TARGETED MYELINATION PATHWAY-BASED GENE ANALYSIS IDENTIFIES RXRA ASSOCIATED WITH WHITE MATTER CHANGES IN MILD COGNITIVE IMPAIRMENT AND ALZHEIMER'S DISEASE

Emrin Horgusluoglu-Moloch,Gaoyu Xiao,Minghui Wang,Bin Zhang

DOI: https://doi.org/10.1016/j.jalz.2019.06.695

2019-01-01

Abstract:Diffusion tensor imaging (DTI) is a very useful neuroimaging technique to detect the abnormal changes in neuron fibers at the micro-structural level in the diagnosis of mild cognitive impairment (MCI) and Alzheimer's Disease (AD). Fractional anisotropy (FA), the measurement of fiber integrity based on DTI, have been demonstrated to be strongly correlated with WM damage and myelination. Recently, we identified top ten regions including sagittal stratum, hippocampus, cingulum, parahippocampal gyrus based on the strength of association between region-specific DTI features and disease status (Alzheimer's&Dement. 2018 with minor revision). For the identification of genetic mechanism underlying WM abnormalities we performed a gene-based association analysis of myelination-pathway related candidate genes with white matter microstructural changes. Myelination-related gene sets (N=49) were created from the existing MSigDB database. Gene-based association analysis was performed using 115 individuals (control=27, significant memory concern= 20, mild cognitive impairment=49, AD=19) from Alzheimer's Disease Neuroimaging Initiative (ADNIGO/2) who have both DTI scans and genotyping data. Two years (baseline-24 month) longitudunal changes of DTI-derived FA feature in sagittal stratum were used as phenotypes, and gender, age, and diagnosis at scanning time used as covariates. For construction of the RXRA centered co-expression gene network we assembled 6 gene expression datasets (the Mount Sinai Brain Bank, ROSMAP and MAYO) from 6 different brain regions. RXRA (Retinoid X Receptor Alpha) gene was most significantly associated with longitudunal changes of DTI-derived FA feature after Bonferroni correction (p-value < 0.001; Table 1) in the saggital stratum. The most significant SNP (rs34133915) in RXRA was most significantly associated with two years changes of fractional anisotropy in the saggital stratum (Figure 1A), hippocampus, cingulum as well as Montreal Cognitive Assessment (MoCA) memory score (Figure 1B). To understand the functional context of RXRA operation in AD pathogenesis, we identified a RXRA centered consensus gene coexpression network. This subnetwork was enriched for Alzheimer's pathway, axon guidance, synaptic signaling and oligodendrocyte differentiation in all six datasets (Figure 2).