Novel mitochondrial-related gene signature predicts prognosis and immunological status in glioma

Abstract

Background: Mitochondria are the center of cellular metabolism. The relationship between mitochondria and diseases has also been studied for a long time. However, the prognostic role of mitochondrial-related genes (MRGs) in patients with glioma and their biological effects are still unclear. The aim of the study was to construct a mitochondria-related model to assess prognosis and potential biological effects like immune infiltration, gene pathway and mutation, and give some predictive chemotherapeutic agents.

Methods: The data of 675 patients from The Cancer Genome Atlas (TCGA) database were used to identify MRG signature and construct a prognostic model. After validating its robustness in Chinese Glioma Genome Atlas (CGGA), two risk groups derived from the prognostic model were then conducted with Gene Set Enrichment Analysis (GSEA), immune status, mutation status and chemotherapeutic agents prediction.

Results: The prognostic model built from six gene signatures can successfully predict the prognosis and reflect clinicopathological characteristics. Patients in high-risk group displayed significantly worse overall survival (OS), immunosuppression effects, and mutation markers with worse prognosis. Twelve chemotherapeutic agents with strongly correlated sensitivity and risk scores were selected as potential agents.

Conclusions: The novel MRG signatures (TYMP, TSFM, MGME1, BOLA3, TRMT5, NDUFA9) can predict prognosis and immunological status in glioma.

Keywords: Mitochondria; glioma; immunotherapy; prognostic model.

Figure 1

Workflow of this study. GTEx, the Genotype-Tissue Expression project; TCGA, The Cancer Genome Atlas; LGG, low grade glioma; GBM, glioblastoma multiforme; DEGs, differentially expressed genes; MITOMAP, the Human Mitochondrial Genome Database; MRGs, mitochondrial-related genes; DEMRGs, differentially expressed mitochondrial-related genes; LASSO, Least Absolute Shrinkage and Selection Operator; CGGA, Chinese Glioma Genome Atlas; ROC, receiver operating characteristic; GSEA, Gene Set Enrichment Analysis.

Figure 2

Identification of prognostic mitochondrial-related DEGs in the TCGA cohort. (A) Twenty-nine mitochondrial-related DEGs between tumor (TCGA) and normal tissue (GTEx) and their categories. (B) PPI network among 28 candidate genes. ATP8A2 does not interact with this network. NDUFA12, NDUFA11, NDUFS7 and COX6A1 are hub genes. (C) Results of univariate Cox regression analysis between candidate gene expression and OS, and 25 genes were identified with P<0.05. (D) LASSO regression and the coefficients of the 25 OS-related genes, and 14 genes were identified for multivariate Cox regression. (E) Multivariate Cox regression for 14 candidate genes, and only 6 genes were identified with coefficients for the construction of prognostic model. (F) Immunostaining results of DEMRGs in different grades of glioma from Human Protein Atlas. The links are provided for TYMP (https://www.proteinatlas.org/ENSG00000025708-TYMP/pathology/glioma), MGME1 (https://www.proteinatlas.org/ENSG00000125871-MGME1/pathology/glioma), BOLA3 (https://www.proteinatlas.org/ENSG00000163170-BOLA3/pathology/glioma), TRMT5 (https://www.proteinatlas.org/ENSG00000126814-TRMT5/pathology/glioma) and NDUFA9 (https://www.proteinatlas.org/ENSG00000139180-NDUFA9/pathology/glioma), respectively. Scale bar, 300 µm. HR, hazard ratio; CI, confidence interval; AIC, Akaike information criterion; Not sig., not significant. DEGs, differentially expressed genes; TCGA, The Cancer Genome Atlas; GTEx, the Genotype-Tissue Expression project; PPI, protein-protein interaction; OS, overall survival; LASSO, Least Absolute Shrinkage and Selection Operator; DEMRGs, differentially expressed mitochondrial-related genes.

Figure 3

Validation of the predictive effects of prognostic models. Kaplan-Meier survival curves for TCGA (A), CGGA_325 (B) and CGGA_693 (C); ROC curves for CGGA_325 (D), TCGA (E) and CGGA_693 (F); (G) nomogram built from TCGA; the calibration plots for nomogram in CGGA_325 (H) and CGGA_693 (I). TCGA, The Cancer Genome Atlas; CGGA, Chinese Glioma Genome Atlas; AUC, area under the curve; OS, overall survival; ROC, receiver operating characteristic.

Figure 4

Correlations between the risk score and some clinicopathological characteristics in CGGA_693 (A) and CGGA_325 (B). ns, not significant; **, P<0.01; ****, P<0.0001. CGGA, Chinese Glioma Genome Atlas; WHO, World Health Organization; IDH, isocitrate dehydrogenase; MGMT, O6-methylguanine-DNA methyltransferase.

Figure 5

Functional enrichment analysis between the high- and low-risk groups in TCGA cohort. The KEGG pathway enrichment analysis (A) and the HALLMARK gene set enrichment analysis (B). The enrichment analysis results in this figure were all statistically significant. abs, absolute value; NES, normalized enrichment score; TCGA, The Cancer Genome Atlas; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Figure 6

The immune status of different risk groups and its correlation with the risk score in TCGA cohort. (A-C) The relationship between the risk score and immune score, stromal score and tumor purity calculated by ESTIMATE. (D) The landscape of immune cell infiltration status analyzed by CIBERSORT between risk groups. R value means the correlation of cell infiltration value and risk score. (E) The landscape of nine candidate immune checkpoints related gene expression between risk groups. R value means the correlation of gene expression value and risk score. (F) The landscape of enrichment degree of 16 immune-related pathways between risk groups. R value means the correlation of the enrichment degree and risk score. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. NK, natural killer; TCGA, The Cancer Genome Atlas.

Figure 7

Analysis results of somatic mutation. (A,B) The top 20 most frequently mutated genes for different risk groups. (C) Fisher’s exact test for genes with mutation rates higher than 5% in two groups. (H) Fisher’s exact test for pathways of mutant gene enrichment. (D-G) Four statistically significant pathways and the gene mutations involved. Not sig., not significant; *, P<0.05; **, P<0.01; ***, P<0.001. OR, odds ratio.

Figure 8

Drug sensitivity prediction. R value means the correlation of predicted drug sensitivity and risk score. An R greater than 0 means that the drug is more sensitive to tumor cells in the high-risk group of patients, and an R less than 0 means that the drug is more sensitive to tumor cells in the low-risk group of patients. abs, absolute value; FDA, Food and Drug Administration.