Total copy number variation as a prognostic factor in adult astrocytoma subtypes

Abstract

Since the discovery that IDH1/2 mutations confer a significantly better prognosis in astrocytomas, much work has been done to identify other molecular signatures to help further stratify lower-grade astrocytomas and glioblastomas, with the goal of accurately predicting clinical outcome and identifying potentially targetable mutations. In the present study, we subclassify 135 astrocytomas (67 IDH-wildtype and 68 IDH-mutant) from The Cancer Genome Atlas dataset (TCGA) on the basis of grade, IDH-status, and the previously established prognostic factors, CDK4 amplification and CDKN2A/B deletion, within the IDH-mutant groups. We analyzed these groups for total copy number variation (CNV), total mutation burden, chromothripsis, specific mutations, and amplifications/deletions of specific genes/chromosomal regions. Herein, we demonstrate that across all of these tumor groups, total CNV level is a relatively consistent prognostic factor. We also identified a trend towards increased levels of chromothripsis in tumors with lower progression-free survival (PFS) and overall survival (OS) intervals. While no significant differences were identified in overall mutation load, we did identify a significantly higher number of cases with mutations in genes with functions related to maintaining genomic stability in groups with higher mean CNV and worse PFS and OS intervals, particularly in the IDH-mutant groups. Our data further support the case for total CNV level as a potential prognostic factor in astrocytomas, and suggest mutations in genes responsible for overall genomic instability as a possible underlying mechanism for some astrocytomas with poor clinical outcome.

Keywords: Astrocytoma; CNV; Copy number variation; GBM; Glioblastoma; Glioma; TCGA.

Fig. 1

Kaplan-Meier survival curves demonstrating a significant difference between IDH-mutant LGGs without CDK4 amplification or CDKN2A/B deletion and both IDH-mutant LGGs with CDK4 or CDKN2A/B alterations (p = 0.0224) and IDH-mutant GBMs (p = 0.0032), but not between IDH-mutant LGGs with CDK4 or CDKN2A/B alterations and IDH-mutant GBMs (p = 0.0769) in terms of progression-free survival (a). There was also a significant difference between IDH-mutant LGGs and both IDH-mutant LGGs with CDK4 or CDKN2A/B alterations (p = 0.0150) and IDH-mutant GBMs (p = 0.0081), but not between IDH-mutant LGGs with CDK4 or CDKN2A/B alterations and IDH-mutant GBMs (p = 0.2892) in terms of overall survival (b). No significant differences are identified between IDH-wildtype LGGs and IDH-wildtype GBMs in terms of progression-free survival (p = 0.2050) (c) or overall survival (p = 0.9351) (d)

Fig. 2

Comparison between IDH-mutant glioblastoma cases with and without amplifications of CDK4 or deletions of CDKN2A/B. There is no significant difference in progression-free survival (p = 0.8406) (a), overall survival (p = 0.1471) (b), total copy number variation burden (p = 0.5326) (c), or total mutation burden (p = 0.6686) (d) between these groups

Fig. 3

Total copy number variation averages demonstrating a significant difference between IDH-mutant LGGs without CDK4 amplification or CDKN2A/B deletion and both IDH-mutant LGGs with CDK4 or CDKN2A/B alterations (p = 0.0003) and IDH-mutant GBMs (p = 0.0078), but not between IDH-mutant LGGs with CDK4 or CDKN2A/B alterations and IDH-mutant GBMs (p = 0.7783) (a); no significant difference was found in total mutation burden between any group of IDH-mutant astrocytoma (b). There was no significant difference between IDH-wildtype LGGs and IDH-wildtype GBMs in terms of overall copy number variation (p = 0.3732) (c) or total mutation burden (p = 0.5627) (d)

Fig. 4

Scatter plots of copy number variation (%) plotted against survival time (months) in grouped IDH-mutant LGGs and IDH-mutant GBMs with Pearson’s R values, illustrating significant inverse correlations between the two data points in terms of (a) progression-free survival (r = − 0.3415; p = 0.0047) and (b) overall survival (r = − 0.3098; p = 0.0102)

Fig. 5

Overall amplification and deletion levels and chromosomal locations in IDH-mutant LGGs without CDK4 amplification or CDKN2A/B deletion (a), IDH-mutant LGGs with either CDK4 amplification or CDKN2A/B deletion (b), and IDH-mutant GBMs (c)

Fig. 6

Overall amplification and deletion levels and chromosomal locations in IDH-wildtype LGGs (a) and IDH-wildtype GBMs (b)

Fig. 7

GISTIC analysis showing the most consistent and relevant cytoband alterations in IDH-mutant LGGs without CDK4 amplification or CDKN2A/B deletion (a), IDH-mutant LGGs with either CDK4 amplification or CDKN2A/B deletion (b), IDH-mutant GBMs (c), IDH-wildtype LGGs (d), and IDH-wildtype GBMs (e). All cytobands shown met the criterion of false discovery rate (FDR) ≤0.25. The annotated cytobands met the criterion of FDR ≤0.05

Fig. 8

Pie charts illustrating (a) the relative frequency of cases with chromothripsis in all 5 astrocytoma subgroups, showing a statistically significant difference between IDH-mut LGGs without CDK4 amplification or CDKN2A/B deletion and IDH-mut GBMs (p = 0.0132) and between IDH-mut LGGs without CDK4 amplification or CDKN2A/B deletion and all IDH-mut tumors with poor clinical outcome (groups 2 + 3; p = 0.0211). Pie charts illustrating (b) the relative frequency of cases with mutations involving genes related to preservation of overall chromosomal stability in all 5 astrocytoma subgroups, showing a statistically significant difference between IDH-mut LGGs without CDK4 amplification or CDKN2A/B deletion and LGGs with those molecular alterations (p = 0.0197) and between IDH-mut LGG without CDK4 amplification or CDKN2A/B deletion and IDH-mut GBMs (p = 0.0086)