The Comprehensive Analysis of m6A-Associated Anoikis Genes in Low-Grade Gliomas

The relationship between N6-methyladenosine (m6A) regulators and anoikis and their effects on low-grade glioma (LGG) is not clear yet. The TCGA-LGG cohort, mRNAseq 325 dataset, and GSE16011 validation set were separately obtained via the Cancer Genome Atlas (TCGA), Chinese Glioma Genome Altas (CGGA), and Gene Expression Omnibus (GEO) databases. In total, 27 m6A-related genes (m6A-RGs) and 508 anoikis-related genes (ANRGs) were extracted from published articles individually. First, differentially expressed genes (DEGs) between LGG and normal samples were sifted out by differential expression analysis. DEGs were respectively intersected with m6A-RGs and ANRGs to acquire differentially expressed m6A-RGs (DE-m6A-RGs) and differentially expressed ANRGs (DE-ANRGs). A correlation analysis of DE-m6A-RGs and DE-ANRGs was performed to obtain DE-m6A-ANRGs. Next, univariate Cox and least absolute shrinkage and selection operator (LASSO) were performed on DE-m6A-ANRGs to sift out risk model genes, and a risk score was gained according to them. Then, gene set enrichment analysis (GSEA) was implemented based on risk model genes. After that, we constructed an independent prognostic model and performed immune infiltration analysis and drug sensitivity analysis. Finally, an mRNA-miRNA-lncRNA regulatory network was constructed. There were 6901 DEGs between LGG and normal samples. Six DE-m6A-RGs and 214 DE-ANRGs were gained through intersecting DEGs with m6A-RGs and ANRGs, respectively. A total of 149 DE-m6A-ANRGs were derived after correlation analysis. Four genes, namely ANXA5, KIF18A, BRCA1, and HOXA10, composed the risk model, and they were involved in apoptosis, fatty acid metabolism, and glycolysis. The age and risk scores were finally sifted out to construct an independent prognostic model. Activated CD4 T cells, gamma delta T cells, and natural killer T cells had the largest positive correlations with risk model genes, while activated B cells were significantly negatively correlated with KIF18A and BRCA1. AT.9283, EXEL.2280, Gilteritinib, and Pracinostat had the largest correlation (absolute value) with a risk score. Four risk model genes (mRNAs), 12 miRNAs, and 21 lncRNAs formed an mRNA-miRNA-lncRNA network, containing HOXA10-hsa-miR-129-5p-LINC00689 and KIF18A-hsa-miR-221-3p-DANCR. Through bioinformatics, we constructed a prognostic model of m6A-associated anoikis genes in LGG, providing new ideas for research related to the prognosis and treatment of LGG.