Multi-ancestry meta-analysis identifies genetic modifiers of age-at-onset of Alzheimer's disease at known and novel loci

Abstract

Introduction: Much of Alzheimer's disease (AD) risk is explained by age, apolipoprotein E (APOE) genotype, and sex. We sought to identify genetic modifiers of age at onset (AAO) of AD while probing the influence of sex and APOE among those with diverse ancestry.

Methods: We performed genome-wide association studies (GWASs) of AAO in two diverse samples followed by meta-analysis, contrasting results with and without adjustment for sex and APOE. Genome-wide significance was set to p < 5×10^-8.

Results: GWASs adjusting for sex, APOE, population structure, and relatedness revealed 17 significant loci including independent associations at AD risk loci and four novel signals. APOE adjustment influenced GWAS effect sizes across the genome while sex adjustment had minimal effect.

Discussion: We identified association signals within a diverse but relatively small sample, replicating loci recently discovered in large European ancestry-only GWASs, and illustrated the power of using a quantitative trait like AAO over a binary diagnosis trait.

Highlights: Survival analysis approach identified known and novel genetic modifiers of Alzheimer's disease (AD). Multi-ancestry analyses revealed independent signals at known AD loci. Apolipoprotein E adjustment influenced variant effects across the genome.

Keywords: age at onset; apolipoprotein E; diversity; multi‐ancestry; sex differences.

FIGURE 1

Genetic diversity within the discovery and replication data sets. X axis: the first principal component (PC). Y axis: the second PC. Each PC is labeled by the amount of genetic variance explained.

FIGURE 2

GWAS for AAO of AD, adjusting for sex, ε2, ε4, population structure, and relatedness. A, GWAS in the discovery data. B, GWAS in the replication data. C, Meta‐analysis of discovery and replication GWASs, adjusted for genomic inflation. AAO, age at onset; AD, Alzheimer's disease; GWAS, genome‐wide association study; LDSC, linkage disequilibrium score regression.

FIGURE 3

Support for heterogeneous meta‐analysis signals. A, Lead variant from meta‐analysis is shown as a purple diamond. Linkage disequilibrium information is specific to 1000 Genomes European samples. Gene labels were restricted to protein coding genes for 11q13.1. B, Minor allele frequency (MAF) variation across 1000 Genomes reference populations. Variants are presented as chromosome:position:reference allele:alternate allele. Reference populations are color coded as gold: admixed Americans; orange: African; magenta: East Asian; purple: European; blue: South Asian.

FIGURE 4

Change in effect size (β) estimates across GWAS models. Top: Contrast effect size estimates from the original GWAS with the first alternative GWAS adjusted for sex but not APOE in panel (A), and the difference between the two values in panel (B). Bottom: Contrast effect size estimates from the first alternative GWAS and the second that did not adjust for sex or APOE in panel (C), and the difference between the two in panel (D). All models adjust for population structure and relatedness. Results on chr19 are excluded to avoid cis effects due to APOE genotype. Dashed line: effect size comparable to the lead variant at 2q14.3 near BIN1. Dotted line: effect size comparable to the lead variant at 6p12.3 near CD2AP. APOE, apolipoprotein E; GWAS, genome‐wide association study.

FIGURE 5

Connections between AAO GWAS hits and AD biology. Detailed annotations are provided in Table S7 in supporting information. A, Lead variants from the AAO GWAS are linked to genes based on their position (“within gene”) and their association with gene expression (“eQTL”), protein expression (“pQTL”), gene splicing (“sQTL”), and gene‐promoter interaction (promoter capture Hi‐C, “PC‐HiC”) as reported in Open Targets Genetics. B, Genes are linked to AD biology based on the strength of their association with AD by the Alzheimer's Disease Sequencing Project's Gene Verification Committee (“AD GWAS”), evidence the gene is differentially expressed (“DEG in AD brain”) or the protein is differentially expressed (“DEP in AD brain”) based on the Accelerating Medicines Partnership Program for Alzheimer's Disease consortium work, and their nomination as a therapeutic target as reported in Agora. C, Distribution of prioritized genes within biological domains linked to AD as reported in Agora, restricted to those supported by 4+ genes. AAO, age at onset; AD, Alzheimer's disease; GWAS, genome‐wide association study.