Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Prognostic value of iron metabolism-related genes in bladder urothelial carcinoma

Prognostic value of iron metabolism-related genes in bladder urothelial carcinoma IntroductionBladder cancer is a common malignant tumour of the urinary system with an incidence ranking 9th globally and 13th in China [1, 2]. Bladder cancer is often caused by occupational exposure to heavy metals, such as arsenic and cadmium, and is influenced by factors such as smoking and race [3], [4], [5]. The clinical manifestations are mainly painless gross haematuria. When the lesions involve the bladder triangle, bladder irritation signs appear, including frequent urination, urgent urination, and pain. The primary screening methods for bladder cancer are urine exfoliation cytology and cystoscopy [6]. Surgical intervention, systemic chemotherapy, and radical resection are preferred standard treatments for most early bladder cancer patients [7]. For patients with advanced bladder cancer or those who are intolerant to surgical treatment, systemic chemotherapy is preferred.An in-depth understanding of the underlying molecular mechanisms of urological cancers can provide entry points for improving treatments and developing new therapeutic approaches. Recently, numerous studies have shown that ferroptosis due to abnormal iron metabolism plays an important role in the occurrence and progression of cancer [8].In our study, IMRGs were obtained from the Molecular Signatures Database. Gene expression profiles and clinical information from 414 BLCA and 19 normal samples were obtained from The Cancer Genome Atlas (TCGA) database. The expression of IMRGs in BLCA was verified in the Gene Expression Omnibus (GEO) database. A BLCA risk-scoring model was established from the identified IMRGs; this model was tested on TCGA-BLCA cohort. Furthermore, the potential functions and mechanisms of IMRGs in BLCA were explored by functional enrichment analysis and GSEA. Our results indicated that the risk-scoring model comprising 14 iron metabolism-related gene markers could effectively predict the overall OS outcomes of BLCA patients.Materials and methodsAcquisition of IMRGsIMRGs were searched in MSigDB (Table S1), and 535 were obtained (Table S2).Acquisition and analysis of gene expression files of BLCA patientsGene expression profiles and clinical data from 414 BLCA and 19 normal samples were obtained from TCGA database. The expression of IMRGs in BLCA was tested in dataset GSE13507. We excluded BLCA patients without available survival data as this study required a prognostic analysis.Screening and identification of IMRGsFirst, “DESeq2” was used to analyse the DEGs between BLCA and normal bladder tissues in TCGA database [9]. The DEGs screening parameters were as follows: adjust p<0.05 and |log 2-fold change| >1. In total, 535 IMRGs were analysed, and the DEGs were displayed in a Venn diagram. A total of 118 IMRGs were selected for subsequent study. GSE13507 was used to test IMRGs expression. We constructed a heatmap of the 118 IMRGs using the “ComplexHeatmap” package. “clusterProfiler”, “org.Hs.eg.db”, and “GOplot” packages were adopted for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses and the visualisation of the 118 IMRGs [10, 11].Creation and testing of risk-scoring systemFirst, we combined the BLCA patients gene expression and OS to conduct Cox analysis. Genes with p<0.05 were included in subsequent analyses. LASSO algorithm was executed on the 25 IMRGs obtained in the above steps using the “glmnet” and “survival” packages to screen variables and build a prognostic model. Fourteen genes significantly related to OS were obtained and used to construct the prognostic model. The risk-scoring model was generated using the following formula: Risk score=∑ni=1 exprgenei coefficientgenei [12]. Gene expression normalisation was adapted using the “edgeR” package. We adopted the “ggplot2” package to generate the risk factor plot. ROC curves were generated using “timeROC” package. High- and low-risk groups were divided by the median risk score. The “survminer” package was used to generate the survival curves. Furthermore, as described by Meng J and Lu X, we verified the prognostic value of the above 14 iron metabolisation-related genes using the external GSE13507 and GSE32894 datasets [13, 14]. We elucidated the correlation between clinicopathological characteristics and risk scores through dot plots generated using “ggplot2”.Construction and assessment of the nomogramWe included the risk group as a parameter and each clinicopathological characteristic in Cox regression analyses to determine the predictive ability of the risk-scoring system for BLCA patients. These clinicopathological characteristics included the following: T stage, N stage, M stage, gender, pathological stage, histological grade, lymphovascular invasion, and subtype. The independent parameters affecting the prognosis of BLCA patients identified above were used to establish a nomogram using the “rms” package. The predictive value of the nomogram was tested by calibration curve and time-ROC analyses.GSEADEGs between risk groups were determined by the criteria and methods described above in TCGA-BLCA cohort. We employed the “clusterProfiler” package for GSEA and the “ggplot2” package to visualise the results.Immune infiltration analysisImmune infiltration analysis was conducted and visualised using the “GSVA” package [15, 16].Statistical analysisR software (version 3.6.3) was employed for data analysis and visualisation in our research. K–M survival analysis was performed using Cox regression analysis. Correlation analysis was performed using Spearman’s rank sum correlation test. The comparison between groups of qualitative data was performed using Chi-square analysis, and the comparison between groups of quantitative data was performed using an independent-samples T test. A p value or adjusted p value<0.05 was considered to represent significance.ResultsScreening of IMRGs in BLCAWe screened 8,702 DEGs in 414 TCGA-BLCA and 19 normal bladder tissues using the DESeq2 algorithm. The 535 IMRGs obtained from MSigDB and the 8,702 DEGs were subjected to Venn analysis. Ultimately, we obtained 118 IMRGs associated with BLCA (Figure 1A). Figure 1B shows the expression of the above 118 IMRGs in the GSE13507 dataset. A comparison of the heatmaps of these 118 genes (Figure S1) revealed consistent change trends across most genes. We conducted an enrichment analysis of the 118 IMRGs, revealing enrichment in DNA polymerase complex, iron ion transport, vacuolar proton-transporting V-type ATPase complex, iron ion homeostasis, transition metal ion transport, heme binding, proton-transporting V-type ATPase complex, iron ion binding, and tetrapyrrole binding (Figure 1D). Arachidonic acid metabolism, chemical carcinogenesis, retinol metabolism, and linoleic acid metabolism were all significantly enriched in the KEGG analysis (Figure 1C). The enrichment analysis results are combined in Table S3.Figure 1:Identification and enrichment analysis of IMRGs in the TCGA-BLCA cohort. (A) Venn diagram of the intersection between IMRGs and DEGs identified by the DESeq2 algorithm. (B) Heatmap of 118 DEGs related to iron metabolism in the GSE13507 data set. Terms of GO enrichment analysis (C) and KEGG pathways (D) related to the 118 IMRGs.Establishment and evaluation of risk-scoring systemFirst, we identified the associations of the 118 IMRGs with the outcomes of BLCA patients using Cox regression analysis. Using Cox p<0.05 as the screening criterion, we screened 25 genes that could potentially predict OS outcomes in BLCA patients (Table S4). Then, variables were further filtered using LASSO regression analysis (Figure 2A, B). We constructed the risk-scoring system using the 14 genes that remained (Table 1). K–M curves of the above 14 genes are shown in Figure S2. We also analysed the tissue expression of the above 14 genes using the Human Protein Atlas (HPA) database immunohistochemical results (Figure S3A–I).Figure 2:Visualization of DEGs with univariate Cox regression p value <0.05. (A) The LASSO regression model of the 25 IMRGs performed by LASSO tenfold cross-validation (lambda.min: lambda 0.023673, Index 18, statistic 11.771, SE 0.17651). (B) The coefficient distribution in the LASSO regression model.Table 1:Fourteen IMRGs identified by univariate Cox regression analysis.GeneDescriptionHazard ratio (95 % CI)p-ValueCYP2W1Cytochrome P450, family 2, subfamily W, polypeptide 11.599 (1.189–2.149)0.002PPEF1Protein phosphatase, EF-hand calcium binding domain 11.487 (1.104–2.002)0.009HEPHHephaestin1.396 (1.036–1.879)0.028FA2HFatty acid 2-hydroxylase0.636 (0.473–0.854)0.003CYP27B1Cytochrome P450, family 27, subfamily B, polypeptide 10.709 (0.529–0.950)0.021ATP6V1B1ATPase, H+ transporting, lysosomal 56/58 kDa, V1 subunit B11.459 (1.086–1.959)0.012PTGISProstaglandin I2 (prostacyclin) synthase1.693 (1.254–2.286)<0.001CCNB1Cyclin B11.411 (1.053–1.892)0.021MYCv-myc avian myelocytomatosis viral oncogene homologue1.610 (1.197–2.165)0.002CYP1B1Cytochrome P450, family 1, subfamily B, polypeptide 11.549 (1.153–2.082)0.004CYP2C9Cytochrome P450, family 2, subfamily C, polypeptide 90.725 (0.541–0.972)0.032CYP2C8Cytochrome P450, family 2, subfamily C, polypeptide 80.683 (0.509–0.917)0.011AIFM3Apoptosis-inducing factor, mitochondrion-associated, 30.579 (0.430–0.780)<0.001ASIC3Acid sensing (proton gated) ion channel 30.731 (0.545–0.981)0.037We calculated a risk score for each patient in TCGA-BLCA cohort based on the abovementioned risk-scoring system. The risk factor plots for TCGA-BLCA cohort, including survival time distribution, risk score distribution, and the expression levels of the 14 genes, are shown in Figure 3A. The time-ROC curves of this risk-scoring system for predicting BLCA patient OS outcomes are shown in Figure 3B (1-, 3-, and 5-year AUCs: 0.705, 0.725, and 0.714). The K–M curve indicates that the prognosis of the high-risk group was significantly poor (p<0.001; Figure 3C). The verification of the prognostic value of the 14 genes in the GSE13507 and GSE32894 datasets is shown in Figure 4, indicating that the risk-scoring system has good predictive performance.Figure 3:Risk score model assessment based on differential expression of 14 IMRGs in TCGA-BLCA patients. (A) Risk factor plot of the 14 IMRGs. (B) Time-ROC curves for the risk scoring model. (C) K–M curve of OS between the risk groups.Figure 4:Validation of prognostic value of the 14 IMRGs in GSE13507 and GSE32894. (A) Risk factor plot of the 14 IMRGs in GSE13507. (B) Time-ROC curves for the risk scoring model in GSE13507. (C) K–M curve of OS between the risk groups in GSE13507. (D) Risk factor plot of the 14 IMRGs in GSE32894. (E) Time-ROC curves for the risk scoring model in GSE32894. (F) K–M curve of OS between the risk groups in GSE32894.Analysis of clinicopathological characteristicsThe association analysis of the clinicopathological characteristics and risk scores of BLCA patients indicated that the clinicopathological characteristics significantly correlated with risk scores were as follows (Figure 5A–H): T stage (p<0.001), n stage (p<0.001), M stage (p<0.001), gender (p<0.05), pathologic stage (p<0.001), histologic grade (p<0.001), lymphovascular invasion (p<0.001), and subtype (p<0.001). Different degrees of bladder cancer malignancy often have different key genes and signalling pathways. We further conducted gene expression differential and pathway enrichment analyses based on BLCA muscle invasion and pathological grade. The correlation analysis results are shown in Figure S4.Figure 5:Correlation between clinicopathological characteristics and risk score in the TCGA-BLCA cohort. (A) T stage (T1 & T2 vs. T3 & T4), (B) n stage (n0 vs. n1 & n2 & n3), (C) M stage (M0 vs. M1), (D) pathologic stage (stage 1 & stage 2 vs. stage 3 & stage 4), (E) gender (female vs. male), (F) histologic grade (high grade vs. low grade), (G) lymphovascular invasion (no vs. yes), and (h) subtype (non-papillary vs. papillary).Nomogram analysisFirst, we included the risk groups and clinicopathological characteristics of BLCA patients in Cox regression analyses (Table 2). The results showed that the T stage, pathological stage, lymphovascular invasion status, and risk group were significantly associated with OS of BLCA patients. The above independent predictors were used to construct the nomogram (Figure 6A; C-index: 0.702, 0.684–0.763). Then, we assessed the predictive effect of this nomogram using time-ROC curves (Figure 6B; 1-, 3-, and 5-year AUCs: 0.741, 0.740, and 0.782). Moreover, the agreement of the nomogram was evaluated using calibration curves, which showed good predictive performance (Figure 6C–E).Table 2:Cox regression analyses between the clinicopathological characteristics and OS in the TCGA-BLCA cohort.CharacteristicsTotal, nUnivariate analysisMultivariate analysisHazard ratio (95 % CI)p-ValueHazard ratio (95 % CI)p-ValueT stage379T1 & T2124ReferenceT3 & T42552.199 (1.515–3.193)<0.0012.687 (1.399–12.048)0.024n stage369n0239Referencen1 & n2 & n31302.289 (1.678–3.122)<0.0011.127 (0.509–2.495)0.767M stage213M0202ReferenceM1113.136 (1.503–6.544)0.0020.921 (0.267–3.176)0.896Pathologic stage411Stage I & stage II134ReferenceStage III & stage IV2772.310 (1.596–3.342)<0.0011.465 (1.795–3.370)0.031Gender413Female109ReferenceMale3040.849 (0.616–1.169)0.316Histologic grade410High grade389ReferenceLow grade210.337 (0.083–1.360)0.126Lymphovascular invasion282No130ReferenceYes1522.294 (1.580–3.328)<0.0011.384 (1.842–4.130)0.017Subtype408Non-papillary275ReferencePapillary1330.690 (0.488–0.976)0.0361.203 (0.605–2.392)0.599Risk group413Low206ReferenceHigh2073.207 (2.329–4.417)<0.0012.790 (1.410–5.523)0.003P-value<0.05 is considered as statistical significance.Figure 6:Prognostic nomogram of BLCA patients in in TCGA-BLCA cohort. (A) Nomogram model constructed by the independent risk factors for OS in BLCA patients screened by multiple Cox regression. (B) The time-ROC curves for the nomogram. (C–E) calibration curves for the nomogram (p values of the Hosmer-Lemeshow test for 1-year, 3-year and 5-year calibration plots were 0.634, 0.687 and 0.872).GSEAGSEA analysis was employed on the DEGs between different risk groups. It showed that these DEGs were mainly enriched in DEVELOPMENTAL BIOLOGY, CELL CYCLE, CELL CYCLE MITOTIC, SIGNALING BY RHO GTPASES, RHO GTPASE EFFECTORS, CELL CYCLE CHECKPOINTS, DNA REPAIR, M PHASE, ECM REGULATORS, and SYSTEMIC LUPUS ERYTHEMATOSUS (Figure 7). This suggested that genes related to iron metabolism may play a role in developmental processes, the cell cycle, mitosis, the RHO GTPase response, DNA repair, and extracellular matrix regulation in BLCA.Figure 7:GSEA of DEGs between risk group in TCGA-BLCA cohort. Top 10 terms of GSEA (REACTOME DEVELOPMENTAL BIOLOGY, REACTOME CELL CYCLE, CELL CYCLE MITOTIC, SIGNALING by RHO GTPASES, RHO GTPASE EFFECTORS, CELL CYCLE CHECKPOINTS, DNA REPAIR, M PHASE, NABA ECM REGULATORS, and KEGG SYSTEMIC LUPUS ERYTHEMATOSUS).Immune infiltration analysisIt indicated that the high-risk group exhibited high levels of natural killer (NK) cells, effector T memory (TEM) cells, central memory T (TCM) cells, and macrophage infiltration (Figure 8A). Furthermore, the risk score positively correlated with the immune infiltrating cells (Figure 8B–F).Figure 8:Analysis of immune infiltration. (A) The box plot shows the infiltrating levels of different immune cells between the risk groups of BLCA patients. Scatter plots between immune cell enrichment and the risk score (B, NK cells; C, TCM; D, TEM; E, Th2 cells; and F, macrophages).DiscussionBladder cancer is characterised by a high incidence, hidden early symptoms, and high heterogeneity. The prognosis of bladder cancer patients depends on the tumour stage [17], which is categorised by the presence or absence of muscle invasion. BLCA patients with muscle-invasive have a poor prognosis [18]. In recent years, IMRGs have gradually become the focus of tumour research due to their important roles in tumorigenesis and development.Here, we constructed a risk-scoring model comprising 14 IMRGs. It was verified to effectively predict the prognosis of BLCA patients. The survival analysis revealed that high expression levels of CYP2W1, PPEF1, HEPH, ATP6V1B1, PTGIS, CCNB1, MYC, and CYP1B1 were associated with poorer OS outcomes, whereas high expression levels of FA2H, CYP27B1, CYP2C9, CYP2C8, AIFM3, and ASIC3 were associated with better OS outcomes (Figure S1). Moreover, BLCA patients with high T, n, or M stage disease, a high pathologic stage and histological grade, lymphovascular invasion, the non-papillary subtype, or were female had higher risk scores. These factors indicate a poor prognosis.Numerous studies shows that IMRGs are associated with many malignant tumors. CYP2W1 encodes a member of the cytochrome P450 superfamily associated with prognosis in many tumours [19]. PPEF1 is important for encoding serine/threonine protein phosphatase family members. Ting Ye et al. showed that PPEF1 promotes breast cancer progression and is a potential noninvasive diagnostic marker [20]. HEPH-encoded protein helps transport dietary iron from the digestive system to the circulatory system. HEPH is an important research focus and prognostic evaluation marker in lung cancer [21]. As a multi-subunit enzyme involved in organelle acidification, vacuolar ATPase (V-ATPase) cannot be synthesised without ATP6V1B1 [22]. Mariko Nishie et al. demonstrated that downregulation of ATP6V1B1 expression can acidify the internal environment of breast cancer cells. And this can lead to resistance to antibody therapy [22]. The protein encoded by CCNB1 is involved in mitosis. Shuxiong Zeng et al. confirmed by transcriptome sequencing that CCNB1 is a prognostic biomarker for BLCA [23]. MYC encodes a multifunctional nucleophosmin that functions in various important biological processes. Different expression states of MYC are associated with various tumours. Guosen Zhang et al. confirmed the value of MYC as a prognostic marker in BLCA [24].FA2H encodes an important protein that catalyses the synthesis of 2-hydroxysphingolipids. Ting Qi et al. found that the chemosensitivity of ovarian cancer to cisplatin is associated with the activation of FA2H, and FA2H can effectively predict the prognosis of ovarian cancer [25]. CYP27B1, CYP2C9, and CYP2C8 encode members of the cytochrome P450 superfamily. Ying Xiong et al. showed that CYP27B1 has good prognostic value in head and neck squamous cell carcinoma [26]. Dong Gui Hu et al. found that CYP2C9 and CYP2C8 are good prognostic markers for liver cancer [27]. ASIC3 is mainly located in peripheral nociceptors, which can rapidly activate and mediate homeostatic currents involved in pain perception. The study by Zhu Shuai et al. manifested that ASIC3 can activate the Ca2+/RhoA pathway and mediate EMT in pancreatic cancer [28].In addition, to better assess OS outcomes in BLCA patients, we included the risk group as an independent parameter together with the T stage, pathologic stage, and lymphovascular invasion status in the nomogram. The nomogram was verified to effectively evaluate the prognosis of BLCA patients and guide the formulation of clinical follow-up plans.GSEA revealed that the DEGs between different risk groups were mainly related to biological processes such as developmental process, DNA repair, mitosis, cell cycle, RHO GTPase reaction, and extracellular matrix regulation of BLCA. In order to meet the needs of rapid growth and proliferation, the intake of iron by tumor cells will be greatly increased. By using an iron-chelating polymer, DFO, Kana Komoto et al. successfully arrested the cell cycle, thus inducing apoptosis and anti-proliferative activity [29]. Iron metabolism is involved in numerous biological processes. Sujani Madhurika Kodagoda Gamage et al. revealed that heme can induce G1 and G2 cell cycle by activating the p53 and β-catenin pathways. This results in genetic and cell cycle changes in colon cancer cells [30]. Frank vom Dorp et al. showed that inhibition of Rho kinase inhibited BLCA tumour cell migration [31]. Alterations in DNA repair capabilities may alter an individual’s susceptibility to cancer. Tayyaba Ahmed et al. elucidated the role of DNA repair genes in BLCA through gene polymorphism studies [32].By analysing immune cell infiltration, we found that risk score was positively correlated with infiltrating immune cells in TCGA-BLCA cohort. Yuhan Sun et al. found that immune infiltration may improve the prognosis of BLCA patients as the NK cell phenotype expanded by IL-2 was most abundant in low- and high-grade BLCA tumours [33]. Benling Xu et al. found that IL-2/S-15/Akti could expand tumour-infiltrating lymphocytes and displayed the highest percentage of TCM cell phenotypes during activation by anti-CD3 antibodies [34]. Yang Shen et al. found that most tumour-infiltrating CD8+CD103+ tissue-resident memory T cells in gastric cancer tissue were TEM cells, unlike in non-tumour tissue, and this TEM cell infiltration was associated with gastric cancer [35]. Li Dehui et al. found that Xihuang Pill could regulate Th1 and Th2 cell levels in rats with precancerous lesions, thereby regulating the Th1/Th2 ratio [36]. Another study demonstrated that the expanded M1-like macrophage signature was enriched in BLCA samples and associated with reduced BLCA patients’ OS times [37]. However, the prognosis of bladder cancer and the response to clinical treatment are also closely influenced by immune activation status. The new classifier developed by Meng J et al. effectively characterises the immune microenvironment in BLCA patients and provides new ideas for immunotherapy [13]. This study also provides a new way to further study the risk-scoring system and the activation state of the immune microenvironment in BLCA patients in the future.ConclusionsIn summary, our prognostic model based on IMRGs can effectively predict BLCA patient prognosis. The nomogram constructed with the risk group as an independent parameter can also effectively predict BLCA patient OS outcomes. The 14 genes identified are potential targets for BLCA therapy and research. Furthermore, GSEA revealed that the DEGs between different risk groups were mainly involved in biological processes such as developmental process, DNA repair, mitosis, cell cycle, RHO GTPase reaction, and extracellular matrix regulation of BLCA. Our comprehensive bioinformatics analysis revealed that IMRGs may become an important focus in basic and clinical BLCA research. Further in-depth studies are required to assess the specific roles of IMRGs in BLCA. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png ONCOLOGIE de Gruyter

Prognostic value of iron metabolism-related genes in bladder urothelial carcinoma

ONCOLOGIE , Volume 25 (4): 13 – Jul 1, 2023

Loading next page...
 
/lp/de-gruyter/prognostic-value-of-iron-metabolism-related-genes-in-bladder-1a65UWpqH7

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
de Gruyter
Copyright
© 2023 the author(s), published by De Gruyter, Berlin/Boston
eISSN
1765-2839
DOI
10.1515/oncologie-2023-0145
Publisher site
See Article on Publisher Site

Abstract

IntroductionBladder cancer is a common malignant tumour of the urinary system with an incidence ranking 9th globally and 13th in China [1, 2]. Bladder cancer is often caused by occupational exposure to heavy metals, such as arsenic and cadmium, and is influenced by factors such as smoking and race [3], [4], [5]. The clinical manifestations are mainly painless gross haematuria. When the lesions involve the bladder triangle, bladder irritation signs appear, including frequent urination, urgent urination, and pain. The primary screening methods for bladder cancer are urine exfoliation cytology and cystoscopy [6]. Surgical intervention, systemic chemotherapy, and radical resection are preferred standard treatments for most early bladder cancer patients [7]. For patients with advanced bladder cancer or those who are intolerant to surgical treatment, systemic chemotherapy is preferred.An in-depth understanding of the underlying molecular mechanisms of urological cancers can provide entry points for improving treatments and developing new therapeutic approaches. Recently, numerous studies have shown that ferroptosis due to abnormal iron metabolism plays an important role in the occurrence and progression of cancer [8].In our study, IMRGs were obtained from the Molecular Signatures Database. Gene expression profiles and clinical information from 414 BLCA and 19 normal samples were obtained from The Cancer Genome Atlas (TCGA) database. The expression of IMRGs in BLCA was verified in the Gene Expression Omnibus (GEO) database. A BLCA risk-scoring model was established from the identified IMRGs; this model was tested on TCGA-BLCA cohort. Furthermore, the potential functions and mechanisms of IMRGs in BLCA were explored by functional enrichment analysis and GSEA. Our results indicated that the risk-scoring model comprising 14 iron metabolism-related gene markers could effectively predict the overall OS outcomes of BLCA patients.Materials and methodsAcquisition of IMRGsIMRGs were searched in MSigDB (Table S1), and 535 were obtained (Table S2).Acquisition and analysis of gene expression files of BLCA patientsGene expression profiles and clinical data from 414 BLCA and 19 normal samples were obtained from TCGA database. The expression of IMRGs in BLCA was tested in dataset GSE13507. We excluded BLCA patients without available survival data as this study required a prognostic analysis.Screening and identification of IMRGsFirst, “DESeq2” was used to analyse the DEGs between BLCA and normal bladder tissues in TCGA database [9]. The DEGs screening parameters were as follows: adjust p<0.05 and |log 2-fold change| >1. In total, 535 IMRGs were analysed, and the DEGs were displayed in a Venn diagram. A total of 118 IMRGs were selected for subsequent study. GSE13507 was used to test IMRGs expression. We constructed a heatmap of the 118 IMRGs using the “ComplexHeatmap” package. “clusterProfiler”, “org.Hs.eg.db”, and “GOplot” packages were adopted for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses and the visualisation of the 118 IMRGs [10, 11].Creation and testing of risk-scoring systemFirst, we combined the BLCA patients gene expression and OS to conduct Cox analysis. Genes with p<0.05 were included in subsequent analyses. LASSO algorithm was executed on the 25 IMRGs obtained in the above steps using the “glmnet” and “survival” packages to screen variables and build a prognostic model. Fourteen genes significantly related to OS were obtained and used to construct the prognostic model. The risk-scoring model was generated using the following formula: Risk score=∑ni=1 exprgenei coefficientgenei [12]. Gene expression normalisation was adapted using the “edgeR” package. We adopted the “ggplot2” package to generate the risk factor plot. ROC curves were generated using “timeROC” package. High- and low-risk groups were divided by the median risk score. The “survminer” package was used to generate the survival curves. Furthermore, as described by Meng J and Lu X, we verified the prognostic value of the above 14 iron metabolisation-related genes using the external GSE13507 and GSE32894 datasets [13, 14]. We elucidated the correlation between clinicopathological characteristics and risk scores through dot plots generated using “ggplot2”.Construction and assessment of the nomogramWe included the risk group as a parameter and each clinicopathological characteristic in Cox regression analyses to determine the predictive ability of the risk-scoring system for BLCA patients. These clinicopathological characteristics included the following: T stage, N stage, M stage, gender, pathological stage, histological grade, lymphovascular invasion, and subtype. The independent parameters affecting the prognosis of BLCA patients identified above were used to establish a nomogram using the “rms” package. The predictive value of the nomogram was tested by calibration curve and time-ROC analyses.GSEADEGs between risk groups were determined by the criteria and methods described above in TCGA-BLCA cohort. We employed the “clusterProfiler” package for GSEA and the “ggplot2” package to visualise the results.Immune infiltration analysisImmune infiltration analysis was conducted and visualised using the “GSVA” package [15, 16].Statistical analysisR software (version 3.6.3) was employed for data analysis and visualisation in our research. K–M survival analysis was performed using Cox regression analysis. Correlation analysis was performed using Spearman’s rank sum correlation test. The comparison between groups of qualitative data was performed using Chi-square analysis, and the comparison between groups of quantitative data was performed using an independent-samples T test. A p value or adjusted p value<0.05 was considered to represent significance.ResultsScreening of IMRGs in BLCAWe screened 8,702 DEGs in 414 TCGA-BLCA and 19 normal bladder tissues using the DESeq2 algorithm. The 535 IMRGs obtained from MSigDB and the 8,702 DEGs were subjected to Venn analysis. Ultimately, we obtained 118 IMRGs associated with BLCA (Figure 1A). Figure 1B shows the expression of the above 118 IMRGs in the GSE13507 dataset. A comparison of the heatmaps of these 118 genes (Figure S1) revealed consistent change trends across most genes. We conducted an enrichment analysis of the 118 IMRGs, revealing enrichment in DNA polymerase complex, iron ion transport, vacuolar proton-transporting V-type ATPase complex, iron ion homeostasis, transition metal ion transport, heme binding, proton-transporting V-type ATPase complex, iron ion binding, and tetrapyrrole binding (Figure 1D). Arachidonic acid metabolism, chemical carcinogenesis, retinol metabolism, and linoleic acid metabolism were all significantly enriched in the KEGG analysis (Figure 1C). The enrichment analysis results are combined in Table S3.Figure 1:Identification and enrichment analysis of IMRGs in the TCGA-BLCA cohort. (A) Venn diagram of the intersection between IMRGs and DEGs identified by the DESeq2 algorithm. (B) Heatmap of 118 DEGs related to iron metabolism in the GSE13507 data set. Terms of GO enrichment analysis (C) and KEGG pathways (D) related to the 118 IMRGs.Establishment and evaluation of risk-scoring systemFirst, we identified the associations of the 118 IMRGs with the outcomes of BLCA patients using Cox regression analysis. Using Cox p<0.05 as the screening criterion, we screened 25 genes that could potentially predict OS outcomes in BLCA patients (Table S4). Then, variables were further filtered using LASSO regression analysis (Figure 2A, B). We constructed the risk-scoring system using the 14 genes that remained (Table 1). K–M curves of the above 14 genes are shown in Figure S2. We also analysed the tissue expression of the above 14 genes using the Human Protein Atlas (HPA) database immunohistochemical results (Figure S3A–I).Figure 2:Visualization of DEGs with univariate Cox regression p value <0.05. (A) The LASSO regression model of the 25 IMRGs performed by LASSO tenfold cross-validation (lambda.min: lambda 0.023673, Index 18, statistic 11.771, SE 0.17651). (B) The coefficient distribution in the LASSO regression model.Table 1:Fourteen IMRGs identified by univariate Cox regression analysis.GeneDescriptionHazard ratio (95 % CI)p-ValueCYP2W1Cytochrome P450, family 2, subfamily W, polypeptide 11.599 (1.189–2.149)0.002PPEF1Protein phosphatase, EF-hand calcium binding domain 11.487 (1.104–2.002)0.009HEPHHephaestin1.396 (1.036–1.879)0.028FA2HFatty acid 2-hydroxylase0.636 (0.473–0.854)0.003CYP27B1Cytochrome P450, family 27, subfamily B, polypeptide 10.709 (0.529–0.950)0.021ATP6V1B1ATPase, H+ transporting, lysosomal 56/58 kDa, V1 subunit B11.459 (1.086–1.959)0.012PTGISProstaglandin I2 (prostacyclin) synthase1.693 (1.254–2.286)<0.001CCNB1Cyclin B11.411 (1.053–1.892)0.021MYCv-myc avian myelocytomatosis viral oncogene homologue1.610 (1.197–2.165)0.002CYP1B1Cytochrome P450, family 1, subfamily B, polypeptide 11.549 (1.153–2.082)0.004CYP2C9Cytochrome P450, family 2, subfamily C, polypeptide 90.725 (0.541–0.972)0.032CYP2C8Cytochrome P450, family 2, subfamily C, polypeptide 80.683 (0.509–0.917)0.011AIFM3Apoptosis-inducing factor, mitochondrion-associated, 30.579 (0.430–0.780)<0.001ASIC3Acid sensing (proton gated) ion channel 30.731 (0.545–0.981)0.037We calculated a risk score for each patient in TCGA-BLCA cohort based on the abovementioned risk-scoring system. The risk factor plots for TCGA-BLCA cohort, including survival time distribution, risk score distribution, and the expression levels of the 14 genes, are shown in Figure 3A. The time-ROC curves of this risk-scoring system for predicting BLCA patient OS outcomes are shown in Figure 3B (1-, 3-, and 5-year AUCs: 0.705, 0.725, and 0.714). The K–M curve indicates that the prognosis of the high-risk group was significantly poor (p<0.001; Figure 3C). The verification of the prognostic value of the 14 genes in the GSE13507 and GSE32894 datasets is shown in Figure 4, indicating that the risk-scoring system has good predictive performance.Figure 3:Risk score model assessment based on differential expression of 14 IMRGs in TCGA-BLCA patients. (A) Risk factor plot of the 14 IMRGs. (B) Time-ROC curves for the risk scoring model. (C) K–M curve of OS between the risk groups.Figure 4:Validation of prognostic value of the 14 IMRGs in GSE13507 and GSE32894. (A) Risk factor plot of the 14 IMRGs in GSE13507. (B) Time-ROC curves for the risk scoring model in GSE13507. (C) K–M curve of OS between the risk groups in GSE13507. (D) Risk factor plot of the 14 IMRGs in GSE32894. (E) Time-ROC curves for the risk scoring model in GSE32894. (F) K–M curve of OS between the risk groups in GSE32894.Analysis of clinicopathological characteristicsThe association analysis of the clinicopathological characteristics and risk scores of BLCA patients indicated that the clinicopathological characteristics significantly correlated with risk scores were as follows (Figure 5A–H): T stage (p<0.001), n stage (p<0.001), M stage (p<0.001), gender (p<0.05), pathologic stage (p<0.001), histologic grade (p<0.001), lymphovascular invasion (p<0.001), and subtype (p<0.001). Different degrees of bladder cancer malignancy often have different key genes and signalling pathways. We further conducted gene expression differential and pathway enrichment analyses based on BLCA muscle invasion and pathological grade. The correlation analysis results are shown in Figure S4.Figure 5:Correlation between clinicopathological characteristics and risk score in the TCGA-BLCA cohort. (A) T stage (T1 & T2 vs. T3 & T4), (B) n stage (n0 vs. n1 & n2 & n3), (C) M stage (M0 vs. M1), (D) pathologic stage (stage 1 & stage 2 vs. stage 3 & stage 4), (E) gender (female vs. male), (F) histologic grade (high grade vs. low grade), (G) lymphovascular invasion (no vs. yes), and (h) subtype (non-papillary vs. papillary).Nomogram analysisFirst, we included the risk groups and clinicopathological characteristics of BLCA patients in Cox regression analyses (Table 2). The results showed that the T stage, pathological stage, lymphovascular invasion status, and risk group were significantly associated with OS of BLCA patients. The above independent predictors were used to construct the nomogram (Figure 6A; C-index: 0.702, 0.684–0.763). Then, we assessed the predictive effect of this nomogram using time-ROC curves (Figure 6B; 1-, 3-, and 5-year AUCs: 0.741, 0.740, and 0.782). Moreover, the agreement of the nomogram was evaluated using calibration curves, which showed good predictive performance (Figure 6C–E).Table 2:Cox regression analyses between the clinicopathological characteristics and OS in the TCGA-BLCA cohort.CharacteristicsTotal, nUnivariate analysisMultivariate analysisHazard ratio (95 % CI)p-ValueHazard ratio (95 % CI)p-ValueT stage379T1 & T2124ReferenceT3 & T42552.199 (1.515–3.193)<0.0012.687 (1.399–12.048)0.024n stage369n0239Referencen1 & n2 & n31302.289 (1.678–3.122)<0.0011.127 (0.509–2.495)0.767M stage213M0202ReferenceM1113.136 (1.503–6.544)0.0020.921 (0.267–3.176)0.896Pathologic stage411Stage I & stage II134ReferenceStage III & stage IV2772.310 (1.596–3.342)<0.0011.465 (1.795–3.370)0.031Gender413Female109ReferenceMale3040.849 (0.616–1.169)0.316Histologic grade410High grade389ReferenceLow grade210.337 (0.083–1.360)0.126Lymphovascular invasion282No130ReferenceYes1522.294 (1.580–3.328)<0.0011.384 (1.842–4.130)0.017Subtype408Non-papillary275ReferencePapillary1330.690 (0.488–0.976)0.0361.203 (0.605–2.392)0.599Risk group413Low206ReferenceHigh2073.207 (2.329–4.417)<0.0012.790 (1.410–5.523)0.003P-value<0.05 is considered as statistical significance.Figure 6:Prognostic nomogram of BLCA patients in in TCGA-BLCA cohort. (A) Nomogram model constructed by the independent risk factors for OS in BLCA patients screened by multiple Cox regression. (B) The time-ROC curves for the nomogram. (C–E) calibration curves for the nomogram (p values of the Hosmer-Lemeshow test for 1-year, 3-year and 5-year calibration plots were 0.634, 0.687 and 0.872).GSEAGSEA analysis was employed on the DEGs between different risk groups. It showed that these DEGs were mainly enriched in DEVELOPMENTAL BIOLOGY, CELL CYCLE, CELL CYCLE MITOTIC, SIGNALING BY RHO GTPASES, RHO GTPASE EFFECTORS, CELL CYCLE CHECKPOINTS, DNA REPAIR, M PHASE, ECM REGULATORS, and SYSTEMIC LUPUS ERYTHEMATOSUS (Figure 7). This suggested that genes related to iron metabolism may play a role in developmental processes, the cell cycle, mitosis, the RHO GTPase response, DNA repair, and extracellular matrix regulation in BLCA.Figure 7:GSEA of DEGs between risk group in TCGA-BLCA cohort. Top 10 terms of GSEA (REACTOME DEVELOPMENTAL BIOLOGY, REACTOME CELL CYCLE, CELL CYCLE MITOTIC, SIGNALING by RHO GTPASES, RHO GTPASE EFFECTORS, CELL CYCLE CHECKPOINTS, DNA REPAIR, M PHASE, NABA ECM REGULATORS, and KEGG SYSTEMIC LUPUS ERYTHEMATOSUS).Immune infiltration analysisIt indicated that the high-risk group exhibited high levels of natural killer (NK) cells, effector T memory (TEM) cells, central memory T (TCM) cells, and macrophage infiltration (Figure 8A). Furthermore, the risk score positively correlated with the immune infiltrating cells (Figure 8B–F).Figure 8:Analysis of immune infiltration. (A) The box plot shows the infiltrating levels of different immune cells between the risk groups of BLCA patients. Scatter plots between immune cell enrichment and the risk score (B, NK cells; C, TCM; D, TEM; E, Th2 cells; and F, macrophages).DiscussionBladder cancer is characterised by a high incidence, hidden early symptoms, and high heterogeneity. The prognosis of bladder cancer patients depends on the tumour stage [17], which is categorised by the presence or absence of muscle invasion. BLCA patients with muscle-invasive have a poor prognosis [18]. In recent years, IMRGs have gradually become the focus of tumour research due to their important roles in tumorigenesis and development.Here, we constructed a risk-scoring model comprising 14 IMRGs. It was verified to effectively predict the prognosis of BLCA patients. The survival analysis revealed that high expression levels of CYP2W1, PPEF1, HEPH, ATP6V1B1, PTGIS, CCNB1, MYC, and CYP1B1 were associated with poorer OS outcomes, whereas high expression levels of FA2H, CYP27B1, CYP2C9, CYP2C8, AIFM3, and ASIC3 were associated with better OS outcomes (Figure S1). Moreover, BLCA patients with high T, n, or M stage disease, a high pathologic stage and histological grade, lymphovascular invasion, the non-papillary subtype, or were female had higher risk scores. These factors indicate a poor prognosis.Numerous studies shows that IMRGs are associated with many malignant tumors. CYP2W1 encodes a member of the cytochrome P450 superfamily associated with prognosis in many tumours [19]. PPEF1 is important for encoding serine/threonine protein phosphatase family members. Ting Ye et al. showed that PPEF1 promotes breast cancer progression and is a potential noninvasive diagnostic marker [20]. HEPH-encoded protein helps transport dietary iron from the digestive system to the circulatory system. HEPH is an important research focus and prognostic evaluation marker in lung cancer [21]. As a multi-subunit enzyme involved in organelle acidification, vacuolar ATPase (V-ATPase) cannot be synthesised without ATP6V1B1 [22]. Mariko Nishie et al. demonstrated that downregulation of ATP6V1B1 expression can acidify the internal environment of breast cancer cells. And this can lead to resistance to antibody therapy [22]. The protein encoded by CCNB1 is involved in mitosis. Shuxiong Zeng et al. confirmed by transcriptome sequencing that CCNB1 is a prognostic biomarker for BLCA [23]. MYC encodes a multifunctional nucleophosmin that functions in various important biological processes. Different expression states of MYC are associated with various tumours. Guosen Zhang et al. confirmed the value of MYC as a prognostic marker in BLCA [24].FA2H encodes an important protein that catalyses the synthesis of 2-hydroxysphingolipids. Ting Qi et al. found that the chemosensitivity of ovarian cancer to cisplatin is associated with the activation of FA2H, and FA2H can effectively predict the prognosis of ovarian cancer [25]. CYP27B1, CYP2C9, and CYP2C8 encode members of the cytochrome P450 superfamily. Ying Xiong et al. showed that CYP27B1 has good prognostic value in head and neck squamous cell carcinoma [26]. Dong Gui Hu et al. found that CYP2C9 and CYP2C8 are good prognostic markers for liver cancer [27]. ASIC3 is mainly located in peripheral nociceptors, which can rapidly activate and mediate homeostatic currents involved in pain perception. The study by Zhu Shuai et al. manifested that ASIC3 can activate the Ca2+/RhoA pathway and mediate EMT in pancreatic cancer [28].In addition, to better assess OS outcomes in BLCA patients, we included the risk group as an independent parameter together with the T stage, pathologic stage, and lymphovascular invasion status in the nomogram. The nomogram was verified to effectively evaluate the prognosis of BLCA patients and guide the formulation of clinical follow-up plans.GSEA revealed that the DEGs between different risk groups were mainly related to biological processes such as developmental process, DNA repair, mitosis, cell cycle, RHO GTPase reaction, and extracellular matrix regulation of BLCA. In order to meet the needs of rapid growth and proliferation, the intake of iron by tumor cells will be greatly increased. By using an iron-chelating polymer, DFO, Kana Komoto et al. successfully arrested the cell cycle, thus inducing apoptosis and anti-proliferative activity [29]. Iron metabolism is involved in numerous biological processes. Sujani Madhurika Kodagoda Gamage et al. revealed that heme can induce G1 and G2 cell cycle by activating the p53 and β-catenin pathways. This results in genetic and cell cycle changes in colon cancer cells [30]. Frank vom Dorp et al. showed that inhibition of Rho kinase inhibited BLCA tumour cell migration [31]. Alterations in DNA repair capabilities may alter an individual’s susceptibility to cancer. Tayyaba Ahmed et al. elucidated the role of DNA repair genes in BLCA through gene polymorphism studies [32].By analysing immune cell infiltration, we found that risk score was positively correlated with infiltrating immune cells in TCGA-BLCA cohort. Yuhan Sun et al. found that immune infiltration may improve the prognosis of BLCA patients as the NK cell phenotype expanded by IL-2 was most abundant in low- and high-grade BLCA tumours [33]. Benling Xu et al. found that IL-2/S-15/Akti could expand tumour-infiltrating lymphocytes and displayed the highest percentage of TCM cell phenotypes during activation by anti-CD3 antibodies [34]. Yang Shen et al. found that most tumour-infiltrating CD8+CD103+ tissue-resident memory T cells in gastric cancer tissue were TEM cells, unlike in non-tumour tissue, and this TEM cell infiltration was associated with gastric cancer [35]. Li Dehui et al. found that Xihuang Pill could regulate Th1 and Th2 cell levels in rats with precancerous lesions, thereby regulating the Th1/Th2 ratio [36]. Another study demonstrated that the expanded M1-like macrophage signature was enriched in BLCA samples and associated with reduced BLCA patients’ OS times [37]. However, the prognosis of bladder cancer and the response to clinical treatment are also closely influenced by immune activation status. The new classifier developed by Meng J et al. effectively characterises the immune microenvironment in BLCA patients and provides new ideas for immunotherapy [13]. This study also provides a new way to further study the risk-scoring system and the activation state of the immune microenvironment in BLCA patients in the future.ConclusionsIn summary, our prognostic model based on IMRGs can effectively predict BLCA patient prognosis. The nomogram constructed with the risk group as an independent parameter can also effectively predict BLCA patient OS outcomes. The 14 genes identified are potential targets for BLCA therapy and research. Furthermore, GSEA revealed that the DEGs between different risk groups were mainly involved in biological processes such as developmental process, DNA repair, mitosis, cell cycle, RHO GTPase reaction, and extracellular matrix regulation of BLCA. Our comprehensive bioinformatics analysis revealed that IMRGs may become an important focus in basic and clinical BLCA research. Further in-depth studies are required to assess the specific roles of IMRGs in BLCA.

Journal

ONCOLOGIEde Gruyter

Published: Jul 1, 2023

Keywords: bioinformatics; bladder urothelial carcinoma; iron metabolism; prognostic value

There are no references for this article.