Progressively exploring and assessing the prognosis of bladder urothelial cancer based on the microenvironment through the integration of multiple databases

Abstract

Background: The heterogeneous prognosis of bladder urothelial carcinoma (BLCA) remains a significant clinical challenge. A multi-factor prognostic model is essential for BLCA, as it not only assesses tumor progression and elucidates underlying molecular mechanisms but also paves the way for timely treatment adjustments and improved clinical decision-making.

Methods: Using R software, we performed immunophenotyping on multiple BLCA cohorts from the GEO database to identify shared immune signatures. Simultaneously, we identified BLCA prognosis-associated genes by analyzing TCGA data. Prognostic genes were further refined via LASSO regression, allowing BLCA patients to be stratified into high- and low-risk groups based on their expression patterns. Quantitative PCR (qPCR) was used to validate gene expression in tumor and matched normal tissues. Finally, we integrated clinical data to construct a prognostic model.

Results: The GSE31684 and GSE48276 cohorts were divided into high immunity (Immunity_H) and low immunity (Immunity_L) groups, and there were significant microenvironment differences between the Immunity_H and Immunity_L of the two cohorts, and there were many common differentially expressed genes (DEGs) between different immune subtypes of the two cohorts, which were mainly involved in immune-related biological processes. In addition, patients in the high-risk BLCA group exhibited significantly worse prognosis than those in the low-risk group. qPCR analysis confirmed that the expression levels of the risk-stratification genes were significantly different between BLCA tumors and matched adjacent normal tissues. The integrated analysis of tumor mutation burden (TMB) and our risk stratification revealed that patients with low-risk scores and high TMB exhibited the most favorable prognosis. Furthermore, the risk score was validated as an independent prognostic factor through both univariate and multivariate Cox regression analyses. Consequently, we constructed a nomogram that incorporates these findings to assist clinicians in prognostic assessment for BLCA patients.

Conclusion: Given that the tumor microenvironment significantly influences BLCA prognosis, our finding that risk stratification serves as an independent prognostic indicator underscores the clinical relevance of our model. This stratification strategy has the potential to improve prognostic assessment and inform personalized treatment planning for BLCA patients.

Keywords: bladder urothelial carcinoma (BLCA); immune subtypes; microenvironment; prognosis; risk stratification.

FIGURE 1

Two different immunophenotypes in BLCA patients. (A,B) Based on ssGSEA results, BLCA patients from GSE31684 (A) and GSE48276 (B) were stratified into Immunity_H and Immunity_L. (C,D) The reliability of BLCA immunophenotypes derived from GSE31684 (C) and GSE48276 (D) cohorts was verified by tSNE.

FIGURE 2

Multilevel comparison of the microenvironment among different immune subtypes of BLCA. (A,B) Comparison of StromalScore, ImmuneScore, and ESTIMATEScore for different immune subtypes from the GSE31684 (A) and GSE48276 (B) cohorts. (C,D) Comparison of HLA gene expression levels from different immune subtypes in GSE31684 (C) and GSE48276 (D) cohorts. (E,F) Microenvironmental landscapes of different immune subtypes in BLCA patients from the GSE31684 (E) and GSE48276 (F) cohorts. *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 3

Comparison of DEGs among different immune subtypes of BLCA. (A,B) Volcano plots displaying the DEGs for different subtypes from the GSE31684 (A) and GSE48276 (B) cohorts. (C,D) A Venn diagram depicting the DEGs that are consistently upregulated (C) or downregulated (D) in the GSE31684 and GSE48276 cohorts (Immunity_L as the control group).

FIGURE 4

Identification of PRGs for BLCA based on TCGA and GEO data, along with the construction of their regulatory network. (A) The forest atlas showed common DEGs in the GSE31684 and GSE48276 cohorts that affected the survival outcomes of TCGA-BLCA patients. (B) A river interaction diagram showcasing the relationships between PRGs and TFs. (C) GO analysis revealed the main biological processes involved in PRGs.

FIGURE 5

Determination and evaluation of risk assessment for patients with BLCA. (A,B) LASSO coefficient profiles (A) and cross-validation plot (B) for gene parameter selection within a lasso model of BLCA. (C) A distribution plot of risk scores for BLCA patients based on 30 genes identified through the lasso model. (D) Correlation analysis between OS and risk scores in patients with BLCA. (E) Kaplan-Meier survival analysis for BLCA across different risk groups. (F) The ROC analysis based on the lasso model yielded AUC values for 1-year, 3-year, and 5-year survival predictions of 0.807, 0.786, and 0.791, respectively. (G) A calibration plot for the risk characteristics of bladder cancer (BLCA) based on genes selected through the lasso model.

FIGURE 6

The impact of the expression levels of genes used for risk stratification on the prognosis of BLCA. (A–C) The influence of the expression levels of the three genes with the highest coefficients—ADCY7 (A), SLC1A6 (B), and NELL2 (C)—on the prognosis of BLCA. (D–F) The influence of the expression levels of the three genes with the smallest coefficients—ITGB7 (D), ZNF823 (E), and CTLA4 (F)—on the prognosis of BLCA.

FIGURE 7

The expression of PRGs in BLCA and adjacent normal tissues. (A–F) The expression of ADCY7 (A), SLC1A6 (B), NELL2 (C), ZNF823 (D), ITGB7 (E) and CTLA4 (F) in BLCA and adjacent tissues in the UALCAN database. (G–L) The qPCR results confirmed the expression of ADCY7 (G), SLC1A6 (H), NELL2 (I), ZNF823 (J), ITGB7 (K) and CTLA4 (L) in BLCA and adjacent tissues. *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 8

Joint analysis of risk stratification and TMB in BLCA. (A) Comparison of TMB across different risk groups in BLCA. (B) Impact of TMB on BLCA prognosis. (C) Evaluating the prognosis of BLCA by combining TMB and risk stratification.

FIGURE 9

Univariate and multivariate analyses of the impact of risk stratification on BLCA, along with the establishment and evaluation of a nomogram model. (A) Analysis of risk stratification and other variables in BLCA using univariate cox regression. (B) Analysis of risk stratification and other variables in BLCA using multivariate cox regression. (C) Construction of a nomogram based on clinical features and risk stratification for BLCA.(D) Assessment of the reliability of the nomogram through ROC analysis. (E) Map of calibration used to compare the nomogram to the ideal model for similarity assessment. *p < 0.05; **p < 0.01; ***p < 0.001.