*Article* **Comprehensive Analysis of CPA4 as a Poor Prognostic Biomarker Correlated with Immune Cells Infiltration in Bladder Cancer**

**Chengcheng Wei 1,†, Yuancheng Zhou 1,†, Qi Xiong 2,†, Ming Xiong <sup>1</sup> , Yaxin Hou <sup>1</sup> , Xiong Yang 1,\* and Zhaohui Chen 1,\***


**Simple Summary:** The overexpression of Carboxypeptidase A4 (CPA4) has been observed in plenty of types of cancer and has been elucidated to promote tumor growth and invasion; however, its role in bladder urothelial carcinoma (BLCA) is still unclear. Therefore, we aimed to show the prognostic role of CPA4 and its relationship with immune infiltrates in BLCA. We confirmed that the overexpression of CPA4 is associated with shorter overall survival, disease-specific survival, progress-free intervals, and higher dead events. Moreover, we found that several infiltrating immune cells (Th1cell, Th2 cell, T cell exhaustion, and Tumor-associated macrophage) were correlated with the expression of CPA4 in bladder cancer using TIMER2 and GEPIA2. In conclusion, CPA4 may be a novel and great prognostic biomarker based on bioinformation analysis in BLCA.

**Abstract:** Carboxypeptidase A4 (CPA4) has shown the potential to be a biomarker in the early diagnosis of certain cancers. However, no previous research has linked CPA4 to therapeutic or prognostic significance in bladder cancer. Using data from The Cancer Genome Atlas (TCGA) database, we set out to determine the full extent of the link between CPA4 and BLCA. We further analyzed the interacting proteins of CPA4 and infiltrated immune cells via the TIMER2, STRING, and GEPIA2 databases. The expression of CPA4 in tumor and normal tissues was compared using the TCGA + GETx database. The connection between CPA4 expression and clinicopathologic characteristics and overall survival (OS) was investigated using multivariate methods and Kaplan–Meier survival curves. The potential functions and pathways were investigated via gene set enrichment analysis. Furthermore, we analyze the associations between CPA4 expression and infiltrated immune cells with their respective gene marker sets using the ssGSEA, TIMER2, and GEPIA2 databases. Compared with matching normal tissues, human CPA4 was found to be substantially expressed. We confirmed that the overexpression of CPA4 is linked with shorter OS, DSF(Disease-specific survival), PFI(Progression-free interval), and increased diagnostic potential using Kaplan–Meier and ROC analysis. The expression of CPA4 is related to T-bet, IL12RB2, CTLA4, and LAG3, among which T-bet and IL12RB2 are Th1 marker genes while CTLA4 and LAG3 are related to T cell exhaustion, which may be used to guide the application of checkpoint blockade and the adoption of T cell transfer therapy.

**Keywords:** CPA4; bladder urothelial carcinoma; immune cells; T cell exhaustion; checkpoint

## **1. Introduction**

Bladder Urothelial Carcinoma (BLCA) is the eighth most prevalent cancer worldwide, with 549,393 new cases reported worldwide in 2018 [1]. Additionally, in the USA alone, there are estimated to be more than 80,000 new cases and 17,000 deaths each year [2].

**Citation:** Wei, C.; Zhou, Y.; Xiong, Q.; Xiong, M.; Hou, Y.; Yang, X.; Chen, Z. Comprehensive Analysis of CPA4 as a Poor Prognostic Biomarker Correlated with Immune Cells Infiltration in Bladder Cancer. *Biology* **2021**, *10*, 1143. https://doi.org/ 10.3390/biology10111143

Academic Editors: Shibiao Wan, Yiping Fan, Chunjie Jiang and Shengli Li

Received: 25 September 2021 Accepted: 4 November 2021 Published: 6 November 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

This disease is particularly heterogeneous [3]. They are classified as high-grade and lowgrade diseases based on standardized histomorphological features, as described by the World Health Organization. The depth of an invasion in the bladder wall determines the tumor stage. Approximately 80% of BLCA patients present non-muscle-invasive bladder cancer (NMIBC) at the time of diagnosis, while the remainder present muscle-invasive bladder cancer (MIBC) or even distant metastases [4]. NMIBCs do not normally pose a threat to patient survival and have a much better prognosis due to effective therapeutic options [5]. However, they almost always relapse, and patients need to repeat intravesical treatments, endoscopic evaluations, and biopsies, which may take an extended period of time, resulting in expensive surgical and surveillance management [6–8]. MIBCs, on the other hand, are clinically aggressive and can progress rapidly to lymph nodes, brain, lungs, liver, and bone metastases, which are often fatal [3]. However, over the past three decades, clinical management and five-year survival rates have seen few substantial advances [9]. Therefore, it is significant to identify novel biomarkers and molecular targets for advancing the prognosis of BLCA.

Carboxypeptidase A4 (CPA4) is a member of the zinc-containing metallocarboxypeptidase family [10], which could specifically catalyze the peptide bonds released from carboxy-terminal amino acids [11,12]. CPA4 was first discovered when screening for upregulated mRNA during cancer cell differentiation induced by sodium butyrate [13]. From the cellular and biochemical characteristics, CPA4 is secreted from cells in the form of soluble proenzyme (pro-CPA4), which might play a role in creating a tumor microenvironment [10]. Previous studies have demonstrated that CPA4 is closely associated with the aggressiveness, growth, and differentiation in cancer cells [14,15]. However, the underlying mechanism of CPA4 in BLCA remains unclear.

Recently, CPA4 has shown the potential to be a biomarker in the early diagnosis for certain cancers. Sun et al. have reported that the higher expression level of CPA4 in pancreatic cancer tissues and serum is related to poor prognosis and higher aggressiveness [13]. Previously studied showed that upregulated mRNA levels of CPA4 in androgen-independent prostate cancer cells is associated with the Histone Hyperacetylation signaling pathway [16]. In liver cancer and lung cancer, studies have also shown that the higher expression of CPA4 was closely associated with early diagnosis and poor prognosis [13,17]. Despite the potential significance of CPA4 expression in plenty types of cancer, no previous studies have ever shown the expression levels of CPA4 in bladder cancer, especially with regard to its potential therapeautic and prognostic values. Additionally, the correlation with immune infiltrates of CPA4 in BLCA remains to be investigated. Shao et al. demonstrated that CPA4 overexpression promotes the progression of aggressive clinical stage in pancreatic cancer and that the downregulation of CPA4 inhibits non-small-cell lung cancer growth [15,18]. Therefore, we hypothesized that the level of CPA4 is associated with the prognosis and immune cell infiltration in BLCA.

To test this hypothesis, our study evaluated the role of CPA4 on tumorigenesis and clinical significance based on The Cancer Genome Atlas (TCGA). We compared the different expression level of BLCA in age; gender; pathologic T, N, and M stage; pathology; subtype; and OS. In this study, we found that CPA4 is upregulated in BLCA. Significantly, the risk factors of CPA4 upregulation are correlated with poor prognosis. Additionally, the correlation with immune infiltrates of CPA4 for BLCA is also evaluated. Eventually, we link high CPA4 levels and poor prognosis in BLCA.

#### **2. Materials and Methods**

#### *2.1. Data Source*

The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/, accessed on 7 September 2021) provides 33 types of clinical and pathological information on cancer for scholars and researchers for free [19]. The expression profiles of CPA4 and clinical information of TCGA cancer data were downloaded from the UCSC Xena (https://xenabrowser.net/datapages/, accessed on 7 September 2021) database. The

TCGA database is available publicly in open access format and is available where ethical approval and informed consent of the patients were not necessary [20].

#### *2.2. CPA4 Methylation Level Analysis*

UALCAN (http://ualcan.path.uab.edu/, accessed on 6 September 2021) is a comprehensive, user-friendly, and interactive web resource for analyzing cancer OMICS data and provides graphs and plots depicting expression profiles and patient survival information for protein-coding, miRNA-coding, and lincRNA-coding genes [21]. The UALCAN online tool was utilized to analyze the CPA4 methylation level in BLCA (TCGA data).

#### *2.3. Analysis of Differentially Expressed Genes (DEGs)*

Through the limma Package by R, patients with different CPA4 expression profiles in the high and low expression groups (HTSeq-TPM) were compared using unpaired Student's *t*-test to identify the DEGs [22]. A |log2Fold Change| > 2 and BH-adjusted *p*-values < 0.05 were considered the threshold for the DEGs in a Gene Ontology (GO) Enrichment Analysis. Metascape (https://metascape.org, accessed on 7 September 2021) is a tool used for gene annotation and pathway analysis [23]. In this study, Metascape was utilized to analyze the enrichment of CPA4-related DEGs in processes and pathways. A *p*-value < 0.01, a minimum count of 3, and an enrichment factor of > 1.5 were regarded as significant [24].

## *2.4. Gene Set Enrichment Analysis (GSEA)*

GSEA was used as a statistical method in order to seek out whether gene exhibits are statistically significant and concordant between two biological states [25]. We used the R package Cluster Profiler to evaluate excessive function and pathway differences between groups with different expression of CPA4 expression [26]. Each analysis of the processes was repeated 1000 times. Adjusted *p*-value < 0.05 and false discovery rate (FDR) < 0.25 were considered statistically significant enrichments [27]. We chose the potential pathway in which FDR < 0.05 with higher NES after analysis.

#### *2.5. Comprehensive Analysis of Protein–Protein Interaction*

The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) website (https://string-db.org/, accessed on 7 September 2021) is a database of known and predicted protein–protein interactions that hosts a collection of integrated and consolidated protein–protein interaction data including direct (physical) and indirect (functional) associations [28]. By importing CPA4 into the online tool STRING, protein–protein interaction (PPI) network information was compiled. Confidence scores > 0.4 were considered median significant.

#### *2.6. Analysis of the Tumor Immune Estimation Resource (TIMER2)*

The Tumor Immune Estimation Resource (TIMER2) is a comprehensive resource including 32 cancer types and incorporates 10,897 samples from the TCGA database for systematically analysis of immune infiltrates across diverse cancer types (http://cistrome. org/TIMER/, accessed on 7 September 2021) [29]. The TIMER2 database is used to evaluate the correlation of the expression of CPA4 in BLCA patients with the six types of infiltrating immune cells (B cells, dendritic cells, CD4 + T cells, CD8 + T cells, macrophages, and neutrophils) and displays the relationship between the expression of the CPA4 gene and the tumor purity.

## *2.7. Univariate and Multivariate Logistic Regression Analysis*

Univariate Cox regression used to calculate the association between OS and patients' CPA4 expression in two cohorts aims at further researching the effect of CPA4 expression. A multivariate analysis was used to assess if CPA4 is an independent prognostic factor for

BLCA patient survival. CPA4 is statistically significant in the Cox regression analysis when the *p*-value is less than 0.05 [30].

#### *2.8. Identification of CPA4 Coexpression Genes and Construction of a Prognostic Nomogram*

cBiopor tal (https://www.cbioportal.org/, accessed on 7 September 2021) (an online tool based on the TCGA database) was used to identify sets of coexpression genes. According to the *p*-value, we select the most relevant genes about CPA4. Then, the clinical factors (T, M, and N stages; radiation therapy; and primary therapy outcome) and the gene expression levels were used to construct a prognostic nomogram to evaluate the probability of 1-, 2-, and 3-year OS for BLCA patients via the R package (https://cran.r-project.org/web/packages/rms/, accessed on 7 September 2021) [31].

## *2.9. Immune Infiltration Analysis by ssGSEA*

Single sample GSEA (ssGSEA) was performed to analyze the state of immune infiltration of BLCA from R package GSVA (version3.6) (http://www.bioconductor.org/ packages/release/bioc/html/GSVA.Html, accessed on 8 September 2021), and we quantified the infiltration levels of 24 immune cell types from gene expression profiles in the literature [32]. In order to discover the correlation between CPA4 and the infiltration levels of 24 immune cells, adjusted *p*-values were established by the Spearman and Wilcoxon rank-sum tests.

#### *2.10. Analysis of the Gene Expression Profiling Interactive Analysis 2*

The Gene Expression Profiling Interactive Analysis2 (GEPIA2) (http://gepia.cancerpku.cn/index.html, accessed on 7 September 2021) is an updated database used for analyzing the RNA sequencing expression data of 9736 tumors and 8587 normal samples from the TCGA and the GTEx projects, which include 60,498 genes and 198,619 isoforms [33]. GEPIA2 database investigated the expression level of CPA4 with various immune cells' markers. TIMER2 was used to identify the gene with a significant correlation with CPA4 expression in the GEPIA2 web.

#### *2.11. Statistical Analysis*

The expression of CPA4 for non-paired and paired samples was analyzed by the Wilcoxon rank-sum test and Wilcoxon signed-rank test, respectively. By using the pROC package, the ROC curve was generated to evaluate the CPA4 expression with diagnostic performance. The relations between the CPA expression and the clinical features were analyzed by the Kruskal–Wallis test, Chi-Squared test, and Wilcoxon signed rank test. The survival curves were generated via the long-rank test for the Kaplan–Meier analysis. *p* < 0.05 was considered statistically significant: \* *p* < 0.05, \*\* *p* < 0.01, and \*\*\* *p* < 0.001; R software was used to process all kinds of statistical analyses (Version 4.0.2). In R, we use *p*adj = *p*.adjust (*p*, method = "BH", *n* = length(*p*)) to correct the *p*-value.

#### **3. Results**

## *3.1. Characteristics of BLCA Patients*

In total, the information for 414 BLCA tumor tissues and 19 normal tissues were collected from the TCGA database including RNA-seq and relative clinical prognostic information in 414 patients. We grouped the BLCA patients into two sets: low (*n* = 207) and high expressions (*n* = 207) of CPA4. The clinical information of BLCA patients includes age, race, gender, pathologic stage, pathologic stage (T, N, or M), pathologic stage, primary therapy outcome, histologic grade, radiation therapy, subtype, smoking status, lymphovascular invasion, and OS event (Table 1).


**Table 1.** Clinical characteristics of two sets of patients with different expressions of CPA4 from the TCGA dataset.

#### *3.2. Tumor Tissues Express Higher CPA4 Than Normal Tissue*

The expression of CPA4 in pan-cancer was analyzed between tumor and normal tissues. From the TCGA + GETx database, the expression level of CPA4 in non-matched patients (*<sup>p</sup>* = 1.6 <sup>×</sup> <sup>10</sup>−<sup>5</sup> ) was significantly higher than that in normal people (Figure 1). The analysis of the correlation between CPA4 expression in BLCA patients and relative clinical information shows that a higher DLEU1 expression level is correlated with OS events

and the subtype papillary. No statistically significant differences were found between the expression levels of CPA4 in BLCA and age; gender; pathological T, N, or M stages; and pathologic stage. the subtype papillary. No statistically significant differences were found between the expression levels of CPA4 in BLCA and age; gender; pathological T, N, or M stages; and pathologic stage.

The expression of CPA4 in pan-cancer was analyzed between tumor and normal tissues. From the TCGA + GETx database, the expression level of CPA4 in non-matched patients (*p* = 1.6 × 10−5) was significantly higher than that in normal people (Figure 1). The analysis of the correlation between CPA4 expression in BLCA patients and relative clinical information shows that a higher DLEU1 expression level is correlated with OS events and

*Biology* **2021**, *10*, x FOR PEER REVIEW 6 of 18

*3.2. Tumor Tissues Express Higher CPA4 Than Normal Tissue* 

**Figure 1.** CPA4 expression and clinicopathological features in BLCA. (**a**) human CPA4 expression levels in different cancer tissues and corresponding normal tissues. (**b**) The expression level of CPA4 in BLCA tissue was significantly higher compared with the normal tissues from the TCGA + GTEx database. (**c**–**g**) No statistically significant differences were found between the expression levels of CPA4 in BLCA and age; gender; and pathological T, N, or M stage. (**h**–**j**) High pathologic stage, higher dead event, and nonpapillary were associated with higher expressions of CPA4 in BLCA. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001 **Figure 1.** CPA4 expression and clinicopathological features in BLCA. (**a**) human CPA4 expression levels in different cancer tissues and corresponding normal tissues. (**b**) The expression level of CPA4 in BLCA tissue was significantly higher compared with the normal tissues from the TCGA + GTEx database. (**c**–**g**) No statistically significant differences were found between the expression levels of CPA4 in BLCA and age; gender; and pathological T, N, or M stage. (**h**–**j**) High pathologic stage, higher dead event, and nonpapillary were associated with higher expressions of CPA4 in BLCA. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; ns: no significance.

#### *3.3. Impact of High CPA4 Expression on the Detection and Prognosis of BLCA Patients*

The expression of CPA4 indicated a significant discriminative power in identifying tumors from normal cells with an AUC value of 0.798 (Figure 2d). The Kaplan–Meier survival analysis showed that BLCA patients with higher CPA4 expressions have shorter overall survival, disease-specific survival, and progress-free intervals (Figure 2a–c). The KM plots show that a higher expression of CPA4 had a worse prognosis than a lower expression. Promoter methylation of CPA4 in the TCGA-BLCA data was significantly

*Biology* **2021**, *10*, x FOR PEER REVIEW 7 of 18

lower than that of normal tissues adjacent to cancer in the UALCAN webpage (*p* < 0.001; Figure 2e). expression. Promoter methylation of CPA4 in the TCGA-BLCA data was significantly lower than that of normal tissues adjacent to cancer in the UALCAN webpage (*p* < 0.001; Figure 2e).

The expression of CPA4 indicated a significant discriminative power in identifying tumors from normal cells with an AUC value of 0.798 (Figure 2d). The Kaplan–Meier survival analysis showed that BLCA patients with higher CPA4 expressions have shorter overall survival, disease-specific survival, and progress-free intervals (Figure 2a–c). The KM plots show that a higher expression of CPA4 had a worse prognosis than a lower

*3.3. Impact of High CPA4 Expression on the Detection and Prognosis of BLCA Patients* 

**Figure 2.** (**a**–**c**) Kaplan–Meier survival curves comparing high and low expressions of CPA4 in BLCA patients. (**a**) overall survival; (**b**) disease-specific survival; (**c**) progression-free interval; (**d**) ROC analysis of CPA4 indicates promising discrimination power between tumor and normal tissues; (**e**) the promoter methylation of CPA4 in tumor tissues (*n* = 418) and normal tissues (*n* = 21) from TCGA-BLCA data. **Figure 2.** (**a**–**c**) Kaplan–Meier survival curves comparing high and low expressions of CPA4 in BLCA patients. (**a**) overall survival; (**b**) disease-specific survival; (**c**) progression-free interval; (**d**) ROC analysis of CPA4 indicates promising discrimination power between tumor and normal tissues; (**e**) the promoter methylation of CPA4 in tumor tissues (*n* = 418) and normal tissues (*n* = 21) from TCGA-BLCA data.

#### *pression Samples*  We analyzed the DEGs in altered expressions of CPA4 including in low and high *3.4. Differentially Expressed Genes and GO Enrichment Analysis in High- and Low-CPA4 Expression Samples*

*3.4. Differentially Expressed Genes and GO Enrichment Analysis in High- and Low-CPA4 Ex-*

samples to explore the potential mechanisms of CPA4 that promote tumor progression. There were 529 DEGs identified, of which 349 genes were upregulated and 180 were downregulated (|log2(FC)| > 2 and *p*.adj < 0.05). The DEGs's expression is shown in a heat We analyzed the DEGs in altered expressions of CPA4 including in low and high samples to explore the potential mechanisms of CPA4 that promote tumor progression. There were 529 DEGs identified, of which 349 genes were upregulated and 180 were downregulated (|log2(FC)| > 2 and *p*.adj < 0.05). The DEGs's expression is shown in a heat map and volcano plot (Figure 3) using GO enrichment analysis to predict the co-expression functions in patients with BLCA. The top GO enrichment items in the biological process (BP), molecular function (MF), and cellular component (CC) groups were epidermal cell differentiation, keratinocyte differentiation, keratinization, intermediate filament cytoskeleton, intermediate filament, cornified envelope, endopeptidase inhibitor activity, peptidase inhibitor activity, peptidase inhibitor activity, serine-type endopeptidase inhibitor activity, metabolism of xenobiotics by cytochrome P450, drug metabolism-cytochrome P450, and retinol metabolism (Figure 4a).

chrome P450, and retinol metabolism (Figure 4a).

**Figure 3.** (**a**) Volcano plot of differentially expressed genes (DEGs) connected with the expression of CPA4; (**b**) heatmap of differentially expressed genes (DEGs) connected with the expression of CPA4. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001. **Figure 3.** (**a**) Volcano plot of differentially expressed genes (DEGs) connected with the expression of CPA4; (**b**) heatmap of differentially expressed genes (DEGs) connected with the expression of CPA4. \*\*\* *p* < 0.001.

#### *3.5. Gene Set Enrichment Analysis for CPA4-Related Signaling Pathways 3.5. Gene Set Enrichment Analysis for CPA4-Related Signaling Pathways*

By the enrichment of MSigDB Collection (c2.all.v7.0.symbols.gmt (curated)), we used the GSEA to identify signaling pathways associated with CPA4 between the different expression levels of CPA4 with significant differences (adjusted *p*-value < 0.05 and FDR < 0.25). The eight pathways included the formation of the cornified envelope, keratinization, immunoregulatory interactions between a lymphoid and a non-lymphoid cell, wp hair follicle development cytodifferentiation part 3 of 3, antigen processing and presentation, assembly of collagen fibrils and other multimeric structures, graft versus host disease, and cytokine–cytokine receptor interaction (Figure 4). By the enrichment of MSigDB Collection (c2.all.v7.0.symbols.gmt (curated)), we used the GSEA to identify signaling pathways associated with CPA4 between the different expression levels of CPA4 with significant differences (adjusted *p*-value < 0.05 and FDR < 0.25). The eight pathways included the formation of the cornified envelope, keratinization, immunoregulatory interactions between a lymphoid and a non-lymphoid cell, wp hair follicle development cytodifferentiation part 3 of 3, antigen processing and presentation, assembly of collagen fibrils and other multimeric structures, graft versus host disease, and cytokine–cytokine receptor interaction (Figure 4).

map and volcano plot (Figure 3) using GO enrichment analysis to predict the co-expression functions in patients with BLCA. The top GO enrichment items in the biological process (BP), molecular function (MF), and cellular component (CC) groups were epidermal cell differentiation, keratinocyte differentiation, keratinization, intermediate filament cytoskeleton, intermediate filament, cornified envelope, endopeptidase inhibitor activity, peptidase inhibitor activity, peptidase inhibitor activity, serine-type endopeptidase inhibitor activity, metabolism of xenobiotics by cytochrome P450, drug metabolism-cyto-

#### *3.6. CPA4 Expression Predicts Poor Prognosis in Different Cancer Stages*

Univariate cox proportional-hazards model analysis showed that high CPA4 expression, high pathologic grade and stage (T, N, and M), and subtype papillary were negative predictors for OS in BLCA patients. Meanwhile, in the multivariate regression analysis, CPA4 expression was an independent factor correlated with OS both in the low-expression set and high-expression set (*p* = 0.003) (Figure 5).

**Figure 4.** (**a**) GO enrichment analysis of differentially expressed genes (DEGs) in high- and low-CPA4 expression samples; (**b**,**c**) enrichment plots from GSEA. Several pathways were differentially enriched in BLCA patients according to different CPA4 expressions; (**b**) formation of the cornified envelope; (**c**) keratinization; (**d**) immunoregulatory interactions between a lymphoid and a non-lymphoid cell; (**e**) WP hair follicle development cytodifferentiation part 3 of 3; (**f**) antigen processing and presentation; (**g**) assembly of collagen fibrils and other multimeric structures; (**h**) graft versus host disease; (**i**) cytokine–cytokine receptor interaction. ES, enrichment score; NES, normalized enrichment score; ADJ *p*-Val, adjusted *p*-value; FDR, false discovery rate. **Figure 4.** (**a**) GO enrichment analysis of differentially expressed genes (DEGs) in high- and low-CPA4 expression samples; (**b**,**c**) enrichment plots from GSEA. Several pathways were differentially enriched in BLCA patients according to different CPA4 expressions; (**b**) formation of the cornified envelope; (**c**) keratinization; (**d**) immunoregulatory interactions between a lymphoid and a non-lymphoid cell; (**e**) WP hair follicle development cytodifferentiation part 3 of 3; (**f**) antigen processing and presentation; (**g**) assembly of collagen fibrils and other multimeric structures; (**h**) graft versus host disease; (**i**) cytokine– cytokine receptor interaction. ES, enrichment score; NES, normalized enrichment score; ADJ *p*-Val, adjusted *p*-value; FDR, false discovery rate.

*3.6. CPA4 Expression Predicts Poor Prognosis in Different Cancer Stages* 

Univariate cox proportional-hazards model analysis showed that high CPA4 expression, high pathologic grade and stage (T, N, and M), and subtype papillary were negative predictors for OS in BLCA patients. Meanwhile, in the multivariate regression analysis, CPA4 expression was an independent factor correlated with OS both in the low-expres-

sion set and high-expression set (*p* = 0.003) (Figure 5).


**Figure 5.** Univariate (**a**) and multivariate (**b**) regression analyses of CPA4 and other clinicopathologic parameters with OS in BLCA patients. **Figure 5.** Univariate (**a**) and multivariate (**b**) regression analyses of CPA4 and other clinicopathologic parameters with OS in BLCA patients.

#### *3.7. Construction of Nomogram for Predicting OS and Validation by Calibration 3.7. Construction of Nomogram for Predicting OS and Validation by Calibration*

We constructed a nomogram for predicting the prognosis of BLCA with relative clinical situation, which integrates the clinical characteristics associated with the survival of BLCA. Based on the multivariate Cox analysis, a nomogram was assigned to the clinical characteristics of a point and the sum of points awarded to each characteristic is a point from 0 to 100. All of the points are accumulated and recorded as the total points. Using the absolute point axis down to the outcome axis, the probability of BLCA survival at 1, 3 and 5 years can be determined (Figure 6a). From the nomogram, the expression of CPA4 contributes many points compared with other relative clinical situations including the T, N, and M stages; radiation therapy; and primary therapy outcome. Meanwhile, the calibration plot indicates great agreement between the predicted and observed values, which are close to the 45-degree line, which is the ideal curve (Figure 6b). We constructed a nomogram for predicting the prognosis of BLCA with relative clinical situation, which integrates the clinical characteristics associated with the survival of BLCA. Based on the multivariate Cox analysis, a nomogram was assigned to the clinical characteristics of a point and the sum of points awarded to each characteristic is a point from 0 to 100. All of the points are accumulated and recorded as the total points. Using the absolute point axis down to the outcome axis, the probability of BLCA survival at 1, 3 and 5 years can be determined (Figure 6a). From the nomogram, the expression of CPA4 contributes many points compared with other relative clinical situations including the T, N, and M stages; radiation therapy; and primary therapy outcome. Meanwhile, the calibration plot indicates great agreement between the predicted and observed values, which are close to the 45-degree line, which is the ideal curve (Figure 6b).

**Figure 6.** The relationship of CPA4 expression with other clinical factors and overall survival (OS). (**a**) Nomogram for predicting the probability of 1-, 3-, and 5-year OS for BLCA patients; (**b**) calibration plot of the nomogram for predicting the OS likelihood. **Figure 6.** The relationship of CPA4 expression with other clinical factors and overall survival (OS). (**a**) Nomogram for predicting the probability of 1-, 3-, and 5-year OS for BLCA patients; (**b**) calibration plot of the nomogram for predicting the OS likelihood. **Figure 6.** The relationship of CPA4 expression with other clinical factors and overall survival (OS). (**a**) Nomogram for predicting the probability of 1-, 3-, and 5-year OS for BLCA patients; (**b**) calibration plot of the nomogram for predicting the OS likelihood.

#### *3.8. CPA4-Interaction Protein Networks in BLCA Tissue 3.8. CPA4-Interaction Protein Networks in BLCA Tissue 3.8. CPA4-Interaction Protein Networks in BLCA Tissue*

CPA4-interaction protein networks were constructed to further explore the necessary proteins for metabolism and the molecular mechanism used by STRING. The PPI network of the CPA4 protein showed the relationship of the CPA4 protein in the progression of BLCA. Ten proteins and corresponding gene names were listed with their annotation scores (Figure 7). The top 10 genes included LXN, CMA1, SGCE, TPSAB1, AGBL2, TPSB2, PEG10, GRB10, TSGA13, and MEST, and LXN had the highest score. CPA4-interaction protein networks were constructed to further explore the necessary proteins for metabolism and the molecular mechanism used by STRING. The PPI network of the CPA4 protein showed the relationship of the CPA4 protein in the progression of BLCA. Ten proteins and corresponding gene names were listed with their annotation scores (Figure 7). The top 10 genes included LXN, CMA1, SGCE, TPSAB1, AGBL2, TPSB2, PEG10, GRB10, TSGA13, and MEST, and LXN had the highest score. CPA4-interaction protein networks were constructed to further explore the necessary proteins for metabolism and the molecular mechanism used by STRING. The PPI network of the CPA4 protein showed the relationship of the CPA4 protein in the progression of BLCA. Ten proteins and corresponding gene names were listed with their annotation scores (Figure 7). The top 10 genes included LXN, CMA1, SGCE, TPSAB1, AGBL2, TPSB2, PEG10, GRB10, TSGA13, and MEST, and LXN had the highest score.

**Figure 7.** CPA4-interaction proteins in BLCA tissue; annotation of CPA4-interacting proteins and their co-expression scores. **Figure 7.** CPA4-interaction proteins in BLCA tissue; annotation of CPA4-interacting proteins and their co-expression scores. **Figure 7.** CPA4-interaction proteins in BLCA tissue; annotation of CPA4-interacting proteins and their co-expression scores.

#### *3.9. Correlation Analysis between CPA4 Expression and Infiltrating Immune Cells 3.9. Correlation Analysis between CPA4 Expression and Infiltrating Immune Cells*

The survival of patients with different cancers including BLCA is associated with the tumor-infiltrating immune cells. From the result, the expression level of CPA4 had significant correlations with CD8+ T cells (r = 0.287, *<sup>p</sup>* = 2.29 <sup>×</sup> <sup>10</sup>−<sup>8</sup> ), B cells (r = 0.218, *<sup>p</sup>* = 8.65 <sup>×</sup> <sup>10</sup>−10), neutrophils (r = 0.196, *<sup>p</sup>* = 1.76 <sup>×</sup> <sup>10</sup>−<sup>4</sup> ), and dendritic cells (r = 0.356, *<sup>p</sup>* = 2.5 <sup>×</sup> <sup>10</sup>−12). *<sup>p</sup>* < 0.05 was considered significant (Figure 8a). Furthermore, we analyzed 24 immune cells including pDC, NK CD56bright cells, DC, cytotoxic cells, TFH, B cells, CD8 T cells, Th17 cells, Treg, T cells, mast cells, iDC, NK cells, Tem, aDC, neutrophils, Th1 cells, NK CD56dim cells, macrophages, eosinophils, Tgd T helper cells, Th2 cells, and Tcm. We analyzed the correlation between the expression of CPA4 and immune infiltration by ssGSEA using Spearman's R. From the result, the expression level of CPA4 was negatively correlated with the infiltration levels of NK CD56bright cells (*p* < 0.001) and positively correlated with cytotoxic cells, T cells, NK cells, idc, Tem, Treg, aDC, Neutrophils, NK CD56dim cells, macrophages, Th2 cells, and Th1 cells (Figure 8). The survival of patients with different cancers including BLCA is associated with the tumor-infiltrating immune cells. From the result, the expression level of CPA4 had significant correlations with CD8+ T cells (r = 0.287, *p* = 2.29 × 10−8), B cells (r = 0.218, *p* = 8.65 × 10−10), neutrophils (r = 0.196, *p* = 1.76 × 10−4), and dendritic cells (r = 0.356, *p* = 2.5 × 10−12). *p* < 0.05 was considered significant (Figure 8a). Furthermore, we analyzed 24 immune cells including pDC, NK CD56bright cells, DC, cytotoxic cells, TFH, B cells, CD8 T cells, Th17 cells, Treg, T cells, mast cells, iDC, NK cells, Tem, aDC, neutrophils, Th1 cells, NK CD56dim cells, macrophages, eosinophils, Tgd T helper cells, Th2 cells, and Tcm. We analyzed the correlation between the expression of CPA4 and immune infiltration by ssGSEA using Spearman's R. From the result, the expression level of CPA4 was negatively correlated with the infiltration levels of NK CD56bright cells (*p* < 0.001) and positively correlated with cytotoxic cells, T cells, NK cells, idc, Tem, Treg, aDC, Neutrophils, NK CD56dim cells, macrophages, Th2 cells, and Th1 cells (Figure 8).

**Figure 8.** The expression level of CPA4 was related to immune infiltration in the tumor microenvironment. (**a**) Correlation of CPA4 expression with infiltrating immune infiltration in BLCA (**b**) The forest plot shows the correlation between CPA4 expression level and 24 immune cells. The size of the dots indicates the absolute value of Spearman's R. (**c**,**d**) The Wilcoxon rank sum test was used to analyze the difference in the macrophage cell infiltration levels between the CPA4 high- and low-expression groups; (**e**,**f**) the correlation between CPA4 expression and NK CD56 bright cell infiltration levels. **Figure 8.** The expression level of CPA4 was related to immune infiltration in the tumor microenvironment. (**a**) Correlation of CPA4 expression with infiltrating immune infiltration in BLCA (**b**) The forest plot shows the correlation between CPA4 expression level and 24 immune cells. The size of the dots indicates the absolute value of Spearman's R. (**c**,**d**) The Wilcoxon rank sum test was used to analyze the difference in the macrophage cell infiltration levels between the CPA4 high- and low-expression groups; (**e**,**f**) the correlation between CPA4 expression and NK CD56 bright cell infiltration levels. \*\*\* *p* < 0.001.
