meta-analysis of neuroblastomas reveals a skewed alk mutation spectrum in tumors with mycn...
TRANSCRIPT
2010;16:4353-4362. Published OnlineFirst August 18, 2010.Clin Cancer Res Sara De Brouwer, Katleen De Preter, Candy Kumps, et al.
AmplificationMYCNMutation Spectrum in Tumors with ALKMeta-analysis of Neuroblastomas Reveals a Skewed
Updated version
10.1158/1078-0432.CCR-09-2660doi:
Access the most recent version of this article at:
Material
Supplementary
lhttp://clincancerres.aacrjournals.org/content/suppl/2011/08/12/1078-0432.CCR-09-2660.DC1.htmAccess the most recent supplemental material at:
Cited Articles
http://clincancerres.aacrjournals.org/content/16/17/4353.full.html#ref-list-1
This article cites by 23 articles, 6 of which you can access for free at:
Citing articles
http://clincancerres.aacrjournals.org/content/16/17/4353.full.html#related-urls
This article has been cited by 17 HighWire-hosted articles. Access the articles at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
To request permission to re-use all or part of this article, contact the AACR Publications
Research. on November 24, 2014. © 2010 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
Research. on November 24, 2014. © 2010 American Association for Cancerclincancerres.aacrjournals.org Downloaded from
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
Hum
MetMut
Sara DArjanGenevJoëlleIsabeHuib
Abst
Neupediatchildrmultimmostmolecthe detarget
AuthorPaediaGhent UDeveloPaediatHuman7DepartMedicaand On9DepartPatholoSpain;ResearStockh
www.a
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
Clinical
Canceresearchan Cancer Biology
a-analysis of Neuroblastomas Reveals a Skewed ALK
R
ation Spectrum in Tumors with MYCN Amplification
e Brouwer1, Katleen De Preter1, Candy Kumps1, Piotr Zabrocki4, Michaël Porcu4, Ellen M. Westerhout6,Lakeman6, Jo Vandesompele1, Jasmien Hoebeeck1, Tom Van Maerken1, Anne De Paepe1,iève Laureys2, Johannes H. Schulte8, Alexander Schramm8, Caroline Van Den Broecke3,Vermeulen1, Nadine Van Roy1, Klaus Beiske9, Marleen Renard5, Rosa Noguera10, Olivier Delattre11,
lle Janoueix-Lerosey11, Per Kogner12, Tommy Martinsson13, Akira Nakagawara14, Miki Ohira14,
Caron7, Angelika Eggert8, Jan Cools4, Rogier Versteeg6, and Frank Speleman1ractPur
neurocy andsignifiExp
were afurtheRes
similaand uR1275F1174occurrtweenphospChromALK e
velopmened therap
s' Affiliatitric Haematniversity Hpmental Grics, UniversGenetics, Ament of Pedl Center, Amcology, Unment of Patgy, Medica11INSERM Uch Unit, Karolm, Swed
acrjourna
wnloaded
pose: Activating mutations of the anaplastic lymphoma kinase (ALK) were recently described inblastoma. We carried out a meta-analysis of 709 neuroblastoma tumors to determine their frequen-mutation spectrum in relation to genomic and clinical parameters, and studied the prognostic
cance of ALK copy number and expression.erimental Design: The frequency and type of ALK mutations, copy number gain, and expressionnalyzed in a new series of 254 neuroblastoma tumors. Data from 455 published cases were used forr in-depth analysis.ults: ALKmutations were present in 6.9% of 709 investigated tumors, and mutations were found inr frequencies in favorable [International Neuroblastoma Staging System (INSS) 1, 2, and 4S; 5.7%]nfavorable (INSS 3 and 4; 7.5%) neuroblastomas (P = 0.087). Two hotspot mutations, at positionsand F1174, were observed (49% and 34.7% of the mutated cases, respectively). Interestingly, themutations occurred in a high proportion ofMYCN-amplified cases (P = 0.001), and this combinedence was associated with a particular poor outcome, suggesting a positive cooperative effect be-both aberrations. Furthermore, the F1174L mutant was characterized by a higher degree of auto-horylation and a more potent transforming capacity as compared with the R1275Q mutant.osome 2p gains, including the ALK locus (91.8%), were associated with a significantly increased
xpression, which was also correlated with poor survival.clusions: ALK mutations occur in equal frequencies across all genomic subtypes, but F1174L mu-
Contants are observed in a higher frequency of MYCN-amplified tumors and show increased transformingcapacity as compared with the R1275Q mutants. Clin Cancer Res; 16(17); 4353–62. ©2010 AACR.
numbstanditumoall casmosowith
roblastoma is the most common solid extracranialric tumor, with an annual incidence of 1 in 100,000en below the age of 15 years (1). Despite intensiveodal treatment, neuroblastoma remains fatal in al-
half of the unfavorable patients. Insights into theular pathogenesis of this disease are required for
t of less toxic and more effective moleculary. Detailed studies of patterns of DNA copy
neuroThe d
ons: 1Center for Medical Genetics, 2Department ofology and Oncology, and 3Department of Pathology,ospital, Ghent, Belgium; 4Department of Molecular andenetics, K.U.Leuven-VIB, and 5Department ofity Hospital Leuven, Leuven, Belgium; 6Department ofcademic Medical Center, University of Amsterdam, andiatric Oncology, Emma Children's Hospital, Academicsterdam, the Netherlands; 8Division of Hematology
iversity Children's Hospital Essen, Essen, Germany;hology, Rikshospitalet, Oslo, Norway; 10Department ofl School of Valencia, University of Valencia, Valencia,830, Institut Curie, Paris, France; 12Childhood Cancerolinska Institutet, Astrid Lindgren Children's Hospital,en; 13Department of Clinical Genetics, Gothenburg
UniversCancer
Note: SResear
S. De B
Corres(CMGGBelgiumspelem
doi: 10
©2010
ls.org
Research. on November 24,clincancerres.aacrjournals.org from
er alterations have been instrumental in our under-ng of the clinical and biological heterogeneity of thisr. Three major genomic subtypes represent >80% ofes, i.e., hyperploid neuroblastoma with whole chro-me gains and losses, near diploid neuroblastoma11q deletions and 17q gain, and MYCN-amplified
blastoma with 1p deletions and 17q gain (2–5).iscovery of rare but recurrent high-level amplificationity, Gothenburg, Sweden; 14Division of Biochemistry and InnovativeTherapeutics, ChibaCancer Center Research Institute, Chiba, Japan
upplementary data for this article are available at Clinical Cancerch Online (http://clincancerres.aacrjournals.org/).
rouwer and K. De Preter contributed equally to this work.
ponding Author: Frank Speleman, Center for Medical Genetics), Ghent University Hospital, De Pintelaan 185, B-9000 Ghent,. Phone: 32-9-332-2451; Fax: 32-9-332-6549; E-mail: franki.
.1158/1078-0432.CCR-09-2660
American Association for Cancer Research.
4353
2014. © 2010 American Association for Cancer
of thetomasneuroin priin theof muhas pdistribent geserieshigh-smentdifferean adical stfindincases (matiothe ALical damutathotspinducformkine-igene esion i
Mate
NeuroIn t
with a
44 sta4S tumStaginnomicmutatmaderefs. 8includ(see SgenotGen
tion k
ALK DFor
lines,remaikinasewas aa mutfrom gTabledirectCycle(Applusing
ArraynumbTo
paratiusingtificiaprevio44K ain neulent Terence(BioPgen).ufactufeaturcessedarrayCref. 12of DN
ALK gGen
Translational Relevance
Our study yielded a number of important new in-sights with clinical implications. First, ALK mutationsare present in similar frequencies in all clinical stagesof neuroblastoma (low as well as high stages). Second,F1174 hotspot mutations are associated with MYCNamplification and their combined occurrence leads tofatal disease outcome in all (except one) patients. Apossible cooperation between the F1174 mutationand MYCN amplification may have implications fortargeted therapy. Third, F1174 mutations have a highertransforming capacity than R1275 mutations. Finally,chromosome 2 copy gain, including the ALK locus, isassociated with an increased ALK expression that wasfound to be associated with a significantly worse out-come in the global population. These findings shed anew and more detailed light on the distribution of ALKmutations in neuroblastoma. This information may beof importance in the light of choice of risk-related ther-apy and development of future targeted therapies.
De Brouwer et al.
Clin C4354
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
ALK gene and a genetic study of familial neuroblas-led to the discovery of activating ALK mutations inblastoma (6–10). The frequency of ALK mutationsmary neuroblastoma varied between 6% and 11%different studies (6–10). The relatively low numbertations described in each of the individual studiesrecluded a thorough analysis of the frequency andution of recurrent ALK mutations across the differ-nomic subtypes. Moreover, in the first published, there was a significant bias towards analysis oftage tumors, thus preventing a more general assess-of frequency and distribution of mutations acrossnt stages (6–8). In our current study, we screenedditional 254 neuroblastoma cases including all clin-ages and genomic subtypes. In a meta-analysis, thesegs were combined with those from 455 published8–10) for which genomic subtype and clinical infor-n were available. This strategy enabled us to analyzeK mutation profile in relation to genomic and clin-ta in 709 neuroblastomas, which revealed a distinction spectrum in relation to genomic subtype. Twoot mutations, F1174L and R1275Q, were shown toe ALK autophosphorylation and were able to trans-interleukin-3 (IL-3)-dependent Ba/F3 cells into cyto-ndependent growth. In addition, we evaluated ALK
xpression levels and showed that high ALK expres- compcustomEBI AdataseST Affymetrix arrays (normalized and summarized at thetranscript level using RMAsketch), and an unpublished
s correlated with poor survival.
rials and Methods
blastoma patients and cell lines
otal, 254 primary untreated neuroblastoma tumorstumor percentage >60%were investigated, including 15 Kostancer Res; 16(17) September 1, 2010
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
ge 1, 30 stage 2, 34 stage 3, 113 stage 4, and 33 stageors [according to the International Neuroblastomag System (INSS); ref. 11]. Patient information and ge-subtypes for 455 published tumors screened for ALKions were retrieved from the publications or wereavailable by the authors (Supplementary Table S1;–10). In addition, 39 neuroblastoma cell lines wereed. The cell lines were obtained from several sourcesupplementary Table S2). All of the cell lines wereyped by DNA fingerprinting (PowerPlex, Promega).omic DNA was isolated using the Qiagen DNA isola-it (Qiagen) or a standard proteinase K/SDS procedure.
NA sequence analysisthe first tumor cohort (146 cases) and the 39 cellall 29 ALK coding exons were analyzed, whereas thening 108 tumors were screened for only the tyrosinedomain. Constitutional DNA from blood samples
vailable and analyzed for 12 of the 17 patients withation in the primary tumor. Exons were amplifiedenomic DNA (primer information in SupplementaryS3). PCR products were subjected to directional or bi-ional sequencing using BigDye Terminator V1.1/V3.1Sequencing chemistry on an ABI3730XL sequenceried Biosystems). Electropherograms were analyzedSeqscape v2.5 software (Applied Biosystems).
comparative genomic hybridization copyer profilingdetermine DNA copy number alterations, array com-ve genomic hybridization (arrayCGH) was done byan in-house developed 1 Mb resolution bacterial ar-l chromosome array (BAC) array (37 samples) asusly described (3) or by using a custom-designedrray enriched for regions with recurrent imbalancesroblastoma (1p, 2p, 3p, 11q, 17; 217 samples; Agi-echnologies). For the latter, 150 ng of tumor and ref-DNA were labeled with Cy3 and Cy5, respectively
rime ArrayCGH Genomic Labeling System, Invitro-Further processing was done according to the man-rer's guidelines. Features were extracted using thee extraction v10.1.0.0.0 software program and pro-with an in-house developed visualization softwareGHbase (http://medgen.ugent.be/arrayCGHbase;), including circular binary segmentation for scoringA copy number alterations (13).
ene expression datae expression data were available for 440 tumors,rising a published dataset of 251 tumors profiled on
Agilent 44k arrays (ref. 14; downloaded from therrayExpress database E-TABM-38), an unpublishedt of 101 tumors profiled on the Human Exon 1.0
er et al., submitted.
Clinical Cancer Research
2014. © 2010 American Association for Cancer
datasU133loadewereform.sion lvalue
TissuFor
represH&E-embe(SuppclassifPathowhichentiatwerechemhumaslideswherecells)cells)the cethe ceculate
WesteLysi
linesof prousingSignalAidaof We
TransBa/
with 1(PeprwerepMSCperimwereamouFor pPBS, aout ILViabiland 9
StatisFish
analyspackacarry
sion apacka
Resu
ALK mtumoIn a
blasto2 ALKtary Fexonsthe tykinasemors.duesof 25thirdF1245mutatdetectthus problasportedR1231of 17tationMu
cell lLAN-6domathat wAll resdancementa
Meta-occurtumoWe
caseslishednomicThe 7The tF1174tary Ftationoveralence wableunfavmas (were aor age1 for
Skewed ALK Mutation Spectrum in MYCN-Amplified Neuroblastoma
www.a
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
et of 88 tumors profiled on the Affymetrix HG-plus2.0 platform (normalized usingMAS5 and down-d from the R2 database).15 Importantly, analysesdone on sets of tumors profiled on the same plat-For Kaplan-Meier and log-rank analysis, ALK expres-evels were digitalized using the median expressionas cutoff.
e microarray and immunohistochemistrythe establishment of a tissue microarray, threeentative areas from each tumor were selected onstained slides from 70 formalin-fixed and paraffin-dded primary untreated neuroblastoma tumorslementary Table S4). Neuroblastoma tumors wereied according to the International Neuroblastomalogy Classification (INPC) scoring system (15),divides tumors into undifferentiated, poorly differ-
ed, and differentiating. Of each tumor, three corespunched into the recipient block. Immunohisto-istry was done using a monoclonal mouse anti-n CD246 ALK antibody (clone ALK1, Dako), andwere scored for immunoreactive neuroblastoma cells0 meant no or weak staining intensity (<10% of the, 1 meant weak staining intensity (10-50% of the, 2 meant medium staining intensity (50-80% oflls), and 3 meant high staining intensity (>80% oflls). The medium score of the three punches was cal-d for Kaplan-Meier and log-rank analysis.
rn blottings of cells was done when the neuroblastoma cellreached 70% confluency. Total cell lysates (50 μgtein) were analyzed by standard procedures (16)anti-phospho-ALK (Tyr1604), anti-ALK (Celling), and anti-ERK2 (Santa Cruz) antibodies. TheImage Analyzer v.4.22 was used for quantificationstern blots.
formation assayF3 cells were cultured in RPMI-1640 supplemented0% fetal bovine serum and 1 ng/mL murine IL-3otech). ALK F1174L and ALK R1275Q constructsgenerated by PCR, cloned into the retroviral vectorV-neo (Clontech), and transduced in Ba/F3 cells. Ex-ents were done in triplicate. Transduced Ba/F3 cellsselected with G418 (500 μg/mL medium). Thent of viable cells was assayed on regular intervals.roliferation curves, Ba/F3 cells were washed withnd 5 × 105 cells were seeded in 5 mL medium with--3. Viable cells were counted using a Vi-CELL Cellity Analyzer (Beckman Coulter) on days 3, 5, 7,.
tical analysiser's exact tests, Mann-Whitney tests, and correlationes were done using R (version 2.8.1). The R Survival
ge was used to generate Kaplan-Meier plots and toout log-rank analyses. Multivariate logistic regres-typesparam
acrjournals.org
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
nalysis was done using the glm function (R-basege).
lts
utation analysis in 254 primary neuroblastomars and 39 neuroblastoma cell linesseries of 254 sporadic, nonfamilial primary neuro-ma tumors, a total of 17 ALK mutations (6.7%) andamplifications (0.8%) were identified (Supplemen-
ig. S1). In a first series of 146 cases, all 29 ALK codingwere analyzed but no mutations were found outsiderosine kinase domain. Therefore, only the tyrosinedomain was analyzed for the remaining 108 tu-The most frequent mutations were located at resi-R1275 and F1174, and were detected in 3.9% (104) and 2.0% (5 of 254) of the cases, respectively. Arecurrent but less frequent mutation affecting residue(6–9) was not detected in our series. One of the
ions previously reported in only one tumor was alsoed as a single case in our series (Y1278S; ref. 10),roviding further evidence for its contribution to neu-toma pathogenesis. In addition to the previously re-mutations we observed one newmissensemutation,Q. Sequence analysis of the constitutional DNAof 12patients with an ALKmutation showed that thesemu-s were somatically acquired.tation analysis was also done in 39 neuroblastomaines. All mutations, except for one (D1091N in), were exclusively found within the tyrosine kinasein. Recurrent mutations were found in two cell linesere previously not analyzed, STA-NB-8 and NB-14.ults of previously reported cell lines were in concor-with published results except for one (see Supple-ry Table S1).
analysis of ALK mutations reveals an increasedrence of the F1174 mutation in MYCN-amplifiedrscarried out a meta-analysis on the new cohort of 254together with 455 previously analyzed and pub-cases to relate the ALK mutations to clinical and ge-data (8–10; Table 1 and Supplementary Table S1).
09 samples had in total 49 ALK mutations (6.9%).wo most frequently occurring mutations were the(34.7%) and R1275 mutations (49%; Supplemen-
ig. S1). When the entire group of different ALK mu-s was taken into account in the evaluation of thel frequency of ALK mutations, no significant differ-as observed for the frequency of mutations in favor-(INSS 1, 2, and 4S; 14 of 245, 5.7%) versusorable (INSS 3 and 4; 33 of 440, 7.5%) neuroblasto-Fisher's exact test P = 0.087). Mutation frequencieslso compared within different stages, MYCN status,, but no significant differences were found (see TableP values). However, when looking at the different
of mutations in relation to the clinical and genomiceters, we noticed a skewed distribution for MYCNClin Cancer Res; 16(17) September 1, 2010 4355
2014. © 2010 American Association for Cancer
statuswild-tmutattumorOf thMYCNMYCNtype Aamplwild-twe seqer thecouldpresenone Rmenta
FrequsubtyTo
tion smors(3). Ftype: ssubtypficatio(n = 1copyered msifiedA c
tion twere
mors1 tumand sumutat(Fig. 1
CorreNo
withrank PpatiencompKaplaenceslargelplificawith tInte
noticeand Fin themutatparedampli
ALK aOve
12 ofvatingfor on0.001
Tableand gcases
MYCN1p-del3p-del11q-de2p-gai17q-gaAgeStagGen
NOTmut*P <†P <‡Data are based on two of the four datasets for which detailed arrayCGH data were available.
De Brouwer et al.
Clin C4356
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
in the F1174 mutated cases versus the tumors withype ALK (P = 0.001; Fig. 1A and Table 1). ALK F1174ions were found in 1.3% of the MYCN-single copys, compared with 6.1% of MYCN-amplified tumors.e 17 tumors with the F1174 mutation, 58.8% had
amplification, compared with a frequency ofamplification of 21.6% in the tumors with wild-LK (Fig. 1A). In contrast, the frequency of MYCNification was similar for R1275 mutated versusype cases. As most cell lines were MYCN amplified,uenced 39 neuroblastoma cell lines to verify wheth-association of F1174 with MYCN amplificationalso be detected. Indeed, F1174 mutations weret in 5 of 27 MYCN-amplified cell lines whereas only1275 mutation was present in this cohort (Supple-ry Table S2).
ency of ALK mutations according to genomicpefurther explore the relationship between ALK muta-tatus and genomic alterations, we classified the tu-into genomic subclasses based on arrayCGH dataor 659 tumors, we could establish the genomic sub-ubtype 1 with numerical imbalances only (n = 218),e 2A with 11q deletion and without MYCN ampli-n (n = 126), subtype 2B with MYCN amplification58), and subtype 3 without any detectable DNA
number alterations (n = 78). This classification cov-ost of the cases (88%), with 79 remaining unclas-(Supplementary Fig. S2).omparison of the ALK mutation frequency in rela-
o genomic subtype revealed that ALK mutationsmost frequently observed in MYCN-amplified tu-plifiedstatist
ancer Res; 16(17) September 1, 2010
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
(subtype 2B; 8.9% mutated), followed by subtypeors (7.3% mutated), subtype 2A tumors (4.0%),btype 3 tumors (1.3%). Interestingly, all infrequentions (6 of 49) were present in subtype 1 tumorsB).
lation of ALK mutation with survivalsignificant survival differences were found in tumorsor without ALK mutations (or amplifications; log-= 0.317; Fig. 2A). However, when the survival ofts with R1275 mutation or wild-type patients wasared with patients with the F1174 mutation type,n-Meier analysis showed significant survival differ-(P = 0.027 and P = 0.002; Fig. 2B and C). This mighty be explained by the high frequency of MYCN am-tion within the F1174 mutated tumors comparedhe R1275 mutated tumors.restingly, although not statistically significant, wed that 9 of 10 patients with a MYCN amplification1174 mutation died of disease, suggesting that with-MYCN-amplified subgroup, patients with the F1174ion may have a particularly poor survival as com-with a 32% 5-year overall survival rate for MYCN-fied cases without the F1174 mutation.
mplificationrall, ALK amplifications could be detected in only709 tumors (1.7%), and none of these carried acti-ALK mutations. All ALK-amplified tumors, excepte, were also found to be MYCN amplified (P <), which accounts for 6.7% of the total ofMYCN-am-
1. Results of Fisher-exact analysis comparing the distribution of genomic alterations, age, stage,
enomic subgrou p in cases with one of t he frequent A LKmutations ortumors. Likeically significa
2014. © 2010
amplification
ALK mutationnt independe
Cli
American As
versus wild
, amplificationnt marker for
nical Cancer R
sociation for C
-type
Amplificvs wild-
ation or mutationtype
Mutatiowild-typ
n vse
Amplificwild-typ
ation vse
F1174 vwild-typ
ersuse
R1275 vwild-typ
ersuse
F1174 vR1275
ersus
status
2.05E-04* 7.74E -02† 6. 49E-07* 1.16 E-03* 6.09E -01 5.31E -02†etion‡
9.94E-02† 8.21E -01 8. 03E-02† 3.57 E-01 7.68E -01 2.91E -01 etion‡ 3.45E-02† 5.54E -02† 1. 00E-00 1.00 E-00 2.36E -01 1.00E -00 letion‡ 5.45E-02† 8.19E -02† 1. 00E-00 1.00 E-00 2.09E -01 5.00E -01 n‡ 7.98E-01 1.00E -00 1. 00E-00 6.02 E-01 1.00E -00 5.26E -01 in‡ 6.85E-01 6.72E -01 1. 00E-00 7.03 E-01 1.00E -00 1.00E -00 > 1 year) 4.07E-01 4.47E -01 7. 73E-01 8.09 E-02† 6.58E -01 9.33E -02† (<e 9.94E-02† 1.66E-01 4.64E-01 6.70E-01 4.56E-01 8.13E-01omic subtype 9.32E-04* 9.35E-02† 5.04E-06* 2.20E-02* 7.71E-01 3.92E-01
E: The last column gives the results of the comparison of the cases with the F1174 mutation versus cases with the R1275ation.0.05.0.1.
is not asurvival
esearch
ancer
whenlogisti
ALK lsurvivLow
whole
Fig. 1.subtypeamplificamplific
Skewed ALK Mutation Spectrum in MYCN-Amplified Neuroblastoma
www.a
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
analyzed in a model with MYCN, stage, and age in ac regression analysis (data not shown).
ow copy number gain, gene expression, andal
copy number gain of chromosome 2p material orFig. 2.amplifie
chromosome 2 gain was detected in 19.3% (49 of
Comparison of the distribution of MYCN status (A) and genomic(B) in tumors with one of the two frequent ALK mutations or ALK
ation versus wild-type tumors. Wt, wild-type ALK; Amp, ALKation; MNA, MYCN amplification.R1275(C). DO
acrjournals.org
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
Kaplan-Meier and log-rank analysis of ALK-mutated and ALK-d tumors versus ALK wild-type cases (A), F1174 mutated versus
mutated cases (B), and F1174 mutated versus wild-type casesD, dead of disease.Clin Cancer Res; 16(17) September 1, 2010 4357
2014. © 2010 American Association for Cancer
254)keepistudiewereextra87 MbIn thewith aimmeFig. S3of paparticnumbcus windicaincreaTo
gain oray extomasnumbcorrelfied sary Fig.compcopy>0.3;Nex
pressithreesamplfiled otion opatiengressiotic regage, sdent plargerintervconfir
ALK iIn a
pressiexpresarraylog-raALK p0.014Fig. 4)ent on(scoreALK re
Forphospas theminedimmuroblasmRNAtein leclearlymRNAwith pestingF1174very lorelativwith twild-t(Supp
TransGiv
and ApacityBa/F3Alth
cellsF1174Ba/F3
Discu
Theroblasparamvationtwo hin MYF1174SeconmutaF1174transfR1275ports,icant4S (5.showencompressia worshowtumo
16 Koste
De Brouwer et al.
Clin C4358
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
and 17.7% (45 of 254) of the cases, respectively, inng with the high occurrence reported in previouss (6, 8–10, 17). No focal low copy number gainsdetected for ALK. In cases with partial 2p gain, thechromosomal segments varied in size from 15 toand included the MYCN gene except for one case.latter, MYCN amplification was present togethermore distal 15 Mb gain with a telomeric breakpointdiately proximal to the ALK locus (Supplementary). The ALK gene was included in all except four casesrtial 2p gain (91.8%). An interesting observation,ularly in view of the high frequency of ALK copyer gain, was that 2p gains encompassing the ALK lo-ere present in only 2 of 17 mutated tumors (11.8%),ting that 2p gain is not a common mechanism forsing mutated ALK copy number.evaluate the possible impact of ALK copy numbern ALK expression, we compared Affymetrix exon ar-pression data and arrayCGH data for 101 neuroblas-. This showed a strong correlation between copyers of the ALK gene and expression levels (Spearmanation coefficient, 0.308; P = 0.002; one ALK-ampli-mple was omitted from this analysis; Supplementa-S4). This was confirmed by Mann-Whitney analysisaring the expression in tumors with normal ALKnumber versus tumors with ALK gain (CGH resultP = 0.001).t, we evaluated the relation between ALK gene ex-on and survival. We analyzed expression levels inindependent datasets including a total of 440 tumores (14).16 Each analysis was done on samples pro-n one particular platform. These showed a correla-f ALK gene expression with survival in the globalt population (log-rank P values for overall and pro-n-free survival <0.05; Fig. 3A-C). Multivariate logis-ression analysis (in a model testing MYCN status,tage, and ALK expression) could show an indepen-rognostic value for ALK mRNA expression in thedataset (ref. 14; odds ratio, 2.94; 95% confidenceal, 1.29-6.69; P = 1.02E-2), but this could not bemed in the other two smaller datasets.
mmunoreactivityddition to the relation between ALK transcript ex-on and survival, we also investigated the ALK proteinsion status and patient survival using a tissue micro-containing 70 primary tumors. Kaplan-Meier andnk analysis show a significant correlation betweenrotein expression and overall survival (OS; P =) and progression-free survival (PFS; P = 0.002;. The three cases with ALKmutation (R1275Q) pres-the tissue microarray had a median expression value
2) whereas in the remaining ALK wild-type tumors,activity varied from low (score 0) to high (score 3).ALK anumbThe
F1174mutatco-ocr et al., submitted.
ancer Res; 16(17) September 1, 2010
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
the evaluation of ALK activity, the level of activated/horylated ALK (p-ALK) was investigated. However,specificity of p-ALK antibodies remains to be deter-, we carried out Western blot analysis rather thannohistochemistry experiments on a panel of 22 neu-toma cell lines. This allowed for comparison of ALKexpression levels (P = 0.011) and native ALK pro-vels (P = 1.55E-07) versus p-ALK protein level. Weshow a significant correlation between both ALKexpression levels and native ALK protein levels
-ALK protein levels (Supplementary Fig. S5). Inter-ly, we could also show that cell lines harboring themutation (except for the SK-N-SH cell line withw ALK expression levels) or ALK amplification haveely more phosphorylated ALK than do cell lineshe other hotspot mutation. Moreover, cell lines withype ALK have very low levels of phosphorylated ALKlementary Figs. S6 and S7).
forming capacity of ALK hotspot mutationsen the observed concordance between mutation typeLK activity, we also compared the transforming ca-of both ALK hotspot mutations in IL-3–dependentcell lines.ough both mutants were able to transform Ba/F3
to IL-3–independent growth, cells expressing ALKL transformed the cells significantly faster than didcells expressing ALK R1275Q (Fig. 5).
ssion
present meta-analysis of ALKmutations of 709 neu-tomas in relation to genomic profiles and clinicaleters resulted in a number of new important obser-s. First, substitutions at residue F1174, one of theotspot mutations, were significantly overrepresentedCN-amplified tumors. Moreover, patients with themutation present with a particularly poor outcome.d, our results indicate that, although both hotspottions have constitutive ALK phosphorylation, theL mutation has stronger autophosphorylation andormation capacity of Ba/F3 cells than does theQ mutation. Third, in contrast to some previous re-we show that ALK mutations also occur in a signif-proportion of tumors with favorable stages 1, 2, and7% versus 7.5% in stage 3 and 4 tumors). Fourth, wethat copy number gain of the chromosome 2 regionpassing ALK is associated with an increased ALK ex-on. Fifth, increased ALK expression is associated withse outcome in the global population. Finally, wethat chromosome 2p is not frequently gained inrs with ALK mutations, indicating that mutatedlleles are not selected for high expression by copyer gain.functional relevance of the high proportion ofmutations in the subset of MYCN-amplified ALK-
ed neuroblastomas remains undetermined, but theircurrence may suggest a cooperative effect between
Clinical Cancer Research
2014. © 2010 American Association for Cancer
bothmightbenefmay e
tientsampli
Fig. 3. K vival (Pa publis s of 10
Skewed ALK Mutation Spectrum in MYCN-Amplified Neuroblastoma
www.a
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
aberrations in these tumors. The F1174 mutationcontribute to an additional growth and survival
aplan-Meier and log-rank analysis for overall (OS) and progression free surhed set of 251 neuroblastoma tumors (A; ref. 14), and in unpublished set
it in MYCN-amplified neuroblastoma cells, whichxplain the particularly poor survival of these pa-
mutattage th
acrjournals.org
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
. The fact that neuroblastoma cell lines with MYCNfications have a relatively high frequency of F1174
FS) comparing tumors with high and low ALKmRNA expression in1 (B) and 88 (C) neuroblastoma tumors.
ions might also point at a particular growth advan-at may have facilitated in vitro growth of these cells.
Clin Cancer Res; 16(17) September 1, 2010 4359
2014. © 2010 American Association for Cancer
Of fuF1174up topartic10 paF1174overalmorsThe
and thal rais
tationGeorgbothF3 celin oncbecau(7). InF3 celR127
Fig. 4. Kaplan-Meier and log-rank analysis for overall (OS) and progression free survival (PFS) comparing tumors with high (expression ≥2) and low(expres
Fig. 5.R1275Qparenta9 days.becaus
De Brouwer et al.
Clin C4360
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
rther interest in this context is the observation thatmutations in the germline have not been reportednow, which could suggest embryonic lethality. Ofular interest was the observation that only 1 of thetients with MYCN amplification together with amutation survived, in contrast to a 32% 5-year
l survival rate in patients with MYCN-amplified tu-without the F1174 mutation.different distribution across the genomic subtypes
sion <2) ALK protein expression. X-axis, number of days.
e adverse impact of the F1174 mutations on surviv-e the question of whether the F1174 and R1275 mu-
sphorservedThe
tion ttumoneckand βof chrknowthe saservedreportOu
studiein boanalysmors,tumoIn a
fromstudierent bIn keemostfrom
A day 9 value for ALK F1174L was not included in the figuree the culture had reached maximal density before this time point.
ancer Res; 16(17) September 1, 2010
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
s may execute distinct effects on tumor biology.e and colleagues have previously also shown thatF1174L and R1275Q mutants could transform Ba/ls, but their analysis did not reveal major differencesogenic potential between these two mutants, in partse the proliferation data were not reported in detailour hands, the F1174L mutant transformed the Ba/ls more efficiently than did the cells expressing the5Q mutant. This correlated with higher autopho-ylation levels of ALK F1174L, which was not ob-for the R1275Q mutant.occurrence of mutations of particular genes in rela-o genomic subgroups has been reported in certainr entities, such as PIK3CA mutations in head andsquamous carcinomas without EGFR amplification-catenin mutations in medulloblastomas with lossomosome 6 (18–20). However, to the best of ourledge, a different distribution for mutations withinme functional domain of one specific gene, as ob-here for the F1174 mutation in ALK, has not beened.r study also showed, in contrast to some of the initials, that ALKmutations occur in fairly equal frequencyth low- and high-stage tumors. Therefore, mutationis should also be done in patients with low-stage tu-and the clinical characteristics and behavior of suchrs should be carefully monitored in further studies.ddition to mutations, ALK activation can also resulthigh level gene amplification as shown by previouss (8–10). Meta-analysis showed that this is a recur-ut raremechanism, detected in only 1.7%of the cases.ping with previous studies, amplification of ALK al-
exclusively occurs in MYCN-amplified tumors. ApartProliferation curve of Ba/F3 cells stably expressing the F1174L orALK mutants. Ba/F3 cells expressing mutant ALK as well as
l Ba/F3 cells were grown in the absence of IL-3 for a period of
such rare ALK high-level amplification, high-stage
Clinical Cancer Research
2014. © 2010 American Association for Cancer
neuro2p thamutatwith 2presensegmegain dber offor otview ogainsmentsALK cneurothe reand thvival.numbwith p(2009ed wifoundmor ctionsPassosion mwith iclinicaIn c
that tin MYin thethe d
F1174tientsneedecouldgree othe Rservedand h
Discl
No p
Ackn
We t
Grant
EU uprogra(01G01of InterMiniste[BOF08Kankersearche0102. CZabrockand J.GhentFlander
Thepaymenadvertiindicate
Refe1. Ma
2002. Va
tionfor200
3. MiclaCh
4. JageOn
5. Brotom
6. Moma930
7. Gepro97
8. Chin n
9. Camurad
Skewed ALK Mutation Spectrum in MYCN-Amplified Neuroblastoma
www.a
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
blastoma tumors often exhibit gain of a large part oft mostly encompasses the ALK locus. In our cohort,ion analysis showed that only a minority of tumorsp gain carried ALK mutations, although 2p gains aret in as much as 19.3% of all neuroblastomas withntal imbalances. This observation indicates that 2poes not act as a mechanism for increased copy num-mutated ALK, in contrast to what has been describedher oncogenes in other tumor entities (21, 22). Inf the variability of centromeric break points for 2pbut almost consistent presence of ALK in these seg-, one could assume that low copy number gain ofould also imply a growth or survival advantage forblastoma cells. To test this hypothesis we analyzedlation between ALK copy number and expressione impact of increased ALK gene expression on sur-ALK gene expression was indeed shown to be copyer sensitive, and increased ALK expression correlatedoor survival. Using immunostaining, Passoni et al.) recently showed that ALK overexpression correlat-th patient survival although no correlation wasbetween mRNA and protein expression in their tu-ohort (23), which is in contrast with our observa-in cell lines. Our findings, together with those ofni et al. (2009), indicate that increased ALK expres-ight be functionally relevant. Therefore, patients
ncreased ALK expression might benefit from futurel trials with ALK inhibitors.onclusion, this meta-analysis shows for the first timehe recurrent F1174 mutation predominantly occursCN-amplified tumors, and clearly shows differences
frequency and distribution of ALK mutations acrossifferent genomic subtypes in neuroblastoma. TheRecepublish
ren H, Abel F, Kogner P, Martinsson T. High incidence of DNAtations and gene amplifications of the ALK gene in advanced spo-ic neuroblastoma tumours. Biochem J 2008;416:153–9.
10. JanacNa
11. Brotiotre
12. MeplaBio
13. OlsmeBio
14. Obmipa20
15. ShTerecCo
16. Liea nPDLe
17. Sp
acrjournals.org
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
mutation might suggest a poor prognosis in pa-with MYCN amplification, but further studies ared to substantiate this hypothesis. Furthermore weshow that the F1174L mutant displayed a higher de-f ALK phosphorylation and tumorigenicity than did1275Q mutant. No significant difference was ob-in the frequency of ALK mutations between low-
igh-stage tumors.
osure of Potential Conflicts of Interest
otential conflicts of interest were disclosed.
owledgments
hank Justine Nuytens for excellent technical support.
Support
nder FP6 (EET-pipeline, nr.037260 and the Kids Cancer Kinomem, nr.6037390), FWO (G.0198.08), IWT (SBO60848), GOA910). This article presents research results of the Belgian programuniversity Poles of Attraction, initiated by the Belgian State, Primer's Office, Science Policy Programming, Methusalem-program/01M01108], FOD (NKP_29_014), Stichting Kindergeneeskundigonderzoek, the KIKA Foundation. K. De Preter is postdoctoral re-r with the FWO. R. Noguera is supported by ICIII RD06/0020/. Kumps and M. Porcu are supported by the IWT, J. Cools and P.i are supported by Foundation against Cancer (SCIE2006-34, J.C.)Hoebeeck is a postdoctoral research supported by a grant of theUniversity (BOF01P07406) and by the fund for Scientific Researchs (KAN1.5.207.08).costs of publication of this article were defrayed in part by thet of page charges. This article must therefore be hereby markedsement in accordance with 18 U.S.C. Section 1734 solely tothis fact.
ived 10/02/2009; revised 05/19/2010; accepted 07/13/2010;ed OnlineFirst 08/24/2010.
rencesris JM, Hogarty MD, Bagatell R, Cohn SL. Neuroblastoma. Lancet7;369:2106–20.
ndesompele J, Baudis M, De Preter K, et al. Unequivocal delinea-of clinicogenetic subgroups and development of a new modelimproved outcome prediction in neuroblastoma. J Clin Oncol5;23:2280–99.
chels E, Vandesompele J, De Preter K, et al. ArrayCGH-basedssification of neuroblastoma into genomic subgroups. Genesromosomes Cancer 2007;46:1098–108.noueix-Lerosey I, Schleiermacher G, Michels E, et al. Overallnomic pattern is a predictor of outcome in neuroblastoma. J Clincol 2009;27:1026–33.deur GM. Molecular basis for heterogeneity in human neuroblas-as. Eur J Cancer 1995;31A:505–10.sse YP, Laudenslager M, Longo L, et al. Identification of ALK as ajor familial neuroblastoma predisposition gene. Nature 2008;455:–5.orge RE, Sanda T, Hanna M, et al. Activating mutations in ALKvide a therapeutic target in neuroblastoma. Nature 2008;455:5–8.en Y, Takita J, Choi YL, et al. Oncogenic mutations of ALK kinaseeuroblastoma. Nature 2008;455:971–4.
oueix-Lerosey I, Lequin D, Brugieres L, et al. Somatic and germlinetivating mutations of the ALK kinase receptor in neuroblastoma.ture 2008;455:967–70.deur GM, Pritchard J, Berthold F, et al. Revisions of the interna-nal criteria for neuroblastoma diagnosis, staging, and response toatment. J Clin Oncol 1993;11:1466–77.nten B, Pattyn F, De Preter K, et al. arrayCGHbase: an analysistform for comparative genomic hybridization microarrays. BMCinformatics 2005;6:124.hen AB, Venkatraman ES, Lucito R, Wigler M. Circular binary seg-ntation for the analysis of array-based DNA copy number data.statistics 2004;5:557–72.erthuer A, Berthold F, Warnat P, et al. Customized oligonucleotidecroarray gene expression-based classification of neuroblastomatients outperforms current clinical risk stratification. J Clin Oncol06;24:5070–8.imada H, Ambros IM, Dehner LP, Hata J, Joshi VV, Roald B.rminology and morphologic criteria of neuroblastic tumors:ommendations by the International Neuroblastoma Pathologymmittee. Cancer 1999;86:349–63.rman E, Michaux L, Beullens E, et al. FIP1L1-PDGFRalpha D842V,ovel panresistant mutant, emerging after treatment of FIP1L1-
GFRalpha T674I eosinophilic leukemia with single agent sorafenib.ukemia 2009;23:845–51.itz R, Oberthuer A, Zapatka M, et al. Oligonucleotide array-basedClin Cancer Res; 16(17) September 1, 2010 4361
2014. © 2010 American Association for Cancer
corevand
18. Cliactsubsis
19. FatmewitJ P
20. Mu
geca
21. Zaamter20
22. Hugabre
De Brouwer et al.
Clin C4362
Do
Published OnlineFirst August 18, 2010; DOI: 10.1158/1078-0432.CCR-09-2660
mparative genomic hybridization (aCGH) of 90 neuroblastomaseals aberration patterns closely associated with relapse patternoutcome. Genes Chromosomes Cancer 2006;45:1130–42.
fford SC, Lusher ME, Lindsey JC, et al. Wnt/Wingless pathwayivation and chromosome 6 loss characterize a distinct molecular-group of medulloblastomas associated with a favorable progno-. Cell Cycle 2006;5:2666–70.tet S, Haberler C, Legoix P, et al. β-Catenin status in paediatricdulloblastomas: correlation of immunohistochemical expression
h mutational status, genetic profiles, and clinical characteristics.athol 2009;218:86–94.rugan AK, Hong NT, Fukui Y, Munirajan AK, Tsuchida N. Onco-23. Palymtien
ancer Res; 16(17) September 1, 2010
Research. on November 24,clincancerres.aacrjournals.org wnloaded from
nic mutations of the PIK3CA gene in head and neck squamous cellrcinomas. Int J Oncol 2008;32:101–11.tkova A, Merk S, Wendehack M, et al. AML/MDS with 11q/MLLplification show characteristic gene expression signature and in-play of DNA copy number changes. Genes Chromosomes Cancer09;48:510–20.G, Chong RA, Yang Q, et al. MTDH activation by 8q22 genomicin promotes chemoresistance and metastasis of poor-prognosisast cancer. Cancer Cell 2009;15:9–20.ssoni L, Longo L, Collini P, et al. Mutation-independent anaplastic
phoma kinase overexpression in poor prognosis neuroblastoma pa-ts. Cancer Res 2009;69:7338–46.Clinical Cancer Research
2014. © 2010 American Association for Cancer