Genome-Wide Profiling of Genetic Alterations in Acute Lymphoblastic

Genome-wide proﬁling of genetic alterations in acute lymphoblastic leukemia: recent insights and future directions

CG Mullighan and JR Downing

Department of Pathology, St Jude Children’s Research Hospital, Memphis, TN, USA

Until recently, our understanding of the genetic factors toxicities. To identify targets for new therapies, a thorough contributing to the pathogenesis of acute lymphoblastic understanding of the genetic lesions contributing to establish- leukemia (ALL) has relied on the detection of gross chromosomal alterations and mutational analysis of individual genes. ment of the leukemic clone and resistance to therapy is required. Although these approaches have identified many important Both B-progenitor and T-lineage ALL are characterized by abnormalities, they have been unable to identify the full recurring chromosomal aneuploidies and translocations, detec- repertoire of genetic alterations in ALL. The advent of high- tion of which is important in diagnosis and risk stratification. resolution, microarray-based techniques to identify DNA copy These include high hyperdiploidy, hypodiploidy, the trans- number alterations and loss-of-heterozygosity in a genome- locations t(12;21)(p13;q22)[ETV6-RUNX1], t(1;19)(q23;p13.3) wide fashion has enabled the identification of multiple novel genetic alterations targeting key cellular pathways, including [TCF3-PBX1], t(9;22)(q34;q11)[BCR-ABL1] and rearrangement lymphoid differentiation, cell cycle, tumor suppression, apop- of the MLL gene at 11q23 in B-progenitor ALL, rearrangement of tosis and drug responsiveness. Recent studies have extended MYC in mature B-cell leukemia/lymphoma, and rearrangement these approaches to examine the biologic basis of high-risk of the TLX1 (HOX11), TLX3 (HOX11L2), LYL1, TAL1 and MLL ALL and treatment relapse. As these techniques continue to genes in T-ALL.8–12 These translocations frequently involve evolve and are integrated with genome-wide epigenetic and transcriptional regulators of hematopoiesis and are important transcriptomic data, we will obtain a comprehensive understanding of the genetic and epigenetic alterations in ALL, and initiating events in leukemogenesis, but the observations that ultimately will be able to translate these findings into the these abnormalities are often detectable years before the onset development of novel therapeutic approaches directed against of leukemia and usually do not alone result in leukemia in rational therapeutic targets. Here, we review recent data experimental models 13,14 suggest that cooperating genetic or obtained from genome-wide profiling studies in ALL, and epigenetic lesions are required. Candidate gene studies have discuss potential avenues for future investigation. identified lesions such as deletion of the CDKN2A/CDKN2B Leukemia (2009) 23, 1209–1218; doi:10.1038/leu.2009.18; (encoding the tumor suppressors p16INK4A, p14ARF and published online 26 February 2009 15–18 Keywords: acute lymphoblastic leukemia; SNP; microarray; p15INK4B) and NOTCH1 in T-ALL, but until recently relapse; PAX5; IKAROS methods to analyze genetic alterations in a genome-wide fashion have been lacking.

Microarray platforms for detection of genetic alterations in cancer Acute lymphoblastic leukemia Several microarray-based techniques for the genome-wide Acute lymphoblastic leukemia (ALL) is the commonest malig- analysis of DNA copy number alterations (CNA) and loss-of- nancy affecting children,1 and is a major cause of morbidity heterozygosity (LOH) have now become well established in the from hematopoietic malignancies in adults. Recent decades 19 study of cancer genomics, and studies of ALL are among the have witnessed tremendous advances in the success of 20 most advanced in the field. The different platforms vary in the treatment of childhood ALL, with over 80% of individuals now SPOTLIGHT nature of the probes used, genomic resolution and the ability to cured.1,2 However, there remains considerable scope for detect LOH independently of copy number state. Array-based improving treatment outcome in ALL. The majority of adult comparative genomic hybridization (CGH) using bacterial patients with ALL are not cured,3 in part, due to an increased artificial chromosomes (BACs) has been highly informative in frequency of unfavorable genetic alterations (most notably the 21–35 many tumor types, including ALL. BAC arrays typically use Philadelphia chromosome encoding BCR-ABL1).3–5 Further- probes derived from large (up to B200 kb) fragments of human more, up to 25% of the children and over half of the adults genomic DNA cloned into BAC vectors. BAC arrays now offer experience relapse, which carries a dismal prognosis.5–7 tiling coverage of the majority of the human genome, and Improvements in these outcomes by refining schedules or generally high signal-to-noise ratio, but due to the large probe escalating doses of existing chemotherapeutic agents is limited 36,37 size have limited ability to detect focal CNA, which, as by toxicity. Novel targeted therapies are urgently required in 38 discussed below, are a hallmark of ALL. ALL to improve cure rates and reduce short- and long-term Oligonucleotide arrays use short (20–100 nucleotide) probes and offer the ability to interrogate the copy number state of Correspondence: Dr CG Mullighan, Assistant Member, St Jude defined genomic regions or entire genomes at extremely high Children’s Research Hospital, 262 Danny Thomas Place, Pathology, resolution, and have become more widely used than BAC Mail Stop 342, Memphis, TN 38105, USA. E-mail: [email protected] arrays. Two types of oligonucleotide array are available: oligo- Received 24 December 2008; accepted 9 January 2009; published CGH arrays, in which test and reference DNA are labeled and online 26 February 2009 hybridized simultaneously to the same array, allowing the Genomic analysis of ALL CG Mullighan and JR Downing 1210 detection of CNA but not copy-neutral LOH; and single Table 1 Key points nucleotide polymorphism (SNP) arrays that interrogate both CNA and LOH. K Genetic characterization of acute lymphoblastic leukemia (ALL) SNP arrays utilize oligonucleotide probes specific for each has traditionally relied on cytogenetic analysis and detection of allele of hundreds of thousands of SNPs, and are widely used as specific translocations or mutations by molecular techniques. a genotyping tool to enable genome-wide association studies These remain important in diagnosis but do not explain the full genetic complexity of ALL. examining associations between SNP genotype and disease K 39 Microarray based analysis of DNA copy number alterations susceptibility or phenotype. As the intensity of probe (CNA) and loss-of-heterozygosity (LOH) can identify genetic hybridization at each SNP also reflects DNA copy number alterations at extremely high resolution. Platforms include state, SNP arrays offer the ability to detect both somatic CNA in bacterial artificial chromosome, oligonucleotide and single tumor samples, and inherited copy number variants (CNV) in nucleotide polymorphism (SNP) microarrays. SNP arrays identify constitutional DNA samples. By comparing SNP genotypes of CNA and copy-neutral LOH. Current platforms have a resolution of B2 million probes per array tumor samples to reference samples (ideally constitutional DNA K Careful simultaneous analysis of both tumor and corresponding obtained from the same individual as the tumor sample) SNP germline DNA is required for optimal CNA/LOH detection arrays can detect regions of LOH. LOH may be caused by CNA K Both B-progenitor and T-lineage ALL harbor recurring CNA (deletion or high-level gain) or may be ‘copy neutral’. Such targeting genes regulating lymphoid development, tumor copy-neutral LOH (CN-LOH), also known as acquired uni- suppression, cell cycle, apoptosis, lymphoid signaling and drug parental disomy may reflect reduplication of a mutated or responsiveness. These alterations are frequently focal (less than 1 Mb) and not evident on cytogenetic analysis. aberrantly methylated region of the genome. Identification of SPOTLIGHT K Alterations in genes regulating B-lymphoid development are CN-LOH is consequently of considerable interest in cancer present in over 60% of B-progenitor ALL, and include CNA, genomics. The ability to detect both CNA and LOH, and the sequence mutations and translocations of lymphoid modest DNA requirement of SNP arrays have resulted in transcriptional factors such as PAX5, IKZF1 and EBF1. Loss-of- increasingly widespread use of this platform in studies profiling function of these genes accelerates leukemogenesis in genetic alterations in cancer. Currently available arrays (for experimental models, suggesting that these genetic alterations contribute to leukemogenesis in human ALL. example, from Affymetrix (Santa Clara, CA, USA)40 and Illumina K Genetic alterations identified in diagnosis leukemic cells, such 41 B (San Diego, CA, USA) ) incorporate up to 1.8 million SNP as deletion/mutation of IKZF1, predict poor outcome in ALL and CNV probes enabling interrogation of CNA at very high K Profiling of CNA at diagnosis and relapse in ALL has identified (sub-kb) resolution. marked evolution of genetic alterations from diagnosis to relapse, and a common clonal origin of diagnosis and relapse leukemic cells in most cases, suggesting that specific genetic changes are selected for during treatment and confer resistance Genome-wide analysis of genetic alterations in ALL to therapy K Future studies will require comprehensive analysis of sequence Numerous studies have used BAC array-CGH analysis in to mutations, epigenetic profiling, and integration of genomic data identify CNA in ALL, many of which are submicroscopic, or not from multiple genomic platforms in order to define the full evident on cytogenetic analysis (Table 1).22,23,25–35,42 These complement of genomic alterations contributing to the studies have delineated regions of deletion (for example, ETV6 pathogenesis of ALL, treatment failure and the identification of new therapeutic targets. in B-ALL), have mapped translocation breakpoints,25,30 have identified new CNA (for example, 9q34 duplication in T-ALL),27 and characterized large and/or complex alterations (for example, gains of 1q and intrachromosomal amplification of regions of homozygosity. Secondly, we developed an array chromosome 21 in B-ALL).26,28 However, the superior resolu- normalization procedure (‘reference normalization’) that uses tion of oligonucleotide array platforms, such as SNP arrays, has SNP data only from diploid chromosomes to guide normal- enabled a more detailed and comprehensive characterization of ization of array data, which was critical to obtain accurate genetic alterations in ALL. copy number inferences from aneuploid samples.38,49 Finally, Irving et al.43 examined a small cohort of pediatric ALL cases we used circular binary segmentation, an automated method (N ¼ 10) using SNP arrays examining 10 000 markers, and of identifying CNA originally developed for array-CGH data, detected LOH in the majority of cases, most commonly to comprehensively identify all lesions.50 involving CDKN2A/B at 9p21.3. This was one of the first studies This study identified over 50 recurring regions of CNA. As to show that SNP arrays could successfully identify LOH in expected, we identified large genomic alterations, such as cancer, although the relatively low resolution of these arrays aneuploidies in hyper- and hypodiploid ALL, and gains of 1q in precluded precise delineation of many of the identified regions high hyperdiploid and unbalanced t(1;19) [TCF3-PBX1] ALL of genetic alteration. (Figure 1a). However, most recurring regions of CNA identified We and others have performed high-resolution SNP array were focal, with the minimal common region of loss or gain analysis of large cohorts of pediatric ALL patients, and identified being less than 1Mb in size, and in the majority of cases multiple novel regions of CNA and LOH in diagnostic ALL targeting a single gene. The majority of focal CNA were samples.38,44–48 Using Affymetrix SNP arrays interrogating over deletional, with few focal amplifications identified (a notable 350 000 markers at an average intermarker resolution of less exception being MYB in T-lineage ALL). The lesions targeted than 5 Kb, we examined 242 diagnosis B-progenitor (N ¼ 192) multiple key cellular pathways, including lymphoid differen- and T-lineage (N ¼ 50) ALL samples.38 Several key methodolo- tiation (PAX5, EBF1, IKZF1, IKZF2, IKZF3, TCF3, LEF1, RAG1/2), gical aspects enabled accurate identification of somatic tumor suppression (NF1, PTEN, RB1), apoptosis (BTG1), (tumor-acquired) lesions in this study. SNP array analysis lymphocyte signaling (BTLA/CD200), microRNAs (mir-15/16) of constitutional DNA samples was performed in conjunction and drug responsiveness (NR3C1), in addition to genes with with tumor samples for the majority of samples, enabling tumor- no known role in lymphoid development or oncogenesis acquired CNA to be definitively distinguished from inherited (for example, ADD3, DMD) and uncharacterized genes copy number variations, and acquired LOH from inherited (C20orf94). Notably, the average number of CNA was low

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1211

Figure 1 DNA copy number alterations and PAX5 alterations in ALL. (a) Log2-ratio copy number heatmap of Affymetrix single nucleotide polymorphism (SNP) array genome-wide copy number data obtained from 126 representative pediatric B-progenitor and T-lineage acute lymphoblastic leukemia (ALL) cases (blue is deletion, red is gain). (b) Copy number heatmap of 61 pediatric ALL cases with copy number alterations (CNA) involving PAX5 at 9p13.2. Reproduced with permission from Seminars in Hematology 2009;46:3–15; copyright: Elsevier. SPOTLIGHT

(mean 6 per case), indicating that in most cases, ALL is not ALL subtypes. Deletions of ADD3, C20orf94, the ETS family characterized by gross genomic instability. Indeed, specific members ERG and ETV6, the fragile histidine triad gene FHIT, mechanisms underlie the acquisition of CNA, as discussed TBL1XR1, a histone cluster at 6p22.22 were common in B-ALL below. Furthermore, there were significant differences in the but rarely, if ever, observed in T-ALL. Conversely, CDKN2A/B frequency of CNA among ALL subtypes. MLL-rearranged ALL deletion was more frequent in T-ALL (72 vs 33.9% of B-ALL cases had less than one CNA per case, most of which occurred cases).38 at the breakpoints of MLL and its fusion partners. This suggests that few cooperating lesions are required in MLL-rearranged leukemia. This is compatible with the behavior of this leukemia, Mutations of genes regulating B-lymphoid development in which typically develops in utero and presents early in infancy. ALL In contrast, ETV6-RUNX1 and BCR-ABL1 leukemia both harbored over six lesions per case, suggesting the requirement Genes regulating normal B-lymphoid development were for multiple additional lesions in leukemogenesis.51 The mutated in over 40% of B-ALL cases (Figures 1b and 2). PAX5, frequency of individual CNA also differed significantly among which encodes a transcription factor required for B-lineage

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1212

Figure 2 Mutations in genes regulating B-lymphoid development in B-progenitor ALL. The stepwise differentiation of hematopoietic stem cells to mature B cells is shown. This process is regulated by a tightly controlled hierarchy of transcription factors that enforce B-lineage commitment and differentiation, and suppress differentiation to alternate (T-lymphoid, myeloid) lineages. These transcription factors are targeted by genetic alterations in over 60% of B-progenitor acute lymphoblastic leukemia (ALL) cases. Notably, the genetic alterations affect transcription factors controlling differentiation up to the pre-B cell stage of development, at which differentiation is arrested in many B-ALL cases. Additional known or putative B-lymphoid regulatory genes targeted less frequently in ALL include BLNK, IKZF2, IKZF3, MEF2C, RAG1/2 and SOX4. SPOTLIGHT commitment and maturation,52–55 was most commonly mutations resulting in truncation of this domain, and are highly involved, and targeted by deletion, focal amplification, trans- deleterious in in vitro assays. A similar frequency and pattern of location or sequence mutation in 31.7% of cases, a finding PAX5 CNA and sequence mutations has been identified in other confirmed in other SNP array studies of ALL (Figure 2).44,46 childhood B-ALL cohorts.67 The most common mutation is Approximately half of the CNA involved the entire PAX5 locus. P80R, which is the most deleterious of the paired domain In the remainder, a subset of exons encoding either the mutations in assays of the DNA-binding and transactivating N-terminal, DNA-binding paired domain and/or C-terminal activity of PAX5. transactivating domains were involved. 40% of deletions In addition to PAX5 alterations, CNA were identified in involved an internal subset of exons, resulting in the expression additional regulators of B-lymphoid development, including the of internally truncated transcripts that encode PAX5 proteins IKAROS family members IKZF1 (IKAROS), IKZF2 (HELIOS) and lacking key functional domains, or frameshifted, hypomorphic IKZF3 (AIOLOS), LEF1, TCF3 and BLNK. A high frequency of alleles. This pattern of very focal deletions involving a subset of deletion of the pre-B-cell receptor gene, VPREB1, was also exons, and resulting in the expression of internally truncated observed, but as this gene is located within the immunoglobulin proteins with aberrant function, has now been observed in lambda light chain locus at 22q11.22 and is frequently deleted multiple genes in B-ALL, including IKZF1 (see below). Several on rearrangement of this locus, the significance of VPREB1 cases had deletions involving C-terminal exons and extending deletion is presently unclear. The frequency of B-pathway distally to PAX5. This pattern was also observed at the alterations varied significantly among ALL subtypes. All breakpoints of known translocation partners in ALL (for hypodiploid ALL cases harbored PAX5 alterations, with the example, MLL, ETV6, RUNX1 and TCF3) suggesting that these majority also having concomitant sequence mutations and other cases may harbor cryptic PAX5 translocations. Three trans- lesions in the pathway. Interestingly, this ALL subtype has a poor locations were identified (PAX5-ETV6,56,57 PAX5-FOXP1, PAX5- prognosis, and subsequent analyses of high-risk B-progenitor ZNF521), each of which fuses the DNA-binding paired domain childhood ALL cohorts have shown that an increasing number of of PAX5 to key functional domains of partner genes that have lesions in this pathway is associated with poor outcome (see known or putative roles in lymphoid development or onco- below).67 One-third of ETV6-RUNX1 cases harbored focal PAX5 genesis.58–62 These translocations are predicted to disrupt CNA (but no sequence mutations), and EBF1 deletion in a normal activity of PAX5 (and possibly the partner gene), and minority of cases, but never deletion of IKZF1. In contrast, in luciferase reporter assays using a CD19 reporter, the fusions deletion of IKZF1 is a near-obligate lesion in BCR-ABL1 ALL.68 were found to act as competitive inhibitors of normal PAX5 Few lesions in this pathway were observed in MLL-rearranged or transactivating activity. Subsequent studies have identified high-hyperdiploid ALL. CNA targeting lymphoid differentiation additional PAX5 fusions in ALL, involving multiple partner are less frequent in T-ALL and involve early lymphoid regulators genes including ASXL1, AUTS2, BRD1, C20orf112, DACH1, (for example, the IKAROS family) rather than B-lineage-specific ELN, HIPK1, JAK2, KIF3B, LOC392027, PML, POM121 and genes. SLCO1B3.47,63–66 The overall frequency of PAX5 translocations The genomic and in vitro functional data described above appears low (B2.6% B-ALL cases),66 although studies compre- suggest that genetic lesions targeting lymphoid development hensively examining PAX5 translocation in large B-ALL cohorts are important in the pathogenesis of ALL. This is supported by are awaited. in vivo studies modeling the role of Pax5 haploinsufficiency Genomic resequencing of PAX5 in the St Jude study described in leukemogenesis. In a murine retroviral bone marrow above also identified 16 missense, insertion/deletion, splice site transplant model of BCR-ABL1 B-progenitor ALL,14,69 Pax5 and frameshift mutations that clustered in the key DNA-binding haploinsufficiency increases the penetrance and reduces the and transactivating domains of PAX5. Notably, most paired latency of B-ALL .70 Additional CNA of Pax5 and Cdkn2a/b, and domain mutations were missense, and each of these (V26G, sequence mutations of Pax5, were observed in the resulting P34Q and P80R) was predicted by structural modeling to murine tumors, further supporting a role for these alterations in impair the interaction of PAX5 with its DNA targets. This leukemogenesis. To further test the hypothesis that PAX5 is a was confirmed using multiple in vitro assays of DNA-binding haploinsufficient tumor suppressor, we performed chemical and and transactivating activity.38 In contrast, mutations involving retroviral mutagenesis of thymectomized mice heterozygous for the transactivating domain were more commonly frameshift a Pax5 null allele71 using retrovirus (Moloney murine leukemia

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1213 virus) or N-ethyl-N-nitrosourea.72 In both approaches, Pax5 loss intense research activity. For example, CN-LOH accompanying was associated with a highly significant increase in the acquisition of homozygous mutations of FLT3 and CEBPA has frequency of leukemia, and a shift from myeloid/T-lymphoid been described in acute myeloid leukemia.84,85 CN-LOH in ALL to pre-B phenotype. These data provide compelling evidence is common, and most frequently involves chromosome 9p,38,46 that genetic alterations resulting in a block in B-lymphoid suggesting that this region may harbor a mutated or methylated development are key events in the pathogenesis of B-ALL. tumor suppressor. However, high-resolution SNP array Furthermore, higher resolution SNP array studies of pediatric studies38,45 have shown that the majority of regions of CN- ALL have identified lesions targeting this pathway in two-third of LOH at 9p harbor a focal deletion of CDKN2A/B that has cases, indicating that these alterations are important in the been reduplicated. Thus, true CN-LOH, in which there is no majority of B-ALL cases.67 discernible CNA in the region of LOH, is uncommon. Moreover the ability to accurately distinguish CN-LOH from inherited regions of homozygosity is dependent on the availability of Genomic analysis of T-lineage ALL matched constitutional samples from the same individuals. Algorithms for inferring LOH from pools of unpaired reference T-lineage ALL is also characterized by a number of sentinel samples are available, but inevitably are a compromise.86 chromosomal translocations, which commonly juxtapose the regulatory regions of T-cell antigen-receptor gene loci to oncogenes, and are associated with distinct gene-expression Genetic determinants of high-risk ALL signatures.11,73 Genomic profiling of CNA has been equally revealing in T-ALL. BAC-array analyses have identified ampli- The data described above indicate that genomic alterations are fication of ABL1, which led to identification of the NUP214- critical in establishing the leukemic clone present at diagnosis. ABL1 fusion that results in constitutive activation of ABL1.74 We Can similar studies identify genetic alterations determining the and others have used SNP, BAC or oligo-array CGH platforms to risk of treatment failure and relapse, and potential new targets identify deletions resulting to dysregulated expression of TAL138 for therapy? Although relapse is most common in BCR-ABL1- and LMO2,75 deletion of RB1,38 deletion and mutation of positive and MLL-rearranged ALL, relapse is observed across the PTEN,38,76,77 deletion or mutation of the U3 ubiquitin ligase spectrum of ALL subtypes, particularly in adult patients, and the FBXW7,76,78 amplification of MYB38,79,80 and fusion of SET to biologic determinants of relapse are poorly understood. Several NUP214.81 T-ALL thus exhibits the similar genetic complexity to recent SNP array studies have also shown that genetic that observed in B-ALL. Studies integrating analysis of CNA, alterations are important in determining the phenotype and risk sequence mutations (for example, NOTCH118 and PTEN 77) will of relapse in high-risk ALL. be important to accurately classify T-ALL and examine associations with treatment outcome. Genomics of BCR-ABL1 leukemia

Mechanism of CNA generation in ALL Expression of the constitutively active tyrosine kinase BCR-ABL1 is the hallmark of two diseases with distinct clinical manifesta- The high resolution of current SNP arrays has enabled PCR tions and therapeutic responsiveness: chronic myeloid leukemia mapping and sequencing of the genomic breakpoints at sites of (CML), a myeloproliferative disease that typically responds well recurring CNA, and this has provided insights into the to kinase inhibition,87 and BCR-ABL1-positive ALL that is highly mechanisms of deletion in ALL. For example, the proximal aggressive and has a poor prognosis.88 This dichotomy has been and distal breakpoints of the most common IKZF1 deletion attributed, at least in part, to the nature of the BCR-ABL1 fusion (D3–6) in B-ALL vary by only a few nucleotides, and heptamer transcript (p210 or p190/p185),89 or the leukemia initiating cell and partially conserved nonamer recombinase activating gene in which the translocation arises.90,91 The role of cooperating (RAG) recognition sequences are present immediately internal genetic lesions in determining the lineage of BCR-ABL1 to the deletion breakpoints.45 This suggests that the IKZF1 leukemia, and the nature of genetic lesions in influencing the deletions arise from the aberrant RAG activity,82 and this is transformation of chronic phase (CP) CML to acute leukemia further supported by the presence of extra nonconsensus (blast crisis), which may be of myeloid or lymphoid lineage, are nucleotides between the breakpoints, compatible with the incompletely understood.92 To examine the genomic basis of SPOTLIGHT action of terminal deoxynucleotidyl transferase. This has lineage and disease progression of BCR-ABL1 leukemia, we previously been described for CDKN2A/B deletions in ALL,83 expanded our previous studies to examine 43 de novo BCR- and we have observed this pattern at the sites of other CNA in ABL1-positive ALL cases (21 pediatric and 22 adult), and 128 B-ALL involving PAX5, IKZF2 and IKZF3, ERG, C20orf94, BTLA/ samples from 90 patients with CML including 64 obtained at CP CD200, and ADD3 (unpublished observations). This suggests (N ¼ 64), 15 at accelerated phase (AP, N ¼ 15), 32 at blast crisis that these deletions arise in a lymphoid progenitor at a stage of (22 myeloid and 9 lymphoid) and germline samples obtained at differentiation during which RAG1/2 are active, and that the disease remission.45,93 Importantly, 25 CML cases had sequen- deletions are subsequently selected for and contribute to tial samples obtained at CP/AP and blast crisis, enabling leukemogenesis. identification of lesions acquired during disease progression. A striking finding was deletion of IKZF1, encoding the early lymphoid transcription factor IKAROS,94 in 36 of 43 (84%) of de Copy-neutral loss-of heterozygosity (acquired uniparental novo BCR-ABL1 ALL cases.45 Several other recurring deletions disomy) in ALL were also identified in BCR-ABL1 ALL, albeit at lower frequency, including CDKN2A/B, PAX5, C20orf94, RB1, MEF2C CN-LOH may indicate duplication of a gene altered by point and EBF1. Similar to deletions of PAX5, the IKZF1 deletions mutations or epigenetic modification within the region of LOH. were usually hemizygous and deleted either the entire IKZF1 Consequently, detection of CN-LOH in leukemia is an area of locus or a subset of exons, most commonly coding exons 3–6,

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1214 resulting in the expression of internally truncated IKZF1 important determinant of the biology and poor outcome of both transcripts. IKAROS is a transcription factor with N-terminal BCR-ABL1-positive and -negative ALL, and also raises the zinc fingers that mediate DNA binding, and C-terminal zinc possibility that these BCR-ABL1-negative cases harbor addi- fingers that mediate dimerization with other IKAROS family tional mutations resulting in kinase activation. members.95 IKAROS isoforms that lack the N-terminal, DNA- If genetic alterations detectable in the predominant clone at binding zinc fingers, and act as dominant negative isoforms had diagnosis influence risk of relapse, what new lesions are previously been identified in ALL,96–102 and had been attributed acquired during progression from diagnosis to relapse? to aberrant posttranscriptional splicing.103 However, we ob- To address this, we performed SNP array analysis of 61 matched served absolute concordance between the extent of IKZF1 diagnosis and relapse B-progenitor (N ¼ 47) and T-lineage deletion and the extent of truncation of IKAROS isoforms on (N ¼ 14) samples.104 Over 92% of cases exhibited differences reverse transcription–PCR and western blotting analysis, in the pattern of CNA between diagnosis and relapse. Analysis of demonstrating that the expression of aberrant IKAROS isoforms the pattern of CNA at antigen-receptor loci demonstrated that in ALL is driven by genetic alterations. most relapse samples were clonally related to the diagnosis Furthermore, the pattern of CNA acquired during the samples, rather than being distinct second leukemias. Remark- progression of CP-CML to lymphoid blast crisis was highly ably, although 34% of cases showed simple clonal evolution similar to de novo BCR-ABL1 ALL. Seven of nine lymphoid blast (that is, preservation of CNA present at diagnosis and acquisition crisis CML cases harbored alterations in genes regulating of new lesions at relapse), over 50% of samples acquired new B-lymphoid development, including IKZF1 (N ¼ 6; two cases lesions and also lost CNA present at diagnosis, yet demonstrated having homozygous deletions and one case a deletion and a common clonal origin. Furthermore, CNA-specific back- SPOTLIGHT frameshift mutation), PAX5 (N ¼ 4) and EBF1 (one case with a tracking using genomic PCR assays showed that in the majority homozygous deletion). CDKN2A/B was deleted in 6 of 9 of cases the relapse clone was present at low levels at diagnosis. lymphoid, but only 2 of 22 myeloid blast crisis cases. Other This indicates that the diagnosis and relapse clones originate CNA observed in de novo BCR-ABL1 ALL were also detected in from a common, prediagnosis ancestral clone, and acquire lymphoid blast crisis CML, including deletions of MEF2C, distinct CNA before emerging as the predominant clones at C20orf94 and HBS1L. These data indicate that cooperating diagnosis or relapse. Pathway analysis of the many genes genomic alterations are critical determinants of the lineage of involved by relapse-acquired CNA demonstrated tumor sup- BCR-ABL1 leukemia and progression of CML to blast crisis. pression (for example, CDKN2A/B) and lymphoid development Studies directly examining the role of IKZF1 alteration and other (for example, IKZF1) to be most frequently involved. Thus, the common lesions (PAX5, CDKN2A/B) in the pathogenesis and genetic basis relapse appears more complex than simple drug treatment resistance of BCR-ABL1 lymphoid leukemia will be of resistance, and further analyses examining additional types of great interest. genetic alteration in this context, such as sequence mutation, will be important to identify potential pathways for therapeutic intervention. Genetic determinants of relapse in ALL

Although treatment failure is more common in subtypes such as Future directions BCR-ABL1, MLL-rearranged and hypodiploid ALL, relapse is observed across the spectrum of ALL, and carries a poor The studies described above have identified a plethora of prognosis. A logical extension of the studies of diagnosis ALL hitherto unsuspected genetic alterations in ALL, and have samples described above was to perform high-resolution transformed our understanding of the genetic basis of this genomic profiling of relapsed ALL, to identify genetic lesions disease. However, much remains to be learned. Microarray associated with treatment failure. Two recent complementary platforms for the detection of CNA have evolved rapidly and studies have provided important insights into the genetic basis of current platforms contain up to 2 million probes that interrogate relapse. In a study with the Children’s Oncology Group as part the genome at sub-kilobase resolutionFa 200-fold increase of the National Cancer Institute TARGET (Therapeutically from the first SNP arrays used ALL.43 It is important to note that Applicable Research to Generate Effective Treatments) project, the coverage of many genes on previous SNP arrays was we used performed Affymetrix SNP array analysis of CNA/LOH, suboptimal (notably IKZF1), and we have observed an increase gene-expression profiling and candidate gene resequencing of a in lesion frequency with each increase in array resolution.105 cohort of 221 childhood B-progenitor ALL cases predicted to be Thus the current estimates of lesions frequency in ALL (and at high risk of relapse.67 This cohort excluded BCR-ABL1 and indeed all tumors) remain a ‘lower limit’, and tiling analysis of hypodiploid ALL, and most cases (77%) lacked a cytogenetic CNA is required to accurately identify all lesions. Moreover, alteration. although we now have a detailed understanding of the texture of Genome-wide analysis of associations between CNA and genetic alterations in diagnostic ALL samples, studies that outcome demonstrated a highly significant association between examine specific subtypes of high-risk ALL, and cases lacking alterations of IKZF1 and relapse. Patients harboring an IKZF1 sentinel chromosomal abnormalities are required to identify deletion or sequence mutation had a 5-year incidence of relapse novel leukemogenic lesions, and potential targets for therapy. of 73.4% compared to 25% for those without an alteration. Relapse is particularly problematic in adult ALL, which has a Notably, this association was confirmed in an independent lower frequency of known chromosomal alterations than cohort of patients treated at St Jude, and remained significant in pediatric ALL.106 CNA profiling data in adult ALL is lim- multivariable analyses incorporating minimal residual levels ited,68,107 and larger studies of this patient group are needed. early in therapy, suggesting that detection of IKZF1 alteration Existing studies have performed genome-wide analysis of only will be useful in risk stratification. An additional finding was that a single class of genetic alterationFcopy number changes- the gene-expression profile of poor outcome, IKZF1-mutated Fand complementary genome-wide analyses of sequence (but BCR-ABL1-negative) cases in this study was highly similar mutations and epigenetic changes in ALL are required.108 This to that of BCR-ABL1-positive ALL. This suggests that IKZF1 is an is not a trivial undertaking, but with the rapid evolution of

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1215 next-generation sequencing technology,109 interrogation of 10 Raimondi SC. Cytogenetics of acute leukemias. In: Pui CH (ed). DNA regions or whole genomes in cancer is now possible. Childhood Leukemias, 2nd edn. Cambridge University Press: Indeed, genome-wide analysis of coding mutations has been Cambridge, 2006, pp 235–271. recently described in acute myeloid leukemia.110 A potential 11 Graux C, Cools J, Michaux L, Vandenberghe P, Hagemeijer A. Cytogenetics and molecular genetics of T-cell acute lympho- benefit of next-generation sequencing approaches is the ability blastic leukemia: from thymocyte to lymphoblast. Leukemia to detect copy number variation, sequence variation and 2006; 20: 1496–1510. structural rearrangements of cancer genomes simultaneously 12 Harrison CJ. Cytogenetics of paediatric and adolescent acute and in a truly genome-wide fashion.111 As next-generation lymphoblastic leukaemia. Br J Haematol 2009; 144: 147–156. sequencing platforms can also perform transcriptional profil- 13 Andreasson P, Schwaller J, Anastasiadou E, Aster J, Gilliland DG. ing112 it is likely that they will eventually supplant microarray The expression of ETV6/CBFA2 (TEL/AML1) is not sufficient for 113 the transformation of hematopoietic cell lines in vitro or the methodologies, however the cost of analyzing large numbers induction of hematologic disease in vivo. Cancer Genet of tumor genomes is currently prohibitive. Integrating massive, Cytogenet 2001; 130: 93–104. complex datasets from multiple complementary genomic plat- 14 Williams RT, Roussel MF, Sherr CJ. Arf gene loss enhances forms (for example, CNA, expression, microRNA and methyla- oncogenicity and limits imatinib response in mouse models of tion) will be a formidable challenge, but necessary to examine Bcr-Abl-induced acute lymphoblastic leukemia. Proc Natl Acad the interaction of different genetic alterations in leukemogenesis Sci USA 2006; 103: 6688–6693. 15 Hebert J, Cayuela JM, Berkeley J, Sigaux F. Candidate tumor- and response to therapy. However, recent studies have made suppressor genes MTS1 (p16INK4A) and MTS2 (p15INK4B) important advances into indentifying new lesions that contribute display frequent homozygous deletions in primary cells from T- to establishment of the leukemic clone, and that influence but not from B-cell lineage acute lymphoblastic leukemias. Blood therapeutic response. Ultimately, we can expect that these new 1994; 84: 4038–4044. findings will be translated into novel, less toxic targeted 16 Ogawa S, Hangaishi A, Miyawaki S, Hirosawa S, Miura Y, therapies for ALL. Takeyama K et al. Loss of the cyclin-dependent kinase 4-inhibitor (p16; MTS1) gene is frequent in and highly specific to lymphoid tumors in primary human hematopoietic malignancies. Blood Acknowledgements 1995; 86: 1548–1556. 17 Okuda T, Shurtleff SA, Valentine MB, Raimondi SC, Head DR, Behm F et al. Frequent deletion of p16INK4a/MTS1 and Supported in part by grants from the Haematology Society of p15INK4b/MTS2 in pediatric acute lymphoblastic leukemia. Australasia, the Royal Australasian College of Physicians, the Blood 1995; 85: 2321–2330. National Health and Medical Research Council (Australia) and the 18 Weng AP, Ferrando AA, Lee W, Morris JPt, Silverman LB, American Lebanese and Syrian Associated Charities of St Jude Sanchez-Irizarry C et al. Activating mutations of NOTCH1 in Children’s Research Hospital. human T cell acute lymphoblastic leukemia. Science 2004; 306: 269–271. 19 Davies JJ, Wilson IM, Lam WL. Array CGH technologies and References their applications to cancer genomes. Chromosome Res 2005; 13: 237–248. 1 Pui CH, Robison LL, Look AT. Acute lymphoblastic leukaemia. 20 Wang Y, Armstrong SA. Genome-wide SNP analysis in cancer: Lancet 2008; 371: 1030–1043. leukemia shows the way. Cancer Cell 2007; 11: 308–309. 2 Pui CH, Sandlund JT, Pei D, Campana D, Rivera GK, Ribeiro RC 21 Pinkel D, Albertson DG. Comparative genomic hybridization. et al. Improved outcome for children with acute lymphoblastic Annu Rev Genomics Hum Genet 2005; 6: 331–354. leukemia: results of Total Therapy Study XIIIB at St Jude 22 Larramendy ML, Huhta T, Heinonen K, Vettenranta K, Mahla- Children’s Research Hospital. Blood 2004; 104: 2690–2696. maki E, Riikonen P et al. DNA copy number changes in 3 Rowe JM, Buck G, Burnett AK, Chopra R, Wiernik PH, Richards childhood acute lymphoblastic leukemia. Haematologica 1998; SM et al. Induction therapy for adults with acute lymphoblastic 83: 890–895. leukemia: results of more than 1500 patients from the interna- 23 Larramendy ML, Huhta T, Vettenranta K, El-Rifai W, Lundin J, tional ALL trial: MRC UKALL XII/ECOG E2993. Blood 2005; 106: Pakkala S et al. Comparative genomic hybridization in 3760–3767. childhood acute lymphoblastic leukemia. Leukemia 1998; 12: 4 Gleissner B, Gokbuget N, Bartram CR, Janssen B, Rieder H, 1638–1644. Janssen JW et al. Leading prognostic relevance of the BCR-ABL 24 Huhta T, Vettenranta K, Heinonen K, Kanerva J, Larramendy ML, translocation in adult acute B-lineage lymphoblastic leukemia: a Mahlamaki E et al. Comparative genomic hybridization and prospective study of the German Multicenter Trial Group and conventional cytogenetic analyses in childhood acute myeloid confirmed polymerase chain reaction analysis. Blood 2002; 99: leukemia. Leuk Lymphoma 1999; 35: 311–315. SPOTLIGHT 1536–1543. 25 Schoumans J, Johansson B, Corcoran M, Kuchinskaya E, 5 Fielding AK, Richards SM, Chopra R, Lazarus HM, Litzow MR, Golovleva I, Grander D et al. Characterisation of dic(9;20)(p11- Buck G et al. Outcome of 609 adults after relapse of acute 13;q11) in childhood B-cell precursor acute lymphoblastic lymphoblastic leukemia (ALL); an MRC UKALL12/ECOG 2993 leukaemia by tiling resolution array-based comparative genomic study. Blood 2007; 109: 944–950. hybridisation reveals clustered breakpoints at 9p13.2 and 6 Einsiedel HG, von Stackelberg A, Hartmann R, Fengler R, 20q11.2. Br J Haematol 2006; 135: 492–499. Schrappe M, Janka-Schaub G et al. Long-term outcome in 26 Strefford JC, van Delft FW, Robinson HM, Worley H, Yianni- children with relapsed ALL by risk-stratified salvage therapy: kouris O, Selzer R et al. Complex genomic alterations and gene results of trial acute lymphoblastic leukemia-relapse study of expression in acute lymphoblastic leukemia with intrachromoso- the Berlin-Frankfurt-Munster Group 87. J Clin Oncol 2005; 23: mal amplification of chromosome 21. Proc Natl Acad Sci USA 7942–7950. 2006; 103: 8167–8172. 7 Rivera GK, Zhou Y, Hancock ML, Gajjar A, Rubnitz J, Ribeiro RC 27 van Vlierberghe P, Meijerink JP, Lee C, Ferrando AA, Look AT, et al. Bone marrow recurrence after initial intensive treatment van Wering ER et al. A new recurrent 9q34 duplication in for childhood acute lymphoblastic leukemia. Cancer 2005; 103: pediatric T-cell acute lymphoblastic leukemia. Leukemia 2006; 368–376. 20: 1245–1253. 8 Harrison CJ, Foroni L. Cytogenetics and molecular genetics 28 Davidsson J, Andersson A, Paulsson K, Heidenblad M, Isaksson of acute lymphoblastic leukemia. Rev Clin Exp Hematol 2002; 6: M, Borg A et al. Tiling resolution array comparative genomic 91–113. hybridization, expression and methylation analyses of dup(1q) 9 Pui CH, Relling MV, Downing JR. Acute lymphoblastic leukemia. in Burkitt lymphomas and pediatric high hyperdiploid acute N Engl J Med 2004; 350: 1535–1548. lymphoblastic leukemias reveal clustered near-centromeric

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1216 breakpoints and overexpression of genes in 1q22-32.3. Hum Mol profiling of childhood ALL reveals novel recurrent genetic lesions Genet 2007; 16: 2215–2225. affecting pathways involved in lymphocyte differentiation and 29 Kuchinskaya E, Nordgren A, Heyman M, Schoumans J, Corcoran cell cycle progression. Leukemia 2007; 21: 1258–1266. M, Staaf J et al. Tiling-resolution array-CGH reveals the pattern 45 Mullighan CG, Miller CB, Phillips LA, Dalton JD, Ma J, Radtke I of DNA copy number alterations in acute lymphoblastic et al. BCR-ABL1 lymphoblastic leukaemia is characterized by the leukemia with 21q amplification: the result of telomere dysfunc- deletion of Ikaros. Nature 2008; 453: 110–114. tion and breakage/fusion/breakage cycles? Leukemia 2007; 21: 46 Kawamata N, Ogawa S, Zimmermann M, Kato M, Sanada M, 1327–1330. Hemminki K et al. Molecular allelokaryotyping of pediatric acute 30 Lundin C, Heidenblad M, Strombeck B, Borg A, Hovland R, Heim lymphoblastic leukemias by high-resolution single nucleotide S et al. Tiling resolution array CGH of dic(7;9)(p11 approximately polymorphism oligonucleotide genomic microarray. Blood 2008; 13;p11 approximately 13) in B-cell precursor acute lympho- 111: 776–784. blastic leukemia reveals clustered breakpoints at 7p11.2 approxi- 47 Kawamata N, Ogawa S, Zimmermann M, Niebuhr B, Stocking C, mately 12.1 and 9p13.1. Cytogenet Genome Res 2007; 118: Sanada M et al. Cloning of genes involved in chromo- 13–18. somal translocations by high-resolution single nucleotide 31 Strefford JC, Worley H, Barber K, Wright S, Stewart AR, Robinson polymorphism genomic microarray. Proc Natl Acad Sci USA HM et al. Genome complexity in acute lymphoblastic leukemia is 2008; 105: 11921–11926. revealed by array-based comparative genomic hybridization. 48 Bungaro S, Dell’Orto MC, Zangrando A, Basso D, Gorletta T, Lo Oncogene 2007; 26: 4306–4318. Nigro L et al. Integration of genomic and gene expression data of 32 Jalali GR, An Q, Konn ZJ, Worley H, Wright SL, Harrison CJ et al. childhood ALL without known aberrations identifies subgroups Disruption of ETV6 in intron 2 results in upregulatory and with specific genetic hallmarks. Genes Chromosomes Cancer insertional events in childhood acute lymphoblastic leukaemia. 2009; 48: 22–38.

SPOTLIGHT Leukemia 2008; 22: 114–123. 49 Pounds SB, Cheng C, Mullighan CG, Raimondi SC, Shurtleff SA, 33 Kuchinskaya E, Heyman M, Nordgren A, Schoumans J, Staaf J, Downing JR. Reference alignment of SNP microarray signals Borg A et al. Array-CGH reveals hidden gene dose changes in for copy number analysis of tumors. Bioinformatics 2009; 25: children with acute lymphoblastic leukaemia and a normal 315–321. or failed karyotype by G-banding. Br J Haematol 2008; 140: 50 Olshen AB, Venkatraman ES, Lucito R, Wigler M. Circular binary 572–577. segmentation for the analysis of array-based DNA copy number 34 Rabin KR, Man TK, Yu A, Folsom MR, Zhao YJ, Rao PH et al. data. Biostatistics 2004; 5: 557–572. Clinical utility of array comparative genomic hybridization 51 Greaves MF, Wiemels J. Origins of chromosome translocations in for detection of chromosomal abnormalities in pediatric childhood leukaemia. Nat Rev Cancer 2003; 3: 639–649. acute lymphoblastic leukemia. Pediatr Blood Cancer 2008; 51: 52 Nutt SL, Heavey B, Rolink AG, Busslinger M. Commitment to the 171–177. B-lymphoid lineage depends on the transcription factor Pax5. 35 Steinemann D, Cario G, Stanulla M, Karawajew L, Tauscher M, Nature 1999; 401: 556–562. Weigmann A et al. Copy number alterations in childhood acute 53 Nutt SL, Eberhard D, Horcher M, Rolink AG, Busslinger M. Pax5 lymphoblastic leukemia and their association with minimal determines the identity of B cells from the beginning to the end of residual disease. Genes Chromosomes Cancer 2008; 47: 471–480. B-lymphopoiesis. Int Rev Immunol 2001; 20: 65–82. 36 Hehir-Kwa JY, Egmont-Petersen M, Janssen IM, Smeets D, van 54 Busslinger M. Transcriptional control of early B cell development. Kessel AG, Veltman JA. Genome-wide copy number profiling on Annu Rev Immunol 2004; 22: 55–79. high-density bacterial artificial chromosomes, single-nucleotide 55 Nutt SL, Kee BL. The transcriptional regulation of B cell lineage polymorphisms, and oligonucleotide microarrays: a platform commitment. Immunity 2007; 26: 715–725. comparison based on statistical power analysis. DNA Res 2007; 56 Cazzaniga G, Daniotti M, Tosi S, Giudici G, Aloisi A, Pogliani E 14: 1–11. et al. The paired box domain gene PAX5 is fused to ETV6/TEL in 37 Lo KC, Bailey D, Burkhardt T, Gardina P, Turpaz Y, Cowell JK. an acute lymphoblastic leukemia case. Cancer Res 2001; 61: Comprehensive analysis of loss of heterozygosity events in 4666–4670. glioblastoma using the 100 K SNP mapping arrays and compar- 57 Strehl S, Konig M, Dworzak MN, Kalwak K, Haas OA. PAX5/ ison with copy number abnormalities defined by BAC array ETV6 fusion defines cytogenetic entity dic(9;12)(p13;p13). comparative genomic hybridization. Genes Chromosomes Can- Leukemia 2003; 17: 1121–1123. cer 2008; 47: 221–237. 58 Warming S, Liu P, Suzuki T, Akagi K, Lindtner S, Pavlakis GN 38 Mullighan CG, Goorha S, Radtke I, Miller CB, Coustan-Smith E, et al. Evi3, a common retroviral integration site in murine B-cell Dalton JD et al. Genome-wide analysis of genetic alterations in lymphoma, encodes an EBFAZ-related Kruppel-like zinc finger acute lymphoblastic leukaemia. Nature 2007; 446: 758–764. protein. Blood 2003; 101: 1934–1940. 39 Xavier RJ, Rioux JD. Genome-wide association studies: a new 59 Bond HM, Mesuraca M, Carbone E, Bonelli P, Agosti V, window into immune-mediated diseases. Nat Rev Immunol Amodio N et al. Early hematopoietic zinc finger protein 2008; 8: 631–643. (EHZF), the human homolog to mouse Evi3, is highly expressed 40 Matsuzaki H, Dong S, Loi H, Di X, Liu G, Hubbell E et al. in primitive human hematopoietic cells. Blood 2004; 103: Genotyping over 100 000 SNPs on a pair of oligonucleotide 2062–2070. arrays. Nat Methods 2004; 1: 109–111. 60 Bohlander SK. ETV6: a versatile player in leukemogenesis. Semin 41 Peiffer DA, Le JM, Steemers FJ, Chang W, Jenniges T, Garcia F Cancer Biol 2005; 15: 162–174. et al. High-resolution genomic profiling of chromosomal aberra- 61 Wlodarska I, Veyt E, De Paepe P, Vandenberghe P, Nooijen P, tions using Infinium whole-genome genotyping. Genome Res Theate I et al. FOXP1, a gene highly expressed in a subset of 2006; 16: 1136–1148. diffuse large B-cell lymphoma, is recurrently targeted by genomic 42 Lilljebjorn H, Heidenblad M, Nilsson B, Lassen C, Horvat A, aberrations. Leukemia 2005; 19: 1299–1305. Heldrup J et al. Combined high-resolution array-based compara- 62 Hu H, Wang B, Borde M, Nardone J, Maika S, Allred L et al. tive genomic hybridization and expression profiling of ETV6/ Foxp1 is an essential transcriptional regulator of B cell develop- RUNX1-positive acute lymphoblastic leukemias reveal a high ment. Nat Immunol 2006; 7: 819–826. incidence of cryptic Xq duplications and identify several putative 63 Nebral K, Konig M, Harder L, Siebert R, Haas OA, Strehl S. target genes within the commonly gained region. Leukemia 2007; Identification of PML as novel PAX5 fusion partner in childhood 21: 2137–2144. acute lymphoblastic leukaemia. Br J Haematol 2007; 139: 43 Irving JA, Bloodworth L, Bown NP, Case MC, Hogarth LA, 269–274. Hall AG. Loss of heterozygosity in childhood acute lympho- 64 Bousquet M, Broccardo C, Quelen C, Meggetto F, Kuhlein E, blastic leukemia detected by genome-wide microarray single Delsol G et al. A novel PAX5-ELN fusion protein identified in nucleotide polymorphism analysis. Cancer Res 2005; 65: B-cell acute lymphoblastic leukemia acts as a dominant negative 3053–3058. on wild-type PAX5. Blood 2007; 109: 3417–3423. 44 Kuiper RP, Schoenmakers EF, van Reijmersdal SV, Hehir-Kwa JY, 65 An Q, Wright SL, Konn ZJ, Matheson E, Minto L, Moorman AV van Kessel AG, van Leeuwen FN et al. High-resolution genomic et al. Variable breakpoints target PAX5 in patients with

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1217 dicentric chromosomes: a model for the basis of unbalanced 85 Mullighan CG, Radtke I, Ma J, Shurtleff SA, Downing JR. High- translocations in cancer. Proc Natl Acad Sci USA 2008; 105: resolution SNP array profiling of relapsed acute leukemia 17050–17054. identifies genomic abnormalities distinct from those present at 66 Nebral K, Denk D, Attarbaschi A, Konig M, Mann G, Haas OA diagnosis [abstract]. Blood 2007; 110: 234. et al. Incidence and diversity of PAX5 fusion genes in childhood 86 Beroukhim R, Lin M, Park Y, Hao K, Zhao X, Garraway LA et al. acute lymphoblastic leukemia. Leukemia 2009; 23: 134–143. Inferring loss-of-heterozygosity from unpaired tumors using high- 67 Mullighan CG, Su X, Zhang J, Radtke I, Phillips LA, Miller CB density oligonucleotide SNP arrays. PLoS Comput Biol 2006; 2: e41. et al. Deletion of IKZF1 and Prognosis in Acute Lymphoblastic 87 Goldman JM, Melo JV. Chronic myeloid leukemiaFadvances in Leukemia. N Engl J Med 2009; 360: 470–480. biology and new approaches to treatment. N Engl J Med 2003; 68 Mullighan CG, Miller CB, Radtke I, Phillips LA, Dalton J, Ma J 349: 1451–1464. et al. BCR-ABL1 lymphoblastic leukaemia is characterized by the 88 Ribeiro RC, Abromowitch M, Raimondi SC, Murphy SB, Behm F, deletion of Ikaros. Nature 2008; 453: 110–114. Williams DL. Clinical and biologic hallmarks of the Philadelphia 69 Hu Y, Liu Y, Pelletier S, Buchdunger E, Warmuth M, Fabbro D chromosome in childhood acute lymphoblastic leukemia. Blood et al. Requirement of Src kinases Lyn, Hck and Fgr for BCR-ABL1- 1987; 70: 948–953. induced B-lymphoblastic leukemia but not chronic myeloid 89 Melo JV. The diversity of BCR-ABL fusion proteins and their leukemia. Nat Genet 2004; 36: 453–461. relationship to leukemia phenotype. Blood 1996; 88: 2375–2384. 70 Miller CB, Mullighan CG, Su X, Ma J, Wang M, Zhang J et al. 90 Huntly BJ, Shigematsu H, Deguchi K, Lee BH, Mizuno S, Duclos Pax5 haploinsufficiency cooperates with BCR-ABL1 to induce N et al. MOZ-TIF2, but not BCR-ABL, confers properties acute lymphoblastic leukemia [abstract]. Blood 2008; 112: p 114. of leukemic stem cells to committed murine hematopoietic 71 Urbanek P, Wang ZQ, Fetka I, Wagner EF, Busslinger M. progenitors. Cancer Cell 2004; 6: 587–596. Complete block of early B cell differentiation and altered 91 Savona M, Talpaz M. Getting to the stem of chronic myeloid patterning of the posterior midbrain in mice lacking Pax5/BSAP. leukaemia. Nat Rev Cancer 2008; 8: 341–350. Cell 1994; 79: 901–912. 92 Melo JV, Barnes DJ. Chronic myeloid leukaemia as a model 72 Dang J, Mullighan CG, Phillips LA, Mehta P, Downing JR. of disease evolution in human cancer. Nat Rev Cancer 2007; 7: Retroviral and chemical mutagenesis identifies Pax5 as a tumor 441–453. suppressor in B-progenitor acute lymphoblastic leukemia [ab- 93 Mullighan CG, Radtke I, Zhang J, Phillips LA, Su X, Ma J et al. stract]. Blood 2008; 112: 256–257. Genome-wide analysis of genetic alterations in chronic myeloid 73 Ferrando AA, Neuberg DS, Staunton J, Loh ML, Huard C, leukemia [abstract]. Blood 2008; 2008: 397–398. Raimondi SC et al. Gene expression signatures define novel 94 Reynaud D, Demarco IA, Reddy KL, Schjerven H, Bertolino E, oncogenic pathways in T cell acute lymphoblastic leukemia. Chen Z et al. Regulation of B cell fate commitment and Cancer Cell 2002; 1: 75–87. immunoglobulin heavy-chain gene rearrangements by Ikaros. 74 Graux C, Cools J, Melotte C, Quentmeier H, Ferrando A, Levine R Nat Immunol 2008; 9: 927–936. et al. Fusion of NUP214 to ABL1 on amplified episomes in 95 Molnar A, Georgopoulos K. The Ikaros gene encodes a family of T-cell acute lymphoblastic leukemia. Nat Genet 2004; 36: functionally diverse zinc finger DNA-binding proteins. Mol Cell 1084–1089. Biol 1994; 14: 8292–8303. 75 Van Vlierberghe P, van Grotel M, Beverloo HB, Lee C, Helgason 96 Sun L, Crotty ML, Sensel M, Sather H, Navara C, Nachman J et al. T, Buijs-Gladdines J et al. The cryptic chromosomal deletion Expression of dominant-negative Ikaros isoforms in T-cell acute del(11)(p12p13) as a new activation mechanism of LMO2 in lymphoblastic leukemia. Clin Cancer Res 1999; 5: 2112–2120. pediatric T-cell acute lymphoblastic leukemia. Blood 2006; 108: 97 Sun L, Goodman PA, Wood CM, Crotty ML, Sensel M, Sather H 3520–3529. et al. Expression of aberrantly spliced oncogenic ikaros isoforms 76 Maser RS, Choudhury B, Campbell PJ, Feng B, Wong KK, in childhood acute lymphoblastic leukemia. J Clin Oncol 1999; Protopopov A et al. Chromosomally unstable mouse tumours 17: 3753–3766. have genomic alterations similar to diverse human cancers. 98 Sun L, Heerema N, Crotty L, Wu X, Navara C, Vassilev A Nature 2007; 447: 966–971. et al. Expression of dominant-negative and mutant isoforms 77 Palomero T, Sulis ML, Cortina M, Real PJ, Barnes K, Ciofani M of the antileukemic transcription factor Ikaros in infant et al. Mutational loss of PTEN induces resistance to NOTCH1 acute lymphoblastic leukemia. Proc Natl Acad Sci USA 1999; inhibition in T-cell leukemia. Nat Med 2007; 13: 1203–1210. 96: 680–685. 78 O’Neil J, Grim J, Strack P, Rao S, Tibbitts D, Winter C et al. FBW7 99 Nakase K, Ishimaru F, Avitahl N, Dansako H, Matsuo K, Fujii K mutations in leukemic cells mediate NOTCH pathway activation et al. Dominant negative isoform of the Ikaros gene in patients and resistance to gamma-secretase inhibitors. J Exp Med 2007; with adult B-cell acute lymphoblastic leukemia. Cancer Res 204: 1813–1824. 2000; 60: 4062–4065. 79 Clappier E, Cuccuini W, Kalota A, Crinquette A, Cayuela JM, Dik 100 Nishii K, Kita K, Miwa H, Shikami M, Taniguchi M, Usui E et al. WA et al. The C-MYB locus is involved in chromosomal Expression of B cell-associated transcription factors in B-cell translocation and genomic duplications in human T-cell acute precursor acute lymphoblastic leukemia cells: association

leukemia (T-ALL), the translocation deﬁning a new T-ALL subtype with PU.1 expression, phenotype, and immunogenotype. SPOTLIGHT in very young children. Blood 2007; 110: 1251–1261. Int J Hematol 2000; 71: 372–378. 80 Lahortiga I, De Keersmaecker K, Van Vlierberghe P, Graux C, 101 Olivero S, Maroc C, Beillard E, Gabert J, Nietfeld W, Chabannon Cauwelier B, Lambert F et al. Duplication of the MYB oncogene C et al. Detection of different Ikaros isoforms in human in T cell acute lymphoblastic leukemia. Nat Genet 2007; 39: leukaemias using real-time quantitative polymerase chain reac- 593–595. tion. Br J Haematol 2000; 110: 826–830. 81 Van Vlierberghe P, van Grotel M, Tchinda J, Lee C, Beverloo HB, 102 Takanashi M, Yagi T, Imamura T, Tabata Y, Morimoto A, van der Spek PJ et al. The recurrent SET-NUP214 fusion as a new Hibi S et al. Expression of the Ikaros gene family in childhood HOXA activation mechanism in pediatric T-cell acute lympho- acute lymphoblastic leukaemia. Br J Haematol 2002; 117: blastic leukemia. Blood 2008; 111: 4668–4680. 525–530. 82 Fugmann SD, Lee AI, Shockett PE, Villey IJ, Schatz DG. The RAG 103 Klein F, Feldhahn N, Herzog S, Sprangers M, Mooster JL, Jumaa H proteins and V(D)J recombination: complexes, ends, and et al. BCR-ABL1 induces aberrant splicing of IKAROS and lineage transposition. Annu Rev Immunol 2000; 18: 495–527. inﬁdelity in pre-B lymphoblastic leukemia cells. Oncogene 2006; 83 Kitagawa Y, Inoue K, Sasaki S, Hayashi Y, Matsuo Y, Lieber MR 25: 1118–1124. et al. Prevalent involvement of illegitimate V(D)J recombination 104 Mullighan CG, Phillips LA, Su X, Ma J, Miller CB, Shurtleff SA in chromosome 9p21 deletions in lymphoid leukemia. J Biol et al. Genomic analysis of the clonal origins of relapsed acute Chem 2002; 277: 46289–46297. lymphoblastic leukemia. Science 2008; 322: 1377–1380. 84 Raghavan M, Smith LL, Lillington DM, Chaplin T, Kakkas I, 105 Mullighan CG, Phillips LA, Kiedrowski LA, Ma J, Williams RT, Molloy G et al. Segmental uniparental disomy is a commonly Shurtleff SA et al. Ultra-high resolution genomic analysis of acquired genetic event in relapsed acute myeloid leukemia. genomic alterations in high-risk acute lymphoblastic leukemia Blood 2008; 112: 814–821. [abstract]. Blood 2008; 112: 715.

Leukemia Genomic analysis of ALL CG Mullighan and JR Downing 1218 106 Moorman AV, Harrison CJ, Buck GA, Richards SM, 109 Shendure J, Ji H. Next-generation DNA sequencing. Nat Secker-Walker LM, Martineau M et al. Karyotype is an Biotechnol 2008; 26: 1135–1145. independent prognostic factor in adult acute lymphoblastic 110 Ley TJ, Mardis ER, Ding L, Fulton B, McLellan MD, Chen K et al. leukemia (ALL): analysis of cytogenetic data from patients treated DNA sequencing of a cytogenetically normal acute myeloid on the Medical Research Council (MRC) UKALLXII/Eastern leukaemia genome. Nature 2008; 456: 66–72. Cooperative Oncology Group (ECOG) 2993 trial. Blood 2007; 111 Campbell PJ, Stephens PJ, Pleasance ED, O’Meara S, Li H, 109: 3189–3197. Santarius T et al. Identiﬁcation of somatically acquired rearrange- 107 Paulsson K, Cazier JB, Macdougall F, Stevens J, Stasevich I, Vrcelj ments in cancer using genome-wide massively parallel N et al. Microdeletions are a general feature of adult and paired-end sequencing. Nat Genet 2008; 40: 722–729. adolescent acute lymphoblastic leukemia: Unexpected 112 Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. similarities with pediatric disease. Proc Natl Acad Sci USA Mapping and quantifying mammalian transcriptomes by 2008; 105: 6708–6713. RNA-Seq. Nat Methods 2008; 5: 621–628. 108 Esteller M. Cancer epigenomics: DNA methylomes and 113 Shendure J. The beginning of the end for microarrays? Nat histone-modiﬁcation maps. Nat Rev Genet 2007; 8: 286–298. Methods 2008; 5: 585–587. SPOTLIGHT

Leukemia