Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Meta-Analysis
. 2017 Oct 9;32(4):520-537.e5.
doi: 10.1016/j.ccell.2017.08.017. Epub 2017 Sep 28.

Integrated Molecular Meta-Analysis of 1,000 Pediatric High-Grade and Diffuse Intrinsic Pontine Glioma

Affiliations
Free PMC article
Meta-Analysis

Integrated Molecular Meta-Analysis of 1,000 Pediatric High-Grade and Diffuse Intrinsic Pontine Glioma

Alan Mackay et al. Cancer Cell. .
Free PMC article

Abstract

We collated data from 157 unpublished cases of pediatric high-grade glioma and diffuse intrinsic pontine glioma and 20 publicly available datasets in an integrated analysis of >1,000 cases. We identified co-segregating mutations in histone-mutant subgroups including loss of FBXW7 in H3.3G34R/V, TOP3A rearrangements in H3.3K27M, and BCOR mutations in H3.1K27M. Histone wild-type subgroups are refined by the presence of key oncogenic events or methylation profiles more closely resembling lower-grade tumors. Genomic aberrations increase with age, highlighting the infant population as biologically and clinically distinct. Uncommon pathway dysregulation is seen in small subsets of tumors, further defining the molecular diversity of the disease, opening up avenues for biological study and providing a basis for functionally defined future treatment stratification.

Keywords: DIPG; exome; genome; glioblastoma; histone; methylation.

Figures

None
Figure 1
Figure 1
Clinicopathological and Molecular Subgroups of pHGG/DIPG (A) Anatomical location of all high-grade glioma cases included in this study, taken from original publications (n = 1,033). Left, sagittal section showing internal structures; right, external view highlighting cerebral lobes. Hemispheric, dark red; non-brainstem midline structures, red; pons, pink. Radius of circle is proportional to the number of cases. Lighter shaded circles represent a non-specific designation of hemispheric, midline, or brainstem. (B) Boxplot showing age at diagnosis of included cases, separated by anatomical location (n = 1,011). The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. ∗∗∗Adjusted p < 0.0001 for all pairwise comparisons, t test. (C) Kaplan-Meier plot of overall survival of cases separated by anatomical location, p value calculated by the log rank test (n = 811). (D) Anatomical location of all cases separated by histone mutation (top, n = 441) and histone WT (bottom, n = 314). Left, sagittal section showing internal structures; right, external view highlighting cerebral lobes. Blue, H3.3G34R/V; green, H3.3K27M; dark green, H3.1K27M. Radius of circle is proportional to the number of cases. Lighter shaded circles represent a non-specific designation of hemispheric, midline, or brainstem. (E) Boxplot showing age at diagnosis of included cases, separated by histone mutation (n = 753). The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. ∗∗∗Adjusted p < 0.0001 for all pairwise comparisons, t test. (F) Kaplan-Meier plot of overall survival of cases separated by histone mutation, p value calculated by the log rank test (n = 693). See also Figure S1 and Table S1.
Figure 2
Figure 2
Methylation-based Subclassification of pHGG/DIPG (A) Unsupervised hierarchical clustering and heatmap representation of β values for 441 samples profiled on the Illumina 450k BeadArray platform (red, high; blue, low). Samples are arranged in columns clustered by most variable 1,381 classifier probes. Age at diagnosis is provided below. Clinicopathological and molecular annotations are provided as bars according to the included key. (B) Anatomical location of methylation-defined PXA-like (n = 43) and LGG-like (n = 27) cases. Left, sagittal section showing internal structures; right, external view highlighting cerebral lobes. Dark gold, PXA-like; tan, LGG-like. Radius of circle is proportional to the number of cases. Lighter shaded circles represent a non-specific designation of hemispheric, midline, or brainstem. (C) Boxplot showing age at diagnosis of included cases, separated by simplified methylation subclass (n = 440). The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. ∗∗∗Adjusted p < 0.0001 for all H3 G34R/V pairwise comparisons, t test; ∗∗adjusted p < 0.01 for LGG-like versus WT, t test. (D) Kaplan-Meier plot of overall survival of cases separated by simplified methylation subclass, p value calculated by the log rank test (n = 307). See also Figure S2 and Table S2.
Figure 3
Figure 3
DNA Copy-Number Aberrations in pHGG/DIPG (A) Heatmap representation of segmented DNA copy number for 834 pHGG/DIPG profiled across one or more of seven different platforms (dark red, amplification; red, gain; dark blue, deletion; blue, loss). Samples are arranged in columns clustered by gene-level data across the whole genome. Age at diagnosis is provided below. Clinicopathological and molecular annotations are provided as bars according to the included key. (B and C) Barplot of all recurrent focal amplifications (B) and deletions (C) across all 834 cases, in order of frequency, and colored independently by both anatomical location and histone mutation. See also Figure S3 and Table S3.
Figure 4
Figure 4
Subgroup-specific Copy-Number Changes in pHGG/DIPG (A) Heatmap representation of segmented DNA copy number for 705 pHGG/DIPG separated for known histone mutation subgroup (dark red, amplification; red, gain; dark blue, deletion; blue, loss). Samples are arranged in columns clustered by gene-level data across the whole genome. Clinicopathological and molecular annotations are provided as bars according to the included key. (B) GISTIC analysis of focal amplifications and deletions for histone mutation subgroups. Log10 values are plotted across the genome for both amplifications (dark red) and deletions (dark blue), with significantly enriched events labeled by likely driver genes. Subgroup-specific genes are highlighted by the appropriate color. (C) Barplot of frequency of whole chromosomal arm gains (red) and losses (blue) for each subgroup. Significantly enriched alterations (p < 0.0001, Fisher’s exact test) are labeled, with subgroup-specific arm changes highlighted by the appropriate color. See also Figure S4 and Table S4.
Figure 5
Figure 5
Alterations Targeting FBXW7 in H3.3G34R/V pHGG and TOP3A in H3.3K27M DIPG (A) Segmented exon-level DNA copy-number heatmaps for 4q loss in H3.3G34R/V tumors (dark red, amplification; red, gain; dark blue, deletion; blue, loss; n = 28). An ideogram of chromosome 4 is provided indicating enlarged genome browser view and genes within common regions targeted across samples (gray). Clinicopathological and molecular annotations are provided as bars according to the included key. (B) Cartoon representation of amino acid position for four somatic mutations found in FBXW7, colored by annotated functional domains and numbers provided for recurrent variants. (C) Boxplots representing gene expression differences between FBXW7 lost/mutated cases (blue) and those with normal copy/WT (gray) in three independent gene expression platform datasets. The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. (D) Segmented exon-level DNA copy-number heatmaps for 17p11.2 amplification in predominantly H3.3K27M DIPG (dark red, amplification; red, gain; dark blue, deletion; blue, loss; n = 17). Chromosome 17 ideogram is provided indicating enlarged genome browser view and genes within common regions targeted across samples (gray). Clinicopathological and molecular annotations are provided as bars according to the included key. (E) Sequencing coverage (top) and log2 ratio plot (bottom) for chromosomes 7, 17, and 20 for two cases, showing complex intra- or inter-chromosomal rearrangements leading to specific copy-number amplification of TOP3A. (F) Boxplots representing gene expression differences between TOP3A amplified cases (red) and those with normal copy (gray) in three independent gene expression platform datasets. The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. See also Figure S5 and Table S5.
Figure 6
Figure 6
Somatic Mutations in pHGG/DIPG (A) Oncoprint representation of an integrated annotation of somatic mutations and DNA copy-number changes for the 30 most frequently altered genes in 326 pHGG/DIPG (n ≥ 6, frequency barplot on the right). Selected common fusion events are also shown where available. Samples are arranged in columns with genes labeled along rows. Age at diagnosis is provided below. Underneath, barplots are provided on a log10 scale for numbers of copy-number aberrations and somatic mutations per case. Clinicopathological and molecular annotations are provided as bars according to the included key. (B) Barplot of all recurrent somatic mutations across all 326 cases, in order of frequency, and colored independently by both anatomical location and histone mutation. See also Figure S6 and Table S6.
Figure 7
Figure 7
Integrated Pathway Analysis of pHGG/DIPG (A) Oncoprint-style representation of an integrated annotation of somatic mutations and DNA copy-number changes in one or more of nine commonly targeted pathways in 326 pHGG/DIPG (n ≥ 6, frequency barplot on the right). Samples are arranged in columns with pathways labeled along rows. Clinicopathological and molecular annotations are provided as bars according to the included key. (B) Pathway enrichment analysis of pHGG/DIPG subgroups. Distinct pathways and biological processes between the subgroups are colored appropriately (FDR q < 0.01). Nodes represent enriched gene sets, which are grouped and annotated by their similarity according to related gene sets. Node size is proportional to the total number of genes within each gene set. The illustrated network map was simplified by manual curation to remove general and uninformative sub-networks. See also Figure S7 and Table S7.
Figure 8
Figure 8
Integrated Analysis of H3/IDH1 WT pHGG/DIPG (A) t Statistic-based stochastic neighbor embedding (t-SNE) projection of the combined 450k methylation dataset (n = 441). The first three projections are plotted in the x, y, and z axes, with samples represented by dots colored by histone H3G34 (blue), H3K27 (green), IDH1 (red), PXA-like (dark gold), LGG-like (tan), and “others” (gray). (B) K means consensus clustering on the H3/IDH1 WT cases highlights three stable clusters (left, black/brown [WT-A], gray/pink [WT-B], and dark cyan [WT-C]) as the most robust subdivision of the data (right, area under the curve analysis for different cluster numbers). (C) Unsupervised hierarchical clustering and attendant heatmap of the H3/IDH1 WT cases (n = 219). Samples are arranged in columns clustered by the most variable 1,521 classifier probes. Age at diagnosis is provided below. Clinicopathological and molecular annotations are provided as bars according to the included key. (D) Oncoprint representation of an integrated annotation of somatic mutations and DNA copy-number changes for the H3/IDH1 WT cases (n = 50). Samples are arranged in columns with genes labeled along rows. Age at diagnosis is provided below. Clinicopathological and molecular annotations are provided as bars according to the included key. (E) Boxplot showing age at diagnosis of H3/IDH1 WT subgroups, separated by anatomical location (n = 190). The thick line within the box is the median, the lower and upper limits of the boxes represent the first and third quartiles, and the whiskers 1.5× the interquartile range. (F) Kaplan-Meier plot of overall survival of H3/IDH1 WT subgroups separated by anatomical location, p value calculated by the log rank test (n = 150). See also Figure S8 and Table S8.

Similar articles

See all similar articles

Cited by 100 articles

See all "Cited by" articles

References

    1. Baker S.J., Ellison D.W., Gutmann D.H. Pediatric gliomas as neurodevelopmental disorders. Glia. 2015;64:879–895. - PMC - PubMed
    1. Barrow J., Adamowicz-Brice M., Cartmill M., MacArthur D., Lowe J., Robson K., Brundler M.A., Walker D.A., Coyle B., Grundy R. Homozygous loss of ADAM3A revealed by genome-wide analysis of pediatric high-grade glioma and diffuse intrinsic pontine gliomas. Neuro Oncol. 2011;13:212–222. - PMC - PubMed
    1. Bax D.A., Mackay A., Little S.E., Carvalho D., Viana-Pereira M., Tamber N., Grigoriadis A.E., Ashworth A., Reis R.M., Ellison D.W. A distinct spectrum of copy number aberrations in pediatric high-grade gliomas. Clin. Cancer Res. 2010;16:3368–3377. - PMC - PubMed
    1. Bender S., Tang Y., Lindroth A.M., Hovestadt V., Jones D.T., Kool M., Zapatka M., Northcott P.A., Sturm D., Wang W. Reduced H3K27me3 and DNA hypomethylation are major drivers of gene expression in K27M mutant pediatric high-grade gliomas. Cancer Cell. 2013;24:660–672. - PubMed
    1. Bjerke L., Mackay A., Nandhabalan M., Burford A., Jury A., Popov S., Bax D.A., Carvalho D., Taylor K.R., Vinci M. Histone H3.3. mutations drive pediatric glioblastoma through upregulation of MYCN. Cancer Discov. 2013;3:512–519. - PMC - PubMed

Publication types

MeSH terms

Feedback