| Background Skin malignant melanoma is a highly heterogeneous and highly aggressive malignant tumor,which progresses quickly and has a high fatality rate,which seriously endangers human health.The treatment of advanced melanoma mainly includes immune checkpoint blockade(ICB)and molecular targeted therapy such as BRAFi/MEKi.Patients with different risks of melanoma have different first-line treatment options.Patients with lower-risk advanced melanoma usually choose BRAFi/MEKi targeted therapy,while patients with higher-risk advanced melanoma usually choose ICB therapy.Therefore,accurately predicting the survival risk of patients is particularly critical to the choice of treatment options,but there is still a lack of effective models for predicting the survival risk of patients.How to accurately predict the survival risk of patients and choose a reasonable treatment plan is a difficult problem.Another problem in the treatment of melanoma is that different individuals have different responses to the same treatment,and each regimen has different sensitive or resistant populations.Inappropriate treatment plan will cause primary drug resistance,delay the patient's optimal treatment opportunity,and cause irreparable losses.How to accurately predict the patient's drug response and formulate a precise and individualized treatment plan is also a difficult problem.Another problem in the treatment of melanoma is that more than half of patients will develop primary or secondary drug resistance.Once a patient develops drug resistance,the existing treatments often have little effect.A key way to solve this problem is to find new molecular therapeutic targets.However,it is also a difficult problem to accurately identify effective molecular targets among thousands of molecules.Based on the above three problems,this research has three main purposes: 1.Construct a nomogram prediction model for predicting the risk of melanoma patients,and provide tools for realizing the risk stratification of melanoma patients.2.Identify the molecular markers related to the treatment response of melanoma,and provide a theoretical basis for formulating individualized treatment plans.3.Identify the key driver genes in the pathogenesis of melanoma and provide new molecular targets for the treatment of melanoma.Methods Part 1 The RNA sequencing data of 472 cases of skin melanoma patients were downloaded from the TCGA database(the Caner Genomic Atlas),and 449 melanoma patients with complete survival data were randomly divided into training set I(n=224)and validation set I(n=225).The RNA sequencing data and survival data of 214 melanoma patients were obtained from the GSE65904 data set of the GEO database(Gene Expression Omnibus),and 210 patients with complete survival data were used as an independent external test set.These data sets are used to identify and validate prognostic markers.Among the 472 melanoma patients from the TCGA database,379 patients had no missing values for their age,sex,overall survival,survival status,and clinical stage.The 379 patients were randomly assigned to training set II(n=189)and validation set II(n=190)to establish and verify a nomogram prognostic model that integrated clinical factors and prognostic markers.Univariate Cox regression analysis,LASSO regression analysis(Least Absolute Shrinkage and Selection Operator,LASSO),and multivariate Cox regression model were used to build predictive models.Area under the ROC curve(AUC),Concordanceindex(C-index),Calibration curve and P value(logrank test)were used for evaluating the predictive power of the prognostic model.Kaplan-Meier analysis and logrank test are used to evaluate the influence of certain factors(such as gene m RNA markers,model scores,tumor immune burden)on the prognosis.The nomogram is used to calculate the risk value of each patient's prognosis and realize risk stratification.Part 2 The immune score,stromal score,and tumor score of each melanoma patient from the TCGA database were downloaded from the ESTIMATE database.5559 human immune genes were downloaded from the Innate DB database.The R package limma and edge R were used to perform differential expression analysis on RNA expression data to calculate differentially expressed genes(DEGs).The selection criteria for DEGs was |log FC| > 1,P < 0.05.Weighted gene co-expression network analysis(WGCNA)was used to identify a subset of functionally interrelated genes based on melanoma RNA sequencing data,and correlate it with external traits(such as immune function,tumor mutation burden,etc.).JAVA software Gene Set Enrichment Analysis(GSEA,version 4.1.0)and the Database for Annotation,Visualization and Integrated Discovery(DAVID,version 6.8)were used to annotate the gene set and identify the gene setrelated physiological functions and signaling pathways.STRING(version 11.0)and JAVA software Cytoscape were used to construct protein-protein interaction network(PPI network).The RNA sequencing data of melanoma cells from the CCLE database(Cancer Cell Line Encyclopedia)and the cell drug sensitivity data from the GDSC database(Genomics of Drug Sensitivity in Cancer)were used to analyze the relationship between immune-related genes and response to multiple agents.Part 3 The tissue samples of human malignant melanoma came from the tissue bank of China-Japan Union Hospital of Jilin University.The human normal epithelial fibroblast cell line BJ1 was a gift from the Research Center of China-Japan Union Hospital of Jilin University.A375 and A874 were donated by the Molecular Biology Laboratory of Jilin University.RT-PCR was used to detect the relative expression of target genes in melanoma tissues and melanoma cell lines.si RNA was used to knock down the expression of target genes in melanoma cell lines.The scratch test was used to detect the migration ability of melanoma cells,and the Cell Counting Kit-8(CCK-8)test was used to detect the proliferation ability of melanoma cells.The TCGA database was used to analyze the clinical significance of target genes.Single-cell sequencing data was used to analyze the relationship between target genes and the immune microenvironment.Results: Part 1 Using univariate Cox regression analysis and LASSO regression analysis,four immune-related genes CLEC7 A,CLEC10A,HAPLN3 and HCP5 were identified from training set I,and a multivariate Cox regression model was used to construct prognostic markers for melanoma.The predictive ability of this marker performed well in the training set,validation set and test set.The AUC values for predicting the 5-year survival rate were 0.68(training set I),0.64(validation set I)and 0.64(test set),respectively.There was a significant difference in survival between the low-risk group and the high-risk group(P <0.05).The nomogram prognostic model that integrated prognostic markers and clinical characteristics(immune score,age,clinical stage,tumor status,Breslow depth and Clark grade)has a better ability to predict the survival of melanoma patients.The consistency indexes of the training set II and the validation set II were 0.853 and 0.736,respectively,and the AUC values were 0.862 and 0.832,which were both significantly higher than 0.5,showing good predictive ability.The prediction model calibration curves of the training set II and the validation set II almost overlapped with the 45° reference line,suggesting that there was a high consistency between the survival rate predicted by the model and the true survival rate.We calculated the risk score of each patient according to the nomogram,divided the training set ? and the validation set ? into the low-risk group and the high-risk group respectively,and used logrank test to analyze the survival of the two groups.The analysis results suggest that the low-risk group has a longer survival time than the high-risk group,and there were significant differences between the groups.These findings all prove that the nomogram prediction model established in this study has good predictive ability.Part 2 We calculated the tumor mutation burden(TMB)of each patient based on the somatic mutation data of melanoma patients.The TMB value of each patient has a large degree of dispersion,reflecting the high heterogeneity of melanoma.The TMB levels was significantly correlated with improved survival outcomes.FLNC,NEXN and TNNT3 were identified as hub genes related to TMB,and their ce RNA network was constructed at the same time,including 5 mi RNAs(HAS-mi R-590-3P,HAS-mi R-374B-5P,HAS-mi R-3127-5P,HAS-mi R-1913 and HAS-mi R-1291)and 31 lnc RNAs(FAM66C,MIAT,NR2F2AS1,etc.).The expression levels of FLNC,NEXN and TNNT3 can reflect the response of melanoma cells to a variety of drugs.Part 3 There are 1499 DEGs between normal skin and primary melanoma(GSE15605);595 DEGs between benign nevus and primary melanoma(GSE112509);209 DEGs between primary melanoma and metastatic melanoma(GSE7553).These three data sets have two common DEGs: AURKA and BUB1.Among them,BUB1 has a significant correlation with the clinical stage,Breslow depth,clarck score and prognosis of patients with melanoma.The methylation and copy number variation of BUB1 gene are significantly correlated with BUB1 expression.In vitro cell experiments found that knocking down BUB1 significantly inhibited the proliferation and migration of melanoma cells(A375 and A875).Single-cell sequencing results showed that the high expression of BUB1 can inhibit the number and function of CD8+ T cells in the tumor microenvironment.Further research results suggest that melanoma cell lines with high BUB1 expression were more responsive to multiple drugs.Conclusions: 1.The nomogram prediction model integrating the gene signature(CLEC7A,CLEC10 A,HAPLN3 and HCP5)and clinical information can effectively predict the survival risk of melanoma patients.2.TMB-related genes FLNC,NEXN and TNNT3 were closely related to the drug response of melanoma cells and were potential molecular markers for predicting the response of melanoma patients.3.BUB1 is an oncogene of melanoma,closely related to the proliferation of melanoma cells and CD8+ T cells,and is a potential molecular therapeutic target. |
PDF Full Text Request