Identification And Validation Of A Metabolism-related Risk Signature And Molecular Classifications Associated With Clinical Prognosis And Therapy In Glioblastoma

Background and Objectives:Glioblastoma(GBM)is the most common and most invasive glioma of the central nervous system in adults with the worst prognosis.GBM itself is highly invasive and invasive.In addition,intra-tumor and inter-tumor heterogeneity leads to drug resistance in the treatment of GBM,which is an important factor hindering the successful treatment of GBM.GBM molecular markers and molecular classification are urgently needed to guide the precise treatment of GBM.Metabolism reprogramming is one of the characteristics of cancer cells,and metabolism abnormalities play an important role in the growth,proliferation,angiogenesis and invasion of cancer cells.Changes in glucose,lipid and glutamine metabolism are prominent characteristics of cancer cells.There is no comprehensive analysis of prognostic risk stratification and molecular classification of metabolism-related genes in GBM.The metabolism heterogeneity of GBM remains to be further explored.Methods:Transcriptome expression profile and clinical information of GBM patients were obtained from the Cancer Genome Atlas(TCGA),Chinese Glioma Genome Atlas(CGGA)and GSE13041.Glucose,lipid and glutamine metabolism-related genes were downloaded from Molecular Signatures Database v7.0(MSigDB)and defined as metabolism-related gene sets.TCGA data set was used as the training dataset,CGGA and GSE13041 were used as the validation datasets for integrated bioinformatics analysis.In addition,the main analysis databases include Connectivity Map(CMap)and Genomics of Drug Sensitivity in Cancer(GDSC).Analysis and statistics were performed on R platform using related R software packages.The first part is to identify and validate the metabolism gene risk scoring model related to GBM prognosis.Univariate Cox analysis was used to screen metabolism genes related to survival,and LASSO regression analysis was used to construct a prognostic risk scoring model.GBM patients were divided into metabolism high-and low-risk group according to the median risk score.Kaplan-meier(KM)method was used to draw KM survival curve to reveal the difference in overall survival(OS)between the high-and low-risk groups.Univariate Cox and multivariate Cox regression analysis were used to determine the prognostic risk factors of GBM and the prognostic independent predictive ability of metabolic risk score model.The differences of metabolism-related gene risk scores among GBM patients with different clinical characteristics and OS of patients with high and low risk scores within the same clinical trait were further evaluated.Receiver Operating characteristic(ROC)analysis was used to evaluate the efficacy of metabolism-related gene risk score in predicting survival of GBM.Gene Set Enrichment Analysis(GSEA)was used to identify the biological processes of GBM in different metabolism risk groups.The second part carried out GBM molecular classification identification and validation for metabolism-related genes.First,unsupervised consistent cluster analysis was performed on GBM based on metabolism-related genes to identify GBM metabolism classification and verify the subtypes using Principal Component analysis(PCA).The difference analysis of OS,metabolism gene risk score,tumor microenvironment(TME),tumor infiltrating immune cell(TIIC)and immune checkpoint(IC)among classifications were further investigated.,GDSC database was used to predict the sensitivity of different GBM classification to chemotherapy drugs,and CMap database was used to screen small molecule drugs targeting metabolism related classification.Results:(1)Based on the transcriptome expression profile of 1395 metabolism genes in GBM patients in TCGA/CGGA/GSE13041 dataset,we constructed a prognostic risk score model including 17 metabolism-related genes associated with GBM prognosis by univariate Cox regression analysis and LASSO regression machine learning method.KM analysis found that patients in the high-and low-risk groups had significantly different OS based on median risk score cutting-off value.Univariate and multivariate Cox regression analysis and ROC curves shower that the risk score model is an independent prognostic factor for GBM.Patients with different clinicopathological factors and molecular marker stratification presented significantly different risk scores,and metabolism risk score model is better than other GBM clinical pathologic factors and molecular markers in prognosis prediction and had stronger ability to predict OS.The above results were further validated in CGGA and GSE13041 datasets,and consistent results were obtained.Gene set enrichment analysis(GSEA)showed that glycolysis gluconeogenesis and oxidative phosphorylation(OxPhos)were significantly enriched in GBM in the high-and low-risk groups,respectively.(2)In the TCGA/CGGA/GSE13041 dataset,GBM was divided into three stable metabolism molecular subtypes by unsupervised consistent cluster analysis based on metabolism-related gene expression profiles.PCA verified the independence of the expression profiles of the three metabolism classifications,and there were different OS and metabolism risk scores among the three metabolism classifications.Further analysis showed that the immunophenotypes of TME,TIICs and IC were different among the three metabolism classifications.Combined with drug sensitivity analysis of GDSC database,three metabolism classifications showed different sensitivity to commonly used GBM chemotherapy drugs,and the classification with poor prognosis were more likely to benefit from erlotinib chemotherapy.Combined with CMap database,54 small molecule compounds targeting metabolism related classification that may have therapeutic effects on GBM were further screened.Conclusion:Based on the GBM metabolism gene expression profiles from multiple databases,we constructed and validated a prognostic risk scoring model based on GBM metabolism genes using machine learning methods,identified two energy metabolism patterns and three different metabolism molecular classifications in GBM,and revealed the metabolism heterogeneity of GBM from multidimensions.It provides a new perspective for risk stratification,molecular classification and precision therapy of GBM.The conclusions were as follows:(1)According to GBM metabolism gene expression profile,metabolism-gene prognostic risk score model could divide GBM into metabolism high risk group and low risk group.The model was an independent survival predictor of GBM and had excellent survival prediction efficacy.(2)There were two different energy metabolism patterns in GBM,the high-risk group was dominated by glycolysis and gluconeogenesis,while the low-risk group was more inclined to OxPhos.(3)GBM had three different metabolism molecular classifications,different classifications had different OS,TME,TIICs,ICs expression,risk scores and sensitivity to different chemotherapy drugs.GBM metabolism classifications with poor prognosis were more sensitive to erlotinib,temozolomide,vincristine,carmustine and rapamycin.In particular,metabolism classification with the worst prognosis may benefit from erlotinib chemotherapy.(4)Drugs targeting energy metabolism may have anti-GBM effects,such as amitriptyline,fluoxetine,imipramine,etc.