency. Heterogeneity and horizontal pleiotropy were evaluated using Cochran's Q statistic and MR-Egger intercept, respectively, and leave-one-out analyses were conducted to assess result robustness. Genetically predicted FI was not causally associated with HV. The IVW analysis showed a non-significant effect estimate (β = −.07, P = .71), which was directionally consistent across other MR methods (βweighted median = −.11, Pweighted median = .58; βMR-Egger = −0.17, PMR-Egger = 0.88). Cochran's Q test indicated no heterogeneity among instrumental variables (IVW Q P = .052) or evidence of directional horizontal pleiotropy as assessed by the MR-Egger intercept (P = .93). Consistent with these findings, no causal effects of genetically predicted HbA1c levels on HV were observed. Leave-one-out analyses demonstrated that no single variant disproportionately influenced the overall causal estimates. This MR study provides genetic evidence that systemic insulin resistance related traits do not have a direct causal effect on HV, a structural MRI-derived trait. These findings indicate that glucose homeostasis may not directly influence hippocampal anatomy, but other, non-glycemic, mechanisms might be driving the associations between insulin resistance and central nervous system outcomes.