(3.3.10)--脑科学与影像新技术.pdf

Intratumor heterogeneity in human glioblastomareflects cancer evolutionary dynamicsAndrea Sottorivaa,b,c,1,Inmaculada Spiterib,1,Sara G.M.Piccirillod,Anestis Touloumisb,e,V.Peter Collinsf,John C.Marionie,Christina Curtisc,Colin Wattsd,g,2,and Simon Tavara,b,h,2aDepartment of Oncology,University of Cambridge,Cambridge CB2 2XZ,United Kingdom;bCancer Research UK,Cambridge Research Institute,Li Ka ShingCentre,Cambridge CB2 0RE,United Kingdom;cDepartment of Preventive Medicine,Keck School of Medicine,University of Southern California,Los Angeles,CA 90033;dDepartment of Clinical Neurosciences,Cambridge Centre for Brain Repair,University of Cambridge,Cambridge CB2 0PY,UnitedKingdom;eEuropean Molecular Biology Laboratory-European Bioinformatics Institute,Wellcome Trust Genome Campus,Cambridge CB10 1SD,UnitedKingdom;fDivision of Molecular Histopathology,Department of Pathology,University of Cambridge,Addenbrookes Hospital,Cambridge CB2 0QQ,United Kingdom;gDivision of Neurosurgery,Department of Clinical Neurosciences,University of Cambridge,Addenbrookes Hospital,Cambridge CB2 0QQ,United Kingdom;andhDepartment of Biological Sciences,University of Southern California,Los Angeles,CA 90089Edited by Dennis A.Carson,University of California at San Diego,La Jolla,CA,and approved January 17,2013(received for review November 14,2012)Glioblastoma(GB)is the most common and aggressiveprimary brainmalignancy,with poor prognosis and a lack of effective therapeuticoptions.Accumulating evidence suggests that intratumor heteroge-neity likely is the key to understanding treatment failure.However,theextentofintratumorheterogeneityasaresultoftumorevolutionis still poorly understood.To address this,we developed a uniquesurgical multisampling scheme to collect spatially distinct tumorfragments from 11 GB patients.We present an integrated genomicanalysis that uncovers extensive intratumor heterogeneity,withmost patients displaying different GB subtypes within the sametumor.Moreover,we reconstructed the phylogeny of the fragmentsfor each patient,identifying copy number alterations in EGFR andCDKN2A/B/p14ARF as early events,and aberrations in PDGFRA andPTEN as later events during cancer progression.We also character-ized the clonal organization of each tumor fragment at the single-moleculelevel,detectingmultiplecoexistingcelllineages.Ourresultsreveal the genome-wide architecture of intratumor variability in GBacross multiple spatial scales and patient-specific patterns of cancerevolution,with consequences for treatment design.tumor progression|high grade gliomaGlioblastoma(GB)is the most common primary brain malig-nancy in adults and one of the most aggressive cancers.Themedian survival in the general patient population is just 4.6 mo.Even in optimally treated patients,the median survival is 14 mo,with a 26%2-y survival rate(1).Considering the average age atdiagnosis,GB typically results in more than 20 y of life lost(2).Since the 1970s,primary treatment has involved surgery followedby radiotherapy(3).Recently,targeted chemotherapy approachessuchasthealkylatingagenttemozolomide(1)alsohavebeenused,although with modest effects on survival.The impossibility of ex-tensive tumor debulking and poor drug delivery in the brain con-tribute significantly to the lack of effective treatment options andpoor prognosis.Insights into the genetic regulatory landscape of GB havebeen achieved through The Cancer Genome Atlas(4)and otherstudies(5).Furthermore,patterns of gene expression have beencollated to identify molecular subgroups with putative prog-nostic or predictive significance(5,6).Nevertheless,the poorprognosis is compounded by the endemic problem of diseaseheterogeneity,which has been reported extensively for othercancer types(710).In glioblastoma,FISH has been used toidentify receptor tyrosine kinase amplifications as markers forthe generation of heterogeneity through clonal evolution(11,12).These data are based on archival material from single tumorsamples.Spatial heterogeneity within an individual tumor masshas not been investigated yet,and the impact of sampling biashas not been addressed.Furthermore,genome-wide studies ofintratumor heterogeneity in GB have yet to be performed.Real-time perioperative tumor sampling of GB may be con-founded by tumor necrosis and the challenge of distinguishing dis-ease from normal brain tissue.To address this,we have developeda fluorescence-guided multiple sampling(FGMS)approach(13)based on 5-aminolevulinic acid administration(14)to improve ob-jective GB sampling in the operating theater.During surgery(Fig.1A),viable tumor tissue can be identified by visible fluorescence(Fig.1B).We adapted this technique to perform multiple samplingof objectively defined(visibly fluorescent)and spatially distinct GBtumor fragments from 11 patients(see Table S1 for sample details).During the operation,between four and six fragments(T1,T2,.)with a volume of 23 mm3each were obtained from the neoplasm,with samples separated by at least 1 cm(Fig.1C).The fragmentswere labeled in order of resection,with superficial fragments takenduring the early stages of tumor debulking(T1,T2),followed bydeeper fragments(T3,T4,.)taken later in the operation.In ad-dition,we collected a further fragment from the bulk of the tumormass(T)and a blood sample as a source of germline DNA to dis-tinguish somatic copy number lesions.Histopathological analysisshowed that all fragments had similar proliferative index,cellularatypia,and vascularization and were devoid of significant necroticareas(as evident by fluorescence-aided resection).This samplingtechnique,performeddirectlyintheoperatingtheater,allowedustocollect a unique dataset to interrogate intratumor heterogeneity atthe genomic level across the malignancy.Here,we show that genome-wide GB intratumor genomicheterogeneity can be decomposed to reveal tumor evolution.Moreover,we report that based upon gene expression levels,tumor fragments from the same patient may be classified intodifferent GB subtypes.Using single-molecule approaches,wealso investigate the clonal composition of single fragments,re-vealing that a hierarchy of mitotic clones coexists within thesame fragment.Our results show that tumor heterogeneityAuthor contributions:A.S.,C.W.,and S.T.designed research;A.S.,I.S.,and A.T.analyzedand interpreted the data;I.S.and S.G.M.P.performed sample collection,processing,anddata generation;C.C.contributed new reagents/analytic tools;V.P.C.performed histo-pathological analysis;C.W.collected the samples in the operating theater;and A.S.,I.S.,S.G.M.P.,A.T.,V.P.C.,J.C.M.,C.C.,C.W.,and S.T.wrote the paper.The authors declare no conflict of interest.This article is a PNAS Direct Submission.Freely available online through the PNAS open access option.Data deposition:Copy number arrays and gene number arrays reported in this paper havebeen deposited in the ArrayExpress Archive,www.ebi.ac.uk/arrayexpress/(accession nos.E-MTAB-1215 and E-MTAB-1129).1A.S.and I.S.contributed equally to this work.2To whom correspondence may be addressed.E-mail:cw209cam.ac.uk or st321cam.ac.uk.This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1219747110/-/DCSupplemental.www.pnas.org/cgi/doi/10.1073/pnas.1219747110PNAS|March 5,2013|vol.110|no.10|40094014MEDICAL SCIENCESrepresents a specific signature that informs on GB evolutionarydynamics at the single-patient level.ResultsSubset of GB Putative Drivers Is Consistently Heterogeneous.Weprofiled genome-wide DNA somatic copy number levels for 38tumor fragments from nine patients from the cohort(see TableS1).To investigate the global patterns of copy number alteration(CNA)within each patient,we took the union of CNAs that oc-curredinatleast one ofthetumorfragments.Weobservedseveralfrequent aberrationsthat havebeen reported in otherGB cohorts,including partial loss of chromosome 10(15)in eight patients andthe focal(60%ofcells)observed.Fig.4Bshowsthe phylogenetic reconstruction of the mitotic clones for threerepresentative tumor fragments,in which leaf thickness indicatestheabundanceofeach clone.This indicates that not only aretheredifferent subclones in each fragment,but they belong to clearlydistinct cell lineages,representative of a complex hierarchy.DiscussionFrom an evolutionary perspective(7,35),the divergent de-velopment of subpopulations of cancer cells within the same tu-mor likely is at the root of therapy failure,the development oftreatment resistance,and ultimately recurrence of the malignancy(3640).A detailed understanding of the evolutionary dynamicsof tumor progression will provide insight into the associatedmoleculargeneticmechanismsandwillallowustoconstructorderfrom apparent chaos.In GB,intratumor heterogeneity at the level of the tyrosinekinases EGFR,PDGFRA,and MET was demonstrated previouslyusing FISH(11).Moreover,cell lines with different EGFR/PDGFRA profiles derived from the same GB showed differentialresponse to growth factors(12).However,a comprehensive ge-nomicanalysisofintratumorheterogeneityandtumorevolutioninGB has not been previously described.More importantly,themechanisms behind intratumor heterogeneity and its conse-quences remain largely unknown.To date,analyses in GB as wellas in many other cancer types are based on single tissue samplesfrom individual patients.Here,we show that an objective multiplesampling scheme is the key to interrogating intratumor heteroge-neity and deconvoluting the underlying cancer dynamics.A singlebiopsy is unlikely to represent the full set of mutations present ina particular cancer because it may underestimate the landscape ofalterations that are present,as recently reported in renal carci-noma(10).For GB,in which the tumor is not resected as a solidmass but in a piecemeal fashion,multiple sampling from the sameneoplasm is challenging and is done most easily at the time ofsurgery.Using FGMS(Fluorescence-Guided Multiple Sampling),we have applied a real-time perioperative multiple sampling ap-proachduringGBresection,whichpartiallyconservesinformationsp56T3sp52Tsp52T1sp52T3sp52T2sp52T4sp49T1sp49T3sp49Tsp49T2sp49T4sp54T1sp54T2sp54T5sp54Tsp54T4sp54T6sp42T2sp42T1sp42Tsp42T3sp50T3sp50T2sp50T1sp50T4sp56Tsp56T1sp56T2sp56T4sp55Tsp55T2sp55T1sp55T3sp55T4sp41T1sp41Tsp41T2sp41T3sp42T4sp54T3sp57T1sp57Tsp57T3sp57T2sp57T4R4T1R4T3R4TR4T2R4T4ProneuralClassicalMesenchymalNeuralABCFig.3.Intratumor heterogeneity at the transcrip-tional level.(A)Clustering based on the gene ex-pression profiles of the full set of 16,811 transcriptsresults in the grouping of most tumor specimens bypatient,with three exceptions(sp42-T4,sp54-T3,andsp56-T3).(B)Despite the overall patient-specific sig-nature,samples taken from the same tumor wereclassified into different GB subtypes in 6 of 10patients.(C)Gene ontology analysis of the genes thatexhibit intratumor heterogeneity in each of the 10patients revealed that they were involved in biologicalprocesses related to neuron generation/development,cell morphogenesis,and tumor angiogenesis.4012|www.pnas.org/cgi/doi/10.1073/pnas.1219747110Sottoriva et al.on the spatial organization of the samples.All patients underwentprimary GB resection with no preoperative treatment that wouldhave introduced biases in the analysis of genetic alterations.Here,we demonstrate that GB exhibits a landscape of hetero-geneous mutations across the whole genome at the copy numberlevel;this represents a signature of the history of the malignancyfrom the first tumor-founding cell(s).By deconvoluting suchasignature,wemayinferthetemporalsequenceofalterationsthathave occurred in the malignancy.Further,using gene expressiondataweobservedthatdifferentsamplesfromthesametumorwereclassified into different GB subtypes(Fig.3A).This shows that theimpact of sampling bias must be considered when establishingmolecular criteria for patient stratification.Supporting this asser-tion,the morphological data available from individual tissue bi-opsies may provide misleading information,resulting in diagnosticerrors(39).Hence,these results support the development of per-sonalized treatments based upon a multimodal approach that usesinformation from multiple samples from the same GB.As an example of how our data might be integrated with thecurrent standard of clinical practice,Fig.5 shows how the ana-tomical positions of the five tumor fragments collected for patientsp42(A)may be coupled with the reconstructed evolution of thetumor mass(B).During the operation,fragments are numbered inorderofresection,approximatelycorrespondingtothedepthofthesample within the brain.This coupling reveals a complex evolu-tionary pattern with the accumulation of malignant traits in dif-ferent parts of the neoplasm(C).Our analysis suggests that thefounder clone displayed amplification/gain of EGFR,CDK6,andMET,and loss/deletion of CDKN2A/B,PTEN,and PARK2(hencetargeting the Ras and Rb pathways),before splitting into twopopulations(subclones1and2),thefirstofwhichgeneratedT2andT3,with T3 gaining a copy of chromosome 3(which containsPIK3CA).The second subclone subsequently acquired furthermutations manifested in T4,which also displayed an altered geneexpressionprofilebelongingtoa mesenchymalsubtype,despite therestofthetumorfragmentsbeingproneural.Subclone2gaverisetoanother independent+PIK3CA subclone(again with chromosome3 gain).The latter then expanded to form T1 and,with the partialloss of chromosome 17(containing NF1 and TP53)and thereforealterationofthep53pathway,generatedT.Ofnote,theoccurrenceofnewlesions,suchaslossofNF1andTP53infragmentT,maynotnecessarilyrepresentafastclonalexpansionevent,butalsoaslowerprocess of selection of a preexisting rare clone that becomesdominant within a fragment and,therefore,detectable by copynumber profiling.Our study presents an integrated analysis of intratumorheterogeneity at the genotype level(copy number),cellular phe-notype(gene expression),and single-molecule mitotic level(mo-lecular clocks).To our knowledge,the multiple sampling schemeand genomic data have never before been integrated in this way todescribe GB evolution at the individual patient level.Taken together,our results shed light upon intratumor hetero-geneity and the clonal evolution of GB.Based on these results,wepropose that patient-specific dynamics of tumor heterogeneityunderlie variation in response to treatment.Specifically,aftertherapy,the surviving population may not be a single resistantcancer clone,but rather a heterogeneous population of malignantcells with genetic aberrations that allow them to survive the initialtreatment.This view extends the concept of clonal evolution byallowing selection to play a role limited only by the spatialstructure of the neoplasm,in which multiple clones with differentfitness coexist within the same cancer(41).Instead of benefittingsingle clones,this would favor the whole tumor population byconverting heterogeneity into an asset to resist treatment(7,42).This hypothesis implies that patterns of heterogeneity might