Abstract
Objectives:
To investigate the prognostic value of ferroptosis-related and immune-related transcriptional features in bladder urothelial carcinoma (BUC) and to evaluate whether their integration provides incremental prognostic information beyond conventional clinical factors.
Methods:
Transcriptomic and clinical data of patients with BUC were obtained from The Cancer Genome Atlas (TCGA) as the training cohort. Ferroptosis and immune scores were constructed based on predefined gene sets using a z-score-based averaging method. Associations with overall survival were evaluated using Kaplan-Meier analysis and multivariable Cox regression with adjustment for age and pathological stage. Model performance was assessed using concordance index (C-index), likelihood ratio testing, and bootstrap internal validation. External validation was performed in independent cohorts (GSE13507 and GSE32894). Nonlinear relationships were examined using restricted cubic spline analysis, and sensitivity analyses were conducted by incorporating additional clinicopathologic variables.
Results:
Neither ferroptosis nor immune scores alone were significantly associated with overall survival in Kaplan-Meier analyses. Multivariable Cox regression showed a directionally protective effect of immune score and a non-significant risk trend for ferroptosis score. Integration of ferroptosis and immune features resulted in modest improvement in prognostic performance compared with clinical models alone. These findings were consistent across external validation cohorts. Restricted cubic spline analysis revealed a nonlinear association for immune score but not for ferroptosis score. Sensitivity analyses incorporating tumor grade yielded similar results, supporting robustness of the findings.
Conclusion:
Ferroptosis and immune transcriptional features provide limited but consistent incremental prognostic information in bladder urothelial carcinoma. These findings are exploratory and hypothesis-generating, suggesting that integration of ferroptosis and immune biology may offer complementary insights but is not yet sufficient for clinical application.
Keywords
Introduction
Bladder urothelial carcinoma is one of the most common malignancies of the urinary system and remains a major cause of cancer-related morbidity and mortality worldwide. 1 Despite advances in surgical techniques, chemotherapy, and immunotherapy, the clinical outcomes of patients with bladder cancer vary substantially, even among those with similar clinicopathological characteristics.2 -4 Conventional prognostic factors, such as age and pathological stage, are insufficient to fully capture the biological heterogeneity of this disease, highlighting the need for additional molecular features that may improve prognostic stratification. 5
Ferroptosis is a distinct form of regulated cell death characterized by iron-dependent lipid peroxidation and has been increasingly recognized as an important biological process in cancer development and treatment response.6 -8 Emerging evidence suggests that ferroptosis may influence tumor growth, therapeutic sensitivity, and disease progression in various malignancies, including bladder cancer.9 -11 However, the prognostic implications of ferroptosis-related features in bladder urothelial carcinoma remain incompletely understood, and existing studies have reported heterogeneous results depending on analytical strategies and patient cohorts.
In parallel, the tumor immune microenvironment plays a critical role in bladder cancer progression and therapeutic responsiveness, particularly in the context of immune checkpoint inhibition. 12 Immune-related signatures derived from transcriptomic data have been associated with patient outcomes across multiple cancer types.13,14 Nevertheless, immune features alone may not fully explain survival variability in bladder cancer, suggesting that interactions between immune processes and other tumor-intrinsic pathways warrant further investigation.
Given the close biological interplay between ferroptosis and immune regulation, integrating ferroptosis-related and immune features may provide complementary prognostic information beyond either dimension alone. 15 In this study, we performed a comprehensive analysis of ferroptosis and immune characteristics in bladder urothelial carcinoma using data from The Cancer Genome Atlas. By combining survival stratification, multivariable Cox regression analyses, and time-dependent prognostic evaluation, we aimed to assess whether showed a trend toward association with overall survival and improves prognostic discrimination beyond conventional clinical factors.
Materials and Methods
Data Acquisition and Preprocessing
Transcriptomic expression data and corresponding clinical information for patients with bladder urothelial carcinoma were obtained from The Cancer Genome Atlas (TCGA) database. 16 Patients with available overall survival data were included in the analysis. Overall survival time was defined as the interval between the date of initial diagnosis and the date of death or last follow-up. Patients without complete survival information were excluded.
Gene expression data were processed at the gene level. Raw gene expression values were log2-transformed when necessary and standardized prior to downstream analyses. Clinical variables extracted for downstream analyses included age at diagnosis and pathological stage. To reduce potential bias caused by sparse subgroups, pathological stage was further categorized into early stage (stage I-II) and late stage (stage III-IV) for multivariable survival analyses.
Construction of Ferroptosis and Immune Scores
Ferroptosis-related genes were curated from published studies and public resources focusing on key regulators of lipid peroxidation, iron metabolism, and oxidative stress. Immune-related genes were selected based on well-established markers of cytotoxic immune activity, including CD8A, CD8B, GZMB, PRF1, IFNG, and CXCL9/10/11. The ferroptosis gene set consisted of 259 genes, and the immune gene set consisted of 13 genes (Supplementary Table S1).
For score construction, gene expression values were first standardized across samples using z-score transformation. The ferroptosis and immune scores were then calculated as the mean z-score of genes within each predefined gene set for each sample. This approach provides a simple and reproducible measure of pathway-level activity without relying on enrichment-based algorithms such as ssGSEA or GSVA.
For stratified survival analyses, patients were dichotomized into high and low ferroptosis or immune groups according to the median value of each score. To further explore the combined effects of ferroptosis and immune features, patients were classified into 4 combined groups based on the joint distribution of ferroptosis and immune scores: ferroptosis high/immune high, ferroptosis high/immune low, ferroptosis low/immune high, and ferroptosis low/immune low. 17
Survival Analysis
Kaplan-Meier survival curves were generated to evaluate differences in overall survival among groups stratified by immune score, ferroptosis score, and combined ferroptosis-immune categories. Survival differences between groups were assessed using the log-rank test.
To quantify the association between ferroptosis and immune features and overall survival, Cox proportional hazards regression analyses were performed. Continuous ferroptosis and immune scores were standardized using z-score transformation and analyzed per standard deviation increase. Multivariable Cox models were constructed with adjustment for age and pathological stage. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were estimated. The proportional hazards assumption was assessed based on Schoenfeld residuals.
To further explore potential nonlinear associations between molecular scores and survival, restricted cubic spline (RCS) analyses were performed using the rms package, with 3 knots specified for continuous variables.
Integrated Prognostic Modeling and Time-Dependent ROC Analysis
To evaluate the prognostic performance of integrated ferroptosis-immune features, a joint Cox regression model incorporating ferroptosis score, immune score, age, and pathological stage was constructed. The linear predictor derived from this model was used as a composite risk score.
Model discrimination was assessed using Harrell’s concordance index (C-index). Incremental prognostic value was evaluated by comparing nested models using likelihood ratio tests (LRT) and by calculating changes in C-index (ΔC-index) between models.
Internal Validation
Internal validation was performed using bootstrap resampling (1000 iterations) to assess model optimism and stability. Optimism-corrected performance metrics, including Dxy and C-index, were estimated using the validate function in the rms package.
External Validation
External validation was conducted using independent Gene Expression Omnibus (GEO) cohorts (GSE13507 and GSE32894). Gene expression data from GEO were processed using the same pipeline as the TCGA cohort, including probe-to-gene mapping, gene-level summarization, and z-score standardization. Ferroptosis and immune scores were calculated using the same predefined gene sets.
Cox regression analyses and model comparisons were repeated in external cohorts to evaluate the reproducibility and generalizability of findings. Kaplan-Meier survival analyses were additionally performed in the validation cohorts as supportive evidence (Supplementary Figure S1).
Sensitivity Analysis
Sensitivity analyses were conducted to evaluate the robustness of the primary findings. Additional clinicopathologic variables available in the TCGA dataset, specifically tumor grade (neoplasm_histologic_grade), were incorporated into multivariable Cox regression models alongside age and pathological stage. The consistency of effect estimates was assessed to determine whether the associations of ferroptosis and immune scores with overall survival were materially altered after extended clinical adjustment.
Time-Dependent ROC Analysis
Time-dependent receiver operating characteristic (ROC) analysis was performed to assess prognostic discrimination at predefined time points (1, 3, and 5 years). The predictive performance of the joint model was compared with that of a clinical baseline model including age and pathological stage alone. Area under the ROC curve (AUC) values were calculated to quantify model discrimination over time.
Statistical Analysis
All statistical analyses were conducted using R software (version 4.5.2; R Foundation for Statistical Computing, Vienna, Austria). Survival analyses were performed using the survival package, and time-dependent ROC analyses were conducted using the timeROC package. Additional analyses were performed using the rms package (for spline modeling and bootstrap validation) and Hmisc package (for C-index estimation). Continuous variables were summarized as medians with interquartile ranges, and categorical variables were summarized as counts and percentages where appropriate. All statistical tests were 2-sided, and a P < .05 was considered statistically significant unless otherwise specified.
Strobe
The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement, and the completed checklist is provided as a supplementary file (Supplementary Table S2).
Results
Baseline Characteristics of the Study Cohort
A total of 430 patients with bladder urothelial carcinoma from the TCGA cohort were included in this study. The baseline clinical and molecular characteristics of the study population are summarized in Table 1. The median age at diagnosis was 67 years, and most patients presented with advanced pathological stage (stage III-IV). Based on transcriptomic profiling, ferroptosis and immune scores were calculated for each patient, and the cohort was evenly stratified into high and low groups according to the median value of each score.
Baseline Characteristics of Patients With Bladder Urothelial Carcinoma From the TCGA Cohort.
Survival Analysis Based on Immune and Ferroptosis Features
To evaluate the prognostic relevance of immune and ferroptosis features, Kaplan-Meier survival analyses were first performed using single-feature stratification. Patients classified into immune high and immune low groups did not show a statistically significant difference in overall survival (Figure 1A). Similarly, stratification based on ferroptosis score alone did not result in a significant survival difference, although a tendency toward poorer survival was observed in the ferroptosis high group compared with the low group (Figure 1B).
Study design and survival analysis based on immune and ferroptosis features: (A) schematic overview of the study design. Transcriptomic and clinical data from the TCGA-BLCA cohort were used to construct ferroptosis and immune scores based on predefined gene sets. Scores were calculated as the mean z-score of genes within each gene set. Associations with overall survival were evaluated using Cox proportional hazards models, with model performance assessed by Harrell’s C-index and LRT. Internal validation was performed using bootstrap resampling, and external validation was conducted in independent GEO cohorts (GSE13507 and GSE32894), (B) Kaplan-Meier survival curves stratified by immune score (high vs low, dichotomized by median). No statistically significant difference in overall survival was observed between groups (log-rank test), and (C) Kaplan-Meier survival curves stratified by ferroptosis score (high vs low). A trend toward poorer survival was observed in the ferroptosis high group, although the difference did not reach statistical significance (log-rank test).
These results suggest that immune or ferroptosis features considered individually may be insufficient to fully capture survival heterogeneity in bladder urothelial carcinoma.
Prognostic Impact of Combined Ferroptosis-Immune Stratification
Given the limited prognostic discrimination observed with single-feature stratification, patients were further classified into 4 groups based on the combined status of ferroptosis and immune scores. Kaplan-Meier analysis demonstrated clearer separation of survival curves among the 4 combined ferroptosis-immune groups (Figure 2A). Although the log-rank test did not reach statistical significance, patients with discordant ferroptosis and immune features tended to exhibit less favorable survival outcomes compared with those in the concordant low-risk groups.
Combined ferroptosis-immune stratification and multivariable Cox analysis: (A) Kaplan-Meier survival curves for overall survival stratified by combined ferroptosis-immune groups: ferroptosis high/immune high, ferroptosis high/immune low, ferroptosis low/immune high, and ferroptosis low/immune low. Although the log-rank test did not reach statistical significance, separation trends were observed across groups, suggesting potential interaction between ferroptosis and immune features and (B) Multivariable Cox proportional hazards regression analysis for overall survival. Ferroptosis score and immune score were analyzed as continuous variables per SD increase. HRs and 95% CIs are shown. Models were adjusted for age and pathological stage. Immune score showed a directionally protective association, whereas ferroptosis score demonstrated a non-significant risk trend.
Multivariable Cox Regression Analysis of Ferroptosis and Immune Scores
To further quantify the association between ferroptosis and immune features and overall survival, multivariable Cox proportional hazards regression analyses were performed using standardized continuous scores (per standard deviation increase), with adjustment for age and pathological stage (Figure 2B).
In the TCGA cohort, the immune score showed a directionally protective association with overall survival (HR = 0.91, P = .189), whereas the ferroptosis score demonstrated a non-significant trend toward increased risk (HR = 1.14, P = .145; Figure 2B). Although neither association reached statistical significance, the effect estimates suggested a relatively stronger contribution from immune-related features.
To further evaluate the incremental prognostic value of these molecular features, model comparison analyses were conducted. The addition of the ferroptosis score resulted in only minimal improvement in model discrimination (ΔC-index <0.01), and likelihood ratio tests comparing nested models were not statistically significant (all P > .05).
In contrast, the immune score showed a comparatively stronger association with overall survival, although its incremental contribution beyond clinical variables remained modest. Consistent patterns were also observed across different modeling strategies (Figure 3A).
Integrated prognostic evaluation of ferroptosis and immune features: (A) forest plot of multivariable Cox regression models incorporating ferroptosis score, immune score, their categorical groupings, and combined ferroptosis-immune classifications. HRs and 95% CIs are presented on a logarithmic scale, demonstrating consistent effect directions across modeling strategies, (B) time-dependent ROC analysis at 3 years comparing the clinical model (age and pathological stage) with the joint model integrating ferroptosis score, immune score, and clinical variables. The joint model showed only modest improvement in discrimination compared with the clinical model (AUC shown in the panel), and (C) Kaplan-Meier survival curves for overall survival stratified by combined ferroptosis-immune groups. Although not statistically significant, the observed separation trends are consistent with the multivariable Cox regression results, supporting a complementary role of ferroptosis and immune features.
Taken together, these findings suggest that the prognostic signal of the integrated model is largely driven by immune-related features, while the independent contribution of ferroptosis is limited.
Integrated Prognostic Evaluation and Predictive Performance
To comprehensively assess the prognostic value of ferroptosis-immune integration, multiple Cox models incorporating continuous scores, categorical groupings, and combined ferroptosis-immune classifications were constructed. Forest plot visualization demonstrated consistent effect directions across different modeling strategies, with immune-related variables showing protective associations and ferroptosis-related variables exhibiting risk-oriented trends (Figure 3A).
The predictive performance of the integrated model was further evaluated using time-dependent receiver operating characteristic analysis. At 3 years, the joint model incorporating ferroptosis score, immune score, and clinical variables (age and pathological stage) showed only modest improvement in prognostic discrimination compared with the clinical model alone (Figure 3B).
Finally, Kaplan-Meier survival analysis based on the combined ferroptosis-immune groups again demonstrated separation trends among the 4 categories, consistent with the multivariable Cox regression results (Figure 3C).
Internal Validation and External Validation Across Independent Cohorts
Internal validation using bootstrap resampling (1000 iterations) demonstrated stable model performance with minimal optimism, supporting the reliability of the Cox models.
External validation was performed in 2 independent GEO cohorts (GSE13507 and GSE32894). Across these datasets, clinical variables (age and stage) consistently showed strong prognostic value, whereas immune and ferroptosis scores did not demonstrate statistically significant independent associations with overall survival.
Model comparison analyses further confirmed that the addition of immune or ferroptosis scores resulted in only marginal improvements in C-index, and likelihood ratio tests remained non-significant across external cohorts (Table 2).
Incremental Prognostic Performance of Clinical, Immune-Augmented, and Integrated Models.
These results indicate that the prognostic contribution of ferroptosis-immune features is reproducible in direction but limited in magnitude across independent datasets.
Kaplan-Meier analyses in the GSE13507 cohort showed no significant survival difference between immune-defined groups (Supplementary Figure S1), consistent with the TCGA results.
Nonlinear Association Analysis Using Restricted Cubic Splines
Restricted cubic spline analyses were performed in the TCGA cohort to explore potential nonlinear associations between continuous molecular scores and survival outcomes (Figure 4).
Nonlinear association between molecular scores and overall survival: (A) RCS analysis of the association between immune score (z-score) and hazard ratio for overall survival. A modest nonlinear association was observed, indicating a non-monotonic association between immune activity and survival risk. The shaded area represents the 95% confidence interval. The model was adjusted for age and pathological stage and (B) RCS analysis for ferroptosis score. No significant linear or nonlinear association with overall survival was observed. The shaded area represents the 95% confidence interval.
A modest nonlinear association was observed for the immune score (overall P = .019; nonlinearity P = .011), demonstrating a modest inverted U-shaped association with hazard risk. In contrast, no significant association or nonlinear pattern was identified for the ferroptosis score (overall P = .715; nonlinearity P = .666).
These findings suggest that the prognostic impact of immune activity may be complex and non-monotonic, whereas ferroptosis-related transcriptional signals exhibit limited prognostic relevance.
Sensitivity Analyses Incorporating Additional Clinicopathologic Variables
Sensitivity analyses incorporating additional clinicopathologic variables yielded consistent findings.
After adjustment for tumor grade, the immune score retained a directionally protective association with overall survival (HR = 0.86, P = .058), while the ferroptosis score remained non-significant (P = .160). Likelihood ratio testing indicated no significant improvement in model fit with the inclusion of ferroptosis score (P = .157).
These results demonstrate that the observed associations are robust to additional clinical adjustment, with no evidence of meaningful incremental prognostic contribution from ferroptosis-related features (Figure 5).
Sensitivity analysis adjusting for tumor grade. Sensitivity analysis evaluating the robustness of the association between molecular scores and overall survival after adjustment for tumor grade. HRs and 95% CIs are shown for both the primary model (adjusted for age and stage) and the grade-adjusted model. Immune score retained a directionally protective association, whereas ferroptosis score remained non-significant, indicating stable effect estimates across models.
Discussion
In this study, we performed an integrated analysis of ferroptosis-related and immune features in bladder urothelial carcinoma using TCGA transcriptomic and clinical data. By combining survival stratification, multivariable Cox regression, and time-dependent prognostic evaluation, we observed that the joint consideration of ferroptosis and immune characteristics was associated with overall survival, but provided only modest incremental prognostic value beyond conventional clinical factors.
Our results showed that stratification based on ferroptosis score or immune score alone did not yield statistically significant differences in overall survival, whereas combined ferroptosis-immune grouping resulted in clearer separation trends of survival curves. This finding suggests that neither ferroptosis-related nor immune-related features alone are sufficient to fully capture the prognostic heterogeneity of bladder cancer. Instead, integrating these 2 biological dimensions may better reflect the complex tumor biology underlying disease progression and patient outcomes.
Multivariable Cox regression analyses further supported the complementary roles of ferroptosis and immune features. After adjustment for age and pathological stage, immune score showed a directionally protective association with overall survival, although statistical significance was not consistently observed across models and cohorts. In contrast, ferroptosis score exhibited a non-significant trend toward increased risk, indicating a more nuanced and context-dependent association. These observations align with emerging evidence that ferroptosis-related pathways may exert heterogeneous effects depending on tumor context and interactions with immune processes. 18
Importantly, model comparison analyses, including likelihood ratio testing and ΔC-index evaluation, consistently demonstrated only minimal improvement in prognostic performance when immune or ferroptosis scores were added to clinical variables. These findings indicate that, while biologically informative, the independent prognostic contribution of these transcriptomic features is limited in magnitude.
The prognostic value of integrating ferroptosis and immune features was further evaluated using time-dependent ROC analysis. The joint model incorporating ferroptosis score, immune score, and clinical variables demonstrated only modest improvement in prognostic discrimination compared with a clinical model alone at 3 years. Although the magnitude of improvement was limited, it is noteworthy given the inherent biological heterogeneity of bladder cancer and the limitations of transcriptome-based prognostic modeling.
External validation across independent GEO cohorts (GSE13507 and GSE32894) further supported the robustness of these findings. While the direction of associations was generally consistent, immune and ferroptosis scores did not demonstrate statistically significant independent prognostic value in external datasets, reinforcing the modest effect size observed in the TCGA cohort.
Restricted cubic spline analyses revealed a modest nonlinear association between immune score and overall survival, suggesting a non-monotonic relationship that may reflect complex biological states within the tumor microenvironment. In contrast, no nonlinear pattern was observed for the ferroptosis score. These findings highlight the potential limitations of assuming linear or dichotomized effects when modeling immune-related transcriptional features.
Sensitivity analyses incorporating additional clinicopathologic variables, including tumor grade and nodal status, yielded consistent effect directions and did not materially alter the primary findings. This supports the robustness of the observed associations and suggests that the results are not substantially confounded by additional clinical factors.
Several limitations of this study should be acknowledged. First, although external validation was performed, all analyses were retrospective and based on publicly available datasets, and prospective validation in well-characterized clinical cohorts is warranted to confirm the generalizability of our findings. Second, ferroptosis and immune scores were derived from predefined gene sets and transcriptomic data, which may not fully capture functional pathway activity at the protein or cellular level. Third, although combined stratification revealed clearer survival trends, some subgroup comparisons did not reach statistical significance, likely reflecting limited sample sizes within specific categories. Finally, the observational nature of this study precludes causal inference regarding the mechanistic interplay between ferroptosis and immune regulation.
Despite these limitations, our study provides a systematic evaluation of ferroptosis-immune integration in bladder urothelial carcinoma and highlights its potential relevance for prognostic assessment. Rather than establishing strong independent prognostic biomarkers, our findings suggest that ferroptosis and immune features may offer complementary but limited incremental information beyond established clinical predictors. Future studies incorporating external validation cohorts, multi-omics data, and functional experiments may help clarify the biological mechanisms underlying these associations and refine their clinical applicability.
Conclusion
In conclusion, this study provides a comprehensive evaluation of ferroptosis-related and immune-related transcriptional features in bladder urothelial carcinoma using integrated transcriptomic analyses. While both ferroptosis and immune scores demonstrated consistent directional associations with overall survival, their independent prognostic contributions were limited after adjustment for established clinical factors. Model comparison analyses, including likelihood ratio testing and concordance index evaluation, revealed only modest incremental predictive value when molecular features were added to clinical models. External validation across independent cohorts further supported the reproducibility of these findings while confirming their limited effect size.
Notably, nonlinear modeling suggested a complex association between immune activity and survival outcomes, highlighting the importance of considering non-monotonic effects in transcriptome-based prognostic analyses. Overall, our findings suggest that ferroptosis and immune features may provide complementary biological insights but are insufficient as standalone prognostic markers. Future studies integrating multi-omics data and functional validation are warranted to better define their clinical utility.
Supplemental Material
sj-pdf-2-cix-10.1177_11769351261453955 – Supplemental material for Integrated Ferroptosis and Immune Features Are Associated With Overall Survival in Bladder Urothelial Carcinoma: A TCGA-Based Analysis
Supplemental material, sj-pdf-2-cix-10.1177_11769351261453955 for Integrated Ferroptosis and Immune Features Are Associated With Overall Survival in Bladder Urothelial Carcinoma: A TCGA-Based Analysis by Lei Zhang and Zhengzuo Sheng in Cancer Informatics
Supplemental Material
sj-tiff-1-cix-10.1177_11769351261453955 – Supplemental material for Integrated Ferroptosis and Immune Features Are Associated With Overall Survival in Bladder Urothelial Carcinoma: A TCGA-Based Analysis
Supplemental material, sj-tiff-1-cix-10.1177_11769351261453955 for Integrated Ferroptosis and Immune Features Are Associated With Overall Survival in Bladder Urothelial Carcinoma: A TCGA-Based Analysis by Lei Zhang and Zhengzuo Sheng in Cancer Informatics
Footnotes
Acknowledgements
The authors would like to thank the TCGA and GEO databases for providing the data used in this study.
Abbreviations
BUC, bladder urothelial carcinoma; TCGA, The Cancer Genome Atlas; GEO, Gene Expression Omnibus; OS, overall survival; HR, hazard ratio; CI, confidence interval; KM, Kaplan-Meier; ROC, receiver operating characteristic; AUC, area under the curve; C-index, concordance index; LRT, likelihood ratio test; ΔC-index, change in concordance index; RCS, restricted cubic spline; IQR, interquartile range; STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.
Ethical Considerations
Ethical approval was obtained by the original data providers, and therefore no additional ethical approval was required for this study.
Author Contributions
ZS conceptualization, data curation, formal analysis, methodology, visualization, writing – original draft. LZ investigation, validation, writing – review and editing. All authors have read and approved the final manuscript.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
All datasets used in this study are publicly available. TCGA-BLCA data can be accessed from The Cancer Genome Atlas (https://portal.gdc.cancer.gov/) and GEO datasets (GSE13507 and GSE32894) are available from the Gene Expression Omnibus database (
).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
