Integrated analysis of genome-wide DNA methylation, gene expression and protein expression profiles in molecular subtypes of WHO II-IV gliomas

Abstract

Background: Glioma is the most common malignant primary brain tumor among adults, among which glioblastoma (GBM) exhibits the highest malignancy. Despite current standard chemoradiation, glioma is still invariably fatal. A further insight into the molecular background of glioma is required to improve patient outcomes.

Method: Previous studies evaluated molecular genetic differences through comparing different grades of glioma. Here, we integrated DNA methylation, RNA sequencing and protein expression data sets of WHO grade II to IV gliomas, to screen for dysregulated genes in subtypes during malignant progression of glioma.

Results: We propose a list of universal genes (UG) as novel glioma biomarkers: 977 up-regulated genes and 114 down-regulated genes, who involved in cell cycle, Wnt receptor signaling pathway and fatty acid metabolic process. Poorer survival was associated significantly with the high expression of 977 up-regulated genes and low expression of 114 down-regulated in UG (P <0.001).

Conclusion: To our knowledge, this was the first study that focused on subtypes to detect dysregulated genes that could contribute to malignant progression. Furthermore, the differentially expressed genes profile may lead to the identification of new therapeutic targets for glioma patients.

Fig. 1

TCGA subtype annotation. Using the predictive 840 gene list, samples were ordered on the basis of subtype predictions

Fig. 2

An integrated omic platform for the characterization of glioma patients in three subtypes. a A, AA and GBM were quantified for DNA methylation, RNAseq and protein expression. Data for each primary GBM and AA, relative to A, were color-encoded as indicated and integrated into six-sided polygons in a symmetrical manner, with AA data in the top three quadrants, and data from the GBM in the mirror-image bottom sections. b The polygons were assembled as a linear genetic map in the vertical direction, which organized by chromosome and by three subtypes in the horizontal direction: IDH-wild-type (IDH-WT) Proneural, IDH-mutated (IDH-Mut) Proneural, Classical. c By this approach 1091 data points were integrated into 3273 polygons. These reflected correlated changes in one or more of the linked genes across the set of individual AA and GBM. Red, high expression; Blue, low expression; White, not significant difference; Grey, data not available

Fig. 3

Gene set enrichment analysis of UG in three subtypes. The cell-cycle enrichment ploted of universally up-regulated genes in (a) IDH-wild-type Proneural AA, (b) IDH-mutated Proneural AA, (c) IDH Classical AA, (d) IDH-wild-type Proneural GBM (e) IDH-mutated Proneural GBM, (f) IDH Classical GBM

Fig. 4

K-means clustering identified four groups of 310 CGGA samples by UG. K-means clustering identified of 310 CGGA samples by 1091 genes whose expression most strongly correlated with grade from TCGA database for k = 4

Fig. 5

Kaplan-Meier estimated of survival for 511 TCGA and 310 CGGA samples. a Among 511 TCGA GBM samples, there was a significant difference in survival between two groups (p = 0.0095) (b) Among 310 CGGA samples, there was a significant difference in survival between four groups (p < 0.0001). c Among GBM patients in 310 CGGA samples, there was a significant difference in survival between two groups (p = 0.0365). Group1, high expression of down-regulated UG; Group2, high expression of up-regulated UG