Abstract
BACKGROUND: Simultaneous consideration of two neuropathological traits related to Alzheimer's disease (AD) has not been attempted in a genome-wide association study. METHODS: We conducted genome-wide pleiotropy analyses using association summary statistics from the Beecham et al. study (PLoS Genet 10:e1004606, 2014) for AD-related neuropathological traits, including neuritic plaque (NP), neurofibrillary tangle (NFT), and cerebral amyloid angiopathy (CAA). Significant findings were further examined by expression quantitative trait locus and differentially expressed gene analyses in AD vs. control brains using gene expression data. RESULTS: Genome-wide significant pleiotropic associations were observed for the joint model of NP and NFT (NP + NFT) with the single-nucleotide polymorphism (SNP) rs34487851 upstream of C2orf40 (alias ECRG4, P = 2.4 × 10(-8)) and for the joint model of NFT and CAA (NFT + CAA) with the HDAC9 SNP rs79524815 (P = 1.1 × 10(-8)). Gene-based testing revealed study-wide significant associations (P ≤ 2.0 × 10(-6)) for the NFT + CAA outcome with adjacent genes TRAPPC12, TRAPPC12-AS1, and ADI1. Risk alleles of proxy SNPs for rs79524815 were associated with significantly lower expression of HDAC9 in the brain (P = 3.0 × 10(-3)), and HDAC9 was significantly downregulated in subjects with AD compared with control subjects in the prefrontal (P = 7.9 × 10(-3)) and visual (P = 5.6 × 10(-4)) cortices. CONCLUSIONS: Our findings suggest that pleiotropy analysis is a useful approach to identifying novel genetic associations with complex diseases and their endophenotypes. Functional studies are needed to determine whether ECRG4 or HDAC9 is plausible as a therapeutic target.