Gene Expression Profiles Identified Novel Urine Biomarkers for Diagnosis and Prognosis of High-Grade Bladder Urothelial Carcinoma

Abstract

Bladder urothelial carcinoma (BC) has been identified as one of the most common malignant neoplasm worldwide. High-grade bladder urothelial carcinoma (HGBC) is aggressive with a high risk of recurrence, progression, metastasis, and poor prognosis. Therefore, HGBC clinical management is still a challenge. We performed the present study to seek new urine biomarkers for HGBC and investigate how they promote HGBC progression and thus affect the prognosis based on large-scale sequencing data. We identified the overlapped differentially expressed genes (DEGs) by combining GSE68020 and The Cancer Genome Atlas (TCGA) datasets. Subsequent receiver operating characteristic (ROC) curves, Kaplan-Meier (KM) curves, and Cox regression were conducted to test the diagnostic and prognostic role of the hub genes. Chi-square test and logistic regression were carried out to analyze the associations between clinicopathologic characteristics and the hub genes. Ultimately, we performed gene set enrichment analysis (GSEA), protein-protein interaction (PPI) networks, and Bayesian networks (BNs) to explore the underlying mechanisms by which ECM1, CRYAB, CGNL1, and GPX3 are involved in tumor progression. Immunohistochemistry based on The Human Protein Atlas and quantitative real-time polymerase chain reaction based on urine samples confirmed the downregulation and diagnostic values of the hub genes in HGBC. In conclusion, our study indicated that CRYAB, CGNL1, ECM1, and GPX3 are potential urine biomarkers of HGBC. These four novel urine biomarkers will have attractive applications to provide new diagnostic methods, prognostic predictors and treatment targets for HGBC, which could improve the prognosis of HGBC patients, if validated by further experiments and larger prospective clinical trials.

Keywords: GEO; TCGA; biomarker; bladder urothelial cancer; diagnosis; prognosis; urine.

Figure 1

Workflow of this study and identification of differentially expressed genes (DEGs) based on GEO GSE68020 and TCGA BLCA datasets. (A) Workflow of this study; (B) Volcano plot for GSE68020 dataset; (C) Volcano plot for TCGA BLCA dataset; (D) Heat map for DEGs of TCGA BLCA dataset; and (E) Venn diagram for overlapped DEGs. Permissions to use the logo of GEO (Gene Expression Omnibus) have been obtained from the copyright holders of GEO. GEO, Gene Expression Omnibus; TCGA, The Cancer Genome Atlas; BLCA, Bladder Urothelial Carcinoma; KM, kaplan-Meier; ROC, receiver operating characteristic; GSEA, gene set enrichment analysis; PPI, protein-protein interaction; THPA, The Human Protein Atlas; qRT-PCR, quantitative real-time polymerase chain reaction.

Figure 2

Receiver operating characteristic (ROC) curves for diagnostic values of ECM1, GPX3, CRYAB, CGNL1, and CRNN. (A) ROC curves of diagnostic value for high-grade bladder urothelial carcinoma (HGBC) based on TCGA BLCA dataset; (B) ROC curves of diagnostic value for HGBC based on GSE68020 dataset; and (C) ROC curves for differential diagnosis between HGBC and low-grade bladder urothelial carcinoma (LGBC) based on TCGA BLCA dataset. AUC, area under the curve.

Figure 3

Kaplan-Meier (KM) curves for prognostic values of ECM1, GPX3, CRYAB, CGNL1, and CRNN. (A) Overall survival (OS) based on TCGA BLCA dataset; (B) OS based on high-grade bladder urothelial carcinoma in TCGA BLCA dataset; and (C) Disease free survival (DFS) based on Gene Expression Profiling Interactive Analysis (GEPIA).

Figure 4

Expression levels of ECM1, GPX3, CRYAB, CGNL1, and CRNN in different clinicopathologic characteristics. (A) Histological grade; (B) pN (pathological N) stage; and (C) UICC (Union for International Cancer Control) stage.

Figure 5

Gene set enrichment analysis (GSEA) analysis of ECM1, GPX3, CRYAB, and CGNL1. (A) KEGG pathway enrichment for ECM1; (B) Gene Ontology (GO) enrichment for ECM1; (C) KEGG pathway enrichment for GPX3; (D) GO enrichment for GPX3; (E) KEGG pathway enrichment for CRYAB; (F) GO enrichment for CRYAB; (G) KEGG pathway enrichment for CGNL1; and (H) GO enrichment for CGNL1. BP, biological process; CC, cell component; MF, molecular function; NES, Normalized Enrichment Score.

Figure 6

Protein-protein interaction (PPI) networks of four modules based on ECM1, GPX3, CRYAB, and CGNL1. (A) CRYAB module PPI network; (B) ECM1 module PPI network; (C) GPX3 module PPI network; and (D) CGNL1 module PPI network.

Figure 7

Bayesian networks (BNs) based on ECM1, GPX3, CRYAB, and CGNL1 with bladder urothelial carcinoma (BC). (A) Histological grade; (B) UICC (Union for International Cancer Control) stage; and (C) pN (pathological N) stage.

Figure 8

Immunohistochemistry from The Human Protein Atlas (THPA) confirmed the downregulation of GPX3 and ECM1 in high-grade bladder urothelial carcinoma (HGBC) tissues by Mann-Whitney U test (P < 0.05). (A) Mann-Whitney U test compared the staining scores of immunohistochemistry between normal bladder tissues and HGBC tissues; (B) Distributions of the four expression levels [negative (–), weakly positive (+), positive (++) and strongly positive (+ + +)] of immunohistochemistry; and (C) Expressions of the 5 genes in TCGA BLCA dataset.

Figure 9

qRT-PCR of urine samples from Tianjin validation cohort. Relative expressions of GPX3, ECM1, CRYAB, and CGNL1 are significantly lower in urine of high-grade bladder urothelial carcinoma (HGBC) patients than in controls (P < 0.05). (A) qRT-PCR of urines samples from controls, low-grade bladder urothelial carcinoma (LGBC) and HGBC; (B) Receiver operating characteristic (ROC) curves of diagnostic value for HGBC based on qRT-PCR; (C) ROC curves for differential diagnosis between HGBC and LGBC based on qRT-PCR; and (D) Expressions of the 5 genes in GSE68020 dataset. AUC, area under the curve.