Skip to main content
NCBI home page
BMC Cancer logoLink to BMC Cancer
. 2025 Nov 21;25:1796. doi: 10.1186/s12885-025-15010-1

Elucidating a novel prognostic signature for bladder cancer by integrating hypoxia and lactate metabolism-related genes: comprehensive bioinformatics analyses and experimental evidence

Yihao Zhao 1, Peixin Li 1, Zhe Shen 1, Shengwen Yao 1, Xiaoyi Zhang 2, Nianzhao Zhang 1,✉, Jun Chen 1,✉
PMCID: PMC12636232  PMID: 41272546

Abstract

Background

Bladder cancer (BLCA) is a highly heterogeneous malignancy with high morbidity and mortality. Massive lactate production and hypoxia are characteristics of the tumor microenvironment (TME). However, our understanding of the clinical value of hypoxia and lactate metabolism (HLM) in BLCA remains limited.

Methods

K-means clustering algorithm was used to classify molecular subtypes. The prognostic model was developed via univariate cox regression, random forest, and stepwise multivariate cox regression analyses. We subsequently systematically correlated the hypoxia and lactate metabolism-related risk score with the TME, BLCA consensus subtypes, and potential predictive value for drug therapy efficacy. Single-cell analyses demonstrated the expression of the modeling genes in various cell subtypes in the TME, and experimental validation was performed to examine the expression and function of GALK1 and TFRC.

Results

The TCGA cohort was classified into two subtypes. A 9-gene signature was established on the basis of genes associated with HLM, which predicted prognosis with exceptional efficacy. Patients with high risk scores had a poor prognosis, abundant infiltration of tumor-promoting immune cells and suppressed immune function. Furthermore, we anticipated that these patients were insensitive to immunotherapy and conventional chemotherapeutic agents. In addition, such patients were more inclined to the basal subtype. The modeling genes GALK1 and TFRC were highly expressed in BLCA and promoted tumor cell proliferation and migration.

Conclusions

Our signature further illustrated heterogeneity of BLCA. This signature could predict prognosis, consensus subtypes and treatment efficacy. We believe that this signature can optimize individual treatment decisions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-15010-1.

Keywords: Bladder cancer, Hypoxia, Lactate metabolism, Tumor microenvironment, Prognosis

Introduction

Bladder cancer (BLCA) is the most prevalent genitourinary neoplasm in China [1]. Based on the depth of tumor infiltration into the bladder wall, BLCA can be classified as non-muscle-invasive bladder cancer (NMIBC) and muscle-invasive bladder cancer (MIBC) [2]. Despite the advancements of treatment modalities such as surgery and chemotherapy, which have led to improvements in patient prognoses [3], a significant proportion of patients continue to experience tumor progression and metastasis following treatment [4]. Advanced BLCA is highly malignant and has a low patient survival rate [5]. Consequently, the development of a reliable biomarker for the precise identification of patients with poor prognosis, as well as for guiding individualized treatment, is of great clinical importance.

The tumor microenvironment (TME) profoundly shapes tumor biology and therapeutic response [6, 7]. Hypoxia, a characteristic feature of the TME in solid tumors, is pivotal in these processes. The rapid proliferation of tumor cells leads to considerable local tissue oxygen consumption [8, 9]. Concurrently, the chaotic abnormal neovascularization is incapable of adequately satisfying the oxygen requirements of the tumor tissue, thereby intensifying the hypoxic state. Hypoxia-inducible factors orchestrate the cellular response to hypoxia, governing key processes such as metabolic reprogramming, proliferation, metastasis, epithelial–mesenchymal transition, and angiogenesis [10–13]. In recent years, a hypoxic state has been shown to induce an immunosuppressive environment, impair immune cell function and differentiation, and promote immunosuppression and tumor immune escape. In a hypoxic environment, immunosensitive tumor cells can be remodeled into drug-resistant cells [14, 15].

Hypoxia drives a shift in tumor metabolism toward glycolysis, and the tumor itself is naturally prone to aerobic glycolysis (Warburg effect), resulting in substantial lactate generation in the TME [16]. Lactate, the byproduct of cellular metabolism, serves as a signaling molecule that modulates cellular phenotypes [17]. Studies have shown that the accumulation of lactate induces macrophage polarization to the M2 type. Lactate also reduces CD8 + T cell infiltration in the TME and upregulates programmed cell death-1 (PD-1) expression in Tregs, ultimately leading to immunotherapy failure [18, 19]. Moreover, lactate could induce histone lactylation modifications that modulate tumor sensitivity to chemotherapeutic drugs [20].

Nevertheless, the clinical significance of hypoxia and lactate metabolism (HLM) in BLCA remains incompletely understood. Owing to tumor-intrinsic heterogeneity and the intricate interplay within the HLM network, we considered both the effects of HLM on BLCA as well as identified and validated a gene signature. Patients stratified according to this signature presented differences in survival outcomes, TME characteristics, immune cell infiltration, immunotherapy response and drug sensitivity. The gene signature discovered in our work has potential for clinical application.

Methods

Data acquisition and processing

RNA-seq data for 412 BLCA samples and 19 normal samples were obtained from the TCGA database, resulting in a training cohort of 403 patients after 9 patients were excluded due to duplicates or absent follow-up data. The validation cohort was established via scRNA-seq data (GSE129845) and bulk RNA-seq data (GSE13507 and GSE32894) were obtained from the GEO database. Hypoxia and lactate metabolism-related genes (HLMRGs) were sourced from the Molecular Signatures Database (MSigDB, https://www.gsea-msigdb.org/gsea/msigdb).Using the keywords “hypoxia,” “lactic,” and “lactate” [21, 22], we identified 20 prioritized gene sets related to HLM in the MSigDB. Following the elimination of duplicate genes, a total of 885 HLMRGs were incorporated into the subsequent study. The details of the datasets were presented in Supplementary Table S1 and S2.

Molecular subtyping

We employed the consensus clustering tool from the R package “ConsuClusterPlus”. The primary parameters were established as follows: distance = “Euclidean”; clusterAlg = “km”; Reps = 1000; pItem = 0.8. On the basis of this approach, we identified different molecular subtypes associated with HLMRGs in BLCA.

Identification of DEGs and DEHLMRGs

The R package “DESeq2” was employed to perform differential expression analysis. Genes with an adjusted P value (P.adj) < 0.05 and |fold change| ≥ 1.5 were classified as differentially expressed genes (DEGs). Differentially expressed hypoxia and lactate metabolism-related genes (DEHLMRGs) were identified by intersecting DEGs with HLMRGs and subsequently visualized via the R package “VennDiagram”.

Construction of a hypoxia and lactate metabolism-related prognostic model

Univariate cox regression analysis was conducted via the R package “survival” to identify DEHLMRGs significantly correlated with prognosis (P < 0.01). To mitigate overfitting and achieve precise gene selection, we utilized random survival forest (RSF) via the R package “randomForestSRC” to further diminish gene dimensionality. The top 15 ranked variable importance (VIMP) and minimum depth genes were preserved for later model development. A stepwise multivariate cox regression analysis was subsequently conducted to develop the hypoxia and lactate metabolism-related risk score (HLMRS) = (0.130679 × ANXA1) + (0.136001 × ACKR3) + (0.1131765 × TFRC) + (−0.391767 × TCIRG1) + (0.223767 × ATAD3A) + (0.394762 × GALK1) + (0.147542 × DTNA) + (−0.261760 × SLC16A8) + (0.235425 × SLC13A5), incorporating a total of nine genes in this prognostic model. The patients were allocated a risk score according to this model and classified into a high risk score group (High RS) and a low risk score group (Low RS) on the basis of the median risk score of the sample population.

Survival analysis and measurement of predictive ability and stratified analysis

Prognostic differences between groups were analyzed via the R package “survival”. The statistical significance of these differences was assessed using the log-rank test, and Kaplan–Meier curves were plotted. The R package “timeROC” was used to construct ROC curves to assess predictive efficacy. Stratified analysis was implemented to evaluate the prognostic value of the HLMRS in diverse clinical subgroups.

Independent prognostic analysis and construction of the prognostic nomogram

We utilized univariate and multivariate Cox regression analyses for the HLMRS, gender, age, pathological stage, T stage and N stage to evaluate independent prognostic variables for BLCA. Then, we created a nomogram utilizing independent prognostic markers. We employed ROC curves, calibration curves, and decision curve analysis (DCA) to evaluate the efficacy of the nomogram.

Functional and pathway enrichment analyses

Gene set enrichment analysis (GSEA) was employed to examine the signaling pathways that are differentially activated between two molecular subtypes. The geneset “h.all.v2022.1.Hs.Symbols.gmt” was acquired from MSigDB. Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) were utilized to analyze variations in pathway activity between High RS and Low RS. The functional enrichment analysis was performed via the R package “clusterProfiler”.

Consensus clustering of the MIBC subtypes

The R package “ConsensusMIBC” was utilized to categorize the consensus subtypes of MIBC within the TCGA cohort. The consensus subtypes were further compared with the High RS and Low RS. According to the findings presented in the consensus subtypes, the subtypes were reclassified as “basal” or “luminal“ [23].

Drug sensitivity analysis

The treatment response of each patient in the TCGA cohort to commonly used chemotherapeutic agents for BLCA was predicted based on GDSC2 using the R package “oncoPredict“ [24].

Immune landscape analysis

Seven independent methods—TIMER, CIBERSORT, CIBERSORT-ABS, quanTIseq, MCPcounter, xCELL, and EPIC—were employed to assess immune cell infiltration levels due to inconsistencies between algorithms and marker gene sets. An investigation of immune function was performed to evaluate the relationship between the HLMRS and immunological function. Data from Tumor Immune Dysfunction and Exclusion (TIDE, http://tide.dfci.harvard.edu/) and The Cancer Immunome Atlas (TCIA, https://tcia.at/home/) were obtained to assess the effectiveness of immunotherapy in patients.

Single-cell analysis

The scRNA-seq matrix was imported via the R package “Seurat.” To improve data quality, we set a series of thresholds for excluding low quality cells. The data were subsequently normalized and standardized via “SCT”. The appropriate resolution was determined via R package “clustree”. Cell clustering analysis was performed using the “FindNeighbors” and “FindClusters” functions. Uniform manifold approximation and projection analysis (UMAP) was conducted via the “RunUMAP”. Correction analysis was executed by applying the “Harmony” package. The marker genes used for single-cell annotation were derived from previously published studies [25].

Patient specimens

Normal and tumor tissues were collected from BLCA patients undergoing radical cystectomy at Qilu Hospital (Jinan, China). The study was approved by the Ethics Committee of Qilu Hospital.

Cell culture

The SV-HUC-1 (CVCL_3798), BIU87 (CVCL_6681), T24 (CVCL_0554), 5637 (CVCL_0126), and UMUC-3 (CVCL_1783) cell Lines were cultured in appropriate medium containing 10% fetal bovine serum (FBS, Gibco, USA) and 1% penicillin-streptomycin (Gibco, USA) at 37 °C under humid conditions with 5% CO₂.

Small interfering RNA interference assay

GALK1 small Interfering RNA (siRNA) and TFRC siRNA were manufactured by General Biol (China). Lipofectamine 3000 (Invitrogen, USA) was used as the transfection reagent. BIU87 cells were transfected with siRNAs according to the manufacturer’s instructions. The details of the siRNA were recorded in Supplementary Table S3.

Quantitative real-time PCR (qRT–PCR) analysis

Total RNA was isolated from cells utilizing the Rapid RNA Extraction Kit (AG21023, Accurate Bioengineering, China) in accordance with the manufacturer’s guidelines. The RNA was reverse transcribed into complementary DNA using a reverse transcription kit (AG11706, Accurate Biology, China). qRT–PCR was conducted utilizing SYBR-Green Mix (AG11701, Accurate Biology, China) in a real-time PCR apparatus (LightCycler 96, Roche, China). Cellular gene expression differences were calculated by the 2-∆∆Ct method, with β-actin used as an internal reference. Primer information was recorded in Supplementary Table S4.

Western blot

Cellular proteins were extracted utilizing RIPA lysis buffer (Beyotime, China). Total protein extracts were resolved using 10% SDS-PAGE gels and subsequently Transferred to PVDF membranes, which were blocked with 5% milk powder at room temperature. The membranes were then incubated overnight at 4 ℃ and then treated with the secondary antibody for 1 h at room temperature (anti-GALK1:AB_2108553, anti-CD71:AB_2240403, anti-β-actin: AB_2687938). The bands were visualized using chemiluminescence imaging (Tanon, China).

Immunohistochemistry (IHC) analysis

Tissue paraffin slices were deparaffinized, rehydrated, antigenically repaired and incubated overnight at 4 °C with anti-GALK1 and anti-CD71 antibodies (Proteintech, China). All the tissue slices were then incubated with secondary antibodies for 30 min at room temperature, followed by incubation with streptavidin-horseradish peroxidase for 30 min. Each tissue sample was ultimately stained with diaminobenzidine and hematoxylin.

Cell counting Kit-8 (CCK8) assay

The cells were inoculated into 96-well plates at a density of 3500 cells/well and incubated at 37 °C for 6 h. After the cells were attached to the wall, CCK-8 reagent (CX001S, Epizyme, China) was added to each well, and the incubation was continued for 2 h. The OD value of each well was measured at 450 nm (Tecan, Switzerland).

Cell colony formation assay

The cells were inoculated into 6-well plates at a density of 1000 cells/well and incubated in an incubator for 2 weeks. The cells were then fixed, stained with crystal violet and photographed.

5-Ethynyl−2’-deoxyuridine Assay

The cells were inoculated into 6-well plates at a density of 3 × 105 cells/well, cultured until they reached 60% density, treated with EDU working solution (CX003, Epizyme, China) for an additional 2 h, fixed, and subsequently processed according to the kit’s instruction manual, culminating in observation and imaging with a fluorescence microscope (Keyence, China).

Transwell migration assay

The cells were resuspended in serum-free medium and then injected into the upper chamber of the Transwell chamber for culture. The lower chamber was filled with the appropriate medium containing 10% FBS. The Transwell chamber was subsequently incubated at 37 °C with 5% CO2 for 24 h. After the incubation period, the cells were fixed, stained with crystal violet, and photographed via an electron microscope (Olympus, Germany).

Statistical analysis

Bioinformatics analyses were performed using R version 4.4.1. Statistical analyses were carried out via GraphPad Prism 9 and ImageJ 1.45. The statistical significance of the differences between the two groups was assessed via Student’s t-test or the Mann-Whitney U test. ANOVA or the Kruskal-Wallis rank-sum test was used to compare the three groups. Significance is indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001.

Results

Construction of molecular subtypes related to hypoxia and lactate metabolism in BLCA

Figure 1 outlined the study workflow. Unsupervised clustering of the TCGA cohort, based on 885 HLMRGs, established a novel molecular classification system for BLCA. The results revealed an optimal classification into two unique subtypes: cluster 1(C1) and cluster 2 (C2) (Fig. 2a and c). A comprehensive evaluation was performed to elucidate the discrepancies between these molecular subtypes. In terms of prognosis, patients in C2 presented a markedly superior prognosis compared with those in C1 (HR = 1.52, P = 0.009, Fig. 2d). GSEA showed that the hypoxia pathway was significantly upregulated in C1 (P < 0.001, Fig. 2e), whereas the oxidative phosphorylation pathway was downregulated (P < 0.001, Fig. 2e). Moreover, C1 demonstrated significant upregulation of KRAS signalling and angiogenic pathways. The downregulation of oxidative phosphorylation, a crucial metabolic step in glycolysis, may indicate a diversion of pyruvate, the final metabolite of intracellular glycolysis, toward the lactate metabolism pathway, thus facilitating metabolic reprogramming within glycolysis. Figure 2f demonstrated markedly higher infiltration of naive B cells, naive CD4⁺ T cells, and activated dendritic cells in C2, whereas C1 were enriched for M0 and M1 macrophages as well as neutrophils.

Fig. 1.

Fig. 1

Open in a new tab

Outline of the analyses performed in this study

Fig. 2.

Fig. 2

Open in a new tab

Novel hypoxia and lactate metabolism-related molecular subtypes of BLCA. a Unsupervised clustering analysis based on 885 HLMRGs. b, c The CDF plot and delta area plot to select the best number of classifications. d Survival analysis of C1 and C2. e The GSEA between C1 and C2. f Infiltration of 22 immune cells between C1 and C2. *P < 0.05, **P < 0.01, ***P < 0.001

Construction of the HLMRS and validation of its predictive ability

The results of this study have revealed distinct activation patterns of hypoxia and lactate metabolic pathways in BLCA, resulting in significantly different patient prognoses and TME characteristics. This observation has prompted the development of a risk score using HLMRGs. The risk score could predict the clinical prognosis for each patient, facilitating precision therapy in BLCA.

First, differential expression analysis revealed 7,216 genes that were significantly altered between BLCA and adjacent normal tissues (Fig. 3a). Intersecting these DEGs with the 885 HLMRGs yielded 362 DEHLMRGs (Fig. 3b). To identify DEHLMRGs that were strongly associated with prognosis (P < 0.01), we performed univariate cox regression analysis and obtained 57 candidate genes. The gene dimensionality was further reduced via RSF, and the top 15 VIMP and minimum depth genes were retained for final model construction (Fig. 3c and d). Finally, we performed stepwise multivariate cox regression analysis based on these 19 genes. This analysis produced a prognostic predictive signature consisting of 9 genes (Fig. 3e). Consistent with its predictive value, patients in the poor-prognosis cluster (C1) exhibited higher HLMRS (P < 0.001, Fig. 3f).

Fig. 3.

Fig. 3

Open in a new tab

Construction and evaluation of the HLMRS. a Volcano plot of the DEGs. b DEGs and HLMRGs took intersection to obtain 362 candidate genes. c, d The genes were ranked by the VIMP and minimal depth. e The forest graph showed the results of stepwise multivariable cox regression analysis. f Comparison of HLMRS between C1 and C2. g Survival analysis of the TCGA cohort. h ROC curves for the TCGA cohort. i Survival analysis of the GSE13507 cohort. j ROC curves for the GSE13507 cohort. k Survival analysis of the GSE32894 cohort. l ROC curves for the GSE32894 cohort. m Mean AUC for the five gene signatures. n GO and KEGG enrichment analysis between the two risk score groups. *P < 0.05, **P < 0.01, ***P < 0.001

To evaluate the model’s validity, the training cohort (TCGA-BLCA) and two validation cohorts from the GEO database (GSE13507, GSE32894) were classified into High RS and Low RS according to the HLMRS. Increased HLMRS correlated with lower overall survival (OS) in BLCA patients across the three datasets. Kaplan–Meier curves demonstrated markedly worse OS in High RS (TCGA-BLCA: HR = 2.79, P < 0.001; GSE13507: HR = 2.46, P < 0.001; GSE32894: HR = 6.39, P < 0.001, Fig. 3g, i and k). The area under the curve (AUC) for predicting 1-year, 3-year, and 5-year OS were all above 0.65, demonstrating the outstanding accuracy (Fig. 3h, j and l). We selected four previously published gene signatures related to hypoxia or energy metabolism in BLCA to enable a thorough comparative analysis with the HLMRS [26–29]. Risk scores were calculated for each patient in the TCGA, GSE13507, and GSE32894 cohorts, utilizing the relevant genes from the four gene signatures. We then employed these signatures to calculate the mean AUC for predicting 1-, 3-, and 5- year OS(Fig. 3m). Our signature achieved the highest AUC across all time points, underscoring its superior predictive performance.

To investigate the differences in pathway activation between the two groups, we conducted GO and KEGG enrichment analyses on these groups. The findings indicated that the divergent pathways between the two groups were predominantly enriched in pathways related to extracellular structural organization, collagen synthesis, collagen containing extracellular matrix, and glycosaminoglycan binding (Fig. 3n). These pathways are clearly associated primarily with the extracellular matrix, which is a pivotal component of the TME. Moreover, the differential expression trends of the modeling genes in the validation cohorts were largely consistent with those in the training cohort (Fig. S1a-S1c).

Stratified analysis of different clinical characteristics of BLCA patients

A stratified study was conducted to further validate the ability of the HLMRS to reliably and independently predict the prognosis of BLCA patients. The TCGA cohort was categorized into subgroups according to clinical characteristics, such as age, gender, pathological grade, TNM stage, pathological stage, and lesion count. Our findings demonstrated that High RS was associated with poorer survival outcomes across all subgroups (Fig. S2a-S2l).

Constructing a nomogram with the HLMRS

In the TCGA cohort, we subjected sex, age, T stage, N stage, pathological stage, and HLMRS to both univariable and multivariable cox regression analyses. The studies revealed that age, T stage, N stage, pathological stage, and HLMRS were independent predictive variables for BLCA (Fig. S3a, S3b). Thereafter, we developed a nomogram utilizing these independent predictive markers (Fig. S3c). The efficacy of the nomogram was later evaluated via the examination of 1-, 3-, and 5-year ROC curves and calibration plots (Fig. S3d, S3e). The findings from the DCA indicated that the integration of HLMRS into the nomogram yielded a much greater net clinical benefit than the clinical characteristics alone (Fig. S3f). Our findings suggest that the HLMRS-based nomogram is a good instrument for predicting both short- and long-term OS in BLCA patients, hence assisting physicians in making clinical care decisions.

Heterogeneity of the TIME between two risk score groups

To more fully reveal the heterogeneity of the tumor immune microenvironment (TIME) between High RS and Low RS, we used seven algorithms to assess the infiltration of immune cells. As shown in the heatmap (Fig. 4a), more M2 macrophages, neutrophils, and tumor-associated fibroblasts infiltrated the TME in High RS than in Low RS. Previous studies have shown that these immune cells were closely related to poor prognosis in BLCA [30]. In addition, immune cells, including CD4+ T cells, CD8+ T cells, NK cells and dendritic cells, which were negatively associated with poor prognosis [31], were more abundant in Low RS’s TIME. In short, BLCA patients with low risk scores presented more “hot” tumor characteristics. The above results suggested extensive heterogeneity in the TIME between these groups classified by the HLMRS, providing a biological rationale for their divergent clinical outcomes.

Fig. 4.

Fig. 4

Open in a new tab

Assessment of differences in immune landscape and immunotherapy response between two risk score groups. a Heatmap of immune cell infiltration analysis using multiple algorithms (including TIMER, CIBERSORT, CIBERSORT-ABS, quanTIseq, MCPcounter, xCELL, and EPIC). b Correlation analysis of the HLMRS with immune functions. c TIDE score between two risk score groups. d The proportion of patients who benefit from immunotherapy in the two risk score groups. e The difference in IPS between two risk score groups. f The HLMRS predicted whether patients can benefit from ICB. *P < 0.05, **P < 0.01, ***P < 0.001

HLMRS assisted precision medicine for BLCA patients

An analysis was performed to ascertain the association between HLMRS and immunological function (Fig. 4b). The results indicated a negative correlation between HLMRS and immunological activity. The functions, which include “APC_co_stimulation,” “HLA,” “inflammation promotion,” and “T_cell_co_stimulation,” proved that patients with elevated HLMRS were susceptible to an immunosuppressed state. Additionally, data from the TIDE and TCIA websites showed that patients with a high risk score exhibited elevated TIDE scores (Fig. 4c). In contrast, the Low RS had a higher immunophenoscore (IPS, ips_ctla4_neg_pd1_neg, P < 0.001; ips_ctla4_neg_pd1_neg, P < 0.001, Fig. 4e) and a greater proportion of patients who responded to immunotherapy (Fig. 4d), suggesting that the Low RS was more likely to benefit from using immune checkpoint inhibitors (ICIs). In summary, HLMRS is a biomarker for predicting the effectiveness of immune checkpoint blockade therapy (Fig. 4f).

MIBC is biologically heterogeneous and encompasses several molecular subtypes with distinct prognoses and therapeutic sensitivities. The most widely recognized subtype classification for MIBC is the Consensus Molecular Classification, which is employed in our study. The TCGA cohort was used to classify patients with MIBC into consensus subtypes, and survival analysis was performed to demonstrate survival differences among the six subtypes in the TCGA-MIBC cohort (Fig. S4a). Furthermore, we observed a greater proportion of LumP and LumU subtypes in Low RS, while Ba/sq, NE-like, and LumNS subtypes were more prevalent in High RS (Fig. S4b). Surprisingly, HLMRS exhibited superior accuracy in predicting MIBC subtypes, with an AUC close to 0.8 (Fig. S4c).

In clinical practice, BLCA patients often require bladder perfusion therapy or systemic chemotherapy, and their sensitivity to chemotherapeutic agents considerably influences their prognosis. Therefore, we predicted the sensitivity to commonly used chemotherapeutic agents in these subgroups based on the GDSC2 database. We observed that Low RS had increased sensitivity to epirubicin, vinblastine, oxaliplatin and cyclophosphamide, whereas High RS had increased sensitivity to docetaxel and paclitaxel. There was no statistically significant difference in drug sensitivity to cisplatin or gemcitabine between the two risk score groups, suggesting that the combination of cisplatin and gemcitabine (GC) may be effective in both groups (Fig. S4d).

HLMRGs hold sound potential in guiding precision medicine approaches for BLCA patients, facilitating tailored treatment plans that align with individual patient profiles.

Single-cell analysis of 9 HLMRGs in the TME

The expression levels of the nine HLMRGs were investigated at the single-cell level in a range of TME-associated cell subtypes. Quality control measures were implemented for single-cell transcriptomic data from three BLCA samples (Fig. 5a). The “Harmony” package was employed to mitigate the impact of batch effects (Fig. 5b). We selected 1.5 as the best resolution (Fig. 5c). Unbiased clustering analysis identified 22 major clusters (Fig. 5d). Cellular annotation was performed using markers specific to different cell types described in previous studies, identifying seven distinct cell subtypes denoted as B cells, T cells, endothelial cells, myeloid cells, epithelial cells, and fibroblasts (Fig. 5e and f). ANXA1 was identified as the most highly expressed gene, SLC16A8 and SLC13A5 had modest expression across all TME-associated cell subtypes. Notably, TFRC and ATAD3A were highly expressed in epithelial cells (malignant cells) (Fig. 5f and g), which was consistent with their role as risk factors in BLCA as predicted by bioinformatics.

Fig. 5.

Fig. 5

Open in a new tab

Single-cell analysis of 9 HLMRGs in the TME. a, b Quality control and removal of batch effects. c Choose the best resolution via clustree. d The UMAP plot colored by the 22 clusters. e Annotation of 22 clusters into 7 cell subtypes by gene markers. f Percentage of 7 cell subtypes. g The expression of 9 modeling genes across 7 cell subtypes

Experimental validation of the expression and biological functions of GALK1 and TFRC

We chose the two genes with the highest (GALK1) and lowest (TFRC) absolute coefficient values in the prognostic model for experimental verification. Paired analysis, RT–qPCR, Western blot, and IHC confirmed that GALK1 and TFRC exhibited increased RNA and protein expression levels in BLCA (Fig. 6a and g). To evaluate the probable function of these genes in BLCA, we selected the BIU87, which had high levels of both RNA and protein expression, for our cellular phenotyping experiments.

Fig. 6.

Fig. 6

Open in a new tab

The expression of GALK1 and TFRC in BLCA. a, c Paired analysis of GALK1 and TFRC mRNA levels in tumor and corresponding normal tissues in TCGA cohort. b, d The expression of GALK1 and TFRC at mRNA levels in normal urothelial cell and BLCA cell lines through qRT-PCR. e, f The expression of GALK1 and TFRC at protein levels in normal urothelial cell and BLCA cell lines through western blot. g GALK1 and TFRC expression in normal and BLCA tissues by IHC. *P < 0.05, **P < 0.01, ***P < 0.001

Transfection of siRNA into the BIU87 led to reduced expression levels of GALK1 and TFRC. We chose the siRNA sequence with the most pronounced knockdown effect for further examination of the cellular phenotype (Fig. 7a and d). Knockdown of GALK1 and TFRC inhibited BIU87 colony formation, proliferation and migration in vitro (Fig. 7e and i). These biological alterations justified our choice of these genes for prognostic modeling.

Fig. 7.

Fig. 7

Open in a new tab

The impact of GALK1 and TFRC on the phenotype of BLCA cell line. a, c Effect of knocking down GALK1 and TFRC at mRNA levels in BIU87 by qRT-PCR. b, d Effect of knocking down GALK1 and TFRC at protein levels in BIU87 by western blot. e, f CCK8 and h Edu assays demonstrating the proliferation capacity after knocking down GALK1 and TFRC in BIU87. g The colony-forming ability of BIU87 after knocking down GALK1 and TFRC. i The migration capacity of BIU87 was assessed by transwell assays after GALK1 and TFRC knockdown. *P < 0.05, **P < 0.01, ***P < 0.001

Discussion

Although pathological staging remains the cornerstone for estimating prognosis and tumor aggressiveness in BLCA [32], its accuracy is limited. Notably, a higher T stage does not invariably translate into a worse outcome, largely because BLCA is biologically heterogeneous and genetically complex [33, 34]. Therefore, predicting the prognosis of BLCA patients remains challenging. In recent years, there has been a surge in studies on the TME, which is closely tied to cancer heterogeneity [35]. HLM are essential components of the TME, which drive tumor progression, drug resistance and TIME reprogramming [21, 22, 36, 37]. Hence, we were committed to exploring the association between these features and BLCA in order to accurately predict the patient prognosis and individually guide treatment.

In this study, we identified a novel hypoxia and lactate metabolism-related molecular subtypes by means of unsupervised clustering analysis. C1 and C2 exhibited distinct differences in prognosis and immune infiltration. Previous studies on TME gene features have predominantly focused on their individual clinical applications, overlooking the intricate interconnections among them [26, 28, 38]. To the best of our knowledge, genetic markers that incorporate genes related to HLM for BLCA prognostic assessment and to aid precision medicine have not yet been identified. We have thus developed an innovative HLMRS that can predict prognosis, immunophenotype, molecular subtyping, and treatment response by employing nine genes. Our signature was a good prognostic tool across multiple gene sets. The AUC values for predicting both long-term and short-term OS were superior, indicating better predictive ability than other hypoxia or metabolism related models.

ICIs, including PD-1/PD-L1 antagonists, have been extensively utilized in the management of advanced BLCA in recent years. Several studies, such as the JAVELIN Bladder 100 phase III trial and the CheckMate 275 phase II single-arm study, have confirmed the effectiveness of ICB [39, 40]. However, the response rate of BLCA to ICB remains only 15–20% [41]. The composition and functional status of immune cells within the TME are the primary determinants of the efficacy of ICIs. Our presentation of the immune landscape revealed that the TME of the High RS was densely infiltrated with tumor-associated fibroblasts, neutrophils, and M2 macrophages, which negatively affected the effectiveness of immunotherapy [19, 42]. We subsequently interrogated public datasets to compare ICIs efficacy between two risk score groups. High RS patients exhibited features consistent with immune evasion and displayed poorer responses to ICI therapy, corroborating earlier findings.

In the current clinical treatment of BLCA, the medical community has also begun to focus on molecular classification characterized by gene expression, in addition to the conventional staging system based on clinicopathological features [43, 44]. To date, seven BLCA molecular classifications have been reported, among which the consensus molecular classification is the most commonly used [23]. In general, the basal subtype typically exhibits high aggressiveness and a limited response to conventional chemotherapy. In contrast, the luminal subtype has a relatively favorable prognosis and frequently results in increased chemosensitivity [45, 46]. However, the complexity of detection methods and the classifications of molecular subtypes hinder the clinical application of molecular typing. By contrast, the HLMRS reproducibly distinguishes basal from luminal subtypes with high accuracy, providing a streamlined and clinically feasible solution. Patients with high HLMRS scores were mostly assigned to the basal subtype, whereas low scores identified the luminal subtype. Subsequent drug-sensitivity analyses further revealed that the High RS group exhibited reduced responsiveness to the most commonly employed chemotherapeutics for BLCA, with the notable exception of paclitaxel analogues. Thus, we inferred that patients exhibiting elevated HLMRS are more likely appropriate candidates for paclitaxel-based chemotherapy or GC. Conversely, individuals with low HLMRS exhibit an elevated response rate to various chemotherapeutic protocols, including the combination of dose-dense methotrexate, vinblastine, doxorubicin, and cisplatin (ddMVAC) or GC [47].

Certain modeling genes within the HLMRS have been shown to be correlated with proliferation, the TME, or the effectiveness of immunotherapy across various cancers [48–50]. In vitro experiments demonstrated the impact of GALK1 and TFRC on the proliferation and migration of BLCA. This not only demonstrated the reliability of our bioinformatics analysis, but also identified new potential drug targets for the clinical treatment of BLCA.

Nonetheless, our study is subject to several limitations. HLMRS was developed via bioinformatics analysis of public databases. Although our conclusions were validated in several external validation cohorts, bias may still exist. Moreover, the molecular mechanisms by which GALK1 and TFRC influence tumor development need to be further explored.

Conclusion

In summary, we performed molecular typing of BLCA on the basis of HLMRGs 

These findings revealed the impact of HLM on patient prognosis and the TME. We also developed and validated a robust gene signature based on the above characteristics, which can be used to predict patient prognosis, MIBC molecular subtype and treatment efficacy. In addition, in vitro experiments confirmed that GALK1 and TFRC could be potential therapeutic targets for BLCA. These findings will help to further reveal the clinical value of TME and improve precision medicine.

Supplementary Information

Supplementary Material 1. (13KB, xlsx)
Supplementary Material 2. (4.3MB, docx)
Supplementary Material 3. (1.4MB, docx)

Acknowledgements

We acknowledged the clinical contributors and the data producers from TCGA and GEO.

Authors’ contributions

Conceptualization: J.C., YH.Z.; Methodology, YH.Z.; Formal analysis and investigation: YH.Z., PX.L.; Writing - original draft preparation: SW.Y.; Writing - review and editing: Z.S.; Funding acquisition: J.C.; Supervision: XY.Z.,NZ.Z.; All authors read and approved the final manuscript.

Funding

This work was funded by the Qingdao Science and Technology Demonstration and Guidance Special Fund for the Benefit of the People (Grant No. 21-1-4-rkjk-7-nsh) and the Natural Science Foundation of Shandong Province (Grant No.ZR2023MH231).

Data availability

The datasets generated and/or analyzed during the current study are available in the TCGA repository, (https://portal.gdc.cancer.gov/) and GEO repository, (https://www.ncbi.nlm.nih.gov/).

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Qilu Hospital of Shandong University (Approval No.: KYLL-202407(X)-Z046-1). Our study obtained informed consent from all participants and was in compliance with the Helsinki Declaration.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Footnotes

Publisher’s Note

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

Contributor Information

Nianzhao Zhang, Email: 18560083906@163.com.

Jun Chen, Email: chenjunxinxiang@163.com.

References

  • 1.Bin RI, Khan AM, Catto JW, Hussain SA. Bladder cancer: shedding light on the most promising investigational drugs in clinical trials. Expert Opin Inv Drug. 2021;30(8):837–55. [DOI] [PubMed] [Google Scholar]
  • 2.Powles T, Bellmunt J, Comperat E, De Santis M, Huddart R, Loriot Y, et al. Bladder cancer: Esmo clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol. 2022;33(3):244–58. [DOI] [PubMed] [Google Scholar]
  • 3.Tran L, Xiao JF, Agarwal N, Duex JE, Theodorescu D. Advances in bladder cancer biology and therapy. Nat Rev Cancer. 2021;21(2):104–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Cumberbatch M, Jubber I, Black PC, Esperto F, Figueroa JD, Kamat AM, et al. Epidemiology of bladder cancer: a systematic review and contemporary update of risk factors in 2018. Eur Urol. 2018;74(6):784–95. [DOI] [PubMed] [Google Scholar]
  • 5.Montazeri K, Bellmunt J. Erdafitinib for the treatment of metastatic bladder cancer. Expert Rev Clin Pharmacol. 2020;13(1):1–6. [DOI] [PubMed] [Google Scholar]
  • 6.Li Z, Li J, Bai X, Huang X, Wang Q. Tumor microenvironment as a complex milieu driving cancer progression: a mini review. Clin Transl Oncol. 2025;27(5):1943–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Guo Y, Zheng Z, Mao S, Yang F, Wang R, Wang H, et al. Metabolic-associated signature and hub genes associated with immune microenvironment and prognosis in bladder cancer. Mol Carcinog. 2023;62(2):185–99. [DOI] [PubMed] [Google Scholar]
  • 8.Guelfi S, Hodivala-Dilke K, Bergers G. Targeting the tumour vasculature: from vessel destruction to promotion. Nat Rev Cancer. 2024;24(10):655–75. [DOI] [PubMed] [Google Scholar]
  • 9.Vaupel P, Mayer A, Hockel M. Tumor hypoxia and malignant progression. Method Enzymol. 2004;381:335–54. [DOI] [PubMed] [Google Scholar]
  • 10.Li Y, Liu F, Cai Q, Deng L, Ouyang Q, Zhang XH, et al. Invasion and metastasis in cancer: molecular insights and therapeutic targets. Signal Transduct Target Ther. 2025;10(1):57. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Lin S, Chai Y, Zheng X, Xu X. The role of Hif in angiogenesis, lymphangiogenesis, and tumor microenvironment in urological cancers. Mol Biol Rep. 2023;51(1):14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.He P, Dai Q, Wu X. New insight in urological cancer therapy: from epithelial-mesenchymal transition (EMT) to application of nano-biomaterials. Environ Res. 2023;229:115672. [DOI] [PubMed] [Google Scholar]
  • 13.Kawahara I, Yoshino H, Fukumoto W, Arima J, Saito S, Li G, et al. Targeting metabolic reprogramming to overcome drug resistance in advanced bladder cancer: insights from gemcitabine- and cisplatin-resistant models. Mol Oncol. 2024;18(9):2196–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Riera-Domingo C, Audige A, Granja S, Cheng WC, Ho PC, Baltazar F, et al. Immunity, hypoxia, and metabolism-the menage a trois of cancer: implications for immunotherapy. Physiol Rev. 2020;100(1):1–102. [DOI] [PubMed] [Google Scholar]
  • 15.Jing X, Yang F, Shao C, Wei K, Xie M, Shen H, et al. Role of hypoxia in cancer therapy by regulating the tumor microenvironment. Mol Cancer. 2019;18(1):157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Zhang W, Xia M, Li J, Liu G, Sun Y, Chen X, et al. Warburg effect and lactylation in cancer: mechanisms for chemoresistance. Mol Med. 2025;31(1):146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Vegran F, Boidot R, Michiels C, Sonveaux P, Feron O. Lactate influx through the endothelial cell monocarboxylate transporter MCT1 supports an NF-kappaB/IL-8 pathway that drives tumor angiogenesis. Cancer Res. 2011;71(7):2550–60. [DOI] [PubMed] [Google Scholar]
  • 18.Feng Q, Liu Z, Yu X, Huang T, Chen J, Wang J, et al. Lactate increases stemness of CD8 + T cells to augment anti-tumor immunity. Nat Commun. 2022;13(1):4981. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Boutilier AJ, Elsawa SF. Macrophage polarization states in the tumor microenvironment. Int J Mol Sci. 2021;22(13):6995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Sun Y, Wang H, Cui Z, Yu T, Song Y, Gao H, et al. Lactylation in cancer progression and drug resistance. Drug Resist Updat. 2025;81:101248. [DOI] [PubMed] [Google Scholar]
  • 21.Wang Y, Wang X, Liu Y, Xu J, Zhu J, Zheng Y, et al. A novel hypoxia- and lactate metabolism-related prognostic signature to characterize the immune landscape and predict immunotherapy response in osteosarcoma. Front Immunol. 2024;15:1467052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Li J, Qiao H, Wu F, Sun S, Feng C, Li C, et al. A novel hypoxia- and lactate metabolism-related signature to predict prognosis and immunotherapy responses for breast cancer by integrating machine learning and bioinformatic analyses. Front Immunol. 2022;13:998140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Kamoun A, de Reyniès A, Allory Y, Sjödahl G, Robertson AG, Seiler R, et al. A consensus molecular classification of muscle-invasive bladder cancer. Eur Urol. 2020;77(4):420–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Maeser D, Gruener RF, Huang RS. Oncopredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform. 2021;22(6):bbab260. [DOI] [PMC free article] [PubMed]
  • 25.Yu Z, Liao J, Chen Y, Zou C, Zhang H, Cheng J, et al. Single-cell transcriptomic map of the human and mouse bladders. J Am Soc Nephrol. 2019;30(11):2159–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Sun X, Zhou Z, Zhang Y, Wang J, Zhao X, Jin L, et al. Identification and validation of a hypoxia-related prognostic and immune microenvironment signature in bladder cancer. Cancer Cell Int. 2021;21(1):251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Sun X, Zhang Y, Chen Y, Xin S, Jin L, Liu X, et al. In silico establishment and validation of novel lipid metabolism-related gene signature in bladder cancer. Oxid Med Cell Longev. 2022;2022:3170950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Shi W, Dong J, Zhong B, Hu X, Zhao C. Predicting the prognosis of bladder cancer patients through integrated multi-omics exploration of chemotherapy-related hypoxia genes. Mol Biotechnol. 2025;67(6):2367–81. [DOI] [PMC free article] [PubMed]
  • 29.Zhou Z, Chai Y, Li Y, Zhang Y, Wang T. Lipid metabolism subtypes reflects the prognosis and immune profiles of bladder cancer patients by Nmf clustering and building related signature. Discover Oncol. 2024;15(1):796. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Jing W, Wang G, Cui Z, Li X, Zeng S, Jiang X, et al. Tumor-neutrophil cross talk orchestrates the tumor microenvironment to determine the bladder cancer progression. Proc Natl Acad Sci U S A. 2024;121(20):e1982112175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Debatin NF, Bady E, Mandelkow T, Huang Z, Lurati M, Raedler JB, et al. Prognostic impact and spatial interplay of immune cells in urothelial cancer. Eur Urol. 2024;86(1):42–51. [DOI] [PubMed] [Google Scholar]
  • 32.Paner GP, Montironi R, Amin MB. Challenges in pathologic staging of bladder cancer: proposals for fresh approaches of assessing pathologic stage in light of recent studies and observations pertaining to bladder histoanatomic variances. Adv Anat Pathol. 2017;24(3):113–27. [DOI] [PubMed] [Google Scholar]
  • 33.Wei Z, Xv Y, Liu H, Li Y, Yin S, Xie Y, et al. A ct-based deep learning model predicts overall survival in patients with muscle invasive bladder cancer after radical cystectomy: a multicenter retrospective cohort study. Int J Surg. 2024;110(5):2922–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Li S, Zheng F, Wang Z, Xiong S, Zeng J, Xu S, et al. Integrated analysis of bulk and single-cell RNA-seq data reveals cell differentiation-related subtypes and a scoring system in bladder cancer. J Cell Mol Med. 2024;28(19):e70111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Grausenburger R, Herek P, Shariat SF, Englinger B. Recent contributions of single-cell and spatial profiling to the understanding of bladder cancer. Curr Opin Urol. 2024;34(4):236–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Sun J, Jiang R, Hou L, Wang L, Li M, Dong H, et al. Identification of a combined hypoxia and lactate metabolism prognostic signature in lung adenocarcinoma. BMC Pulm Med. 2024;24(1):323. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Huang A, Sun Z, Hong H, Yang Y, Chen J, Gao Z, et al. Novel hypoxia- and lactate metabolism-related molecular subtyping and prognostic signature for colorectal cancer. J Transl Med. 2024;22(1):587. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Deng Y, Zhu G, Mi X, Jing X. Prognostic implication of a novel lactate score correlating with immunotherapeutic responses in pan-cancer. Aging (Albany Ny). 2024;16(1):820–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Powles T, Park SH, Voog E, Caserta C, Valderrama BP, Gurney H, et al. Avelumab maintenance therapy for advanced or metastatic urothelial carcinoma. New Engl J Med. 2020;383(13):1218–30. [DOI] [PubMed] [Google Scholar]
  • 40.Sharma P, Retz M, Siefker-Radtke A, Baron A, Necchi A, Bedke J, et al. Nivolumab in metastatic urothelial carcinoma after platinum therapy (checkmate 275): a multicentre, single-arm, phase 2 trial. Lancet Oncol. 2017;18(3):312–22. [DOI] [PubMed] [Google Scholar]
  • 41.Osipov A, Saung MT, Zheng L, Murphy AG. Small molecule immunomodulation: the tumor microenvironment and overcoming immune escape. J Immunother Cancer. 2019;7(1):224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Liu H, Zhao H, Zhou M, Zhao X, Lu Y. Neutrophils in cancer drug resistance: roles and therapeutic opportunities. Cancer Lett. 2024;611:217417. [DOI] [PubMed] [Google Scholar]
  • 43.Robertson AG, Kim J, Al-Ahmadie H, Bellmunt J, Guo G, Cherniack AD, et al. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell. 2018;174(4):1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Lindgren D, Frigyesi A, Gudjonsson S, Sjodahl G, Hallden C, Chebil G, et al. Combined gene expression and genomic profiling define two intrinsic molecular subtypes of urothelial carcinoma and gene signatures for molecular grading and outcome. Cancer Res. 2010;70(9):3463–72. [DOI] [PubMed] [Google Scholar]
  • 45.Lindskrog SV, Prip F, Lamy P, Taber A, Groeneveld CS, Birkenkamp-Demtroder K, et al. An integrated multi-omics analysis identifies prognostic molecular subtypes of non-muscle-invasive bladder cancer. Nat Commun. 2021;12(1):2301. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Comprehensive molecular characterization of urothelial bladder carcinoma. Nature. 2014;507(7492):315–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Pfister C, Gravis G, Flechon A, Chevreau C, Mahammedi H, Laguerre B, et al. Dose-dense methotrexate, vinblastine, doxorubicin, and cisplatin or gemcitabine and cisplatin as perioperative chemotherapy for patients with nonmetastatic muscle-invasive bladder cancer: results of the getug-afu v05 vesper trial. J Clin Oncol. 2022;40(18):2013–22. [DOI] [PubMed] [Google Scholar]
  • 48.Tang R, Tay SS, Sharbeen G, Herrmann D, Youkhana J, Timpson P, et al. Bystander expression of atypical chemokine receptor 2 protects t cells from chemoattraction towards cancer-associated fibroblasts. Eur J Immunol. 2025;55(2):e202451215. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Wang X, Zhou Y, Ning L, Chen J, Chen H, Li X. Knockdown of anxa10 induces ferroptosis by inhibiting autophagy-mediated Tfrc degradation in colorectal cancer. Cell Death Dis. 2023;14(9):588. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Liu Z, Sun L, Zheng B, Wang H, Qin X, Zhang P, et al. The value of ATAD3A as a potential biomarker for bladder cancer. Cancer Med. 2023;12(24):22395–406. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1. (13KB, xlsx)
Supplementary Material 2. (4.3MB, docx)
Supplementary Material 3. (1.4MB, docx)

Data Availability Statement

The datasets generated and/or analyzed during the current study are available in the TCGA repository, (https://portal.gdc.cancer.gov/) and GEO repository, (https://www.ncbi.nlm.nih.gov/).

Articles from BMC Cancer are provided here courtesy of BMC

RESOURCES