Construction And Validation of a Combined Immune and Metabolism-Related Genes Model for Prognosis Prediction in Glioma

Asare K

Published on: 2025-05-02

Abstract

Background: Glioma, characterized by its highly malignant nature and low survival rates, poses a significant challenge in terms of prognosis due to its complex etiology and high heterogeneity. Hence, there is a pressing need to identify new, reliable prognostic models for these tumors.

Methods: A comprehensive analysis was conducted by using public glioma gene expression datasets to identify immune and metabolism-related differentially expressed genes (IMRDEGs). The IMRDEGs were used for clustering glioma samples via non-negative matrix factorization (NMF) analysis, followed by differential analysis of the resulting clusters. A prognostic model was constructed using Cox regression and least absolute shrinkage and selection operator (LASSO) analysis. Internal and external validation of the model was performed using The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) datasets. Additionally, a nomogram was developed to assess the predictive performance of the model. The biological functions of this model were elucidated through gene set enrichment analysis (GSEA). Furthermore, the expression levels of the five genes in gliomas were validated using the Human Protein Atlas (HPA) database and reverse transcription quantitative polymerase chain reaction (qRT-PCR) in glioma tissues.

Results: A predictive model was developed based on five IMRDEGs: AK7, RDH10, DTYMK, PLXNA3, and NMNAT2. This model exhibited robust prognostic prediction capabilities across training, test, and GEO datasets. Both univariate and multivariate Cox regression analyses confirmed the model as a novel independent predictor of outcomes in glioma patients. Additionally, a nomogram demonstrated excellent predictive performance in accurately estimating survival rates at 1, 3, and 5 years for glioma patients. Compared to existing models, our model demonstrated superior performance in distinguishing overall survival (OS) between high and low-risk patients. GSEA revealed enrichment of biological functions related to the cell cycle and ECM receptor interactions within this model. Furthermore, qRT-PCR analysis of glioma tissues indicated high expression levels of DTYMK, RDH10, AK7, and PLXNA3, while NMNAT2 expression was found to be low in glioma tissues.

Conclusion: The immune and metabolism-related genes (IMRGs) prognosis risk model and their nomogram have important value for survival prediction and risk assessment of glioma patients.