Multivariate Data Fusion Analysis For Neurologic Diseases

Alzheimer's disease(AD)is a progressive neurodegenerative disorder with characteristic neuropathological changes.It is the most common form of dementia in the elderly,which are embodied as memory loss and cognitive decline and the genetic risk factors for AD are not yet clear except APOE-e4 gene.Imaging genomics,a new burgeoning and interdisciplinary research field,can extract the image characteristic of gene polymorphism or gene expression by integrating and analyzing image data and genome data.Thus,early diagnostic method based on image features is promising.In this work,we mainly focused on the Imaging Genetics in Alzheimer's disease by relational analysis between Single Nucleotide Polymorphism(SNP)and multiple linear regression models which was constructed based on the the difference of brain gray matter volume in different cohorts.Using a voxel-based morphological analysis method,the brain gray matter volumes of one nomal control group(NC)and two test groups(Mild Cognitive Impairment,MCI and Alzheimer's disease,AD)were pairwise compared.The AD and MCI groups had significant atrophy in multiple brain regions compared to the NC group: hippocampus,parahippocampal gyrus,fusiform gyrus,and amygdala are always associated with gray matter volume decline in the progression of AD,which are the most severe atrophied brain area.However,there was a significant difference between the NC and MCI groups in Insula,but no significant difference was noted between MCI and AD groups.While there was a significant difference between the atrophied condition of some areas of occipital lobes,parietal lobes and cingulate gyrus,no significant difference was present between the NC and MCI groups.In addition,gray matter volume differences were found in frontal lobe only in comparison between MCI and AD groups.It is also found that the left brain atrophy of AD patients was slightly more severe than that of the right hemisphere while the atrophy of MCI patients was more symmetrical,which suggested that in the process of development from MCI to AD,the atrophy of the brain area is asymmetric,but relatively symmetrical for the whole brain.SNPs which are strongly related with the volume of gray matter in the brain region were found using single variable correlation analysis in which a multiple linear regression model was applied on the gene features and imaging features of brain region.And there were significant differences in the age,education level,the frequency of the APOEe4 allele,the level of clinical cognition and the volume of gray matter of three groups of NC,MCI and AD.The SPM12 tool kit was used to calculate the volume of gray matter in 16 key brain regions and the Plink software was used to select 506 SNPs of the first 30 genes from ALZGENE(FIELD SYNOPSIS OF GENETIC ASSOCIATION STUDIES IN AD the Alzheimer Gene Research Database,www.alzgene.org).A multiple linear regression model was constructed to analyze the correlation between the SNP loci and the gray matter volume of key brain areas in the whole cohorts,in which the size of gray matter in key brain regions as the outcome variable,the SNP locus as explanatory variable,and age,gender,education level,and APOEe4 status as covariates.we found that 11 SNP loci are still significantly associated with the key brain regions(bilateral hippocampus,bilateral amygdala,medial left temporal lobe,left parahippocampal gyrus,left entorhinal cortex,and right fusiform gyrus.)after FDR correction,and among them,three SNPs are strongly associated with multiple brain regions: rs405509 was significantly associated with left hippocampus,left amygdala,and left entorhinal cortex;rs9314349 was significantly associated with left medial temporal lobe and right fusiform gyrus;rs11218322 was significantly associated with left parahippocampal gyrus and right amygdala.