Expression of bone marrow stromal antigen 2 gene in patients with glioblastoma and its correlation with non-CpG island DNA methylation and prognosis / 肿瘤研究与临床

ABSTRACT

Objective:To investigate the expression level of bone marrow stromal antigen 2 (BST2) in patients with glioblastoma (GBM) gene and its correlation with DNA methylation level and prognosis.Methods:The datasets of GBM samples from the Cancer Genome Atlas (TCGA) database (35 cases), GSE22891 cohort (50 cases) within Gene Expression Omnibus (GEO) database, and China Glioma Genome Atlas (CGGA) database (105 cases), and non-tumor brains (NTB) (10 cases in TCGA database, 6 cases in GSE22891 cohort, 5 cases in CGGA database were used to make comparisons of gene expression level; 25 cases in GSE63347 cohort, 4 cases in GSE22891 cohort, 8 cases in CGGA database were used to make comparisons of methylation data). Based on gene expression chip and DNA methylation microarray data from public GBM databases, the expression level of BST2 gene in GBM and the association with non-CpG island DNA methylation, GBM molecular subtypes [CpG island methylation phenotype (G-CIMP) and non-G-CIMP proneural, neural, classical, mesenchymal], overall survival (OS) time and functional gene expression profiles were obtained by using intra-group comparison of BST2 gene expression level between GBM and NTB, survival analysis, and bioinformatic analysis.Results:Compared with NTB samples, BST2 mRNA was highly expressed in GBM (mRNA expression data were based on Z-score standardization) (TCGA database vs. GSE63347 cohort -0.97±1.14 vs. -2.32±0.21, t = 3.74, P < 0.05; GSE22891 cohort 9.03±1.28 vs. 7.18±0.22, t = 3.42, P < 0.05; CGGA database -0.43±1.11 vs. -0.62±0.35, t = 2.09, P < 0.05). In TCGA database, BST2 mRNA was highly expressed in different tumor molecular subgroups (G-CIMP -1.96±0.94; non-G-CIMP proneural -1.74±0.88; neural -0.83±0.98; classical 0.71±1.18; mesenchymal -0.55±1.01), which were compared with NTB samples (-2.32±0.21), and the differences were statistically significant (all P < 0.05). There were no statistically significant differences in the expressions of BST2 mRNA in the neural, classical, mesenchymal tumors (all P > 0.05), but the expression of BST2 mRNA in above three groups was higher than that in G-CIMP group and non-G-CIMP proneural group (all P < 0.05). Correlation analysis showed that non-CpG island DNA methylation level of BST2 was negatively correlated with the expression of its mRNA (TCGA database r = -0.30, P < 0.05; GSE22891 cohort r = -0.54, P < 0.05; CGGA database r = -0.29, P > 0.05). Survival analysis showed that BST2 mRNA expression level of non-G-CIMP and non-proneural patients was negatively associated with OS ( HR = 1.18, 95% CI 1.00-1.39, P < 0.05); among those tumors with G-CIMP or proneural subtypes, BST2 mRNA expression was not associated with OS ( HR = 1.08, 95% CI 0.84-1.40, P > 0.05). Bioinformatic analysis showed that among non-G-CIMP and non-proneural samples of TCGA database, GBM samples with higher BST2 expression were enriched with functional gene sets related to negative regulation of immune responses and activation of nuclear factor κB (NF-κB) pathway. Conclusions:The upregulated expression of BST2 gene in GBM may be associated with non-CpG island DNA hypomethylation alteration. BST2 gene may become a potential prognostic biomarker for non-G-CIMP and non-proneural GBM.