NOVEL CANDIDATE AD-RISK LOCI IDENTIFIED THROUGH WHOLE EXOME SEQUENCING IN AFRICAN-AMERICANS

Alzheimer's disease (AD) is twice as prevalent in African-American individuals, yet most AD studies have been conducted in Caucasians. Whole exome sequence data from 112 African-American AD patients and 105 non-demented African-American elderly controls were tested for association with disease status, both at the single variant level using a linear regression model adjusted for appropriate covariates, and at the gene level using SKAT-O. The SKAT-O analysis was performed using default weights and various minor allele frequency (MAF) upper and lower bounds (none, MAF > 5%; 5% > MAF > 0.5%, MAF < 0.5%), and the test was only performed when 2 or more variants could be analyzed in each gene. We followed up in an independent set of African-American AD cases and controls (74 AD vs. 247 controls) associations that met any of the following criteria: 1) 25 variants that have the smallest AD association P-values; 2) AD association with P<0.05, prioritizing 25 variants with the largest CADD Phred score; 3) variants not observed in our controls, prioritizing 25 variants with the largest minor allele counts in ADs and that have a MAF<0.001 in public datasets; 4) variants within the gene that had the best P-value for gene-based association. We identified several variants with nominally significant associations with AD status. A non-synonymous variant in the gene encoding the junction-mediating and -regulatory protein (JMY) had the best P-value in the single variant analysis in the discovery cohort, and remained significant when the discovery and follow-up cohorts were analyzed jointly (p=0.004 and 0.0002, odds ratio=4.29 and 2.92, respectively). Interestingly, JMY has been determined to be a modulator of neuritogenesis. Follow-up of the gene-based associations are underway and will be presented. Employing whole exome sequencing in a cohort of African-Americans, we have identified associations with potentially functional coding variants at genes not previously associated with AD-risk. Our results warrant replication in independent cohorts, given the paucity of studies in African-Americans, and the need to identify risk factors that may be more relevant to AD in this minority population.