Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Timing of mutations in a patient with ES. The schematic shows genetic alterations in tumors at
and other genic disruptions
symptoms before clinical disease presentation.
▪
prediagnosis, diagnosis, and relapse. In many cases, the fusion gene that drives tumorigenesis
(EWSR1-FLI1 or EWSR1-ERG) emerges via a sudden burst of genomic rearrangements involving The list of author affiliations is available in the full article online.
multiple chromosomes and genes. This event, called chromoplexy (indicated by the starburst), *These authors contributed equally to this work.
happens early in the evolution of the disease in a prediagnostic lesion. After this event, the diagnostic †Corresponding author. Email: adam.shlien@sickkids.ca (A.S.);
sb31@sanger.ac.uk (S.B.); david.malkin@sickkids.ca (D.M.)
and relapsed tumors evolve in parallel. In this model, the clone that would ultimately become the Cite this article as N. D. Anderson et al., Science 361,
relapsed tumor was already present at the time of initial diagnosis, although it was undetectable. eaam8419 (2018). DOI: 10.1126/science.aam8419
Rearrangement bursts generate of patients with ES, and overall low number of
mutations, we found that the tumors contained
at least seven distinct signatures, all of which
canonical gene fusions in bone and matched patterns seen in adult cancers [Cata-
logue of Somatic Mutations in Cancer (COSMIC)
soft tissue tumors signatures 1, 2, 5, 8, 13, 18, and 31; Fig. 1B and
fig. S2A] (11, 12). Two of these (signatures 1 and
5) were nearly universal and associated with
Nathaniel D. Anderson1,2, Richard de Borja1*, Matthew D. Young3*, Fabio Fuligni1*, patient age. Signature 1 generated a steady rate
Andrej Rosic1, Nicola D. Roberts3, Simon Hajjar1†, Mehdi Layeghifard1, of seven mutations per gigabase per year, which
Ana Novokmet1†, Paul E. Kowalski1, Matthew Anaka1, Scott Davidson4, Mehdi Zarrei5, is similar to that of adult ovarian and breast
Badr Id Said1, L. Christine Schreiner1, Remi Marchand1, Joseph Sitter1, Nalan Gokgoz6, cancers (fig. S2B) (13). An overview of the so-
Ledia Brunga1, Garrett T. Graham7, Anthony Fullam3, Nischalan Pillay8,9, matic architecture and mutational signatures of
Jeffrey A. Toretsky7, Akihiko Yoshida10, Tatsuhiro Shibata11,12, Markus Metzler13, each tumor in our discovery cohort is shown in
Gino R. Somers2,14, Stephen W. Scherer1,5,15,16, Adrienne M. Flanagan9,10, Fig. 1, A to C (left panels, Toronto cohort).
G
Research Institute, Sinai Health System, Toronto, Ontario,
enomic rearrangements (structural var- The starting point of our investigation was Canada. 7Department of Oncology and Pediatrics,
Georgetown University, Washington, DC, USA. 8University
iants) are a ubiquitous source of somatic Ewing sarcoma (ES), a bone and soft tissue College London Cancer Institute, Huntley Street,
mutation in human cancer. They arise cancer predominantly diagnosed in adolescents London, UK. 9Histopathology, Royal National Orthopaedic
from breaks in chromosomes, which are and young adults. It represents the prototypical Hospital NHS Trust, Stanmore, Middlesex, UK. 10Department
then aberrantly rejoined. Rearrangements fusion-driven sarcoma, defined by fusions be- of Pathology and Clinical Laboratories, National Cancer
Center Hospital, Tokyo, Japan. 11Division of Cancer
may occur in isolation or in the context of com- tween EWSR1, a gene encoding an RNA binding Genomics, National Cancer Center Research Institute, Tokyo,
plex genomic catastrophes that shatter chromo- protein, and genes encoding E26 transformation- Japan. 12Laboratory of Molecular Medicine,
somes [chromothripsis (1)] or join chromosomes specific (ETS) transcription factors, including Human Genome Center, The Institute of Medical Sciences,
in chains or loop structures [chromoplexy (2)]. FLI1 and ERG (5). Although the downstream The University of Tokyo, Minato-ku, Tokyo, Japan.
13
Department of Pediatrics and Adolescent Medicine,
Rearrangements can generate cancer-driving mu- consequences of EWSR1-ETS fusion genes are University Hospital Erlangen, Erlangen, Germany.
tations through several mechanisms, including well established (6), the timing and mechanism 14
Department of Pathology, Hospital for Sick Children,
the formation of gene fusions. Typically, fusions by which they arise are unknown. University of Toronto, Toronto, Ontario, Canada.
15
are fashioned by translocations that are often Department of Molecular Genetics, University of Toronto,
reciprocal. An exception is the prostate cancer Burden and signatures of substitution Toronto, Ontario, Canada. 16The McLaughlin Centre,
mutations in ES University of Toronto, Toronto, Ontario, Canada.
fusion gene, TMPRSS2-ERG, which can occur 17
Department of Haematology, University of Cambridge,
through chromoplexy (2). We sequenced either the gene-containing por- Cambridge, UK. 18Departments of Pediatrics and Oncological
Oncogenic gene fusions are particularly com- tions or the whole genomes of 50 ES tumors Sciences, Huntsman Cancer Institute, University of Utah, Salt
Lake City, UT, USA. 19Department of Cellular and Molecular
mon in leukemia and bone and soft tissue tumors and their matched normal DNA (complete Medicine and Department of Bioengineering and Moores
(3), often acting as the sole driver mutation and sequencing details in table S1). We used a Cancer Center, University of California, La Jolla, San Diego,
delineating clinically relevant tumor entities and conventional analysis pipeline to call somatic CA, USA. 20University Musculoskeletal Oncology Unit, Mount
subgroups. In leukemia, recombination-activating substitutions and rearrangements, with addi- Sinai Hospital, Toronto, Ontario, Canada. 21Division of
Orthopaedic Surgery, Department of Surgery, University of
gene (RAG)–mediated structural variation has tional custom software to remove recurrent Toronto, Toronto, Ontario, Canada. 22Division of Hematology-
been identified as the leading mutational pro- artifacts and sources of false positives (see Oncology, The Hospital for Sick Children, Toronto, Ontario,
cess that creates canonical gene fusions and methods and fig. S1). Overall, and consistent Canada. 23Department of Pediatrics, University of Toronto,
drives oncogenesis through translocations and with previous reports (7–10), the ES genome Ontario, Canada. 24Department of Paediatrics, University of
Cambridge, Cambridge, UK.
deletions (4). Here we sought to investigate pro- is genetically quiet, with few somatic substitu- *These authors contributed equally to this work. †Deceased
cesses and timing of oncogenic fusions in hu- tions identified (median of <1 mutation/Mb; ‡Corresponding author. Email: adam.shlien@sickkids.ca (A.S.);
man bone and soft tissue tumors. Fig. 1A). sb31@sanger.ac.uk (S.B.); david.malkin@sickkids.ca (D.M.)
distance of <10 kilobase pairs (kbp) (see meth- patient outcomes, and known markers of worse through chromoplexy (n = 168 gene disruptions,
ods). Using a computational data structure that prognosis. We found that higher overall genomic and n = 47 fusions; Fig. 3C). Given that chro-
modeled adjacent breakpoints as vertices and complexity, a marker of aggressive ES (10, 15), was moplexy appeared to mark an aggressive form
interconnected rearrangements as edges in a almost completely explained by chromoplectic re- of ES, we wondered if its gene expression pro-
graph, we uncovered several distinct configu- arrangements (Fig. 1F). By contrast, there was no gram was globally different—above and beyond
rations of rearrangement clusters (Fig. 1D). As difference in the burden of nonchromoplectic rear- the immediate, focal structural consequences
expected, one configuration of rearrangement rangements. Similarly, TP53 mutations, another listed above. We identified 504 differentially ex-
clusters was a result of reciprocal rearrange- established marker of poor prognosis (10, 16), pressed genes in chromopletic ES compared to
ments, where there is an equal exchange of were enriched in chromoplexy ES (16 versus simple ES (P < 0.001, Student’s t test; Fig. 3D).
genetic material and overlapping breakpoints. 3%, P < 0.05, Fisher’s exact test). There was no Gene set enrichment analysis of well-curated
These were isolated rearrangements that oc- enrichment for CDKN2A or STAG2 mutations pathways (18) uncovered a significant enrich-
curred without additional breakpoints nearby. (fig. S7). Finally, and consistent with the above, ment of dysregulated genes in established can-
A second configuration, seen in 14/24 tumor patients with chromoplexy ES were more likely cer hallmark pathways (table S3).
genomes, was a distinctive pattern of focal clus- to relapse (54 versus 30%, P < 0.05, Fisher’s exact Both prostate cancer and ES loops were
tered events with nearly overlapping junctions, test), strongly suggesting that it marks a more characterized by focal intrachromosomal rear-
organized as closed loops (distance of <30 bp; aggressive variant of ES. rangements, so-called deletion bridges (2), that
Fig. 1D, red distribution). That is, if one follows acted as local mediators of large-scale loops
these complex rearrangements across their mul- Chromoplexy generates the key fusion (illustrated in fig. S11). We found deletion bridges
tiple constituent chromosomes, one is ultimately in other bone and soft tissue tumors in ~60% (30/52) of chromoplectic ES. In contrast
brought back to the point of departure. Impor- We next widened our search across four dif- to the bridges observed in prostate cancer, more
tantly, the loops were nearly always centered on ferent benign and malignant bone and soft tis- than a third of bridges are utilized in ES in a
contained the chromoplexy-associated fusion, increase of COSMIC signature 31, which has been Early divergence and parallel evolution
whether it was from a metastasis at the time of recently associated with exposure to platinum of ES tumors
diagnosis or a relapse arising later (Fig. 5B). The therapy in chronic myelomonocytic leukemia A predominant model for tumor progression
pattern of point mutations was also distinct. (24). Notably, our patient had been treated with posits that metastases originate directly from
There was an enrichment of signatures 8 and/or carboplatin for retinoblastoma 3 years before the primary tumor. The metastatic lesions may
18, in addition to the clocklike signature seen at diagnosis of ES. At least three other patients in have acquired new mutations, but, because they
diagnosis, suggesting that new processes drive the validation cohort had a similar signature in are thought to be derived by linear clonal evo-
relapse and metastatic ES (Fig. 5B). For example, their ES, which may also have been treatment lution, most of the genomic properties of the pri-
in one patient’s tumor, we found a pronounced induced (Fig. 1B). mary tumor will be found in the metastasis (25).
gender, tumor site, stage, etc.) were obtained were also obtained from Universitätsklinikum except for a 37-year-old individual who was in-
from the corresponding institutional tumor Erlangen, Erlangen, Germany. All metastatic or stead found to have a FUS-ERG translocation.
banks (table S5). Overall, the patients’ clinical relapse ES tumors were collected from the This patient’s gene expression profile (by RNA-
features and demographics were typical of ES: SickKids tumor bank or the SickKids cancer se- seq) was also discrepant, so they were re-
the average age of diagnosis was 14.8 years (2.8 to quencing program (KiCS). Detailed information on moved from subsequent analyses (fig. S19). One
36.6 years); the male to female ratio was 1.38:1; KiCS is available at https://www.kicsprogram.com. additional genome was removed due to poor
and 14 patients had relapsed, with 13 having died Of the 25 high-coverage genomes sequenced, sequencing quality. We also performed low pass
from their disease. Additional samples (n = 3) EWSR1-ETS fusions were detected in all patients (~10X) rearrangement screens on 19 ES samples.
However, as we required breakpoint resolution, sequenced using the same technology and to a moplexy was already known from the whole
all but one of the rearrangement screens were similar depth of coverage (n = 133). We removed genome and uncovered new instances in sam-
excluded from this study due to insufficient cov- any putative rearrangements if present in ≥2 ples that had not been whole-genome sequenced
erage (see table S1, orange row). Taken together, normals. For rearrangements, breakpoints must (n = 7 and 4, respectively). Each tumor had three
our discovery cohort consisted of individuals exist on both sides of the junction within a 1-kb or four rearrangements validated using the panel.
from which 23 standard genomes (30X to 60X) window. We found that as we increased the All had the same breakpoint (as found by the
and one rearrangement screen genome (20X) number of normals in our panel, our specificity whole-genome sequence) and were found to
were sequenced. The validation cohort consisted increased (fig. S1C). harbor looped rearrangements are on the same
of individuals from which 119 tumor-normal sam- 3. Quality-control filtering. Putative rear- derivative chromosomes.
ples were sequenced by Tirode et al. (10), which rangements were removed if supported by reads
we downloaded from the European Genome- with MAPQ < 30. Putative rearrangements were Validation by FISH, G-banding,
phenome Archive (accessions: EGAS00001000855 removed if they met any two of the following or spectral karyotyping
and EGAS00001000839). Of these, 19 patient criteria: We further validated these looped rearrange-
samples were omitted either because the EWSR1- (i) Non-unique mapping. <70% of the reads ments by karyotyping ESs using standard G-
ETS fusion was not detected by our pipeline and at the locus map uniquely. banding as well as spectral karyotyping (n = 17
manual inspection of the aligned reads, or be- (ii) Multimapping clusters. At the same locus and 3; fig. S4). By cytogenetics we found addi-
cause they harbored an excess of artefactual small (200 bp up and downstream), a pattern of mul- tional complexity—beyond the canonical chr22-
inversions or deletions. tiple overlapping groups of discordant reads chr11 translocation—in eight cases. Of these, six
whose paired-ends align to different chromo- tumors had been sequenced and found to be
Code availability somes (>3 reads in each group, mapping to complex. Additionally, there were five cases for
Custom code described here is available at >4 chromosomes). Seen in both the tumor which chromoplexy was detected by genome
polymorphic copy number variants (CNVs) (36). multiple independent fusion algorithms, and samples were from diagnostic biopsies taken
As expected, the BAF of germline CNV deletions then removed those found in normal tissue. before chemotherapy. The raw MIP data from
followed a bimodal distribution with peaks at Putative fusions were validated by de novo the completed assay was loaded into Nexus
0.5 and 1.0, for heterozygous and homozygous assembly. A total of 1277 normal (nonneoplastic) Copy Number (BioDiscovery, Inc., El Segundo,
rearrangements, respectively (fig. S14B, green samples from 43 different tissues were obtained California) for copy number detection using
line). We then compared the BAF of somatic from the NHGRI GTEx consortium (database default settings.
rearrangements in chains to those without. version 4) and used to remove artifacts. All
Chained rearrangements had higher BAFs than fusions were visually inspected if one or both
REFERENCES AND NOTES
simple structural variants (fig. S14B, red versus genes involved chromoplexy or were adjacent
1. P. J. Stephens et al., Massive genomic rearrangement acquired
blue line), confirming that chromoplectic rear- (up to 1 Mbp). Fusions were further filtered by in a single catastrophic event during cancer development. Cell
rangements are in fact earlier. quality of the realigned transcript, breakpoint 144, 27–40 (2011). doi: 10.1016/j.cell.2010.11.055;
coverage, and gene expression. pmid: 21215367
Detection of breakpoint clusters 2. S. C. Baca et al., Punctuated evolution of prostate cancer
of chained rearrangements Detection of gene expression genomes. Cell 153, 666–677 (2013). doi: 10.1016/
j.cell.2013.03.021; pmid: 23622249
Using their interbreakpoint distance, we iden- Gene expression for fusions, differential gene 3. F. Mitelman, B. Johansson, F. Mertens, The impact of
tified rearrangements within 10 kbp of one expression analysis, and principal component translocations and gene fusions on cancer causation. Nat. Rev.
Cancer 7, 233–245 (2007). doi: 10.1038/nrc2091;
another. Using these, we created an undirected analysis was performed using HT-seq (39) to
pmid: 17361217
graph in which two rearrangement breakpoints count the reads aligning to every gene. PCR 4. E. Papaemmanuil et al., RAG-mediated recombination is the
within 10 kbp of one another (a breakpoint duplicates and reads mapping to ribosomal predominant driver of oncogenic rearrangement in ETV6-
cluster) were represented as a vertex and con- RNA, miRNA, and small nucleolar RNA were RUNX1 acute lymphoblastic leukemia. Nat. Genet. 46, 116–125
(2014). doi: 10.1038/ng.2874; pmid: 24413735
nected to other breakpoint clusters (rearrange- removed. We used the trimmed mean of M-value
5. S. Sankar, S. L. Lessnick, Promiscuous partnerships in Ewing’s
21. J. O. Korbel, P. J. Campbell, Criteria for inference of Proc. Natl. Acad. Sci. U.S.A. 108, E1128–E1136 (2011). carcinomas. J. Pathol. 242, 10–15 (2017). doi: 10.1002/
chromothripsis in cancer genomes. Cell 152, 1226–1236 doi: 10.1073/pnas.1110574108; pmid: 22065754 path.4880; pmid: 28127763
(2013). doi: 10.1016/j.cell.2013.02.023; pmid: 23498933 32. A. Shlien et al., Combined hereditary and somatic mutations of
22. R. S. Hansen et al., Sequencing newly replicated DNA reveals replication error repair genes result in rapid onset of ultra- AC KNOWL ED GME NTS
widespread plasticity in human replication timing. Proc. Natl. hypermutated cancers. Nat. Genet. 47, 257–262 (2015). We dedicate this manuscript to Simon Hajjar, a brilliant and fearless
Acad. Sci. U.S.A. 107, 139–144 (2010). doi: 10.1073/ doi: 10.1038/ng.3202; pmid: 25642631 friend, colleague, and student in our lab, and to Ana Novokmet, a
pnas.0912402107; pmid: 19966280 33. H. Li, Toward better understanding of artifacts in variant friend and colleague who held all of us to high standards and was
23. J. Sima, D. M. Gilbert, Complex correlations: Replication timing calling from high-coverage samples. Bioinformatics 30, committed to helping children with cancer. Simon and Ana both
and mutational landscapes during cancer and genome 2843–2851 (2014). doi: 10.1093/bioinformatics/btu356; died from cancer as this work neared completion. We thank The
evolution. Curr. Opin. Genet. Dev. 25, 93–100 (2014). pmid: 24974202 Centre for Applied Genomics (TCAG) NGS and Biobanking facility for
doi: 10.1016/j.gde.2013.11.022; pmid: 24598232 34. A. Morgulis, E. M. Gertz, A. A. Schäffer, R. Agarwala, A fast and sequencing services. We thank J. Lees, N. Alon, G. Collord, and
24. J. Merlevede et al., Mutation allele burden remains unchanged symmetric DUST implementation to mask low-complexity DNA R. Arnold for their contributions toward this work. We thank N. Tan for
in chronic myelomonocytic leukaemia responding to sequences. J. Comput. Biol. 13, 1028–1040 (2006). her work on the illustration used in the manuscript’s summary
hypomethylating agents. Nat. Commun. 7, 10767 (2016). doi: 10.1089/cmb.2006.13.1028; pmid: 16796549 page. Funding: This research project was conducted with support
doi: 10.1038/ncomms10767; pmid: 26908133 35. P. Rice, I. Longden, A. Bleasby, EMBOSS: The European from C17 and partially funded by Ewings Cancer Foundation of Canada
25. J. W. Gray, Evidence emerges for early metastasis and Molecular Biology Open Software Suite. Trends Genet. 16, and Childhood Cancer Canada Foundation. A.S. and D.M. received
parallel evolution of primary and metastatic tumors. Cancer 276–277 (2000). doi: 10.1016/S0168-9525(00)02024-2; financial support from the SickKids Foundation through the Garron
Cell 4, 4–6 (2003). doi: 10.1016/S1535-6108(03)00167-3; pmid: 10827456 Family Cancer Centre. N.D.A is personally supported by an NSERC
pmid: 12892707 36. M. Zarrei, J. R. MacDonald, D. Merico, S. W. Scherer, A copy Canada Graduate Scholarship (Master’s) and a SickKids Restracomp
26. J. F. Brasme, M. Chalumeau, O. Oberlin, D. Valteau-Couanet, number variation map of the human genome. Nat. Rev. Genet. Award. S.B. receives personal Fellowships from Wellcome and the
N. Gaspar, Time to diagnosis of Ewing tumors in children and 16, 172–183 (2015). doi: 10.1038/nrg3871; pmid: 25645873 St. Baldrick’s Foundation. Author contributions: A.S., S.B., and D.M.
adolescents is not associated with metastasis or survival: A 37. J. Cohen, The statistical power of abnormal-social designed the study. N.D.A., R.D., M.D.Y., M.L., A.R., N.D.R., F.F.,
prospective multicenter study of 436 patients. J. Clin. Oncol. psychological research: A review. J. Abnorm. Soc. Psychol. 65, S.H., P.E.K., B.I., L.C.S., R.M., and J.S. performed experiments. N.D.A.,
32, 1935–1940 (2014). doi: 10.1200/JCO.2013.53.8058; 145–153 (1962). doi: 10.1037/h0045186; pmid: 13880271 R.D., M.D.Y., A.R., F.F., M.A., M.S., and L.B.A. collected and
pmid: 24841977 38. J. Cohen, A power primer. Psychol. Bull. 112, 155–159 (1992). analyzed data. L.B., S.D., N.G., A.N., A.F., N.P., A.Y., T.S., M.M., G.R.S.,
27. P. Roberts et al., Ploidy and karyotype complexity are powerful doi: 10.1037/0033-2909.112.1.155; pmid: 19565683 S.W.S., A.M.F., M.S., J.S.W., I.L.A., P.J.C., J.D.S., D.M., S.B., and A.S.
SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/08/29/361.6405.eaam8419.DC1
MATERIALS
REFERENCES This article cites 43 articles, 6 of which you can access for free
http://science.sciencemag.org/content/361/6405/eaam8419#BIBL
PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions
Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights reserved; exclusive
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science is a registered trademark of AAAS.