Predicting Clinical Prognosis and Treatment Response in Glioblastoma Based on Gene Replication Stress-Related...

Glioblastoma (GBM) is a frequent malignant glioma among astrocytic tumors. Traditional pathological diagnostic criteria inadequately capture the underlying biological heterogeneity, highlighting the need for novel biomarkers to improve the diagnostic and prognostic accuracy of GBM.

Based on the GBM-related data from The Cancer Genome Atlas (TCGA), Replication Stress-related Genes (RSGs) were collected. The RSG feature scores were calculated by ssGSEA and combined with WGCNA to screen target module genes. Enrichment analysis of the modular signature genes was conducted using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. Prognostic genes were screened by univariate Cox and multivariate Cox regression analyses to establish a prognostic model for GBM. The Tumor Microenvironment (TME) was assessed using MCPcounter and TIMER methods. Immunotherapeutic responses and drug sensitivity responses were evaluated using the TIDE method. Finally, the role of PDCL3 in GBM was explored via immunohistochemistry, qRT-PCR, wound-healing, and transwell assays.

WGCNA identified RSG-associated module genes enriched in DNA replication, DNA-dependent ATPase activity, and other pathways. Four characterized genes (PDCL3, STEAP2, NXPH4, MPST) were selected to establish a Riskscore model for GBM. The risk score of the model was significantly related to the infiltration of myeloid dendritic cells, endothelial cells, and fibroblasts. The Riskscore was significantly positively correlated with the TIDE score, indicating higher immune escape potential in the high-risk group. The high-risk group was sensitive to Vinorelbine and Crizotinib. Through in vitro experiments, it was observed that knocking down PDCL3 noticeably inhibited the viability, migration, and invasion capacities of U87 cells.

This study validated the correlation between Oncogene-induced Replication Stress (ORS) characteristics and the prognosis of GBM, and screened RSGs to develop a Riskscore for GBM applying multiple types of regression analyses.

We constructed and verified a four-gene ORS-based prognosis model for GBM, linking replication stress to immune evasion and drug sensitivity for the first time. Experimental validation confirmed the pro-tumorigenic role of PDCL3, offering potential biomarkers and therapeutic targets.