Genetic Evidence for Causal Association Between Atrial Fibrillation and Dementia: A Mendelian Randomization Study

Figure 1. Overall design of the Mendelian randomization study. The association between exposure (atrial fibrillation) and outcomes (dementia and dementia subtypes) is influenced by both causal effect and confounding effects of common risk factors. Mendelian randomization analysis evaluates the causal association of genetically predicted atrial fibrillation and dementia outcomes under 3 critical assumptions: (1) the relevance assumption: genetic instruments must be robustly associated with atrial fibrillation; (2) the independence assumption: genetic instruments must not be associated with potential confounders; and (3) the exclusivity assumption: the association between the genetic instruments and the outcome must be achieved only through the exposure. Multivariable Mendelian randomization analysis allows investigation of direct causal effect of atrial fibrillation on dementia with adjustment for multiple candidate mediators. CSVD indicates cerebral small‐vessel disease.

Figure 2. Mendelian randomization estimates for the causal effect of genetically predicted AF on overall dementia and dementia subtypes. Forest plot of the main inverse variance–weighted analyses indicating association of genetically predicted AF with risk of overall dementia, vascular dementia, Alzheimer dementia, Lewy body dementia, and frontotemporal dementia. AF indicates atrial fibrillation; FDR, false discovery rate; GWAS, genome‐wide association study; OR, odds ratio per unit increase of log odds of genetically predicted atrial fibrillation; and SNP, single‐nucleotide polymorphism.

Figure 3. UVMR and MVMR results of the causal association between genetically predicted atrial fibrillation and dementia. Blue plot represents the causal effect of genetically predicted atrial fibrillation on dementia in UVMR analysis. Red plots represent the causal effect of genetically predicted atrial fibrillation on dementia, with adjustment for 10 candidate mediators separately in MVMR analyses. MVMR indicates multivariable Mendelian randomization; OR, odds ratio per unit‐increase in log odds of genetically predicted atrial fibrillation; and UVMR, univariable Mendelian randomization.

Figure 4. Main design and findings of the Mendelian randomization study. We investigated the association of genetic predisposition to atrial fibrillation with overall dementia and multiple dementia subtypes. Genetically predicted atrial fibrillation was modestly associated with an increased risk of overall dementia and strongly associated with vascular dementia. The causal effect of genetically predicted atrial fibrillation on dementia was potentially mediated by ischemic stroke and low cardiac output. The figure was partly generated using Servier Medical Art, provided by Servier, licensed under a Creative Commons Attribution 3.0 unported license.