Identification of heterogeneous subtypes and a prognostic model for gliomas based on mitochondrial dysfunction and oxidative stress-related genes

OBJECTIVE: Mitochondrial dysfunction and oxidative stress are known to involved in tumor occurrence and progression. This study aimed to explore the molecular subtypes of lower-grade gliomas (LGGs) based on oxidative stress-related and mitochondrial-related genes (OMRGs) and construct a prognostic model for predicting prognosis and therapeutic response in LGG patients. METHODS: A total of 223 OMRGs were identified by the overlap of oxidative stress-related genes (ORGs) and mitochondrial-related genes (MRGs). Using consensus clustering analysis, we identified molecular subtypes of LGG samples from TCGA database and confirmed the differentially expressed genes (DEGs) between clusters. We constructed a risk score model using LASSO regression and analyzed the immune-related profiles and drug sensitivity of different risk groups. The prognostic role of the risk score was confirmed using Cox regression and Kaplan-Meier curves, and a nomogram model was constructed to predict OS rates. We validated the prognostic role of OMRG-related risk score in three external datasets. Quantitative real-time PCR (qRT-PCR) and immunohistochemistry (IHC) staining confirmed the expression of selected genes. Furthermore, wound healing and transwell assays were performed to confirm the gene function in glioma. RESULTS: We identified two OMRG-related clusters and cluster 1 was significantly associated with poor outcomes (P<0.001). The mutant frequencies of IDH were significantly lower in cluster 1 (P<0.05). We found that the OMRG-related risk scores were significantly correlated to the levels of immune infiltration and immune checkpoint expression. High-risk samples were more sensitive to most chemotherapeutic agents. We identified the prognostic role of OMRG-related risk score in LGG patients (HR=2.665, 95%CI=1.626-4.369, P<0.001) and observed that patients with high-risk scores were significantly associated with poor prognosis (P<0.001). We validated our findings in three external datasets. The results of qRT-PCR and IHC staining verified the expression levels of the selected genes. The functional experiments showed a significant decrease in the migration of glioma after knockdown of SCNN1B. CONCLUSION: We identified two molecular subtypes and constructed a prognostic model, which provided a novel insight into the potential biological function and prognostic significance of mitochondrial dysfunction and oxidative stress in LGG. Our study might help in the development of more precise treatments for gliomas.