Development Of A Clinical Prediction Model Based On Tumor Cell Stemness And Construction Of Temozolomide-resistant Glioma Cell Lines And Their Characteristics Analysis

Background:Glioma is the most common primary intracranial tumor,among which glioblastoma(GBM)is the most lethal one with a very poor prognosis,and limited treatment options.Temozolomide(TMZ)is currently the only first-line chemotherapy agent.However,there are still many patients become resistant to TMZ.Glioma stem cell(GSC)is thought to play an important role in TMZ resistance of GBM.Thus,our study tried to carry out some word mainly in this direction.Methods:One-class logistic regression algorithm was performed on GBM patients from The Cancer Genome Atlas(TCGA)in this study,dividing them into high stemness subgroup and low stemness subgroup,and Kaplan-Meier survival analysis was used to determine their prognosis.Then,differentially expressed genes in the two subgroups were analyzed,and functional enrichment analyses were performed.By using data from GDSC database,we predict the sensitivity of the patients.Then,7 genes that can predict tumor stemness were figured out by machine learning and we used these 7 genes to establish a prediction model.Finally,the above results were verified by using data from The Chinese Glioma Genome Atlas(CGGA)and 38 GBM samples from Peking Union Medical College Hospital.Furthermore,to explore the mechanism of TMZ resistance in GBM,we treated two GBM cell lines,with gradient increasing concentration of TMZ.MTT assay was used to measure the half maximal inhibitory concentration(IC50)of the cells.The quantitative proteomics analysis,the RNA sequencing analysis and RNA m6A-immunoprecipitation were performed in these cells.Results:According to the mRNA stemness index,518 GBM patients in TCGA database were divided into high stemness subgroup and low stemness subgroup.It was found that the patients with high stemness had poorer PFS,but the OS was better.Then,we identified 130 differentially expressed genes among different subgroups and based on them,we further dividing all GBM patients into two different subgroups,subgroup Ⅰ and subgroup Ⅱ,finding that subgroup Ⅰ had the prognostic characteristics of patients with high stemness and subgroup Ⅱ had that of patients with low stemness and patients in subgroup Ⅰ were more prone to be resistance to TMZ and might have a better response to immunotherapy.Then,machine learning were used to find 7 genes which can distinguish these two subgroups of patients and a prediction model was constructed based on them,and CGGA and PUMCH data were used for the validation.Furthermore,we induced two TMZ-resistant GBM cell lines(U251TR and U118TR)by increasing TMZ concentration.And proteome and transcriptome sequencing and also RNA m6A modification were analysis.Conclusions:The stemness of tumor has a definite predictive value for the prognosis of GBM patients,and patients with different level of stemness have different sensitivity to TMZ.In this study,a prediction model of stemness level based on 7 related genes was developed,which can effectively distinguish patients with high stemness from patients with low stemness,so as to better guide clinical diagnosis and treatment.In order to further explore the mechanism of TMZ resistance,we successfully induced two TMZ resistant cell lines and analyzed the differences in proteome and transcriptome level,laying a foundation for further research.