| Objective:Bladder cancer(BC)is the common malignancy in urinary system,whose prognosis is undesirable.Recently,lncRNA have been found to play an important role in the tumorigenesis and development of tumors.This study aimed at screening the prognostic lncRNAs and clinical factors.By using a computational framework,we tried to develop a nomogram which combined the above factors and could effectively predict survival for BC patients.We also sought to explore the functional roles of these prognostic genes in the development of bladder cancer.1.The RNA expression profiles and clinical data for BC patients and normal control were downloaded from The Cancer Genome Atlas(TCGA)database.Differently expressed long non-coding RNAs(differently expressed lncRNAs,DELs)were screened by the "DESeq2" package of R.Then,we performed univariate Cox proportional hazards regression(CPHR)analysis and Kaplan-Meier analysis to identify candidate prognostic lncRNAs.2.In the training stage,prognostic lncRNAs were selected from a multivariate CPHR analysis(stepwise model)and integrated into a risk-score model based on their risk coefficients.Kaplan-Meier analysis and time-dependent receiver operating characteristic curve(ROC)were used to evaluate the prognostic value of this model.The predictive performance of the model was validated in the validation dataset.Next,multivariate CPHR analysis and stratified analysis were performed to test whether the risk-score model was independent from common clinical risk factors,such as age and TNM stage.3.We evaluated the prognostic value of the lncRNA risk-score model and common clinical risk factors by univariate and multivariate CPHR analysis.Using R software to visualize,a nomogram was constructed to predict the 3-year and 5-year survival for each BC patients.For internal validation,concordance index(C-index),calibration curves and time-dependent ROC curves were used to evaluate the discrimination and calibration abilities of the nomogram.4.In another independent BC dataset,quantitative real-time polymerase chain reaction,(qRT-PCR)was used to detect the expression of the prognostic lncRNAs between BC tissues and adjacent normal tissues.5.The functional enrichment analysis was used to reveal the potential biological processes and signaling pathway of the prognostic lncRNAs.According to its potential function and the expression level,we chose the best-fit prognostic lncRNA for further study.The specific siRNA was performed to silence the best-fit prognostic lncRNA in the BC cell lines.6.The wound-healing assays and Transwell assays were used to detect the cell migration and invasion after silenced or overexpressed the best-fit prognostic lncRNA.Western blot was then performed to explore how the best-fit prognostic lncRNA regulated the migration and invasion of BC cells.Results:1.We obtained the RNA expression profiles and clinical data of 414 BC patients and 19 normal control from the TCGA.826 DELs with |log2(fold change)|>2 and P<0.01 were screened by the "DESeq2" package of R.Through univariate CPHR analysis and Kaplan-Meier analysis,11 DELs were significantly associated with the OS of BC patients(all P<0.05)and considered as candidate prognostic lncRNAs.2.In the training set,we filtered candidate lncRNAs through multivariate CPHR analysis(stepwise model)and found three lncRNAs(RNF144A-AS1,AC019211.1 and ST8SIA6-AS1)with the largest likelihood ratios.The three lncRNAs were combined into a prognostic model:Risk Score=(0.228 × ExpressionRNF144A-AS1)+(0.436 × ExpressionAC019211.1)+(0.116 × ExpressionST8S1A6-AS1).Kaplan-Meier analysis showed that the model could distinguish high-risk patients,which had a shorter OS than low-risk patients(P=3.1E-04).The area under ROC curve(AUC)reached 0.703(95%confidence interval[CI]:0.593-0.814)at 3-year and 0.696(95%CI:0.563-0.829)at 5-year.3.In the validation dataset,the AUC of the 3-lncRNA risk-score model was 0.675(95%CI:0.593-0.759)at 3-year and 0.678(95%CI:0.576-0.781)at 5-year,indicating a great predictive performance.Multivariate CPHR analysis revealed that the model could be an independent risk factor(HR=1.856,P=0.002).A risk-stratified analysis showed the 3-lncRNA model was even able to distinguish high-risk from low-risk BC patients in the TNM stage-II,TNM stage-III and age>65 subgroups(all P<0.05).4.In addition to the 3-lncRNA risk-score model,univariate and multivariate CPHR analysis revealed that age and TNM stage were also the independent risk factors.A nomogram was then constructed by integrating above three factors using R software.Internal validation showed that the C-index was 0.682 for the training dataset and 0.688 for the validation dataset.The 3-year and 5-year survival probabilities predicted by the nomogram were close to actual probabilities.More importantly,the AUC of the nomogram at 3-year was 0.739(95%CI:0.663-0.818),superior to the 3-lncRNA model(AUC=0.675,95%CI:0.592-0.759),TNM stage(AUC=0.696,95%CI:0.618-0.775)and age(AUC=0.559,95%CI:0.469-0.649).5.In another independent BC dataset,three prognostic lncRNAs were significantly overexpressed in the bladder cancer tissues by using qRT-PCR(all P<0.05).Using correlation analysis,184 differently expressed mRNAs(DEMs)that co-expressed with at least one of the prognostic lncRNAs were obtained.GO and KEGG enrichment analysis indicated that the co-expressed DEMs were enriched in the extracellular structure organization and extracellular matrix organization,indicating the prognostic lncRNAs might be involve in the metastasis of BC.6.According to the fold-change of the prognostic lncRNA and the number of co-expressed DEMs,RNF144A-AS1 was chosen for further function study.The transfection of the specific siRNA could generate RNF144A-AS1 knockdown BC cell lines.Functional expression assays showed that the migration and invasion of BC cells were dramatically attenuated after knockdown RNF144A-AS1.Compared with the scrambled group,the expression of epithelial markers(E-cadherin and ZO-1)increased,while the expression of mesenchymal markers(N-cadherin and Vimentin)decreased in RNF 144A-AS1-knockdown group.Conclusions:1.In the present study,we used a rigorous computational framework to screen prognostic lncRNAs between BC patients and normal samples from TCGA.Using multivariate CPHR analysis,we identified a 3-lncRNA risk-score model that efficiently distinguished high-risk from low-risk BC patients.2.By combining the 3-lncRNA model and two conventional clinical factors(age and TNM stage),a prognostic nomogram was constructed to quantify the individual's 3-year and 5-year survival of OS for BC patients,which could potentially be used for individualized management of such patients.3.The prognostic lncRNAs might be involved in the extracellular matrix organization RNF 144A-AS1,one member of the 3-lncRNA model,were found to overexpressed in BC and could enhance the migration and invasion abilities of BC cells in vivo. |
PDF Full Text Request