scRNA-seq reveals epithelial heterogeneity in bladder cancer and establishes a cancer stem cell prognostic model

scRNA-seq reveals epithelial heterogeneity in bladder cancer and establishes a cancer stem cell prognostic model | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article scRNA-seq reveals epithelial heterogeneity in bladder cancer and establishes a cancer stem cell prognostic model Biao Zhang, Yi Liu, Fei Yang, Man Yang, Yu Pan, Yingpin Lei, Chunhong Li, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7536468/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Bladder cancer (BLCA) is a common malignancy with increasing incidence globally. Epithelial cells play crucial roles in tumor development and metastasis. Single-cell RNA sequencing (scRNA-seq) enables investigation of cellular heterogeneity. This study aims to analyze epithelial cell heterogeneity in BLCA, identify biomarkers, and develop prognostic models using machine learning to explore their clinical significance. Methods We integrated scRNA-seq and bulk RNA-seq data from TCGA and GEO databases, including muscle-invasive bladder cancer (MIBC), non-muscle-invasive bladder cancer (NMIBC), and adjacent tissue samples. Data processing included Seurat clustering, Harmony batch correction, copy number variation analysis, KEGG enrichment analysis, CellChat intercellular communication analysis, and Monocle3 trajectory analysis. A cancer stem cell-related prognostic index (CSCRPI) was constructed using four machine learning algorithms based on cancer stem cell (CSC) subpopulation marker genes. Results We identified 39 clusters, 9 cell types, and 11 epithelial cell subtypes, including a cancer stem cell subpopulation (Epi9_CSC). Epi9_CSC was highly enriched in MIBC and promoted invasion and recurrence. Pathway analysis revealed that Epi9_CSC secretes collagen proteins that interact with integrin and CD44 receptors, activating downstream signaling pathways including Focal adhesion, PI3K-Akt, NF-κB, and MAPK pathways. Drug sensitivity analysis identified AZD1208, IAP inhibitors, and Nutlin-3a as potential therapeutic agents targeting Epi9_CSC. The CSCRPI scoring system, based on 5 key feature genes from Epi9_CSC, accurately predicted BLCA patient prognosis and provided clinical guidance. Conclusions This study identified Epi9_CSC as a highly invasive cancer stem cell subpopulation that drives BLCA malignancy through extracellular matrix remodeling and multiple oncogenic pathway activation. The CSCRPI system offers valuable prognostic insights and potential therapeutic targets for precision medicine in BLCA treatment. BLCA Single-cell RNA sequencing Cancer stem cells Targeted therapy Prognostic model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Background BLCA is a malignant tumor that originates in the bladder lining and is one of the most common cancers in the urinary system. The exact cause of BLCA remains unclear, but various factors are believed to increase its risk, including smoking, exposure to certain chemicals (such as aniline dyes and aromatic amines), chronic bladder inflammation or infection, and genetic susceptibility [ 1 ]. Clinically, the most prominent early symptom of BLCA is painless hematuria, and patients may also experience urinary frequency, urgency, and dysuria [ 2 ], which often lead them to seek medical attention. As the ninth most common cancer globally, BLCA accounted for approximately 614,000 new cases and 220,000 deaths in 2022, with a significantly higher burden and incidence in men than in women [ 3 ]. Studies suggest that in the next decade, the incidence and mortality rates of BLCA in China will further increase [ 4 ], placing a substantial burden on the country’s economy. The overall five-year survival rate for BLCA is approximately 77% [ 5 ], with cancer staging being a crucial factor in prognosis and treatment. As the tumor stage progresses, the survival rate of BLCA decreases [ 6 ]. Tumor staging is closely linked to the TME [ 7 ]. The TME influences tumor progression through various complex mechanisms. Its main components include epithelial cells, immune cells, endothelial cells, fibroblasts, and ECM, which interact closely with tumor cells to promote tumor growth, angiogenesis, and metastasis [ 8 ]. In fact, most malignant tumors are composed of several subpopulations of tumor cells with different phenotypes, a fact that has been widely confirmed by research over the past years [ 9 ]. These studies have revealed the degree and complexity of tumor cell heterogeneity, providing a deeper understanding of cancer. Today, the phenotypic diversity of tumor cells within the tumor is recognized as a major driver of therapeutic resistance, drawing increasing attention [ 10 ]. In the tumor microenvironment, epithelial cells, as one of the initial cell types that form tumors, influence tumor progression and metastasis through various mechanisms. One key mechanism is EMT, during which epithelial cells lose polarity and tight cell junctions, acquiring enhanced migratory and invasive abilities, thus facilitating tumor metastasis [ 11 ]. Additionally, the cytokines and growth factors secreted by epithelial cells can regulate the behavior of immune cells, recruiting immunosuppressive cells such as tumor-associated macrophages and regulatory T cells, helping tumor cells evade immune surveillance [ 12 ][ 13 ]. Meanwhile, epithelial cells promote angiogenesis by secreting vascular endothelial growth factor (VEGF) [ 14 ], providing the tumor with sufficient oxygen and nutrients to support its rapid growth. In this complex environment, tumor stem cells also play an important role. Tumor stem cells, with their self-renewal capacity, not only drive tumor growth and recurrence but also interact with epithelial cells, further enhancing the malignancy of the tumor. These interactions make tumors more difficult to treat with traditional therapies [ 15 ]. Therefore, therapeutic strategies targeting the EMT process in epithelial cells, their angiogenic capacity, and their interaction with tumor stem cells could provide new directions for effectively inhibiting tumor progression and metastasis. In traditional transcriptome sequencing (bulk RNA-seq), researchers typically sequence mixed samples from large populations of cells to obtain overall gene expression information. However, this approach may obscure the specific expression patterns and functions of epithelial cells within the tumor microenvironment. In contrast, scRNA-seq enables gene expression analysis at the single-cell level, offering a more precise way to reveal cellular heterogeneity. This technological advancement provides a new perspective for studying the specific roles of epithelial cells in various cancers, especially in revealing their diversity and complexity within the tumor microenvironment [ 16 ]. This study will utilize single-cell sequencing data to analyze the characteristics of epithelial cells in BLCA and construct a detailed atlas of epithelial cells within the BLCA tumor microenvironment. By identifying specific tumor cell subpopulations and applying machine learning methods to develop and validate prognostic models for these specific tumor cells, the study will further explore their clinical significance and biological characteristics. This will help gain a deeper understanding of the role of epithelial cells in the onset and progression of BLCA and provide new insights for precision and personalized treatment strategies. This study will utilize single-cell sequencing data to analyze epithelial cell characteristics in BLCA and construct a detailed atlas of epithelial cells within the BLCA tumor microenvironment. By identifying specific tumor cell subpopulations and combining machine learning methods to develop and validate prognostic models targeting these specific tumor cells, we will further explore their clinical significance and biological characteristics. This will help deepen our understanding of the role of epithelial cells in BLCA occurrence and progression, and provide new insights for precision therapy and personalized treatment strategies. Materials and methods Data Acquisition The scRNA-seq datasets for BLCA, including GSE192575, GSE145137, GSE135337, GSE222315, and GSE129845, were downloaded from the GEO database. Tumor samples were classified into MIBC for grades T2 and above, and NMIBC for grades below T2. The dataset comprises samples from 5 muscle-invasive bladder cancer (MIBC), 14 non-muscle-invasive bladder cancer (NMIBC), and 8 adjacent normal (Para) tissues. Detailed sample information is provided in Supplementary Table S1 .The bulk RNA-seq datasets for BLCA were obtained from the TCGA and GEO databases under the following accession IDs: TCGA-BLCA and GSE31684. The TCGA-BLCA dataset includes 372 tumor samples with complete clinical information, which were used as the training cohort. The bulk RNA-seq expression matrix from GSE31684 was normalized and used as an external validation cohort. The corresponding clinical information for these bulk RNA-seq datasets includes overall survival (OS), survival status, age, gender, pathological stage, and TNM stage. The drug sensitivity training dataset from GDSC2 was downloaded from the GDSC database ( https://www.cancerrxgene.org/ ). Furthermore, the genomic annotation file for copy number variation analysis, hg38_gencode_v27, was downloaded from the Broad Institute server ( https://data.broadinstitute.org/Trinity/CTAT/cnv ). Data Quality Control and Filtering The scRNA-seq expression matrix obtained in the previous step was imported into R. Data quality control was performed using the Seurat package (version 3.2.2). The mitochondrial gene ratio (mt_percent) for each cell was calculated by identifying genes that start with "MT-". The erythrocyte gene ratio (HB_percent) for each cell was calculated using a known list of erythrocyte genes. Cells with a mitochondrial gene ratio greater than 35% and cells with an erythrocyte gene ratio greater than 3% were filtered out. Only cells with gene and transcript counts within the range of the mean ± 2 standard deviations were retained. Data Normalization and Feature Gene Selection The Harmony package (version 1.2.3) was used to correct for batch effects across samples using the RunHarmony function. The sample name in the metadata was chosen as the integration variable. The Parameters were set with a theta value of 3, a lambda value of 0.6, and a maximum of 20 iterations. Data Integration The Seurat package's NormalizeData function was applied to normalize the data, with the normalization.method set to "LogNormalize" and the scale.factor set to 10,000. The FindVariableFeatures function was used to select feature genes for subsequent analysis, with the selection.method set to "vst" and nfeatures set to 2,000. The ScaleData function was then applied to normalize the integrated data. Dimensionality Reduction and Cell Clustering Principal component analysis (PCA) was performed using the Seurat package's RunPCA function, with the number of principal components (npcs) set to 30. The data was visualized and reduced in dimensionality using the RunTSNE and RunUMAP functions, with dims set to 1:20. Cell clustering was performed using the FindNeighbors and FindClusters functions, with dims set to 1:20 and resolution set to 0.8. Differentially expressed genes for each cluster were identified using the FindAllMarkers function, with min.pct set to 0.5, logfc.threshold set to 0.5, and the test.use set to "wilcox". Based on the reference literature for cell cluster-specific marker genes, each cell cluster was defined according to its characteristic gene expression profile. Data Processing and Cell Clustering of Epithelial Cell Clusters After extracting the epithelial cell clusters, the data was processed using the same methods as described above, including normalization, feature gene selection, scaling, dimensionality reduction, and cell clustering. The epithelial cell subpopulations were annotated using BLCA molecular subtype-related marker genes and tumor stem cell genes. CNV Analysis CNV analysis on epithelial cell subgroups using the infercnv package (version 1.22.0). Randomly select 800 plasma cells and 800 endothelial cells from all samples as the reference group for CNV analysis. Set the cutoff value to 0.2 to filter out low expression data, choose 'ward.D2' as the hierarchical clustering method, and enable the denoising step to reduce technical noise. KEGG Analysis To identify expression characteristics of each epithelial cell subpopulation, genes were selected based on the following criteria: an average log fold change (logFC) greater than 2, an expression proportion (pct.1) exceeding 30% within the subpopulation, and an expression proportion (pct.2) below 10% in other subpopulations. These selected genes were used as gene sets for KEGG enrichment analysis, performed using the compareCluster function from the clusterProfiler package. The Parameters were set as follows: fun = "enrichKEGG", organism = "hsa", and pvalueCutoff = 0.05. KEGG pathways with P-values less than 0.05 were retained and exported as enrichment result tables. All genes from the Epi9_CSC-enriched pathways that were differentially expressed compared to other subpopulations were uploaded to KEGG Mapper for visualization. Differential Gene Comparison Analysis Between Groups Differentially expressed genes between different sample sources were identified using the FindMarkers function from the Seurat package. The log fold change threshold was set to 1, min.pct was set to 0.3, and the test.use was selected as "wilcox." Subsequently, a volcano plot was generated using the ggplot2 package. Drug Sensitivity Test Analysis of Tumor Stem Cell Subpopulations After extracting the tumor stem cell subpopulations, drug sensitivity analysis was performed using the oncoPredict package (version 1.2). The ComBat method was applied to batch-correct the expression data for both training and test sets to reduce batch effect. The removeLowVaryingGenes Parameter was set to 0.2, removing genes with low variability (below the 20th percentile) from the original data to simplify the model and minimize noise interference. A boxplot of drug sensitivity scores was generated using the ggplot2 package, with specific drugs marked in red for easier identification. The drug sensitivity analysis table was also exported. Pseudotime Analysis The Monocle3 package (version 1.3.7) was used to read the epithelial cell cluster data object constructed by the Seurat package and convert it to Monocle3 format using the new_cell_data_set() function. The UMAP coordinates computed by Seurat were directly used for subsequent analysis. In constructing the cell trajectory, the learn_graph() function was called with a series of Parameters set: euclidean_distance_ratio was set to 380 to control the ratio of Euclidean distance between two tree apex nodes to the maximum path distance; geodesic_distance_ratio was set to 400 to control the ratio of geodesic distance to the tree diameter path length; minimal_branch_len was set to 70 to determine the minimum branch length to retain; prune_graph was set to TRUE to remove insignificant small branches; and nn.k was set to 3 to define the number of nearest neighbors calculated in the reverse graph embedding process. Finally, the order_cells() function was used to arrange the cells in pseudotime and construct and label the cell trajectory. Cell-Cell Interaction Analysis The cell cluster data object constructed by the Seurat package was imported into the R language's CellChat package. The "Secreted Signaling" reference database was selected for the analysis. The computeCommunProb function was applied to assign a probability value to each interaction and perform permutation tests to infer biologically significant cell-to-cell interactions. The computeCommunProbPathway function was used to infer the interactions between cells at the signaling pathway level. The aggregateNet function was applied to calculate the number of links or aggregate communication probabilities, thereby constructing an aggregated communication network between cells. Machine Learning-Based CSC-Related Gene Integration Method The CSC subgroup marker genes obtained from the FindAllMarkers function were filtered based on the condition of avg_log2FC > 2, and univariate Cox regression analysis was performed on the resulting genes, yielding 182 CSC subgroup marker genes associated with prognosis. Four machine learning algorithms were then used to identify core CSC prognosis-related genes. The four algorithms included Random Forest (RF), Lasso, XGBoost, and Decision Tree (DT). The TCGA cohort was set as the training cohort, and GSE31684 was used as the validation cohort. The results obtained from the four algorithms were combined using a weighted average method to select the top 10 genes as the central CSC prognosis-related genes. Construction and Validation of Tumor Stem Cell-Related Prognostic Index (CSCRPI) The top 10 central CSC prognosis-related genes selected by the weighted average method were subjected to multivariate Cox analysis to determine the most reliable independent prognostic factors. The CSCRPI for patients was calculated using the following formula:CSCRPI = coefficient(gene1) × expression(gene1) + coefficient(gene2) × expression(gene2) + ... + coefficient(genen) × expression(genen).In this formula, expression(gene n) represents the expression level of a specific gene, and coefficient(gene n) refers to the coefficient obtained from the multivariate Cox analysis. Based on the median CSCRPI, BLCA patients were divided into low-CSCRPI and high-CSCRPI groups. Kaplan-Meier analysis was performed using the "survival" R package to investigate the relationship between the survival status of BLCA patients and CSCRPI, with further validation conducted using the GSE31684 cohort. Nomogram Construction By combining clinical features of BLCA patients, such as age, gender, clinical pathological stage, and CSCRPI, through multivariate Cox and stepwise regression analysis, a prognostic nomogram was created. The nomogram and calibration plots were visualized using the "rms" package. ROC curve analysis was performed to evaluate the performance of CSCRPI in predicting the 1-year, 2-year, and 3-year overall survival rates of BLCA patients. Subsequently, decision curve analysis (DCA) was used to assess the net benefit of combining the nomogram with a model containing only clinical features. Furthermore, correlation and stratified analyses of CSCRPI based on the clinical features from the TCGA dataset were also performed. Tumor Immune Microenvironment Analysis and Immune Therapy Efficacy Evaluation To quantify the immune infiltration status in BLCA patients, the "IOBR" R package was used for analysis, employing nine different algorithms, including the MCP-counter, EPIC, xCell, CIBERSORT, IPS, quanTIseq, ESTIMATE, TIMER, and ssGSEA algorithms. Additionally, the Cancer Immunome Atlas (TCIA) was utilized to evaluate the potential response of BLCA patients to checkpoint immunotherapies. We also conducted a comprehensive analysis to explore the correlation between immune phenotype scores (IPS) for anti-PD-1 and anti-CTLA4 treatments and MPCDI in BLCA patients. Drug Sensitivity Analysis for High-CSCRPI and Low-CSCRPI Groups To personalize treatment, the "oncoppredict" R package was used to predict the chemotherapy sensitivity of BLCA patients with different MPCDI scores. The expression profile of patient tissues was matched with gene expression profiles from cancer cell lines, and the half-maximal inhibitory concentration (IC50) was calculated. The Wilcoxon test was applied to explore the differences in drug IC50 between the two groups, and a P-value of < 0.05 was considered statistically significant. To improve the accuracy of drug sensitivity analysis, GSCALite platforms' GDSC and CellMiner databases were used for further analysis. Statistical Analysis In this study, we performed statistical analyses using R software (version 4.4.0). Batch effects between samples were corrected using the Harmony algorithm. For differential gene expression analysis, we employed the Wilcoxon rank-sum test. Cell trajectory analysis was conducted using the Monocle3 software, and the significance of gene expression was assessed using the Wilcoxon test, with the Benjamini-Hochberg method applied to adjust for false positives. Additionally, permutation tests were used to validate the robustness and statistical significance of the trajectory model. The cell interaction network was also evaluated using permutation tests. Survival curves were generated using the Kaplan-Meier method. To compare multiple or two groups of data, we used the Wilcoxon test or Kruskal-Wallis test. Correlation assessments were conducted using Spearman correlation analysis. Results with a p-value less than 0.05 were considered statistically significant. For visualization purposes, a p -value less than 0.05 was represented as *, less than 0.01 as **, and less than 0.001 as ***. Results Identification of BLCA Cell Subtypes First, the data from cells originating from MIBC, NMIBC, and Para organizations were integrated, followed by batch effect removal, PCA analysis, dimensional reduction, clustering, and differential gene identification. A total of 160,155 cells were obtained, classified into 39 clusters. Based on the differentially expressed genes highly expressed in each cluster and known cell-specific marker genes (Table 1 ), 14 major cell types were identified: Urothelial cells, Tregs, Memory T cells, Effector T cells, Activated Memory B cells, Activated B cells, Plasma cells, Monocyte-macrophage, iCAF, matCAF, Smooth muscle cells, Angiogenic ECs, Inflammatory ECs, and Mast Cells (Fig. 1A). Figure 1D displays the marker gene expression patterns for each cell type. Further analysis of the cell proportions in MIBC, NMIBC, and adjacent normal tissue samples (Fig. 1B, D) revealed that urothelial cells comprise 66.79% of the total cells in MIBC, 50.60% in NMIBC, and 23.45% in Para-originating samples. Table 1 Marker genes of cell types in BLCA and adjacent normal tissue samples Cell Type Name Marker Genes Urothelial cells KRT8、TACSTD2、KRT19 Tregs CD3D, CD4, FOXP3, CTLA4, IKZF2 Memory T cells CD3D, CD4,IL7R, CCR7, SELL Effector T cells CD3D, CD8A, CD8B, IFNG, GZMB, PRF1 Activated Memory B cells CD19, CD79A, CD79B,MS4A1, CD27 Activated B cells CD19, CD79A, CD79B,CD69, CD80, CD40 Plasma cells CD19, CD79A, CD79B,IGLL5, IGHG1, IGHA1 Monocyte-macrophage CD14, CD68, MS4A7, FCGR3A iCAF IL6, CXCL1, LIF matCAF COL1A1, COL3A1, FN1 Smooth muscle cells TAGLN, ACTA2, MYLK Angiogenic ECs VWF, CDH5, PECAM1, VEGFA, KDR, DLL4 Inflammatory ECs VWF, CDH5, PECAM1,ICAM1, VCAM1, SELE Mast Cells KIT, TPSAB1, TPSB2 Urothelial Cell Clustering and Definition After extracting the urothelial cell population, re-clustering analysis was performed. After removing contaminating cells, a total of 83,672 cells were obtained, which were divided into 11 clusters. The sample source of each cluster was heterogeneous, indicating that the urothelial cell subpopulations exhibit heterogeneity under different pathological conditions. Clusters 1, 3, 4, and 5 predominantly originated from NMIBC, clusters 6, 8, and 10 primarily from MIBC, and cluster 9 mainly from MIBC and Paraffin-embedded tissues. The remaining clusters were found in MIBC, NMIBC, and Para (Fig. 2A,B). CNV analysis of each urothelial cell cluster revealed copy number amplifications or deletions in all clusters, with CNV scores being generally high, suggesting the presence of tumor cells in all clusters (Fig. 2E). Based on BLCA molecular subtype-related marker genes and tumor stem cell genes (Table 2 ), each cluster was named [ 17 ][ 18 ]. Clusters 0, 2, 3, and 10 were mixed clusters of multiple subtypes, clusters 1, 4, and 5 mainly expressed HER2 subtype marker genes, clusters 6 and 8 mainly expressed HER2 and luminal subtype marker genes, cluster 7 primarily expressed mesenchymal subtype marker genes, and cluster 9 primarily expressed tumor stem cell marker genes. These clusters were thus named as follows: Epi0_Mix, Epi1_HER2, Epi2_Mix, Epi3_Mix, Epi4_HER2, Epi5_HER2, Epi6_Luminal_HER2, Epi7_Mesenchymal, Epi8_Luminal_HER2, Epi9_CSC, and Epi10_Mix(Fig. 2C).In addition, using the tumor stem cell gene set (CD44, SOX2, IGF1R, SOX4, ARRB1, ARRB2, ALDH1A1, POU5F1, CDK1, DCLK1, NANOG) [ 18 ] and the EMT gene set (VIM, CDH2, FOXC2, SNAI1, SNAI2, TWIST1, FN1, ITGB6, MMP2, MMP3, MMP9, SOX10) [ 19 ], gene set scoring was performed on Urothelial cells. The results suggest that the tumor stem cell and EMT gene set scores for Epi9_CSC are high(Fig. 2D), further confirming that cluster 9 consists of tumor stem cells and that the CSC subpopulation exhibits an EMT phenotype [ 20 ]. Table 2 Characteristic genes related to molecular subtypes of BLCA and tumor stem cell genes Cell Type Name Marker Genes Luminal KRT20、UPK3A、FOXA1 Her2 ERBB2、GRB7、MUC1 Squamous KRT5、KRT14、TP63 Mesenchymal AXL、VIM、CDH2 Papillary FGFR3、TACC3 Neural WNT3A、SYP CSC CD44 KEGG Analysis of Urothelial Cell Subpopulations To understand the phenotypic diversity of tumor cells in BLCA, KEGG analysis was performed on the marker genes of urothelial cell subpopulations (Fig. 3A), including Epi1_HER2, Epi4_HER2, Epi5_HER2, Epi6_Luminal_HER2, and Epi9_CSC. The enrichment results for Epi1_HER2, Epi4_HER2, and Epi5_HER2 were: ribosome and Coronavirus disease - COVID-19. The enrichment results for Epi5_HER2 were: Epstein-Barr virus infection. The enrichment results for Epi9_CSC included: Focal adhesion, ECM-receptor interaction, Cytoskeleton in muscle cells, Proteoglycans in cancer, Human papillomavirus infection, PI3K-Akt signaling pathway, and Prion disease. We uploaded all genes with differential expression between the Epi9_CSC enrichment pathways and the other subpopulations to KEGG Mapper. The results showed that the four pathways—Focal adhesion, ECM-receptor interaction, Proteoglycans in cancer, and PI3K-Akt signaling pathway—are closely interconnected( refer to Supplementary Figures S1 -S4). Pseudotime Analysis of Cancer Stem Cells and Mesenchymal Subpopulations CSCs are a distinct subpopulation of tumor cells characterized by self-renewal, multipotency, and resistance to therapy. They are considered key drivers of tumor heterogeneity, invasiveness, and recurrence [ 21 ]. In BLCA, the mesenchymal-like subtype is the most aggressive and associated with the poorest prognosis [ 17 ]. Its hallmark is the pronounced activation of EMT, leading to high migratory capacity, drug resistance, and metastatic potential [ 22 ]. Recent studies have revealed a strong link between CSCs and the mesenchymal subtype: CSCs may evolve into mesenchymal-like tumor cells through EMT or other mechanisms, promoting the progression of BLCA toward a more invasive phenotype [ 23 ].In this study, CSCs and mesenchymal-like cells were extracted for pseudotime analysis to reconstruct the continuous differentiation trajectory from CSCs to the mesenchymal subtype. This approach enabled us to identify key transitional stages in cell state evolution and to explore the dynamic EMT process and its core regulatory factors.During the differentiation of BLCA CSCs into the mesenchymal subtype, dynamic changes in gene expression revealed critical biological mechanisms. Downregulated genes included NT5E (CD73), SERPINA1, TGFBI, LAMC2, DCBLD2, CMTM3, and COL7A1, while upregulated genes included COL1A2, COL6A2, and COL3A1 (Fig. 3B). Drug Sensitivity Analysis of Cancer Stem Cell Subpopulations To explore the sensitive drugs for BLCA tumor stem cells, this study conducted GDSC drug sensitivity analysis on the tumor stem cell subpopulations. The results showed that the tumor stem cell subpopulations were sensitive to Bortezomib, Dactinomycin, Docetaxel, Daporinad, Sepantronium bromide, Vinblastine, Eg5_9814, Vinorelbine, Staurosporine, Dinaciclib, Paclitaxel, and other drugs (Fig. 3C). Additionally, we explored the targets and pathways of these sensitive drugs using the GDSC database (Table 3 ), and evaluated the expression of target genes of certain drugs in CSCs (Fig. 3D). Table 3 Drug Targets and Pathways Sensitive to Cancer Stem Cell Subpopulations. Drug Name Drug ID Drug Target Target Pathway Dactinomycin 1911 RNA polymerase Other Docetaxel 1007 Microtubule stabiliser Mitosis Daporinad 1248 NAMPT Metabolism Sepantronium bromide 1941 BIRC5 Apoptosis regulation Vinblastine 1004 Microtubule destabiliser Mitosis Eg5_9814 1712 KSP11 Other Vinorelbine 2048 Microtubule destabiliser Mitosis Staurosporine 1034 Broad spectrum kinase inhibitor RTK signaling Dinaciclib 1180 CDK1, CDK2, CDK5, CDK9 Cell cycle Paclitaxel 1080 Microtubule stabiliser Mitosis Bortezomib 1191 Proteasome Protein stability and degradation Comparative Analysis of Urothelial cells in MIBC, NMIBC, and Adjacent Normal Tissues Due to differences in the tumor microenvironment between MIBC and NMIBC, this study conducted differential gene analysis of Urothelial cells in MIBC, NMIBC, and adjacent normal tissues, with a focus on the top 20 genes to identify potential clinical diagnostic and therapeutic targets(Fig. 4 A). In the comparison between MIBC and NMIBC, upregulated genes in Urothelial cells included immune-related genes (e.g., LYZ, LEAP2, IGKC, HLA-DRB5, DEFB1, IFI27, IFI6, and IFITM3), metabolism and synthesis-related genes (e.g., ECH1, INSIG1, and MRPS12), and cell structure and function-related genes (e.g., PHGR1, LGALS4, BMP2, and TMEM19). Downregulated genes were involved in energy metabolism (e.g., ATP5MJ, ATP5MC2, ATP5ME, ATP5F1E, and NDUFAF8), chromatin structure and gene expression regulation (e.g., H2AZ1, H3-3A, H3-3B, H4C3, and H1-2), signal transduction and protein degradation (e.g., RACK1 and ELOC), as well as RNA regulation (e.g., SNHG29 and DANCR).In the comparison between MIBC and Para group, upregulated genes mainly included immune-related genes (e.g., LYZ, IFI27, LEAP2, ISG15, and IFI6) and metabolism-related genes (e.g., SULT1E1, SCD, and DHCR24). Downregulated genes were related to cell adhesion (e.g., ITGA2, ITGA6, LMO7), inflammation and immunity (e.g., CCL20, PTGS2), and signaling pathways (e.g., DKK1 and AREG).In the comparison between NMIBC and Para group, upregulated genes were related to immunity and inflammation (e.g., S100A9, PSORS1C2, and IFI27), cellular energy metabolism (e.g., ATP5F1C, ATP6V0C, and ATP5MC1), while downregulated genes were associated with the extracellular matrix (e.g., THBS1), cell migration and tissue lubrication (e.g., HAS3), and signaling pathway regulation (e.g., DKK1). Additionally, in the differential analysis of Epi9_CSC between the MIBC and NMIBC groups(Fig. 4 B), upregulated genes were primarily associated with the extracellular matrix and structure (e.g., TFPI2, TGFBI, COL8A1, and SPOCK1), which may reflect an enhancement of MIBC functionality in these aspects. Inference of Epithelial Cell-Cell Communication To gain deeper insights into how Urothelial cells in different subpopulations interact with other cells and with themselves through ligands and receptors, we focused on the molecular interactions between these cells. Specifically, Urothelial cells from each subpopulation secrete specific ligands that bind to receptors on target cells, thereby activating various signaling pathways and regulating the functions of epithelial cells. In this study, we used the CellChat tool to identify intercellular communication pathways and explore how Urothelial cells from different subpopulations promote tumor initiation and progression through ligand-receptor interactions. We first compared the total number and strength of cell-cell communication interactions in Urothelial cells across different pathological states. The results showed that MIBC had the highest number and strongest interactions, followed by NMIBC, with the Para group showing the fewest and weakest interactions (Fig. 5 A). Moreover, in the MIBC pathological state, Epi9_CSC served as the primary sender and receiver of communication; in contrast, in the NMIBC and Para pathological states, Epi7_Mesenchymal was the major communicator (Fig. 5 B). Since Epi9_CSC plays a central role in tumor development, we further examined the complex intercellular communication pathways involving Epi9_CSC by calculating the strength of outgoing and incoming interactions for each signaling pathway in different pathological states. Notably, in the MIBC pathological state, the main outgoing signaling pathways from Epi9_CSC were collagen and laminin pathways, which were widely received by various epithelial cell subpopulations. Additionally, the primary incoming signaling pathways for Epi9_CSC also included its own collagen and laminin pathways (Fig. 5 C). In the NMIBC pathological state, the primary outgoing signaling pathways from Epi9_CSC were the CypA and MIF pathways, while the main incoming signaling pathway was collagen from Epi7_Mesenchymal(Fig. 5 D). In the Para pathological state, the main outgoing signaling pathways from Epi9_CSC were collagen and APP pathways, with the primary incoming signaling pathway still being collagen from Epi7_Mesenchymal(Fig. 5 E). By combining the primary outgoing and incoming pathways for Epi9_CSC in different pathological states, we infer that the collagen pathway is central to the interactions between Epi9_CSC and other cells, as well as its self-interaction. Given the central role of the collagen pathway in intercellular communication of Epi9_CSC, we further explored the major ligand-receptor pairs involved in this pathway between Epi9_CSC and other epithelial cell subpopulations. The results showed that the strongest communication probabilities in the collagen pathway occurred between the ligands [collagen types I and IV (COL1 and COL4)] and the receptors [syndecans 1 and 4 (SDC1 and SDC4), CD44, and integrins α2β1 and α3β1 (ITGA2_ITGB1, ITGA3_ITGB1)](Fig. 6 A,B). In conclusion, these findings suggest that Epi9_CSC primarily mediates specific ligand-receptor interactions via the collagen pathway, regulating cell adhesion, signal transduction, and microenvironment adaptation. This not only enhances cell adhesion properties and signal transmission but also plays a crucial role in tissue structural support, barrier function, and cell fate determination. Development and Validation of CSCRPI Using the FindAllMarkers function with the condition avg_log2FC > 2, a total of 631 CSC subpopulation marker genes were identified. These genes were then subjected to univariate Cox regression analysis, resulting in the identification of 182 CSC subpopulation marker genes significantly associated with prognosis (See Supplementary Table S13 ). Subsequently, four machine learning algorithms—Random Forest, Lasso, XGBoost, and Decision Tree—were employed to score and rank these 182 core prognostic genes (see Supplementary Materials Table S14 ). Through a weighted averaging method, the top 10 core prognostic genes were determined (Fig. 7 A). In the TCGA-BLCA training cohort, multivariable Cox regression analysis further identified five key prognostic genes: TUBB6, CERCAM, MAP7D3, SEL1L3, and HSD17B1. Based on the expression profiles and regression coefficients of these genes, the following CSCRPI formula was constructed:CSCRPI = − 0.1483420×TUBB6 + 0.2376760×CERCAM + 0.4180103× MAP7D3 − 0.1687422×SEL1L3 + 0.1378119×HSD17B1.BLCA patients were categorized into low CSCRPI and high CSCRPI groups based on the median CSCRPI value. In the TCGA-BLCA training cohort, high CSCRPI patients demonstrated significantly poorer prognosis compared to low CSCRPI patients (Fig. 7 B); this conclusion was consistently observed in the independent validation cohort GSE31684 (Fig. 7 C).Further analysis in the TCGA-BLCA training cohort led to the development of a nomogram model for predicting 1-year, 3-year, and 5-year overall survival (OS) based on CSCRPI, age, gender, and clinical stage through multivariable Cox and stepwise regression analyses (Fig. 7 D). Calibration curves demonstrated that the model's predictions for 1-year, 3-year, and 5-year survival were highly consistent with actual survival rates (Fig. 7 E). Decision curve analysis (DCA) indicated that the nomogram provided a greater net clinical benefit compared to other clinical variables (Fig. 7 F). In the TCGA-BLCA cohort, the area under the curve (AUC) values for the nomogram at 1-year, 3-year, and 5-year were 0.74, 0.71, and 0.72, respectively; in the independent validation cohort GSE31684, the AUC values were 0.77, 0.80, and 0.77, respectively (Fig. 7 G and 7 H). The results further revealed that, except for SEL1L3, the other four key prognostic genes were highly expressed in the high CSCRPI group, suggesting they might act as risk factors, while SEL1L3 might serve as a protective factor. This trend was further validated in the independent cohort GSE31684. These findings indicate that CSCRPI not only effectively distinguishes high-risk from low-risk BLCA patient groups but also serves as a reliable tool for predicting short-term and long-term survival, providing precise quantitative guidance for personalized clinical management. Clinical Relevance Assessment of CSCRPI To further validate the prognostic value of CSCRPI in subgroups with different clinical variables, we performed a stratified analysis. Patients were categorized into subgroups based on age, gender, and pathological stage, followed by survival analysis. The results revealed that CSCRPI levels were significantly elevated in deceased patients, high-risk patients, and those with stage IV disease, indicating a strong association between elevated CSCRPI levels and the progression of BLCA (Fig. 8 A-C). Furthermore, Kaplan-Meier survival analysis demonstrated that patients with higher CSCRPI levels had significantly worse prognoses across subgroups with varying clinical characteristics, whereas patients with lower CSCRPI levels exhibited better outcomes (Fig. 8 D-I). These findings strongly support the critical role of CSCRPI in predicting the prognosis of BLCA patients. Tumor Immune Microenvironment and Evaluation of Immunotherapy Efficacy To assess the immune cell infiltration in BLCA (bladder urothelial carcinoma) patients, we used nine different algorithms to calculate immune scores and stromal cell scores. According to the ESTIMATE algorithm, the high CSCRPI group exhibited increased stromal score, immune score, and ESTIMATE score, along with decreased tumor purity (Fig. 9 A). The two CSCRPI groups showed significant differences in immune infiltration, with the high CSCRPI group exhibiting significantly higher overall immune infiltration than the low CSCRPI group (Fig. 9 B). Additionally, we found that five characteristic prognostic genes in CSCRPI were highly correlated with tumor-infiltrating immune cells, including TUBB6, CERCAM, and MAP7D3, which were positively correlated with most tumor-infiltrating immune cells, while SEL1L3 and HSD17B1 were generally negatively correlated with tumor-infiltrating immune cells (Fig. 9 C). Immune checkpoint inhibitors (ICIs) have brought benefits to patients in clinical immunotherapy for various malignancies. TIDE (Tumor Immune Dysfunction and Exclusion) analysis showed that BLCA patients with high CSCRPI had increased TIDE scores and T-cell rejection scores, while there was no difference in microsatellite instability (MSI) score and T-cell dysfunction score between the two groups, indicating that BLCA patients with high CSCRPI are more likely to undergo immune escape, and thus the potential efficacy of ICI therapy may be poor (Fig. 9 D). Similarly, the IPS (Immunogenicity Prediction Score) results showed that BLCA patients with high CSCRPI would benefit less from immune checkpoint inhibitor therapy (Fig. 9 E). Using the ssGSEA algorithm, we obtained the scores for immune-related pathways. The high CSCRPI group showed enhanced activity in many immune-related pathways (Fig. 9 F), suggesting an overactivation of the immune response state or enhanced tumor-driven immune escape mechanisms in the high CSCRPI group. The study also observed that, except for the HSD17B1 gene, the remaining four characteristic prognostic genes in CSCRPI were positively correlated with most immune checkpoints (Fig. 9 G). The expression of these immune checkpoint genes was significantly higher in the high CSCRPI group (Fig. 9 H). Drug Sensitivity Analysis of High and Low CSCRPI Groups To further explore the clinical significance of CSC characteristic prognostic genes in BLCA for precision therapy, we comprehensively evaluated the sensitivity of patients in both CSCRPI groups to target drugs (Daporinad, Vinblastine, Vinorelbine, Eg5_9814, Paclitaxel, Sepantronium bromide, Staurosporine, Dinaciclib) using the GDSC database combined with the results from our previous cancer stem cell subgroup drug sensitivity analysis. Based on IC50 results, we found that patients in both CSCRPI groups were sensitive to these drugs, which was consistent with our previous results. Moreover, for most of these drugs, BLCA patients in the low CSCRPI group showed greater sensitivity than those in the high CSCRPI group, indicating that the CSCRPI score could serve as a guidance for chemotherapy in BLCA patients (Fig. 10 A). To better explore the association between the expression profiles of the 5 characteristic prognostic genes and drug sensitivity, we performed analysis using the CellMiner database. We generated correlation histograms for selected genes with selected target drugs. Positive correlation indicates that increased gene expression is associated with increased drug sensitivity, while negative correlation indicates that increased gene expression confers drug resistance. Notably, upregulation of TUBB6, CERCAM, and HSD17B1 enhanced resistance to some drugs, while upregulation of MAP7D3 enhanced sensitivity to Staurosporine (Fig. 10 B). Discussion Epithelial cells play a crucial role in tumor initiation and progression. A comprehensive and systematic understanding of the characteristics and functions of epithelial cells within the tumor microenvironment will help reveal the molecular mechanisms underlying tumor progression and drive the development of novel therapeutic strategies. scRNA-seq can uncover the compositional features, heterogeneity, dynamic changes, and key roles of tumor-associated epithelial cells in tumor initiation and progression, providing important insights for a deeper understanding of the tumor microenvironment. Epi9_CSC Promotes High Recurrence and Invasion in MIBC, with Paracancerous Presence Suggesting Its Origin from Normal Tissue or Its Role in Metastasis The results of this study indicate the presence of various epithelial cell subtypes in BLCA, with distinct distributions of these subtypes in the MIBC, NMIBC, and Para groups. Additionally, we identified a critical subpopulation of CSCs.CSC characteristics include self-renewal, differentiation potential, and the ability to promote tumor recurrence and metastasis. Self-renewal enables CSCs to maintain their population and drive tumor growth through symmetric or asymmetric division, while their differentiation potential allows them to generate other types of tumor cells, sustaining the tumor's high heterogeneity. Furthermore, CSCs possess strong migratory and invasive abilities, enabling them to induce metastatic tumor formation in new microenvironments, thereby worsening disease progression and complicating treatment. Our findings show that there are different proportions of CSCs in tumor and Paracancerous tissues, with the highest proportion of CSCs observed in MIBC. This feature may be a key factor contributing to the high recurrence and incidence rates of MIBC in clinical practice. Recent studies have further revealed the critical role of CSCs in the initiation and progression of MIBC.Moreover, we observed the presence of CSCs in Paracancerous tissues, suggesting that CSCs may originate from normal bladder tissue, or that CSCs play a pivotal role in driving cancer metastasis. Epi9_CSC Promotes Tumor Cell Survival, Invasion, and Self-Renewal through ECM Activation of Downstream Pathways KEGG enrichment analysis of Epi9_CSC reveals that pathways such as focal adhesion, ECM-receptor interaction, proteoglycans in cancer, and the PI3K-Akt signaling pathway play complex, synergistic roles in the initiation and progression of BLCA by CSCs.The ECM consists of collagen, laminin, fibronectin, and proteoglycans, which provide structural integrity and strength to tissues. Collagen contributes to tissue structure and strength, laminin affects cell adhesion and differentiation, and fibronectin regulates cell migration. The ECM serves as an important bridge between cells and their external environment, influencing cell adhesion, migration, and signal transduction through interactions with cell surface receptors such as integrins and CD44. In the ECM-receptor interaction pathway map from KEGG Mapper, our results show that ECM components, including collagen, laminin, and fibronectin, are highly expressed. These ECM components activate several downstream pathways (focal adhesion, proteoglycans in cancer, and the PI3K-Akt signaling pathway) by binding to integrin receptors or CD44 on the cell membrane.Focal adhesion, as a critical point of cell-ECM connection, plays an important role in this process. Through binding to ECM molecules, focal adhesion can activate the regulation of the actin cytoskeleton, NF-kB, and PI3K-Akt signaling pathways. Notably, the PI3K-Akt pathway controls cell proliferation, survival, and metabolism, and its abnormal activation can drive tumor progression. Our results show that ECM, through integrin receptor binding at focal adhesions, activates the NF-kB and PI3K-Akt signaling pathways, enhancing tumor cell proliferation and survival. Additionally, ECM, through integrin receptor binding at focal adhesions, can also activate the regulation of the actin cytoskeleton, promoting cytoskeletal reorganization and enhancing tumor cell motility and migration.Furthermore, our results suggest that ECM may activate the MAPK signaling pathway by binding to CD44 receptors on the cell membrane, aiding in the self-renewal of CSCs.On the other hand, intercellular communication results for Epi9_CSC indicate that Epi9_CSC primarily mediates communication with itself and other epithelial cells through collagen ligands in the collagen pathway, binding to integrin receptors, which is consistent with our KEGG findings. In summary, during the EMT process, Epi9_CSC secretes collagen that binds to integrins and CD44 receptors on the cell membranes of both Epi9_CSC itself and other epithelial cells, thereby activating downstream pathways including focal adhesion, PI3K-Akt signaling, regulation of the actin cytoskeleton, NF-kB, proteoglycans in cancer, and MAPK. The activation of these signaling pathways collectively enhances tumor (stem) cell proliferation and survival, increases their motility and invasiveness, and promotes their self-renewal. Molecular Mechanisms of Epi9_CSC Differentiation into Mesenchymal Subtypes During the differentiation of Epi9_CSC into mesenchymal subtypes, genes that are downregulated include NT5E (CD73), SERPINA1, TGFBI, LAMC2, DCBLD2, CMTM3, and COL7A1.NT5E (encoding CD73) is an important gene involved in immune regulation within the tumor microenvironment by converting extracellular AMP into adenosine. Adenosine plays an inhibitory role in immune escape by reducing the activity of T cells and NK cells, thereby promoting tumor immune evasion. Downregulation of CD73 during the differentiation of CSCs into mesenchymal subtypes may reduce adenosine production, weakening the immunosuppressive effect, and reflecting the transition from a stem cell state to an invasive mesenchymal state.SERPINA1 (encoding α1-antitrypsin) is a serine protease inhibitor involved in inflammation and the stability of the ECM. Downregulation of SERPINA1 may reduce protease inhibition, increasing ECM degradation and promoting CSC migration and invasion, which is consistent with the high invasiveness of mesenchymal subtypes.TGFBI (TGF-β induced gene) encodes an ECM protein closely related to the TGF-β signaling pathway, a major inducer of EMT. Downregulation of TGFBI may affect TGF-β-mediated EMT, indicating dynamic changes in the regulation of the TGF-β signaling pathway during CSC differentiation into mesenchymal subtypes. LAMC2 encodes the γ2 chain of laminin, which is a component of laminin-332 involved in cell adhesion to the basement membrane. Downregulation of LAMC2 may disrupt cell adhesion to the ECM, promoting cell detachment and migration, a feature particularly evident in mesenchymal subtype CSCs. Huang et al. (2020) noted that downregulation of LAMC2 (γ2 chain) promotes EMT in pancreatic ductal adenocarcinoma (PDAC) cells, further enhancing tumor invasion and metastasis.DCBLD2 (Discoidin, CUB, and LCCL domain-containing protein 2) is closely associated with its activation in EMT, supporting the invasiveness and metastatic potential of CSCs.CMTM3 (CKLF-like MARVEL transmembrane domain-containing 3) is involved in cell membrane structure and function and may affect cell polarity or signaling. Downregulation of CMTM3 may result in the loss of cell polarity, which is a hallmark of EMT and mesenchymal transformation. Yuan et al. (2016) reported that downregulation of CMTM3 in gastric cancer is associated with tumor invasion and metastasis.COL7A1 encodes type VII collagen, an essential component of the basement membrane that maintains epithelial tissue integrity. Downregulation of COL7A1 may disrupt the integrity of the basement membrane, promoting cell invasion and migration, which is consistent with the high invasiveness of mesenchymal subtype CSCs. Guerra et al. (2017) found that downregulation of COL7A1 leads to basement membrane integrity disruption, providing a less resistant barrier for tumor cells to infiltrate, thus promoting tumor expansion. During the differentiation of Epi9_CSC into mesenchymal subtypes, the expression of COL1A2, COL6A2, and COL3A1 is upregulated. These genes encode different types of collagen and are major components of the ECM. Their upregulation suggests that the cells are remodeling their extracellular environment, increasing the rigidity and structure of the matrix to support cell migration and invasion. This matrix remodeling is often associated with tumor fibrosis and sclerosis, which aids in tumor progression.By analyzing these gene expression changes in detail, we can gain a deeper understanding of the molecular mechanisms involved in the transition of BLCA stem cells to a more invasive phenotype, providing new perspectives for developing more effective therapeutic strategies. Dinaciclib as Potential Sensitive Therapeutic Agents for Epi9_CSC The drug sensitivity analysis results indicate that Epi9_CSC exhibits significant sensitivity to several chemotherapeutic agents, including Bortezomib, Dactinomycin, Docetaxel, Daporinad, Sepantronium bromide, Vinblastine, Eg5_9814, Vinorelbine, Staurosporine, Dinaciclib, and Paclitaxel.In the expression analysis of potential drug target genes in Epi9_CSC, we observed high expression levels of members of the cyclin-dependent kinase (CDK) family (CDK5, CDK9) within CSCs. These CDKs may play crucial roles in regulating cell cycle processes, influencing tumor cell proliferation and survival, and participating in cell signaling and gene expression modulation mechanisms that contribute to the self-renewal and drug resistance of CSCs [ 40 ]. Dinaciclib, a small-molecule inhibitor, exerts its antitumor effects predominantly by targeting CDKs, particularly CDK1, CDK2, CDK5, and CDK9. By inhibiting CDK9, Dinaciclib affects the functionality of the transcription elongation factor P-TEFb, leading to a reduction in the phosphorylation of RNA polymerase II. This ultimately decreases mRNA synthesis and inhibits the expression of short-lived anti-apoptotic proteins such as MCL-1, thereby resulting in tumor cell apoptosis and cell cycle arrest [ 41 ].In a study by Masahiro Shimizu et al., it was found that the oncogenic RAS signaling pathway enhances the activity of CDK1, promoting the generation of CSCs. This process involves an increase in the expression and activity of CSC-related factors, such as SOX2. However, Dinaciclib, as a CDK1 inhibitor, effectively disrupts this mechanism. By inhibiting CDK1 activity, Dinaciclib impairs the proliferation and self-renewal capabilities of CSCs, leading to a reduction in their population [ 42 ]. This mechanism underscores the significant role of CDKs in the formation and maintenance of CSCs. Given our drug sensitivity analysis, the sensitivity of Epi9_CSC to Dinaciclib suggests that CDK inhibitors may suppress the self-renewal of BLCA CSCs by targeting the highly expressed CDK family members (such as CDK5 and CDK9) within these cells. Therefore, we speculate that Dinaciclib represents a potential therapeutic agent for targeting BLCA CSC. Differential Marker Genes in BLCA Tissues Under Different Pathological Conditions In the comparative analysis of MIBC and NMIBC, upregulated genes in Urothelial cells include immune-related genes (such as LYZ, LEAP2, IGKC, HLA-DRB5, DEFB1, IFI27, IFI6, and IFITM3), indicating enhanced innate immunity, antimicrobial defense, active adaptive immune responses, and an enhanced antiviral response. Additionally, metabolism-related genes (such as ECH1, INSIG1, and MRPS12) reflect tumor cell metabolic reprogramming, supporting rapid proliferation through altered fatty acid oxidation and cholesterol metabolism [ 43 ]. Genes involved in cell structure and function (such as PHGR1, LGALS4, BMP2, and TMEM19) promote cell adhesion, signal transduction, and tissue development, potentially enhancing the invasiveness and migratory capacity of tumor cells [ 44 ]. In contrast, downregulated energy metabolism-related genes (such as ATP5MJ, ATP5MC2, ATP5ME, ATP5F1E, and NDUFAF8) suggest a reliance on glycolysis due to the inhibition of mitochondrial oxidative phosphorylation, which aids cancer cell survival under hypoxic conditions [ 45 ]. Downregulation of chromatin-related genes (such as H2AZ1, H3-3A, H3-3B, H4C3, and H1-2) may reduce the open chromatin state, affecting gene expression and promoting genetic instability [ 46 ]. Furthermore, the downregulation of genes related to protein degradation (e.g., RACK1 and ELOC) and RNA regulation (e.g., SNHG29 and DANCR) may disrupt critical signaling and regulatory networks, enhancing tumor invasiveness and uncontrolled growth [ 47 , 48 ]. When comparing MIBC with the Para group, the upregulation of immune-related genes (such as LYZ, IFI27, and LEAP2) suggests a strong immune response, alongside enhanced antimicrobial and antiviral defenses. Additionally, the upregulation of metabolism-related genes (SULT1E1, SCD, and DHCR24) indicates metabolic reprogramming to support rapid tumor growth [ 49 , 50 ]. These changes highlight the adaptive strategies of MIBC cells to evade immune surveillance, adjust metabolism, and promote invasiveness. On the other hand, the Para group shows no such upregulation, maintaining strong cell adhesion, cytoskeletal stability, and immune responses, which help limit tumor progression. The downregulation of adhesion-related genes (ITGA2, ITGA6, LMO7) and immune evasion-related genes (CCL20 and PTGS2) in MIBC further promotes invasiveness and metastasis [ 51 , 52 ]. The expression of signaling pathway-related genes (e.g., DKK1 and AREG) supports the higher invasiveness in MIBC, while the Para group shows lower malignancy traits, reinforcing its protective role against tumor development. In the comparison between NMIBC and Para, upregulated genes like S100A9, PSORS1C2, and IFI27 indicate alterations in tumor metabolism and microenvironment, promoting tumor initiation and progression [ 53 ]. The upregulation of ATP5F1C, ATP6V0C, and ATP5MC1 suggests active oxidative phosphorylation, fueling tumor cell proliferation [ 54 ]. Conversely, downregulation of genes such as THBS1 and HAS3 in NMIBC impairs extracellular matrix remodeling and cell migration, while the downregulation of DKK1 may enhance Wnt signaling, promoting cell proliferation and tumor progression [ 55 – 57 ]. These molecular alterations in NMIBC highlight significant changes in metabolism, extracellular matrix remodeling, and immune regulation, offering insights into potential therapeutic targets. Finally, the differential analysis of Epi9_CSC in MIBC and NMIBC reveals distinct gene expression patterns. MIBC tumor stem cells show upregulation of genes like TFPI2, TGFBI, and COL8A1, which are involved in extracellular matrix remodeling and intercellular interactions, likely contributing to their enhanced migratory and invasive capabilities [ 58 , 59 ]. The upregulation of SPOCK1 further supports matrix remodeling, suggesting that MIBC stem cells exhibit stronger invasiveness and migration ability compared to NMIBC. These findings underscore the critical role of extracellular matrix dynamics and cellular interactions in MIBC progression, enabling tumor cells to breach the basement membrane and invade surrounding tissues. In conclusion, MIBC demonstrates enhanced invasiveness, immune evasion, and metabolic reprogramming compared to NMIBC and Para group. These adaptive features involve immune enhancement, metabolic reprogramming, and extracellular matrix remodeling. NMIBC, in contrast, exhibits metabolic activity and microenvironment regulation with relatively weaker invasiveness, while the Para group maintains strong cell adhesion and immune responses, limiting malignant progression. Additionally, MIBC tumor stem cells exhibit enhanced invasiveness and migration, likely due to extracellular matrix remodeling and intercellular interactions. Clinical Translation Value and Multidimensional Functional Analysis of the CSCRPI Prognostic Risk Assessment Model Based on the identified molecular characteristics of the Epi9_CSC subpopulation, we successfully established the CSCRPI prognostic risk assessment model. This model demonstrates good correlation with clinicopathological parameters and exhibits excellent performance in predicting patient survival prognosis. More importantly, we conducted systematic analysis of patients in high and low CSCRPI risk groups, including clinical feature correlation assessment, tumor immune microenvironment status analysis, immunotherapy response potential evaluation, and drug sensitivity profile analysis, establishing a solid theoretical foundation for comprehensively elucidating the biological behavioral patterns of the Epi9_CSC subpopulation and its clinical translation value. Specifically, among the 5 characteristic genes constituting CSCRPI, CERCAM, MAP7D3, and HSD17B1 have positive coefficients, indicating that high expression of these genes leads to higher CSCRPI scores, resulting in poor prognosis in the high CSCRPI group, which is consistent with their mechanisms of action in cancer. CERCAM is a cell adhesion molecule primarily localized to the cell membrane, playing a role in intercellular interactions and cell-matrix binding. Research has found that CERCAM is abnormally highly expressed in BLCA tissues and cells, and its overexpression can significantly promote BLCA cell proliferation, DNA synthesis, and invasive capacity, and induce EMT, manifesting as upregulation of PCNA, vimentin, Twist, and N-cadherin, while inhibiting E-cadherin and cleaved caspase-3 expression; in vivo, CERCAM silencing can inhibit xenograft tumor growth in nude mice[ 60 ]. MAP7D3 is a microtubule-associated protein involved in microtubule assembly and stability regulation. Studies have shown that MAP7D3 enhances cancer stem cell characteristics, increases TNBC cell resistance to chemotherapy, and thereby promotes tumor metastasis and progression. Specifically, high expression of MAP7D3 promotes cell attachment and detachment from the matrix by upregulating extracellular matrix remodeling-related proteins such as integrin α6, ABCG2, and ALDH1A1, thereby increasing cell migration and invasion capacity in the matrix[ 61 ]. HSD17B1 is an important steroid metabolic enzyme that plays a key role in various cancers. Research shows it regulates active estrogen biosynthesis and promotes breast cancer cell proliferation[ 62 ]. Although the molecular mechanisms of HSD17B1 in BLCA pathogenesis require further elucidation, loss-of-function experiments have shown that knocking down HSD17B1 can significantly reduce migration and invasion activity of BLCA cells[ 63 ]. Therefore, all three genes above have effects that lead to poor BLCA prognosis. Particularly noteworthy is that two of these genes are related to EMT, which is consistent with our previous pathogenic mechanism analysis results for the Epi9_CSC subpopulation. Among the other two genes, TUBB6 is an important component of the cytoskeleton that forms microtubule structures by binding with α-tubulin, participating in maintaining cell morphology and cell division processes in normal cells. In bladder urothelial carcinoma, TUBB6 expression is abnormally upregulated, promoting tumor cell motility and invasive capacity by remodeling cytoskeletal structure. Additionally, TUBB6 may also promote local tumor invasion and distant metastasis by regulating the distribution and function of cell adhesion molecules, weakening intercellular connections. Studies have shown that inhibiting TUBB6 expression can significantly reduce migration and invasion activity of BLCA cells, suggesting its value as a potential therapeutic target[ 64 ]. SEL1L3 is an endoplasmic reticulum-localized transmembrane protein that mainly participates in endoplasmic reticulum-associated protein degradation pathways, playing an important role in maintaining intracellular protein homeostasis and regulating stress responses. However, current research on SEL1L3 is relatively limited, and its specific molecular mechanisms in tumor development and progression remain unclear. Based on its negative coefficient characteristics and high expression in the low-risk group, we speculate that SEL1L3 may function as a tumor suppressor with protective effects. Additionally, TUBB6 shows a negative coefficient in the prognostic model, which contradicts its biological mechanism of promoting cell motility and invasion metastasis. This may stem from disease stage-dependent effects or the influence of complex interactions between multiple genes. In the tumor immune microenvironment and immunotherapy efficacy assessment analysis, this study revealed the complex characteristics of the immune microenvironment in high-risk CSCRPI BLCA patients through multidimensional immune analysis. ESTIMATE scores showed that high-risk patients had significantly elevated stromal scores, immune scores, and composite scores, accompanied by reduced tumor purity. This phenomenon initially seemed contradictory to poor prognosis. However, in-depth immune cell subpopulation analysis provided a reasonable explanation for this paradox: the high-risk group showed significantly elevated scores across 24 immune cell types, including anti-tumor effector cells such as activated B cells, effector memory T cells, and NK cells, as well as immunosuppressive cells such as regulatory T cells, MDSCs, and immature dendritic cells. This "pan-lineage" increase in immune cell infiltration reflects a highly activated state of the tumor immune microenvironment, but the simultaneous increase in effector and suppressor cells suggests profound immune dysfunction. TIDE and IPS analyses further elucidated the molecular basis of poor immunotherapy responsiveness in high-risk patients. The significant increase in TIDE scores was mainly driven by T cell exclusion scores rather than T cell dysfunction, suggesting that high-risk tumors primarily achieve immune escape by preventing effector T cells from approaching tumors rather than inducing T cell exhaustion. Correspondingly, MSI scores showed no significant difference between groups, indicating that immune therapy resistance in high-risk patients does not stem from insufficient tumor mutational burden[ 65 ]. IPS analysis results were highly consistent with TIDE predictions, showing that high-risk patients derive lower benefit from immune checkpoint inhibitor therapy, collectively confirming that T cell exclusion-type immune escape is a key mechanism leading to immunotherapy failure. At the molecular level, the high-risk group showed systemic activation of immunosuppressive signaling pathways, with significantly enhanced activity of HLA, CCR, T cell co-inhibitory, and immune checkpoint pathways, while T cell co-stimulatory pathway activity showed no significant change, forming an "inhibitory signal-dominant" imbalanced immune regulation pattern[ 66 ]. More importantly, 34 key immune checkpoint genes were generally upregulated in the high-risk group, including the classical PD-1/PD-L1 axis (PDCD1, CD274, PDCD1LG2), CTLA-4, TIM-3 (HAVCR2), and multiple TNF/TNFR family members, constructing a multi-layered, redundant immunosuppressive network. This "checkpoint storm" phenomenon may be an important reason for the limited efficacy of single-target immunotherapy[ 67 , 68 ]. The association between the 5 characteristic genes in CSCRPI and the immune microenvironment presents interesting differentiation patterns. TUBB6, CERCAM, and MAP7D3 are positively correlated with most tumor-infiltrating immune cells and positively correlated with immune checkpoint expression, suggesting these genes may activate immunosuppressive mechanisms while promoting immune cell recruitment, forming a "recruitment-suppression" vicious cycle. Conversely, SEL1L3 and HSD17B1 are negatively correlated with immune cell infiltration and may participate in immune regulation through different mechanisms. Particularly noteworthy is that TUBB6 shows a negative coefficient in the prognostic model, which presents an apparent contradiction with its known function of promoting cell motility and invasion metastasis. This phenomenon may reflect the functional transition of TUBB6 at different disease stages or composite effects under multi-gene network interactions, requiring further mechanistic studies for clarification. Our research results have important guiding significance for precision immunotherapy in BLCA. Although high CSCRPI patients have abundant immune cell infiltration, they present a typical "immune desert" functional state, suggesting that traditional single immune checkpoint inhibitor therapy may have limited efficacy[ 69 ]. Based on the T cell exclusion-dominant immune escape mechanism, these patients may be more suitable for combination therapy strategies: immune checkpoint inhibitors combined with anti-angiogenic therapy to promote T cell infiltration through vascular normalization[ 70 , 71 ]; CAR-T or TCR-T cell therapy to directly deliver effector cells bypassing T cell exclusion mechanisms[ 72 ]. Our drug sensitivity analysis results indicate that CSCRPI scores have important clinical value in predicting chemotherapy responses in BLCA patients. Patients in the low CSCRPI group showed higher sensitivity to most targeted drugs, which is consistent with previous research results showing that tumor stem cell characteristics are closely related to chemotherapy resistance. Previous studies have confirmed that enhancement of tumor stem cell-like characteristics is usually accompanied by increased multidrug resistance, primarily achieved through activating DNA repair mechanisms, upregulating drug efflux pumps, and inhibiting cell apoptosis[ 73 ]. The association between our identified characteristic genes and drug sensitivity has important significance. TUBB6, as a member of the tubulin family, has been confirmed to be associated with Paclitaxel resistance through its overexpression, possibly affecting drug efficacy by altering microtubule dynamics[ 74 ]. HSD17B1, as a key enzyme in steroid hormone metabolism, is abnormally expressed in various cancers. Research shows that its expression level is closely related to tumor cell proliferation and survival capacity, possibly affecting chemotherapy sensitivity by regulating cell cycle and apoptosis processes[ 75 ]. In summary, the CSCRPI scoring system constructed based on the Epi9_CSC subpopulation integrates expression information from 5 key characteristic genes and can not only accurately predict prognostic risk in BLCA patients but, more importantly, provides multidimensional guidance for clinical treatment decisions: in immunotherapy, this system reveals T cell exclusion-dominant immune escape mechanisms in high-risk patients, providing theoretical basis for formulating combination therapy strategies; in chemotherapy selection, CSCRPI scores can effectively predict patient drug sensitivity, helping achieve precise formulation of individualized chemotherapy regimens. The establishment of this multifunctional prognostic assessment tool lays a solid foundation for precision medicine practice in BLCA. Conclusions Overall, we identified 11 subtypes of Urothelial cells, each exhibiting varying degrees of copy number variation. Among these Urothelial cells, MIBC demonstrates greater invasiveness, immune evasion, and metabolic adaptability compared to NMIBC and Para groups. These features are associated with immune enhancement, metabolic reprogramming, and extracellular matrix remodeling. NMIBC is metabolically active and significantly regulates the microenvironment, but it has weaker invasiveness. In contrast, the Para group limits malignant progression by enhancing cell adhesion and immune response. We also identified a crucial population of cancer stem cells (Epi9_CSC), which secrete collagen during the EMT process. This collagen interacts with integrins and CD44 receptors on the cell membranes of Epi9_CSCs and other epithelial cells, activating downstream pathways such as Focal adhesion, PI3K-Akt signaling pathway, Regulation of actin cytoskeleton, NF-kB, Proteoglycans in cancer, and MAPK. The activation of these signaling pathways collectively enhances tumor (stem) cell proliferation and survival, boosts their motility and invasiveness, and promotes self-renewal (Fig. 11 ). Compared to NMIBC, MIBC tumor stem cells exhibit stronger invasiveness and migration abilities. Based on the expression profiles of potential sensitive drug targets for Epi9_CSC, Dinaciclib are identified as promising therapeutic agents for BLCA CSC. The CSCRPI scoring system, based on the Epi9_CSC subpopulation and integrating expression information from 5 key characteristic genes, not only accurately predicts prognostic risk in BLCA patients but also provides comprehensive guidance value for clinical treatment decision-making. Abbreviations GEO Gene Expression Omnibus DEGs Differentially Expressed Genes scRNA-seq Single-cell RNA Sequencing RNA-seq RNA Sequencing KEGG Kyoto Encyclopedia of Genes and Genomes CSC Cancer Stem Cells Epi Epithelial Cells CNV Copy Number Variation ECM Extracellular Matrix TME Tumor Microenvironment EMT Epithelial-Mesenchymal Transition CSCRPI Cancer Stem Cell-Related Prognostic Index Declarations Acknowledgements I sincerely thank my advisors for their guidance and support. I am also grateful to my team members for their invaluable assistance. Your contributions have been essential to this project. Authors’ contributions Supervision, funding acquisition: Xianliang Hou, Songbai Liao. Data collection: Biao Zhang, Yi Liu. Fei Yang. Formal analysis: Biao Zhang, Man Yang, Yu Pan. Methodology: Chunhong Li and Chune Mo. Writing–original draft: Biao Zhang. Writing–review and editing: Xianliang Hou,Yingpin Lei, and Jiahua Hu. All authors read and approved the final manuscript. Funding This work was supported by grants from the Guangxi Natural Science Foundation (2024GXNSFAA010096), Guangdong Basic and Applied Basic Research Foundation (2025A1515012661), Guilin Science Research and Technology Development Project (20220139-13-2), Guangdong Province Medical Science and Technology Research Fundation (B2025207), Guangxi Medical and Health Appropriate Technology Development and Promotion Project (S2024075), Innovation Training Program for College Students (202410601013, S202410601182), Guangxi Key Laboratory of Tumor Immunology and Microenvironmental Regulation (2023KF006, 3030302213). Availability of data and materials scRNA-seq datasets for BLCA were obtained from GEO (GSE192575, GSE145137, GSE135337, GSE222315, GSE129845). Drug sensitivity data came from the GDSC2 database (cancerrxgene.org). Copy number variation analysis used the hg38_gencode_v27 annotation file from the Broad Institute (broadinstitute.org). These datasets are accessible following their respective usage policies. Ethics approval and consent to participate The patient sample data used in this study were approved by the Ethics Committees of all participating hospitals, including SYSMH, GDPPH, STCH, ZJH, SMUTH, SYUTH, and NFH. Written informed consent was obtained from all participants at the time of initial sample collection for analyses including scRNA-seq. The bioinformatics analysis conducted in this study utilized only de-identified secondary data analysis and did not involve new human sample collection or direct patient interaction. Therefore, no additional ethical approval was required for this analysis. We are committed to adhering to data use agreements and relevant laws and regulations to ensure the privacy of participants is protected. All analyses were conducted in accordance with the Declaration of Helsinki and the Terms of Use of each public database. Competing interests The authors declare no competing interests. References Burger M, Catto JW, Dalbagni G, Grossman HB, Herr H, Karakiewicz P, Kassouf W, Kiemeney LA, La VC, Shariat S, Lotan Y. Epidemiology and risk factors of urothelial BLCA. Eur Urol. 2013;63(2):234–41. Ahmadi H, Duddalwar V, Daneshmand S. Diagnosis and Staging of BLCA. Hematol Oncol Clin NorthAm. 2021;35(3):531–41. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024 May-Jun;74(3):229–63. Huang Q, Zi H, Luo L, Li X, Zhu C, Zeng X. Secular trends of morbidity and mortality of prostate, bladder, and kidney cancers in China, 1990 to 2019 and their predictions to 2030. BMC Cancer. 2022;22(1):1164. Richters A, Aben KKH, Kiemeney LALM. The global burden of urinary BLCA: an update. World JUrol. 2020;38(8):1895–904. Chan ES, Yip SK, Hou SM, Cheung HY, Lee WM, Ng CF. Age, tumour stage, and preoperative serum albumin level are independent predictors of mortality after radical cystectomy for treatment of BLCA in HongKong Chinese. Hong Kong Med J. 2013;19(5):400–6. Lee YC, Lam HM, Rosser C, Theodorescu D, Parks WC, Chan KS. The dynamic roles of the bladder tumourmicroenvironment. Nat Rev Urol. 2022;19(9):515–33. Arneth B. Tumor Microenvironment. Med (Kaunas). 2019;56(1):15. Marusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity: a looking glass for cancer? Nat Rev Cancer. 2012;12(5):323–34. Dagogo-Jack I, Shaw AT. Tumour heterogeneity and resistance to cancer therapies. Nat Rev Clin Oncol. 2018;15(2):81–94. Lambert AW, Weinberg RA. Linking EMT programmes to normal and neoplastic epithelial stem cells. Nat Rev Cancer. 2021;21(5):325–38. Mantovani A, Marchesi F, Malesci A, Laghi L, Allavena P. Tumour-associated macrophages as treatment targets in oncology. Nat Reviews Clin Oncol. 2017;14(7):399–416. Nishikawa H, Sakaguchi S. Regulatory T cells in tumor immunity. Int J Cancer. 2010;127(4):759–67. Yoon C, Lu J, Jun Y, et al. KRAS activation in gastric cancer stem-like cells promotes tumor angiogenesis and metastasis [retracted in. BMC Cancer. 2024;24(1):392. Zeng Z, Fu M, Hu Y, Wei Y, Wei X, Luo M. Regulation and signaling pathways in cancer stem cells: implications for targeted therapy for cancer. Mol Cancer. 2023;22(1):172. Published 2023 Oct 18. Liu Z, Li H, Dang Q, Weng S, Duo M, Lv J, Han X. Integrative insights and clinical applications of single-cell sequencing in cancer immunotherapy. Cell Mol Life Sci. 2022;79(11):577. Tan TZ, Rouanne M, Tan KT, Huang RY, Thiery JP. Molecular Subtypes of Urothelial BLCA: Results from a Meta-cohort Analysis of 2411 Tumors. Eur Urol. 2019;75(3):423–32. Hamid ARAH, Syadza YZ, Yausep OE, Christanto RBI, Muharia BHR, Mochtar CA. The expression of stem cells markers and its effects on the propensity for recurrence and metastasis in BLCA: A systematic review. PLoS ONE. 2023;18(5):e0269214. Published 2023 May 17. Gibbons DL, Creighton CJ. Pan-cancer survey of epithelial-mesenchymal transition markers across the Cancer Genome Atlas. Dev Dyn. 2018;247(3):555–64. Tanabe S. Epithelial-Mesenchymal Transition and Cancer Stem Cells. Adv Exp Med Biol. 2022;1393:1–49. Batlle E, Clevers H. Cancer stem cells revisited. Nat Med. 2017;23(10):1124–34. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in development and disease. Cell. 2009;139(5):871–90. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, Brooks M, Reinhard F, Zhang CC, Shipitsin M, Campbell LL, Polyak K, Brisken C, Yang J, Weinberg RA. The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell. 2008;133(4):704–15. El-Tanani M, Rabbani SA, Satyam SM, Rangraze IR, Wali AF, El-Tanani Y, Aljabali AAA. Deciphering the Role of Cancer Stem Cells: Drivers of Tumor Evolution, Therapeutic Resistance, and Precision Medicine Strategies. Cancers (Basel). 2025;17(3):382. Shin K, Lim A, Odegaard JI, Honeycutt JD, Kawano S, Hsieh MH, Beachy PA. Cellular origin of bladder neoplasia and tissue dynamics of its progression to invasive carcinoma. Nat Cell Biol. 2014;16(5):469–78. Ohishi T, Koga F, Migita TBLCA. Stem-Like Cells: Their Origin and Therapeutic Perspectives. Int J Mol Sci. 2016;17:43. Lawson DA, Bhakta NR, Kessenbrock K, Prummel KD, Yu Y, Takai K et al. Single-cell analysis reveals a stem-cell program in human metastatic breast cancer cellsNature., 526 (7571) (2015), pp. 131–5. Hynes RO. The extracellular matrix: not just pretty fibrils. Science. 2009;326(5957):1216–9. Frantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J Cell Sci. 2010;123(Pt 24):4195–200. Vivanco I, Sawyers CL. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer. 2002;2(7):489–501. Gibieža P, Petrikaitė V. The regulation of actin dynamics during cell division and malignancy. Am J Cancer Res. 2021;11(9):4050–69. Gao ZW, Wang HP, Lin F, Wang X, Long M, Zhang HZ, Dong K. CD73 promotes proliferation and migration of human cervical cancer cells independent of its enzyme activity. BMC Cancer. 2017;17(1):135. Xiang S, Yang L, He Y, Ding F, Qiao S, Su Z, Chen Z, Lu A, Li F. Alpha-1 Antitrypsin as a Regulatory Protease Inhibitor Modulating Inflammation and Shaping the Tumor Microenvironment in Cancer. Cells. 2025;14(2):88. Huang H, Tang Q, Li S, Qin Y, Zhu G. TGFBI: A novel therapeutic target for cancer. Int Immunopharmacol. 2024;134:112180. Huang C, Chen J. Laminin–332 mediates proliferation, apoptosis, invasion, migration and epithelial–to–mesenchymal transition in pancreatic ductal adenocarcinoma. Mol Med Rep. 2021;23(1):11. Xie P, Liu JY, Yan H, Wang ZB, Jiang SL, Li X, Liu ZQ. Pan-cancer analyses identify DCBLD2 as an oncogenic, immunological, and prognostic biomarker. Front Pharmacol. 2022;13:950831. Yuan W, Liu B, Wang X, Li T, Xue H, Mo X, Yang S, Ding S, Han W. CMTM3 decreases EGFR expression and EGF-mediated tumorigenicity by promoting Rab5 activity in gastric cancer. Cancer Lett. 2017;386:77–86. Guerra L, Odorisio T, Zambruno G, Castiglia D. Stromal microenvironment in type VII collagen-deficient skin: The ground for squamous cell carcinoma development. Matrix Biol. 2017;63:1–10. Xu S, Xu H, Wang W, Li S, Li H, Li T, Zhang W, Yu X, Liu L. The role of collagen in cancer: from bench to bedside. J Transl Med. 2019;17(1):309. Malumbres M. Cyclin-dependent kinases. Genome Biol. 2014;15(6):122. Zhang M, Zhang L, Hei R, Li X, Cai H, Wu X, Zheng Q, Cai C. CDK inhibitors in cancer therapy, an overview of recent development. Am J Cancer Res. 2021;11(5):1913–35. Shimizu M, Shibuya H, Tanaka N. Enhanced O-GlcNAc modification induced by the RAS/MAPK/CDK1 pathway is required for SOX2 protein expression and generation of cancer stem cells. Sci Rep. 2022;12(1):2910. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–74. Li TT, Lai YW, Han X, Niu X, Zhang PX. BMP2 as a promising anticancer approach: functions and molecular mechanisms. Invest New Drugs. 2022;40(6):1322–32. Pavlova NN, Thompson CB. The Emerging Hallmarks of Cancer Metabolism. Cell Metab. 2016;23(1):27–47. Kiri S, Ryba T. Cancer, metastasis, and the epigenome. Mol Cancer. 2024;23(1):154. Ding F, Xiao H, Wang M, Xie X, Hu F. The role of the ubiquitin-proteasome pathway in cancer development and treatment. Front Biosci (Landmark Ed). 2014;19(6):886–95. Statello L, Guo CJ, Chen LL, Huarte M. Gene regulation by long non-coding RNAs and its biological functions. Nat Rev Mol Cell Biol. 2021;22(2):96–118. Galon J, Bruni D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat Rev Drug Discov. 2019;18(3):197–218. Han C, Hu C, Liu T, Sun Y, Hu F, He Y, Zhang J, Chen J, Ding J, Fan J, Zhang X, Wang J, Qiao X, Jiang D, Yang K, Yang S. IGF2BP3 enhances lipid metabolism in cervical cancer by upregulating the expression of SCD. Cell Death Dis. 2024;15(2):138. Cruz SP, Zhang Q, Devarajan R, Paia C, Luo B, Zhang K, Koivusalo S, Qin L, Xia J, Ahtikoski A, Vaarala M, Wenta T, Wei GH, Manninen A. Dampened Regulatory Circuitry of TEAD1/ITGA1/ITGA2 Promotes TGFβ1 Signaling to Orchestrate Prostate Cancer Progression. Adv Sci (Weinh). 2024;11(11):e2305547. Wang H, Shi H, He X, Liao A. Downregulation of Chemokine CCL20 Involved in Myeloma Cells Resistant to Elotuzumab and Lenalidomide. Onco Targets Ther. 2021;14:2789–95. Chen Y, Ouyang Y, Li Z, Wang X, Ma J. S100A8 and S100A9 in Cancer. Biochim Biophys Acta Rev Cancer. 2023;1878(3):188891. Galber C, Acosta MJ, Minervini G, Giorgio V. The role of mitochondrial ATP synthase in cancer. Biol Chem. 2020;401(11):1199–214. Li P, Wang K, Song J, Chen Z, Li Y, Chen Z. THBS1 knockdown suppresses pancreatic cancer progression through JAK2/STAT3 signaling pathway. Mol Cell Probes. 2025;79:102003. Chang IW, Liang PI, Li CC, Wu WJ, Huang CN, Lin VC, Hsu CT, He HL, Wu TF, Hung CH, Li CF. HAS3 underexpression as an indicator of poor prognosis in patients with urothelial carcinoma of the upper urinary tract and urinary bladder. Tumour Biol. 2015;36(7):5441–50. Jumpertz S, Hennes T, Asare Y, Schütz AK, Bernhagen J. CSN5/JAB1 suppresses the WNT inhibitor DKK1 in colorectal cancer cells. Cell Signal. 2017;34:38–46. Liang W, Chen J, Li H, Zhang P, Guan H, Li Y. High expression of COL8A1 predicts poor prognosis and promotes EMT in papillary thyroid cancer. Endocr Connect. 2024;13(12):e240279. Sun LR, Li SY, Guo QS, Zhou W, Zhang HM. SPOCK1 Involvement in Epithelial-to-Mesenchymal Transition: A New Target in Cancer Therapy? Cancer Manag Res. 2020;12:3561–9. Zuo Y, Xu X, Chen M, Qi L. The oncogenic role of the cerebral endothelial cell adhesion molecule (CERCAM) in bladder cancer cells in vitro and in vivo. Cancer Med. 2021;10(13):4437–50. Kuo WH, Chu PY, Wang CC, Huang PS, Chan SH. MAP7D3, a novel prognostic marker for triple-negative breast cancer, drives cell invasiveness and cancer-initiating cell properties to promote metastatic progression. Biol Direct. 2023;18(1):44. Kristanti AN, Aminah NS, Siswanto I, Manuhara YSW, Abdjan MI, Wardana AP, Aung EE, Takaya Y. Anticancer potential of β-sitosterol and oleanolic acid as through inhibition of human estrogenic 17beta-hydroxysteroid dehydrogenase type-1 based on an in silico approach. RSC Adv. 2022;12(31):20319–29. Zheng F, Wang Z, Li S, Xiong S, Yuan Y, Zeng J, Tan Y, Liu X, Xu S, Fu B. Development of a propionate metabolism-related gene-based molecular subtypes and scoring system for predicting prognosis in bladder cancer. Eur J Med Res. 2024;29(1):393. Kim B, Jung M, Moon KC, Han D, Kim K, Kim H, Yang S, Lee D, Jun H, Lee KM, Lee CH, Nikas IP, Yang S, Lee H, Ryu HS. Quantitative proteomics identifies TUBB6 as a biomarker of muscle-invasion and poor prognosis in bladder cancer. Int J Cancer. 2023;152(2):320–30. Jiang P, Gu S, Pan D, Fu J, Sahu A, Hu X, Li Z, Traugh N, Bu X, Li B, Liu J, Freeman GJ, Brown MA, Wucherpfennig KW, Liu XS. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018;24(10):1550–8. Chen L, Flies DB. Molecular mechanisms of T cell co-stimulation and co-inhibition. Nat Rev Immunol. 2013;13(4):227–42. Rui R, Zhou L, He S. Cancer immunotherapies: advances and bottlenecks. Front Immunol. 2023;14:1212476. Sharma P, Hu-Lieskovan S, Wargo JA, Ribas A. Primary, Adaptive, and Acquired Resistance to Cancer Immunotherapy. Cell. 2017;168(4):707–23. Spranger S, Gajewski TF. Impact of oncogenic pathways on evasion of antitumour immune responses. Nat Rev Cancer. 2018;18(3):139–47. Tian L, Goldstein A, Wang H, Ching Lo H, Sun Kim I, Welte T, Sheng K, Dobrolecki LE, Zhang X, Putluri N, Phung TL, Mani SA, Stossi F, Sreekumar A, Mancini MA, Decker WK, Zong C, Lewis MT, Zhang XH. Mutual regulation of tumour vessel normalization and immunostimulatory reprogramming. Nature. 2017;544(7649):250–4. Mpekris F, Baish JW, Stylianopoulos T, Jain RK. Role of vascular normalization in benefit from metronomic chemotherapy. Proc Natl Acad Sci U S A. 2017;114(8):1994–9. June CH, O'Connor RS, Kawalekar OU, Ghassemi S, Milone MC. CAR T cell immunotherapy for human cancer. Science. 2018;359(6382):1361–5. Phi LTH, Sari IN, Yang YG, Lee SH, Jun N, Kim KS, Lee YK, Kwon HY. Cancer Stem Cells (CSCs) in Drug Resistance and their Therapeutic Implications in Cancer Treatment. Stem Cells Int. 2018;2018:5416923. Gan PP, Pasquier E, Kavallaris M. Class III beta-tubulin mediates sensitivity to chemotherapeutic drugs in non small cell lung cancer. Cancer Res. 2007;67(19):9356–63. Aka JA, Mazumdar M, Chen CQ, Poirier D, Lin SX. 17beta-hydroxysteroid dehydrogenase type 1 stimulates breast cancer by dihydrotestosterone inactivation in addition to estradiol production. Mol Endocrinol. 2010;24(4):832–45. Additional Declarations No competing interests reported. Supplementary Files SupplementalMaterial.doc TableS1SampleGroupingInformation.csv TableS2CNVResultsofUrothelialCellSubgroups.csv TableS3CharacteristicGeneResultsofUrothelialCellSubgroups.csv TableS4KEGGAnalysisResultsofUrothelialCellSubgroups.csv TableS5PseudotimeanalysisofCSCsdifferentiationintomesenchymalsubtypesusingMonocle3.csv TableS6DrugsensitivityscoringresultsforCSCsusingoncoPredict..csv TableS7DifferentialanalysisresultsofurothelialcellsinMIBCNMIBCandPara.csv TableS8DifferentialanalysisresultsofEpi9CSCbetweentheMIBCandNMIBC.csv TableS9CellChatInteractionCountandWeight.csv TableS10CellChatMIBC.csv TableS11CellChatNMIBC.csv TableS12CellChatPara.csv TableS13182prognosticCSCmarkergenesbyunivariateCoxregression.csv TableS14Resultsoffourmachinelearningweightedaveragingmethods.csv TableS15CoregeneexpressionprofileandclinicaldataofCSCRPIpatients.csv TableS16CSCRPIESTIMATEresults.csv TableS17CSCRPIssgsearesults.csv TableS18CSCRPIGDSC2.csv TableS19CSCRPICellMiner.csv Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7536468","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":531774905,"identity":"1b5db63d-3ba5-4ced-b47d-ceb2e239e9cf","order_by":0,"name":"Biao Zhang","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Biao","middleName":"","lastName":"Zhang","suffix":""},{"id":531774907,"identity":"a284527e-e6a4-436b-8ce7-e1d130d98bea","order_by":1,"name":"Yi Liu","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Liu","suffix":""},{"id":531774909,"identity":"9d3d4773-8f45-441f-8eb1-043ac25c7575","order_by":2,"name":"Fei Yang","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Yang","suffix":""},{"id":531774910,"identity":"d6ac876b-a839-4145-851f-40330cc0e225","order_by":3,"name":"Man Yang","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Man","middleName":"","lastName":"Yang","suffix":""},{"id":531774911,"identity":"e30f5669-2938-47b1-993a-50ac4c830cac","order_by":4,"name":"Yu Pan","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Pan","suffix":""},{"id":531774912,"identity":"62d8978f-e994-429e-882b-615b729f8f23","order_by":5,"name":"Yingpin Lei","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yingpin","middleName":"","lastName":"Lei","suffix":""},{"id":531774913,"identity":"a649f741-7894-42b8-b96a-03f664f3b1b2","order_by":6,"name":"Chunhong Li","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chunhong","middleName":"","lastName":"Li","suffix":""},{"id":531774914,"identity":"a99b02c9-9109-490b-9e25-cfce812620f1","order_by":7,"name":"Jiahua Hu","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiahua","middleName":"","lastName":"Hu","suffix":""},{"id":531774915,"identity":"7145782e-f295-4e18-9e6a-6daf16cac1c1","order_by":8,"name":"Chune Mo","email":"","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chune","middleName":"","lastName":"Mo","suffix":""},{"id":531774919,"identity":"7eb350d2-c94f-4b45-80ab-a36c704900a4","order_by":9,"name":"Songbai Liao","email":"","orcid":"","institution":"the Nanxishan Hospital of Guangxi Zhuang Autonomous Region","correspondingAuthor":false,"prefix":"","firstName":"Songbai","middleName":"","lastName":"Liao","suffix":""},{"id":531774923,"identity":"283e28de-8bcc-4ef8-a97a-f77a28b82161","order_by":10,"name":"Xianliang Hou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtklEQVRIiWNgGAWjYBACAxDxwECinp+Z+eAD4rUkFFgkSLazJRuQoOVDRYLBeR4zAaK0mLP3mEkkGEjkGR9mMGNgqLGJJqjFsueMsQFQS7HZYYa0BwzH0nIbCDrsRo7hA6AWxm2HGY4bMDYcJkqLwQGQls3NjG0SxGoB25K4gZmZjUgtZ44Vg/xiLHGYjdkggSi/HG/eJvHhT50cf//5jw8+1NgQ1oIKEkhTPgpGwSgYBaMAFwAAdQA9IQ7yy9gAAAAASUVORK5CYII=","orcid":"","institution":"the Second Affiliated Hospital of Guilin Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xianliang","middleName":"","lastName":"Hou","suffix":""}],"badges":[],"createdAt":"2025-09-04 13:08:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7536468/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7536468/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94139210,"identity":"a91cea67-7c7c-4770-b911-9db42894add8","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":69110889,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.doc","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/ec3421dd03572a002814c95e.doc"},{"id":94139191,"identity":"12d82111-04ef-463a-a17d-5d20477f22d4","added_by":"auto","created_at":"2025-10-22 19:31:49","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11773,"visible":true,"origin":"","legend":"","description":"","filename":"e4ff9004cdab4da49851c56be9b4b782.json","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/200cb3fb3dd95af0e8ec5eef.json"},{"id":94140299,"identity":"b70769e4-a32c-4772-99c7-105fb2bd2c80","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"doc","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2790622,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalMaterial.doc","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/182f0cd1c7fb5bf0c9262437.doc"},{"id":94139193,"identity":"c38710b4-141f-40d3-b703-fdfc28fee262","added_by":"auto","created_at":"2025-10-22 19:31:49","extension":"csv","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":413195,"visible":true,"origin":"","legend":"","description":"","filename":"TableS10CellChatMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/2524bab8fe595b0cd7034c9b.csv"},{"id":94138047,"identity":"88096024-5247-4a6d-9c72-b1cc53ba43bb","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":191537,"visible":true,"origin":"","legend":"","description":"","filename":"TableS11CellChatNMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/badd7bfb18e0a7bad7d4122e.csv"},{"id":94138073,"identity":"7ead79fb-7bcc-4a9c-8ef9-a7c28fff20a4","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":142777,"visible":true,"origin":"","legend":"","description":"","filename":"TableS12CellChatPara.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/ff97198e71df29d4b9e6daf3.csv"},{"id":94139198,"identity":"53fffbd1-4466-4e19-845e-e03439155f28","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"csv","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19222,"visible":true,"origin":"","legend":"","description":"","filename":"TableS13182prognosticCSCmarkergenesbyunivariateCoxregression.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/78fefcaa2edcc6baad7f4370.csv"},{"id":94140303,"identity":"b5b9edb8-b19f-4e61-9580-9cb8e4a5e0c6","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"csv","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14696,"visible":true,"origin":"","legend":"","description":"","filename":"TableS14Resultsoffourmachinelearningweightedaveragingmethods.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/9bc094ec1f1b98b65d81e377.csv"},{"id":94138052,"identity":"2c21bac6-b498-46f2-a055-1d71bb4028ab","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65542,"visible":true,"origin":"","legend":"","description":"","filename":"TableS15CoregeneexpressionprofileandclinicaldataofCSCRPIpatients.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/9b1dea74aeefdc48aaf3c14b.csv"},{"id":94138064,"identity":"529553fb-ca85-4f99-85cf-dd10f8b4f6bb","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":33806,"visible":true,"origin":"","legend":"","description":"","filename":"TableS16CSCRPIESTIMATEresults.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/12990eabc1bfa9a413298f59.csv"},{"id":94138068,"identity":"7a50e471-c71a-4a4f-80b2-c0bca569557e","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":204118,"visible":true,"origin":"","legend":"","description":"","filename":"TableS17CSCRPIssgsearesults.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/eb351f91bbbe1b990eacef25.csv"},{"id":94139216,"identity":"c6196fe4-415f-4aa3-8cd8-7c1778ce99ba","added_by":"auto","created_at":"2025-10-22 19:31:51","extension":"csv","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1271561,"visible":true,"origin":"","legend":"","description":"","filename":"TableS18CSCRPIGDSC2.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/3eea02855262bfb6ddcd081b.csv"},{"id":94138069,"identity":"2da436ea-9289-4542-a7d9-bba624713539","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":37587,"visible":true,"origin":"","legend":"","description":"","filename":"TableS19CSCRPICellMiner.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/59318405db48a6e69e85c662.csv"},{"id":94138061,"identity":"a2ec38c6-9bff-447e-8f08-c07dfd45f96c","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10614,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1SampleGroupingInformation.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/1bd4c372a8424e112722ab90.csv"},{"id":94140811,"identity":"cfc47b3e-72c2-4ace-bb79-d37d707ab0fd","added_by":"auto","created_at":"2025-10-22 19:47:50","extension":"csv","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":597,"visible":true,"origin":"","legend":"","description":"","filename":"TableS2CNVResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/7a425a9cf6015f5082a3fdf7.csv"},{"id":94138070,"identity":"a9371c6b-22d8-4ac9-9201-f4b11c8f4d29","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1901668,"visible":true,"origin":"","legend":"","description":"","filename":"TableS3CharacteristicGeneResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/a0da76a0f35c4184dc89634e.csv"},{"id":94140304,"identity":"34a1735b-e91a-48f6-81ff-50c7e94cb03c","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"csv","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7867,"visible":true,"origin":"","legend":"","description":"","filename":"TableS4KEGGAnalysisResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/576ba5152c64bc22e10468e6.csv"},{"id":94138091,"identity":"035f8b02-46f4-485f-aa91-7921c6c7c4ed","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"csv","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2310045,"visible":true,"origin":"","legend":"","description":"","filename":"TableS5PseudotimeanalysisofCSCsdifferentiationintomesenchymalsubtypesusingMonocle3.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/596e9985d887e9eb594a2391.csv"},{"id":94138082,"identity":"c2aaaa42-e015-4a4d-8bb2-9376f31fae25","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"csv","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1285885,"visible":true,"origin":"","legend":"","description":"","filename":"TableS6DrugsensitivityscoringresultsforCSCsusingoncoPredict..csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/2c641857308ad396320fd81f.csv"},{"id":94138103,"identity":"6408ac49-9523-4f38-8ef2-3d88f89f6221","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"csv","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":174334,"visible":true,"origin":"","legend":"","description":"","filename":"TableS7DifferentialanalysisresultsofurothelialcellsinMIBCNMIBCandPara.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/c4d5287f46ad8d7fbc445549.csv"},{"id":94139225,"identity":"13a3fa29-3da5-475a-8f82-0d0a584756b2","added_by":"auto","created_at":"2025-10-22 19:31:51","extension":"csv","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":809611,"visible":true,"origin":"","legend":"","description":"","filename":"TableS8DifferentialanalysisresultsofEpi9CSCbetweentheMIBCandNMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/3ef154a6a2e43276a5b40c71.csv"},{"id":94139208,"identity":"c9fc4dac-51a3-4104-a3ca-8735d95c5c16","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"csv","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":110,"visible":true,"origin":"","legend":"","description":"","filename":"TableS9CellChatInteractionCountandWeight.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/f592167a5edc3c0e5ac37d2e.csv"},{"id":94139223,"identity":"9aa77aa4-4f2e-4e91-99cd-0a5dc53fef30","added_by":"auto","created_at":"2025-10-22 19:31:51","extension":"xml","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":216548,"visible":true,"origin":"","legend":"","description":"","filename":"e4ff9004cdab4da49851c56be9b4b7821enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/8ddd0f57524410c7c9447365.xml"},{"id":94138095,"identity":"631e5a50-d6d0-4ca5-bb5d-472a06ee5ab9","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":175675,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/ed0b07530b7736316d5c99f1.png"},{"id":94140313,"identity":"15348aeb-1e28-4222-bd45-faa5f9dc894b","added_by":"auto","created_at":"2025-10-22 19:39:51","extension":"png","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":75731,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/d439e7a3e55a3ecd3c48d217.png"},{"id":94140310,"identity":"068e1a3b-903c-46dc-9ded-904db976331d","added_by":"auto","created_at":"2025-10-22 19:39:51","extension":"png","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":168715,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/9ec4c423883785cf47d7e772.png"},{"id":94138102,"identity":"bbbc8a72-2c57-4467-bf28-feba26fcf7e3","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":192403,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/937503530d5611347d49181b.png"},{"id":94138092,"identity":"b6f57a5c-2cf8-4579-9ed1-78a51c763360","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":38,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":228617,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/bb7e72b2aa1db8745b411cce.png"},{"id":94139220,"identity":"f0230fe4-6c68-4ddf-a806-1b4f59ce815b","added_by":"auto","created_at":"2025-10-22 19:31:51","extension":"png","order_by":39,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":463544,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/6410cb42b317328f9d874c9d.png"},{"id":94140812,"identity":"30954d49-5182-4378-85a5-73cca401703b","added_by":"auto","created_at":"2025-10-22 19:47:51","extension":"png","order_by":40,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":604687,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/48854edaab73101ea4112dd1.png"},{"id":94138100,"identity":"96caa6f7-aa81-479e-9d87-7b86db348b64","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":41,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":157371,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/fac1febc093e0203a24c2c5b.png"},{"id":94140814,"identity":"a08bb691-0e06-4aef-a28d-a23957933f96","added_by":"auto","created_at":"2025-10-22 19:47:51","extension":"png","order_by":42,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":148238,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/764871f9935899184e612dab.png"},{"id":94139212,"identity":"6902c991-c7b1-4bd3-bf63-200b7411e170","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"png","order_by":43,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":97783,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/b4e65eb2317b4f9f8d4165d5.png"},{"id":94138099,"identity":"0648f3d5-1505-49ff-8b92-5d1d6bff1365","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":44,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":227940,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/cd671431f5d4ba2570bdd475.png"},{"id":94138101,"identity":"6e13914f-fa85-4629-96c8-501bbbfe5dc1","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"xml","order_by":45,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":213318,"visible":true,"origin":"","legend":"","description":"","filename":"e4ff9004cdab4da49851c56be9b4b7821structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/d7fc0f1b14463b23d8e06d2d.xml"},{"id":94140315,"identity":"fcede973-8137-466f-a9c1-45e4200671c0","added_by":"auto","created_at":"2025-10-22 19:39:51","extension":"html","order_by":46,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":229257,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/44516b60bbe3b9cbf17cbc11.html"},{"id":94138037,"identity":"e833fb56-ef69-43e1-8901-5634c83cb738","added_by":"auto","created_at":"2025-10-22 19:23:49","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":636441,"visible":true,"origin":"","legend":"\u003cp\u003eCharacterization and distribution of single-cell subpopulations in BLCA.(A) Cell clustering plot. Each dot represents a single cell in a two-dimensional UMAP space.(B) Proportions of cell clusters derived from tumor (MIBC/NMIBC) and adjacent normal (Para) tissues.(C)Sample origin plot showing the tissue source of each cell. MIBC: Muscle-invasive BLCA; NMIBC: Non-muscle-invasive BLCA; Para: Adjacent normal tissue .(D) Dot plot showing the expression levels of representative marker genes across identified cell clusters.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/15e601dd31bd2166f637e8fa.jpeg"},{"id":94138038,"identity":"ee9c9372-4d2b-48af-9f4f-cb61e6e0a2ef","added_by":"auto","created_at":"2025-10-22 19:23:49","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":910152,"visible":true,"origin":"","legend":"\u003cp\u003eCharacterization of Urothelial cells and their genomic features. (A) Epithelial cell clustering plot. Each dot represents a single cell. (B) Sample composition plot, showing the distribution of cells across different tissue types and illustrating whether there are distribution differences between them. MIBC: Muscle-invasive BLCA; NMIBC: Non-muscle-invasive BLCA; Para: Adjacent normal tissue. (C) Dot plot showing the expression of characteristic genes in different clusters of urothelial cells. (D) Dot plot of gene set expression for CSC and EMT marker proteins and functional molecules. (E) CNV distribution of urothelial cells. The upper section represents the reference group, and the lower section shows urothelial cells. The x-axis denotes chromosomal positions, and the y-axis represents urothelial cell clusters. Red indicates gene amplification, and green indicates gene deletion.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/78a4e467ec46f65eb6930763.jpeg"},{"id":94139194,"identity":"ced909db-4e0d-4473-b73f-77eb5d607e86","added_by":"auto","created_at":"2025-10-22 19:31:49","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1264441,"visible":true,"origin":"","legend":"\u003cp\u003eKEGG Analysis, Gene Expression Dynamics, and Drug Sensitivity of BLCA Tumor Stem Cell Subpopulations.(A) KEGG Analysis: Shows the KEGG pathway analysis of partial urothelial cell subclusters, revealing their functional characteristics in biological pathways.(B) Gene Expression Dynamics: Displays the pseudotime gene expression dynamics from Epi9_CSC to Epi7_Mesenchymal.(C) Drug Sensitivity Analysis: Shows the drug sensitivity analysis of CSC subpopulations, with the y-axis representing -log10(IC50). A lower IC50 indicates higher drug efficacy, resulting in a higher -log10(IC50) value.(D) Drug Target Gene Expression: Presents the expression levels of specific drug target genes in the tumor stem cells.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/5a8f65394912295390e1f684.jpeg"},{"id":94140298,"identity":"8fcdcd16-2524-4228-b81c-de78a93d611d","added_by":"auto","created_at":"2025-10-22 19:39:49","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":848876,"visible":true,"origin":"","legend":"\u003cp\u003eDifferential Gene Volcano Plot. (A) Differential gene analysis of overall Urothelial cells in MIBC, NMIBC, and Para. (B) Differential gene analysis between Epi9_CSC in MIBC and NMIBC.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/9f28e8ac60ed28dbc05d3fa0.png"},{"id":94138046,"identity":"b41994cc-c8d4-4b80-a438-58ecaf0cff3b","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1743061,"visible":true,"origin":"","legend":"\u003cp\u003eComprehensive Analysis of Cellular Communication and Signaling Pathways in Urothelial cells.(A) The total number and interaction strength of cell-cell communications among Urothelial cells under different pathological conditions. (B) The main communicators and receivers of cell-cell communications among Urothelial cells under different pathological conditions. (C)The main signaling pathways emitted and received by Epi9_CSC in MIBC. (D)The main signaling pathways emitted and received by Epi9_CSC in NMIBC. (E)The main signaling pathways emitted and received by Epi9_CSC in Para.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4ae87d65fd841867baad4da0.png"},{"id":94138055,"identity":"adb8d852-98d1-4270-8429-2bf1de076633","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":849596,"visible":true,"origin":"","legend":"\u003cp\u003eCellChat Analysis of COLLAGEN Pathways in CSC. (A) Outgoing ligand-receptor interactions of COLLAGEN pathways in CSC. (B) Incoming ligand-receptor interactions of COLLAGEN pathways in CSC.\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/a9db3d69a6ac2aa3bcf1dd99.jpeg"},{"id":94139196,"identity":"50a5c894-d623-42bf-893b-a562b62e456b","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"jpeg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":838692,"visible":true,"origin":"","legend":"\u003cp\u003eCSCRPI's Construction, Validation, and Clinical Application for Prognosis Prediction in BLCA.(A) Bar chart ranking the top 10 prognostic genes by weighted scores from four machine learning algorithms. (B) Kaplan-Meier curves for high vs. low CSCRPI groups in the TCGA-BLCA cohort.(C) Kaplan-Meier curves for high vs. low CSCRPI groups in the GSE31684 cohort.(D) Nomogram predicting 1, 3, and 5year survival using CSCRPI, age, sex, and clinical stage.(E) Calibration plot comparing predicted and actual survival at 1, 3, and 5 years. (F) Decision curve showing the nomogram's clinical benefit over other variables.(G) In the TCGA-BLCA cohort, ROC curves show the nomogram's accuracy for 1, 3, and 5-year OS. The heatmap of MPCDI distribution, survival status, and the expression of six key DEPCDRGs highlight the link between predictions and outcomes.(H)In the GSE31684 cohort, ROC curves validate the nomogram's 1, 3, and 5year OS predictions. The MPCDI heatmap, survival status, and six key DEPCDRG expressions confirm the model's performance in an independent dataset.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/2f0fca619113e98e1fb08a99.jpeg"},{"id":94138048,"identity":"b20ce4d3-3294-4e33-99e9-f4429bfbf551","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"jpeg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":442867,"visible":true,"origin":"","legend":"\u003cp\u003eClinical relevance assessment of the CSCRPI. (A-C) Differences in CSCRPI between groups with different clinical features. (D-I)OS KM curves for CSCRPI in the two groups stratified by clinicopatho[1]logic factors .\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4a0e9d7b6bdc0296b1c9741c.jpeg"},{"id":94140307,"identity":"94cad74d-0383-461a-ac7c-f63afb784802","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"jpeg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":1140871,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis of Tumor Immune Microenvironment, Prediction of Immunotherapy Efficacy.(A) Stroma score, Immune score, ESTIMATE score, and Tumor purity in the two CSCRPI groups.(B) ssGSEA score of immune cell infiltration in the two CSCRPI groups.(C) The correlation between each key molecule and each TME infiltration cell type.(D) TIDE score, T-cell rejection score, T-cell dysfunction score, and microsatellite instability (MSI) score in the two groups.(E) Comparisons of the IPS in the two CSCRPI groups.(F) Immune-related pathways' activity showing a significant difference between the high CSCRPI group and the low CSCRPI group.(G) The five key prognostic genes were associated with common immune checkpoints.(H) Expression levels of immune checkpoints in high and low CSCRPI groups.\u003c/p\u003e","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/418abe64bb221549ca84ba38.jpeg"},{"id":94139202,"identity":"e24c7340-e020-4ea8-8710-e39a5121983a","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"jpeg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":371028,"visible":true,"origin":"","legend":"\u003cp\u003eDrug sensitivity analysis between the two CSCRPI groups. (A)Differences in response to target drugs between the high and low CSCRPI groups. (B)Correlation between the constructed CSCRPI genes and sensitivity to selected target drugs.\u003c/p\u003e","description":"","filename":"floatimage10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/466b0b4f006ec9b96a6d4259.jpeg"},{"id":94138076,"identity":"2fcd71ec-daf0-469f-a8df-c5bc12118631","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":402383,"visible":true,"origin":"","legend":"\u003cp\u003eEpi9_CSC enhances the proliferation, survival, motility, invasion, and self-renewal capabilities of tumor stem cells by secreting collagen that binds to integrins and CD44 receptors on the cell membrane, activating multiple signaling pathways during the EMT process.\u003c/p\u003e","description":"","filename":"floatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4a5b1ae8c3ccb27516ab065f.png"},{"id":97664559,"identity":"550613a9-9a2f-4f17-bd6e-cdade980ce81","added_by":"auto","created_at":"2025-12-08 09:10:01","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10180344,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4c35d5a8-108e-4ef1-829e-b6b1dd378f8b.pdf"},{"id":94138051,"identity":"5a34677b-8768-4d09-b33d-7659e1b04ab8","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"doc","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":2790622,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementalMaterial.doc","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/5e6108cc1ffcf6306ec1dd82.doc"},{"id":94140297,"identity":"b755d512-1a8b-4773-ad72-914ec03f7841","added_by":"auto","created_at":"2025-10-22 19:39:49","extension":"csv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":10614,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1SampleGroupingInformation.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/6e277ad48a734d049b65222c.csv"},{"id":94138042,"identity":"50313177-f2ce-4e56-a9cd-8930b96436a0","added_by":"auto","created_at":"2025-10-22 19:23:49","extension":"csv","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":597,"visible":true,"origin":"","legend":"","description":"","filename":"TableS2CNVResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/585c2dae34549514274233c7.csv"},{"id":94140808,"identity":"aab1d001-a81f-4094-b33e-c1763919b3ab","added_by":"auto","created_at":"2025-10-22 19:47:50","extension":"csv","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":1901668,"visible":true,"origin":"","legend":"","description":"","filename":"TableS3CharacteristicGeneResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/6e7f95e23f0c778fa86407ca.csv"},{"id":94138043,"identity":"a4b6c835-e2b0-457e-9a4e-745d37025573","added_by":"auto","created_at":"2025-10-22 19:23:49","extension":"csv","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":7867,"visible":true,"origin":"","legend":"","description":"","filename":"TableS4KEGGAnalysisResultsofUrothelialCellSubgroups.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/19885816bcc77901a99118c6.csv"},{"id":94138059,"identity":"9e3c6505-6d72-4333-969f-3da34c483f0f","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":2310045,"visible":true,"origin":"","legend":"","description":"","filename":"TableS5PseudotimeanalysisofCSCsdifferentiationintomesenchymalsubtypesusingMonocle3.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/0ad1d4cb0307cbea6c0c2083.csv"},{"id":94140302,"identity":"ab639a3b-69a5-464a-8782-aff32df782b2","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"csv","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":1285885,"visible":true,"origin":"","legend":"","description":"","filename":"TableS6DrugsensitivityscoringresultsforCSCsusingoncoPredict..csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/649f9bdcf1d651aefbb6076f.csv"},{"id":94138074,"identity":"199b4f33-9c67-4564-bcfc-da6586339b6d","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":174334,"visible":true,"origin":"","legend":"","description":"","filename":"TableS7DifferentialanalysisresultsofurothelialcellsinMIBCNMIBCandPara.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/5b0895325c2dd84376e6eb84.csv"},{"id":94140810,"identity":"8fcb40e8-9d49-493d-a298-5e0005acc8a8","added_by":"auto","created_at":"2025-10-22 19:47:50","extension":"csv","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":809611,"visible":true,"origin":"","legend":"","description":"","filename":"TableS8DifferentialanalysisresultsofEpi9CSCbetweentheMIBCandNMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/3cd423f5a95b117c7ad7bef9.csv"},{"id":94140301,"identity":"e59921f4-e43c-4af5-9364-ba76e02cf106","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"csv","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":110,"visible":true,"origin":"","legend":"","description":"","filename":"TableS9CellChatInteractionCountandWeight.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4da514c6cbd9e26be2b53c68.csv"},{"id":94140809,"identity":"3e411b75-fdf9-45a1-b031-34986cea8c6b","added_by":"auto","created_at":"2025-10-22 19:47:50","extension":"csv","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":413195,"visible":true,"origin":"","legend":"","description":"","filename":"TableS10CellChatMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/5e53c0001b91317f93b9b0c1.csv"},{"id":94141233,"identity":"1907a224-9077-4d5e-a2f8-a690981ef8a3","added_by":"auto","created_at":"2025-10-22 19:55:51","extension":"csv","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":191537,"visible":true,"origin":"","legend":"","description":"","filename":"TableS11CellChatNMIBC.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/4f77a1917e68a6d525c31f44.csv"},{"id":94139206,"identity":"db03021a-5f2e-4bf9-aa9d-623fae86ca0c","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"csv","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":142777,"visible":true,"origin":"","legend":"","description":"","filename":"TableS12CellChatPara.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/973482568dea5f88c572c795.csv"},{"id":94140305,"identity":"4b83938d-e7f2-4534-8482-1a6cdbf29272","added_by":"auto","created_at":"2025-10-22 19:39:50","extension":"csv","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":19222,"visible":true,"origin":"","legend":"","description":"","filename":"TableS13182prognosticCSCmarkergenesbyunivariateCoxregression.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/8fd92522007cc315b72d414e.csv"},{"id":94138072,"identity":"63f9210b-d420-4e69-8ec9-bad0d80b5430","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":14696,"visible":true,"origin":"","legend":"","description":"","filename":"TableS14Resultsoffourmachinelearningweightedaveragingmethods.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/2d6ac1abe6f08f249c03adb0.csv"},{"id":94139214,"identity":"59067b97-39d2-4327-a17c-470a98906d7a","added_by":"auto","created_at":"2025-10-22 19:31:51","extension":"csv","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":65542,"visible":true,"origin":"","legend":"","description":"","filename":"TableS15CoregeneexpressionprofileandclinicaldataofCSCRPIpatients.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/c8a797eb6dd9c5d7182d51a4.csv"},{"id":94138087,"identity":"69940ac6-c6fe-4503-8c89-d4fe7a587754","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"csv","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":33806,"visible":true,"origin":"","legend":"","description":"","filename":"TableS16CSCRPIESTIMATEresults.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/9b6e3b4aa07cd1fa9fccc8eb.csv"},{"id":94138071,"identity":"50e95a16-9bdb-4145-ac68-fecc6f566de5","added_by":"auto","created_at":"2025-10-22 19:23:50","extension":"csv","order_by":17,"title":"","display":"","copyAsset":false,"role":"supplement","size":204118,"visible":true,"origin":"","legend":"","description":"","filename":"TableS17CSCRPIssgsearesults.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/253aeb27939a181e3b44a15e.csv"},{"id":94138086,"identity":"4f116f5f-3c91-4490-8436-50e3a1874808","added_by":"auto","created_at":"2025-10-22 19:23:51","extension":"csv","order_by":18,"title":"","display":"","copyAsset":false,"role":"supplement","size":1271561,"visible":true,"origin":"","legend":"","description":"","filename":"TableS18CSCRPIGDSC2.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/f192750b8f360fc772f7b09e.csv"},{"id":94139209,"identity":"83f0a925-039a-435d-9cee-a7b115521106","added_by":"auto","created_at":"2025-10-22 19:31:50","extension":"csv","order_by":19,"title":"","display":"","copyAsset":false,"role":"supplement","size":37587,"visible":true,"origin":"","legend":"","description":"","filename":"TableS19CSCRPICellMiner.csv","url":"https://assets-eu.researchsquare.com/files/rs-7536468/v1/3fb004651fd2646b9c18ed76.csv"}],"financialInterests":"No competing interests reported.","formattedTitle":"scRNA-seq reveals epithelial heterogeneity in bladder cancer and establishes a cancer stem cell prognostic model","fulltext":[{"header":"Background","content":"\u003cp\u003eBLCA is a malignant tumor that originates in the bladder lining and is one of the most common cancers in the urinary system. The exact cause of BLCA remains unclear, but various factors are believed to increase its risk, including smoking, exposure to certain chemicals (such as aniline dyes and aromatic amines), chronic bladder inflammation or infection, and genetic susceptibility [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Clinically, the most prominent early symptom of BLCA is painless hematuria, and patients may also experience urinary frequency, urgency, and dysuria [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e], which often lead them to seek medical attention. As the ninth most common cancer globally, BLCA accounted for approximately 614,000 new cases and 220,000 deaths in 2022, with a significantly higher burden and incidence in men than in women [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Studies suggest that in the next decade, the incidence and mortality rates of BLCA in China will further increase [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e], placing a substantial burden on the country\u0026rsquo;s economy. The overall five-year survival rate for BLCA is approximately 77% [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], with cancer staging being a crucial factor in prognosis and treatment. As the tumor stage progresses, the survival rate of BLCA decreases [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Tumor staging is closely linked to the TME [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. The TME influences tumor progression through various complex mechanisms. Its main components include epithelial cells, immune cells, endothelial cells, fibroblasts, and ECM, which interact closely with tumor cells to promote tumor growth, angiogenesis, and metastasis [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. In fact, most malignant tumors are composed of several subpopulations of tumor cells with different phenotypes, a fact that has been widely confirmed by research over the past years [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. These studies have revealed the degree and complexity of tumor cell heterogeneity, providing a deeper understanding of cancer. Today, the phenotypic diversity of tumor cells within the tumor is recognized as a major driver of therapeutic resistance, drawing increasing attention [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn the tumor microenvironment, epithelial cells, as one of the initial cell types that form tumors, influence tumor progression and metastasis through various mechanisms. One key mechanism is EMT, during which epithelial cells lose polarity and tight cell junctions, acquiring enhanced migratory and invasive abilities, thus facilitating tumor metastasis [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Additionally, the cytokines and growth factors secreted by epithelial cells can regulate the behavior of immune cells, recruiting immunosuppressive cells such as tumor-associated macrophages and regulatory T cells, helping tumor cells evade immune surveillance [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e][\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Meanwhile, epithelial cells promote angiogenesis by secreting vascular endothelial growth factor (VEGF) [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e], providing the tumor with sufficient oxygen and nutrients to support its rapid growth. In this complex environment, tumor stem cells also play an important role. Tumor stem cells, with their self-renewal capacity, not only drive tumor growth and recurrence but also interact with epithelial cells, further enhancing the malignancy of the tumor. These interactions make tumors more difficult to treat with traditional therapies [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Therefore, therapeutic strategies targeting the EMT process in epithelial cells, their angiogenic capacity, and their interaction with tumor stem cells could provide new directions for effectively inhibiting tumor progression and metastasis.\u003c/p\u003e\u003cp\u003eIn traditional transcriptome sequencing (bulk RNA-seq), researchers typically sequence mixed samples from large populations of cells to obtain overall gene expression information. However, this approach may obscure the specific expression patterns and functions of epithelial cells within the tumor microenvironment. In contrast, scRNA-seq enables gene expression analysis at the single-cell level, offering a more precise way to reveal cellular heterogeneity. This technological advancement provides a new perspective for studying the specific roles of epithelial cells in various cancers, especially in revealing their diversity and complexity within the tumor microenvironment [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThis study will utilize single-cell sequencing data to analyze the characteristics of epithelial cells in BLCA and construct a detailed atlas of epithelial cells within the BLCA tumor microenvironment. By identifying specific tumor cell subpopulations and applying machine learning methods to develop and validate prognostic models for these specific tumor cells, the study will further explore their clinical significance and biological characteristics. This will help gain a deeper understanding of the role of epithelial cells in the onset and progression of BLCA and provide new insights for precision and personalized treatment strategies.\u003c/p\u003e\u003cp\u003eThis study will utilize single-cell sequencing data to analyze epithelial cell characteristics in BLCA and construct a detailed atlas of epithelial cells within the BLCA tumor microenvironment. By identifying specific tumor cell subpopulations and combining machine learning methods to develop and validate prognostic models targeting these specific tumor cells, we will further explore their clinical significance and biological characteristics. This will help deepen our understanding of the role of epithelial cells in BLCA occurrence and progression, and provide new insights for precision therapy and personalized treatment strategies.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eData Acquisition\u003c/h2\u003e\u003cp\u003eThe scRNA-seq datasets for BLCA, including GSE192575, GSE145137, GSE135337, GSE222315, and GSE129845, were downloaded from the GEO database. Tumor samples were classified into MIBC for grades T2 and above, and NMIBC for grades below T2. The dataset comprises samples from 5 muscle-invasive bladder cancer (MIBC), 14 non-muscle-invasive bladder cancer (NMIBC), and 8 adjacent normal (Para) tissues. Detailed sample information is provided in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.The bulk RNA-seq datasets for BLCA were obtained from the TCGA and GEO databases under the following accession IDs: TCGA-BLCA and GSE31684. The TCGA-BLCA dataset includes 372 tumor samples with complete clinical information, which were used as the training cohort. The bulk RNA-seq expression matrix from GSE31684 was normalized and used as an external validation cohort. The corresponding clinical information for these bulk RNA-seq datasets includes overall survival (OS), survival status, age, gender, pathological stage, and TNM stage. The drug sensitivity training dataset from GDSC2 was downloaded from the GDSC database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cancerrxgene.org/\u003c/span\u003e\u003cspan address=\"https://www.cancerrxgene.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Furthermore, the genomic annotation file for copy number variation analysis, hg38_gencode_v27, was downloaded from the Broad Institute server (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://data.broadinstitute.org/Trinity/CTAT/cnv\u003c/span\u003e\u003cspan address=\"https://data.broadinstitute.org/Trinity/CTAT/cnv\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eData Quality Control and Filtering\u003c/h3\u003e\n\u003cp\u003eThe scRNA-seq expression matrix obtained in the previous step was imported into R. Data quality control was performed using the Seurat package (version 3.2.2). The mitochondrial gene ratio (mt_percent) for each cell was calculated by identifying genes that start with \"MT-\". The erythrocyte gene ratio (HB_percent) for each cell was calculated using a known list of erythrocyte genes. Cells with a mitochondrial gene ratio greater than 35% and cells with an erythrocyte gene ratio greater than 3% were filtered out. Only cells with gene and transcript counts within the range of the mean\u0026thinsp;\u0026plusmn;\u0026thinsp;2 standard deviations were retained.\u003c/p\u003e\n\u003ch3\u003eData Normalization and Feature Gene Selection\u003c/h3\u003e\n\u003cp\u003eThe Harmony package (version 1.2.3) was used to correct for batch effects across samples using the RunHarmony function. The sample name in the metadata was chosen as the integration variable. The Parameters were set with a theta value of 3, a lambda value of 0.6, and a maximum of 20 iterations.\u003c/p\u003e\n\u003ch3\u003eData Integration\u003c/h3\u003e\n\u003cp\u003eThe Seurat package's NormalizeData function was applied to normalize the data, with the normalization.method set to \"LogNormalize\" and the scale.factor set to 10,000. The FindVariableFeatures function was used to select feature genes for subsequent analysis, with the selection.method set to \"vst\" and nfeatures set to 2,000. The ScaleData function was then applied to normalize the integrated data.\u003c/p\u003e\n\u003ch3\u003eDimensionality Reduction and Cell Clustering\u003c/h3\u003e\n\u003cp\u003ePrincipal component analysis (PCA) was performed using the Seurat package's RunPCA function, with the number of principal components (npcs) set to 30. The data was visualized and reduced in dimensionality using the RunTSNE and RunUMAP functions, with dims set to 1:20. Cell clustering was performed using the FindNeighbors and FindClusters functions, with dims set to 1:20 and resolution set to 0.8. Differentially expressed genes for each cluster were identified using the FindAllMarkers function, with min.pct set to 0.5, logfc.threshold set to 0.5, and the test.use set to \"wilcox\". Based on the reference literature for cell cluster-specific marker genes, each cell cluster was defined according to its characteristic gene expression profile.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eData Processing and Cell Clustering of Epithelial Cell Clusters\u003c/h2\u003e\u003cp\u003eAfter extracting the epithelial cell clusters, the data was processed using the same methods as described above, including normalization, feature gene selection, scaling, dimensionality reduction, and cell clustering. The epithelial cell subpopulations were annotated using BLCA molecular subtype-related marker genes and tumor stem cell genes.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eCNV Analysis\u003c/h3\u003e\n\u003cp\u003eCNV analysis on epithelial cell subgroups using the infercnv package (version 1.22.0). Randomly select 800 plasma cells and 800 endothelial cells from all samples as the reference group for CNV analysis. Set the cutoff value to 0.2 to filter out low expression data, choose 'ward.D2' as the hierarchical clustering method, and enable the denoising step to reduce technical noise.\u003c/p\u003e\n\u003ch3\u003eKEGG Analysis\u003c/h3\u003e\n\u003cp\u003eTo identify expression characteristics of each epithelial cell subpopulation, genes were selected based on the following criteria: an average log fold change (logFC) greater than 2, an expression proportion (pct.1) exceeding 30% within the subpopulation, and an expression proportion (pct.2) below 10% in other subpopulations. These selected genes were used as gene sets for KEGG enrichment analysis, performed using the compareCluster function from the clusterProfiler package. The Parameters were set as follows: fun = \"enrichKEGG\", organism = \"hsa\", and pvalueCutoff\u0026thinsp;=\u0026thinsp;0.05. KEGG pathways with P-values less than 0.05 were retained and exported as enrichment result tables. All genes from the Epi9_CSC-enriched pathways that were differentially expressed compared to other subpopulations were uploaded to KEGG Mapper for visualization.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eDifferential Gene Comparison Analysis Between Groups\u003c/h2\u003e\u003cp\u003eDifferentially expressed genes between different sample sources were identified using the FindMarkers function from the Seurat package. The log fold change threshold was set to 1, min.pct was set to 0.3, and the test.use was selected as \"wilcox.\" Subsequently, a volcano plot was generated using the ggplot2 package.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eDrug Sensitivity Test Analysis of Tumor Stem Cell Subpopulations\u003c/h2\u003e\u003cp\u003eAfter extracting the tumor stem cell subpopulations, drug sensitivity analysis was performed using the oncoPredict package (version 1.2). The ComBat method was applied to batch-correct the expression data for both training and test sets to reduce batch effect. The removeLowVaryingGenes Parameter was set to 0.2, removing genes with low variability (below the 20th percentile) from the original data to simplify the model and minimize noise interference. A boxplot of drug sensitivity scores was generated using the ggplot2 package, with specific drugs marked in red for easier identification. The drug sensitivity analysis table was also exported.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003ePseudotime Analysis\u003c/h2\u003e\u003cp\u003eThe Monocle3 package (version 1.3.7) was used to read the epithelial cell cluster data object constructed by the Seurat package and convert it to Monocle3 format using the new_cell_data_set() function. The UMAP coordinates computed by Seurat were directly used for subsequent analysis. In constructing the cell trajectory, the learn_graph() function was called with a series of Parameters set: euclidean_distance_ratio was set to 380 to control the ratio of Euclidean distance between two tree apex nodes to the maximum path distance; geodesic_distance_ratio was set to 400 to control the ratio of geodesic distance to the tree diameter path length; minimal_branch_len was set to 70 to determine the minimum branch length to retain; prune_graph was set to TRUE to remove insignificant small branches; and nn.k was set to 3 to define the number of nearest neighbors calculated in the reverse graph embedding process. Finally, the order_cells() function was used to arrange the cells in pseudotime and construct and label the cell trajectory.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eCell-Cell Interaction Analysis\u003c/h2\u003e\u003cp\u003eThe cell cluster data object constructed by the Seurat package was imported into the R language's CellChat package. The \"Secreted Signaling\" reference database was selected for the analysis. The computeCommunProb function was applied to assign a probability value to each interaction and perform permutation tests to infer biologically significant cell-to-cell interactions. The computeCommunProbPathway function was used to infer the interactions between cells at the signaling pathway level. The aggregateNet function was applied to calculate the number of links or aggregate communication probabilities, thereby constructing an aggregated communication network between cells.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eMachine Learning-Based CSC-Related Gene Integration Method\u003c/h2\u003e\u003cp\u003eThe CSC subgroup marker genes obtained from the FindAllMarkers function were filtered based on the condition of avg_log2FC\u0026thinsp;\u0026gt;\u0026thinsp;2, and univariate Cox regression analysis was performed on the resulting genes, yielding 182 CSC subgroup marker genes associated with prognosis. Four machine learning algorithms were then used to identify core CSC prognosis-related genes. The four algorithms included Random Forest (RF), Lasso, XGBoost, and Decision Tree (DT). The TCGA cohort was set as the training cohort, and GSE31684 was used as the validation cohort. The results obtained from the four algorithms were combined using a weighted average method to select the top 10 genes as the central CSC prognosis-related genes.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\u003ch2\u003eConstruction and Validation of Tumor Stem Cell-Related Prognostic Index (CSCRPI)\u003c/h2\u003e\u003cp\u003eThe top 10 central CSC prognosis-related genes selected by the weighted average method were subjected to multivariate Cox analysis to determine the most reliable independent prognostic factors. The CSCRPI for patients was calculated using the following formula:CSCRPI\u0026thinsp;=\u0026thinsp;coefficient(gene1) \u0026times; expression(gene1)\u0026thinsp;+\u0026thinsp;coefficient(gene2) \u0026times; expression(gene2) + ... + coefficient(genen) \u0026times; expression(genen).In this formula, expression(gene n) represents the expression level of a specific gene, and coefficient(gene n) refers to the coefficient obtained from the multivariate Cox analysis. Based on the median CSCRPI, BLCA patients were divided into low-CSCRPI and high-CSCRPI groups. Kaplan-Meier analysis was performed using the \"survival\" R package to investigate the relationship between the survival status of BLCA patients and CSCRPI, with further validation conducted using the GSE31684 cohort.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\u003ch2\u003eNomogram Construction\u003c/h2\u003e\u003cp\u003eBy combining clinical features of BLCA patients, such as age, gender, clinical pathological stage, and CSCRPI, through multivariate Cox and stepwise regression analysis, a prognostic nomogram was created. The nomogram and calibration plots were visualized using the \"rms\" package. ROC curve analysis was performed to evaluate the performance of CSCRPI in predicting the 1-year, 2-year, and 3-year overall survival rates of BLCA patients. Subsequently, decision curve analysis (DCA) was used to assess the net benefit of combining the nomogram with a model containing only clinical features. Furthermore, correlation and stratified analyses of CSCRPI based on the clinical features from the TCGA dataset were also performed.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\u003ch2\u003eTumor Immune Microenvironment Analysis and Immune Therapy Efficacy Evaluation\u003c/h2\u003e\u003cp\u003eTo quantify the immune infiltration status in BLCA patients, the \"IOBR\" R package was used for analysis, employing nine different algorithms, including the MCP-counter, EPIC, xCell, CIBERSORT, IPS, quanTIseq, ESTIMATE, TIMER, and ssGSEA algorithms. Additionally, the Cancer Immunome Atlas (TCIA) was utilized to evaluate the potential response of BLCA patients to checkpoint immunotherapies. We also conducted a comprehensive analysis to explore the correlation between immune phenotype scores (IPS) for anti-PD-1 and anti-CTLA4 treatments and MPCDI in BLCA patients.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003ch2\u003eDrug Sensitivity Analysis for High-CSCRPI and Low-CSCRPI Groups\u003c/h2\u003e\u003cp\u003eTo personalize treatment, the \"oncoppredict\" R package was used to predict the chemotherapy sensitivity of BLCA patients with different MPCDI scores. The expression profile of patient tissues was matched with gene expression profiles from cancer cell lines, and the half-maximal inhibitory concentration (IC50) was calculated. The Wilcoxon test was applied to explore the differences in drug IC50 between the two groups, and a P-value of \u0026lt;\u0026thinsp;0.05 was considered statistically significant. To improve the accuracy of drug sensitivity analysis, GSCALite platforms' GDSC and CellMiner databases were used for further analysis.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eIn this study, we performed statistical analyses using R software (version 4.4.0). Batch effects between samples were corrected using the Harmony algorithm. For differential gene expression analysis, we employed the Wilcoxon rank-sum test. Cell trajectory analysis was conducted using the Monocle3 software, and the significance of gene expression was assessed using the Wilcoxon test, with the Benjamini-Hochberg method applied to adjust for false positives. Additionally, permutation tests were used to validate the robustness and statistical significance of the trajectory model. The cell interaction network was also evaluated using permutation tests. Survival curves were generated using the Kaplan-Meier method. To compare multiple or two groups of data, we used the Wilcoxon test or Kruskal-Wallis test. Correlation assessments were conducted using Spearman correlation analysis. Results with a p-value less than 0.05 were considered statistically significant. For visualization purposes, a \u003cem\u003ep\u003c/em\u003e-value less than 0.05 was represented as *, less than 0.01 as **, and less than 0.001 as ***.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\n \u003ch2\u003eIdentification of BLCA Cell Subtypes\u003c/h2\u003e\n \u003cp\u003eFirst, the data from cells originating from MIBC, NMIBC, and Para organizations were integrated, followed by batch effect removal, PCA analysis, dimensional reduction, clustering, and differential gene identification. A total of 160,155 cells were obtained, classified into 39 clusters. Based on the differentially expressed genes highly expressed in each cluster and known cell-specific marker genes (Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e), 14 major cell types were identified: Urothelial cells, Tregs, Memory T cells, Effector T cells, Activated Memory B cells, Activated B cells, Plasma cells, Monocyte-macrophage, iCAF, matCAF, Smooth muscle cells, Angiogenic ECs, Inflammatory ECs, and Mast Cells (Fig. 1A). Figure 1D displays the marker gene expression patterns for each cell type. Further analysis of the cell proportions in MIBC, NMIBC, and adjacent normal tissue samples (Fig. 1B, D) revealed that urothelial cells comprise 66.79% of the total cells in MIBC, 50.60% in NMIBC, and 23.45% in Para-originating samples.\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eMarker genes of cell types in BLCA and adjacent normal tissue samples\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCell Type Name\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eMarker Genes\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUrothelial cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKRT8、TACSTD2、KRT19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTregs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD3D, CD4, FOXP3, CTLA4, IKZF2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMemory T cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD3D, CD4,IL7R, CCR7, SELL\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEffector T cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD3D, CD8A, CD8B, IFNG, GZMB, PRF1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eActivated Memory B cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD19, CD79A, CD79B,MS4A1, CD27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eActivated B cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD19, CD79A, CD79B,CD69, CD80, CD40\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePlasma cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD19, CD79A, CD79B,IGLL5, IGHG1, IGHA1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMonocyte-macrophage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD14, CD68, MS4A7, FCGR3A\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eiCAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIL6, CXCL1, LIF\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ematCAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCOL1A1, COL3A1, FN1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmooth muscle cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTAGLN, ACTA2, MYLK\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAngiogenic ECs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVWF, CDH5, PECAM1, VEGFA, KDR, DLL4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eInflammatory ECs\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVWF, CDH5, PECAM1,ICAM1, VCAM1, SELE\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMast Cells\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKIT, TPSAB1, TPSB2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\n \u003ch2\u003eUrothelial Cell Clustering and Definition\u003c/h2\u003e\n \u003cp\u003eAfter extracting the urothelial cell population, re-clustering analysis was performed. After removing contaminating cells, a total of 83,672 cells were obtained, which were divided into 11 clusters. The sample source of each cluster was heterogeneous, indicating that the urothelial cell subpopulations exhibit heterogeneity under different pathological conditions. Clusters 1, 3, 4, and 5 predominantly originated from NMIBC, clusters 6, 8, and 10 primarily from MIBC, and cluster 9 mainly from MIBC and Paraffin-embedded tissues. The remaining clusters were found in MIBC, NMIBC, and Para (Fig. 2A,B). CNV analysis of each urothelial cell cluster revealed copy number amplifications or deletions in all clusters, with CNV scores being generally high, suggesting the presence of tumor cells in all clusters (Fig. 2E). Based on BLCA molecular subtype-related marker genes and tumor stem cell genes (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e), each cluster was named [\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e][\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e]. Clusters 0, 2, 3, and 10 were mixed clusters of multiple subtypes, clusters 1, 4, and 5 mainly expressed HER2 subtype marker genes, clusters 6 and 8 mainly expressed HER2 and luminal subtype marker genes, cluster 7 primarily expressed mesenchymal subtype marker genes, and cluster 9 primarily expressed tumor stem cell marker genes. These clusters were thus named as follows: Epi0_Mix, Epi1_HER2, Epi2_Mix, Epi3_Mix, Epi4_HER2, Epi5_HER2, Epi6_Luminal_HER2, Epi7_Mesenchymal, Epi8_Luminal_HER2, Epi9_CSC, and Epi10_Mix(Fig.\u0026nbsp;2C).In addition, using the tumor stem cell gene set (CD44, SOX2, IGF1R, SOX4, ARRB1, ARRB2, ALDH1A1, POU5F1, CDK1, DCLK1, NANOG) [\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e] and the EMT gene set (VIM, CDH2, FOXC2, SNAI1, SNAI2, TWIST1, FN1, ITGB6, MMP2, MMP3, MMP9, SOX10) [\u003cspan class=\"CitationRef\"\u003e19\u003c/span\u003e], gene set scoring was performed on Urothelial cells. The results suggest that the tumor stem cell and EMT gene set scores for Epi9_CSC are high(Fig.\u0026nbsp;2D), further confirming that cluster 9 consists of tumor stem cells and that the CSC subpopulation exhibits an EMT phenotype [\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eCharacteristic genes related to molecular subtypes of BLCA and tumor stem cell genes\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCell Type Name\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eMarker Genes\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLuminal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKRT20、UPK3A、FOXA1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHer2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eERBB2、GRB7、MUC1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSquamous\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKRT5、KRT14、TP63\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMesenchymal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAXL、VIM、CDH2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePapillary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFGFR3、TACC3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNeural\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWNT3A、SYP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCSC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCD44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\n \u003ch2\u003eKEGG Analysis of Urothelial Cell Subpopulations\u003c/h2\u003e\n \u003cp\u003eTo understand the phenotypic diversity of tumor cells in BLCA, KEGG analysis was performed on the marker genes of urothelial cell subpopulations (Fig. 3A), including Epi1_HER2, Epi4_HER2, Epi5_HER2, Epi6_Luminal_HER2, and Epi9_CSC. The enrichment results for Epi1_HER2, Epi4_HER2, and Epi5_HER2 were: ribosome and Coronavirus disease - COVID-19. The enrichment results for Epi5_HER2 were: Epstein-Barr virus infection. The enrichment results for Epi9_CSC included: Focal adhesion, ECM-receptor interaction, Cytoskeleton in muscle cells, Proteoglycans in cancer, Human papillomavirus infection, PI3K-Akt signaling pathway, and Prion disease. We uploaded all genes with differential expression between the Epi9_CSC enrichment pathways and the other subpopulations to KEGG Mapper. The results showed that the four pathways\u0026mdash;Focal adhesion, ECM-receptor interaction, Proteoglycans in cancer, and PI3K-Akt signaling pathway\u0026mdash;are closely interconnected( refer to Supplementary Figures \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e-S4).\u003c/p\u003e\n \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\n \u003ch2\u003ePseudotime Analysis of Cancer Stem Cells and Mesenchymal Subpopulations\u003c/h2\u003e\n \u003cp\u003eCSCs are a distinct subpopulation of tumor cells characterized by self-renewal, multipotency, and resistance to therapy. They are considered key drivers of tumor heterogeneity, invasiveness, and recurrence [\u003cspan class=\"CitationRef\"\u003e21\u003c/span\u003e]. In BLCA, the mesenchymal-like subtype is the most aggressive and associated with the poorest prognosis [\u003cspan class=\"CitationRef\"\u003e17\u003c/span\u003e]. Its hallmark is the pronounced activation of EMT, leading to high migratory capacity, drug resistance, and metastatic potential [\u003cspan class=\"CitationRef\"\u003e22\u003c/span\u003e]. Recent studies have revealed a strong link between CSCs and the mesenchymal subtype: CSCs may evolve into mesenchymal-like tumor cells through EMT or other mechanisms, promoting the progression of BLCA toward a more invasive phenotype [\u003cspan class=\"CitationRef\"\u003e23\u003c/span\u003e].In this study, CSCs and mesenchymal-like cells were extracted for pseudotime analysis to reconstruct the continuous differentiation trajectory from CSCs to the mesenchymal subtype. This approach enabled us to identify key transitional stages in cell state evolution and to explore the dynamic EMT process and its core regulatory factors.During the differentiation of BLCA CSCs into the mesenchymal subtype, dynamic changes in gene expression revealed critical biological mechanisms. Downregulated genes included NT5E (CD73), SERPINA1, TGFBI, LAMC2, DCBLD2, CMTM3, and COL7A1, while upregulated genes included COL1A2, COL6A2, and COL3A1 (Fig.\u0026nbsp;3B).\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\n \u003ch2\u003eDrug Sensitivity Analysis of Cancer Stem Cell Subpopulations\u003c/h2\u003e\n \u003cp\u003eTo explore the sensitive drugs for BLCA tumor stem cells, this study conducted GDSC drug sensitivity analysis on the tumor stem cell subpopulations. The results showed that the tumor stem cell subpopulations were sensitive to Bortezomib, Dactinomycin, Docetaxel, Daporinad, Sepantronium bromide, Vinblastine, Eg5_9814, Vinorelbine, Staurosporine, Dinaciclib, Paclitaxel, and other drugs (Fig. 3C). Additionally, we explored the targets and pathways of these sensitive drugs using the GDSC database (Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e), and evaluated the expression of target genes of certain drugs in CSCs (Fig.\u0026nbsp;3D).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\n \u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDrug Targets and Pathways Sensitive to Cancer Stem Cell Subpopulations.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDrug Name\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDrug ID\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDrug Target\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTarget Pathway\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDactinomycin\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1911\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRNA polymerase\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDocetaxel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMicrotubule stabiliser\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMitosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDaporinad\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1248\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNAMPT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMetabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSepantronium bromide\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1941\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBIRC5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eApoptosis regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVinblastine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMicrotubule destabiliser\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMitosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEg5_9814\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1712\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKSP11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOther\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVinorelbine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2048\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMicrotubule destabiliser\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMitosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eStaurosporine\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBroad spectrum kinase inhibitor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRTK signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDinaciclib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1180\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCDK1, CDK2, CDK5, CDK9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCell cycle\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePaclitaxel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1080\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMicrotubule stabiliser\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMitosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBortezomib\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1191\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eProteasome\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eProtein stability and degradation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e\n \u003ch2\u003eComparative Analysis of Urothelial cells in MIBC, NMIBC, and Adjacent Normal Tissues\u003c/h2\u003e\n \u003cp\u003eDue to differences in the tumor microenvironment between MIBC and NMIBC, this study conducted differential gene analysis of Urothelial cells in MIBC, NMIBC, and adjacent normal tissues, with a focus on the top 20 genes to identify potential clinical diagnostic and therapeutic targets(Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA).\u003c/p\u003e\n \u003cp\u003eIn the comparison between MIBC and NMIBC, upregulated genes in Urothelial cells included immune-related genes (e.g., LYZ, LEAP2, IGKC, HLA-DRB5, DEFB1, IFI27, IFI6, and IFITM3), metabolism and synthesis-related genes (e.g., ECH1, INSIG1, and MRPS12), and cell structure and function-related genes (e.g., PHGR1, LGALS4, BMP2, and TMEM19). Downregulated genes were involved in energy metabolism (e.g., ATP5MJ, ATP5MC2, ATP5ME, ATP5F1E, and NDUFAF8), chromatin structure and gene expression regulation (e.g., H2AZ1, H3-3A, H3-3B, H4C3, and H1-2), signal transduction and protein degradation (e.g., RACK1 and ELOC), as well as RNA regulation (e.g., SNHG29 and DANCR).In the comparison between MIBC and Para group, upregulated genes mainly included immune-related genes (e.g., LYZ, IFI27, LEAP2, ISG15, and IFI6) and metabolism-related genes (e.g., SULT1E1, SCD, and DHCR24). Downregulated genes were related to cell adhesion (e.g., ITGA2, ITGA6, LMO7), inflammation and immunity (e.g., CCL20, PTGS2), and signaling pathways (e.g., DKK1 and AREG).In the comparison between NMIBC and Para group, upregulated genes were related to immunity and inflammation (e.g., S100A9, PSORS1C2, and IFI27), cellular energy metabolism (e.g., ATP5F1C, ATP6V0C, and ATP5MC1), while downregulated genes were associated with the extracellular matrix (e.g., THBS1), cell migration and tissue lubrication (e.g., HAS3), and signaling pathway regulation (e.g., DKK1).\u003c/p\u003e\n \u003cp\u003eAdditionally, in the differential analysis of Epi9_CSC between the MIBC and NMIBC groups(Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB), upregulated genes were primarily associated with the extracellular matrix and structure (e.g., TFPI2, TGFBI, COL8A1, and SPOCK1), which may reflect an enhancement of MIBC functionality in these aspects.\u0026nbsp;\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec28\" class=\"Section2\"\u003e\n \u003ch2\u003eInference of Epithelial Cell-Cell Communication\u003c/h2\u003e\n \u003cp\u003eTo gain deeper insights into how Urothelial cells in different subpopulations interact with other cells and with themselves through ligands and receptors, we focused on the molecular interactions between these cells. Specifically, Urothelial cells from each subpopulation secrete specific ligands that bind to receptors on target cells, thereby activating various signaling pathways and regulating the functions of epithelial cells. In this study, we used the CellChat tool to identify intercellular communication pathways and explore how Urothelial cells from different subpopulations promote tumor initiation and progression through ligand-receptor interactions.\u003c/p\u003e\n \u003cp\u003eWe first compared the total number and strength of cell-cell communication interactions in Urothelial cells across different pathological states. The results showed that MIBC had the highest number and strongest interactions, followed by NMIBC, with the Para group showing the fewest and weakest interactions (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eA). Moreover, in the MIBC pathological state, Epi9_CSC served as the primary sender and receiver of communication; in contrast, in the NMIBC and Para pathological states, Epi7_Mesenchymal was the major communicator (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB). Since Epi9_CSC plays a central role in tumor development, we further examined the complex intercellular communication pathways involving Epi9_CSC by calculating the strength of outgoing and incoming interactions for each signaling pathway in different pathological states. Notably, in the MIBC pathological state, the main outgoing signaling pathways from Epi9_CSC were collagen and laminin pathways, which were widely received by various epithelial cell subpopulations. Additionally, the primary incoming signaling pathways for Epi9_CSC also included its own collagen and laminin pathways (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eC). In the NMIBC pathological state, the primary outgoing signaling pathways from Epi9_CSC were the CypA and MIF pathways, while the main incoming signaling pathway was collagen from Epi7_Mesenchymal(Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eD). In the Para pathological state, the main outgoing signaling pathways from Epi9_CSC were collagen and APP pathways, with the primary incoming signaling pathway still being collagen from Epi7_Mesenchymal(Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eE). By combining the primary outgoing and incoming pathways for Epi9_CSC in different pathological states, we infer that the collagen pathway is central to the interactions between Epi9_CSC and other cells, as well as its self-interaction.\u003c/p\u003e\n \u003cp\u003eGiven the central role of the collagen pathway in intercellular communication of Epi9_CSC, we further explored the major ligand-receptor pairs involved in this pathway between Epi9_CSC and other epithelial cell subpopulations. The results showed that the strongest communication probabilities in the collagen pathway occurred between the ligands [collagen types I and IV (COL1 and COL4)] and the receptors [syndecans 1 and 4 (SDC1 and SDC4), CD44, and integrins \u0026alpha;2\u0026beta;1 and \u0026alpha;3\u0026beta;1 (ITGA2_ITGB1, ITGA3_ITGB1)](Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eA,B).\u003c/p\u003e\n \u003cp\u003eIn conclusion, these findings suggest that Epi9_CSC primarily mediates specific ligand-receptor interactions via the collagen pathway, regulating cell adhesion, signal transduction, and microenvironment adaptation. This not only enhances cell adhesion properties and signal transmission but also plays a crucial role in tissue structural support, barrier function, and cell fate determination.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec29\" class=\"Section2\"\u003e\n \u003ch2\u003eDevelopment and Validation of CSCRPI\u003c/h2\u003e\n \u003cp\u003eUsing the FindAllMarkers function with the condition avg_log2FC\u0026thinsp;\u0026gt;\u0026thinsp;2, a total of 631 CSC subpopulation marker genes were identified. These genes were then subjected to univariate Cox regression analysis, resulting in the identification of 182 CSC subpopulation marker genes significantly associated with prognosis (See Supplementary Table \u003cspan class=\"InternalRef\"\u003eS13\u003c/span\u003e). Subsequently, four machine learning algorithms\u0026mdash;Random Forest, Lasso, XGBoost, and Decision Tree\u0026mdash;were employed to score and rank these 182 core prognostic genes (see Supplementary Materials Table \u003cspan class=\"InternalRef\"\u003eS14\u003c/span\u003e). Through a weighted averaging method, the top 10 core prognostic genes were determined (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eA). In the TCGA-BLCA training cohort, multivariable Cox regression analysis further identified five key prognostic genes: TUBB6, CERCAM, MAP7D3, SEL1L3, and HSD17B1. Based on the expression profiles and regression coefficients of these genes, the following CSCRPI formula was constructed:CSCRPI\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.1483420\u0026times;TUBB6\u0026thinsp;+\u0026thinsp;0.2376760\u0026times;CERCAM\u0026thinsp;+\u0026thinsp;0.4180103\u0026times;\u003c/p\u003e\n \u003cp\u003eMAP7D3\u0026thinsp;\u0026minus;\u0026thinsp;0.1687422\u0026times;SEL1L3\u0026thinsp;+\u0026thinsp;0.1378119\u0026times;HSD17B1.BLCA patients were categorized into low CSCRPI and high CSCRPI groups based on the median CSCRPI value. In the TCGA-BLCA training cohort, high CSCRPI patients demonstrated significantly poorer prognosis compared to low CSCRPI patients (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eB); this conclusion was consistently observed in the independent validation cohort GSE31684 (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eC).Further analysis in the TCGA-BLCA training cohort led to the development of a nomogram model for predicting 1-year, 3-year, and 5-year overall survival (OS) based on CSCRPI, age, gender, and clinical stage through multivariable Cox and stepwise regression analyses (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eD). Calibration curves demonstrated that the model\u0026apos;s predictions for 1-year, 3-year, and 5-year survival were highly consistent with actual survival rates (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eE). Decision curve analysis (DCA) indicated that the nomogram provided a greater net clinical benefit compared to other clinical variables (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eF).\u003c/p\u003e\n \u003cp\u003eIn the TCGA-BLCA cohort, the area under the curve (AUC) values for the nomogram at 1-year, 3-year, and 5-year were 0.74, 0.71, and 0.72, respectively; in the independent validation cohort GSE31684, the AUC values were 0.77, 0.80, and 0.77, respectively (Fig. \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eG and \u003cspan class=\"InternalRef\"\u003e7\u003c/span\u003eH). The results further revealed that, except for SEL1L3, the other four key prognostic genes were highly expressed in the high CSCRPI group, suggesting they might act as risk factors, while SEL1L3 might serve as a protective factor. This trend was further validated in the independent cohort GSE31684. These findings indicate that CSCRPI not only effectively distinguishes high-risk from low-risk BLCA patient groups but also serves as a reliable tool for predicting short-term and long-term survival, providing precise quantitative guidance for personalized clinical management.\u003c/p\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eClinical Relevance Assessment of CSCRPI\u003c/h3\u003e\n\u003cp\u003eTo further validate the prognostic value of CSCRPI in subgroups with different clinical variables, we performed a stratified analysis. Patients were categorized into subgroups based on age, gender, and pathological stage, followed by survival analysis. The results revealed that CSCRPI levels were significantly elevated in deceased patients, high-risk patients, and those with stage IV disease, indicating a strong association between elevated CSCRPI levels and the progression of BLCA (Fig. \u003cspan class=\"InternalRef\"\u003e8\u003c/span\u003eA-C). Furthermore, Kaplan-Meier survival analysis demonstrated that patients with higher CSCRPI levels had significantly worse prognoses across subgroups with varying clinical characteristics, whereas patients with lower CSCRPI levels exhibited better outcomes (Fig. \u003cspan class=\"InternalRef\"\u003e8\u003c/span\u003eD-I). These findings strongly support the critical role of CSCRPI in predicting the prognosis of BLCA patients.\u003c/p\u003e\n\u003cdiv id=\"Sec31\" class=\"Section2\"\u003e\n \u003ch2\u003eTumor Immune Microenvironment and Evaluation of Immunotherapy Efficacy\u003c/h2\u003e\n \u003cp\u003eTo assess the immune cell infiltration in BLCA (bladder urothelial carcinoma) patients, we used nine different algorithms to calculate immune scores and stromal cell scores. According to the ESTIMATE algorithm, the high CSCRPI group exhibited increased stromal score, immune score, and ESTIMATE score, along with decreased tumor purity (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eA). The two CSCRPI groups showed significant differences in immune infiltration, with the high CSCRPI group exhibiting significantly higher overall immune infiltration than the low CSCRPI group (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eB). Additionally, we found that five characteristic prognostic genes in CSCRPI were highly correlated with tumor-infiltrating immune cells, including TUBB6, CERCAM, and MAP7D3, which were positively correlated with most tumor-infiltrating immune cells, while SEL1L3 and HSD17B1 were generally negatively correlated with tumor-infiltrating immune cells (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eC). Immune checkpoint inhibitors (ICIs) have brought benefits to patients in clinical immunotherapy for various malignancies. TIDE (Tumor Immune Dysfunction and Exclusion) analysis showed that BLCA patients with high CSCRPI had increased TIDE scores and T-cell rejection scores, while there was no difference in microsatellite instability (MSI) score and T-cell dysfunction score between the two groups, indicating that BLCA patients with high CSCRPI are more likely to undergo immune escape, and thus the potential efficacy of ICI therapy may be poor (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eD). Similarly, the IPS (Immunogenicity Prediction Score) results showed that BLCA patients with high CSCRPI would benefit less from immune checkpoint inhibitor therapy (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eE). Using the ssGSEA algorithm, we obtained the scores for immune-related pathways. The high CSCRPI group showed enhanced activity in many immune-related pathways (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eF), suggesting an overactivation of the immune response state or enhanced tumor-driven immune escape mechanisms in the high CSCRPI group. The study also observed that, except for the HSD17B1 gene, the remaining four characteristic prognostic genes in CSCRPI were positively correlated with most immune checkpoints (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eG). The expression of these immune checkpoint genes was significantly higher in the high CSCRPI group (Fig. \u003cspan class=\"InternalRef\"\u003e9\u003c/span\u003eH).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec32\" class=\"Section2\"\u003e\n \u003ch2\u003eDrug Sensitivity Analysis of High and Low CSCRPI Groups\u003c/h2\u003e\n \u003cp\u003eTo further explore the clinical significance of CSC characteristic prognostic genes in BLCA for precision therapy, we comprehensively evaluated the sensitivity of patients in both CSCRPI groups to target drugs (Daporinad, Vinblastine, Vinorelbine, Eg5_9814, Paclitaxel, Sepantronium bromide, Staurosporine, Dinaciclib) using the GDSC database combined with the results from our previous cancer stem cell subgroup drug sensitivity analysis. Based on IC50 results, we found that patients in both CSCRPI groups were sensitive to these drugs, which was consistent with our previous results. Moreover, for most of these drugs, BLCA patients in the low CSCRPI group showed greater sensitivity than those in the high CSCRPI group, indicating that the CSCRPI score could serve as a guidance for chemotherapy in BLCA patients (Fig. \u003cspan class=\"InternalRef\"\u003e10\u003c/span\u003eA). To better explore the association between the expression profiles of the 5 characteristic prognostic genes and drug sensitivity, we performed analysis using the CellMiner database. We generated correlation histograms for selected genes with selected target drugs. Positive correlation indicates that increased gene expression is associated with increased drug sensitivity, while negative correlation indicates that increased gene expression confers drug resistance. Notably, upregulation of TUBB6, CERCAM, and HSD17B1 enhanced resistance to some drugs, while upregulation of MAP7D3 enhanced sensitivity to Staurosporine (Fig. \u003cspan class=\"InternalRef\"\u003e10\u003c/span\u003eB).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eEpithelial cells play a crucial role in tumor initiation and progression. A comprehensive and systematic understanding of the characteristics and functions of epithelial cells within the tumor microenvironment will help reveal the molecular mechanisms underlying tumor progression and drive the development of novel therapeutic strategies. scRNA-seq can uncover the compositional features, heterogeneity, dynamic changes, and key roles of tumor-associated epithelial cells in tumor initiation and progression, providing important insights for a deeper understanding of the tumor microenvironment.\u003c/p\u003e\u003cp\u003e\u003cb\u003eEpi9_CSC Promotes High Recurrence and Invasion in MIBC, with Paracancerous Presence Suggesting Its Origin from Normal Tissue or Its Role in Metastasis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe results of this study indicate the presence of various epithelial cell subtypes in BLCA, with distinct distributions of these subtypes in the MIBC, NMIBC, and Para groups. Additionally, we identified a critical subpopulation of CSCs.CSC characteristics include self-renewal, differentiation potential, and the ability to promote tumor recurrence and metastasis. Self-renewal enables CSCs to maintain their population and drive tumor growth through symmetric or asymmetric division, while their differentiation potential allows them to generate other types of tumor cells, sustaining the tumor's high heterogeneity. Furthermore, CSCs possess strong migratory and invasive abilities, enabling them to induce metastatic tumor formation in new microenvironments, thereby worsening disease progression and complicating treatment.\u003c/p\u003e\u003cp\u003eOur findings show that there are different proportions of CSCs in tumor and Paracancerous tissues, with the highest proportion of CSCs observed in MIBC. This feature may be a key factor contributing to the high recurrence and incidence rates of MIBC in clinical practice. Recent studies have further revealed the critical role of CSCs in the initiation and progression of MIBC.Moreover, we observed the presence of CSCs in Paracancerous tissues, suggesting that CSCs may originate from normal bladder tissue, or that CSCs play a pivotal role in driving cancer metastasis.\u003c/p\u003e\u003cdiv id=\"Sec34\" class=\"Section2\"\u003e\u003ch2\u003eEpi9_CSC Promotes Tumor Cell Survival, Invasion, and Self-Renewal through ECM Activation of Downstream Pathways\u003c/h2\u003e\u003cp\u003eKEGG enrichment analysis of Epi9_CSC reveals that pathways such as focal adhesion, ECM-receptor interaction, proteoglycans in cancer, and the PI3K-Akt signaling pathway play complex, synergistic roles in the initiation and progression of BLCA by CSCs.The ECM consists of collagen, laminin, fibronectin, and proteoglycans, which provide structural integrity and strength to tissues. Collagen contributes to tissue structure and strength, laminin affects cell adhesion and differentiation, and fibronectin regulates cell migration. The ECM serves as an important bridge between cells and their external environment, influencing cell adhesion, migration, and signal transduction through interactions with cell surface receptors such as integrins and CD44.\u003c/p\u003e\u003cp\u003eIn the ECM-receptor interaction pathway map from KEGG Mapper, our results show that ECM components, including collagen, laminin, and fibronectin, are highly expressed. These ECM components activate several downstream pathways (focal adhesion, proteoglycans in cancer, and the PI3K-Akt signaling pathway) by binding to integrin receptors or CD44 on the cell membrane.Focal adhesion, as a critical point of cell-ECM connection, plays an important role in this process. Through binding to ECM molecules, focal adhesion can activate the regulation of the actin cytoskeleton, NF-kB, and PI3K-Akt signaling pathways. Notably, the PI3K-Akt pathway controls cell proliferation, survival, and metabolism, and its abnormal activation can drive tumor progression. Our results show that ECM, through integrin receptor binding at focal adhesions, activates the NF-kB and PI3K-Akt signaling pathways, enhancing tumor cell proliferation and survival. Additionally, ECM, through integrin receptor binding at focal adhesions, can also activate the regulation of the actin cytoskeleton, promoting cytoskeletal reorganization and enhancing tumor cell motility and migration.Furthermore, our results suggest that ECM may activate the MAPK signaling pathway by binding to CD44 receptors on the cell membrane, aiding in the self-renewal of CSCs.On the other hand, intercellular communication results for Epi9_CSC indicate that Epi9_CSC primarily mediates communication with itself and other epithelial cells through collagen ligands in the collagen pathway, binding to integrin receptors, which is consistent with our KEGG findings.\u003c/p\u003e\u003cp\u003eIn summary, during the EMT process, Epi9_CSC secretes collagen that binds to integrins and CD44 receptors on the cell membranes of both Epi9_CSC itself and other epithelial cells, thereby activating downstream pathways including focal adhesion, PI3K-Akt signaling, regulation of the actin cytoskeleton, NF-kB, proteoglycans in cancer, and MAPK. The activation of these signaling pathways collectively enhances tumor (stem) cell proliferation and survival, increases their motility and invasiveness, and promotes their self-renewal.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eMolecular Mechanisms of Epi9_CSC Differentiation into Mesenchymal Subtypes\u003c/h3\u003e\n\u003cp\u003eDuring the differentiation of Epi9_CSC into mesenchymal subtypes, genes that are downregulated include NT5E (CD73), SERPINA1, TGFBI, LAMC2, DCBLD2, CMTM3, and COL7A1.NT5E (encoding CD73) is an important gene involved in immune regulation within the tumor microenvironment by converting extracellular AMP into adenosine. Adenosine plays an inhibitory role in immune escape by reducing the activity of T cells and NK cells, thereby promoting tumor immune evasion. Downregulation of CD73 during the differentiation of CSCs into mesenchymal subtypes may reduce adenosine production, weakening the immunosuppressive effect, and reflecting the transition from a stem cell state to an invasive mesenchymal state.SERPINA1 (encoding α1-antitrypsin) is a serine protease inhibitor involved in inflammation and the stability of the ECM. Downregulation of SERPINA1 may reduce protease inhibition, increasing ECM degradation and promoting CSC migration and invasion, which is consistent with the high invasiveness of mesenchymal subtypes.TGFBI (TGF-β induced gene) encodes an ECM protein closely related to the TGF-β signaling pathway, a major inducer of EMT. Downregulation of TGFBI may affect TGF-β-mediated EMT, indicating dynamic changes in the regulation of the TGF-β signaling pathway during CSC differentiation into mesenchymal subtypes.\u003c/p\u003e\u003cp\u003eLAMC2 encodes the γ2 chain of laminin, which is a component of laminin-332 involved in cell adhesion to the basement membrane. Downregulation of LAMC2 may disrupt cell adhesion to the ECM, promoting cell detachment and migration, a feature particularly evident in mesenchymal subtype CSCs. Huang et al. (2020) noted that downregulation of LAMC2 (γ2 chain) promotes EMT in pancreatic ductal adenocarcinoma (PDAC) cells, further enhancing tumor invasion and metastasis.DCBLD2 (Discoidin, CUB, and LCCL domain-containing protein 2) is closely associated with its activation in EMT, supporting the invasiveness and metastatic potential of CSCs.CMTM3 (CKLF-like MARVEL transmembrane domain-containing 3) is involved in cell membrane structure and function and may affect cell polarity or signaling. Downregulation of CMTM3 may result in the loss of cell polarity, which is a hallmark of EMT and mesenchymal transformation. Yuan et al. (2016) reported that downregulation of CMTM3 in gastric cancer is associated with tumor invasion and metastasis.COL7A1 encodes type VII collagen, an essential component of the basement membrane that maintains epithelial tissue integrity. Downregulation of COL7A1 may disrupt the integrity of the basement membrane, promoting cell invasion and migration, which is consistent with the high invasiveness of mesenchymal subtype CSCs. Guerra et al. (2017) found that downregulation of COL7A1 leads to basement membrane integrity disruption, providing a less resistant barrier for tumor cells to infiltrate, thus promoting tumor expansion.\u003c/p\u003e\u003cp\u003eDuring the differentiation of Epi9_CSC into mesenchymal subtypes, the expression of COL1A2, COL6A2, and COL3A1 is upregulated. These genes encode different types of collagen and are major components of the ECM. Their upregulation suggests that the cells are remodeling their extracellular environment, increasing the rigidity and structure of the matrix to support cell migration and invasion. This matrix remodeling is often associated with tumor fibrosis and sclerosis, which aids in tumor progression.By analyzing these gene expression changes in detail, we can gain a deeper understanding of the molecular mechanisms involved in the transition of BLCA stem cells to a more invasive phenotype, providing new perspectives for developing more effective therapeutic strategies.\u003c/p\u003e\n\u003ch3\u003eDinaciclib as Potential Sensitive Therapeutic Agents for Epi9_CSC\u003c/h3\u003e\n\u003cp\u003eThe drug sensitivity analysis results indicate that Epi9_CSC exhibits significant sensitivity to several chemotherapeutic agents, including Bortezomib, Dactinomycin, Docetaxel, Daporinad, Sepantronium bromide, Vinblastine, Eg5_9814, Vinorelbine, Staurosporine, Dinaciclib, and Paclitaxel.In the expression analysis of potential drug target genes in Epi9_CSC, we observed high expression levels of members of the cyclin-dependent kinase (CDK) family (CDK5, CDK9) within CSCs. These CDKs may play crucial roles in regulating cell cycle processes, influencing tumor cell proliferation and survival, and participating in cell signaling and gene expression modulation mechanisms that contribute to the self-renewal and drug resistance of CSCs [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eDinaciclib, a small-molecule inhibitor, exerts its antitumor effects predominantly by targeting CDKs, particularly CDK1, CDK2, CDK5, and CDK9. By inhibiting CDK9, Dinaciclib affects the functionality of the transcription elongation factor P-TEFb, leading to a reduction in the phosphorylation of RNA polymerase II. This ultimately decreases mRNA synthesis and inhibits the expression of short-lived anti-apoptotic proteins such as MCL-1, thereby resulting in tumor cell apoptosis and cell cycle arrest [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].In a study by Masahiro Shimizu et al., it was found that the oncogenic RAS signaling pathway enhances the activity of CDK1, promoting the generation of CSCs. This process involves an increase in the expression and activity of CSC-related factors, such as SOX2. However, Dinaciclib, as a CDK1 inhibitor, effectively disrupts this mechanism. By inhibiting CDK1 activity, Dinaciclib impairs the proliferation and self-renewal capabilities of CSCs, leading to a reduction in their population [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. This mechanism underscores the significant role of CDKs in the formation and maintenance of CSCs.\u003c/p\u003e\u003cp\u003eGiven our drug sensitivity analysis, the sensitivity of Epi9_CSC to Dinaciclib suggests that CDK inhibitors may suppress the self-renewal of BLCA CSCs by targeting the highly expressed CDK family members (such as CDK5 and CDK9) within these cells. Therefore, we speculate that Dinaciclib represents a potential therapeutic agent for targeting BLCA CSC.\u003c/p\u003e\u003cdiv id=\"Sec37\" class=\"Section2\"\u003e\u003ch2\u003eDifferential Marker Genes in BLCA Tissues Under Different Pathological Conditions\u003c/h2\u003e\u003cp\u003eIn the comparative analysis of MIBC and NMIBC, upregulated genes in Urothelial cells include immune-related genes (such as LYZ, LEAP2, IGKC, HLA-DRB5, DEFB1, IFI27, IFI6, and IFITM3), indicating enhanced innate immunity, antimicrobial defense, active adaptive immune responses, and an enhanced antiviral response. Additionally, metabolism-related genes (such as ECH1, INSIG1, and MRPS12) reflect tumor cell metabolic reprogramming, supporting rapid proliferation through altered fatty acid oxidation and cholesterol metabolism [\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Genes involved in cell structure and function (such as PHGR1, LGALS4, BMP2, and TMEM19) promote cell adhesion, signal transduction, and tissue development, potentially enhancing the invasiveness and migratory capacity of tumor cells [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. In contrast, downregulated energy metabolism-related genes (such as ATP5MJ, ATP5MC2, ATP5ME, ATP5F1E, and NDUFAF8) suggest a reliance on glycolysis due to the inhibition of mitochondrial oxidative phosphorylation, which aids cancer cell survival under hypoxic conditions [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e]. Downregulation of chromatin-related genes (such as H2AZ1, H3-3A, H3-3B, H4C3, and H1-2) may reduce the open chromatin state, affecting gene expression and promoting genetic instability [\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. Furthermore, the downregulation of genes related to protein degradation (e.g., RACK1 and ELOC) and RNA regulation (e.g., SNHG29 and DANCR) may disrupt critical signaling and regulatory networks, enhancing tumor invasiveness and uncontrolled growth [\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eWhen comparing MIBC with the Para group, the upregulation of immune-related genes (such as LYZ, IFI27, and LEAP2) suggests a strong immune response, alongside enhanced antimicrobial and antiviral defenses. Additionally, the upregulation of metabolism-related genes (SULT1E1, SCD, and DHCR24) indicates metabolic reprogramming to support rapid tumor growth [\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. These changes highlight the adaptive strategies of MIBC cells to evade immune surveillance, adjust metabolism, and promote invasiveness. On the other hand, the Para group shows no such upregulation, maintaining strong cell adhesion, cytoskeletal stability, and immune responses, which help limit tumor progression. The downregulation of adhesion-related genes (ITGA2, ITGA6, LMO7) and immune evasion-related genes (CCL20 and PTGS2) in MIBC further promotes invasiveness and metastasis [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. The expression of signaling pathway-related genes (e.g., DKK1 and AREG) supports the higher invasiveness in MIBC, while the Para group shows lower malignancy traits, reinforcing its protective role against tumor development.\u003c/p\u003e\u003cp\u003eIn the comparison between NMIBC and Para, upregulated genes like S100A9, PSORS1C2, and IFI27 indicate alterations in tumor metabolism and microenvironment, promoting tumor initiation and progression [\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e]. The upregulation of ATP5F1C, ATP6V0C, and ATP5MC1 suggests active oxidative phosphorylation, fueling tumor cell proliferation [\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. Conversely, downregulation of genes such as THBS1 and HAS3 in NMIBC impairs extracellular matrix remodeling and cell migration, while the downregulation of DKK1 may enhance Wnt signaling, promoting cell proliferation and tumor progression [\u003cspan additionalcitationids=\"CR56\" citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e]. These molecular alterations in NMIBC highlight significant changes in metabolism, extracellular matrix remodeling, and immune regulation, offering insights into potential therapeutic targets.\u003c/p\u003e\u003cp\u003eFinally, the differential analysis of Epi9_CSC in MIBC and NMIBC reveals distinct gene expression patterns. MIBC tumor stem cells show upregulation of genes like TFPI2, TGFBI, and COL8A1, which are involved in extracellular matrix remodeling and intercellular interactions, likely contributing to their enhanced migratory and invasive capabilities [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. The upregulation of SPOCK1 further supports matrix remodeling, suggesting that MIBC stem cells exhibit stronger invasiveness and migration ability compared to NMIBC. These findings underscore the critical role of extracellular matrix dynamics and cellular interactions in MIBC progression, enabling tumor cells to breach the basement membrane and invade surrounding tissues.\u003c/p\u003e\u003cp\u003eIn conclusion, MIBC demonstrates enhanced invasiveness, immune evasion, and metabolic reprogramming compared to NMIBC and Para group. These adaptive features involve immune enhancement, metabolic reprogramming, and extracellular matrix remodeling. NMIBC, in contrast, exhibits metabolic activity and microenvironment regulation with relatively weaker invasiveness, while the Para group maintains strong cell adhesion and immune responses, limiting malignant progression. Additionally, MIBC tumor stem cells exhibit enhanced invasiveness and migration, likely due to extracellular matrix remodeling and intercellular interactions.\u003c/p\u003e\u003cdiv id=\"Sec38\" class=\"Section3\"\u003e\u003ch2\u003eClinical Translation Value and Multidimensional Functional Analysis of the CSCRPI Prognostic Risk Assessment Model\u003c/h2\u003e\u003cp\u003eBased on the identified molecular characteristics of the Epi9_CSC subpopulation, we successfully established the CSCRPI prognostic risk assessment model. This model demonstrates good correlation with clinicopathological parameters and exhibits excellent performance in predicting patient survival prognosis. More importantly, we conducted systematic analysis of patients in high and low CSCRPI risk groups, including clinical feature correlation assessment, tumor immune microenvironment status analysis, immunotherapy response potential evaluation, and drug sensitivity profile analysis, establishing a solid theoretical foundation for comprehensively elucidating the biological behavioral patterns of the Epi9_CSC subpopulation and its clinical translation value.\u003c/p\u003e\u003cp\u003eSpecifically, among the 5 characteristic genes constituting CSCRPI, CERCAM, MAP7D3, and HSD17B1 have positive coefficients, indicating that high expression of these genes leads to higher CSCRPI scores, resulting in poor prognosis in the high CSCRPI group, which is consistent with their mechanisms of action in cancer. CERCAM is a cell adhesion molecule primarily localized to the cell membrane, playing a role in intercellular interactions and cell-matrix binding. Research has found that CERCAM is abnormally highly expressed in BLCA tissues and cells, and its overexpression can significantly promote BLCA cell proliferation, DNA synthesis, and invasive capacity, and induce EMT, manifesting as upregulation of PCNA, vimentin, Twist, and N-cadherin, while inhibiting E-cadherin and cleaved caspase-3 expression; in vivo, CERCAM silencing can inhibit xenograft tumor growth in nude mice[\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e]. MAP7D3 is a microtubule-associated protein involved in microtubule assembly and stability regulation. Studies have shown that MAP7D3 enhances cancer stem cell characteristics, increases TNBC cell resistance to chemotherapy, and thereby promotes tumor metastasis and progression. Specifically, high expression of MAP7D3 promotes cell attachment and detachment from the matrix by upregulating extracellular matrix remodeling-related proteins such as integrin α6, ABCG2, and ALDH1A1, thereby increasing cell migration and invasion capacity in the matrix[\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e]. HSD17B1 is an important steroid metabolic enzyme that plays a key role in various cancers. Research shows it regulates active estrogen biosynthesis and promotes breast cancer cell proliferation[\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e]. Although the molecular mechanisms of HSD17B1 in BLCA pathogenesis require further elucidation, loss-of-function experiments have shown that knocking down HSD17B1 can significantly reduce migration and invasion activity of BLCA cells[\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e]. Therefore, all three genes above have effects that lead to poor BLCA prognosis. Particularly noteworthy is that two of these genes are related to EMT, which is consistent with our previous pathogenic mechanism analysis results for the Epi9_CSC subpopulation. Among the other two genes, TUBB6 is an important component of the cytoskeleton that forms microtubule structures by binding with α-tubulin, participating in maintaining cell morphology and cell division processes in normal cells. In bladder urothelial carcinoma, TUBB6 expression is abnormally upregulated, promoting tumor cell motility and invasive capacity by remodeling cytoskeletal structure. Additionally, TUBB6 may also promote local tumor invasion and distant metastasis by regulating the distribution and function of cell adhesion molecules, weakening intercellular connections. Studies have shown that inhibiting TUBB6 expression can significantly reduce migration and invasion activity of BLCA cells, suggesting its value as a potential therapeutic target[\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e]. SEL1L3 is an endoplasmic reticulum-localized transmembrane protein that mainly participates in endoplasmic reticulum-associated protein degradation pathways, playing an important role in maintaining intracellular protein homeostasis and regulating stress responses. However, current research on SEL1L3 is relatively limited, and its specific molecular mechanisms in tumor development and progression remain unclear. Based on its negative coefficient characteristics and high expression in the low-risk group, we speculate that SEL1L3 may function as a tumor suppressor with protective effects. Additionally, TUBB6 shows a negative coefficient in the prognostic model, which contradicts its biological mechanism of promoting cell motility and invasion metastasis. This may stem from disease stage-dependent effects or the influence of complex interactions between multiple genes.\u003c/p\u003e\u003cp\u003eIn the tumor immune microenvironment and immunotherapy efficacy assessment analysis, this study revealed the complex characteristics of the immune microenvironment in high-risk CSCRPI BLCA patients through multidimensional immune analysis. ESTIMATE scores showed that high-risk patients had significantly elevated stromal scores, immune scores, and composite scores, accompanied by reduced tumor purity. This phenomenon initially seemed contradictory to poor prognosis. However, in-depth immune cell subpopulation analysis provided a reasonable explanation for this paradox: the high-risk group showed significantly elevated scores across 24 immune cell types, including anti-tumor effector cells such as activated B cells, effector memory T cells, and NK cells, as well as immunosuppressive cells such as regulatory T cells, MDSCs, and immature dendritic cells. This \"pan-lineage\" increase in immune cell infiltration reflects a highly activated state of the tumor immune microenvironment, but the simultaneous increase in effector and suppressor cells suggests profound immune dysfunction. TIDE and IPS analyses further elucidated the molecular basis of poor immunotherapy responsiveness in high-risk patients. The significant increase in TIDE scores was mainly driven by T cell exclusion scores rather than T cell dysfunction, suggesting that high-risk tumors primarily achieve immune escape by preventing effector T cells from approaching tumors rather than inducing T cell exhaustion. Correspondingly, MSI scores showed no significant difference between groups, indicating that immune therapy resistance in high-risk patients does not stem from insufficient tumor mutational burden[\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e]. IPS analysis results were highly consistent with TIDE predictions, showing that high-risk patients derive lower benefit from immune checkpoint inhibitor therapy, collectively confirming that T cell exclusion-type immune escape is a key mechanism leading to immunotherapy failure. At the molecular level, the high-risk group showed systemic activation of immunosuppressive signaling pathways, with significantly enhanced activity of HLA, CCR, T cell co-inhibitory, and immune checkpoint pathways, while T cell co-stimulatory pathway activity showed no significant change, forming an \"inhibitory signal-dominant\" imbalanced immune regulation pattern[\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e]. More importantly, 34 key immune checkpoint genes were generally upregulated in the high-risk group, including the classical PD-1/PD-L1 axis (PDCD1, CD274, PDCD1LG2), CTLA-4, TIM-3 (HAVCR2), and multiple TNF/TNFR family members, constructing a multi-layered, redundant immunosuppressive network. This \"checkpoint storm\" phenomenon may be an important reason for the limited efficacy of single-target immunotherapy[\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe association between the 5 characteristic genes in CSCRPI and the immune microenvironment presents interesting differentiation patterns. TUBB6, CERCAM, and MAP7D3 are positively correlated with most tumor-infiltrating immune cells and positively correlated with immune checkpoint expression, suggesting these genes may activate immunosuppressive mechanisms while promoting immune cell recruitment, forming a \"recruitment-suppression\" vicious cycle. Conversely, SEL1L3 and HSD17B1 are negatively correlated with immune cell infiltration and may participate in immune regulation through different mechanisms. Particularly noteworthy is that TUBB6 shows a negative coefficient in the prognostic model, which presents an apparent contradiction with its known function of promoting cell motility and invasion metastasis. This phenomenon may reflect the functional transition of TUBB6 at different disease stages or composite effects under multi-gene network interactions, requiring further mechanistic studies for clarification. Our research results have important guiding significance for precision immunotherapy in BLCA. Although high CSCRPI patients have abundant immune cell infiltration, they present a typical \"immune desert\" functional state, suggesting that traditional single immune checkpoint inhibitor therapy may have limited efficacy[\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e]. Based on the T cell exclusion-dominant immune escape mechanism, these patients may be more suitable for combination therapy strategies: immune checkpoint inhibitors combined with anti-angiogenic therapy to promote T cell infiltration through vascular normalization[\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e, \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e]; CAR-T or TCR-T cell therapy to directly deliver effector cells bypassing T cell exclusion mechanisms[\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eOur drug sensitivity analysis results indicate that CSCRPI scores have important clinical value in predicting chemotherapy responses in BLCA patients. Patients in the low CSCRPI group showed higher sensitivity to most targeted drugs, which is consistent with previous research results showing that tumor stem cell characteristics are closely related to chemotherapy resistance. Previous studies have confirmed that enhancement of tumor stem cell-like characteristics is usually accompanied by increased multidrug resistance, primarily achieved through activating DNA repair mechanisms, upregulating drug efflux pumps, and inhibiting cell apoptosis[\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e]. The association between our identified characteristic genes and drug sensitivity has important significance. TUBB6, as a member of the tubulin family, has been confirmed to be associated with Paclitaxel resistance through its overexpression, possibly affecting drug efficacy by altering microtubule dynamics[\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e]. HSD17B1, as a key enzyme in steroid hormone metabolism, is abnormally expressed in various cancers. Research shows that its expression level is closely related to tumor cell proliferation and survival capacity, possibly affecting chemotherapy sensitivity by regulating cell cycle and apoptosis processes[\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn summary, the CSCRPI scoring system constructed based on the Epi9_CSC subpopulation integrates expression information from 5 key characteristic genes and can not only accurately predict prognostic risk in BLCA patients but, more importantly, provides multidimensional guidance for clinical treatment decisions: in immunotherapy, this system reveals T cell exclusion-dominant immune escape mechanisms in high-risk patients, providing theoretical basis for formulating combination therapy strategies; in chemotherapy selection, CSCRPI scores can effectively predict patient drug sensitivity, helping achieve precise formulation of individualized chemotherapy regimens. The establishment of this multifunctional prognostic assessment tool lays a solid foundation for precision medicine practice in BLCA.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eOverall, we identified 11 subtypes of Urothelial cells, each exhibiting varying degrees of copy number variation. Among these Urothelial cells, MIBC demonstrates greater invasiveness, immune evasion, and metabolic adaptability compared to NMIBC and Para groups. These features are associated with immune enhancement, metabolic reprogramming, and extracellular matrix remodeling. NMIBC is metabolically active and significantly regulates the microenvironment, but it has weaker invasiveness. In contrast, the Para group limits malignant progression by enhancing cell adhesion and immune response.\u003c/p\u003e\u003cp\u003eWe also identified a crucial population of cancer stem cells (Epi9_CSC), which secrete collagen during the EMT process. This collagen interacts with integrins and CD44 receptors on the cell membranes of Epi9_CSCs and other epithelial cells, activating downstream pathways such as Focal adhesion, PI3K-Akt signaling pathway, Regulation of actin cytoskeleton, NF-kB, Proteoglycans in cancer, and MAPK. The activation of these signaling pathways collectively enhances tumor (stem) cell proliferation and survival, boosts their motility and invasiveness, and promotes self-renewal (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCompared to NMIBC, MIBC tumor stem cells exhibit stronger invasiveness and migration abilities. Based on the expression profiles of potential sensitive drug targets for Epi9_CSC, Dinaciclib are identified as promising therapeutic agents for BLCA CSC.\u003c/p\u003e\u003cp\u003eThe CSCRPI scoring system, based on the Epi9_CSC subpopulation and integrating expression information from 5 key characteristic genes, not only accurately predicts prognostic risk in BLCA patients but also provides comprehensive guidance value for clinical treatment decision-making.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eGEO\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eGene Expression Omnibus\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eDEGs\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eDifferentially Expressed Genes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003escRNA-seq\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eSingle-cell RNA Sequencing\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eRNA-seq\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eRNA Sequencing\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eKEGG\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eKyoto Encyclopedia of Genes and Genomes\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCSC\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eCancer Stem Cells\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eEpi\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eEpithelial Cells\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCNV\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eCopy Number Variation\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eECM\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eExtracellular Matrix\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eTME\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eTumor Microenvironment\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eEMT\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eEpithelial-Mesenchymal Transition\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv class=\"DefinitionListEntry\"\u003e\u003cdiv class=\"Term\"\u003eCSCRPI\u003c/div\u003e\u003cdiv class=\"Description\"\u003e\u003cp\u003eCancer Stem Cell-Related Prognostic Index\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eI sincerely thank my advisors for their guidance and support. I am also grateful to my team members for their invaluable assistance. Your contributions have been essential to this project.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026rsquo; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSupervision, funding acquisition: Xianliang Hou, Songbai Liao. Data collection: Biao Zhang, Yi Liu. Fei Yang. Formal analysis: Biao Zhang, Man Yang, Yu Pan. Methodology: Chunhong Li and Chune Mo. Writing\u0026ndash;original draft: Biao Zhang. Writing\u0026ndash;review and editing: Xianliang Hou,Yingpin Lei, and Jiahua Hu. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by grants from the Guangxi Natural Science Foundation (2024GXNSFAA010096), Guangdong Basic and Applied Basic Research Foundation (2025A1515012661), Guilin Science Research and Technology Development Project (20220139-13-2), Guangdong Province Medical Science and Technology Research Fundation (B2025207), Guangxi Medical and Health Appropriate Technology Development and Promotion Project (S2024075), Innovation Training Program for College Students (202410601013, S202410601182), Guangxi Key Laboratory of Tumor Immunology and Microenvironmental Regulation (2023KF006, 3030302213).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003escRNA-seq datasets for BLCA were obtained from GEO (GSE192575, GSE145137, GSE135337, GSE222315, GSE129845). Drug sensitivity data came from the GDSC2 database (cancerrxgene.org). Copy number variation analysis used the hg38_gencode_v27 annotation file from the Broad Institute (broadinstitute.org). These datasets are accessible following their respective usage policies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe patient sample data used in this study were approved by the Ethics Committees of all participating hospitals, including SYSMH, GDPPH, STCH, ZJH, SMUTH, SYUTH, and NFH. Written informed consent was obtained from all participants at the time of initial sample collection for analyses including scRNA-seq. The bioinformatics analysis conducted in this study utilized only de-identified secondary data analysis and did not involve new human sample collection or direct patient interaction. Therefore, no additional ethical approval was required for this analysis. We are committed to adhering to data use agreements and relevant laws and regulations to ensure the privacy of participants is protected. All analyses were conducted in accordance with the Declaration of Helsinki and the Terms of Use of each public database.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBurger M, Catto JW, Dalbagni G, Grossman HB, Herr H, Karakiewicz P, Kassouf W, Kiemeney LA, La VC, Shariat S, Lotan Y. Epidemiology and risk factors of urothelial BLCA. Eur Urol. 2013;63(2):234\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhmadi H, Duddalwar V, Daneshmand S. Diagnosis and Staging of BLCA. Hematol Oncol Clin NorthAm. 2021;35(3):531\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024 May-Jun;74(3):229\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang Q, Zi H, Luo L, Li X, Zhu C, Zeng X. Secular trends of morbidity and mortality of prostate, bladder, and kidney cancers in China, 1990 to 2019 and their predictions to 2030. BMC Cancer. 2022;22(1):1164.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRichters A, Aben KKH, Kiemeney LALM. The global burden of urinary BLCA: an update. World JUrol. 2020;38(8):1895\u0026ndash;904.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChan ES, Yip SK, Hou SM, Cheung HY, Lee WM, Ng CF. Age, tumour stage, and preoperative serum albumin level are independent predictors of mortality after radical cystectomy for treatment of BLCA in HongKong Chinese. Hong Kong Med J. 2013;19(5):400\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLee YC, Lam HM, Rosser C, Theodorescu D, Parks WC, Chan KS. The dynamic roles of the bladder tumourmicroenvironment. Nat Rev Urol. 2022;19(9):515\u0026ndash;33.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eArneth B. Tumor Microenvironment. Med (Kaunas). 2019;56(1):15.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMarusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity: a looking glass for cancer? Nat Rev Cancer. 2012;12(5):323\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDagogo-Jack I, Shaw AT. Tumour heterogeneity and resistance to cancer therapies. Nat Rev Clin Oncol. 2018;15(2):81\u0026ndash;94.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLambert AW, Weinberg RA. Linking EMT programmes to normal and neoplastic epithelial stem cells. Nat Rev Cancer. 2021;21(5):325\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMantovani A, Marchesi F, Malesci A, Laghi L, Allavena P. Tumour-associated macrophages as treatment targets in oncology. Nat Reviews Clin Oncol. 2017;14(7):399\u0026ndash;416.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNishikawa H, Sakaguchi S. Regulatory T cells in tumor immunity. Int J Cancer. 2010;127(4):759\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoon C, Lu J, Jun Y, et al. KRAS activation in gastric cancer stem-like cells promotes tumor angiogenesis and metastasis [retracted in. BMC Cancer. 2024;24(1):392.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZeng Z, Fu M, Hu Y, Wei Y, Wei X, Luo M. Regulation and signaling pathways in cancer stem cells: implications for targeted therapy for cancer. Mol Cancer. 2023;22(1):172. Published 2023 Oct 18.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu Z, Li H, Dang Q, Weng S, Duo M, Lv J, Han X. Integrative insights and clinical applications of single-cell sequencing in cancer immunotherapy. Cell Mol Life Sci. 2022;79(11):577.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTan TZ, Rouanne M, Tan KT, Huang RY, Thiery JP. Molecular Subtypes of Urothelial BLCA: Results from a Meta-cohort Analysis of 2411 Tumors. Eur Urol. 2019;75(3):423\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHamid ARAH, Syadza YZ, Yausep OE, Christanto RBI, Muharia BHR, Mochtar CA. The expression of stem cells markers and its effects on the propensity for recurrence and metastasis in BLCA: A systematic review. PLoS ONE. 2023;18(5):e0269214. Published 2023 May 17.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGibbons DL, Creighton CJ. Pan-cancer survey of epithelial-mesenchymal transition markers across the Cancer Genome Atlas. Dev Dyn. 2018;247(3):555\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTanabe S. Epithelial-Mesenchymal Transition and Cancer Stem Cells. Adv Exp Med Biol. 2022;1393:1\u0026ndash;49.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBatlle E, Clevers H. Cancer stem cells revisited. Nat Med. 2017;23(10):1124\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eThiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal transitions in development and disease. Cell. 2009;139(5):871\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, Brooks M, Reinhard F, Zhang CC, Shipitsin M, Campbell LL, Polyak K, Brisken C, Yang J, Weinberg RA. The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell. 2008;133(4):704\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEl-Tanani M, Rabbani SA, Satyam SM, Rangraze IR, Wali AF, El-Tanani Y, Aljabali AAA. Deciphering the Role of Cancer Stem Cells: Drivers of Tumor Evolution, Therapeutic Resistance, and Precision Medicine Strategies. Cancers (Basel). 2025;17(3):382.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShin K, Lim A, Odegaard JI, Honeycutt JD, Kawano S, Hsieh MH, Beachy PA. Cellular origin of bladder neoplasia and tissue dynamics of its progression to invasive carcinoma. Nat Cell Biol. 2014;16(5):469\u0026ndash;78.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOhishi T, Koga F, Migita TBLCA. Stem-Like Cells: Their Origin and Therapeutic Perspectives. Int J Mol Sci. 2016;17:43.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLawson DA, Bhakta NR, Kessenbrock K, Prummel KD, Yu Y, Takai K et al. Single-cell analysis reveals a stem-cell program in human metastatic breast cancer cellsNature., 526 (7571) (2015), pp. 131\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHynes RO. The extracellular matrix: not just pretty fibrils. Science. 2009;326(5957):1216\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFrantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J Cell Sci. 2010;123(Pt 24):4195\u0026ndash;200.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVivanco I, Sawyers CL. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer. 2002;2(7):489\u0026ndash;501.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGibieža P, Petrikaitė V. The regulation of actin dynamics during cell division and malignancy. Am J Cancer Res. 2021;11(9):4050\u0026ndash;69.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGao ZW, Wang HP, Lin F, Wang X, Long M, Zhang HZ, Dong K. CD73 promotes proliferation and migration of human cervical cancer cells independent of its enzyme activity. BMC Cancer. 2017;17(1):135.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXiang S, Yang L, He Y, Ding F, Qiao S, Su Z, Chen Z, Lu A, Li F. Alpha-1 Antitrypsin as a Regulatory Protease Inhibitor Modulating Inflammation and Shaping the Tumor Microenvironment in Cancer. Cells. 2025;14(2):88.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang H, Tang Q, Li S, Qin Y, Zhu G. TGFBI: A novel therapeutic target for cancer. Int Immunopharmacol. 2024;134:112180.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHuang C, Chen J. Laminin\u0026ndash;332 mediates proliferation, apoptosis, invasion, migration and epithelial\u0026ndash;to\u0026ndash;mesenchymal transition in pancreatic ductal adenocarcinoma. Mol Med Rep. 2021;23(1):11.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXie P, Liu JY, Yan H, Wang ZB, Jiang SL, Li X, Liu ZQ. Pan-cancer analyses identify DCBLD2 as an oncogenic, immunological, and prognostic biomarker. Front Pharmacol. 2022;13:950831.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYuan W, Liu B, Wang X, Li T, Xue H, Mo X, Yang S, Ding S, Han W. CMTM3 decreases EGFR expression and EGF-mediated tumorigenicity by promoting Rab5 activity in gastric cancer. Cancer Lett. 2017;386:77\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuerra L, Odorisio T, Zambruno G, Castiglia D. Stromal microenvironment in type VII collagen-deficient skin: The ground for squamous cell carcinoma development. Matrix Biol. 2017;63:1\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu S, Xu H, Wang W, Li S, Li H, Li T, Zhang W, Yu X, Liu L. The role of collagen in cancer: from bench to bedside. J Transl Med. 2019;17(1):309.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMalumbres M. Cyclin-dependent kinases. Genome Biol. 2014;15(6):122.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang M, Zhang L, Hei R, Li X, Cai H, Wu X, Zheng Q, Cai C. CDK inhibitors in cancer therapy, an overview of recent development. Am J Cancer Res. 2021;11(5):1913\u0026ndash;35.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShimizu M, Shibuya H, Tanaka N. Enhanced O-GlcNAc modification induced by the RAS/MAPK/CDK1 pathway is required for SOX2 protein expression and generation of cancer stem cells. Sci Rep. 2022;12(1):2910.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi TT, Lai YW, Han X, Niu X, Zhang PX. BMP2 as a promising anticancer approach: functions and molecular mechanisms. Invest New Drugs. 2022;40(6):1322\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePavlova NN, Thompson CB. The Emerging Hallmarks of Cancer Metabolism. Cell Metab. 2016;23(1):27\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKiri S, Ryba T. Cancer, metastasis, and the epigenome. Mol Cancer. 2024;23(1):154.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDing F, Xiao H, Wang M, Xie X, Hu F. The role of the ubiquitin-proteasome pathway in cancer development and treatment. Front Biosci (Landmark Ed). 2014;19(6):886\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eStatello L, Guo CJ, Chen LL, Huarte M. Gene regulation by long non-coding RNAs and its biological functions. Nat Rev Mol Cell Biol. 2021;22(2):96\u0026ndash;118.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGalon J, Bruni D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat Rev Drug Discov. 2019;18(3):197\u0026ndash;218.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHan C, Hu C, Liu T, Sun Y, Hu F, He Y, Zhang J, Chen J, Ding J, Fan J, Zhang X, Wang J, Qiao X, Jiang D, Yang K, Yang S. IGF2BP3 enhances lipid metabolism in cervical cancer by upregulating the expression of SCD. Cell Death Dis. 2024;15(2):138.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCruz SP, Zhang Q, Devarajan R, Paia C, Luo B, Zhang K, Koivusalo S, Qin L, Xia J, Ahtikoski A, Vaarala M, Wenta T, Wei GH, Manninen A. Dampened Regulatory Circuitry of TEAD1/ITGA1/ITGA2 Promotes TGFβ1 Signaling to Orchestrate Prostate Cancer Progression. Adv Sci (Weinh). 2024;11(11):e2305547.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang H, Shi H, He X, Liao A. Downregulation of Chemokine CCL20 Involved in Myeloma Cells Resistant to Elotuzumab and Lenalidomide. Onco Targets Ther. 2021;14:2789\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen Y, Ouyang Y, Li Z, Wang X, Ma J. S100A8 and S100A9 in Cancer. Biochim Biophys Acta Rev Cancer. 2023;1878(3):188891.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGalber C, Acosta MJ, Minervini G, Giorgio V. The role of mitochondrial ATP synthase in cancer. Biol Chem. 2020;401(11):1199\u0026ndash;214.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi P, Wang K, Song J, Chen Z, Li Y, Chen Z. THBS1 knockdown suppresses pancreatic cancer progression through JAK2/STAT3 signaling pathway. Mol Cell Probes. 2025;79:102003.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChang IW, Liang PI, Li CC, Wu WJ, Huang CN, Lin VC, Hsu CT, He HL, Wu TF, Hung CH, Li CF. HAS3 underexpression as an indicator of poor prognosis in patients with urothelial carcinoma of the upper urinary tract and urinary bladder. Tumour Biol. 2015;36(7):5441\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJumpertz S, Hennes T, Asare Y, Sch\u0026uuml;tz AK, Bernhagen J. CSN5/JAB1 suppresses the WNT inhibitor DKK1 in colorectal cancer cells. Cell Signal. 2017;34:38\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiang W, Chen J, Li H, Zhang P, Guan H, Li Y. High expression of COL8A1 predicts poor prognosis and promotes EMT in papillary thyroid cancer. Endocr Connect. 2024;13(12):e240279.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSun LR, Li SY, Guo QS, Zhou W, Zhang HM. SPOCK1 Involvement in Epithelial-to-Mesenchymal Transition: A New Target in Cancer Therapy? Cancer Manag Res. 2020;12:3561\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZuo Y, Xu X, Chen M, Qi L. The oncogenic role of the cerebral endothelial cell adhesion molecule (CERCAM) in bladder cancer cells in vitro and in vivo. Cancer Med. 2021;10(13):4437\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKuo WH, Chu PY, Wang CC, Huang PS, Chan SH. MAP7D3, a novel prognostic marker for triple-negative breast cancer, drives cell invasiveness and cancer-initiating cell properties to promote metastatic progression. Biol Direct. 2023;18(1):44.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKristanti AN, Aminah NS, Siswanto I, Manuhara YSW, Abdjan MI, Wardana AP, Aung EE, Takaya Y. Anticancer potential of β-sitosterol and oleanolic acid as through inhibition of human estrogenic 17beta-hydroxysteroid dehydrogenase type-1 based on an in silico approach. RSC Adv. 2022;12(31):20319\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng F, Wang Z, Li S, Xiong S, Yuan Y, Zeng J, Tan Y, Liu X, Xu S, Fu B. Development of a propionate metabolism-related gene-based molecular subtypes and scoring system for predicting prognosis in bladder cancer. Eur J Med Res. 2024;29(1):393.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim B, Jung M, Moon KC, Han D, Kim K, Kim H, Yang S, Lee D, Jun H, Lee KM, Lee CH, Nikas IP, Yang S, Lee H, Ryu HS. Quantitative proteomics identifies TUBB6 as a biomarker of muscle-invasion and poor prognosis in bladder cancer. Int J Cancer. 2023;152(2):320\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang P, Gu S, Pan D, Fu J, Sahu A, Hu X, Li Z, Traugh N, Bu X, Li B, Liu J, Freeman GJ, Brown MA, Wucherpfennig KW, Liu XS. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018;24(10):1550\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen L, Flies DB. Molecular mechanisms of T cell co-stimulation and co-inhibition. Nat Rev Immunol. 2013;13(4):227\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRui R, Zhou L, He S. Cancer immunotherapies: advances and bottlenecks. Front Immunol. 2023;14:1212476.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSharma P, Hu-Lieskovan S, Wargo JA, Ribas A. Primary, Adaptive, and Acquired Resistance to Cancer Immunotherapy. Cell. 2017;168(4):707\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSpranger S, Gajewski TF. Impact of oncogenic pathways on evasion of antitumour immune responses. Nat Rev Cancer. 2018;18(3):139\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTian L, Goldstein A, Wang H, Ching Lo H, Sun Kim I, Welte T, Sheng K, Dobrolecki LE, Zhang X, Putluri N, Phung TL, Mani SA, Stossi F, Sreekumar A, Mancini MA, Decker WK, Zong C, Lewis MT, Zhang XH. Mutual regulation of tumour vessel normalization and immunostimulatory reprogramming. Nature. 2017;544(7649):250\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMpekris F, Baish JW, Stylianopoulos T, Jain RK. Role of vascular normalization in benefit from metronomic chemotherapy. Proc Natl Acad Sci U S A. 2017;114(8):1994\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJune CH, O'Connor RS, Kawalekar OU, Ghassemi S, Milone MC. CAR T cell immunotherapy for human cancer. Science. 2018;359(6382):1361\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePhi LTH, Sari IN, Yang YG, Lee SH, Jun N, Kim KS, Lee YK, Kwon HY. Cancer Stem Cells (CSCs) in Drug Resistance and their Therapeutic Implications in Cancer Treatment. Stem Cells Int. 2018;2018:5416923.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGan PP, Pasquier E, Kavallaris M. Class III beta-tubulin mediates sensitivity to chemotherapeutic drugs in non small cell lung cancer. Cancer Res. 2007;67(19):9356\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAka JA, Mazumdar M, Chen CQ, Poirier D, Lin SX. 17beta-hydroxysteroid dehydrogenase type 1 stimulates breast cancer by dihydrotestosterone inactivation in addition to estradiol production. Mol Endocrinol. 2010;24(4):832\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"BLCA, Single-cell RNA sequencing, Cancer stem cells, Targeted therapy, Prognostic model","lastPublishedDoi":"10.21203/rs.3.rs-7536468/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7536468/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e Bladder cancer (BLCA) is a common malignancy with increasing incidence globally. Epithelial cells play crucial roles in tumor development and metastasis. Single-cell RNA sequencing (scRNA-seq) enables investigation of cellular heterogeneity. This study aims to analyze epithelial cell heterogeneity in BLCA, identify biomarkers, and develop prognostic models using machine learning to explore their clinical significance.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e We integrated scRNA-seq and bulk RNA-seq data from TCGA and GEO databases, including muscle-invasive bladder cancer (MIBC), non-muscle-invasive bladder cancer (NMIBC), and adjacent tissue samples. Data processing included Seurat clustering, Harmony batch correction, copy number variation analysis, KEGG enrichment analysis, CellChat intercellular communication analysis, and Monocle3 trajectory analysis. A cancer stem cell-related prognostic index (CSCRPI) was constructed using four machine learning algorithms based on cancer stem cell (CSC) subpopulation marker genes.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e We identified 39 clusters, 9 cell types, and 11 epithelial cell subtypes, including a cancer stem cell subpopulation (Epi9_CSC). Epi9_CSC was highly enriched in MIBC and promoted invasion and recurrence. Pathway analysis revealed that Epi9_CSC secretes collagen proteins that interact with integrin and CD44 receptors, activating downstream signaling pathways including Focal adhesion, PI3K-Akt, NF-κB, and MAPK pathways. Drug sensitivity analysis identified AZD1208, IAP inhibitors, and Nutlin-3a as potential therapeutic agents targeting Epi9_CSC. The CSCRPI scoring system, based on 5 key feature genes from Epi9_CSC, accurately predicted BLCA patient prognosis and provided clinical guidance.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e This study identified Epi9_CSC as a highly invasive cancer stem cell subpopulation that drives BLCA malignancy through extracellular matrix remodeling and multiple oncogenic pathway activation. The CSCRPI system offers valuable prognostic insights and potential therapeutic targets for precision medicine in BLCA treatment.\u003c/p\u003e","manuscriptTitle":"scRNA-seq reveals epithelial heterogeneity in bladder cancer and establishes a cancer stem cell prognostic model","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-22 19:23:44","doi":"10.21203/rs.3.rs-7536468/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"da799775-e3e1-45e2-8404-3a44f537dddd","owner":[],"postedDate":"October 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-03T08:39:52+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-22 19:23:44","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7536468","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7536468","identity":"rs-7536468","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}