Causal association between insulin sensitivity index and Alzheimer's disease

Evidence from observational and Mendelian randomization (MR) studies suggested that insulin resistance (IR) was associated with Alzheimer's disease (AD). However, the causal effects of different indicators of IR on AD remain inconsistent. Here, we aim to assess the causal association between the insulin sensitivity index (ISI), a measure of post-prandial IR, and the risk of AD. We first conducted primary and secondary univariable MR analyses. We selected 8 independent genome-wide significant (p < 5E-08, primary analyses) and 61 suggestive (p < 1E-05, secondary analyses) ISI genetic variants from large-scale genome-wide association studies (GWAS; N = 53 657), respectively, and extracted their corresponding GWAS summary statistics from AD GWAS, including IGAP2019 (N = 63 926) and FinnGen_G6_AD_WIDE (N = 412 181). We selected five univariable MR methods and used heterogeneity, horizontal pleiotropy test, and leave-one-out sensitivity analysis to confirm the stability of MR estimates. Finally, we conducted a meta-analysis to combine MR estimates from two non-overlapping AD GWAS datasets. We further performed multivariable MR (MVMR) to assess the potential mediating role of type 2 diabetes (T2D) on the association between ISI and AD using two MVMR methods. In univariable MR, utilizing 8 genetic variants in primary analyses, we found a significant causal association of genetically increased ISI with decreased risk of AD (OR = 0.79, 95% CI: 0.68-0.92, p = 0.003). Utilizing 61 genetic variants in secondary analyses, we found consistent findings of a causal effect of genetically increased ISI on the decreased risk of AD (OR = 0.89, 95% CI: 0.82-0.96, p = 0.003). Heterogeneity, horizontal pleiotropy test, and leave-one-out sensitivity analysis ensured the reliability of the MR estimates. In MVMR, we found no causal relationship between ISI and AD after adjusting for T2D (p > 0.05). We provide genetic evidence that increased ISI is significantly and causally associated with reduced risk of AD, which is mediated by T2D. These findings may inform prevention strategies directed toward IR-associated T2D and AD.