Identification of tumor microenvironment-related genes in lower-grade gliomas by mining TCGA database

BACKGROUND: Lower-grade gliomas (LGGs) are ubiquitous and fatal branches of brain neoplasm. Finding biomarkers related to diagnosis and treatment is essential for the treatment of LGG. It is possible to reveal the potential links between tumor microenvironment and overall survival (OS) in LGG by mining the mRNA expression profile from the TCGA database. Our primary purpose was to explore key genes that can be applied for diagnosis or treatment in LGG microenvironment. METHODS: Based on the ESTIMATE algorithm, the immune and stromal scores were calculated to measure the extent of infiltration of immune cells and stromal cells, respectively. The LGG samples from TCGA database were assigned into high- or low-score groups per the immune and stromal scores and differentially expressed genes (DEGs) were selected by comparing gene expression levels in the two groups. Functional enrichment analysis and protein-protein interaction (PPI) networks were performed to analyze DEGs. Finally, selected DEGs were validated using another independent LGG cohort from CGGA dataset. RESULTS: The results indicated that immune/stromal scores correlated with LGG prognosis. Furtherly, survival analysis conducted for each subtype shown that immune/stromal scores were only significantly associated with the prognosis of astrocytoma, IDH-wildtype, and there was no significant statistical difference in the other subtypes. Functional enrichment analysis and protein-protein interaction (PPI) networks further showed that the upregulated DEGs were primarily involved in immune response, plasma membrane, and cytokine binding. Accordingly, a series of genes that have significant impacts on prognosis and are significantly associated with the tumor microenvironment were obtained. CONCLUSIONS: Based on the ESTIMATE algorithm, we first explored the relationship between immune/stromal scores and prognosis in different subtypes of LGG and the result shown that the scores were only strongly associated with the prognosis of astrocytoma, IDH-wildtype. Furtherly, a comprehensive bioinformatics analysis of the gene expression profiles of astrocytoma, IDH-wildtype patients was conducted, CASP8, TRIM6, TRIM38, PARP9, NMI, EPSTI1, DTX3L and AGBL2 were identified as tumor microenvironment-related genes, may be involved in the occurrence, development, and invasion of LGG.