Objectives Bladder cancer is a highly heterogeneous malignancy characterized by complex immune interactions that influence disease progression and patient outcomes. Identifying immune-related molecular determinants with prognostic and therapeutic relevance is therefore critical. This study aimed to systematically investigate immune-related genes involved in bladder cancer and to identify key predictive molecular prognostic biomarkers, with particular emphasis on cell-cycle–associated immune mechanisms. Methods RNA expression profiles from bladder tumor and corresponding normal tissues were analyzed across five independent cohorts (GSE13507, GSE19915, GSE40355, GSE42089, and TCGA-Bladder Cancer). Common differentially expressed immune-related genes (DEIRGs) were identified using GEO2R and TCGA analyses. Functional enrichment and pathway analyses were performed using Metascape and DAVID, while hub genes were identified through protein–protein interaction networks using MCODE, CytoHubba, and ClueGO. Expression patterns, prognostic relevance, and immune associations were validated using UALCAN, the Human Protein Atlas, TIMER2.0, and PrognoScan databases. Results A total of 300 common DEIRGs were identified across datasets. Network analysis revealed 21 significant immune-related hub genes, several of which were consistently enriched in the G2M checkpoint pathway. Eleven hub genes were significantly associated with reduced overall survival in bladder cancer patients. Notably, six G2M checkpoint–related genes ( TOP2A, CCNA2, CDK1, BIRC5, PLK1, and E2F1 ) demonstrated strong prognostic relevance across multiple cohorts. Conclusions This comprehensive in silico analysis highlights the critical role of G2M checkpoint–associated immune-related genes in bladder cancer progression. These genes represent predictive molecular prognostic immune biomarkers and potential therapeutic targets, providing a foundation for the development of personalized immunotherapeutic strategies in bladder cancer.
Objective In randomized clinical trials, prediction models can be used to explore the relationships between patients’ variables and treatment effect magnitude. Our aim was to evaluate flexible machine learning models that incorporate interactions and nonlinear effects from high-dimensional data to estimate individualized treatment recommendations in trials with time-to-event outcomes. Methods We compared neural networks (CoxCC and CoxTime) and random survival forests, including Interaction Forests (IF) and Causal Survival Forest (CSF) against a Cox proportional hazards model with an adaptive LASSO (ALASSO) penalty as a benchmark. For individualized treatment recommendations, we adapted metrics originally designed for binary outcomes to accommodate time-to-event data with censoring, using both survival-probability and restricted mean survival time estimates of treatment benefit. Metrics included the C-for-Benefit, E50-for-Benefit, Integrated Calibration Index for benefit, and root-mean-squared error for treatment benefit. We also evaluated decision-making utility using the Area Under the Targeting Operator Characteristic and Population Benefit (PB). An extensive simulation study used two data generation processes incorporating nonlinearity and interactions. The models were applied to three cancer clinical trial datasets. Results In the first data generation process, neural networks outperformed ALASSO in calibration while IF showed superior C-for-benefit performance. In the second one, machine learning methods outperformed the benchmark linear ALASSO method across discrimination, calibration, and RMSE. In the cancer trial datasets, the machine learning methods often performed better than ALASSO: neural networks consistently achieved the best calibration, IF achieved the highest C-for-Benefit, and CSF showed strong calibration and the highest PB in the trastuzumab dataset. Conclusion Machine learning-based survival models can efficiently estimate conditional average treatment effects from randomized trials when strong nonlinear and interaction effects are expected. These flexible approaches offer valuable alternatives to traditional regression methods for developing treatment recommendation rules from clinical trials with time-to-event outcomes and medium to high-dimensional data set sizes.
Background Glioblastoma (GBM) is among the most aggressive primary brain tumors and carries a dismal prognosis. Hypoxia and endoplasmic reticulum (ER) stress have been proven to promote the development of GBM, but their combined role in the prognosis of GBM has yet to be clarified. Objective The goal of the present study was to characterize the molecular signature arising from the interplay of hypoxia and ER stress in GBM and to evaluate its prognostic value. Methods Differentially expressed genes (DEGs) between GBM and normal tissues were identified from the GSE116520 dataset using the limma package. Hypoxia- and ER stress-related DEGs were intersected to obtain a focused gene set. A prognostic model was constructed using LASSO Cox regression in the TCGA-GBM cohort. Weighted gene co-expression network analysis (WGCNA) was performed to identify key modules, and hub genes were further selected using Support Vector Machine (SVM) and Random Forest (RF) algorithms. The hub genes were validated through transcription factor analysis, methylation profiling, immune infiltration assessment, and single-cell RNA-seq analysis. Results A total of 265 HEDEGs were identified. A four-gene prognostic model comprising SPP1, CYGB, PLAUR, and HSPB1 stratified patients into groups with different overall survival outcomes in the training and internal validation datasets. Consensus clustering identified two molecular subtypes with distinct immune characteristics and survival outcomes. Integrated WGCNA and machine-learning analyses identified KCNMA1, ANGPT2, and XBP1 as candidate hub genes. Higher ANGPT2 and XBP1 expression was associated with poorer overall survival, whereas higher KCNMA1 expression was associated with better survival. Single-cell analysis revealed distinct cell-type-specific expression patterns of these genes within the GBM microenvironment. Conclusion This study developed a four-gene prognostic model and identified three candidate hub genes associated with hypoxia and ER stress in GBM. These findings provide a basis for further investigation of stress-related molecular heterogeneity and prognostic stratification, although independent external validation and experimental confirmation are required.
Background Whole-exome sequencing is a widely used technology to identify pathogenic variants in cancer. Although sequencing itself has become increasingly accessible, downstream analysis remains computationally complex, presenting a challenge for many researchers. Existing pipelines lack integrated support for somatic and germline variant detection and still require significant computational resources. Methods We developed GATES (GATK Automated Tool for Exome Sequencing), a lightweight pipeline that automates data preprocessing, variant calling, and variant annotation directly from raw paired-end FASTQ files through a simplified command-line interface. GATES implements the GATK Best Practices for somatic and germline variant detection and leverages Ensembl’s Variant Effect Predictor for functional annotation, outputting the results in a human-readable tab-separated values (TSV) file. We evaluated the pipeline’s performance using the SEQC-II benchmarking dataset and demonstrated its application using a clinical sample harboring known pathogenic germline and somatic variants. Results GATES was run on a standard laptop and performed end-to-end variant analysis for each sample within a few hours. In benchmarking with SEQC-II samples, germline and tumor-normal somatic variant calling modes demonstrated high concordance with their respective truth sets. Tumor-only somatic mode showed decreased accuracy, consistent with expected germline contamination. GATES demonstrated high performance across various hardware configurations and compared to the established nf-core/sarek pipeline. GATES further successfully performed somatic and germline analysis of a >100X clinical sample in under 7 hours. Importantly, the pipeline accurately distinguished the known KRAS p.G12V and KEAP1 p.S338L as somatic and germline, respectively. Conclusion By lowering the technical barriers to exome sequencing analysis, GATES provides a practical solution for pathogenic variant discovery for researchers both with and without computational expertise.
Objective Oral squamous cell carcinoma (OSCC) is characterized by high recurrence rates and an unclear underlying molecular basis, highlighting the need for improved early diagnostic strategies to enhance treatment outcomes and reduce healthcare burden. Given the regulatory role of microRNAs (miRNAs) in oral cancer progression through post-transcriptional control of target genes, this study aimed to investigate the diagnostic potential of miR-4303 and miR-4291 in OSCC. Methods Using TargetScan, MAPK1 was predicted as a common target of both miR-4303 and miR-4291. Expression levels of miR-4303, miR-4291, and MAPK1 were quantified by qRT-PCR in 30 paired OSCC and adjacent normal tissue samples. Diagnostic performance was evaluated using receiver operating characteristic (ROC) curve analysis. Results MAPK1 and both miRNAs (miR-4303 and miR-4291) were significantly upregulated in OSCC tissues compared to adjacent normal tissues (P < 0.0001). MAPK1 expression showed a significant association with higher tumor grade (P = 0.003) and advanced clinical stage (P = 0.005). ROC curve analysis demonstrated strong diagnostic performance, with AUC values of 0.92 for MAPK1, 0.731 for miR-4303 (P = 0.0001), and 0.707 for miR-4291 (P < 0.0001), indicating their potential utility in distinguishing OSCC from normal tissue. Conclusion These findings suggest that miR-4303 and miR-4291, together with MAPK1, may serve as promising diagnostic biomarkers in OSCC. However, further validation in larger, independent cohorts and functional studies is required to confirm their clinical applicability.
