7-ACVR1.1.Pdf
Total Page:16
File Type:pdf, Size:1020Kb
LETTERS Genomic analysis of diffuse intrinsic pontine gliomas identifies three molecular subgroups and recurrent activating ACVR1 mutations Pawel Buczkowicz1–3,24, Christine Hoeman4,24, Patricia Rakopoulos2,3, Sanja Pajovic2, Louis Letourneau5, Misko Dzamba6, Andrew Morrison2, Peter Lewis7, Eric Bouffet8, Ute Bartels8, Jennifer Zuccaro2, Sameer Agnihotri2, Scott Ryall2, Mark Barszczyk2,3, Yevgen Chornenkyy2,3, Mathieu Bourgey5, Guillaume Bourque5, Alexandre Montpetit5, Francisco Cordero4, Pedro Castelo-Branco2, Joshua Mangerel2, Uri Tabori2,8, King Ching Ho2, Annie Huang2,8, Kathryn R Taylor9, Alan Mackay9, Anne E Bendel10, Javad Nazarian11, Jason R Fangusaro12, Matthias A Karajannis13, David Zagzag13, Nicholas K Foreman14, Andrew Donson14, Julia V Hegert15, Amy Smith15, Jennifer Chan16, Lucy Lafay-Cousin16, Sandra Dunn17, Juliette Hukin17, Chris Dunham17, Katrin Scheinemann18, Jean Michaud19, Shayna Zelcer20, David Ramsay20, Jason Cain21, Cameron Brennan22, Mark M Souweidane22, Chris Jones9, C David Allis7, Michael Brudno6,23, Oren Becher4,25 & Cynthia Hawkins1–3,25 Diffuse intrinsic pontine glioma (DIPG) is a fatal brain cancer death in children. Diagnosis of DIPG is based on a combination of that arises in the brainstem of children, with no effective clinical and radiological findings; a tissue biopsy is rarely acquired. treatment and near 100% fatality. The failure of most therapies Radiation is the mainstay of therapy but offers only symptom con- can be attributed to the delicate location of these tumors and trol, and, so far, chemotherapy has shown no benefit4. A potential to the selection of therapies on the basis of assumptions that contributor to the failure of DIPG clinical trials is the use of agents Nature America, Inc. All rights reserved. Inc. Nature America, DIPGs are molecularly similar to adult disease. Recent studies targeting the genetic alterations of adult glioblastomas (GBMs)5. 4 have unraveled the unique genetic makeup of this brain cancer, A number of recent studies have reported differences at both the copy with nearly 80% found to harbor a p.Lys27Met histone H3.3 number and expression levels that distinguish pediatric DIPGs from © 201 or p.Lys27Met histone H3.1 alteration. However, DIPGs are both their adult and pediatric supratentorial GBM counterparts6–9, still thought of as one disease, with limited understanding indicating that they may be separate biological entities requiring their of the genetic drivers of these tumors. To understand what own therapeutic strategies. Recent identification of frequent histone drives DIPGs, we integrated whole-genome sequencing H3 gene mutations (encoding p.Lys27Met) in DIPGs has suggested with methylation, expression and copy number profiling, that both genetic and epigenetic mechanisms may be important driv- discovering that DIPGs comprise three molecularly distinct ers of these tumors10–12. However, the complete genetic and epigenetic subgroups (H3-K27M, silent and MYCN) and uncovering a new landscapes of DIPG and the functional role that histone modifications recurrent activating mutation affecting the activin receptor may have in DIPG remain unknown. These data are critical for the gene ACVR1 in 20% of DIPGs. Mutations in ACVR1 were development of better therapies for affected children. constitutively activating, leading to SMAD phosphorylation We integrated deep sequencing analysis of 36 tumor-normal pairs and increased expression of the downstream activin signaling (20 whole-genome sequencing (Illumina HiSeq 2000) and 16 whole- targets ID1 and ID2. Our results highlight distinct molecular exome sequencing (Applied Biosystems SOLiD 5500xl)) with compre- subgroups and novel therapeutic targets for this incurable hensive methylation (28 DIPGs; Illumina Infinium450k methylation pediatric cancer. array), copy number (45 DIPGs; Affymetrix SNP6.0) and expression (35 DIPGs; Illumina HT-12 v4) data (Supplementary Table 1). All Brain tumors are the largest group of solid tumors and the leading coding somatic single-nucleotide variants (SNVs) identified in the cause of cancer-related deaths in childhood1. The most devastating of combined cohort are listed in Supplementary Table 2, including these is DIPG, which is almost universally fatal2,3. Despite collabora- new mutations in ACVR1. We did not find mutations in BRAF or tive efforts to improve treatments, survival has remained static over IDH1 or structural rearrangements in FGFR1 or MYB in our DIPG decades, and DIPGs are now the main cause of brain tumor–related cohort. Verification of somatic alterations was conducted on all 36 A full list of author affiliations appears at the end of the paper. Received 24 April 2013; accepted 5 March 2014; published online 6 April 2014; doi:10.1038/ng.2936 NATURE GEnETICS ADVANCE ONLINE PUBLICATION 1 LETTERS a MYCN Silent H3-K27M b � D01 D29 D52 D09 D18 D26 D63 D19 D62 D60 D06 D57 D28 D24 D27 D25 D08 D30 D67 D58 D64 D78 D05 D07 D04 D56 D03 D61 values Normal brain Silent 1 DNA methylation 16 H3-K27M 10 Methylated 4 –2 –8 ( n –14 PC2 0.5 = 1,0000 probes) –20 –26 Hemimethylated –32 –38 MYCN –44 PC3 1835 –29.3 –23.6 –17.9 –12.2 –6.5 –0.8 4.9 10.6 16.3 22 0 Unmethylated PC1 c d k = 3 1.0 1.2 0.8 0.98 1.0 0.6 0.8 CDF 0.4 0.6 0.94 2 clusters 0.4 0.2 3 clusters Change in Gini Cophenetic correlation 4 clusters 5 clusters 0.2 2.0 3.0 4.0 5.0 Samples 0 k groups 0 0.2 0.4 0.6 0.8 1.0 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Consensus index value Number of clusters Figure 1 Methylation profiling identifies three molecular subgroups of DIPG. (a) Heat map of methylation levels in three DIPG subgroups identified by unsupervised hierarchical clustering. (b–d) Subrouping was supported by principal-components analysis (b), non-negative matrix factorization (cophenetic coefficient = 0.9934, k = 3) (c) and consensus clustering represented by a cumulative distribution function (CDF) and change in Gini coefficient (d). PC, principal component. sample pairs using Fluidigm Array Ion Torrent chip sequencing 0.19–24) in the other two groups; P = 0.05 (Supplementary Table 4)). (Supplementary Table 3). Further, for key SNVs, we also tested a All DIPGs with low-grade astrocytoma (LGA) histology were from validation cohort of an additional 25 tumors. This analysis shows that, this group, although, interestingly, there was no difference in overall although previously considered to be one disease, DIPG represents survival in comparison with the other subgroups. Affected individuals Nature America, Inc. All rights reserved. Inc. Nature America, three distinct subgroups with different methylation, expression, copy in this group were diagnosed at a significantly younger age (4.81 ± 4 number alteration (CNA) and mutational profiles. 1.64 years for the children from the silent group versus 6.89 ± 2.62 Unsupervised subgrouping of DIPG cases on the basis of CpG island years for those not in the silent group; P = 0.04). The p.Lys27Met his- © 201 methylation (Online Methods) resulted in three distinct subgroups; tone H3 alteration was present in 44% of tumors (p.Lys27Met histone ‘MYCN’, ‘silent’ and ‘H3-K27M’ (Fig. 1a). This subgrouping was sup- H3.3 in 33% and p.Lys27Met histone H3.1 in 11%), but there were no ported by multiple analyses, including principal-components analysis recurrent copy number changes as observed in the MYCN and H3- (Fig. 1b), non-negative matrix factorization (Fig. 1c) and consensus K27M subgroups. At the expression level, the silent subgroup showed clustering (Fig. 1d). Subgroup-specific differences were supported by overexpression of WNT pathway genes (Supplementary Table 5), the integration of mutation, structural, expression and clinical data as well as MDM2, MSMP and ADAM33, compared with the other (Fig. 2). The identification of DIPG subgroups will have key implica- two subgroups (Table 1). Notably, DIPGs in neither the MYCN nor tions for the design of appropriate therapy for these tumors. the silent subgroup had receptor tyrosine kinase gene amplification, The MYCN subgroup had no recurrent mutations but was instead suggesting that the group of inhibitors targeting these kinases will be characterized by hypermethylation, high-grade histology and chro- less effective in patients with these DIPG subtypes. mothripsis on chromosome 2p leading to recurrent high-level ampli- H3-K27M subgroup DIPGs were highly mutated in either histone fication of MYCN and ID2 (Table 1 and Supplementary Figs. 1–3). H3.3 (H3F3A) or H3.1 (HIST1H3B and HIST1H3C). This group had Tumors in this group overexpressed MYCN by 4-fold and 8-fold highly unstable genomes (segmentation analysis of SNP6.0 data; 497 and ID2 by 2.5-fold and 5-fold relative to the H3-K27M and silent CNAs per genome versus 300 CNAs per genome in the silent group; groups, respectively. The top most overexpressed genes in the MYCN P = 0.04). Alternative lengthening of telomeres (ALT; detected by telo group included FAP, HRSP12 and DYX2. Therapies aimed at tar- mere restriction fragment assay13, the presence of C-circles14 and/or geting altered histone modifications would not be effective in this telomere length from whole-genome sequencing data) is exclusively subgroup. Rather, children with this subtype of DIPG will potentially associated with the H3-K27M subgroup; only two of the tumors in benefit from therapies targeting MYCN or possibly ID2. Tumors in this subgroup harbored ATRX mutations. ALT-positive cases were the silent subgroup had silent genomes on the basis of both whole- significantly older at the time of diagnosis (P < 0.001; Supplementary genome sequencing, structural and SNP6.0 copy number analysis and Fig. 4). Similarly, PVT1, MYC and PDGFRA gains or amplifications had a lower mutation rate than tumors in the other two subgroups (Fig. 2a) and structural variants (Supplementary Fig. 5) were exclu- (median of 0.11 mutations per megabase (range of 0.02–0.19) in the sive to this group.