Introduction
Bladder cancer (BLCA) is a common urologic malignancy that is three to four times more common in men than in women and has a higher mortality and morbidity rate1,2. BLCA is also one of the most common cancers worldwide, with approximately 550,000 new cases each year3. However, the outcome of treatment for advanced BLCA remains unsatisfactory. Patients with BLCA are vulnerable to multiple recurrence and require intervention4. Exploring and developing a safe and potent targeted therapeutic drug with minimal side effects to combat the poor prognosis associated with BLCA is imperative due to the high tumor heterogeneity and the tumor resistance caused by inhibitory tumor microenvironment (TME). Tumor angiogenesis plays a crucial role in influencing the anti-tumor immune response by regulating the interaction between tumor cells and the TME5. Therefore, identifying angiogenesis-related features is crucial for assessing the TME and enhancing the therapeutic efficacy of immunotherapy for BLCA patients. Tumor angiogenesis is associated with cell invasion, migration, drug resistance, and immune escape6.
Antibody-drug conjugates (ADCs) are commonly created through the covalent linkage of monoclonal antibodies (mAbs) with cytotoxic drugs, facilitated by chemical linkers7. These conjugates combine the benefits of highly specific targeting and potent cytotoxicity, allowing for the precise and efficient eradication of cancer cells. Consequently, ADCs have gained significant attention in the field of cancer drug development. Artificial intelligence-driven drug design (AIDD) is an emerging field that combines artificial intelligence (AI), machine learning (ML), and deep learning (DL) techniques to handle large volumes of biological data, thereby reducing the time and costs associated with the drug development process8. Therefore, the utilization of AIDD technology for the development of highly cytotoxic companion agents for ADCs targeting BLCA holds immense feasibility and research value. Therefore, utilizing ML and DL within an AIDD framework to develop a novel angiogenesis-targeted ADC for BLCA therapy holds significant promise and appears highly feasible.
This study employs an integrated ML system to construct angiogenesis-related gene signatures (ARGS) for unveiling the association between tumor angiogenesis and TME remodeling in BLCA, as well as evaluating prognostic biomarkers and chemotherapy drug sensitivity in BLCA patients. Furthermore, leveraging AIDD technology and target protein homology modeling, we seek to discover a new anti-angiogenic ADC with low biological toxicity and high targeting efficacy for BLCA treatment.
Materials and methods
Flowchart of the research in this study
This study aims to construct an ARGS model using an integrated learning system based on publicly available databases and openly accessible data from the literature. The model will be used to evaluate the prognostic survival, chemotherapy drug sensitivity, and the association between angiogenesis and TME remodeling in different subgroups of BLCA patients. Additionally, novel anti-angiogenic ADCs will be designed and screened using AIDD technology for the treatment of BLCA. Furthermore, a multi-dimensional ML approach will be employed to identify a prognostic biomarker for BLCA and assess the therapeutic efficacy of the drug treatment (Fig. 1).
[See PDF for image]
Fig. 1
Workflow diagram of this study.
Data acquisition
RNA transcriptome sequencing data, somatic mutation data and corresponding clinical information of BLCA patients were obtained from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/). The TCGA-BLCA cohort was used as the training cohort in this study. To validate the results of the training set, we downloaded the independent dataset of BLCA from the Gene Expression Omnibus website (https://www.ncbi.nlm.nih.gov/geo/). Therefore, the GSE19915, GSE3167, GSE13507, GSE31189, GSE52519, GSE65635 and GSE149582 cohort were used as validation set for this study. The 668 angiogenesis-related genes were retrieved from Molecular Signatures Database (M41711, M12982, M160, M12975, M39556, and M14493) (https://www.gsea-msigdb.org/gsea/msigdb)9 and GeneCards database (https://pathcards.genecards.org/).
The transcription factors associated with cancer were downloaded from CistromeDB10. High cytotoxic drugs that are structurally known and associated with the immune checkpoints were collected from existing studies11. These drugs (Pseudomonas aeruginosa exotoxin A, MMAE, Mertansine, Exotoxin A, Azithromycin, Exatecan) derivative for the ADCs, MMAF, IRDye700DX, 7-ethyl-10-hydroxycamptothecin, Pyrrolobenzodiazepine7,12. The structures of these compounds were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Since crystal structures of the identified core therapeutic targets were not available, the amino acid sequences of these proteins were retrieved from the UniProt database (https://www.uniprot.org/). All data used in this study are obtained from publicly available databases and publicly accessible data from published articles. No ethical approval is required, and all data have been obtained with proper permissions for use.
Integrating ML framework for ARGS scoring and functional annotation
We performed differential expression analysis of 19 normal and 412 tumor tissues in the TCGA-BLCA cohort to identify angiogenesis-related genes and transcription factors involved in BLCA progression. Genes with |log2-fold change (FC) | > 1 and false discovery rate (FDR) < 0.05 were considered as diffeomorphically exposed genes. To further understand the biological functions and pathways of differentially expressed angiogenesis-related genes and transcription factors, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis using the “clusterprofiler” package in R (version 4.3.1)13,14. To avoid overfitting, we performed Unicox, Multicox and LASSO regression (iteration = 1000) using the “glmnet” package in the R software. The ARGS score was calculated using the following formula: ARGS score = [expressivity limit of gene1 × the factor of gene1 + expressivity limit of gene2 × the factor of gene2 +. + expression level of genen] × the factor of genen, where the coefficients (weights) for each gene were derived from the LASSO regression analysis. The median ARGS score across the entire TCGA-BLCA cohort (n = 412) was computed as the midpoint value separating the higher half from the lower half of the ranked ARGS scores. Using the cohort median ARGS score as cutoff, high-risk (score > median) and low-risk (score ≤ median) groups were defined. The median-based cutoff was selected to avoid arbitrary thresholding and ensure balanced subgroup sizes for downstream analyses.
Validation of ARGS and construction of prognostic model
The prognostic capacity of the ARGS in BLCA was systematically evaluated through three complementary analytical approaches: (1) receiver operating characteristic (ROC) curve analysis to quantify predictive accuracy for survival outcomes, (2) principal component analysis (PCA) to assess dimensionality reduction and feature space separation, and (3) t-distributed stochastic neighbor embedding (t-SNE) for nonlinear visualization of risk group stratification patterns. Time-dependent ROC curve was generated to assess 1-year, 3-year, and 5-year overall survival (OS) prediction accuracy in the BLCA cohort. To assess the clinical prognostic value of ARGS in BLCA patients, Kaplan‒Meier (KM) survival analysis was performed using R software to compare the OS between the high-risk and low-risk groups. Additionally, Cox proportional hazards regression models were applied to determine whether ARGS served as an independent predictor of OS. To investigate the impact of ARGS on BLCA progression, the association between ARGS and clinical pathological factors (TNM staging, pathological grading, age, gender, etc.) was analyzed.
Nomogram-based prognostic modeling with algorithm validation
We constructed column nomograms based on the ARGS and important clinicopathological parameters to predict the 1-year, 3-year, and 5-year survival of BLCA patients. Calibration curves for assessing whether there is congruence and forecasting value between predicted prognostic OS and actual OS in patients with BLCA. ROC curves over time were applied to assess the specificity and sensitivity of the model. Using 113 algorithms from 18 machine learning methods [glmBoost, LASSO, RF, GBM, Ridge, Stepglm, Enet, LDA, NaiveBayes, plsRglm, SVM, XGBoost, CoxBoost, RSF, plsRcox, StepCox, survivalSVM, SuperPC] to compute and construct ROC models for secondary evaluation of the accuracy of BLCA angiogenesis models. Unicox and Multicox regression analyses were conducted to determine whether ARGS could serve as an independent risk factor for survival and clinical characteristics of BLCA patients. Forest plots were generated to visualize the results.
Enrichment analysis of GO, KEGG, and gene set variation analysis (GSVA)
To identify signaling pathways and signature gene sets associated with ARGS, pre-ranked gene set enrichment analysis (GSEA version 4.3.2) was performed using sample groupings and expression profiles. Subsequently, GSVA was constructed to compare enrichment patterns between high-risk and low-risk patient cohorts. To elucidate key pathways regulating tumor angiogenesis in the ARGS model, gene and transcription factor enrichment analyses were performed separately on the 12 genes associated with the prognosis of BLCA patients using GO and KEGG enrichment analysis.
Expression, prognosis, and functional analysis of ARGS model genes in BLCA
This study systematically analyzed the expression and prognostic significance of ARGS model genes in the TCGA-BLCA dataset using R language (v4.3.1). Survival analysis: Based on the “Survival” and “Survminer” packages, patients were divided into high/low expression groups according to the median gene expression. Kaplan Meier method was used to plot survival curves, and log rank test was used to evaluate inter group differences (significance threshold P < 0.05). Functional enrichment analysis: The correlation between gene expression and survival status was calculated using the limma package, sorted by log2 (fold change), and then subjected to GSEA analysis on the MSigDB gene set using the clusterProfiler package (count-nPerm = 1000, FDR < 0.25).
Retrieval of immunohistochemistry (IHC) images for ARGS model genes from the HPA database
To validate the protein expression patterns of the 12 ARGS genes comprising our prognostic model, we systematically retrieved IHC data from the HPA database15,16. For each gene, we accessed the Tissue Atlas module, with priority given to BLCA specimens in the Pathology Atlas section, while normal bladder tissue data were included when tumor-specific images were unavailable. High-resolution IHC images were downloaded in PNG format along with their corresponding semi-quantitative staining intensity scores (negative, weak, moderate, or strong).
Eight immune cell algorithms for mining TME remodeling features
Patients were stratified into Immune-H and Immune-L groups based on immune scoring calculated using the ESTIMATE and ssGSEA algorithm. To characterize the immune cell profile of BLCA samples, we used multiple methods (XCELL, TIMER, QUANTISEQ, EPIC, MCPCOUNTER, CIBERSORT, and CIBERSORT-ABS) to study the correlation of immune cells with patient risk scores in the TCGA-BLCA group17. A comparison between the two ARGS subgroups was analyzed to further investigate the impact of the ARGS risk score on immunotherapy to assess the differences in immune checkpoints and immune function. In a separate effort, we conducted a TCIA for therapy in both the high-risk and low-risk cohorts of BLCA patients.
Screening for potential anti-angiogenic drugs and biomarkers
The drugs used for drug sensitivity analysis are from the research cohort of Paul Geeleher et al. and the developed pRRophetic software package, which can be used to predict clinical drug responses to many cancer drugs18, 19–20. We estimated half maximal inhibitory concentration (IC50) values for common clinical/preclinical anti-tumor drugs in low- and high-risk groups based on ARGS scores (R < −0.4). Based on the XSum algorithm, perform high-throughput virtual screening of small molecule compounds from the CMap database (https://clue.io/data) to identify small molecule drugs with anti-BLCA angiogenesis activity. An ensemble of ML algorithms [support vector machine (SVM), random forest (RF), batch gradient descent for deep learning (BATCH), weighted gene co-expression network analysis (WGCNA), LASSO] was then employed to perform dimensionality reduction and jointly select biomarkers that meet BLCA features. The GSE65635 dataset was used as a test set to evaluate the accuracy of the ML models. GSVA was used to analyze the biological functions of the BLCA biomarkers.
AIDD simulation screening of ADC drugs based on risk-consensus targets
The Discovery Studio (DS) was used for ligand-based pharmacophore modeling and molecular docking to explore the main pharmacological features of these highly cytotoxic drugs. Risk genes obtained from previous analysis were subjected to differential gene expression analysis in BLCA using the whole genome (logFC = 1, P < 0.05), resulting in potential therapeutic target genes associated with risk in BLCA. These genes were imported into the STRING database (https://string-db.org/) for protein-protein interaction (PPI) analysis to identify core therapeutic targets in BLCA. Homology modeling of human crystal structures of these proteins were generated using the DS software, utilizing the SWISS-MODEL (https://swissmodel.expasy.org/) and SAVES (https://saves.mbi.ucla.edu/).
Compounds potentially targeting the identified core therapeutic targets were collected from the HIT 2.0 database (http://www.badd-cao.net:2345/), and their absorption, distribution, metabolism, and excretion (ADME) properties were analyzed by using the SWISS-ADME21,22. The pharmacophore models of the ADC combinations with high cytotoxicity were used to overlay the molecular pharmacological features of compounds that potentially target the core therapeutic targets in BLCA. Molecular docking was used to assess whether compounds can stably bind to the potential therapeutic targets. The DS and AutoDock Vina software were employed for molecular docking, with the protein structure being the homology-modeled human crystal structure. CDOCKER and LibDock were used to evaluate the potential interactions between small molecule ligands and the protein binding site.
Prediction of toxicological properties of SSD based on FDA & NTP library
This study systematically predicted the toxicity characteristics of saikosaponin D (SSD) using the Toxicity Prediction by Komputer Assisted (TOPKA) module of DS. TOPKAT uses a quantitative structure-activity relationship (QSAR) model, combined with Bayesian probability scoring and Mahalanobis Distance algorithm, to evaluate the potential risks of compounds in 12 toxicity endpoints, including biodegradability, Ames-mutagenicity, rodent carcinogenicity (NTP and FDA models), acute toxicity (IC50 and LD50), and chronic toxicity (LOAEL).
In the specific operation, the two-dimensional molecular structure of SSD is input into software, which automatically generates three-dimensional conformation and physicochemical parameters. Then, the model determines toxicity classification (positive/negative) through Bayesian scoring, and evaluates the reliability of the prediction results by using the Mahalanobis distance and its corresponding P-value (reflecting the consistency between sample and training set distribution).
Statistical analysis
All data processing, analysis, and visualization were conducted in R software. Regression analysis of the data was performed using both the univariate and multivariate modules. Spearman’s rank correlation test was used for correlation analysis of the ARGS score, gene co-mutations, immune infiltrating cells, and drug sensitivity. In the present study, P < 0.05 was considered statistically significant. Libdockscore, Vina score, and CDOCKER energy are counted in units of kcal/mol. Vina score < −8 and Libdockscore > 100 are considered indicative of good targetability and stable binding ability of the drug molecule. FitValue is used to assess the similarity between the molecule and the pharmacological features, and a FitValue > 2 suggests that the investigational drug has similar chemical actions and functions as known drugs.
Results
Establishment of ARGS model and regulation mechanism of BLCA angiogenesis
Differentially expressed analysis identified 147 angiogenesis-related genes (98 downregulated, 49 upregulated). A heatmap subsequently visualized ARGS expression patterns in BLCA tissues and normal bladder tissues (Fig. 2A). The differential expression of down and upregulated angiogenesis-related genes is presented in the volcano map (Fig. 2B). GO enrichment analysis revealed that these genes are involved in the regulation of endothelial cell, epithelial cell, cell migration, and tumor angiogenesis. These pathways showed a downregulation trend, suggesting a high regulation of tumor angiogenesis and cell migration (Fig. 2C). KEGG enrichment analysis indicated a decreased trend in pathways related to PI3K-AKT signaling pathway and ECM-receptor interaction (Fig. 2D). Unicox regression analysis identified 41 risk genes associated with BLCA prognosis (Fig. 2E), and LASSO regression further reduced the dimensionality to 12 genes used to construct the ARGS scoring model (Fig. 2F). Among them, CALR, COL14A1, EMP1, ENO1, HMGA1, PDGFRA, SLC3A2, and TCF4 exhibited higher expression levels in the high-risk group, while GNG5, HSPE1, VHL, and MST1R exhibited higher expression levels in the low-risk group (Fig. 2G).
[See PDF for image]
Fig. 2
Functional characterization, screening, and modeling of DEGs in BLCA for ARGS model construction. A, B Heatmaps illustrating differential expression levels of DEGs between BLCA samples and normal bladder tissues; Volcano plots depicting expression fold-changes of DEGs. C, D Functional enrichment analysis of DEGs using GO and KEGG. E, FUnicox and LASSO regression were used to screen prognostic risk genes for BLCA, which were then utilized to construct the ARGS model. G Stratification of BLCA patients into high-risk and low-risk subgroups based on prognostic genes, with comparative analysis of gene expression patterns between subgroups.
Training of the ARGS model and construction of the BLCA prognostic model
ARGS risk stratification revealed significantly shorter OS in high-risk patients versus low-risk counterparts (P < 0.05) (Fig. 3A). Risk scores demonstrated a significant negative correlation with OS duration. The area under curve (AUC) for 1, 3, and 5-year survival indicated that in both the training and validation cohorts, the ARGS model effectively assessed the impact of angiogenesis on survival in BLCA patients (AUC > 0.6, Risk > 0.6) (Fig. 3B). The clinical relevance analysis of angiogenesis and the analysis of independent risk factors yielded results consistent with the AUC model, and the differences were statistically significant in the training, control, and validation models (P < 0.001) (Figs. 3E and 4D). PCA and t-SNE results demonstrated that the ARGS model effectively stratified BLCA patients, with a well-dispersed pattern (Fig. 3C). KM survival analysis revealed that patients in the high-risk group had significantly shorter OS compared to patients in the low-risk group (P < 0.001) (Fig. 3D). Consistent results were observed in both the training and validation cohorts.
[See PDF for image]
Fig. 3
ARGS model was used to evaluate the prognosis and correlation analysis of BLCA patients. (A) ARGS model was utilized to distinguish high-risk and low-risk subgroups of BLCA patients. (B) The AUC values for clinical characteristics risk demonstrated good concordance between the two models. (C) PCA and t-SNE analysis confirming distinct sample stratification and ARGS model efficacy. (D) KM survival curves demonstrating significant OS disparity between ARGS-defined risk subgroups. (E) Correlation heatmap of ARGS risk scores with clinicopathological characteristics.
[See PDF for image]
Fig. 4
Development of the ARGS model using ensemble learning and identification of angiogenesis-associated independent prognostic factors. A Ensemble learning validation of ARGS prognostic power. B, C Nomogram performance evaluation. D Forest plot of multivariate Cox regression showing ARGS as independent risk factor.
Prognostic significance of the ARGS
Taking the median ARGS risk score as the threshold, we divided the BLCA patients into two groups, the high-risk and low-risk groups (Fig. 3C). The heatmap shows the differential expression of 12 essential genes between two ARGS subgroups (Fig. 3A). The risk scores and surveillance profiles of each BLCA patient are shown in Fig. 3B. KM analysis demonstrated that in both the treatment and validation cohorts, the OS and meditation duration of survival of patients in the high-risk cohort were lower than those in the low-risk cohort (P < 0.05) (Fig. 3D). The AUC values at 1, 3 and 5 years were 0.695, 0.696 and 0.738 in the TCGA cohort, as indicated in Fig. 3C. In the GEO cohort, the AUC values at 1, 3 and 5 years were 0.712, 0.729 and 0.731. The KM analysis showed that the patients had worse OS and median survival times than those in the low-risk cohort (P < 0.05) (Fig. 3D). The correlation between ARGS scores and clinicopathological factors was elucidated in the TCGA cohort (Fig. 4D). These results suggest that the construction of the ARGS score model plays a crucial role in the progression for BLCA and is a good predictor of patient prognosis in BLCA.
Nomogram for survival probability in BLCA patients
To accurately predict the accuracy of OS models for different subgroups of patients, risk scores and other clinicopathologic characteristics of BLCA patients, including age, sex, grade, and stage, were integrated by constructing a column-line graph of prognostic risk-related models (Fig. 4B). We assessed the effectiveness of the KM survival curves and the validity of the ROC phantom based on the prophylactic risk score and the nomogram risk score at 1, 3, and 5 years, respectively (Fig. 4C). The calibration curves showed that the predicted and actual values fit well, and the model exhibited good predictive validity for the survival of actual BLCA patients. The nomogram AUC was 0.773 and 0.851 and the risk AUC was 0.747 and 0.786 for the trial and validation groups (Fig. 4C), respectively, and the results indicated that the column line plot was effective in predicting BLCA survival rates (Fig. 3B). The ARGS scoring model still demonstrates robust predictive ability for the prognosis of BLCA patients across various machine learning algorithms (Fig. 4A).
Enrichment analysis were conducted to explore the functional implications of the ARGS model
In GSVA, the risk score of each sample showed that the low-risk group was mainly enriched in the peroxisome, linoleic acid metabolism, and alpha-linolenic acid metabolism, which are pathways primarily related to metabolism. In contrast, the high-risk group was enriched in ECM-receptor interaction, B cell receptor signaling pathway, T-cell receptor signaling pathway, Toll-like receptor signaling pathway, and other immune-related pathways (Fig. 5A). In GO enrichment, the primary BP of prognostic risk angiogenesis-related genes is involved in peptide antigen assembly with MHC class I protein complex (GO:0002502) and phenylalanine transport (GO:0015823). For CCs, angiogenesis-related genes were mainly present in the collagen type XIV trimer (GO:0005596), the apical pole of neurons (GO:0044225), and the cell pole (GO:0060187). The MF of angiogenesis-related genes were mainly enriched in platelet-derived growth factor alpha-receptor activity (GO:0005018), structural constituent of chromatin (GO:0030527), and macrophage colony-stimulating factor receptor activity (GO:0005011) (Fig. 5B, C). KEGG analysis showed that prognosis-related risk angiogenesis-related genes were associated with the HIF-1 signaling pathway (ko04066), RAS signaling pathway (ko04014), and PI3K-AKT signaling pathway (ko04151) (Fig. 5B, C). The above results suggest that prognostic risk angiogenesis-related genes are mainly associated with immunity, cell proliferation, and vascular growth. Therefore, it can be inferred that in BLCA, tumor angiogenesis may be induced by risk genes regulating the HIF-1/PI3K-AKT/RAS signaling pathway, and the tumor immune microenvironment is closely associated with this process (Fig. 5B, C). T cells and B cells are likely to be the major participants in this process. The 18 transcription factors (KLF9, KLF2, HAND2, HLF, GATA6, FOXF1, EGR3, EGR2, EBF1, ZEB1, ZBTB16, SOX5, SOX17, SOX10, RUNX1T1, NR4A3, MYOCD) associated with ARGS are primarily related to tumor cell proliferation (Fig. 5C).
[See PDF for image]
Fig. 5
Functional enrichment analysis of BLCA samples derived from the ARGS model for high- and low-risk groups. A, B GSVA, GO, and KEGG analysis of prognostic genes. C Functional enrichment analysis of transcription factors associated with angiogenesis risk genes.
Expression, prognostic value, and GSEA analysis of ARGS genes in BLCA
The KM survival analysis showed that overexpression of PDGFRA, TCF4, EMP1 and low expression of MST1R, HSPE1, VHL suggest poor prognosis in BLCA patients (P < 0.05) (Figure S1), consistent with Fig. 2E. PDGFRA, TCF4, and EMP1 are risk factors for BLCA, while MST1R, HSPE1, and VHL are protective factors for BLCA. However, no significant differences were observed in the impact of mRNA expression levels of SLC3A2, HMGA1, GNG5, ENO1, and CALR on the prognosis of BLCA patients. The IHC map showed that except for PDGFRA, VHL, EMP1, and GNG5, which did not obtain slices, the expression levels of other genes in tissues were similar to those obtained from KM survival analysis and risk gene analysis. CALR, COL14A1, SLC3A2, ENO1, and TCF4 were highly expressed in BLCA tissues, while HMGA1, MST1R, and HSPE1 were highly expressed in normal bladder and urinary tract epithelial tissues (Figure S2).
GSEA identified significant associations between ARGS genes and critical oncogenic pathways (Figure S3). CALR, PDGFRA, and EMP1 were enriched in cytokine-cytokine receptor interaction and ECM-receptor interaction, suggesting their roles in tumor microenvironment remodeling and immune modulation. Notably, COL14A1, MST1R, SLC3A2, TCF4, and ENO1 clustered at the leading edge of ECM-receptor interaction, implicating their potential involvement in extracellular matrix dysregulation and tumor invasiveness.
Conversely, GNG5 showed inverse enrichment in natural killer cell-mediated cytotoxicity and xenobiotic metabolism, hinting at immune evasion and chemoresistance mechanisms. HMGA1 and VHL were linked to cell cycle, cytokine signaling, and JAK-STAT pathways, aligning with their known roles in proliferation and cell-cycle dysregulation. Additionally, HSPE1 was highly enriched in oxidative phosphorylation, possibly reflecting metabolic reprogramming in BLCA progression. These findings highlight the multifaceted regulatory roles of ARGS genes in BLCA pathogenesis, spanning immune suppression, ECM remodeling, and metabolic adaptation.
Immune cell correlation analysis
Taking the median risk score of BLCA patients as a threshold, risk correlation analysis of immune-related cells and TME scores was performed with various software, and 12 types of immune cells were highly correlated with risk scores(Fig. 6A). The results showed that increased levels of the infiltration of immune cells, such as M0 macrophages, neutrophils, and M2 macrophages, were strongly associated with poor prognosis in patients in the high-risk group. In ARGS, there are mainly cancer associated fibroblasts, M1 macrophages, M2 macrophages, macrophages, monocytes, monocytes, activated myeloid dendritic cells, myeloid dendritic cells, neutrophils, CD4 + Th2 T cells, CD8 + T cells and regulatory T cells (Tregs) (Fig. 6D), and the increased level of infiltration of immune cells described above in the high-risk group suggests that the poorer prognosis of BLCA patients in the high-risk group correlates with the increased level of infiltration of immune cells in the TME (Fig. 6F).
[See PDF for image]
Fig. 6
The contribution of ARGS model to TME remodeling in BLCA and prediction of immunotherapy efficacy. (A) Immune functional activation status in angiogenesis-related high-immune and low-immune subtypes of BLCA. (B) The t-SNE sample distribution plots for different immune level subgroups. (C) The degree of TME reshaping, as well as the differences in immune scores and stromal scores, among different immune subgroups. (D) Evaluation of immune cell infiltration levels and their association with angiogenesis risk using seven immune cell algorithms. (E) The heatmap depicting the association between angiogenesis risk and the activation/blockade status of immune checkpoints. (F) The contribution of angiogenesis risk to the activation/inhibition of immune-related functions in the TME. (G) Boxplots illustrating the association between angiogenesis risk and C1-C6 immune subtypes in BLCA. (H) Calculating IPS to predict the efficacy of immunotherapy in different subgroups of BLCA patients. *P < 0.05, **P < 0.01, ***P < 0.001.
The t-SNE results showed that the grouping model could evaluate the differences between the high-immunity group and the low-immunity group very well (Fig. 6B). Furthermore, both the immunity score and stromal score were lower in the low immunity score cohort than in the high immunity score cohort (Fig. 6C). Significant immunological features were found in the high and low immunity scoring cohorts, with statistically meaningful variations, as demonstrated in Fig. 6A. Immunophenotyping analysis was performed on all samples, and ssGSEA scores based on 29 immune genomes were categorized into four subtypes (referred to as C1-C4) (Fig. 6G), including wound healing (C1), IFN-γ dominance (C2), inflammation (C3), and lymphocyte depletion (C4). Immunological infiltration decreases in the decreasing sequence: C2 > C1 > C3 > C4. For example, C2 is dominant. C1 was mainly clustered in the high-risk group of patients. C3 and C4, however, were mainly clustered in the low-risk group of patients (Fig. 6G). Accordingly, patients in the low-risk category are better equipped to be immunotherapy beneficiaries. While patients in the high-risk bracket are refractory to immunotherapy. As far as the high-immunity subgroup and low-immunity subgroup are involved, the immunotherapy outcomes and prognosis of the high-immunity subgroup are better as compared to the patients in the low-immunity subgroup.
After analyzing the immune checkpoints, we found that TNFSF15, TNFRSF25, TNFRSF14, TMIGD2, and LGALS9 were negatively correlated with the risk scores. This suggests that the costimulatory activation of the TNF receptor and ligand superfamily was suppressed (Fig. 6E), and the downregulation of VHL, HSPE1, and MST1R in the tumor tissues was strongly correlated with suppression of TNF receptor and ligand superfamily. According to TCIA immunotherapy analysis data (https://tcia.at/)23, when CTLA-4 and PD-1 were blocked, a comparison of immunocyte positive scores (IPS) revealed that patients in the low-risk group had a better immunotherapy outcome and prognosis than patients in the high-risk group. The percentage of tumor-associated immune cells was higher than any intensity of PD-L1 membrane- and cytoplasm-stained tumor-associated immune cells, as determined by the percentage of all tumor-associated immune cells. Patients with BLCA in the high-risk group received both CTLA-4 and PD-1 blockade as the optimal immunotherapy strategy. Moreover, patients in the low-risk group have a better immunotherapy outcome than patients in the high-risk group when PD-1 alone is blocked, and a poor immunotherapy outcome is a crucial factor in the poor prognosis of patients in the high-risk group (Fig. 6H).
Drug sensitivity analysis and evaluation of biomarker selection
To identify therapeutic candidates for BLCA, we conducted drug sensitivity analysis using the prognostic ARGS risk signature. We finally obtained a total of 18 known drugs with the potential to reverse poor prognosis in BLCA patients (R > 0.4), such as TGX221, Saracatinib, and pazopanib. WH-4-023 is the most likely to reverse poor prognosis in BLCA patients (Fig. 7A). STOCK1N-35,696, was identified as a novel potential therapy for BLCA (Fig. 7B). Drug sensitivity analysis showed that chemotherapy is less effective in patients in the high-risk group than those in the low-risk group.
[See PDF for image]
Fig. 7
ARGS model evaluation of chemotherapy drug sensitivity in high-risk BLCA patients and screening of potential anti angiogenic therapy targets based on machine learning. (A) Drug library screening for sensitive chemotherapy drugs in high-risk group patients. (B) The small molecule compound STOCK1N-35,696 shows the most potential for anti-angiogenic treatment in BLCA. (C) The BATCH algorithm was used to remove batch effects from samples in different datasets. D–G Integration of ML for the screening of biomarkers to evaluate the therapeutic efficacy of anti-angiogenic treatment in BLCA.
The PCA model displayed the results of batch normalization for five groups of BLCA and normal bladder tissue datasets, indicating good separation of the samples (Fig. 8A). The BATCH algorithm was used to select genes with significant differences between BLCA tissue and normal tissue, using a logFC = 1 and P = 0.05 threshold for multiple learning screenings (Fig. 7C). LASSO regression identified 18 potential biomarkers associated with BLCA (Fig. 7D), while SVM selected 25 potential biomarkers with the smallest selection error (ERROR) for change point screening (Fig. 7E). RF selected 45 potential biomarkers with a statistical count of 1000 and gene importance score > 2 (Fig. 7F). WGCNA exhibited a stable topological fit index reaching 0.9 when the soft threshold was set to 7 (Fig. 7G). Key modules were selected, and it was found that the blue module had the strongest correlation with BLCA, encompassing 33 key module genes (Fig. 7D-G). The intersection of the potential biomarkers obtained from the four machine learning screenings yielded 5 common genes (MYH11, FGF9, BIN1, SRPX, JAM3) (Fig. 8B). The AUC curves for the control and validation groups were 0.834 and 1.000, respectively (Fig. 8C), with high predictive validity. The expression changes of these genes in the samples were represented by line graphs, showing significant differences in the expression of potential biomarkers between the control group and the experimental group (Fig. 8E). Figure 8D shows that, in the training group queue, the AUC for MYH11 was 0.782 (95% CI = 0.729–0.827), for FGF9 it was 0.782 (95% CI = 0.732–0.830), for BIN1 it was 0.782 (95% CI = 0.733–0.830), for SRPX it was 0.763 (95% CI = 0.708–0.813), and for JAM3 it was 0.767 (95% CI = 0.717–0.812). Among them, MYH11 had the highest cutoff value, and FGF9 had the highest model sensitivity.
[See PDF for image]
Fig. 8
Confirmation of biomarkers for evaluating the therapeutic efficacy of anti-angiogenic treatment and their functional analysis. (A) PCA analysis to examine the dispersion of samples after batch removal using the BATCH algorithm. (B) Venn diagram illustrating the genes with the highest potential to serve as biomarkers for evaluating treatment efficacy. C, D ROC models and AUC for the training and validation cohorts, as well as the model genes. E Violin plots depicting the expression level of model genes in the training and validation cohorts. F Functional enrichment analysis of biomarkers in BLCA using GSVA. *P < 0.05, **P < 0.01, ***P < 0.001.
Functional evaluation of biomarkers for anti-angiogenic therapy in BLCA
Consequently, MYH11 was found to be primarily involved in pathways such as vascular smooth muscle contraction, ECM-receptor interaction, and focal adhesion in BLCA. These pathways showed a negative correlation with MYH11 expression, suggesting that the downregulation of MYH11 is associated with tumor proliferation, migration, and angiogenesis in BLCA (Fig. 8F). Similarly, SRPX was found to be involved in signaling pathways such as complement and coagulation cascades, regulation of actin cytoskeleton, and focal adhesion, sharing similar functions with MYH11 and primarily associated with tumor proliferation and migration in BLCA. The downregulation of BIN1 was related to signaling pathways such as complement and coagulation cascades and cell adhesion molecules, which are mostly involved in cell proliferation and migration. FGF9 and JAM3 were also implicated in tumor proliferation and migration in BLCA. These biomarkers can serve as evaluation indicators for the therapeutic efficacy of anti-angiogenic treatment in BLCA, with MYH11 showing the highest sensitivity.
Exploring novel ADC-based treatments for BLCA using homology modeling and pharmacophore modeling
Using the Degree value as a screening criterion, the PPI network revealed the top 20 key genes that induce BLCA angiogenesis. Among them, FN1 is the most likely protein for BLCA angiogenesis and has high potential as a target for BLCA anti angiogenic therapy (Table 1). The protein crystal model of FN1 was derived from the Bos taurus (Bovine).
To identify the critical chemical features of highly cytotoxic ADC payloads targeting angiogenesis in BLCA, we constructed pharmacophore models incorporating three key interaction features: hydrogen bond acceptors, hydrophobic regions, and aromatic ring centers. Among the 10 generated models, Model 1 demonstrated superior alignment with known ADC payload mechanisms, achieving the highest rank (79.289) and FitValue (3.99953). The thermodynamic stability of Model 1 was further supported by favorable energy values (absolute energy = 596.419 kcal/mol; clean energy = 707.921 kcal/mol). These pharmacophore features correspond to empirically validated characteristics of ADC payloads, where hydrogen bond networks mediate target engagement while hydrophobic domains influence payload retention and bystander effects (Table 2). Model 1 included features of hydrogen bond acceptor interaction force and hydrophobic feature interaction force (Fig. 9A).
[See PDF for image]
Fig. 9
Screening and validation of novel anti-angiogenesis ADCs using AIDD technology. (A) Known pharmacophore receptor feature models for ADCs. (B) Homology modeling of the protein FN1’s tertiary structure and analysis of the existence of amino acid residue conformations as well as residue Chi1-Chi2 analysis. C, D Analysis of the rationality of the FN1 protein tertiary structure and Ramachandran plot of the protein model. E ADME property analysis of potential natural compounds targeting FN1 and anti-angiogenesis effects.
Using SWISS-MODEL, we generated 10 homology models of human FN1 based on the bovine FN1 template (PDB:1FNF). Model 9 showed superior structural quality, with 97.23% of residues in Ramachandran-favored regions (vs. 70.11–95.97% for other models) and excellent validation scores (MolProbity = 0.77, QMEANDisCo = 0.81 ± 0.05). Despite moderate GMQE (0.13) due to FN1’s complex domain architecture, Model 9 preserved key functional motifs including the integrin-binding RGD domain (Table 3). The overall quality factor analysis of FN1 yielded a coefficient of 80, indicating good predictive ability and a stable protein crystal structure (Fig. 9D). The Ramachandran plot occupancy was 97.23% (> 90%), indicating reasonable dihedral angles for the target protein’s Cα (Fig. 9C). The energy landscape maps for the full residue and Chi1-Chi2 displayed active red residues that matched the amino acid residues involved in the molecular docking, further confirming the potential of FN1 as a cancer target (Fig. 9B).
ADME property analysis revealed that Ouabain, Rutin, SSD, Panaxoside A and Baicalin have poor synthetic accessibility and do not comply with Lipinski’s rule (Fig. 9E; Table 5). Except for Wogonin, the remaining compounds exhibit good solubility and lipophilicity. All compounds have low polarity, indicating stability (Table 5). Among them, Ceastrol has high drug-likeness and low levels of toxic side effects. Its low synthetic accessibility suggests ease of artificial synthesis or extraction.
All compounds targeting FN1 will be retained for the subsequent pharmacophore overlay of ADC companion agents. Comprehensive toxicity evaluation of potential ADC payloads (Table 4) revealed significant variation in safety profiles. Gypenoside III showed the highest toxicity risk (Acute Tox. = 4, STOT RE = 4), followed by Panaxoside A (scores = 4/4) and Genistein (scores = 4/2). In contrast, SSD exhibited favorable toxicity characteristics (scores = 3/3), supporting its selection as the lead candidate. These findings were consistent with molecular docking results, where higher toxicity compounds demonstrated stronger binding affinity but poorer safety margins. The toxicity hierarchy (Gypenoside III > Panaxoside A > Genistein > SSD) informed subsequent prioritization of compounds for ADC development.
Table 1. PPI network scoring (Top 20).
Name | EigenVector | Degree | Bridging | Closeness | Betweenness | Eccentricity |
|---|---|---|---|---|---|---|
FN1 | 0.34 | 42 | 8.07 | 0.0053 | 2555 | 0.25 |
MMP2 | 0.29 | 28 | 10.19 | 0.0048 | 797 | 0.25 |
ICAM1 | 0.24 | 26 | 9.10 | 0.0047 | 962 | 0.25 |
COL1A2 | 0.20 | 25 | 9.27 | 0.0044 | 1001 | 0.25 |
SPP1 | 0.25 | 24 | 38.67 | 0.0050 | 1915 | 0.33 |
PDGFRB | 0.24 | 24 | 12.86 | 0.0046 | 75 | 0.25 |
BGN | 0.19 | 23 | 7.94 | 0.0043 | 654 | 0.25 |
CD274 | 0.19 | 19 | 12.56 | 0.0044 | 642 | 0.25 |
FCGR3B | 0.15 | 19 | 12.35 | 0.0043 | 500 | 0.2 |
CSF1R | 0.18 | 19 | 10.03 | 0.0045 | 413 | 0.25 |
CSF1 | 0.20 | 18 | 6.82 | 0.0044 | 199 | 0.25 |
CXCL10 | 0.19 | 17 | 10.73 | 0.0045 | 217 | 0.25 |
TAGLN | 0.15 | 14 | 20.44 | 0.0041 | 349 | 0.25 |
COMP | 0.15 | 14 | 3.52 | 0.0040 | 62 | 0.25 |
TEK | 0.16 | 13 | 10.25 | 0.0042 | 168 | 0.25 |
FMOD | 0.14 | 13 | 1.85 | 0.0038 | 27 | 0.2 |
PPARG | 0.15 | 12 | 42.82 | 0.0042 | 667 | 0.25 |
HCK | 0.09 | 12 | 10.89 | 0.0037 | 163 | 0.2 |
ASPN | 0.13 | 12 | 4.79 | 0.0037 | 62 | 0.2 |
ITGA5 | 0.14 | 11 | 10.19 | 0.0041 | 92 | 0.25 |
Table 2. Construction of pharmacophore models.
Name | Rank | FitValue | Absoulte Energy | Clean Energy | Conf Number |
|---|---|---|---|---|---|
Model 1 | 79.289 | 3.99953 | 596.419 | 707.921 | 4 |
Model 2 | 78.183 | 3.99967 | 609.356 | 707.921 | 30 |
Model 3 | 78.077 | 3.99973 | 609.356 | 707.921 | 30 |
Model 4 | 77.612 | 3.99947 | 596.419 | 707.921 | 4 |
Model 5 | 77.611 | 3.9994 | 596.419 | 707.921 | 4 |
Model 6 | 77.237 | 3.99906 | 611.47 | 707.921 | 34 |
Model 7 | 77.237 | 3.99901 | 611.47 | 707.921 | 34 |
Model 8 | 76.966 | 3.99954 | 596.419 | 707.921 | 4 |
Model 9 | 76.933 | 3.9998 | 611.47 | 707.921 | 34 |
Model 10 | 75.885 | 3.9996 | 611.53 | 707.921 | 35 |
Table 3. Homology modeling of FN1.
Name | Ramachandran Favoured | Seq Identity | GMQE | QMEAN | MolProbity Score | QMEANDisCo Global |
|---|---|---|---|---|---|---|
Model 1 | 88.38% | 83.13% | 0.69 | −2.97 | 1.73 | 0.92 ± 0.05 |
Model 2 | 95.64% | 91.67% | 0.10 | 0.66 | 1.91 | 0.72 ± 0.05 |
Model 3 | 95.62% | 91.64% | 0.10 | 0.62 | 1.91 | 0.72 ± 0.05 |
Model 4 | 94.66% | 70.11% | 0.11 | 0.69 | 1.87 | 0.77 ± 0.05 |
Model 5 | 94.52% | 85.01% | 0.11 | 0.40 | 1.26 | 0.81 ± 0.05 |
Model 6 | 94.49% | 84.93% | 0.11 | 0.49 | 1.27 | 0.81 ± 0.05 |
Model 7 | 95.97% | 72.21% | 0.10 | 0.68 | 1.65 | 0.78 ± 0.05 |
Model 8 | 96.63% | 70.11% | 0.09 | −0.36 | 1.59 | 0.72 ± 0.05 |
Model 9 | 97.23% | 74.86% | 0.13 | 1.63 | 0.77 | 0.85 ± 0.05 |
Model 10 | 97.19% | 75.42% | 0.12 | 1.72 | 0.83 | 0.84 ± 0.05 |
Table 4. Potential high cytotoxicity companion agents for adcs.
Name | PubChem CID | Molecular formula | Hydrogen bond | Acute Tox. | STOT RE |
|---|---|---|---|---|---|
Genistein | 5,280,961 | C15H10O5 | 8 | 4 | 2 |
Ouabain | 439,501 | C29H44O12 | 20 | 3 | 2 |
Rutin | 5,280,805 | C27H30O16 | 26 | - | 3 |
Wogonin | 5,281,703 | C16H12O5 | 7 | - | - |
Oridonin | 5,321,010 | C20H28O6 | 10 | - | - |
Saikosaponin D | 107,793 | C42H68O13 | 21 | - | 3 |
Baicalin | 64,982 | C21H18O11 | 17 | - | 3 |
Celastrol | 122,724 | C29H38O4 | 6 | 3 | - |
Panaxoside A | 441,923 | C42H72O14 | 24 | 4 | - |
Gypenoside III | 9,898,279 | C54H92O23 | 38 | 4 | 4 |
Table 5. ADME properties of potential high cytotoxicity companion agents for adcs.
Name | TPSA§ | Solubility* | Lipinski | Bioavailability Score# | Synthetic accessibility† |
|---|---|---|---|---|---|
Genistein | 90.90 | Soluble | Yes | 55% | 2.87 |
Ouabain | 206.60 | Soluble | No | 17% | 7.13 |
Rutin | 269.43 | Soluble | No | 17% | 6.52 |
Wogonin | 79.90 | Moderately soluble | Yes | 55% | 3.15 |
Oridonin | 107.22 | Very soluble | Yes | 55% | 6.68 |
Saikosaponin D | 207.99 | Poorly soluble | No | 17% | 9.86 |
Baicalin | 187.12 | Soluble | No | 11% | 5.09 |
Celastrol | 74.60 | Poorly soluble | Yes | 85% | 6.28 |
Panaxoside A | 239.22 | Soluble | No | 17% | 9.21 |
Gypenoside III | – | – | – | – | – |
§The unit of Topological Polar Surface Area (TPSA) is Ų.
*Solubility class: Log S scale.
Insoluble< −10 < Poorly< −6 < Moderately< −4 < Soluble< −2 < Very < 0 < Highly. Estimated Solubility (ESOL), Ali’s Rule of Five (Ali), and SILICOS-IT are different algorithms related to solubility. SILICOS-IT is a machine learning-based algorithm that predicts the solubility of compounds.
#Bioavailability Score: Probability of F > 10%.
†Synthetic accessibility: 1 (very easy) ~ 10 (very difficult).
Virtual screening of cytotoxic ADC payload candidates using pharmacophore models and molecular docking
Based on Model 1 pharmacophore model, a screening was conducted to identify potential natural compounds with high cytotoxicity for ADC. SSD, Gypenoside III, Panaxoside A, and Celastrol were identified as potential natural compounds targeting FN1 that could potentially serve as high cytotoxicity ADCs. SSD has a FitValue of 2.46255, Gypenoside III has a FitValue of 2.40565, Panaxoside A has a FitValue of 2.23743, and Celastrol has a FitValue of 1.52411, indicating that SSD has the highest potential as an ADC companion agent. After molecular docking, CDOCKER energy analysis revealed that the small molecule ligand SSD had the lowest binding energy (−15.186 kcal/mol) when bound to the receptor protein FN1, indicating the most stable ligand-protein complex (Table 6). In conclusion, SSD has the highest potential as an ADC companion agent and can form stable complexes with FN1. Therefore, SSD is considered to have the ability to act as an ADC companion agent against BLCA angiogenesis.
Table 6. Results of molecular Docking energy and conformational analysis.
Ligand | FitValue | Libdock score | CDOCKER energy* | Binding energy* |
|---|---|---|---|---|
Saikosaponin D | 2.46255 | 113.527 | − 32.02 | − 15.186 |
Gypenoside III | 2.40565 | 104.948 | − 16.74 | − 14.525 |
Panaxoside A | 2.23743 | 86.663 | − 22.45 | − 11.216 |
Celastrol | 1.52411 | 92.305 | − 21.09 | − 8.791 |
*The units for CDOCKER energy and CDOCKER interaction energy are kcal/mol.
TOPKA analysis of SSD
The prediction results show that SSD exhibits low-risk characteristics in most toxicity endpoints (Table S1). SSD was clearly predicted as a “non-mutagenic substance” (Ames-test, Bayesian score = − 31.60, P < 0.05) and a “non-carcinogenic substance” (all rodent models, P < 0.05), with Bayesian scores for carcinogenicity in rats and mice ranging from − 8.05 to −0.89, showing significant statistical significance. Acute and chronic toxicity analysis showed that the LD50 of SSD in rats was 4.83 mg/(g·BW·day) (P < 0.05), and the chronic LOAEL was 2.62 mg/kg·BW (P < 0.05), indicating its high safety under acute and long-term exposure. In addition, the biodegradability of SSD was predicted to be “degradable” (Bayesian score = 5.69, P < 0.05), indicating that it may be easily metabolized in the environment and has a low ecological accumulation risk. The MD of all endpoints was significantly higher than the training set heart value (MD > 10), and the P-value was much lower than 0.05, further supporting the high confidence of the prediction results.
Discussion
The remodeled TME selects for highly invasive tumor cell populations, thereby promoting tissue invasiveness and metastatic potential5. Angiogenesis is the main prerequisite for tumor invasion and growth. This is highly likely to lead to poor prognosis and death of patients due to the angiogenesis of tumor tissues and the microenvironment selection of tumor invasive growth that mutually promote each other. The latest research shows that multi-omics analysis driven by integrated ML and genetic algorithms has strong predictive power for immune checkpoints and treatment outcomes of immunotherapy24. In this study, we constructed a novel ARGS based on angiogenesis-related genes in this study. In addition, the ARGS was shown to be of great value in predicting the TME in BLCA.
The ARGS consists of 12 key genes, including CALR, COL14A1, EMP1, ENO1, GNG5, HMGA1, HSPE1, MST1R, PDGFRA, SLC3A2, TCF4, and VHL. CALR is an endoplasmic reticulum-resident protein that is involved in a range of cellular processes25. CALR overexpression stimulates NF-κB activation, while NF-κB response is reduced in CALR-deficient lung cancer cells, as evidenced in a pulmonary carcinoma study26. Due to the activation of NF-κB and its association with inflammation response in the TME, CALR has a certain contribution to TME reshaping in BLCA. COL14A1 is closely related to core proteoglycans, small leucine-rich proteoglycans that have been progressively shown to strongly inhibit the growth of various tumor cells27. EMP1 is a biomarker of gefitinib resistance in lung cancer and contributes to prednisolone resistance in patients with acute lymphoblastic leukemia. EMP1 can inhibit cancer cell proliferation, angiogenesis, metastasis and induce apoptosis, 28. ENO1 is a glycolytic enzyme. Genolytic capabilities of ENO1 include dysregulation of cellular energy, sustaining tumor multiplication, and the inhibition of apoptosis in cancer cells. Moreover, ENO1 can help tumors be protected from immune destructiveness29. Therefore, in the contribution of the ARGS model to TME reshaping, ENO1 and EMP1 are likely to be major factors, which may be highly associated with tumor immune evasion and angiogenesis. This association is favorable for tumor proliferation and metastasis, indicating their role as prognostic risk factors. GNG5 is a recently discovered novel prognostic biomarker, but no studies have shown its specific mechanism. HMGA1 reprograms somatic cells to induced pluripotent stem cells and functions as a paracrine transcription factor, contributing to cancer development and metastasis30,31. HSPE1 has not yet been studied as a potential biomarker for cancer development and progression, but MST1R/RON receptor tyrosine kinase is a more widely known homolog of the MET receptor. Like MET, RON coordinates cellular signaling pathways that promote tumorigenesis and cancer cell survival, and this gene has potential as a therapeutic target32. Steroid receptor coactivator-1 (SRC-1), a ribonucleoprotein which modulates transcriptional excitation, is implicated in both PDGFRA and VHL. PDGFRA and VHL frequently demonstrate copy number alterations (CNAs), and their most pivotal carcinogenic properties are mitotic G1-G1/S phase restriction, gene transcription, apoptosis, and the PI3K-AKT pathway. However, the role of the 12 essential genes in BLCA is unclear. Therefore, the model is avant-garde for predicting the prognosis of patients with BLCA, the association of BLCA with TME and the contribution of ARGS to TME remodeling. It can be inferred that ARGS is closely associated with tumor metastasis, immune evasion, and TME reshaping in BLCA patients. As the risk increases, patients tend to have poor prognosis and treatment outcomes. The development of a learning feature using ML technology for evaluating the response rate of tumor immunotherapy and chemotherapy by TME has broad prospects33,34.
Sophisticated reprogramming of the TME impacts tumor progression and responsiveness to both chemotherapeutic agents and the validity of immunotherapeutic modalities35. TME differs considerably between the two ARGS subgroups, especially in the TME. The two groups differed in immune functional status, including cytolytic activity, MHC class I expression, HLA, and IFN response. The cytolytic activity of immune cells reflects the ability to kill tumor cells. Long-term management of tumors by CD8 + T cells requires sustained integration of IFN-I/IFN-II signaling as a prerequisite36. CD8 + T cells induce IFNγ-dependent repolarization of M1-like/IFNβ-producing macrophages37. Increased MHC-I expression with high T-cell infiltration favors the prognosis of patients with BLCA. In our analysis, MHC-I expression was higher in patients in the high-risk group. Therefore, the impact of MHC-I expression on the prognosis of BLCA patient needs to be clarified. Additionally, activation of the IFN response is an essential aspect of anti-tumor immunity38. This suggests that patients in the low-risk group have better anti-tumor immune activity. The TCIA immunotherapy endpoint diagnosis, a significantly better immune prospective treatment efficacy was found in patients with tumors in the low-risk cohort than patients with tumors in the high-risk cohort. Therefore, improving the immune microenvironment of the TME is necessary to reverse poor prognosis of patients in the high-risk group. The immunophenotyping analysis identified the high-risk group as mainly C1 and C2, indicating that patients in the high-risk group were more sensitive to immunotherapy. However, remodeling of the TME leads to poor immunotherapy results. Tumor infiltrating immune cells are one of the most critical components of the TME, and tumor cells are highly correlated with stemness, angiogenesis, and EMT in TNBC39. These dysfunctional vessels lead to a hypoxic environment where tumors thrive and eventually metastasize to secondary sites40. The Tregs are a critical agent of immune homogenization. Correlation with VEGF signaling and Treg viability and proplasia specifically within tumors is essential41. Intratumor vasculature that is intricate, jumbled, erratic, and leaky causes hypoxia and precludes anti-tumor drugs from being delivered productively throughout the TME. Alternatively, vascular endothelial growth factor may directly impact cancer cells or cancer stem cells. Isolated angiogenesis experiments have indicated that B cell mediated tumor vasculogenesis is predominantly contingent on the triggering of pro-angiogenic gene expression, which necessitates STAT3 signalling in B cells42. An important type of immune cells that infiltrate tumors are tumor-associated macrophages (TAMs). These cells can be classified into two subtypes: M1 macrophages, which are classically activated, and M2 macrophages, which are alternatively activated43. Macrophages with a preference for the M2 phenotype have a higher angiogenic potential than other subpopulations44. It has been demonstrated in many studies that both M1 and M2 macrophages can induce tumor angiogenesis in different settings, which undoubtedly leads to poor patient prognosis. Our findings are in agreement with these conclusions. Low-risk group uncharacterized cells, myeloid dendritic cells, plasma cells, and class-switched memory B cells exhibited higher levels of infiltration. On the other hand, the high-risk group had higher myeloid dendritic cell, B-cell, mast cell, monocyte, M1 macrophage, and M2 macrophage levels. The most important of these cell types was the cancer-associated fibroblasts, which were heavily infiltrated in patients in the high-risk group. This indicates that poor prognosis in patients in the high-risk group was highly correlated with tumor-associated fibroblasts (P < 0.05, R > 0.4). Our findings suggest that the immune cells of patients in the high-risk cohort were compromised through angiogenesis. The TME is more inclined to be in an immunosuppressed state.
In addition to the accumulation of immune cells that negatively regulate immune activity, the immunosuppressive TME was associated with the upregulation of the expression of suppressive immune checkpoints45. We further explored the differences in suppressor molecule expression patterns between the ARGS subgroups. In the risk-prognosis model, most immune checkpoints, such as CD274, CD86, and CTLA-4, were positively correlated with risk scores. Moreover, only TNF superfamily expression was significantly negatively correlated with risk score, suggesting that poor prognosis of patients in the high-risk group was associated with blocked deletion of expression of most immune checkpoints. Recently, immunotherapy targeting suppressive immune checkpoints has shown promising efficacy in treating BLCA46. Based on our results, we hypothesize that patients in the low-risk group may benefit more from immunotherapy and immunotherapy effects in the high-risk group tend to indicate poor prognosis due to checkpoint suppression caused by massive immune cell infiltration and remodeling of the TME47. In this study, BLCA patients with high-risk scores had low suppressive immune checkpoint expression and low IPS and were more tolerant to immunotherapy. For comparison, patients in the low-risk cohort were characterized by high inhibitory immune checkpoint resistance and high IPS. In summary, TNFSF15, TNFRSF25, and TNFRSF14 can potentially be immunotherapy targets for improving poor prognosis in patients in the high-risk group. Thus, the ARGS may be of good value to precisely predict which patients are likely to respond to immunotherapy.
The results obtained from the integrated algorithms glmBoost, LASSO, RF, GBM, Ridge, Stepglm, Enet, LDA, NaiveBayes, plsRglm, SVM, XGBoost, CoxBoost, RSF, plsRcox, StepCox, survivalSVM, and SuperPC are consistent with the AUC and calibration curves of the angiogenesis model. This further establishes that the BLCA angiogenesis model we have developed can accurately assess patient prognosis and demonstrates the significance of angiogenesis in BLCA progression and the remodeling of the BLCA TME. The study by Nan Zhang et al. indicated that the integrated algorithms possess strong testing capabilities, and the computational framework exhibits novelty48. By combining SVM, WGCNA, RF, and LASSO machine learning methods, we conducted screening and prediction of diagnostic and prognostic biomarkers for BLCA using the GEO database BLCA dataset. MYH11, FGF9, JAM3, BIN1, and SRPX were identified as potential diagnostic markers for BLCA. The model suggests that these genes can effectively assess the progression and prognosis of BLCA patients . GSVA enrichment analysis reveals that the downregulation of MYH11, FGF9, JAM3, BIN1, and SRPX gene expression is associated with enhanced tumor cell proliferation, migration, and differentiation, indicating poor prognosis in BLCA patients. Among them, MYH11 shows the most potential as a novel diagnostic and prognostic biomarker for BLCA. Another study, utilizing KM survival analysis and Cox regression, also suggested that MYH11 may be a potential biomarker for BLCA49. In a large-scale lung cancer sequencing study, MYH11 + αSMA + cancer-associated fibroblasts (CAFs) were found to be present in early-stage tumors and formed a single-cell layer known as cancer nests. These CAFs drive a distinct molecular program that leads to T cell exclusion. Targeting these CAFs is expected to enhance the effectiveness of immunotherapy in patients with T cell-excluded tumors50. In summary, MYH11 holds the potential to serve as a biomarker for early diagnosis and clinical prognosis of BLCA. Additionally, MYH11 may be associated with TME remodeling and the predictive efficacy of immunotherapy.
Accurate analysis and acquisition of pharmacokinetic parameters are crucial for the clinical application of drugs. Currently, the determination of these parameters for new drug development primarily relies on methods such as data analysis and physiological model construction. However, the obtained results often deviate significantly from the actual situation, requiring substantial human and material resources51. Through risk differential analysis and the PPI network, FN1 has the highest correlation with the risk coefficient of BLCA angiogenesis. Previous studies have indicated that FN1 is closely associated with immune and inflammatory response characteristics in BLCA and can serve as a high-precision predictive biomarker for advanced clinical features, as well as for the response to immunotherapy and chemotherapy52. Another study on gastric cancer revealed that the 3’-UTR (3’-untranslated region) of FN1 exhibits a stronger oncogenic effect compared to the FN1 protein. Therefore, the FN1 3’-UTR may be a more suitable therapeutic target for developing gastric cancer-targeted drugs53. FN1 is associated with the growth and metastasis of BLCA. In addition, FN1 has strong potential for predicting immunotherapy54,55. Therefore, confirming FN1 as a therapeutic target for BLCA holds promising research prospects. Consistent with our research findings, FN1 is associated with tumor cell proliferation, migration, and differentiation.
ADCs are a novel therapeutic modality that has the potential to revolutionize the paradigm of cancer chemotherapy56. ADCs consist of humanized or human monoclonal antibodies chemically linked to highly cytotoxic small molecules, known as payloads57. A pharmacophore model was constructed based on the known ADC drug payload combinations, focusing primarily on hydrogen bond acceptor and hydrophobic features. Pharmacophore model suggests that ADC payload combinations tend to exhibit higher polarity and cellular retention. ADCs currently face challenges such as heterogeneity, bystander effect, protein aggregation, low internalization efficiency or poor tumor cell penetration, narrow therapeutic index, and the emergence of resistance58. The hydrophobic feature is closely associated with the bystander effect of ADCs. The pharmacophore and ADME-integrated molecular docking technique explored the targeting inhibition of FN1 by ten natural compounds and their potential as ADC payload combinations. The results showed that the small-molecule ligand SSD exhibited the lowest binding energy when binding to the receptor protein FN1. The ligand-protein complex structure was the most stable with SSD, indicating its potential as a targeted inhibitor of FN1 and as a suitable ADC payload combination. The anti-tumor effects of SSD are multi-targeted and can be achieved through various mechanisms, including inhibiting proliferation, invasion, metastasis, and angiogenesis, as well as inducing cell apoptosis, autophagy, and differentiation59. However, the ADME properties of SSD in BLCA and its specific mechanisms of action against BLCA are not yet clear. ADME analysis revealed that the synthesis and extraction of SSD pose certain challenges. However, this does not hinder its potential as a promising drug targeting FN1 for anti-angiogenesis in BLCA. In addition, TOPKAT utilizes QSAR models trained on a wide range of toxicology databases to predict potential risks for multiple endpoints60, 61–62. This study systematically predicted the toxicity characteristics of SSD using the TOPKAT model, and the results showed that SSD exhibited low-risk characteristics in the multiple key toxicity endpoints, providing preliminary theoretical support for its safety as a candidate drug. Previous studies have shown that the toxic effects of SSD mainly include hepatotoxicity, neurotoxicity, hemolysis, and cardiotoxicity, making it a potentially effective and relatively safe natural anti-tumor substance63,64.
The ADC binds specifically to the antigen, which is overexpressed on the surface of cancer cells, resulting in a reduced bystander effect and an increased therapeutic index65. Pharmacophore simulation showed that among the four selected compounds, SSD primarily exhibited hydrogen bond acceptor features and had fewer hydrophobic features. Previous studies have indicated that SSD is a bioactive compound derived from plant secondary metabolism and possesses potent anti-tumor abilities. It has been shown to block the cell cycle and activate cell death receptors and the mitochondrial apoptotic pathway66. In addition, the latest research shows that SSD has strong potential for anti-angiogenesis, demonstrating promising research prospects63,67. Therefore, SSD demonstrated good cellular retention and a reduced bystander effect.
Conclusion
In summary, in this study, we constructed a novel ARGS with high value in predicting prognosis and reflecting the TME in BLCA. ARGS is closely related to clinical outcomes and can be an independent prognostic indicator. In addition, patients with different ARGS subgroup scores had different TME statuses, including the degree of stromal cell and immune cell infiltration, immune activity, immune checkpoint expression, and drug sensitivity. Therefore, ARGS is a promising biomarker for inferring the molecular and immune characteristics of BLCA, which may provide new therapeutic strategies for BLCA treatment. Using machine learning, we explored the novel biomarker MYH11 for BLCA diagnosis and prognosis. Additionally, we applied AIDD techniques to identify and validate the natural compound SSD as a potential next-generation targeted inhibitor of FN1 and as an ADC payload combination.
Acknowledgements
We would like to express our sincere thanks to American Journal Experts for polishing the manuscript.
Author contributions
YTL and HWL designed the study. YTL and LZ contributed to the data analysis and manuscript writing. XYS, XD and YYH performed data collection and analysis of public datasets. KZ and XD conducted initial data analysis and visualization. HWL reviewed and modified the manuscript. All authors reviewed the manuscript and agreed to its publication.
Funding
This study was financially supported by Guangdong Basic and Applied Basic Research Foundation (2022A1515012195, 2024A1515012742), Medical Scientific Research Foundation of Guangdong Province (A2023290, A2024489), Zhanjiang Science and Technology Plan Project (2021A05091, 2022A01141), High-level talents scientific research start-up funds of the Affiliated Hospital of Guangdong Medical University (GCC2024021), the Clinical and Basic Science Innovation Special Program of Guangdong Medical University (GDMULCJC2024054), and Guangdong Medical University Undergraduate Innovation and Entrepreneurship Training Program (GDMUCX2024268, JDXM2024040F, 202510571017).
Data availability
The datasets analyzed during this study are available in the TCGA and GEO databases at https://portal.gdc.cancer.gov/and https://www.ncbi.nlm.nih.gov/geo/. The collected data used in the research are available from the corresponding author upon enquiry. The illustrations used in the workflow diagrams are from Bioicons (https://bioicons.com/) and are used with permission.
Declarations
Ethics approval and consent to participate
No ethical approval was required for this study.
Consent for publication
The published version of the manuscript has been read and approved by all authors.
Competing interests
The authors declare no competing interests.
Conflict of interest
The authors state that this study was conducted without commercial or financial relationships that could be interpreted as potential conflicts of interest.
Data sharing and data accessibility
The datasets analyzed during this study are available in the TCGA and GEO databases at https://portal.gdc.cancer.gov/ and https://www.ncbi.nlm.nih.gov/geo/. The collected data used in the research are available from the corresponding author upon enquiry. The illustrations used in the workflow diagrams are from Bioicons (https://bioicons.com/) and are used with permission.
Abbreviations
Bladder cancer
Tumor microenvironment
Angiogenesis-related gene signatures
Antibody-drug conjugates
Artificial intelligence-driven drug design
Monoclonal antibodies
Artificial intelligence
Machine learning
Deep learning
The Cancer Genome Atlas
False discovery rate
Gene Ontology
Kyoto Encyclopedia of Genes and Genomes
Receiver operating characteristic
t-distributed stochastic neighbor embedding
Kaplan-Meier
Overall survival
Principal component analysis
Gene set variation analysis
Half maximal inhibitory concentration
Immunohistochemistry
Support vector machine
Random forest
Batch gradient descent for deep learning
Weighted gene co-expression network analysis
Discovery Studio
Protein-protein interaction
Absorption, distribution, metabolism, and excretion
Area under curve
Regular T cells
Immunocyte positive scores
Topological polar surface area
Estimated solubility
Ali’s rule of five
Copy number alterations
Tumor-associated macrophages
Cancer-associated fibrolasts
Saikosaponin D
Mahalanobis distance
Toxicity Prediction by Komputer Assisted
Quantitative structure-activity relationship
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
1. Dyrskjøt, L et al. Bladder cancer. Nat. Rev. Dis. Primers; 2023; 9, 58. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37884563][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11218610]
2. Lopez-Beltran, A; Cookson, MS; Guercio, BJ; Cheng, L. Advances in diagnosis and treatment of bladder cancer. BMJ; 2024; 384, e076743.1:CAS:528:DC%2BB2MXlt12rsLs%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38346808]
3. Richters, A; Aben, KKH; Kiemeney, L. The global burden of urinary bladder cancer: an update. World J. Urol.; 2020; 38, pp. 1895-1904. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31676912]
4. Liu, L et al. Impact of tumour stroma-immune interactions on survival prognosis and response to neoadjuvant chemotherapy in bladder cancer. EBioMedicine; 2024; 104, 105152.1:CAS:528:DC%2BB2cXhtVyhsb7E [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38728838][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11090066]
5. Qing, X et al. Molecular characteristics, clinical significance, and cancer immune interactions of angiogenesis-associated genes in gastric cancer. Front. Immunol.; 2022; 13, 843077.1:CAS:528:DC%2BB38XhtFOktb7O [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35273618][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8901990]
6. Majidpoor, J; Mortezaee, K. Angiogenesis as a hallmark of solid tumors-clinical perspectives. Cell. Oncol. (Dordr); 2021; 44, pp. 715-737.1:CAS:528:DC%2BB3MXhtF2nsLbP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33835425]
7. Fu, Z; Li, S; Han, S; Shi, C; Zhang, Y. Antibody drug conjugate: the biological missile for targeted cancer therapy. Signal. Transduct. Target. Ther.; 2022; 7, 93.1:CAS:528:DC%2BB38XhsVOmt7jM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35318309][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8941077]
8. Vemula, D; Jayasurya, P; Sushmitha, V; Kumar, YN; Bhandari, V. CADD, AI and ML in drug discovery: A comprehensive review. Eur. J. Pharm. Sci.; 2023; 181, 106324.1:CAS:528:DC%2BB38XjtF2rtrjL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36347444]
9. Liberzon, A et al. The molecular signatures database (MSigDB) hallmark gene set collection. Cell. Syst.; 2015; 1, pp. 417-425.1:CAS:528:DC%2BC2sXhtFaltLc%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26771021][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4707969]
10. Zheng, R et al. Cistrome data browser: expanded datasets and new tools for gene regulatory analysis. Nucleic Acids Res.; 2019; 47, pp. D729-d735.2019pgrb.book...Z1:CAS:528:DC%2BC1MXhs1Gqsb3P [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30462313]
11. Beck, A; Goetsch, L; Dumontet, C; Corvaïa, N. Strategies and challenges for the next generation of antibody-drug conjugates. Nat. Rev. Drug Discov; 2017; 16, pp. 315-337.1:CAS:528:DC%2BC2sXkt1Onsbc%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28303026]
12. Alas, M; Saghaeidehkordi, A; Kaur, K. Peptide-drug conjugates with different linkers for cancer therapy. J. Med. Chem.; 2021; 64, pp. 216-232.1:CAS:528:DC%2BB3MXitV0%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33382619]
13. Kanehisa, M; Furumichi, M; Sato, Y; Matsuura, Y; Ishiguro-Watanabe, M. KEGG: Biological systems database as a model of the real world. Nucleic Acids Res.; 2025; 53, pp. D672-d677. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39417505]
14. Kanehisa, M. Toward Understanding the origin and evolution of cellular organisms. Protein Sci.; 2019; 28, pp. 1947-1951.1:CAS:528:DC%2BC1MXhslaisL3I [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31441146][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6798127]
15. Zheng, H et al. Comprehensive pan-cancer analysis reveals NUSAP1 is a novel predictive biomarker for prognosis and immunotherapy response. Int. J. Biol. Sci.; 2023; 19, pp. 4689-4708.1:CAS:528:DC%2BB3sXitlans7fO [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37781040][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10535697]
16. Gul, S et al. Stemness signature and targeted therapeutic drugs identification for triple negative breast Cancer. Sci. Data; 2023; 10, 815.1:CAS:528:DC%2BB3sXisFGjur%2FO [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37985782][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10662149]
17. Lu, H; Wu, J; Liang, L; Wang, X; Cai, H. Identifying a novel defined Pyroptosis-Associated long noncoding RNA signature contributes to predicting prognosis and tumor microenvironment of bladder Cancer. Front. Immunol.; 2022; 13, 803355.1:CAS:528:DC%2BB38XhtFemtbrP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35154117][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8828980]
18. Geeleher, P; Cox, N; Huang, RS. pRRophetic: an R package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PLoS One; 2014; 9, e107468.2014PLoSO..9j7468G [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25229481][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4167990]
19. Zhao, F; Chen, M; Wu, T; Ji, M; Li, F. Integration of single-cell and bulk RNA sequencing to identify a distinct tumor stem cells and construct a novel prognostic signature for evaluating prognosis and immunotherapy in LUAD. J. Transl Med.; 2025; 23, 222.1:CAS:528:DC%2BB2MXks1Gntbg%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39987127][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11847374]
20. Shi, Y et al. Crosstalk of ferroptosis regulators and tumor immunity in pancreatic adenocarcinoma: novel perspective to mRNA vaccines and personalized immunotherapy. Apoptosis; 2023; 28, pp. 1423-1435.1:CAS:528:DC%2BB3sXhtlGqur%2FJ [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37369808][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10425492]
21. Zheng, Y et al. Bao Yuan Decoction alleviates fatigue by restraining inflammation and oxidative stress via the AMPK/CRY2/PER1 signaling pathway. J. Ethnopharmacol.; 2024; 328, 118058.1:CAS:528:DC%2BB2cXhtVaqsrfI [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38513778]
22. Li, Y. et al. Kushenol O regulates GALNT7/NF-κB axis-Mediated M2 macrophage polarization and efferocytosis in papillary thyroid carcinoma. Phytomedicine 15:6373. (2025).
23. Charoentong, P et al. Pan-cancer Immunogenomic analyses reveal Genotype-Immunophenotype relationships and predictors of response to checkpoint Blockade. Cell. Rep.; 2017; 18, pp. 248-262.1:CAS:528:DC%2BC2sXjvFKntw%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28052254]
24. Ye, B et al. iMLGAM: integrated machine learning and genetic Algorithm-driven multiomics analysis for pan-cancer immunotherapy response prediction. Imeta; 2025; 4, e70011.1:CAS:528:DC%2BB2MXlslynsLw%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/40236779][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11995183]
25. Liang, C et al. Hypericin photoactivation induces triple-negative breast cancer cells pyroptosis by targeting the ROS/CALR/Caspase-3/GSDME pathway. J. Adv. Res.; 2025; 96, 32.
26. Fucikova, J; Spisek, R; Kroemer, G; Galluzzi, L. Calreticulin and cancer. Cell. Res.; 2021; 31, pp. 5-16.1:CAS:528:DC%2BB3cXhsV2murrP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32733014]
27. Jiang, Y et al. RNA-Binding protein COL14A1, TNS1, NUSAP1 and YWHAE are valid biomarkers to predict peritoneal metastasis in gastric Cancer. Front. Oncol.; 2022; 12, 830688.1:CAS:528:DC%2BB3sXhtlGqsbnL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35515139][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9062106]
28. Cañellas-Socias, A et al. Metastatic recurrence in colorectal cancer arises from residual EMP1(+) cells. Nature; 2022; 611, pp. 603-613.2022Natur.611.603C [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36352230][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7616986]
29. Huang, CK; Sun, Y; Lv, L; Ping, Y. ENO1 and Cancer. Mol. Ther. Oncolytics; 2022; 24, pp. 288-298.1:CAS:528:DC%2BB38XhtVyht7zI [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35434271][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8987341]
30. Wang, L., Wang, Y., Wang, J., Li, L. & Bi, J. Identification of a prognosis-related risk signature for bladder cancer to predict survival and immune landscapes. J Immunol Res 201:3236384. (2021).
31. Wang, Y; Hu, L; Zheng, Y; Guo, L. HMGA1 in cancer: Cancer classification by location. J. Cell. Mol. Med.; 2019; 23, pp. 2293-2302.1:CAS:528:DC%2BC1MXmtlWgs7k%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30614613][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6433663]
32. Cazes, A; Childers, BG; Esparza, E; Lowy, AM. The MST1R/RON tyrosine kinase in cancer: Oncogenic functions and therapeutic strategies. Cancers (Basel); 2022; 33, 14.
33. Zhang, P et al. Novel post-translational modification learning signature reveals B4GALT2 as an immune exclusion regulator in lung adenocarcinoma. J. Immunother. Cancer; 2025; 96, 13.
34. Ye, B et al. Navigating the immune landscape with plasma cells: A pan-cancer signature for precision immunotherapy. Biofactors; 2025; 51, e2142.1:CAS:528:DC%2BB2cXisVShurbM [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39495620]
35. Li, Y; Song, X; Huang, Y; Zhou, S; Zhong, L. Genetic associations of plasma metabolites with immune cells in hyperthyroidism revealed by Mendelian randomization and GWAS-sc-eQTLs xQTLbiolinks analysis. Sci. Rep.; 2025; 84, 15.
36. Farhood, B; Najafi, M; Mortezaee, K. CD8(+) cytotoxic T lymphocytes in cancer immunotherapy: A review. J. Cell. Physiol.; 2019; 234, pp. 8509-8521.1:CAS:528:DC%2BC1MXjtlGgsLY%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30520029]
37. Leblond, MM et al. CD40 agonist restores the antitumor efficacy of Anti-PD1 therapy in Muscle-Invasive bladder Cancer in an IFN I/II-Mediated manner. Cancer Immunol. Res.; 2020; 8, pp. 1180-1192.1:CAS:528:DC%2BB3cXit1Kns7jL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32661095]
38. Takahashi, M et al. The tumor suppressor kinase DAPK3 drives tumor-intrinsic immunity through the STING-IFN-β pathway. Nat. Immunol.; 2021; 22, pp. 485-496.1:CAS:528:DC%2BB3MXnt1Okt7Y%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33767426][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8300883]
39. Gao, J et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal.; 2013; 6, pl1. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23550210][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4160307]
40. Kabir, AU et al. Dual role of endothelial Myct1 in tumor angiogenesis and tumor immunity. Sci. Transl. Med.; 2021; 89, 13.
41. Lužnik, Z; Anchouche, S; Dana, R; Yin, J. Regulatory T cells in angiogenesis. J. Immunol.; 2020; 205, pp. 2557-2565. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33168598]
42. Hu, S et al. Glycoprotein PTGDS promotes tumorigenesis of diffuse large B-cell lymphoma by MYH9-mediated regulation of Wnt-β-catenin-STAT3 signaling. Cell. Death Differ.; 2022; 29, pp. 642-656.1:CAS:528:DC%2BB38Xktlyju7Y%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34743203]
43. Pan, Y; Yu, Y; Wang, X; Zhang, T. Tumor-Associated macrophages in tumor immunity. Front. Immunol.; 2020; 11, 583084.1:CAS:528:DC%2BB3MXitVCisb0%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33365025][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7751482]
44. Najafi, M et al. Macrophage Polarity in cancer: A review. J. Cell. Biochem.; 2019; 120, pp. 2756-2765.1:CAS:528:DC%2BC1cXhvVahsbbL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30270458]
45. Sangro, B; Sarobe, P; Hervás-Stubbs, S; Melero, I. Advances in immunotherapy for hepatocellular carcinoma. Nat. Rev. Gastroenterol. Hepatol.; 2021; 18, pp. 525-543. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33850328][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8042636]
46. Hu, J et al. Siglec15 shapes a non-inflamed tumor microenvironment and predicts the molecular subtype in bladder cancer. Theranostics; 2021; 11, pp. 3089-3108.1:CAS:528:DC%2BB3MXmtF2js7g%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33537076][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7847675]
47. Martínez-Reyes, I; Chandel, NS. Cancer metabolism: looking forward. Nat. Rev. Cancer; 2021; 21, pp. 669-680. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34272515]
48. Zhang, N et al. Machine learning-based identification of tumor-infiltrating immune cell-associated LncRNAs for improving outcomes and immunotherapy responses in patients with low-grade glioma. Theranostics; 2022; 12, pp. 5931-5948.1:CAS:528:DC%2BB38Xit12nur%2FE [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35966587][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9373811]
49. Hu, J et al. The identification of new biomarkers for bladder cancer: A study based on TCGA and GEO datasets. J. Cell. Physiol.; 2019; 234, pp. 15607-15618.2019fpts.book...H1:CAS:528:DC%2BC1MXjtFOkt74%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30779109]
50. Grout, JA et al. Spatial positioning and matrix programs of Cancer-Associated fibroblasts promote T-cell exclusion in human lung tumors. Cancer Discov; 2022; 12, pp. 2606-2625.1:CAS:528:DC%2BB3sXht1emtrs%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36027053][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9633420]
51. Wang, Y; Zhan, Y; Liu, C; Zhan, W. Application of machine learning technology in the prediction of ADME- related Pharmacokinetic parameters. Curr. Med. Chem.; 2023; 30, pp. 1945-1962.1:CAS:528:DC%2BB3sXotFelsLo%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35993465]
52. Chang, Z et al. Distinct immune and inflammatory response patterns contribute to the identification of poor prognosis and advanced clinical characters in bladder cancer patients. Front. Immunol.; 2022; 13, 1008865.1:CAS:528:DC%2BB38XivFOis7bO [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36389789][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9646535]
53. Pan, S et al. FN1 mRNA 3’-UTR supersedes traditional fibronectin 1 in facilitating the invasion and metastasis of gastric cancer through the FN1 3’-UTR-let-7i-5p-THBS1 axis. Theranostics; 2023; 13, pp. 5130-5150.1:CAS:528:DC%2BB3sXitFOhsbjK [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37771777][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10526670]
54. Xiao, Z; Cai, Z; Deng, D; Tong, S; Zu, X. An EMT-based risk score thoroughly predicts the clinical prognosis, tumor immune microenvironment and molecular subtypes of bladder cancer. Front. Immunol.; 2022; 13, 1000321.1:CAS:528:DC%2BB38Xis1WksbfJ [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36211349][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9540509]
55. Li, B et al. Comprehensive analysis of RNA methylation-related genes to identify molecular cluster for predicting prognosis and immune profiles in bladder cancer. Sci. Rep.; 2025; 15, 9147.1:CAS:528:DC%2BB2MXot1aht7c%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/40097551][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11914693]
56. Liu, K et al. A review of the clinical efficacy of FDA-approved antibody–drug conjugates in human cancers. Mol. Cancer; 2024; 23, 62.1:CAS:528:DC%2BB2cXnt1Sls7Y%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38519953][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10960395]
57. Tsuchikama, K; Anami, Y; Ha, SYY; Yamazaki, CM. Exploring the next generation of antibody-drug conjugates. Nat. Rev. Clin. Oncol.; 2024; 21, pp. 203-223.1:CAS:528:DC%2BB2cXmvVWhsg%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38191923]
58. Maiti, R et al. Antibody drug conjugates as targeted cancer therapy: past development, present challenges and future opportunities. Arch. Pharm. Res.; 2023; 46, pp. 361-388.2023omte.book...M1:CAS:528:DC%2BB3sXotFKntbY%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37071273][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11345756]
59. Li, X; Li, X; Huang, N; Liu, R; Sun, R. A comprehensive review and perspectives on Pharmacology and toxicology of saikosaponins. Phytomedicine; 2018; 50, pp. 73-87.1:CAS:528:DC%2BC1cXhvVKrs7vN [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30466994][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7126585]
60. Zhang, A et al. Development of an MCL-1-related prognostic signature and inhibitors screening for glioblastoma. Front. Pharmacol.; 2023; 14, 1162540.1:CAS:528:DC%2BB3sXhs1KqsrfL [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37538176][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10394558]
61. Benfenati, E; Gini, G. Computational predictive programs (expert systems) in toxicology. Toxicology; 1997; 119, pp. 213-225.1:CAS:528:DyaK2sXjtlGrs7w%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9152017]
62. Bhhatarai, B; Wilson, DM; Parks, AK; Carney, EW; Spencer, PJ. Evaluation of TOPKAT, toxtree, and Derek Nexus in Silico models for ocular irritation and development of a Knowledge-Based framework to improve the prediction of severe irritation. Chem. Res. Toxicol.; 2016; 29, pp. 810-822.1:CAS:528:DC%2BC28XkvVyjtL4%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27018716]
63. Manoharan, S; Deivendran, B; Perumal, E. Chemotherapeutic potential of Saikosaponin D: Experimental evidence. J. Xenobiot; 2022; 12, pp. 378-405.1:CAS:528:DC%2BB3sXnt1Ohtw%3D%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36547471][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9782205]
64. Zhou, P et al. Saikosaponin D: review on the antitumour effects, toxicity and pharmacokinetics. Pharm. Biol.; 2021; 59, pp. 1480-1489.1:CAS:528:DC%2BB3MXisVSlsL7N [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34714209]
65. Samantasinghar, A et al. A comprehensive review of key factors affecting the efficacy of antibody drug conjugate. Biomed. Pharmacother; 2023; 161, 114408.1:CAS:528:DC%2BB3sXktVaitro%3D [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36841027]
66. Tang, TT et al. Saikosaponin D exerts cytotoxicity on human endometrial cancer Ishikawa cells by inducing apoptosis and inhibiting metastasis through MAPK pathways. Food Chem. Toxicol.; 2023; 177, 113815.1:CAS:528:DC%2BB3sXhtVKhtLzP [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37209937]
67. Wu, Y; Zhang, J; Wang, X; Xu, Y; Zheng, J. Saikosaponin-d regulates angiogenesis in idiopathic pulmonary fibrosis through angiopoietin/Tie-2 pathway. Acta Histochem.; 2023; 125, 152100.1:CAS:528:DC%2BB3sXitFKksr3N [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37837833]
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© The Author(s) 2025. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Bladder cancer (BLCA) is a prevalent urological malignancy that exhibits a high degree of tumor heterogeneity and morbidity. Tumor angiogenesis, a vital hallmark of cancer, greatly influences the tumor microenvironment (TME). The emergence of anti-angiogenic drugs has provided a new turning point in cancer treatment. An integrated machine learning system was constructed to build the angiogenesis-related gene signatures (ARGS). ARGS was used to assess TME status in BLCA. Pharmacophore construction was employed to construct pharmacophore features of highly cytotoxic drug payload combinations for antibody-drug conjugates (ADCs). In addition, we developed a natural compound using artificial intelligence-driven drug design technology. This compound exhibits anti-angiogenic effects in BLCA and serves as a highly cytotoxic drug payload for ADCs. Multi-dimensional machine learning was used to screen biomarkers for evaluating the post-treatment effects of drug therapy in BLCA. The ARGS consists of 12 angiogenesis-related genes associated with prognostic risk in BLCA. The ARGS divides BLCA patients into high-risk and low-risk groups. Significant TME remodeling was identified in the high-risk BLCA cohort and demonstrated a strong association with tumor angiogenesis. Expression levels of key immune checkpoint markers significantly differed between BLCA risk groups. Saikosaponin D (SSD) shows promising potential as a novel ADC drug for anti-angiogenic treatment in BLCA. Multi-dimensional machine learning results indicate that MYH11 is the most likely biomarker for evaluating the post-treatment effects of SSD therapy. SSD may potentially treat tumors by regulating angiogenesis in BLCA. The detection of MYH11 can be used to assess the therapeutic effectiveness of SSD in BLCA.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
1 Affiliated Hospital of Guangdong Medical University, Laboratory of Urology, Zhanjiang, China (GRID:grid.410560.6) (ISNI:0000 0004 1760 3078)
2 The Second Affiliated Hospital of Guangdong Medical University, Department of Traditional Chinese Medicine, Zhanjiang, China (GRID:grid.410560.6) (ISNI:0000 0004 1760 3078)