P1‐050: Genome‐wide association study in different clinical stages of Alzheimer's disease

Previous genome-wide association (GWA) studies have considered endophenotypes to understand neuropathological mechanisms of Alzheimer's disease (AD). However, comparison of the influence of genetic susceptibility factors at pre-clinical and clinical stages of AD has not been fully investigated. We conducted GWA studies for cerebrospinal fluid(CSF) levels of amyloid beta-42(AB42), the tau/AB42 ratio, hippocampal volume(HPV), and logical memory test(LMT) in subjects with AD (n=190), mild cognitive impairment (MCI, n=581), and clinically normal (CN, n=305) from Alzheimer's Disease Neuroimaging Initiative(ADNI) study who were genotyped using two different platforms. A total of 8,082,822 genotyped and imputed single nucleotide polymorphisms (SNPs) were tested for association in a linear regression model using the normalized phenotypes after accounting for age, sex, and population substructure. GWA studies were performed separately for subjects genotyped on the same platform, and the results were combined by meta-analysis. Top-ranked SNPs(p<5x10−6) in meta-analysis were examined for potential function and enriched pathways using IPA software. Genome-wide significant (GWS, p<5x10−8) association was observed with LOC286114 and APOE for AB42 in the total sample. Subgroup analysis revealed GWS SNPs for AB42 from GRIN2B (rs74442473: p-value in meta-analysis[p] =3x10−8) and NACA2 (rs2378873: p=2x10−8), and FAM19A5 for AB42/tau (rs5769666: p=4x10−9) in CN subjects, for LMT from NRG1 in MCI subjects (rs10095844: p=2x10−9), and for HPV from AKAP9 in AD subjects (rs149454736: p=7x10−9). The GWS SNPs in GRIN2B, NACA2, FAM19A5, and NRG1 were significant cis-acting eSNPs (p<0.05) in hippocampus or frontal cortex. Pathway analysis demonstrated that top-ranked genes were involved in function of neuronal cells and significantly enriched in nNOS signaling, reelin signaling, and glutamate receptor signaling pathways (p<6.9x10−3). We identified association of several AD endophenotypes with genes not previously associated with AD (NACA2 and FAM19A5), as well as with several known AD risk genes (GRIN2B, NRG1, and AKAP9). Our results suggest that some AD-related traits are influenced by unique sets of genetic factors at different states of cognitive ability, but the collective set of genes are enriched for a role in neuron-specific pathways. Genes associated with AD-related phenotypes in pre-clinical (CN) or early stage (MCI) of AD may be potential early biomarkers for AD.