Construction Of MiRNA Prognostic Model For Nasopharyngeal Carcinoma Based On GEO Database

Objective:In this study,we will construct the prognostic risk model of NPC based on micro RNAs(miRNAs)which express differentially between NPC tissues and normal tissues using bioinformatics,and further predict the related target genes,so as to explore the pathways of the pathogenesis of nasopharyngeal carcinoma and provide direction and theoretical basis for future researches.Methods:Data sets of miRNA expression in nasopharyngeal carcinoma and clinical follow-up information were downloaded from Gene Expression Omnibus(GEO)official website.limma package of R language was used to screen differentially expressed miRNA with the threshold value(FDR < 0.05 and | log2 FC | > 1.5).GSE32960 data set was randomly divided into training set and validation set.Univariate Cox regression analysis was performed on the training set to screen out candidate genes,and then the target miRNA were further screen by Least Absolute Shrinkage and Selection Operator(LASSO)regression and step-by-step regression analysis.With operations mentioned above,a prognostic risk model for nasopharyngeal carcinoma patients was constructed.The Receive operating characteristic curve(ROC)and Kaplan-Meier(K-M)survival analysis were used to evaluate the prognostic risk model and validate internal and external data sets.Then,univariate and multivariate Cox regression analysis was performed on the clinical characteristics and prognostic risk model to evaluate whether the risk score calculated by the model could be an independent prognostic factor.According to the results of univariate and multivariate analysis,the clinical features and models were used to construct the histogram model and the Decision curve analysis(DCA)diagram of the Normogram.Finally,10 tools,including Micro T,miRanda and miRcode,were used for the prediction of the target genes of differentially expressed miRNAs.Then,protein-protein interaction(PPI)analysis and Kyoto Encyclopedia of Genes and Gen omes(KEGG)analysis of target genes were performed by Network Analyst 3.0 and STRING.In addition,in order to verify the prognostic evaluation efficiency of the model in the patient population,paraffin tissue samples from 6 patients with nasopharyngeal carcinoma and 6 normal healthy people were collected to verify the expression of miRNA in the mode.Results:1.In this study,160 differentially expressed miRNAs were screened based on the GSE32960 dataset.With univariate Cox regression analysis,11 miRNAs were screened out(P < 0.05).2.Five miRNAs obtained from LASSO regression analysis and stepwise regression were used to construct a prognostic risk model for nasopharyngeal carcinoma.The ROC curve and K-M survival analysis results proved that the miRNA prognostic model had good predictive efficacy in different data sets,and thereinto the prognosis of the high-risk group was worse than that of the low-risk group(P < 0.05).3.According to univariate and multivariate Cox regression analysis results,this model can be used as an independent prognostic factor,and the Normogram is constructed based on age,N stage,metastatic,relapse and risk score,which proves that this method has good predictive performance.The results of DCA chart showed that the Normogram and prognostic risk model had good predictive effect.4.In the prediction on target genes of 5 miRNA,13 target genes were obtained.Two subnetworks were obtained by performing the PPI function analysis and KEGG analysis of target genes.5.The results of the validation experiment showed that the expression levels of hsa-let-7e-5p and hsa-miR-93-3p in NPC patients group showed up-regulated expression which were higher than those in the normal control group,showing tissue differences.It is verified that the model has certain predictive value.Conclusion:Based on GEO database,a prognostic risk model composed of five miRNAs was constructed,which could better predict the survival and prognosis of patients with NPC providing certain reference value for the prognosis evaluation of patients with NPC,and high-risk groups screening,so as to guide the development of individualized treatment plans.