Multi-ancestry exome-wide study identifies variants associated with Alzheimer's disease protection

BackgroundPrevious whole exome and whole genome sequencing (WES/WGS) studies identified genome-wide significant associations for late-onset Alzheimer's disease (AD) with rare variants but highlighted the need for larger samples.ObjectiveIdentify associations of rare coding variants with AD risk in a large-scale, multi-ancestry exome-wide.MethodsWe combined non-overlapping portions of the Alzheimer's Disease Sequencing Project (ADSP) WES (n = 18 717) and WGS (n = 35 014) datasets obtaining a sample (n = 34 202) including participants ages ≥ 60 from four genomic similarity clusters consistent with European ancestry (EA, 9 744 AD cases and 9 095 controls), African American (AA, 1 944 AD cases and 4 215 controls), Caribbean Hispanic (CH 2 344 AD cases and 3 465 controls), and Native American Hispanic (NAH 743 AD cases and 2 652 AD controls) populations. Association of AD with 253,421 bi-allelic variants with minor allele count ≥ 20 in the total sample and each population group was evaluated using GENESIS. Gene-based tests comprising predicted moderate and high-impact variants were performed using SAIGE.ResultsNovel study-wide significant associations (p < 1.97 × 10^-7) were identified with variants enriched among the Amish in <i>BTBD8</i> (rs927193859, p = 1.58 x10^-10), <i>LINGO1</i> (rs150289554, p = 8.60 × 10^-8), and <i>KCNG2</i> (rs140218057, p = 1.77 × 10^-7) in the total sample. Population-specific analyses confirmed significant associations with <i>APOE</i> in all groups and detected an association in CH individuals with the <i>PSEN1</i> missense mutation G206A (p = 3.07 × 10^-7), a variant that was previously linked to early-onset AD in Hispanics. Gene-based tests highlighted significant associations with <i>LINGO1</i> (p = 2.75 × 10^-8), <i>DYNLT4</i> (p = 2.80 × 10^-7), and <i>ADCY4</i> (p = 7.43 × 10^-7).ConclusionsWe identified rare variants in novel genes which provide new insights into AD pathogenesis.