Screening And Validation Of Potential Biomarkers For Atrial Fibrillation Based On Weighted Gene Co-expression Network Analysis

Background:Atrial fibrillation（AF）,is a global medical problem.The prevalence of AF among residents aged≥35years was 0.7%in China,the prevalence of AF among residents aged≥75 years was 2.4%.The prevalence of AF increases with age,and the disease burden of AF is a major public health problem.Weighted gene co-expression network analysis（WGCNA）is a systematic biology method.WGCNA can be used to identify potential genes or biomarkers by partitioning gene co-expression networks in complex biological processes into several characteristic modules and correlating feature modules with clinical traits.Mendelian randomization（MR）study is a method that use genetic variants strongly correlated with exposure factors as instrument variables（IV）to investigate the causal relationship between exposure and outcome,which can avoid the issues of causal timing and confounding factors in the traditional epidemiological study designs.In this study,the key genes associated with the risk of atrial fibrillation were screened by WGCNA method,and the MR study design was used to explore the causal association between the key gene（Apolipoprotein E,ApoE）and AF riskObjectives:In this study,the key gene（ApoE）related to AF risk was screened out based on the WGCNA method.Then,MR study was used to design the ApoE gene rs405509 locus as an instrumental variable to infer the association between ApoE levels and AF risk.Methods:1.Weighted gene co-expression network analysis method（WGCNA）:According to the GEO online database,the following words that“atrial fibrillation and homo sapiens”were used to screen the miRNA databases or gene databases related to atrial fibrillation（GSE68475 and GSE62871）.And the original data were read in R software,then,some preprocessing were performed,such as gene probe conversion and data normalization.Differentially expression analysis was performed using the LIMMA package,and differentially expressed genes（DEGs）were identified with|log₂FC|>0.5 and P<0.05 as the threshold.The gene modules of DEGs and miRNA modules of GSE62871 were constructed by the WGCNA software package,and the gene modules and miRNA modles associated with AF most significantly were selected as the key modules of AF for subsequent analysis.Candidate target genes of the miRNA modules were obtained using the Target Scan target prediction database and Mi RNet target predition database.Common genes were obtained by visual analysis of candidate target genes and the genes in gene modules using Veen plots.The STRING database was performed to construct a PPI network for common genes.The key genes were obtained by cyto Hubba,a Cytoscape plugin.The GO database,KEGG database and Fun Rich database were used for functional enrichment analysis of common genes.2.Mendelian randomization study:A total of 1157 patients with AF and 1206 control group were included for epidemiological investigation and blood sample collection.The serum ApoE level was determined by ELISA.Then,DNA was extracted from blood samples,and the RFLP-PCR was used to detect the polymorphism of the ApoE gene rs405509（G/T）.ApoE gene rs405509（G/T）was used as an instrumental variable,combined with serologically related phenotype ApoE level,to verify the association between ApoE genotype and ApoE level,ApoE genotype and atrial fibrillation,and then infer the causal association between ApoE levels and the risk of AF.Results:1.A total of 31 common genes were obtained by WGCNA.Ten key genes were obtained by PPI network analysis,including GABBR2,ApoE,CD68,C1QB,CIQC,SSTR2,SUCNR1,IL-13,SLC6A4,and ADAP2.Gene biological pathways were mainly enriched in biological pathways such as high-density lipoprotein mediated lipid transport,immune system regulation,chylomicron-mediated lipid transport,and activation of ion channels.2.Mendelian randomization studies（MR）included 2363 subjects,including 1157 atrial fibrillation cases and 1206 control subjects.The average level of ApoE in the case group（9.89±2.81 mg/d L）was higher than that in the control group（9.53±2.60 mg/d L）,and the difference in ApoE level between the two groups was statistically significant（P=0.018）.The association between the ApoE gene rs405509（G/T）locus mutation and the risk of atrial fibrillation was statistically significant,and the regression coefficient(β_ZY)of the association between the ApoE gene rs405509（G/T）locus mutation and atrial fibrillation was 0.231(OR=e^0.231=1.26,95%CI:1.035-1.533),which meant that compared with individuals with the TT genotype,an individual had one G mutation of the allele,the risk of AF increased by 26%.The correlation coefficient(β_ZX)between the rs405509（G/T）locus mutation of the ApoE gene and the ApoE level was 1.386（95%CI:1.024-1.749）,which meant that compared with individuals with the TT genotype,each additional G mutant allele（GT/GG）in an individual,ApoE levels increased by 1.386 mg/d L.The regression coefficient of the association between ApoE level and atrial fibrillation risk inferred by MR was 0.167(β_XY=β_ZY/β_ZX=0.231/1.386=0.167),which meant that after the natural exponential transformation,with the increased of the ApoE level by 1 mg/d L,subjects had a 18%increased risk of AF(OR=e^0.167=1.18).The association coefficient(β’_XY)between the ApoE level and the risk of AF obtained by traditional epidemiological methods was 0.057,indicating that with the increased risk of ApoE level by 1 mg/d L,subjects had a 6%increased risk of AF(OR=e^0.057=1.06,95%CI:1.015-1.104),which was slightly lower than that of MR study,suggesting that the causal association between increased ApoE levels and AF risk in MR studies was more reliable.Conclusions:1.Ten key genes were obtained by WGCNA,among which the expression products of GABBR2,ApoE,CD68 and C1QB genes may be potential biomarkers related to AF.2.The causal association between ApoE levels and atrial fibrillation risk was verified by Mendelian randomization study.