Shared genetic susceptibility between diabetes and dementia in the Million Veteran Program.

Abstract

Diabetes and dementia are diseases of high healthcare burden worldwide, and there is evidence that diabetes is a risk factor for dementia. Compared to those without diabetes, the relative risk of dementia in individuals with diabetes ranges from 1.4 to 2.2. Our goal was to evaluate shared genetic susceptibility for diabetes and dementia. Using imputed genotype data from MVP, we built a weighted genetic risk score (GRS) for T2D using 331 variants and their effect sizes from published genome-wide association studies of T2D. We evaluated associations with four nested definitions of dementia: strict Alzheimer's Disease (AD), AD plus non-specific dementias (AD-Plus), AD-related dementia (AD-RD), and all-cause dementia. We used logistic regression to assess associations of the GRS with dementia sub-types, stratified by European, African, and Hispanic ancestries, using a significance threshold of 0.0033 to account for multiple testing. As a sensitivity test, we removed the rs429358 variant near the APOE gene from the GRS as it has known genome-wide significant associations with AD. In 19,562 cases and 252,893 controls of European ancestry, we found associations of the T2D GRS with all-cause dementia (p=1.30e-07, OR 1.06), and in 14,455 cases and 258,000 controls with AD-RD (p=3.35e-07, OR = 1.07), but not with AD-Plus or strict AD. The T2D GRS was not associated with any of the dementia outcomes in participants of African (p =0.27 to 0.48) or Hispanic (p = 0.15 to 0.94) ancestries. When removing the APOE locus, we found that the effect size estimates for the GRS increased and the probability of association by chance decreased: all-cause dementia (p=2.08e-12, OR = 1.08), AD-RD (p=1.37e-12, OR = 1.10), and AD-Plus (p=3.97e-06, OR = 1.08). We found evidence of shared genetic susceptibility between diabetes and dementia. Future work will evaluate causality of the diabetes-dementia association using Mendelian Randomization. This research is based on data from the Million Veteran Program, Office of Research and Development, Veterans Health Administration, and was supported by awards MVP009 and MVP022. This publication does not represent the views of the Department of Veteran Affairs or the United States Government.