The evolutionary trajectory of epigenomics in adult glioma

Document Type

Conference Proceeding

Publication Date

7-1-2021

Publication Title

Cancer Res

Abstract

Background: Glioma is the most common malignant tumor of the central nervous system, often behaving very aggressively. Recently, The Cancer Genome Atlas (TCGA) and others have shown that epigenomic alterations in primary glioma tumors have prognostic and predictive roles, but there is a gap in knowledge of the molecular alterations after glioma treatment. In order to fill this gap, the Glioma Longitudinal AnalySiS (GLASS) Consortium, a multi-national collaboration from 13 institutions, is investigating genome-wide molecular data from primary and recurrent matched pairs. The current data freeze has DNA methylation data for 266 primary-recurrent pairs fromIllumina 450K and EPIC array platforms. Methods: We hypothesize that there will be evidence of aggressivity in the DNA methylation profiles of recurrent tumors relative to their matched primary, and we explored this through tumor subtyping, patterns of differentially methylated CpGs (DMPs), and epigenomic aging. Results: The subtype classification in the primary tumors was as follows for IDH wildtype tumors - 43.8%Classical, 47.3% Mesenchymal, 8.7% PA-like, none LGm6-GBM and for IDH mutant tumors 19.1% Codel, 79.4% G-CIMP-high, 1.3% G-CIMP-low. We observed that among IDH wildtype tumors 29.8% changed subtype, 47.1% of which shifted to the more aggressive Mesenchymal-like subtype. In IDH mutant tumors, 26.0% changed at recurrence, of which 57.9% shifted to the aggressive G-CIMP-low subtype. Patterns of DMPs in IDH mutant-code tumors (15 pairs) showed a loss of methylation upon recurrence, with 651 DMPs identified (paired Wlcoxon test, FDR < 0.05). In unsupervised clustering, recurrent Codel tumors thus move away from IDH mutant tumors and align more closely to IDH wildtype tumors. We do not see the same increases in hypomethylation upon recurrence among IDHwt tumors, with only 17 DMPs at FDR < 0.05 in the classical subtype (24 pairs). It has been shown that biological age estimates of the tumor using DNA methylation can predict aggressivity of certain tumor types. We examined epigenetic aging changes between primary and recurrent tumors, relative to the patients chronological age, with Horvath (tissue-based) and epiTOC (mitotic-based) DNA methylation clocks. Both clocks showed shifts in age acceleration, that is increased biological age in relation to chronological age. For the Horvath clock, we found increased age acceleration in Classical-like tumors (primary=39.5±8.9, recurrent 41.7± 19.7) and Codel tumors (primary= 57.0±22.3, recurrent= 64.2±31.8). Conclusions: Collectively, we observed glioma tumor epigenetic changes from the primary to recurrent state and these differences tend to reflect a shift to a more aggressive phenotype. Future work will explore the relationship of these findings with clinical treatments received between primary and recurrent states.

Volume

81

Issue

13 SUPPL

Find It @ Sladen

Share

COinS