# Integrative analysis of EMT-driving genes identifies a prognostic signature and GJB2 as a potential biomarker in glioblastoma

## TL;DR

This study identifies genes linked to tumor progression in glioblastoma and finds GJB2 as a potential biomarker for prognosis and treatment.

## Contribution

A novel EMT-based prognostic signature and the identification of GJB2 as a key driver in glioblastoma progression.

## Key findings

- 145 EMT-driving genes were identified by integrating survival data and EMT-related gene databases.
- An EMT-based prognostic signature was validated across five GBM cohorts and stratified patients into high- and low-risk groups.
- GJB2 was experimentally validated as an oncogenic driver promoting GBM cell proliferation and migration.

## Abstract

Glioblastoma (GBM) is an aggressive malignancy of the central nervous system characterized by rapid progression, therapeutic resistance, and poor prognosis. Epithelial–mesenchymal transition (EMT) contributes to tumor plasticity and immune remodeling in GBM. Identification of EMT-driving genes (EDGs) with prognostic and therapeutic relevance may provide insights into disease progression and precision management. This study aimed to systematically identify prognostic EDGs and to construct a robust EMT-based prognostic model for GBM.

Univariate Cox regression analyses were performed across five GBM cohorts (TCGA, CGGA-693, CGGA-325, GSE83300, and GSE74187) to rank genes according to hazard ratios, followed by Gene Set Enrichment Analysis (GSEA), which identified EMT as the pathway most strongly associated with adverse survival. EMT-related genes from MSigDB and dbEMT2.0 were intersected with TCGA-derived prognostic genes to obtain 145 EDGs. An EMT-related prognostic signature was generated using LASSO Cox regression and validated in independent datasets. Functional analyses, including Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), weighted gene co-expression network analysis (WGCNA), and GSEA, elucidated biological processes associated with the risk signature. Immune infiltration analyses (ESTIMATE, CIBERSORT) and drug sensitivity analyses characterized the tumor microenvironment and therapeutic response. Random Forest analysis and in vitro assays identified and validated GJB2 as a key mediator of GBM progression.

A total of 145 EDGs were identified by integrating survival-associated genes from TCGA with EMT-related genes from MSigDB and dbEMT2.0. An EMT-related prognostic signature was developed and validated across five GBM cohorts. The risk score stratified patients into high- and low-risk groups with significantly different survival outcomes and remained an independent prognostic factor in multivariate Cox analyses. A nomogram integrating the risk score with age, IDH mutation status, and MGMT promoter methylation demonstrated strong predictive performance. Immune profiling revealed that the high-risk group exhibited an “immune-inflamed yet immunosuppressed” phenotype characterized by elevated macrophage and regulatory T-cell infiltration. Drug sensitivity analyses suggested that high-risk GBM may respond better to paclitaxel and tamoxifen. Random Forest modeling and in vitro experiments identified GJB2 as an oncogenic driver that promotes GBM cell proliferation and migration.

Our findings provide a clinically applicable EMT-based prognostic framework that links transcriptional plasticity to patient outcomes in GBM and identify GJB2 as a promising therapeutic target.