Construction Of A Multi-omics Based Survival And Prognosis Model For Clear Cell Renal Cell Carcinoma

ObjectiveClear cell renal cell carcinoma is a malignant tumor of the urinary system.At present,the traditional prognostic model has some limitations,such as TNM staging method,single omics prognostic model and so on.In this study,multi-omics data such as full transcripts,epigenetics and functional genomic data of cc RCC patients were integrated,and clinical bioinformatics analysis methods were used to screen multi-dimensional biomarkers that were significantly related to the survival and prognosis of cc RCC patients,to construct a multi-omics risk prognosis model of cc RCC patients,and to draw a diagram of clinical quantifiable use,so as to provide guidance for laboratory basic research,clinical treatment and services of cc RCC.MethodsUse the keyword "Clear Cell Renal Cell Carcinoma" to retrieve the relevant data in the TCGA database.On the basis of integrating the full transcripts of patients,the differentially expressed mi RNA,lnc RNA and m RNA were obtained by differential expression analysis.Then,the relationship files of mi RNA,lnc RNA and m RNA were constructed based on mi Rcode,mi RDB,mi RTar Base and Target Scan databases,and Cytoscape was used to construct the ce RNA network.Cox regression analysis and Lasso regression analysis were used to screen nc RNA associated with patient survival and construct a prognostic nc RNA risk prognosis model.Subsequently,on the basis of integrating patients’ DNA methylation levels and total transcriptome data,differential methylation driver genes were screened by Methyl Mix and their gene ontology functions were examined by GO enrichment analysis.Cox regression analysis and Lasso regression analysis were used to screen the methylation driver genes associated with survival and construct a prognostic risk model of prognostic methylation driver genes.Then we integrated the expression profile of m6 A methylation regulatory factor.After differential expression analysis,tumor molecular typing analysis was used to verify whether m6 A methylation regulatory factor could reflect patient survival at transcriptome level.Cox regression analysis and Lasso regression analysis were used to screen m6 A methylation regulatory factors related to patient survival and construct a prognostic m6 A methylation regulatory factor risk prognosis model.Finally,we integrated all the above data,and constructed a multi-omics combined risk prognosis model and drew a nomogram after removing multicollinearity characteristic genes.Cox risk regression analysis,Kaplan-Meier survival curve,log-rank test and ROC analysis were used to evaluate the predictive accuracy of risk scores of each model and whether they could be used as independent prognostic risk factors compared with other clinical characteristics of patients.ResultsA total of 2322 differential m RNAs,1483 differential lnc RNAs,and 173 differential mi RNAs were obtained by whole-transcriptional differential analysis of cc RCC patients.There were 198 nodes and 618 edges in the ce RNA network.Cox and Lasso regression analysis showed that there were 16 nc RNAs significantly correlated with patient survival.Combined with DNA methylation data difference analysis,31 methylation driver genes were screened and obtained.GO enrichment analysis showed that methylation driver genes were mainly involved in the regulation of immune cell differentiation and proliferation,interferon production,positive regulation of interleukin-1 and other biological processes.Cox and Lasso regression analysis showed that 8 methylation driver genes were significantly correlated with survival.Differential analysis of expression profiles of m6 A methylation regulatory factors showed that 18 m6 A methylation regulatory factors were differentially expressed in cc RCC.Tumor molecular typing showed that cc RCC patients were divided into 3 groups,and the survival of patients in each group was significantly different(P<0.05).Cox and Lasso regression analysis showed that there were 5 m6 A methylation regulatory factors significantly correlated with survival.On the basis of normalization,the above data were integrated to build a multi-omics combined risk prognosis model with 19 characteristic genes.Respectively,LINC00887,PPP1R18,HHLA2,MAL,MYH14,TMEM173,HIST3H2 A,PPP1R36,HSA-Mir-21,HSA-Mir-223,LINC00460,ARHGAP26-AS1,LINC00443 and HO TTIP,PSORS1C3,HNRNPA2B1,IGF2BP2,IGF2BP3,EIF3 A.Multivariate Cox risk regression analysis,Kaplan-Meier survival curve and log-rank test showed that the above four risk prognostic models could distinguish patients with high and low risk of cc RCC,and the risk scores of each model could be used as independent predictors of patients’ prognostic risk compared with other clinical characteristics.ROC analysis showed that the AUC value of the multi-omics combined prognostic model was the highest(AUC=0.85),followed by the nc RNA risk prognostic model(AUC=0.788),followed by the methylation drive gene risk prognostic model(AUC=0.743).The AUC value of m6 A methylation regulatory factor risk prognosis model was the lowest(AUC=0.703).ConclusionThe 19 characteristic genes of the multi-omics combined risk prognosis model were closely related to the survival prognosis of cc RCC patients,and could be used as multi-omics biomarkers of cc RCC occurrence,development,metastasis,drug resistance and prognosis.