LETTER doi:10.1038/nature10738 Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing Li Ding.2+,Timothy J.Ley.3.4*,David E.Larson',Christopher A.Miller,Daniel C.Koboldt',John S.Welch3,Julie K.Ritchey3, Margaret A.Young Tamara Lamprecht,Michael D.McLellan',Joshua F.McMichael',John W.Wallis2,Charles Lu',Dong Shen' Christopher C.Harris David J.Dooling 2,Robert S.Fulton,Lucinda L.Fulton2,Ken Chen2,Heather Schmidt', Joelle Kalicki-Veizer,Vincent J.Magrini2,Lisa Cook',Sean D.McGrath',TammiL.Vickery',MichaelC.Wendl2,Sharon Heath2 Mark A.WatsonDanielC.Link4,Michael H.Tomasson3.4,William D.Shannon,Jacqueline E.Payton,Shashikant Kulkarni2.4.5 Peter Westervelt Matthew J.WalterTimothy A.GraubertElaineR.MardisRichard K.Wilson&JohnF.DiPersio Most patients with acute myeloid leukaemia(AML)die from pro- 1 to 3 were validated (see ref.7 for tier designations;Supplementary gressive disease after relapse,which is associated with clonal evolu- Fig.1a and Supplementary Tables 4a and 5).Of these,78 mutations tion at the cytogenetic level2.To determine the mutational were relapse-specific(63 point mutations,1 dinucleotide mutation,13 spectrum associated with relapse,we sequenced the primary tumour indels and 1 translocation;relapse-specific criteria described in Sup- and relapse genomes from eight AML patients,and validated plementary Information and shown in Supplementary Fig 1b),5 point hundreds of somatic mutations using deep sequencing;this allowed mutations were primary-tumour-specific,and 330(317 point mutations us to define clonality and clonal evolution patterns precisely at and 13 indels)were shared between the primary tumour and relapse relapse.In addition to discovering novel,recurrently mutated genes samples (Fig.1a,b and Supplementary Fig.2).The skin sample was (for example,WAC SMC3,DIS3,DDX4I and DAXX)in AML,we contaminated with leukaemic cells for this case (peripheral white blood also found two major clonal evolution patterns during AML relapse: cell count was 105,000 cells mm3 when the skin sample was banked), (1)the founding clone in the primary tumour gained mutations and with an estimated tumour content in the skin sample of 29% evolved into the relapse clone,or(2)a subclone of the founding (Supplementary Information).In addition to the ten somatic non- clone survived initial therapy,gained additional mutations and synonymous mutations originally reported for the primary tumour expanded at relapse.In all cases,chemotherapy failed to eradicate sample,we identified one deletion that was not detected in the original the founding clone.The comparison of relapse-specific versus analysis(DNMT3A L723fs(ref.8))and three mis-sense mutations previ- primary tumour mutations in all eight cases revealed an increase ously misclassified as germline events(SMC3 G662C,PDXDCI E421K in transversions,probably due to DNA damage caused by cytotoxic and TTNE14263K)(Fig.1b,Table 1 and Supplementary Table 4b). chemotherapy.These data demonstrate that AML relapse is asso- A total of 169 tier 1 coding mutations(approximately 21 per case) ciated with the addition of new mutations and clonal evolution, were identified in the eight patients (Table 1 and Supplementary which is shaped,in part,by the chemotherapy that the patients Tables 4b and 6),of which 19 were relapse-specific.In addition to receive to establish and maintain remissions. mutations in known AML genes such as DNMT3A(ref.8),FLT3 To investigate the genetic changes associated with AML relapse,and (ref.9),NPMI(ref.10),IDHI(ref.7),IDH2(ref.11),WTI(ref.12) to determine whether clonal evolution contributes to relapse,we per- RUNXI(refs 13,14),PTPRT(ref.3),PHF6(ref.15)and ETV6(ref.16) formed whole-genome sequencing of primary tumour-relapse pairs in these eight patients,we also discovered novel,recurring mutations and matched skin samples fromeight patients,including unique patient in WAC,SMC3,DIS3,DDX41 and DAXX using 200 AML cases whose identifier(UPN)933124,whose primary tumour mutations were previ- exomes were sequenced as part of the Cancer Genome Atlas AML ously reported.Informed consent explicit for whole-genome sequen- project(Table 1,Supplementary Table 4b and Supplementary Fig.3; cing was obtained for all patients on a protocol approved by the T.J.L.,R.K.W.and The Cancer Genome Atlas working group on AML, Washington University Medical School Institutional Review Board. unpublished data).Details regarding the novel,recurrently mutated We obtained >25X haploid coverage and >97%diploid coverage for genes are provided in Table 1,Supplementary Tables 4b and 7 and each sample(Supplementary Table 1 and Supplementary Information). Supplementary Figs 3 and 4.Structural and functional analyses of These patients were from five different French-American-British structural variants are presented in the Supplementary Information haematological subtypes,with elapsed times of 235-961 days between (Supplementary Figs 5-10 and Supplementary Tables 2,8 and 9). initial diagnosis and relapse(Supplementary Table 2a,b). The generation ofhigh-depth sequencing data allowed us to quantify Candidate somatic events in the primary tumour and relapse genomes accurately mutant allele frequencies in all cases,permitting estimation were identifiedss and selected for hybridization capture-based validation of the size of tumour clonal populations in each AML sample.On the using methods described in Supplementary Information.Deep sequen- basis of mutation clustering results,we inferred the identity of four cing of the captured target DNAs from skin (the matched normal tissue), clones having distinct sets of mutations (clusters)in the primary primary tumour and relapse tumour specimense (Supplementary tumour of AML1/UPN 933124(Supplementary Information).The Table 3)yielded a median of 590-fold coverage per site.The average median mutant allele frequencies in the primary tumour for clusters number of mutations and structural variants was 539(range 118- 1to 4 were 46.86%,24.89%,16.00%and 2.39%,respectively (Fig.1band 1,292)per case (Fig.1a). Supplementary Table 5c).Clone 1 is the 'founding'clone(that is,the The general approach for relapse analysis is exemplified by the first other subclones are derived from it),containing the cluster I mutations; sequenced case(UPN 933124).A total of 413 somatic events from tiers assuming that nearly all of these mutations are heterozygous,they must The Genome Institute,Washington University,St Louis,Missouri63108,USA2Department of Genetics,Washington University,StLouis,Missouri63110,USADepartmentof Internal Medicine,Division of Oncology,Washington University,St Louis,Missouri63110.USASiteman Cancer Center,Washington University,St Louis Missouri63110.USASDepartment of Pathology and Immunology.Washington University,St Louis,Missouri 63110,USA Division of Biostatistics,Washington University,St Louis,Missouri 63110,USA. .These authors contributed equally to this work 506 NATURE I VOL JANUARY 2012 2012 Macmillan Publishers Limited.All rights reservedLETTER doi:10.1038/nature10738 Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing Li Ding1,2*, Timothy J. Ley1,3,4*, David E. Larson1 , Christopher A. Miller1 , Daniel C. Koboldt1 , John S. Welch3 , Julie K. Ritchey3 , Margaret A. Young3 , Tamara Lamprecht3 , Michael D. McLellan1 , Joshua F. McMichael1 , John W. Wallis1,2, Charles Lu1 , Dong Shen1 , Christopher C. Harris1 , David J. Dooling1,2, Robert S. Fulton1,2, Lucinda L. Fulton1,2, Ken Chen1,2, Heather Schmidt1 , Joelle Kalicki-Veizer1 , Vincent J. Magrini1,2, Lisa Cook1 , Sean D. McGrath1 , Tammi L. Vickery1 , Michael C.Wendl1,2, Sharon Heath3 , Mark A.Watson5 , Daniel C. Link3,4, Michael H. Tomasson3,4,William D. Shannon6 , Jacqueline E. Payton5 , Shashikant Kulkarni2,4,5, Peter Westervelt3,4, Matthew J.Walter3,4, Timothy A. Graubert3,4, Elaine R. Mardis1,2,4, Richard K. Wilson1,2,4 & John F. DiPersio3,4 Most patients with acute myeloid leukaemia (AML) die from pro￾gressive disease after relapse, which is associated with clonal evolu￾tion at the cytogenetic level1,2. To determine the mutational spectrum associated with relapse, we sequenced the primary tumour and relapse genomes from eight AML patients, and validated hundreds of somatic mutations using deep sequencing; this allowed us to define clonality and clonal evolution patterns precisely at relapse. In addition to discovering novel, recurrently mutated genes (for example, WAC, SMC3, DIS3, DDX41 and DAXX) in AML, we also found two major clonal evolution patterns duringAML relapse: (1) the founding clone in the primary tumour gained mutations and evolved into the relapse clone, or (2) a subclone of the founding clone survived initial therapy, gained additional mutations and expanded at relapse. In all cases, chemotherapy failed to eradicate the founding clone. The comparison of relapse-specific versus primary tumour mutations in all eight cases revealed an increase in transversions, probably due to DNA damage caused by cytotoxic chemotherapy. These data demonstrate that AML relapse is asso￾ciated with the addition of new mutations and clonal evolution, which is shaped, in part, by the chemotherapy that the patients receive to establish and maintain remissions. To investigate the genetic changes associated with AML relapse, and to determine whether clonal evolution contributes to relapse, we per￾formed whole-genome sequencing of primary tumour–relapse pairs and matched skin samples from eight patients, including unique patient identifier (UPN) 933124, whose primary tumour mutations were previ￾ously reported3 . Informed consent explicit for whole-genome sequen￾cing was obtained for all patients on a protocol approved by the Washington University Medical School Institutional Review Board. We obtained .253 haploid coverage and .97% diploid coverage for each sample (Supplementary Table 1 and Supplementary Information). These patients were from five different French–American–British haematological subtypes, with elapsed times of 235–961 days between initial diagnosis and relapse (Supplementary Table 2a, b). Candidate somatic events in the primary tumour and relapse genomes were identified4,5 and selected for hybridization capture-based validation using methods described in Supplementary Information. Deep sequen￾cing of the captured target DNAsfrom skin (the matched normal tissue), primary tumour and relapse tumour specimens6 (Supplementary Table 3) yielded a median of 590-fold coverage per site. The average number of mutations and structural variants was 539 (range 118– 1,292) per case (Fig. 1a). The general approach for relapse analysis is exemplified by the first sequenced case (UPN 933124). A total of 413 somatic events from tiers 1 to 3 were validated (see ref. 7 for tier designations; Supplementary Fig. 1a and Supplementary Tables 4a and 5). Of these, 78 mutations were relapse-specific (63 point mutations, 1 dinucleotide mutation, 13 indels and 1 translocation; relapse-specific criteria described in Sup￾plementary Information and shown in Supplementary Fig. 1b), 5 point mutations were primary-tumour-specific, and 330 (317 pointmutations and 13 indels) were shared between the primary tumour and relapse samples (Fig. 1a, b and Supplementary Fig. 2). The skin sample was contaminated with leukaemic cells for this case (peripheral white blood cell count was 105,000 cells mm23 when the skin sample was banked), with an estimated tumour content in the skin sample of 29% (Supplementary Information). In addition to the ten somatic non￾synonymous mutations originally reported for the primary tumour sample3 , we identified one deletion that was not detected in the original analysis (DNMT3AL723fs (ref. 8)) and three mis-sense mutations previ￾ously misclassified as germline events (SMC3 G662C, PDXDC1 E421K and TTN E14263K) (Fig. 1b, Table 1 and Supplementary Table 4b). A total of 169 tier 1 coding mutations (approximately 21 per case) were identified in the eight patients (Table 1 and Supplementary Tables 4b and 6), of which 19 were relapse-specific. In addition to mutations in known AML genes such as DNMT3A (ref. 8), FLT3 (ref. 9), NPM1 (ref. 10), IDH1 (ref. 7), IDH2 (ref. 11), WT1 (ref. 12), RUNX1 (refs 13, 14), PTPRT (ref. 3), PHF6 (ref. 15) and ETV6 (ref. 16) in these eight patients, we also discovered novel, recurring mutations in WAC, SMC3, DIS3, DDX41 and DAXX using 200 AML cases whose exomes were sequenced as part of the Cancer Genome Atlas AML project (Table 1, Supplementary Table 4b and Supplementary Fig. 3; T.J.L., R.K.W. and The Cancer Genome Atlas working group on AML, unpublished data). Details regarding the novel, recurrently mutated genes are provided in Table 1, Supplementary Tables 4b and 7 and Supplementary Figs 3 and 4. Structural and functional analyses of structural variants are presented in the Supplementary Information (Supplementary Figs 5–10 and Supplementary Tables 2, 8 and 9). The generation of high-depth sequencing data allowed us to quantify accurately mutant allele frequencies in all cases, permitting estimation of the size of tumour clonal populations in each AML sample. On the basis of mutation clustering results, we inferred the identity of four clones having distinct sets of mutations (clusters) in the primary tumour of AML1/UPN 933124 (Supplementary Information). The median mutant allele frequencies in the primary tumour for clusters 1 to 4 were 46.86%, 24.89%, 16.00% and 2.39%, respectively (Fig. 1b and Supplementary Table 5c). Clone 1 is the ‘founding’ clone (that is, the other subclones are derived from it), containing the cluster 1 mutations; assuming that nearly all of these mutations are heterozygous, they must 1 The Genome Institute, Washington University, St Louis, Missouri 63108, USA. 2 Department of Genetics, Washington University, St Louis, Missouri 63110, USA. 3 Department of Internal Medicine, Division of Oncology, Washington University, St Louis, Missouri 63110, USA. 4 Siteman Cancer Center, Washington University, St Louis, Missouri 63110, USA. 5 Department of Pathology and Immunology, Washington University, St Louis, Missouri 63110, USA. 6 Division of Biostatistics, Washington University, St Louis, Missouri 63110, USA. *These authors contributed equally to this work. 506 | NATURE | VOL 481 | 26 JANUARY 2012 ©2012 Macmillan Publishers Limited. All rights reserved