You are viewing the site in preview mode

Skip to main content
Download PDF

Prognostic feature based on androgen-responsive genes in bladder cancer and screening for potential targeted drugs

BioData Mining volume 17, Article number: 59 (2024) Cite this article

Abstract

Objectives

Bladder cancer (BLCA) is a tumor that affects men more than women. The biological function and prognostic value of androgen-responsive genes (ARGs) in BLCA are currently unknown. To address this, we established an androgen signature to determine the prognosis of BLCA.

Methods

Sequencing data for BLCA from the TCGA and GEO datasets were used for research. The tumor microenvironment (TME) was measured using Cibersort and ssGSEA. Prognosis-related genes were identified and a risk score model was constructed using univariate Cox regression, LASSO regression, and multivariate Cox regression. Drug sensitivity analysis was performed using Genomics of drug sensitivity in cancer (GDSC). Real-time quantitative PCR was performed to assess the expression of representative genes in clinical samples.

Results

ARGs (especially the CDK6, FADS1, PGM3, SCD, PTK2B, and TPD52) might regulate the progression of BLCA. The different expression patterns of ARGs may lead to different immune cell infiltration. The risk model indicates that patients with higher risk scores have a poorer prognosis, more stromal infiltration, and an enrichment of biological functions. Single-cell RNA analysis, bulk RNA data, and PCR analysis support the reliability of this risk model, and a nomogram was also established for clinical use. Drug prediction analysis showed that high-risk patients had a better response to fludarabine, AZD8186, and carmustine.

Conclusion

ARGs played an important role in the progression, immune infiltration, and prognosis of BLCA. The ARGs model has high accuracy in predicting the prognosis of BLCA patients and provides more effective medication guidelines.

Peer Review reports

Introduction

Bladder cancer (BLCA) has a significantly higher global mortality and morbidity rate, with approximately 380,000 new cases diagnosed and 150,000 deaths annually [1]. The majority of BLCA are urothelial carcinomas [2]. Men are at a much higher risk of developing BLCA than women, with a difference in incidence of approximately tenfold [1,2,3,4]. Occupational exposures and smoking may contribute to the difference in BLCA incidence between genders [5, 6]. Although environmental and lifestyle factors have been considered, men are still three to four times more likely to develop BLCA than women [5,6,7]. Thus, intrinsic factors that are specific to sex differences may play a crucial role in BLCA. Several studies have shown that the androgen receptor signaling pathway could affect the development and progression of BLCA [8]. Additionally, postmenopausal women have a higher incidence of BLCA compared to premenopausal women. This suggests that the androgen receptor signaling pathway may be involved in BLCA development [9].

Androgens are steroid hormones secreted by various organs, including the testes, ovaries, adrenal cortex, adipose tissue, skin, and endometrium [10, 11]. The testis can convert androgens into dihydrotestosterone, a pure androgenic steroid found exclusively in men. This conversion is carried out through the action of 5α-reductase [12]. Androgens bind to the androgen responsor (AR) in target cells to produce physiological effects [13]. The androgen receptor (AR) is a transcription factor that belongs to the steroid hormone receptor superfamily and is activated by ligands. It plays a role in target cell differentiation and proliferation [13, 14]. Several studies have confirmed a strong link between androgens, AR, and the development and occurrence of BLCA [11,12,13,14,15]. Studies have shown that androgen and AR signals can promote the proliferation of BLCA cells. Androgen blockade, AR silencing, or antagonism can inhibit BLCA growth and metastasis [16,17,18]. Studies suggest that patients with BLCA who are AR-positive may have higher rates of recurrence and metastasis than those who are AR-negative [19, 20]. Previous studies suggest that AR may have an important impact on the prognosis of BLCA [16,17,18,19,20]. However, there is not enough research on the role and prognosis of androgen-responsive genes (ARGs) in BLCA. Investigating the function of ARGs in BLCA may provide new insights and strategies for predicting prognosis and targeting treatment in BLCA.

In this study, we screened for ARGs in BLCA and examined potential subtypes implicated in the disease. We analyzed the relationship between ARGs and various types of BLCAs in the tumor microenvironment. Additionally, we developed a risk score model using screened ARGs to predict patient prognosis and response to chemotherapy. The model’s credibility and rationality were strengthened by using external data. The study used bioinformatics techniques to investigate the expression and function of ARGs in BLCA. The aim was to better understand the sex differences in BLCA and to develop new markers and therapies for the BLCA.

Methods

Data collection

RNA-seq transcriptome expression profiles and clinical information for 405 BLCA patients were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/). We obtained the transcripts per million (TPM) expression profile and the data was transformed into Log2 (expression value + 1). In addition, 101 ARGs were obtained from the MsigDB database (https://www.gsea-msigdb.org/gsea/msigdb) to investigate the potential involvement of ARGs in BLCA. To ensure the robustness of the results, we obtained the GSE13507 dataset from the Gene Expression Omnibus (GEO) database as external data for cross-validation. The dataset consisted of 165 primary BLCA samples. We also downloaded copy number variations (CNVs) from the TCGA database.

Characterization screening of androgen response genes

We performed univariate Cox regression analysis to determine the ARGs associated with clinical prognosis and to examine the interaction between different ARGs with p value less than 0.05. To perform dimensionality reduction clustering on the data, we used the non-negative matrix decomposition algorithm. From the results of the consistent clustering, we proceeded with Principal Component Analysis (PCA) to confirm the segmentation of the ARGs into distinct clusters. The clinical prognosis of the different clusters was investigated by Kaplan-Meier survival analysis using the R package ‘survminer’. Genetic variation of the 32 ARGs was also analyzed using the R package.

Construction of prognostic models

The prognostic value of genes was assessed by calculating univariate Cox regression coefficients. Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis was then performed to identify additional genes associated with predicting prognosis and survival in BLCA. From the features identified in the LASSO analysis, the optimal prognostic risk score model was constructed by applying regression coefficients through multivariate Cox regression. To differentiate the training and test sets, the TCGA dataset was systematically randomized in a 1:1 ratio. Within-group validation of survival curves was used to assess the reliability of the risk score model on both the training and test sets. The survival rates of patients with BLCA at 1, 3, and 5 years were predicted using the training and test sets. The combination of risk scores and clinical characteristics was used to obtain clinical data on BLCA survival and prognosis. Cross-validation was performed with the external GSE13507 dataset to further assess the reliability of the risk-scoring model.

Enrichment and survival analysis

To determine the likely biological significance of the ARGs in the subtypes, we performed a KEGG enrichment analysis using the R package ClusterProfiler (p < 0.05). We also performed single-sample gene set enrichment analysis (ssGSEA) using the GSVA package. To elucidate the biological significance of ARGs’ risk score groups with different prognoses, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. Kaplan-Meier survival curves were constructed to observe differences in survival between the divided groups in different datasets. The predictive value of our model was validated by receiver operating characteristic (ROC) analysis, in which we calculated the area under the curve (AUC). Cox regression analysis was performed on clinical characteristics and risk scores to determine whether the risk score could independently predict survival prognosis.

Construction and characteristics of the risk scoring model nomogram

The nomogram was created using the R package “RMS” by combining prognostic and clinical features of BLCA. The risk-scoring models and nomograms were evaluated using calibration curves and time-dependent ROC curves. To illustrate the cumulative probability curve of BLCA incidence, we analyzed the risk difference between high-risk and low-risk groups using the R package time ROC.

ScRNA-seq data analysis

To obtain a thorough understanding of the expression of the six genes in cells associated with bladder cancer, we obtained single-cell RNA sequencing data for bladder cancer from the GEO database for external validation. The GSE135337 dataset for BLCA comprises 7 primary tumor samples and 1 para-cancerous tissue sample. This analysis was executed using the Seurat package (version 3.1.1) in R. The resulting Seurat objects underwent data quality control and standardization, screening for hypervariable genes, clustering with cell dimensionality reduction, and identification of differentially expressed genes between clusters. Subsequently, the cell types were annotated according to the original text annotation method [15].

Tumor immune microenvironment and drug sensitivity analysis

BLCA tumor purity, stromal score, immune score, and ESTIMATE score were calculated using the R package ESTIMATE. The proportion of immune cells in the sample was determined using the R package ‘CIBERSORT’. The ssGSEA algorithm was then used to investigate the level of immune infiltration between high-risk and low-risk groups based on different immune cell types and functions. We performed drug sensitivity analysis using the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org/). We gathered drug response data and pathways targeted by drugs from GDSC. We then obtained the drugs associated with the risk scores by performing Pearson correlation analysis.

Clinical sample collection and quantitative real-time polymerase chain reaction analysis

Cancerous and para-cancerous tissues from 25 pairs of men with BLCA were collected from the Second Affiliated Hospital of the Army Military Medical University and the Affiliated Hospital of Guangdong Medical from March 2022 to June 2023. This study was approved by the Ethics Committee of the Affiliated Hospital of Guangdong Medical (YJYS2023179) and conducted following the Helsinki Declaration. The BLCA tissue was obtained from patients undergoing radical cystectomy. BLCA was graded by senior pathologists from the Department of Pathology, Army Military Medical University. Total RNA was extracted from all tissues using Trizol (Invitrogen, 15,596,026). The SYBR Green One-Step qRT-PCR kit (Invitrogen, 11,736,059) was used to measure total RNA (100 ng). PrimeScript RT Master Mix (Shenggong, Shanghai, China) was used for cDNA synthesis. This study was approved by the Committee of the Affiliated Hospital of Guangdong Medical of China (Approval No. YJYS2023179) and was conducted following the Declaration of Helsinki. The PCR primers are listed below:

PYK2B, forward: 5′- AATGCACTTGACAAGAAGTC-3′;

PYK2B, reverse: 5′- GCTTTAAGTTCTCCTGCATC-3′;

SCD, forward: 5′-GCCATCGGTCCTACAAAGC-3′;

SCD, reverse: 5′-AGCCAATGTGGGAGAAGAAA-3′;

β-actin, forward:5′-CAACGAGCGGTTCAGGTGT-3′;

β-actin, reverse: 5′-TGGAGTTGAAGGTGGTCTCG-3′;

FADS1, forward: 5′-GTGGCTAGTGATCGACCGTAA-3′;

FADS1, reverse: 5′- ATTCT TGGTG GGCTC AAAGC-3′;

TPD52, forward: 5′-AACAGAACATTGCCAAAGGGTG-3′;

TPD52, reverse: 5′- TGACTGAGCCAACAGACGAAA-3′;

CDK6, forward: 5′-GCTGACCAGCAGTACGAATG-3′;

CDK6, reverse: 5′-GCACACATCAAACAACCTGACC-3′;

PGM3, forward: 5′-GCAGAGAGTGCTTATTGACATCA-3′;

PGM3, reverse: 5′-TGTGAAAGTTTCTCACTGCTGG-3′;

The 2-ΔΔCt method was conducted to calculate the RNA expression. Student’s t-test was used to compare the expression level of each RNA between different groups.

Statistical analysis

Statistical analysis was conducted using the R software, with a significance level of p < 0.05. Continuous variables were evaluated using the t-test, while categorical variables were compared using the χ2 test. Furthermore, multivariate analyses were conducted to determine the independent prognostic significance of immune-related risk factors. The mean ± SD of three independent experiments is presented. ns, not significant, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

Screening and genetic variation of ARGs in BLCA

Univariate Cox regression analysis identified 32 ARGs. The clinical prognostic value of these ARGs was depicted in a network diagram (Fig. 1A). Notably, TPD52, KRT8, and PTK2B were found to be highly expressed in patients with a relatively positive prognosis for BLCA, whereas the remaining ARGs were highly expressed in patients with a relatively negative prognosis. The BLCA patients were divided into 2 subtypes using an unsupervised clustering algorithm based on the expression patterns of 32 genes (Fig. 1B). These two clusters showed different expression patterns of the 32 ARGs. To confirm the validity of these subtypes, the BLCA samples were clustered using PCA principal component analysis, which revealed a clear division into cluster 1 (A) and cluster 2 (B) (Fig. 1C). To determine the clinical prognostic significance of Cluster A and Cluster B, the Kaplan-Meier survival curve (Fig. 1D) shows that the survival prognosis of Cluster 1 is worse than that of Cluster 2. This suggests that distinct expression patterns of ARGs result in significantly different prognoses in BLCA patients. From the box plot, the expression levels of ARGs in cluster A exceed those in cluster B (Fig. 1E). CNV-associated mutations were found to be widespread by examining the genetic copy variation of genes in 32 ARGs. Five genes, namely NDRG1, IDI1, TPD52, VAPA, and DHCR24, showed extensive CNV amplification. On the other hand, four genes, namely, TSC22D1, PGM3, PTPN21, and INSIG1, showed CNV loss (Fig. 1F). Using 32 subtypes of ARGs’ expression, BLCA samples were classified into subtypes that developed different prognostic features and mutations.

Fig. 1
figure 1

Screening and genetic variation of androgen-responsive genes (ARGs) in bladder cancer (BLCA). (A) The network diagram illustrates the interaction between 32 ARGs in BLCA. The size of the circles indicates the p-value of each gene for survival prognosis. The red dots indicate risk factors, while the green dots indicate favorable factors. The thickness of the lines represents the correlation between genes, with red and blue lines representing positive and negative correlations respectively. (B) Molecular subtype clustering of bladder cancer patients based on ARGs. (C) Principal component analysis of clusters A and B. (D) Assessment of overall survival differences between cluster A and cluster B. Cluster A is shown in blue and cluster B in yellow. (E) Differences in the expression of 32 androgen-responsive genes between cluster A and cluster B. (F) Top frequency of CNV variants from the 32 ARGs

Full size image

Analysis of immune infiltration and gene enrichment

To investigate the correlation between 32ARGs and clinical features of BLCA, the expression of ARGs in cluster A and cluster B was mapped using a heat map. The two clusters showed different expression patterns of ARGs (Fig. 2A). To investigate the potential functions of the two groups, a KEGG analysis was performed (Fig. 2B). The results indicated significant activation of the following pathways for cluster A: cytokine receptor interaction, focal adhesion, and ECM receptor interaction, among others. Conversely, cluster B showed a marked activation of linoleic acid metabolism. As variations in the tumor microenvironment may influence tumor progression, an analysis of immune infiltration was then performed via CIBERSORT algorithm. Between cluster A and cluster B, cluster A was found to have a high infiltration of different immune cells, including B cells, CD4 T cells, and CD8 T cells. This infiltration was greater than that of cluster B (Fig. 2C). These differences in the enrichment of immune cells and signaling pathways may conducted to the difference in prognosis.

Fig. 2
figure 2

Enrichment analysis and immune infiltration analysis of 33 androgen-responsive genes (ARGs). (A) Sample annotation of 33 ARGs by stage, grade, sex, and cluster. (B) KEGG analysis was performed on androgen response genes of two gene clusters. (C) sGSEA analysis performed on clusters A and B for immune cell infiltration

Full size image

Risk score construction and validation

To identify the BLCA-related genes, we used the LASSO algorithm to screen the previously obtained 33 ARGs. This led to the identification of six ARGs, namely CDK6, PTK2B, SCD, FADS1, PGM3, and TPD52 (Fig. 3A). The TCGA dataset was divided into training set (n = 203) and validation set (n = 202) randomly. We then used multivariate Cox analysis to establish a risk model using these six ARGs to predict survival and prognosis in the training set of BLCA patients. The TCGA validation set and external validation set (GSE13507) were used to explore the predictive efficiency of this model. Based on the risk scores, BLCA samples were divided into a high-risk group (risk score above the median) and a low-risk group (risk score below the median). The sensitivity and specificity of the prognostic models were assessed using the time-dependent receiver operating characteristic (ROC) approach. The ROC curve area evaluation results show that the training set had AUCs of 0.764, 0.727, and 0.759 for 1, 3, and 5 years respectively. The test set displayed AUCs of 0.674, 0.670, and 0.640, while the external validation set showed AUCs of 0.752, 0.635, and 0.637 (Fig. 3B). The risk model showed optimal effectiveness in predicting survival, as indicated by ROC analysis. In addition, Kaplan-Meier (KM) analysis showed a negative correlation between higher risk scores and survival rates in the training, test, and external validation sets. The results were statistically significant (P < 0.0001) (Fig. 3C). KM analysis indicates that our model successfully discriminates the probability of survival in BLCA patients. We also trained the model in the whole TCGA dataset. In the hole TCGA set, the heatmap showed high expression levels of CDK6, SCD, FADS1, and PGM3 in the high-risk group, whereas PTK2B and TPD52 were expressed at low levels in the high-risk group (Fig. 3D). Of these, CDK6, SCD, FADS1, and PGM3 are risk factors, while PTK2B and TPD52 are protective factors. In the Sankey diagram (Fig. 3E), the specific genes in the two clusters were found to contribute to different risks and ultimately to different prognoses. In the barplot of surv status percents, the low risk group had higher mortality (Fig. 3F). In this section, different algorithms have been used to create a risk model for bladder cancer. The results demonstrate the ability of the model to effectively discriminate and predict the prognosis of BLCA patients.

Fig. 3
figure 3

Construction and validation of risk scores for 33 androgen-responsive genes (ARGs). (A) A Least Absolute Shrinkage and Selection Operator (LASSO) analysis was conducted on ARGs associated with prognosis. (B) Time-dependent receiver operating characteristics (ROC) were calculated for the training, test, and validation sets. (C) Kaplan-Meier curves were plotted for the training set (p < 0.001, log-rank test), test set (p < 0.001, log-rank test), and validation set (p < 0.001, log-rank test). (D) Heatmap showing the expression levels of the six ARGs (E) Sankey diagram showing the relationship between risk factors and clinical prognosis of clusters A and B. (F) The survival status percents difference of high-risk group and low-risk group

Full size image

Gene enrichment analysis of risk groups

The ‘ClusterProfiler’ package in R was used for GO enrichment and KEGG pathway enrichment analysis of differentially expressed genes in high and low-risk groups of BLCA. The results of both groups were presented using heat maps. Figure 4B shows that in the high-risk BLCA group, the differentially expressed genes are mainly associated with biological processes such as nucleobase metabolism, amino sugar biosynthesis, and pyrimidine nucleobase metabolism. In terms of cellular components, they are primarily associated with microtubule plus ends and nucleocytoplasmic transport complexes. Additionally, the Molecular Function category is largely associated with cytoskeletal motor activity and microtubule binding. The results of the KEGG pathway enrichment analysis (Fig. 4A) indicate that genes abnormally expressed in the high-risk BLCA group are enriched in pathways related to cell types, regulation of the actin cytoskeleton, long-term potentiation, and gap junctions, among others. This suggests that genes that are abnormally expressed in high-risk BLCA groups may affect cell proliferation and tumor growth in BLCA through these pathways. These key enriched pathways had a significant impact on survival in BLCA patients.

Fig. 4
figure 4

Gene enrichment analysis of androgen-responsive genes (ARGs) in high and low-risk groups. (A) KEGG enrichment analysis of ARGs was performed on high and low-risk groups. (B) GO enrichment analysis of ARGs was performed on high and low-risk groups

Full size image

Nomogram based on BLCA risk score and clinical features

To gain a more complete understanding of the prognostic features, the patient’s likelihood of survival was plotted using a nomogram that combined our risk model and clinical signature. The nomogram we created was able to predict survival probabilities for BLCA at 1, 3, and 5 years based on different risk levels, age, gender, tumor grade, and tumor stage (Fig. 5A). Each variable was given a point based on their corresponding interval range, and all the scores are then summed to produce a total score corresponding to its relative viability prediction. Our predicted probability of survival was consistent with the actual figures (Fig. 5B). The nomogram was also able to discriminate between low-risk and high-risk patients, as assessed by the KM curve. The high-risk group exhibited lower survival ability and a worse prognosis (Fig. 5C). The risk factors in the prognostic model were evaluated using decision curve analysis (DCA), which indicates that the nomogram has sufficient validity (Fig. 5D). In summary, by combining the risk score and the clinical factor, we established a nomogram that had a paragon effect on predicting prognosis.

Fig. 5
figure 5

A nomogram based on BLCA risk score and clinical features. A nomogram based on BLCA risk score and clinical features. (A) A nomogram is generated based on an individual’s sex, age, risk score, pathological grade, and clinical stage. (B) Calibration curve showing the agreement between predicted survival at 1, 3, and 5 years and actual survival using the bias-adjusted prognostic nomogram. (C) The baseline cumulative hazards identified for each event number. (D) The decision curve analysis (DCA) of the train set and the test set containing the nomogram

Full size image

Immune microenvironment analysis

The tumor immune microenvironment consists mainly of tumor cells, immune infiltrating cells, stromal cells, and some extracellular components [22]. Non-tumor cells, such as immune cells and stromal cells in the microenvironment, are highly relevant to tumor immunotherapy and survival prognosis [23]. Therefore, we used the ESTIMATE algorithm to calculate the immune score, stromal score, and purity of tumor cell infiltration in BLCA tissues. We aimed to identify any differences in the composition of TME between high-risk and low-risk groups. As observed in Fig. 6A, the stromal score of the high-risk group was noticeably greater than that of the low-risk group, which leads to lower level of tumor purity in high-risk group (Fig. 6B). Conversely, the ESTIMATE score of the high-risk group was significantly superior to that of the low-risk group. Based on this, we present the ratio of immune infiltrating cells between the two groups (Fig. 6B). The correlation between different immune cells and the risk score was then examined. The results of our bioinformatics analysis suggested that regulatory T cells, CD8 T cells, monocytes, naive B cells, plasma cells, and activated dendritic cells were negatively associated with the risk score, whereas macrophage M2, macrophage M0, and CD4 resident memory T cells were positively correlated with the risk score (Fig. 6C-K). Inhibiting CD8 T cells and activating M2 macrophages may be critical determinants of poor outcomes in the high-risk group.

Fig. 6
figure 6

Analysis of the immune microenvironment in BLCA. (A) Analysis of immune infiltration using the Estimate algorithm. (B) The tumor purity conducted by ESTIMATE algorithm. (C) Determine the immune cell ratio using the Cibersort method. (D) Investigate the correlation between the risk score and regulatory T cells. (E) Examine the correlation between the risk score and CD8 T cells. (F) Explore the correlation between the risk score and CD4 resident memory T cells. (G) Examine the correlation between the Risk Score and M2 macrophages. (H) Correlation between risk score and M0 macrophages. (I) Correlation between risk score and monocytes. (J) Correlation between risk score and naive B cells. (K) Correlation between risk score and activated dendritic cells. (L) Correlation between risk score and plasma cells

Full size image

Six ARGs in clinical samples

We analyzed the expression of six ARGs in all clinical samples (Fig. 7). The results showed that the mRNA levels of CDK6, FADS1, PGM3, and SCD were significantly higher in primary tumors compared to normal samples (Fig. 7). In contrast, PTK2B was expressed at a lower level in primary tumors than in normal tumors (Fig. 7). These results suggest that CDK6, FADS1, PGM3, and SCD are risk factors, whereas PTK2B is a protective factor, which is consistent with our risk model. TPD52 expression did not show a statistically significant difference between the two groups as shown in Fig. 7.

Fig. 7
figure 7

Expression levels of signature genes between normal and BLCA tissues were experimentally verified by qRT-PCR

Full size image

Validation of the single-cell data

To investigate the cell type-specific expression of the model genes, we performed an analysis of a single-cell dataset. By analyzing the GSE135337 dataset, we identified five unique cell clusters using UMAP plots. These clusters include urothelial cells, myeloid/macrophages, T cells, fibroblasts, and endothelial cells as shown in Fig. 8A. For a detailed investigation of the expression of six ARGs in each cluster, we generated both violin and UMAP plots, as shown in Fig. 8B and C. Notably, CDK6, FADS1, PGM3, PTK2B, and SCD were found to be highly expressed in urothelial cells, suggesting that these five ARGs function primarily by directly affecting urothelial cells. TPD52 expression was increased in all five cell types, with a notable increase in T cells. This suggests that TPD52 acts as a protective factor, influencing the prognosis of BLCA patients through T-cell-related functions. In addition, this result may explain the non-differential trend in TPD52 expression observed in the Q-PCR analysis.

Fig. 8
figure 8

Single-cell data validation analysis of six ARGs. (A) Five cell clusters (urothelial cells, endothelial cells, fibroblasts, myeloid/macrophages, T cells) were identified on the UMAP map. (B) Violin plot of six ARGs in each cluster. (C) UMAP plot of six ARGs in each cluster

Full size image

Drug sensitivity analysis and screening

To investigate the response of different risk scores to chemotherapy drugs in different types of cancer, we identified 68 drugs from the GDSC database. We then assessed the semi-maximum inhibitory concentration (IC50) of these chemotherapeutic drugs between the low-risk and high-risk groups to explore their potential use in BLCA chemotherapy. The results show lower IC50 values for fludarabine, AZD8186, and carmustine in the high-risk group, suggesting a potential benefit for patients in this group (Fig. 9A-F). In contrast, the low-risk group had lower IC50 values for trametinib, selumetinib, and VX-11e, indicating that they may benefit more from these treatments (Fig. 9A-F).

Fig. 9
figure 9

Drug sensitivity analysis. (A) Sensitivity of trametinib between high-risk and low-risk groups. (B) Sensitivity of temozolomide between high and low-risk groups. (C) Sensitivity of VX-11e between high-risk and low-risk groups. (D) Sensitivity of fludarabine between high and low-risk groups. (E) Sensitivity of AZD8186 between high and low-risk groups. (F) Sensitivity of carmustine between high and low-risk groups

Full size image

Discussion

BLCA is a common genitourinary tumor with significant gender differences. The incidence of BLCA is significantly higher in men than in women [24]. Numerous experimental studies suggest that the androgen receptor (AR) accelerates the growth and metastasis of BLCA and thus plays a critical role in the development and progression of BLCA [25, 26]. The bladder and prostate both originate from the urogenital sinus during embryonic development. In addition, the androgen receptor (AR) has been detected in BLCA tissue, strongly suggesting a potential role for both androgen and AR in BLCA [24,25,26,27]. However, our understanding of the mechanism and prognostic significance of ARGs in BLCA is currently limited.

At present, some studies have shown that the presence of androgen-response genes in different types of tumors is closely linked to their occurrence [28]. The 14-3-3ζ gene is a novel androgen-responsive gene that is upregulated in prostate cancer and promotes prostate cancer cell proliferation and survival [29]. Miyamoto et al. used the AR knockout mouse model and found that 92% of wild-type male mice and 42% of wild-type female mice eventually developed BLCA when stimulated with N-butyl-N-(4-hydroxybutyl-N) nitrosamine (BBN), while the AR knockout male and female mice did not develop BLCA [20]. Further research has shown that blocking androgen or interfering with AR expression levels can reduce or delay the onset of BBN-induced BLCA [30, 31]. Recent studies have shown that AR and EGF-EGFR also regulate each other and that their interaction promotes the progression of BLCA [32]. At the same time, AR regulates DNA damage by regulating the p53 PCNA DNA repair signal, thereby promoting tumorigenesis of normal bladder epithelium and the development of BLCA [17]. In addition, androgens can increase beta-transcription levels of catenin through AR signaling to promote Wnt/beta-catenin-induced BLCA [33]. In conclusion, these studies show that androgens and AR play an important role in the development of BLCA and partly explain the reasons for the differences in BLCA between the sexes. However, there are few studies on the impact of ARGs on BLCA invasion and drug resistance, and their role in predicting BLCA prognosis.

In this study, a thorough understanding of the expression profile of androgen-associated genes in BLCA was achieved. 33 prognostically valuable ARGs were selected using univariate Cox regression. BLCA patients were categorized using an unsupervised consensus clustering algorithm, resulting in two expression subgroups. Of the two subtypes, cluster A showed a notable adverse survival prognosis in BLCA patients. In addition, immune cell infiltration was assessed using ssGSEA to identify differences in immune cell composition between the two subgroups. The results showed that cluster A had an increased level of immune infiltration, particularly MDSC and neutrophils. MDSC was identified as playing a role in relapse and consequently influencing the poor prognosis of BLCA [34]. Neutrophils can be divided into several subtypes that contribute to tumor progression, development, and poor survival [35]. We then used the LASSO algorithm to select 6 ARGs with optimal prognostic value. Among these, CDK6, SCD, FADS1, and PGM3 were classified as risk factors, while PTK2B and TPD52 were identified as protective factors. The Q-PCR results showed that CDK6, SCD, FADS1, and PGM3 were expressed at higher levels in tumor tissue compared to normal tissue, while PTK2B was expressed at higher levels in normal tissue. These results are consistent with previous analyses. We observed that there was no significant difference in TPD52 expression levels between the tumor and normal tissue. Consequently, we decided to conduct a single-cell RNA-sequencing analysis and identified five genes, CDK6, FADS1, PGM3, PTK2B, and SCD, that exhibited high expression levels in urothelial cells, indicating a potential direct role in promoting or inhibiting tumor cell growth. TPD52 expression was detected in all five cell clusters, with particularly high expression in macrophages and T cells. These findings suggest that the protective role of TPD52 may be related to immune regulation of BLCA, rather than direct inhibition of BLCA cells. This also explains why the expression of TPD52 does not differ significantly in overall tissue expression, while still having a protective prognostic significance. Previous research has shown that increased expression of FADS1, CDK6, and SCD is strongly correlated with BLCA incidence, progression, and poor prognosis, and our results confirm these findings [36,37,38]. The significance of PGM3, PTK2B, and TPD52 in bladder cancer remains largely unexplored. Further research is needed to clarify and analyze the role of these proteins in BLCA.

We then created a risk-scoring model based on ARGs’ expression and analyzed and verified the survival and prognosis of patients with BLCA. Our study further demonstrated the efficacy of our model in differentiating and predicting the prognosis of BLCA patients using multiple algorithms. The validation and external validation sets, single-cell RNA-sequencing set, and Q-PCR analysis all supported the efficacy of our model. In addition, the potential function between high and low-risk groups was investigated using GO and KEGG. Abundant enrichments were observed for molecular functions such as Wnt, cell cycle, and gap junction in the high-risk group. Studies suggest that an abnormal WNT pathway may lead to poor outcomes and a lower response rate to neoadjuvant chemotherapy or immunotherapy in BLCA [39].

Cell cycle and gap junction dysfunction can lead to tumor proliferation and migration in BLCA [40, 41]. In addition, our research examined the immune microenvironment of tumors in different prognosis groups. Our results showed a positive correlation between the risk score and M2 macrophages and a negative correlation with CD8 + T cells, suggesting increased M2 macrophage infiltration and decreased CD8 + T cell infiltration in high-risk patients. There is a strong correlation between M2 macrophages and the epithelial-mesenchymal transition that leads to tumor progression [42]. CD8 + T cells can differentiate into cytotoxic CD8 + T cells capable of killing tumor cells and exerting anti-tumor activity [43]. The imbalance between CD8 + T cells and M2 macrophages was fundamental to the development of a prognostic model. By combining our risk model with clinical information, we have developed a nomogram. Validation using the KM curve demonstrated the strong predictive power of the nomogram and its contribution to our study. When we analyzed the treatment outcomes of people with BLCA based on their risk scores, we discovered a correlation between drug sensitivity and the scores. It was observed that BLCA patients classified as high risk had an increased sensitivity to chemotherapy drugs such as fludarabine, AZD8186, and carmustine, while those classified as low risk had an increased sensitivity to trametinib, selumetinib, and VX-11e. Our research has the potential to provide guidance and a point of reference for the development of BLCA drugs in the future.

Conclusion

We found that androgen-responsive genes play important roles in prognosis, immune infiltration, and tumor biological functions in BLCA patients. Based on the androgen-responsive genes, we developed a prognostic model that could predict survival, immune cell infiltration, and drug candidates, and further explain the potential prognosis-related molecular mechanism of BLCA patients. Our model was shown to be particularly effective and provides novel strategies for the treatment and management of patients with BLCA in future clinical practice.

Data availability

The databases used for constructing datasets in this study were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/) and the GSE13507 and GSE135337 dataset from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/).

References

  1. Lenis AT, Lec PM, Chamie K, et al. Bladder Cancer: Rev JAMA. 2020;324(19):1980–91.

    CAS  Google Scholar 

  2. Lobo N, Shariat SF, Guo CC, et al. What is the significance of variant histology in Urothelial Carcinoma? Eur Urol Focus. 2020;6(4):653–63.

    Article  PubMed  Google Scholar 

  3. Cai Z, Liu Q. Understanding the Global Cancer statistics 2018: implications for cancer control. Sci China Life Sci. 2021;64(6):1017–20.

    Article  PubMed  Google Scholar 

  4. Sung H, Ferlay J, Siegel RL, et al. Global Cancer statistics 2020: GLOBOCAN estimates of incidence and Mortality Worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49.

    Article  PubMed  Google Scholar 

  5. Cumberbatch MG, Cox A, Teare D, et al. Contemporary occupational carcinogen exposure and bladder Cancer: a systematic review and Meta-analysis. JAMA Oncol. 2015;1(9):1282–90.

    Article  PubMed  Google Scholar 

  6. Cumberbatch MGK, Jubber I, Black PC, et al. Epidemiology of bladder Cancer: a systematic review and contemporary update of risk factors in 2018. Eur Urol. 2018;74(6):784–95.

    Article  PubMed  Google Scholar 

  7. Li P, Chen J, Miyamoto H. Androgen receptor signaling in bladder Cancer. Cancers (Basel). 2017;9(2):20.

    Article  PubMed  Google Scholar 

  8. Schweizer MT, Yu EY. AR-Signaling in Human malignancies: prostate Cancer and Beyond. Cancers. 2017;9(1):7.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Burger M, Catto JW, Dalbagni G, et al. Epidemiology and risk factors of urothelial bladder cancer. Eur Urol. 2013;63(2):234–41.

    Article  PubMed  Google Scholar 

  10. Brzozowska M, Lewiński A. Changes of androgens levels in menopausal women. Prz Menopauzalny. 2020;19(4):151–4.

    CAS  PubMed  Google Scholar 

  11. Martínez-Rojo E, Berumen LC, García-Alcocer G, et al. The role of androgens and androgen receptor in human bladder Cancer. Biomolecules. 2021;11(4):594.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Yang L, Huang W, Bai X, et al. Androgen dihydrotestosterone promotes bladder cancer cell proliferation and invasion via EPPK1-mediated MAPK/JUP signaling. Cell Death Dis. 2023;14(6):363.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. Jordan CL, Doncarlos L. Androgens in health and disease: an overview. Horm Behav. 2008;53(5):589–95.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Gao L, Schwartzman J, Gibbs A, et al. Androgen receptor promotes ligand-independent prostate cancer progression through c-Myc upregulation. PLoS ONE. 2013;8(5):e63563.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Teng XY, Liu GQ, Diao XL, et al. CAG repeats in the androgen receptor gene are shorter in patients with pulmonary, esophageal, or bladder carcinoma and longer in women with uterine leiomyoma. Oncol Rep. 2010;23(3):811–8.

    CAS  PubMed  Google Scholar 

  16. Shiota M, Takeuchi A, Yokomizo A, et al. Androgen receptor signaling regulates cell growth and vulnerability to doxorubicin in bladder cancer. J Urol. 2012;188(1):276–86.

    Article  CAS  PubMed  Google Scholar 

  17. Hsu JW, Hsu I, Xu D, et al. Decreased tumorigenesis and mortality from bladder cancer in mice lacking urothelial androgen receptors. Am J Pathol. 2013;182(5):1811–20.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Johnson AM, O’Connell MJ, Miyamoto H et al. Androgenic dependence of exophytic tumor growth in a transgenic mouse model of bladder cancer: a role for thrombospondin-1. BMC Urol 2008(23):87.

  19. Mashhadi R, Pourmand G, Kosari F. Role of steroid hormone receptors in formation and progression of bladder carcinoma: a case-control study. Urol J. 2014;11(6):1968–73.

    PubMed  Google Scholar 

  20. Miyamoto H, Yang Z, Chen YT, et al. Promotion of bladder cancer development and progression by androgen receptor signals. J Natl Cancer Inst. 2007;99(7):558–68.

    Article  CAS  PubMed  Google Scholar 

  21. Lai H, Cheng X, Liu Q, et al. Single-cell RNA sequencing reveals the epithelial cell heterogeneity and invasive subpopulation in human bladder cancer. Int J Cancer. 2021;149(12):2099–115.

    Article  CAS  PubMed  Google Scholar 

  22. Anderson NM, Simon MC. The tumor microenvironment. Curr Biol. 2020;30(16):R921–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. Lin Y, Xiao Y, Liu S, et al. Role of a lipid metabolism-related lncRNA signature in risk stratification and immune microenvironment for colon cancer. BMC Med Genomics. 2022;15(1):221.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Chen J, Huang CP, Quan C, et al. The androgen receptor in bladder cancer. Nat Rev Urol. 2023;20(9):560–74.

    Article  CAS  PubMed  Google Scholar 

  25. Jamroze A, Chatta G, Tang DG. Androgen receptor (AR) heterogeneity in prostate cancer and therapy resistance. Cancer Lett. 2021;518:1–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Tan MH, Li J, Xu HE, et al. Androgen receptor: structure, role in prostate cancer and drug discovery. Acta Pharmacol Sin. 2015;36(1):3–23.

    Article  CAS  PubMed  Google Scholar 

  27. Sanguedolce F, Cormio L, Carrieri G, et al. Role of androgen receptor expression in non-muscle-invasive bladder cancer: a systematic review and meta-analysis. Histol Histopathol. 2020;35(5):423–32.

    CAS  PubMed  Google Scholar 

  28. Quan J, Bode AM, Luo X. ACSL family: the regulatory mechanisms and therapeutic implications in cancer. Eur J Pharmacol. 2021;909:174397.

    Article  CAS  PubMed  Google Scholar 

  29. Murata T, Takayama K, Urano T, et al. 14-3-3zeta, a novel androgen-responsive gene, is upregulated in prostate cancer and promotes prostate cancer cell proliferation and survival. Clin Cancer Res. 2012;18(20):5617–27.

    Article  CAS  PubMed  Google Scholar 

  30. Imada S, Akaza H, Ami Y, et al. Promoting effects and mechanisms of action of androgen in bladder carcinogenesis in male rats. Eur Urol. 1997;31(3):360–4.

    Article  CAS  PubMed  Google Scholar 

  31. Hsu JW, Hsu I, Xu D, et al. Decreased tumorigenesis and mortality from bladder cancer in mice lacking urothelial androgen receptor. Am J Pathol. 2013;182(5):1811–20.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Zheng Y, Izumi K, Yao JL, et al. Dihydrotestosterone upregulates the expression of epidermal growth factor receptor and ERBB2 in androgen receptor-positive bladder cancer cells. Endocr Relat Cancer. 2011;18(4):451–64.

    Article  CAS  PubMed  Google Scholar 

  33. Lin C, Yin Y, Stemler K, et al. Constitutive β-catenin activation induces male-specific tumorigenesis in the bladder urothelium. Cancer Res. 2013;73(19):5914–25.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Chevalier MF, Trabanelli S, Racle J, et al. ILC2-modulated T cell-to-MDSC balance is associated with bladder cancer recurrence. J Clin Invest. 2017;127(8):2916–29.

    Article  PubMed  PubMed Central  Google Scholar 

  35. Xue R, Zhang Q, Cao Q, et al. Liver tumor immune microenvironment subtypes and neutrophil heterogeneity. Nature. 2022;612(7938):141–7.

    Article  CAS  PubMed  Google Scholar 

  36. Li Q, Wang S, Wu Z, et al. DDX11-AS1 exacerbates bladder cancer progression by enhancing CDK6 expression via suppressing miR-499b-5p. Biomed Pharmacother. 2020;127:110164.

    Article  CAS  PubMed  Google Scholar 

  37. Lu Y, Shao J, Shu X, et al. ADS1 is a prognostic biomarker in bladder Cancer: a study based on TCGA Data. Comb Chem High Throughput Screen. 2021;24(8):1197–204.

    Article  CAS  PubMed  Google Scholar 

  38. Piao C, Cui X, Zhan B, et al. Inhibition of stearoyl CoA desaturase-1 activity suppresses tumor progression and improves prognosis in human bladder cancer. J Cell Mol Med. 2019;23(3):2064–76.

    Article  CAS  PubMed  Google Scholar 

  39. Wu G, Weng W, Xia P, et al. Wnt signaling pathway in bladder cancer. Cell Signal. 2021;79:109886.

    Article  CAS  PubMed  Google Scholar 

  40. Nishiyama H, Watanabe J, Ogawa O. p53 and chemosensitivity in bladder cancer. Int J Clin Oncol. 2008;13(4):282–6.

    Article  CAS  PubMed  Google Scholar 

  41. Tschernig T. Connexins and gap junctions in Cancer of the urinary tract. Cancers (Basel). 2019;11(5):704.

    Article  CAS  PubMed  Google Scholar 

  42. Sharifi L, Nowroozi MR, Amini E, et al. A review on the role of M2 macrophages in bladder cancer; pathophysiology and targeting. Int Immunopharmacol. 2019;76:105880.

    Article  CAS  PubMed  Google Scholar 

  43. Zheng X, Zhou X, Xu H, et al. A Novel Immune-Gene pair signature revealing the Tumor Microenvironment features and Immunotherapy Prognosis of muscle-invasive bladder Cancer. Front Genet. 2021;12:764184.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This research was supported by the Guangdong Provincial Natural Science Foundation General Project (2414050006122 and 2024A1515010342) and Guangdong Provincial Key Laboratory of Autophagy and Major Chronic Non-Communicable Diseases (2022B1212030003), Open funding from Affiliated Hospital of Guangdong Medical University—research of autophagy and diseases (KF202305) and National key R&D program (2020YFA0908800) and 2024 Additional Special Funding of Guangdong Medical Universit (4SG24016G) and Start-up funds for scientific research of high- level talents of discipline leaders of the Affiliated Hospital of Guangdong Medical University (1035Z20230005) and Central University basic research business fee special project (2022CDJYGRH-020).

Author information

Author notes

  1. Jiang Zhao, Qian Zhang and Cunle Zhu contributed equally to this work.

Authors and Affiliations

  1. Department of Urology, Guangdong Provincial Key Laboratory of Autophagy and Major Chronic Non-Communicable Diseases, Affiliated Hospital of Guangdong Medical University, Zhanjiang, 524001, China

    Jiang Zhao, Cunle Zhu, Wu Yuqi, Guohui Zhang, Qianliang Wang & Xiangwei Wang

  2. Department of Urology, General Hospital of Northern Theater Command, Shenyang, 110016, China

    Qian Zhang

  3. Department of Urology, Second Affiliated Hospital, Army Medical University, Chongqing, 400037, China

    Jiang Zhao

  4. Department of Urology, South China Hospital Affiliated to Shenzhen University, Shenzhen, 518000, China

    Wu Yuqi & Xingyou Dong

  5. Department of Urology, People ’ s Hospital of Shapingba District, Chongqing, 400030, China

    Jiang Zhao & Xingyou Dong

  6. Department of Urology, The University of Kansas Medical Center, Kansas City, KS, 66160, USA

    Benyi Li

Authors
  1. Jiang Zhao
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Qian Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Cunle Zhu
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Wu Yuqi
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Guohui Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Qianliang Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Xingyou Dong
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Benyi Li
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Xiangwei Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Conceptualization: J.Z, Q.Z. Data curation: C.L.Z, W.Y.Q. Formalanalysis: J.Z, G.H.Z. Funding acquisition: J.Z. Investigation: J.Z, Q.L.W. Methodology: J.Z, Q.Z. Project administration: J.Z, Q.Z. Resources: Q.L.W. Software: X.Y.D. Supervision: B.Y.L, X.W.W. Validation: B.Y.L, X.W.W. Visualization: B.Y.L, X.W.W. Writing – original draft: J.Z. Writing – review & editing: all authors.

Corresponding authors

Correspondence to Xingyou Dong, Benyi Li or Xiangwei Wang.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Committee of the Affiliated Hospital of Guangdong Medical of China (Approval No. YJYS2023179) and was conducted following the Declaration of Helsinki. Written informed consent was obtained from individual or guardian participants.

Consent for publication

All authors have read and agreed to the published version of the manuscript.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, J., Zhang, Q., Zhu, C. et al. Prognostic feature based on androgen-responsive genes in bladder cancer and screening for potential targeted drugs. BioData Mining 17, 59 (2024). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13040-024-00377-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13040-024-00377-x

Keywords

BioData Mining

ISSN: 1756-0381