Spatial and temporal diversity in genomic instability processes defines lung cancer evolution

DOI: 10.1126/science.1253462

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for yourcolleagues, clients, or customers by clicking here.

Permission to republish or repurpose articles or portions of articles can be obtained byfollowing the guidelines here.

The following resources related to this article are available online at

www.sciencemag.org (this information is current as of October 9, 2014 ):

Updated information and services, including high-resolution figures, can be found in the onlineversion of this article at:

http://www.sciencemag.org/content/346/6206/251.full.html

Supporting Online Material can be found at:

http://www.sciencemag.org/content/suppl/2014/10/08/346.6206.251.DC1.html

A list of selected additional articles on the Science Web sites related to this article can befound at:

http://www.sciencemag.org/content/346/6206/251.full.html#related This article cites 62 articles, 21 of which can be accessed free:http://www.sciencemag.org/content/346/6206/251.full.html#ref-list-1 This article has been cited by 1 articles hosted by HighWire Press; see:http://www.sciencemag.org/content/346/6206/251.full.html#related-urls This article appears in the following subject collections:Medicine, Diseases

http://www.sciencemag.org/cgi/collection/medicine

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright2014 by the American Association for the Advancement of Science; all rights reserved. The title Science is aregistered trademark of AAAS.

Downloaded from www.sciencemag.org on October 10, 2014RESEARCH|REPORTS

thatameiosisI–specificfactorfrombuddingyeast,monopolin,generateskinetochoreswithmoremicrotubule-bindingelementsandgreaterstrength.Thesefindingsprovidedirectevidencethatsisterkinetochorefusionunderliesthecose-gregationofsisterchromatidsduringmeiosisI.

REFERENCESANDNOTES

LUNGCANCEREVOLUTION1.2.3.4.5.6.7.8.9.10.11.12.13.14.15.16.17.18.19.20.21.22.23.24.

K.D.Corbettetal.,Cell142,556–567(2010).

X.Li,R.K.Dawe,Nat.CellBiol.11,1103–1108(2009).L.S.Goldstein,Cell25,591–602(1981).

L.V.Paliulis,R.B.Nicklas,J.CellBiol.150,1223–1232(2000).

F.Monje-Casas,V.R.Prabhu,B.H.Lee,M.Boselli,A.Amon,Cell128,477–490(2007).

M.Winey,G.P.Morgan,P.D.Straight,T.H.GiddingsJr.,D.N.Mastronarde,Mol.Biol.Cell16,1178–1188(2005).B.H.Lee,A.Amon,Science300,482–486(2003).B.Akiyoshietal.,Nature468,576–579(2010).

MaterialsandmethodsareavailableassupplementarymaterialsonScienceOnline.

T.M.Carlile,A.Amon,Cell133,280–291(2008).

K.R.Benjamin,C.Zhang,K.M.Shokat,I.Herskowitz,GenesDev.17,1524–1539(2003).

A.P.Joglekar,D.C.Bouck,J.N.Molk,K.S.Bloom,E.D.Salmon,Nat.CellBiol.8,581–585(2006).

J.Lawrimore,K.S.Bloom,E.D.Salmon,J.CellBiol.195,573–582(2011).

A.Hochwagen,W.H.Tham,G.A.Brar,A.Amon,Cell122,861–873(2005).

J.H.Cocker,S.Piatti,C.Santocanale,K.Nasmyth,J.F.Diffley,Nature379,180–182(1996).

T.Goldfarb,M.Lichten,PLOSBiol.8,e1000520(2010).S.Keeney,C.N.Giroux,N.Kleckner,Cell88,375–384(1997).

M.A.Shonn,R.McCarroll,A.W.Murray,Science289,300–303(2000).

M.Petronczkietal.,Cell126,1049–1064(2006).K.P.Rabitschetal.,Dev.Cell4,535–548(2003).A.Tóthetal.,Cell103,1155–1168(2000).

K.D.Corbett,S.C.Harrison,CellReports1,583–589(2012).

S.Sarkaretal.,PLOSGenet.9,e1003610(2013).

K.T.Jones,S.I.Lane,Development140,3719–3730(2013).

SpatialandtemporaldiversityingenomicinstabilityprocessesdefineslungcancerevolutionElzaC.deBruin,1*NicholasMcGranahan,2,3*RichardMitter,2*MaxSalm,2*DavidC.Wedge,4*LucyYates,4,5†MariamJamal-Hanjani,1†SeemaShafi,1NirupaMurugaesu,1AndrewJ.Rowan,2EvaGrönroos,2MadihaA.Muhammad,1StuartHorswell,2MarcoGerlinger,2IgnacioVarela,6DavidJones,4JohnMarshall,4ThierryVoet,4,7PeterVanLoo,4,7DorisM.Rassl,8RobertC.Rintoul,8SamM.Janes,9Siow-MingLee,1,10MartinForster,1,10TanyaAhmad,10DavidLawrence,10MaryFalzon,10ArrigoCapitanio,10TimothyT.Harkins,11ClarenceC.Lee,11WarrenTom,11EnockTeefe,11Shann-ChingChen,11SharminBegum,2AdamRabinowitz,2BenjaminPhillimore,2BradleySpencer-Dene,2GordonStamp,2ZoltanSzallasi,12,13NikMatthews,2AengusStewart,2PeterCampbell,4CharlesSwanton1,2‡Spatialandtemporaldissectionofthegenomicchangesoccurringduringtheevolutionofhumannon–smallcelllungcancer(NSCLC)mayhelpelucidatethebasisforitsdismalprognosis.Wesequenced25spatiallydistinctregionsfromsevenoperableNSCLCsandfoundevidenceofbranchedevolution,withdrivermutationsarisingbeforeandaftersubclonaldiversification.Therewaspronouncedintratumorheterogeneityincopynumberalterations,translocations,andmutationsassociatedwithAPOBECcytidinedeaminaseactivity.Despitemaintainedcarcinogenexposure,tumorsfromsmokersshowedarelativedecreaseinsmoking-relatedmutationsovertime,accompaniedbyanincreaseinAPOBEC-associatedmutations.Intumorsfromformersmokers,genome-doublingoccurredwithinasmoking-signaturecontextbeforesubclonaldiversification,whichsuggestedthatalongperiodoftumorlatencyhadprecededclinicaldetection.Theregionallyseparateddrivermutations,coupledwiththerelentlessandheterogeneousnatureofthegenomeinstabilityprocesses,arelikelytoconfoundtreatmentsuccessinNSCLC.ACKNOWLEDGMENTS

WethankA.Hoskins,M.Miller,N.Umbreit,andE.Yuskoforhelpfulcomments.WethankA.DesaiforprovidingantibodiesandA.HoskinsforSNAPfandCLIPfplasmids.TheWellcomeTrustsupportedthisworkthroughaSirHenryWellcomeFellowshiptoE.D.(096078),SeniorResearchFellowshipstoA.L.M.(090903)andJ.R.(084229),andtwoWellcomeTrustCentreCoreGrants(077707and092076)andaninstrumentgrant(091020).TheworkwasalsosupportedbyNIHgrantstoC.L.A.(RO1GM079373andS10RR026406),S.B.(R01GM064386),andK.D.C.(R01GM104141)andbyaPackardFellowshiptoC.L.A.(2006-30521).K.D.C.alsoacknowledgessupportfromtheLudwigInstituteforCancerResearchandtheSidneyKimmelFoundation.Additionaldatadescribedinthisworkcanbefoundinthesupplementarymaterials.E.D.,K.K.S.,S.B.,A.L.M.,andC.L.A.conceivedtheexperiments.E.D.generatednewyeaststrainsandisolatedkinetochoreparticles.K.K.S.performedlasertrapexperiments.Y.D.performedthefluorescencemeasurements.F.d.L.A.andJ.R.performedproteomicsanalysis.Q.Y.andK.D.C.purifiedmonopolin.K.N.O.andS.C.optimizedfluorescentlabelingofkinetochoreparticles.E.D.,K.K.S.,A.L.M.,andC.L.A.preparedthemanuscript.

SUPPLEMENTARYMATERIALS

ungcanceristheleadingcauseofcancer-relatedmortality(1,2).Understandingthepathogenesisandevolutionoflungcancermayleadtogreaterinsightintotumorini-tiationandmaintenanceandmayguidetherapeuticinterventions.Previousworkchar-acterizingthegenomeofnon–smallcelllungcancer(NSCLC)hasdemonstratedthatNSCLCgenomesexhibithundredsofnonsilentmuta-tionstogetherwithcopynumberaberrationsandgenomedoublings(3–9).Althoughsubclonalpop-CancerResearchUKLungCancerCentreofExcellence,UniversityCollegeLondonCancerInstitute,LondonWC1E6BT,UK.2CancerResearchUKLondonResearchInstitute,LondonWC2A3LY,UK.3CentreforMathematicsandPhysicsintheLifeScienceandExperimentalBiology

(CoMPLEX),UniversityCollegeLondon,LondonWC1E6BT,UK.4WellcomeTrustSangerInstitute,Hinxton,CB101SA,UK.5UniversityofCambridge,CambridgeCB21TN,UK.6InstitutodeBiomedicinayBiotecnologíadeCantabria(CSIC-UC-Sodercan),DepartamentodeBiologíaMolecular,UniversidaddeCantabria,Santander,Spain.7DepartmentofHumanGenetics,UniversityofLeuven,3000Leuven,Belgium.8PapworthHospitalNHSFoundationTrust,CambridgeCB233RE,UK.9LungsforLivingResearchCentre,University

CollegeLondon,LondonWC1E6BT,UK.10UniversityCollegeLondonHospitals,LondonNW12BU,UK.11ThermoFisherScientific,Carlsbad,CA92008,USA.12TechnicalUniversityofDenmark,2800KongensLyngby,Denmark.13Children’s

HospitalInformaticsProgram,HarvardMedicalSchool,Boston,MA02115,USA.

*Theseauthorscontributedequallytothiswork.†Theseauthorscontributedequallytothiswork.‡Correspondingauthor.E-mail:charles.swanton@cancer.org.uk

1Lwww.sciencemag.org/content/346/6206/248/suppl/DC1MaterialsandMethodsFigs.S1toS8TablesS1toS3References(25–42)MovieS1

AdditionalDataTablesS1toS3

30May2014;accepted1September2014Publishedonline11September2014;10.1126/science.1256729

ulationshavebeenidentifiedwithinsinglebi-opsies(9),theextentofgenomicdiversitywithinprimaryNSCLCsremainsunclear.Moreover,al-thoughbothexogenousmutationalprocesses,suchassmoking(10–12),andendogenouspro-cesses,suchasup-regulationofAPOBECcyti-dinedeaminases(13–15),havebeenfoundtocontributetothelargemutationalburdeninNSCLC,thetemporaldynamicsoftheseprocessesandtheircontributiontodriversomaticaberra-tionsovertimeremainunknown.

Toinvestigatelungcancerevolution,weper-formedmultiregionwhole-exomeand/orwhole-genomesequencing(M-seqWES/WGS)onatotalof25tumorregions,collectedfromsevenNSCLCpatientswhounderwentsurgicalresectionbeforereceivingadjuvanttherapy.ThemajorNSCLChistologicalsubtypes,includingadenocarcinoma(LUAD)andsquamouscellcarcinoma(LUSC),wererepresented(tableS1).SequencingoftumorandnormalDNAtomeancoveragedepthsof107×and54×forM-seqWESandM-seqWGS,respectively(tableS2),identified1884nonsilentand76,129silentmutations(16).

Toevaluatetheintratumorheterogeneityofnonsilentmutations,weclassifiedeachmutationasubiquitous(presentinalltumorregions)orheterogeneous(presentinatleastone,butnotall,regions).SpatialintratumorheterogeneitywasidentifiedinallsevenNSCLCs,withame-dianof30%heterogeneousmutations(range4

10OCTOBER2014•VOL346ISSUE6206

SCIENCEsciencemag.org251

RESEARCH|REPORTS

to63%)(Fig.1Aandfig.S1).Intheadenosqua-moustumorfrompatientL002,heterogeneousmutationsseparatedconcordantwithLUAD(regionsR1andR2)orLUSC(regionsR3andR4)histopathologies(fig.S2).PatientsL003andL008eachpresentedwithtwotumorsinseparatelobesofthelung.M-seqWESrevealed74%ubiquitousmutationsinthetumorsfromL008,whichindicatedaclonalorigin.However,inL003,onlyasinglemutation(EGFRL858R,theepidermalgrowthfactorreceptorinwhichLeu858isreplacedwithArg)wasdetectedinbothtu-mors(Fig.1A).GiventhatEGFRL858Risahighlyrecurrentmutation(17)andalsothatnosilentmutationswereshared,weconcludedthatthetumorsinL003wereofindependentclonalorigin,withtheevolutionofidenticaloncogeniceventsinparallel.ToresolvetheextentofgenomicdiversityinNSCLCandtoinfertheancestralrelationsbe-tweentumorregions,weestimatedthefractionoftumorcellswithineachregionharboringeachmutation(16,18).Almostallubiquitousmutations(>99%)wereclassifiedasfullyclonalwithineachregion.Moreover,inmostregions,themajorityofheterogeneousmutationswasclonaland,thus,presentinallcellswithintheregion(Fig.1Bandfig.S3).However,certainregionsdisplayedcon-siderablesubclonaldiversity.Forexample,>75%ofheterogeneousmutationspresentinL004R5weresubclonal(Fig.1B),andthisregionconsistedoftwodistinctsubclonalpopulations.Thesubclonalstructureofeachtumorregionwasthenusedtoconstructphylogenetictrees,byusingbothmaximumparsimonyandun-weightedpair-groupmethods.Wealsotookintoaccountregionalcopynumberlossesthatre-sultedinsharedtruncalmutationsbecomingheterogeneouslaterintumorevolution(16),suchasasegmentofchromosome6inLS01(fig.S4)andthePAX7mutationinthelymphnodeofL001(Fig.1A).Notably,allsevenNSCLCsshowedevidenceofbranchedtumorevolution(fig.S5).WenextevaluatedtheregionalheterogeneityofpotentialNSCLCdrivermutations,classifiedintothreecategoriesonthebasisofcurrentevi-dencesupportingdrivermutationstatus(16).Everytumorshowedevidenceforubiquitous,aswellasheterogeneous,drivermutations,manyofwhichwereclonallydominantinasubsetoftumorregionsandentirelyabsentinothers(Fig.1B,fig.S3,andtableS3).Notethattheprob-abilityofmissingacategory1“high-confidence”drivergenebyanalyzingasingleregionforeach

Fig.1.IntratumorheterogeneityofsomaticmutationsinhumanNSCLC.(A)Heatmapsshowtheregionaldistributionofallnonsilentmutations;presence(blue)orabsence(gray)ofeachmutationisindicatedforeverytumorregion.Cartoonsdepictthelocationofeachtumor.Columnnexttoheatmapshowstheintratumorheterogeneity;mutationpresentinallregions(blue),inmorethanonebutnotall(yellow),orinoneregion(red).Mutationsareorderedontumordrivercategorywithcategories1to3indicatedintherightcolumninblack,darkgray,andlightgray,respectively(detailsintableS3).Totalnumberofnonsilentmutationsisprovidedbeloweachtumorwithpercentageofheterogeneousmutationsinbrackets.InL001,themutationmarkedbyanasterisk(*)isadditionaltothegermlineMEN1mutation.LN,lymphnode;R,region.(B)Two-dimensionalDirichletplotsshowthecancercellfraction(CCF)ofthemutationsinallregionsoftumorsL004;increasingintensityofredindicatesthelocationofahighposteriorprobabilityofacluster.InregionR5,themajorityofheterogeneousmutationsaresubclonal,andaclusterofmutationswithaCCFbelow1canbeobserved.

sciencemag.orgSCIENCE

252

10OCTOBER2014•VOL346ISSUE6206

RESEARCH|REPORTS

tumorwasonaverage42%(range0to67%)and83%(range67to100%)forallpotentialdrivergenes(categories1to3),whichhighlightsthepotentiallimitationsofassessingsingletumorregions.Nevertheless,category1and2drivermutationsweresignificantlymoreoftentruncalcomparedwithmutationsinnondrivergenesinourM-seqanalysis(P=0.04).Consistentwiththesedata,inTheCancerGenomeAtlas(TCGA)cohort,previouslyreporteddrivergenes(5,19,20)weresignificantlyenrichedforclonalmutations(P<0.001)(fig.S6).Thesedataindicatethat,inNSCLC,mostknowndrivermutationsoccurearlyintu-morevolution.

Todeterminetheintratumorheterogeneityofcopynumberaberrations,weestimatedin-tegerDNAcopynumbersforeachtumorregion(7,16,21,22).Alargefractionofthegenomehadundergonealterationsinalltumors,andge-nomicprofilesweremoresimilarwithintumorsthanbetweendifferenttumors(fig.S7).Toeval-uatethespatialheterogeneityofpotentialtumordrivercopynumberaberrations,weexploredtheregionaldistributionofchromosomalsegmentsidentifiedasrecurrentlygainedorlostinTCGALUADorLUSCtumors.Mostsegmentswereidentifiedasaberrantinatleastonetumorre-gion,andmanyrecurrentgainsandlosseswerefoundtobeheterogeneousinatleastonetumor(Fig.2A).Forexample,inL001,afocalEGFRam-plification(chr7p11.2),aswellasdeletionsofchro-mosomalsegmentsharboringCDKN2A(chr9p21.3)andPTEN(chr10q23.31),wasobservedinallre-gions,whereas,inL008,weobservedheteroge-neouscopynumberlossesinvolvingCDKN2AandPTEN.Insupportofcopynumberaberra-tionsoccurringlaterintumordevelopment,wealsoidentifiedsubclonalcopynumberaberra-tionswithintumorregions.Forinstance,morethan15%ofthegenomeinregionR1ofL008wassubjecttosubclonalcopynumberalterations(fig.S8).Consistentwithevidenceofsubclonalcopynumberaberrations,centromericfluorescenceinsituhybridizationanalysesconfirmednumericalchromosomaldiversitywithinindividualtumorregions(fig.S9),whichsuggestedthatchromo-somalinstabilitymayprovideasubstrateforsubclonalcompetition.

Thehigh-coverageM-seqWGS(mean96×)forL002andL008enabledustoinvestigatetheregionalseparationoflarge-scalegenomiceventsinthesesamples.FortheadenosquamoustumorL002,weidentified30structuralvariants,mostofwhichwerefoundeitherintheLUADregionR1ortheLUSCregionR3,butnotboth(tableS4),whichsuggestedthattheyoccurredaftersub-clonaldiversification(Fig.2B).Bycontrast,forL008,48ofthe52identifiedstructuralvariantsweresharedbetweenthetwotumorregionsfromdifferentlobesofthelung,whichsuggestedthatthemajorityofthesevariantsoccurredbe-foretumormetastasistotheotherlobe(Fig.2B).Notably,inL008,“chains”oftranslocationswithhighlyclusteredbreakpointswerefoundbetweenchromosomes14and17,aswellaschromo-somes17and19(fig.S10andtableS4),whichdisruptedtheFANCMandNF1tumorsuppres-sorgenes.Breakpointhomologyprofilingsug-gestsinvolvementofeithernonhomologousoralternativeend-joining(23,24),indicativeofdouble-strandbreakevents.Thislesionpatternisconsistentwithchromoanagenesis(25)andindicatesapunctuatedevolutionpatternwheremultipleoncogeniceventsmayhaveoccurredsimultaneously(26).

Fourtumorsdisplayedevidenceforwhole-genome–doublingevents(16).Inthreetumors(L001,L004,andL008),thegenome-doublingeventwassharedacrosseverytumorregion;itoccurredbeforediversification,withthemajorityoftruncalmutations(84to88%)presentatploidy≥2,indicativeofalargemutationalburdenbe-foregenomedoubling.Inonetumor,L002,themajorityofheterogeneousmutationswerealsopresentatploidy≥2,indicativeoftwoindepen-dentgenome-doublingevents:oneintheLUADregionandoneintheLUSCregion(fig.S11).Notably,everytruncaldrivermutationlikelyoccurredbeforegenomedoubling.

Tofurtherexplorethedynamicsofthemuta-tionalprocessesshapinglungcancergenomesovertime,thespectraofpointmutationsineachtumorweretemporallydissected.Early(truncal)mutationslikelyreflectprocessesinvolvedbe-foreandduringtumorinitiationandearlyde-velopment,whereaslate(branched)mutationsFig.2.IntratumorheterogeneityofchromosomalalterationsinhumanNSCLC.(A)Distributionofpotentialtumordrivercopynumberalterationsisindicatedforeachtumorregion.Theupperheatmapsshowtheregionaldis-tributionofrecurrentlyamplified(left)ordeleted(right)chromosomalseg-mentsbasedonTCGALUADdata,andthelowerheatmapsshowtheregionaldistributionofrecurrentlyamplifiedordeletedchromosomalsegmentsbasedonTCGALUSCdata.Foreachregion,gain(red)orloss(blue)wasdeterminedSCIENCEsciencemag.org

relativetothemeanploidy.(B)Circosplotsdepictinginter-andintrachromo-somaltranslocations,aswellasdeletionsandinsertionsforregionsR1andR3forL002(upper)andL008(lower);sharedeventsareindicatedinblue,eventsprivatetoregionR1areindicatedinred,andprivatetoregionR3ingreen.TheoutercirclerepresentstheintegercopynumberdataforR1andtheinnercircleforR3foreachtumorsample;copynumbersegmentswithanintegervaluegreaterthanmeanploidyareinredandthoselessthanmeanploidyinblue.10OCTOBER2014•VOL346ISSUE6206

253

RESEARCH|REPORTS

revealmutationalprocessesshapingthegenomeduringtumormaintenanceandprogression,in-cludingthosecontributingtointratumorhetero-geneity.ForL002,weanalyzedregionsR1andR3separatelytoallowcomparisonsofLUADandLUSChistologieswithinthesametumor.

Inalltumors,weobservedstatisticallysignifi-cantshiftsinthemutationspectraovertime(P<0.05allcases)(Fig.3A).Furthermore,everytu-morexhibitedastatisticallysignificantdecreaseintheproportionofC>Atransversionsinlatecomparedwithearlymutations(P<0.05)(Fig.3A),althoughthiswasmorepronouncedintheLUADcases[meanoddsratio:LUAD3.13(range2.07to5.55)andLUSC1.34(range1.21to1.46)].BecauseC>Atransversionsareassociatedwiththemutageniceffectsoftobaccosmoke(12),adecreaseintheproportionofC>AtransversionsindicatesarelativedecreaseinthemutationalburdenattributabletosmokingduringLUADdevelopment,inbothformersmokersandcur-rentsmokers.

TovalidatetheseobservationsinalargerNSCLCcohort,mutationsinTCGALUADandLUSCsam-plesweretemporallydissected(16).ConsistentwithourM-seqanalyses,bothTCGALUADandLUSCsmokersandformersmokersexhibitedade-creaseintheproportionofC>Atransversionsinlatemutations(LUADcurrentsmokers,P<0.0001;formersmokers,P<0.0001;never-smokers,P=0.147;LUSCcurrentsmokers,P=0.003;formersmokers,P<0.0001;andnever-smokers,P=0.673)(Fig.3B).Similarly,theleast-pronounceddecreasewasobservedinLUSCcurrentsmokers;25%ofLUSCdisplayednodecreaseinC>Atrans-versions,comparedwithlessthan10%inLUAD.ThemutationalfootprintofsmokingexhibitsastrandbiaswithC>Atransversionsaccumulatingpreferentiallyonthetranscribedstrand(10,12).BothLUADandLUSCformersmokersrevealedastatisticallysignificantdecreaseinstrandbiasinlate,comparedwithearly,C>AtransversionsFig.3.TemporalandspatialdissectionofmutationspectrainLUADandLUSCsamples.(A)Fractionofearlymutations(trunk)andlatemutations(branch)accountedforbyeachofthesixmutationtypesinallM-seqsam-ples.(B)BeeswarmplotsshowingthefractionofearlymutationsandlatemutationsaccountedforbyeachofthesixmutationtypesineveryTCGAformersmokerorcurrentsmokerwithbothearlyandlatemutations.Sig-nificanceisindicated.(C)APOBECmutationenrichmentoddsratioforearly(trunk,bluebars)andlate(branch,redbars)mutationsforM-seqsamples.TheAPOBECsignatureencompassesC>TandC>GmutationsinaTpCcontext(16).The95%confidenceintervalsforFisher’sexacttestareindicated.(DandE)Threemutationtypes(C>A;C>GandC>T)atall16possibletrinucleotidecontextsforL002(D)andL008(E).Forbothsam-ples,trunkmutationsaswellasbranchmutationsfromtworegionsaredepicted.sciencemag.orgSCIENCE

254

10OCTOBER2014•VOL346ISSUE6206

RESEARCH|REPORTS

(LUAD,P=0.00354;LUSC,P=0.046),consist-entwithanancestralfootprintofsmokingonthesegenomes.Conversely,nostatisticallysig-nificantdifferencewasobservedbetweenearlyandlatemutationsincurrentsmokers(LUAD,P=0.23;LUSC,P=0.22).

InthemajorityofM-seqtumors,thedecreasedproportionofC>Amutationswasaccompa-niedbyanincreaseinC>TandC>GmutationsatTpCsites,indicativeofAPOBECcytidinede-aminaseactivity(13–15).MutationsconsistentwithAPOBEC-mediatedmutagenesisweremorepronouncedonthebranchesthanthetrunkinfouroutoffiveLUADM-seqsamples(Fig.3C).Onaverage31%(8to41%)ofnonsilentbranchmutationsoccurredinanAPOBEC-mutationcontextcomparedwith11%(7to16%)oftrun-calnonsilentmutations.BrancheddrivergenesPIK3CA,EP300,TGFBR1,PTPRD,andAKAP9harboredmutationsinanAPOBECcontext,whichindicatedapossiblefunctionalimpactofAPOBECactivityonsubclonalexpansion.Likewise,TCGALUADtumorswithdetectableAPOBECmuta-tionalsignaturesshowedsignificantenrichmentinlate,comparedwithearly,APOBECmutations(P<0.001)(fig.S12),and20%ofsubclonaldrivermutationswerefoundtooccurinanAPOBECcontext,comparedwith11%ofclonaldrivermu-tations.However,forTCGALUSCtumorswithdetectableAPOBECmutationalsignatures,tem-poraldissectionofAPOBECmutationsdidnotrevealsuchacleartrend(fig.S12),whichindicatedpotentialdifferencesinthetemporaldynamicsofAPOBEC-mediatedmutagenesisbetweenhis-tologicalsubtypes.Inadditiontotemporalhet-erogeneity,spatialheterogeneityinboththeproportionofAPOBEC-associatedmutations(Fig.3,DandE)andAPOBECmRNAexpressionwasobservedintheM-seqtumors(fig.S13).TogainadeeperunderstandingofNSCLCevolution,wefocusedonthetwotumorswithhigh-coverageM-seqWGSandtemporallyplacedthegenomicinstabilityprocessesrelativetotheemergenceofthemost-recentcommonancestor(Fig.4).InpatientL002,acurrentsmoker,to-baccocarcinogensplayedasignificantroleearlyintumordevelopment,withC>Atransversionsrepresenting39%oftruncalmutations(Fig.4A).Earlymutationsincludedmultipledrivergenes,suchasTP53andCHD8.UpondiversificationintoaLUADsubcloneandaLUSCsubclone,copynumberalterations(fig.S7)anddrivermutationswereacquiredindependentlyinbothsubclones,suchasastop-gainmutationinthetumorsuppres-sorgeneFAT1ontheLUSCbranchandmuta-tionsaffectingTGFBR1,ZFHX4,ARHGAP35,andPTPRDintheLUADregion.APOBEC-associatedmutationswereelevatedspecificallyintheLUADregion,whichincludedthedrivermutationsinTGFBR1andPTPRD,andthehighestAPOBEC3BmRNAexpressionwasdetectedinthisregion(fig.S13).

ThetumorsfrompatientL008alsodisplayedtruncalC>Atransversionsandspatialheteroge-neityinAPOBECenrichment,withamorepro-nouncedAPOBECsignatureinthetumorofthemiddlelobecomparedwiththeupperlobe(Fig.4B).InL008,wegainedfurthertemporalresolutionbyexploringthemutationsbeforeandafterthetruncalgenome-doublingevent.Alltruncaldrivermutationswerefoundtooc-curbeforegenomedoubling.However,atobaccosmokesignatureofC>Atransversionswasob-servedinmorethan30%oftruncalmutationsbothbeforeandafterdoubling,andonlyin21%and9%ofheterogeneousmutationsinthetworegionsR1andR3fromseparatelobesofthelung.BecauseL008ceasedsmokingmorethan20yearsbeforesurgery(tableS1),thesedatasuggestthatthegenome-doublingeventandtruncaldrivermutationsoccurredwithinasmokingcarcinogeniccontextmorethan20yearsago.Similarly,thegenome-doublingeventandtruncaldrivermutationsinformersmokerL001alsoappearedtooccurbeforesmokingcessationmorethan20yearsbeforesurgery(fig.S14).Thesedatasuggestaprolongedtumorlatencyperiodaftergenomedoublingandbeforeclin-icaldetectioninNSCLC.

Throughsequencingmultiplesurgicallyre-sectedtumorregions,wewereabletounravelboththeextentofgenomicheterogeneityandtheevolutionaryhistoryofsevenNSCLCs.Incon-trasttothesituationinclearcellrenalcellcar-cinoma(ccRCC)(27,28),knowndrivermutationstypicallyoccurredearlyinNSCLCdevelopment,andthemajorityofhigh-confidencedrivereventswerefullyclonal.Conceivably,thisexplainstheprogression-freesurvivalbenefitsassociatedwithNSCLConcogenicdrivertargeting(29).However,likeccRCC(27,28),allNSCLCsexhibitedheter-ogeneousdrivermutationsand/orrecurrentcopynumberaberrationsandmanyheteroge-neousmutationsgavethe“illusionofclonal-ity,”astheyarepresentinallcellsfromcertainregionsbutundetectablewithinotherregions.Notably,althoughourmultiregionalsamplingapproachallowedustoevaluatespatialhetero-geneity,onlyasmallpartoftheentiretumorwassampled(onaverage<5%),whichindicatesthatwemightbeunderestimatingthefullextentofheterogeneityinthesetumors.

Conceivably,intratumorheterogeneitymaycompromisetheabilityofasinglebiopsytode-finealldrivereventscomprehensivelyforopti-maltumorcontrol.Forinstance,L008presentedwithanactivatingBRAF(G469A)mutation(30)inallregionsandanactivatingPIK3CA(E542K)mutation(31)onlyinregionR3.Thus,abiopsytakenfromR3mightsuggesttreatmentwithaninhibitorofthephosphatidylinositol3-kinase–mammaliantargetofrapamycin(PI3K/mTOR)sig-nalingaxisandcombinationtherapy.Conversely,Fig.4.AmodeloftheevolutionaryhistoryofNSCLC.EvolutionaryhistoriesoftumorsfrompatientsL002(A)andL008(B)aredepicted.GenomicinstabilityprocessesdefiningNSCLCevolutionhavebeenplacedontheirphylogenetictrees.DrivermutationsoccurringinanAPOBECcontextarehighlightedwithabluebox,andthoseoccurringinasmokingcontextwithagraybox.Ineachcase,thetimingofgenome-doublingeventsisindicatedwithanarrow.CIN,chromosomalinstability;muts,mutations.SCIENCEsciencemag.org

10OCTOBER2014•VOL346ISSUE6206

255

RESEARCH|REPORTS

asinglebiopsyfromanyotherregionwouldsug-gesttreatmentwithaBRAFinhibitor,forwhichthetumorcellsfromR3mightberesistantbe-causeofthePIK3CAmutation(32).

OurstudyalsoshedslightonthedivergentgenomicinstabilityprocessesinvolvedinNSCLCevolutionandtheirdynamicsovertime.Evidenceforspatialdiversityingenomicinstabilitypro-cessessuggeststhatopportunitiestoexploitsuchmechanismstherapeuticallymaybelimitedinthisdisease(33).Inthreetumors,wedetectedgenome-doublingeventsoccurringbeforesub-clonaldiversificationbutafteracquisitionofdrivermutations,consistentwithfindingsincolorectalcancerthatgenomedoublingmayacceleratecancergenomeevolution(34).Therelationofchromosomalinstabilitywithdrugresistanceandearlytumorrecurrence(35,36)suggeststhattargetingtruncaldrivereventsmaybecompromisedbytheinitiationofchromo-somalinstabilitylaterintumorevolution.Theseresults,coupledwiththeobservationthatNSCLCtumorsmayhaveprolongedlatencyperiods,sup-portcontinuedeffortstooptimizemethodsforearlierdetection.

Unexpectedly,wefoundthatdespitecontinu-ousexposuretothemutagensintobaccosmoke,tumorsfromsmokersshowedevidencethatanadditionalgenomicinstabilityprocess(APOBEC-associatedmutagenesis)likelycontributestotu-morprogression.AlargeproportionofsubclonaldrivermutationswerefoundtooccurinanAPOBECcontext,whichsuggeststhatthedif-ferencesinmutationspectraovertimeandspacemayreflecttheactivityoftheprocessgeneratingthemutations,aswellastheselectiveadvantageoftheacquiredmutations.

Thepresenceofsubclonal,regionallysepa-rateddriverevents,coupledwiththerelentlessanddynamicnatureofgenomicinstabilityprocessesobservedinthisstudy,highlightthetherapeuticchallengesassociatedwithNSCLC.Engaginganadaptableimmunesystemmaypresentatracta-bleapproachtomanagethedynamiccomplexityinNSCLC(37).Longitudinalstudieswillbere-quiredtodecipherdriversofsubclonalexpansion,identifytheoriginsofsubclonescontributingtometastaticrecurrence,andresolvetheevolution-aryprinciplesthatunderpinthedismaloutcomeassociatedwiththisdisease.

REFERENCESANDNOTES

17.S.V.Sharma,D.W.Bell,J.Settleman,D.A.Haber,Nat.Rev.

Cancer7,169–181(2007).

18.N.Bollietal.,Nat.Commun.5,2997(2014).19.M.Imielinskietal.,Cell150,1107–1120(2012).20.M.S.Lawrenceetal.,Nature505,495–501(2014).

21.P.VanLooetal.,Proc.Natl.Acad.Sci.U.S.A.107,16910–16915

(2010).

22.S.Nik-Zainaletal.,Cell149,994–1007(2012).

23.A.Malhotraetal.,GenomeRes.23,762–776(2013).

24.S.F.Bunting,A.Nussenzweig,Nat.Rev.Cancer13,443–454

(2013).

25.C.Z.Zhang,M.L.Leibowitz,D.Pellman,GenesDev.27,

2513–2530(2013).

26.S.C.Bacaetal.,Cell153,666–677(2013).

27.M.Gerlingeretal.,Nat.Genet.46,225–233(2014).

28.M.Gerlingeretal.,N.Engl.J.Med.366,883–892(2012).29.M.Reck,D.F.Heigener,T.Mok,J.C.Soria,K.F.Rabe,Lancet

382,709–719(2013).

30.H.Daviesetal.,Nature417,949–954(2002).

31.S.Kang,A.G.Bader,P.K.Vogt,Proc.Natl.Acad.Sci.U.S.A.

102,802–807(2005).

32.H.Shietal.,CancerDiscov.4,80–93(2014).

33.R.A.Burrell,N.McGranahan,J.Bartek,C.Swanton,Nature

501,338–345(2013).

34.S.M.Dewhurstetal.,CancerDiscov.4,175–185(2014).

35.R.Sotillo,J.M.Schvartzman,N.D.Socci,R.Benezra,Nature

464,436–440(2010).

36.A.J.Leeetal.,CancerRes.71,1858–1870(2011).

37.J.C.Soriaetal.,Eur.J.Cancer49(suppl.2),3408(2013).

ACKNOWLEDGMENTS

bytheCambridgeBiomedicalResearchCentreandCancerResearchUKCancerCentre;P.V.L.isapostdoctoralresearcheroftheResearchFoundation—Flanders(FWO);S.M.J.isaWellcomeSeniorFellowinClinicalScience;andC.S.isaseniorCancerResearchUKclinicalresearchfellowandisfundedbyCancerResearchUK,theRosetreesTrust,EuropeanUnionFrameworkProgramme7(projectsPREDICTandRESPONSIFY,ID:259303),theProstateCancerFoundation,theEuropeanResearchCouncilandtheBreastCancerResearchFoundation.ThisresearchissupportedbytheNationalInstituteforHealthResearchUniversityCollegeLondonHospitalsBiomedicalResearchCentre.WewouldliketothankServicioSantanderSupercomputaciónfortheirsupport.P.C.isapaidconsultantforandholdsequityin14MGenomicsLtd.TheresultspublishedhereareinpartbasedupondatageneratedbytheCancerGenomeAtlaspilotprojectestablishedbytheNationalCancerInstituteandNationalHumanGenomeResearchInstitute,NIH.InformationaboutTCGAandtheinvestigatorsandinstitutionswhoconstitutetheTCGAresearchnetworkcanbefoundathttp://cancergenome.nih.gov/.ThedatawereretrievedthroughdbGaPauthorization(accessionno.phs000178.v5.p5).SequencedatahavebeendepositedattheEuropeanGenome-PhenomeArchive(EGA,www.ebi.ac.uk/ega/),whichishostedbytheEuropeanBionformaticsInstitute(EBI),underaccessionnumbersEGAS00001000809,EGAS00001000888andEGAS00001000889.

SUPPLEMENTARYMATERIALS

E.B.isaRosetreesTrustfellow;M.J.H.hasaCancerResearchUKfellowship;N.Mu.receivedfundingfromtheRosetreesTrust;M.G.isfundedbytheUKMedicalResearchCouncil;I.V.isfundedbySpanishMinisteriodeEconomíayCompetitividad

subprogramaRamónyCajal;R.C.R.andD.M.R.arepartlyfundedwww.sciencemag.org/content/346/6206/251/suppl/DC1MaterialsandMethodsFigs.S1toS14TablesS1toS4References(38–64)

17March2014;accepted4July201410.1126/science.1253462LUNGCANCEREVOLUTIONIntratumorheterogeneityinlocalizedlungadenocarcinomasdelineatedbymultiregionsequencingJianjunZhang,1,2JunyaFujimoto,3JianhuaZhang,4DavidC.Wedge,5XingzhiSong,4JiexinZhang,6SahilSeth,4Chi-WanChow,3YuCao,1CurtisGumbs,1KathrynA.Gold,2NedaKalhor,7LatashaLittle,1HarshadMahadeshwar,4CesarMoran,7AlexeiProtopopov,4HuandongSun,4JiabinTang,4*XifengWu,8YuanqingYe,8WilliamN.William,2J.JackLee,9JohnV.Heymach,2,10WaunKiHong,2StephenSwisher,11IgnacioI.Wistuba,3P.AndrewFutreal1,12†Cancersarecomposedofpopulationsofcellswithdistinctmolecularandphenotypicfeatures,aphenomenontermedintratumorheterogeneity(ITH).ITHinlungcancershasnotbeenwellstudied.Weappliedmultiregionwhole-exomesequencing(WES)on11localizedlungadenocarcinomas.AlltumorsshowedclearevidenceofITH.Onaverage,76%ofallmutationsand20outof21knowncancergenemutationswereidentifiedinallregionsofindividualtumors,whichsuggestedthatsingle-regionsequencingmaybeadequatetoidentifythemajorityofknowncancergenemutationsinlocalizedlungadenocarcinomas.Withamedianfollow-upof21monthsaftersurgery,threepatientshaverelapsed,andallthreepatientshadsignificantlylargerfractionsofsubclonalmutationsintheirprimarytumorsthanpatientswithoutrelapse.Thesedataindicatethatalargersubclonalmutationfractionmaybeassociatedwithincreasedlikelihoodofpostsurgicalrelapseinpatientswithlocalizedlungadenocarcinomas.1.WorldHealthOrganization,www.who.int/cancer/en/(2013).2.R.Siegel,D.Naishadham,A.Jemal,CACancerJ.Clin.63,

11–30(2013).

3.H.Tanakaetal.,CancerRes.67,6007–6011(2007).4.B.A.Weiretal.,Nature450,893–898(2007).5.L.Dingetal.,Nature455,1069–1075(2008).6.Z.Kanetal.,Nature466,869–873(2010).7.S.L.Carteretal.,Nat.Biotechnol.30,413–421

(2012).

8.T.I.Zacketal.,Nat.Genet.45,1134–1140(2013).9.R.Govindanetal.,Cell150,1121–1134(2012).

10.E.D.Pleasanceetal.,Nature463,184–190(2010).11.W.Leeetal.,Nature465,473–477(2010).

12.G.P.Pfeifer,P.Hainaut,Mutat.Res.526,39–43(2003).13.S.A.Robertsetal.,Nat.Genet.45,970–976(2013).

14.M.B.Burns,N.A.Temiz,R.S.Harris,Nat.Genet.45,977–983

(2013).

15.L.B.Alexandrovetal.,Nature500,415–421(2013).

16.DetailedinformationonmethodsisavailableonScienceonline.

256

10OCTOBER2014•VOL346ISSUE6206

Intratumorheterogeneity(ITH)mayhaveim-pactsontumorbiopsystrategy,characteriza-tionofactionabletargets,treatmentplanning,anddrugresistance(1–6).ITHhasrecentlybeenelucidatedinsubstantialdetailinseve-ralcancertypeswiththeuseofnext-generationsequencing(NGS)approaches(7–14).Recentevidencesupportsamodelofbranchedevolu-tionleadingtovariableITHindifferenttumors(9,13,15,16).Studiesinclearcellrenalcarcinoma

sciencemag.orgSCIENCE