A Hypoxia and ER Stress-Related Gene Signature for Prognostic Prediction in Glioblastoma: An Integrated Machine Learning Approach

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.