A genome-wide association meta-analysis of all-cause and vascular dementia

Abstract

Introduction: Dementia is a multifactorial disease with Alzheimer's disease (AD) and vascular dementia (VaD) pathologies making the largest contributions. Yet, most genome-wide association studies (GWAS) focus on AD.

Methods: We conducted a GWAS of all-cause dementia (ACD) and examined the genetic overlap with VaD. Our dataset includes 800,597 individuals, with 46,902 and 8702 cases of ACD and VaD, respectively. Known AD loci for ACD and VaD were replicated. Bioinformatic analyses prioritized genes that are likely functionally relevant and shared with closely related traits and risk factors.

Results: For ACD, novel loci identified were associated with energy transport (SEMA4D), neuronal excitability (ANO3), amyloid deposition in the brain (RBFOX1), and magnetic resonance imaging markers of small vessel disease (SVD; HBEGF). Novel VaD loci were associated with hypertension, diabetes, and neuron maintenance (SPRY2, FOXA2, AJAP1, and PSMA3).

Discussion: Our study identified genetic risks underlying ACD, demonstrating overlap with neurodegenerative processes, vascular risk factors, and cerebral SVD.

Highlights: We conducted the largest genome-wide association study of all-cause dementia (ACD) and vascular dementia (VaD). Known genetic variants associated with AD were replicated for ACD and VaD. Functional analyses identified novel loci for ACD and VaD. Genetic risks of ACD overlapped with neurodegeneration, vascular risk factors, and cerebral small vessel disease.

Keywords: Alzheimer's disease; GWAS meta‐analysis; all‐cause dementia; cross‐ancestry; genome‐wide association study (GWAS); vascular dementia.

FIGURE 1

Manhattan plot of ACD GWAS. In addition to variants in APOE region, we identified five new genetic loci associated with VaD. Blue and red lines correspond to p value of 5e⁻⁷ and 5e⁻⁸ for genome‐wide suggestive and significant SNPs, respectively. Manhattan plots for the cross‐ancestry meta‐analysis. Each dot represents a SNP, the x‐axis shows the chromosomes where each SNP is located, and the y‐axis shows −log10 p value of the association of each SNP with ACD in the cross‐ancestry meta‐analysis. The red horizontal line shows the genome‐wide significant threshold (p value = 5e‐8; −log10 p value = 7.30). The nearest gene to the most significant SNP in each locus has been labeled.

FIGURE 2

Manhattan plot of VaD GWAS. In addition to variants in the APOE region, we identified five new genetic loci associated with VaD. Blue and red lines correspond to a p value of 5e⁻⁷ and 5e⁻⁸ for genome‐wide suggestive and significant SNPs, respectively. Manhattan plots for cross‐ancestry meta‐analysis. Each dot represents a SNP, the x‐axis shows the chromosomes where each SNP is located, and the y‐axis shows the −log10 p value of the association of each SNP with VaD in the cross‐ancestry meta‐analysis. The red horizontal line shows the genome‐wide significant threshold (p value = 5e‐8; −log10 p value = 7.30). The gene closest to the most significant SNP in each locus has been labeled.

FIGURE 3

Q‐Q plots of ACD (left) and VaD (right) GWASs. The expected p values (x‐axis) are plotted against the observed p values (y‐axis). The units of the axes are the −log10 of the p value. The red and blue curves represent the plots with MAF ≥ 0.05 and 0.01, respectively. The diagonal line of the null hypothesis and its 95% confidence interval are plotted in gray based on the p values without the previously reported SNPs. The red dotted line represents the cutoff for genome‐wide significance. MAF, minor allele frequency.

FIGURE 4

Shared genetic contribution between ACD/VaD and related risk factors. Contributions determined by LD score regression analysis (LDSR) (top), and Genetic Covariation Analyzer (GNOVA) (bottom). Effect sizes (rg) and significance levels (logp) are represented by color and symbol size. AD, Alzheimer's disease; GCF, general cognitive function; all stroke (AS) and its subtypes (ischemic, IS; cardioembolic, CES; small vessel, SVS; large artery, LAS); WMH, white matter hyperintensity burden; DBP, diastolic blood pressure; SBP, systolic blood pressure; PP, pulse pressure; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein. Diamond shape: non‐significant. Cross‐significant causal effect estimates from MR‐LAP analysis.