Construction Of Survival Prediction Model Of Glioblastoma Based On Bioinformatics

Objective: Glioblastoma(GBM)is the most aggressive subtype of gliomas.It is highly malignant,aggressive and has a poor prognosis.The average survival time is 10-12 months.Considering that the overall survival rate(OS)of each GBM patient is a key factor for individual treatment,this study aims to analyze the gene expression profile of GBM,search for differential genes related to poor prognosis of GBM,and evaluate the risk of differential genes,so as to reveal genes related to tumor diagnosis and prognosis of GBM.Methods:Firstly,the glioblastoma dataset was downloaded from the Cancer Genome Atlas(TCGA)database,and the gene expression and clinical information of 174 patients were obtained.We used limma software package in R language to analyze the differential expression of gliomas and corresponding normal samples.Then,we used the online enrichment tool(http://enrich.shbio.com/)to enrich the function of the differentially expressed args,It includes the following items:1)Gene Ontology(go): molecular function(MF),cell composition(CC)and biological process(BP);2)Kyoto Encyclopedia of genes and genomes(KEGG).Then the survival of R /Bioconductor software was used for one-to-one univariate survival analysis.Firstly,Cox single factor analysis was used to analyze the differentially expressed genes,and the genes with P value less than 0.05 were selected.After that,multivariate regression was used to analyze the variables that would significantly affect the survival and prognosis of patients.The prognostic risk score model was established based on the following expression: risk score =expression level of Gene1 × β1 + expression level of Gene2 × β2 +…+expression level of Genen × βn;β is regression coefficient model calculated by multivariate Cox regression.Ubsequently,a prognostic risk score was generated for each patient.All TCGA GBM patients were divided into high risk(high risk score)and low risk(low risk score)groups according to the median of their risk score.At the same time,In order to verify the effectiveness of the model,the patients were randomly divided into the training group and the verification group.The number of the two groups was consistent.The training group was divided into high-risk group and low-risk group based on the risk level.Then,the corresponding software kits are used to determine the group boundary values,and the Kaplan Meier curves of the two groups are established based on the results.Then the curve was used to analyze the prognosis,and the survival difference between high-risk group and low-risk group was evaluated by bilateral log rank test.Based on the median risk score,160 patients were divided into high-risk group and low-risk group(survival analysis was conducted by K-M curve,and the overall survival rate was compared on this basis).After that,K-M survival analysis was conducted for the two groups of patients,and K-M survival curves were drawn for the comparison of OS between the two groups.Then,K-M survival analysis was performed on the screened genes to evaluate the relationship between the expression levels of prognostic genes and OS.The selected target genes were then included into the gene expression profile interaction analysis(GEPIA)database and the human protein map for verification.Finally,according to the selected gene expression level,the R software package "RMS" was used to generate a prognosis histogram to evaluate its contribution to the prognosis of patients.Results:We identified eight prognostic genes in the model,which have significant prognostic value.The results showed that TBX15,TGFBI,steam3,NELL1 and clec5 a were up-regulated in GBM,but best3 and HAS1 were down regulated in GBM.Then,the K-M survival curve was constructed to evaluate the association between the expression level of prognosis related genes and OS.The results showed that NELL1、TBX15、TMEM233、TGFBI、STEAP3、CLEC5A and HAS1,BEST3 high expression group(P < 0.05)had relatively good prognosis.Conclusion:The prognostic gene model composed of NELL1,TBX15,TMEM233,TGFBI,HAS1,STEAM3,BEST3 and CLEC5 8 genes constructed in this study can be used as prognostic markers for GBN patients,It can better predict the overall survival rate of patients,and has higher sensitivity than a single genome,All of these can be used as potential target genes for GBM therapy and provide a good reference for cancer researchers.