An EMT-related gene signature for the prognosis of human bladder cancer

Abstract

The transition from non-muscle-invasive bladder cancer (NMIBC) to muscle-invasive bladder cancer (MIBC) is detrimental to bladder cancer (BLCA) patients. Here, we aimed to study the underlying mechanism of the subtype transition. Gene set variation analysis (GSVA) revealed the epithelial-mesenchymal transition (EMT) signalling pathway with the most positive correlation in this transition. Then, we built a LASSO Cox regression model of an EMT-related gene signature in BLCA. The patients with high risk scores had significantly worse overall survival (OS) and disease-free survival (DFS) than those with low risk scores. The EMT-related gene signature also performed favourably in the accuracy of prognosis and in the subtype survival analysis. Univariate and multivariate Cox regression analyses demonstrated that the EMT-related gene signature, pathological N stage and age were independent prognostic factors for predicting survival in BLCA patients. Furthermore, the predictive nomogram model was able to effectively predict the outcome of BLCA patients by appropriately stratifying the risk score. In conclusion, we developed a novel EMT-related gene signature that has tumour-promoting effects, acts as a negative independent prognostic factor and might facilitate personalized counselling and treatment in BLCA.

Keywords: Cox; EMT; GEO; LASSO; TCGA; bladder cancer; signature.

Figure 1

The EMT signalling pathway is dramatically activated in the subtype transition from NMIBC to MIBC. A, GSVA of the http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32548, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32894 and http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48075 data sets. B, Venn diagram of the activated (a) and suppressed (b) gene sets in the indicated data sets. C, GSEA plot of ‘HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION’ of http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32548, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32894 and http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48075. D, KM survival analysis of NMIBC and MIBC patients in http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32548 and http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32894

Figure 2

KM survival analysis, risk score assessment by the EMT‐related gene signature and time‐dependent ROC curve in the TCGA‐BLCA training set. A, KM survival analysis of high‐ and low‐risk samples in the TCGA‐BLCA data set. B, Relationship between the survival status/risk score rank and survival time (days)/risk score rank. C, Time‐dependent ROC curve for OS of the TCGA‐BLCA data set. The AUC was assessed at 1, 3 and 5 y

Figure 3

KM survival analysis, risk score assessment by the EMT‐related gene signature and time‐dependent ROC curves in the OS validation data sets. A, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507, B, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48075. a. KM survival analysis of high‐ and low‐risk samples. b. Relationship between the survival status/risk score rank and survival time (days)/risk score rank. c. Time‐dependent ROC curve for overall survival of the validation data sets. The AUC was assessed at 1, 3 and 5 y

Figure 4

KM survival analysis, risk score assessment by the EMT‐related gene signature and time‐dependent ROC curves in the DFS validation data sets. A, TCGA‐BLCA, B, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13507, C, http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE32894. a. KM survival analysis of high‐ and low‐risk samples. b. Relationship between the survival status/risk score rank and survival time (days)/risk score rank. c. Time‐dependent ROC curve for overall survival of the validation data sets. The AUC was assessed at 1, 3 and 5 y

Figure 5

KM survival subgroup analysis of all patients with BLCA according to the EMT‐related gene signature stratified by clinical characteristics. A, Age ≤65 y. B, Age >65 y. C, Female. D, Male. E, Early stage (stage I/II). F, Late stage (stage III/IV). G, Non‐papillary. H, Papillary. I, N0. J, N+. K, Pathology T0‐2. L, Pathology T3‐4. M, Lymphovascular invasion −. N, Lymphovascular invasion +. O, Number of positive lymphonodes by HE = 0. P, Number of positive lymphonodes by HE > 0

Figure 6

Nomogram to predict 3‐ or 5‐year OS in the TCGA‐BLCA training set. A, Nomogram for predicting the 3‐ or 5‐year OS time in patients. B, Calibration curve for the prediction of 3‐ or 5‐year overall survival. C, DCA curve