Advertisement

Pediatric leukemia: Moving toward more accurate models

  • Thomas Milan
    Affiliations
    Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, QC, Canada
    Search for articles by this author
  • Author Footnotes
    1 Current address: The Chan Zuckerberg Biohub, 499 Illinois Street, San Francisco, CA 94158.
    Hera Canaj
    Footnotes
    1 Current address: The Chan Zuckerberg Biohub, 499 Illinois Street, San Francisco, CA 94158.
    Affiliations
    Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, QC, Canada
    Search for articles by this author
  • Chloe Villeneuve
    Affiliations
    Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, QC, Canada
    Search for articles by this author
  • Author Footnotes
    2 Current address: Electrical Engineering and Computer Sciences, University of California, Berkeley, CA 94720-1770.
    Aditi Ghosh
    Footnotes
    2 Current address: Electrical Engineering and Computer Sciences, University of California, Berkeley, CA 94720-1770.
    Affiliations
    Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, QC, Canada
    Search for articles by this author
  • Frédéric Barabé
    Affiliations
    Centre de recherche en infectiologie du CHUL, Centre de recherche du CHU de Québec, Quebec City, QC, Canada

    CHU de Québec Hôpital Enfant-Jésus, Quebec City, QC, Canada

    Department of Medicine, Université Laval, Quebec City, QC, Canada
    Search for articles by this author
  • Sonia Cellot
    Affiliations
    Division of Hematology, Department of Pediatrics, Ste-Justine Hospital, Montréal, Université de Montréal, Montréal, QC, Canada
    Search for articles by this author
  • Brian T. Wilhelm
    Correspondence
    Offprint requests to: Brian T. Wilhelm, Institute for Research in Immunology and Cancer, Université de Montréal, PO Box 6128, Station Centre-Ville, Montreal, QC H3C 3J7, Canada
    Affiliations
    Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, QC, Canada

    Department of Medicine, Université de Montréal, Montréal, QC, Canada
    Search for articles by this author
  • Author Footnotes
    1 Current address: The Chan Zuckerberg Biohub, 499 Illinois Street, San Francisco, CA 94158.
    2 Current address: Electrical Engineering and Computer Sciences, University of California, Berkeley, CA 94720-1770.
Open AccessPublished:May 30, 2019DOI:https://doi.org/10.1016/j.exphem.2019.05.003

      Highlights

      • Various models that functionally characterize pediatric leukemias are described.
      • It has been found through studies of pediatric AML cohorts that pediatric AML is genetically distinct from adult AML.
      • The strengths and limitations of human models versus PDXs are discussed.
      Leukemia is a complex genetic disease caused by errors in differentiation, growth, and apoptosis of hematopoietic cells in either lymphoid or myeloid lineages. Large-scale genomic characterization of thousands of leukemia patients has produced a tremendous amount of data that have enabled a better understanding of the differences between adult and pediatric patients. For instance, although phenotypically similar, pediatric and adult myeloid leukemia patients differ in their mutational profiles, typically involving either chromosomal translocations or recurrent single–base-pair mutations, respectively. To elucidate the molecular mechanisms underlying the biology of this cancer, continual efforts have been made to develop more contextually and biologically relevant experimental models. Leukemic cell lines, for example, provide an inexpensive and tractable model but often fail to recapitulate critical aspects of tumor biology. Likewise, murine leukemia models of leukemia have been highly informative but also do not entirely reproduce the human disease. More recent advances in the development of patient-derived xenografts (PDXs) or human models of leukemias are poised to provide a more comprehensive, and biologically relevant, approach to directly assess the impact of the in vivo environment on human samples. In this review, the advantages and limitations of the various current models used to functionally define the genetic requirements of leukemogenesis are discussed.
      Leukemia is a cancer of the blood, originating from hematopoietic stem and progenitor cells that lose their capacity for proper self-renewal, differentiation and apoptosis. While typically diagnosed in older adults (median age of ∼68 [
      • Appelbaum FR
      • Gundacker H
      • Head DR
      • et al.
      Age and acute myeloid leukemia.
      ,
      • Estey E
      • Dohner H
      Acute myeloid leukaemia.
      ,
      • Juliusson G
      • Antunovic P
      • Derolf A
      • et al.
      Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry.
      ,
      • Phekoo KJ
      • Richards MA
      • Moller H
      • Schey SA
      South Thames Haematology Specialist Committee
      The incidence and outcome of myeloid malignancies in 2,112 adult patients in southeast England.
      ]) this complex genetic disease still remains one of the most common cancer during childhood, representing almost one third of all cancer diagnoses in children under the age of 15 [
      • Steliarova-Foucher E
      • Colombet M
      • Ries LAG
      • et al.
      International incidence of childhood cancer, 2001–10: a population-based registry study.
      ,
      • Siegel DA
      • King J
      • Tai E
      • Buchanan N
      • Ajani UA
      • Li J
      Cancer incidence rates and trends among children and adolescents in the United States, 2001–2009.
      ,
      • Belson M
      • Kingsley B
      • Holmes A
      Risk factors for acute leukemia in children: a review.
      ]. Large-scale genomics studies of adult Acute Myeloid Leukemia (AML) patient cohorts have shown that it is a genetically heterogeneous disease, with a complex mutational landscape that has complicated efforts to develop broadly applicable targeted therapies [
      • TJ Ley
      • Miller C
      • et al.
      Cancer Genome Atlas Research Network
      Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia.
      ]. Evidence suggests that in adult patients, the gradual acquisition of random mutations, some of which have oncogenic activity, over several decades can convert a normal hematopoietic stem cell into an unregulated leukemic blast [
      • Welch JS
      • Ley TJ
      • Link DC
      • et al.
      The origin and evolution of mutations in acute myeloid leukemia.
      ]. Recently, similar studies of large cohorts of pediatric acute myeloid leukemia patients have confirmed past observations that, in contrast to the gradual accumulation of mutations seen in adult AML, in younger AML patients, chromosomal translocations are most often responsible for the development of the disease [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ], in contrast to pediatric vs adult ALL [
      • Liu YF
      • Wang BY
      • Zhang WN
      • et al.
      Genomic profiling of adult and pediatric B-cell acute lymphoblastic leukemia.
      ]. Even mutations that have been shown to be recurrent in both adult and pediatric AML patients occur at highly divergent frequencies [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ]. Taken together, current evidence suggests that despite the common phenotypes of pediatric and adult AML, they largely represent distinct genetic diseases with chromosomal rearrangements likely representing the initiating (or driver) event in a large fraction of childhood leukemia development.
      Patient outcomes for childhood leukemia have significantly improved in the last 40 years [
      • Mussai FJ
      • Yap C
      • Mitchell C
      • Kearns P
      Challenges of clinical trial design for targeted agents against pediatric leukemias.
      ], largely through improved treatment modulation; however, they still remain lower in AML than in most other pediatric cancers [
      National Cancer Institute (NCI)
      SEER cancer statistics review.
      ]. In addition, patient outcomes are highly dependent on the specific subgroup of leukemia involved [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ], suggesting that distinct mutational profiles, and therefore therapeutic targets, exist. As noted previously, individual genetic heterogeneity is observed in all leukemia subgroups, confounding efforts to define the essential genetic elements required for leukemias with specific translocations or other genetic driver events. It is therefore critical to develop fully representative models of each leukemia type, such that the genetic background noise that comes with studying individual patient samples can be effectively removed.
      Among the first experimental tools developed to model the biology underlying the disease were cell lines established from the in vitro culture of primary patient samples. After their initial establishment, leukemic cell lines represented a relatively simple and cost-effective way to study the disease, and have formed the backbone of much of the experimental work performed to date. More recently, large-scale studies of leukemic cell lines have also provided deeper insight into drug sensitivities [
      • Barretina J
      • Caponigro G
      • Stransky N
      • et al.
      The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity.
      ] and genetic networks that are active within cell lines [
      • Yang J
      • Li A
      • Li Y
      • Guo X
      • Wang M
      A novel approach for drug response prediction in cancer cell lines via network representation learning.
      ,
      • Zhang F
      • Wang M
      • Xi J
      • Yang J
      • Li A
      A novel heterogeneous network-based method for drug response prediction in cancer cell lines.
      ]. Despite their utility, all cell lines have undergone selection for their ability to grow in an in vitro environment, typically without growth factors or other signals primary leukemias typically receive in the bone marrow niche. As a result, the gene expression patterns and underlying molecular biology often are not representative of the disease seen in patients.
      In parallel efforts, concerted studies aimed at better characterizing mouse hematopoietic stem/progenitor cells (HSPCs) have subsequently allowed the development of murine leukemic models to explore disease mechanisms in vivo [
      • Challen GA
      • Boles N
      • Lin KK
      • Goodell MA
      Mouse hematopoietic stem cell identification and analysis.
      ,
      • Dykstra B
      • Ramunas J
      • Kent D
      • et al.
      High-resolution video monitoring of hematopoietic stem cells cultured in single-cell arrays identifies new features of self-renewal.
      ,
      • Okada S
      • Nakauchi H
      • Nagayoshi K
      • Nishikawa S
      • Miura Y
      • Suda T
      In vivo and in vitro stem cell function of c-kit- and Sca-1-positive murine hematopoietic cells.
      ]. Such models represent an extremely powerful tool, not only for defining the genetic dependencies of leukemias, but also for providing an experimental system in which to evaluate therapeutic approaches. Although a range of leukemic models have been developed, many of these involve the use of the potently oncogenic gene fusions typically seen in pediatric leukemias [
      • Dang J
      • Nance S
      • Ma J
      • et al.
      AMKL chimeric transcription factors are potent inducers of leukemia.
      ,
      • Bijl J
      • Sauvageau M
      • Thompson A
      • Sauvageau G
      High incidence of proviral integrations in the Hoxa locus in a new model of E2a-PBX1-induced B-cell leukemia.
      ,
      • Forster A
      • Pannell R
      • Drynan LF
      • et al.
      Engineering de novo reciprocal chromosomal translocations associated with Mll to replicate primary events of human cancer.
      ,
      • Corral J
      • Lavenir I
      • Impey H
      • et al.
      An Mll–AF9 fusion gene made by homologous recombination causes acute leukemia in chimeric mice: a method to create fusion oncogenes.
      ]. While potentially closer to the human disease than cell lines because of the lack of selection for in vitro growth, differences in genome structure, life span, and transformability of mouse cells mean that some of these models still do not fully recapitulate the human disease. Clearly, the ability to generate human models of pediatric leukemia would represent an optimal experimental system; however, the practical development of such models has been hampered by technical challenges in isolating and transforming the correct HSPC population and identifying the correctly optimized culture conditions and the availability of suitable mouse strains. Once again, these road blocks have been overcome by transducing HSPCs with potently oncogenic fusion genes such as KMT2A–MLLT3 (MLL–AF9), allowing engineered human leukemias to be produced [
      • Barabe F
      • Kennedy JA
      • Hope KJ
      • Dick JE
      Modeling the initiation and progression of human acute leukemia in mice.
      ].
      In this review, we discuss the various benefits and drawbacks of the various different model systems available to study pediatric leukemias, including cell lines and human models. Additionally, we present descriptions of the current public patient-derived xenograft (PDX) resources available, and describe their impact on the development of novel therapeutics.

      Mining patient cohorts to understand the disease

      In the past decade, rapid advances in DNA sequencing technology have fueled the field of cancer genomics, enabling the comprehensive sequencing of large cohorts of patient tumors and matched normal tissue [
      • TJ Hudson
      • Anderson W
      • et al.
      International Cancer Genome Consortium
      International network of cancer genome projects.
      ,
      • Stratton MR
      • Campbell PJ
      • Futreal PA
      The cancer genome.
      ]. The vast data sets created through these efforts has opened a powerful window to understanding the genetics of cancer biology including recurrent mutations, gene expression, and chromatin accessibility [
      • Hoadley KA
      • Yau C
      • Hinoue T
      • et al.
      Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer.
      ]. Comparisons of the mutational burden in different cancer types have shown that acute myeloid leukemias have significantly fewer somatic mutations present relative to other cancers [
      • Alexandrov LB
      • Nik-Zainal S
      • Wedge DC
      • et al.
      Signatures of mutational processes in human cancer.
      ], with pediatric AML having even lower levels than adult AMLs [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ]. Despite this fact, high levels of genetic heterogeneity exist between all cancer patients, including pediatric leukemia patient samples. Such heterogeneity includes not only recurrent somatic mutations, but also the 200–300 [
      • Shen H
      • Li J
      • Zhang J
      • et al.
      Comprehensive characterization of human genome variation by high coverage whole-genome sequencing of forty four Caucasians.
      ,
      • Genomes Project C
      • Abecasis GR
      • Altshuler D
      • et al.
      A map of human genome variation from population-scale sequencing.
      ] loss-of-function variants and thousands of structural variations [
      • Sudmant PH
      • Rausch T
      • Gardner EJ
      • et al.
      An integrated map of structural variation in 2,504 human genomes.
      ] estimated to be present in each human genome. Clearly this genetic variation has the potential to influence the development of leukemia, though likely through complex interactions of other variants or somatic mutations, which may be patient specific. Although studies of patient cohorts provide an important “average view” of a cancer, a comprehensive understanding of the contribution of genetic drivers in individual patients (i.e., how all of their somatic and germline variants contribute to the leukemia) is often missing.
      Several large-scale genomics studies, focusing on pediatric cancers [
      • Grobner SN
      • Worst BC
      • Weischenfeldt J
      • et al.
      The landscape of genomic alterations across childhood cancers.
      ] or specifically on leukemias [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ,
      • Schwartz JR
      • Ma J
      • Lamprecht T
      • et al.
      The genomic landscape of pediatric myelodysplastic syndromes.
      ,
      • Hsu CH
      • Nguyen C
      • Yan C
      • et al.
      Transcriptome profiling of pediatric core binding factor AML.
      ,
      • Andersson AK
      • Ma J
      • Wang J
      • et al.
      The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias.
      ,
      • Jaffe JD
      • Wang Y
      • Chan HM
      • et al.
      Global chromatin profiling reveals NSD2 mutations in pediatric acute lymphoblastic leukemia.
      ,
      • Roberts KG
      • Morin RD
      • Zhang J
      • et al.
      Genetic alterations activating kinase and cytokine receptor signaling in high-risk acute lymphoblastic leukemia.
      ], have been published in which the frequency of recurrent translocations is typically high. One landmark project, the TARGET (Therapeutically Applicable Research to Generate Effective Treatments) program, is managed by the National Cancer Institute, the Children's Oncology group, St. Jude Children's Hospital, and others and has the goal of using a multi-omics approach to comprehensively define all the molecular changes that are specifically implicated in the initiation and progression of childhood cancer. The data from this project, much of it publicly available, have been of significant value in investigating common disease mechanisms and identifying novel therapeutic strategies for hard-to-treat pediatric cancers. For instance, a pan-cancer genome analysis of 1,699 pediatric leukemias reported 142 potential driver genes, of which only 45% were also found in adult patients, highlighting the need for the development of distinct treatment strategies for young patients [
      • Ma X
      • Liu Y
      • Liu Y
      • et al.
      Pan-cancer genome and transcriptome analyses of 1,699 paediatric leukaemias and solid tumours.
      ]. Another more specifically focused study characterized nearly 1,000 young and adult pediatric patients and provided an outstanding molecular landscape of AML [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ]. Supporting previous evidence on the differences between adult and pediatric leukemia, the authors provided clear evidence for age-specific differences in the spectrum of mutations and genetic variations present. Mutations that are typically found at high frequency in adult patients (e.g., FLT3, NPM1, IDH1, and IDH2) are rare in pediatric patients, whereas the frequency of mutations in specific signaling pathways (e.g., KRAS, NRAS, KIT, WT1) is significantly higher [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ]. Moreover, the incidence of recurrent translocations in AML also follows an age-specific pattern, a fact likely linked to the early development of leukemia. Of all gene fusions, those involving the KMT2A (mixed lineage leukemia [MLL]) gene are the most common translocations seen in infants and older children [
      • Bolouri H
      • Farrar JE
      • Triche Jr, T
      • et al.
      The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
      ,
      • Brunner AM
      • Graubert TA.
      Genomics in childhood acute myeloid leukemia comes of age.
      ]. With respect to recurrent fusions in pediatric B-cell (B-ALL) and T-cell (T-ALL) acute lymphoblastic leukemia, many of these have been identified and functionally characterized as well. For instance, the ETV6–RUNX1 (TEL/AML1) fusion resulting from the often cryptic t(12;21) translocation is one of the most common fusions and is found in 25% of pediatric ALL patients [
      • Zelent A
      • Greaves M
      • Enver T
      Role of the TEL–AML1 fusion gene in the molecular pathogenesis of childhood acute lymphoblastic leukaemia.
      ], where its occurrence during pregnancy may provide a “first hit” for the leukemia [
      • Mori H
      • Colman SM
      • Xiao Z
      • et al.
      Chromosome translocations and covert leukemic clones are generated during normal fetal development.
      ]. The well-documented Philadelphia chromosome resulting from the fusion between ABL1 and BCR (t(9;22)(q34;q11)), which is often observed in adult B-ALL, is much less frequent in pediatric cases of B-ALL (∼25% vs. 4% of cases [
      • Bernt KM
      • Hunger SP.
      Current concepts in pediatric Philadelphia chromosome-positive acute lymphoblastic leukemia.
      ]). Interestingly however, in Philadelphia chromosome-like B-ALL (representing ∼20% of all pediatric patients [
      • Reshmi SC
      • Harvey RC
      • Roberts KG
      • et al.
      Targetable kinase gene fusions in high-risk B-ALL: a study from the Children's Oncology Group.
      ]), fusions of the CRLF2 gene were seen in 43% of patients, whereas those without CRLF2 fusions were frequently characterized by JAK2/EPOR fusions (∼8% of cases) [
      • Reshmi SC
      • Harvey RC
      • Roberts KG
      • et al.
      Targetable kinase gene fusions in high-risk B-ALL: a study from the Children's Oncology Group.
      ]. The most common gene fusion in pediatric T-ALL involves the STIL–TAL1 genes leading to an overexpression of TAL1 (along with frequent overexpression of SLC17A9) not seen to the same extent in adult T-ALL [
      • Chen B
      • Jiang L
      • Zhong ML
      • et al.
      Identification of fusion genes and characterization of transcriptome features in T-cell acute lymphoblastic leukemia.
      ], while a diverse collection of other recurrent fusions have been identified [
      • Papenhausen P
      • Kelly CA
      • Zhang Z
      • Tepperberg J
      • Burnside RD
      • Schwartz S
      Multidisciplinary analysis of pediatric T-ALL: 9q34 gene fusions.
      ]
      Another large-scale genomics effort, the Pediatric Cancer Genome Project (PCPG) cooperatively developed in 2010 by St. Jude Children's Research Hospital and Washington University analyzed the genomes of childhood cancer patients and identified somatic mutations that drive cancer [
      • Downing JR
      • Wilson RK
      • Zhang J
      • et al.
      The Pediatric Cancer Genome Project.
      ]. With roughly 5,000 patient samples from more than a dozen cancer types, including more than 2,300 leukemic samples, these data represent another highly informative source for pediatric cancers, complementing others available through the US National Human Genome Research Institute (NHGRI), National Cancer Institute (NCI), The Cancer Genome Atlas (TCGA), the International Cancer Genome Consortium (ICGC), and the TARGET project. Importantly, the findings from all of these studies have emphasized the challenges involved in analyzing the complex genetic data, where age-dependent differences complicate comparisons between adult and pediatric patients. For example, although the frequency of somatic mutation is lower in pediatric AML patients than in adult patients, there may be a greater role for germline variants that predispose pediatric patients to AML. Such predisposition variants have been identified in pediatric ALL [
      • Churchman ML
      • Qian M
      • Te Kronnie G
      • et al.
      Germline genetic IKZF1 variation and predisposition to childhood acute lymphoblastic leukemia.
      ,
      • Qian M
      • Cao X
      • Devidas M
      • et al.
      TP53 germline variations influence the predisposition and prognosis of B-cell acute lymphoblastic leukemia in children.
      ], AML [
      • Phillips CL
      • Gerbing R
      • Alonzo T
      • et al.
      MDM2 polymorphism increases susceptibility to childhood acute myeloid leukemia: a report from the Children's Oncology Group.
      ], and multiple myeloma [
      • Wei X
      • Calvo-Vidal MN
      • Chen S
      • et al.
      Germline lysine-specific demethylase 1 (LSD1/KDM1A) mutations confer susceptibility to multiple myeloma.
      ] patients and are estimated to occur in ∼10% of all pediatric cancers [
      • Zhang J
      • Walsh MF
      • Wu G
      • et al.
      Germline mutations in predisposition genes in pediatric cancer.
      ]. Given that no comprehensive list of such variants exists for pediatric AML, and that the 10% frequency reported may be an underestimate [
      • Brodeur GM
      • Nichols KE
      • Plon SE
      • Schiffman JD
      • Malkin D
      Pediatric cancer predisposition and surveillance: an overview, and a tribute to Alfred G.
      ], comparing the genetics of pediatric AML and adult AML through recurrently mutated genes is problematic. This, in turn, has highlighted the critical need for better models of the disease to be able to functionally study the role of specific mutations in the disease and to develop genotype-specific treatments.

      Using cell lines as model systems for pediatric leukemias

      Long before the advent of recombinant DNA technologies, immortal cell lines derived from patient tumors had already been developed as a model to try to better understand the biology of cancer. Over the years, hundreds of cell lines have been established from a range of different tumor types, and like current cancer patient samples, NGS approaches have recently been applied to provide a detailed molecular characterization of large sets of these cells [
      • Barretina J
      • Caponigro G
      • Stransky N
      • et al.
      The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity.
      ,
      • Iorio F
      • Knijnenburg TA
      • Vis DJ
      • et al.
      A landscape of pharmacogenomic interactions in cancer.
      ,
      • Klijn C
      • Durinck S
      • Stawiski EW
      • et al.
      A comprehensive transcriptional portrait of human cancer cell lines.
      ,
      • Garnett MJ
      • Edelman EJ
      • Heidorn SJ
      • et al.
      Systematic identification of genomic markers of drug sensitivity in cancer cells.
      ]. With respect to pediatric leukemias, a number of cell lines have been established from patients [
      • Drexler HG
      • Quentmeier H
      • MacLeod RA
      Malignant hematopoietic cell lines: in vitro models for the study of MLL gene alterations.
      ], many of which contain recurrent gene fusions seen in many patient leukemias. At the same time, this effort has been complicated by the fact that primary leukemic cells will rapidly differentiate or die when placed into in vitro culture. Even those cells that do not immediately die must often be cultured for months [
      • Munker R
      • Nordberg ML
      • Veillon D
      • et al.
      Characterization of a new myeloid leukemia cell line with normal cytogenetics (CG-SH).
      ] before a proliferating clone emerges, resulting in a highly inefficient process for generating cell lines. Once established, such cell lines have, however, provided a useful, inexpensive, and simple model to study defects in cell differentiation and self-renewal leading to the development of a specific leukemia. They also represent a powerful system for high-throughput chemical or genetic screens and for testing the impact of specific mutations. Moreover, the large number of different cell lines representing various tissues provides the possibility of not only assessing the consistency and reproducibility of results but also examining the tissue-specific aspects of experiments.
      Despite their immense value, virtually all cell lines are in fact far removed, biologically speaking, from the actual human disease for a number of reasons. For instance, by virtue of how they are established, all tumor cells are selected for the “unnatural” characteristic of being able to grow in vitro. For leukemic cells, basic in vitro culture conditions typically lead to rapid cell differentiation and death without the addition of additional growth factors or stromal support cells [
      • Pabst C
      • Krosl J
      • Fares I
      • et al.
      Identification of small molecules that support human leukemia stem cell activity ex vivo.
      ]. As a result, these stringent growth conditions can rapidly select for any subclones harboring mutations that allow them to survive, even if these would have no advantage in vivo. Moreover, the constant passage of cells also has the potential to apply selective pressure for faster growth, leading to the accumulation of additional mutations or genomic instability not seen in the primary disease, and is often reflected in their highly variable karyotypes. There is clear evidence of the dramatic effects of this process in other cancer cell lines [
      • Ben-David U
      • Siranosian B
      • Ha G
      • et al.
      Genetic and transcriptional evolution alters cancer cell line drug response.
      ], and although evidence for similar genetic heterogeneity has not yet been described for AML cell lines, an immortal and continually growing cell line with finite DNA replication fidelity must inevitably acquire novel mutations. In addition, the karyotypic abnormalities seen in large numbers of AML cell lines [
      • Moy C
      • Oleykowski CA
      • Plant R
      • et al.
      High chromosome number in hematological cancer cell lines is a negative predictor of response to the inhibition of Aurora B and C by GSK1070916.
      ] likely also reflects an inherent level of genomic instability that can affect the phenotype of the cell lines. For instance, analysis of the gene expression patterns of ∼200 hematopoietic cell lines reveals that their morphological classification does not always correlate well with their clustering based on expression patterns of different leukemia subtypes (Figure 1). As to whether any specific AML cell line may be a more representative model of pediatric AML, it is difficult to say, and this likely depends on the presence of specific mutations or biological activities that are being examined.
      Figure 1
      Figure 1Hierarchical clustering of hematopoietic cell lines. Published gene expression data from the Cancer Cell Line Encyclopedia analysis of (∼200) established hematopoietic cell lines was used to perform hierarchical clustering based on the expression levels of the 1,000 most variable genes across all cell lines. Cell type annotation derived from published descriptions as well as information from commercial cell repositories was used to color the dendrogram branches for the cell types noted in the legend.
      Attempts to overcome these inherent limitations of cell lines have largely focused on using large numbers of cells to observe the general trends despite the genetic noise present. Several large-scale studies, using hundreds of cell lines, have been performed to test tumor response to drugs to identify novel therapeutic targets [
      • Barretina J
      • Caponigro G
      • Stransky N
      • et al.
      The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity.
      ,
      • Iorio F
      • Knijnenburg TA
      • Vis DJ
      • et al.
      A landscape of pharmacogenomic interactions in cancer.
      ]. While similar caveats remain (e.g., that drugs showing therapeutic effects in vitro may not have the same consequences on cancerous cells because of differences in the in vivo microenvironment [
      • Chang YT
      • Hernandez D
      • Alonso S
      • et al.
      Role of CYP3A4 in bone marrow microenvironment-mediated protection of FLT3/ITD AML from tyrosine kinase inhibitors.
      ]), these data can still be informative for understanding the relevance of specific biological pathways. Human AML cells lines bearing gene fusions commonly seen in pediatric leukemias have been the focus of a number of high-throughput studies aimed at establishing a catalog of genetic vulnerabilities. Approaches using both shRNA- [
      • Heshmati Y
      • Turkoz G
      • Harisankar A
      • et al.
      The chromatin-remodeling factor CHD4 is required for maintenance of childhood acute myeloid leukemia.
      ,
      • Li H
      • Mar BG
      • Zhang H
      • et al.
      The EMT regulator ZEB2 is a novel dependency of human and murine acute myeloid leukemia.
      ,
      • Porter CC
      • Kim J
      • Fosmire S
      • et al.
      Integrated genomic analyses identify WEE1 as a critical mediator of cell fate and a novel therapeutic target in acute myeloid leukemia.
      ] and CRISPR-Cas9–based [
      • Wang T
      • Yu H
      • Hughes NW
      • et al.
      Gene essentiality profiling reveals gene networks and synthetic lethal interactions with oncogenic Ras.
      ,
      • Tzelepis K
      • Koike-Yusa H
      • De Braekeleer E
      • et al.
      A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia.
      ] screening approaches have provided useful insight into the biology of pediatric leukemia despite clear differences caused by cell line–specific genetic heterogeneity (e.g. both differences between cell lines and subclonal heterogeneity with cell lines). Taken together, even though cell lines provide a useful resource to study the genetics of leukemias, they still present potential limitations for specific biological questions, and studies using individual cell lines may not provide accurate data for attempts to recapitulate the disease in vivo.

      Murine models of pediatric AML

      Given the inherent limitations of cell line models of pediatric leukemia, the generation of novel models starting from primary cells has been of great interest. Such models have in turn been dependent on parallel technical advances in the ability to isolate and characterize murine progenitor cells that are competent to be transformed. Initial efforts to transform murine cells involved the prototypical BCR–ABL fusion seen in adult chronic myelogenous leukemia (CML) and used retroviral transduction of murine bone marrow (BM) cells [
      • Daley GQ
      • Van Etten RA
      • Baltimore D
      Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome.
      ,
      • Kelliher MA
      • McLaughlin J
      • Witte ON
      • Rosenberg N
      Induction of a chronic myelogenous leukemia-like syndrome in mice with v-abl and BCR/ABL.
      ]. These models illustrated that it was feasible to introduce specific oncogenes into murine cells to transform the cells into a state that can mimic the human disease. Subsequent models established through the homologous recombination of common pediatric gene fusions such as the KMT2A gene were performed using murine embryonic stem (ES) cells and indicated that, phenotypically, the transformed cells resembled human AML [
      • Corral J
      • Lavenir I
      • Impey H
      • et al.
      An Mll–AF9 fusion gene made by homologous recombination causes acute leukemia in chimeric mice: a method to create fusion oncogenes.
      ]. The high frequency of KMT2A translocations in pediatric leukemias [
      • Winters AC
      • Bernt KM.
      MLL-rearranged leukemias—an update on science and clinical approaches.
      ], its astonishing range of fusion partners [
      • Meyer C
      • Burmeister T
      • Groger D
      • et al.
      The MLL recombinome of acute leukemias in 2017.
      ], and the success of initial studies all serve to explain the numerous subsequent efforts to study similar models with KMT2A fusions with ENL and AF4, among others. Interestingly, a number of oncogenic fusions seen in patient leukemias are insufficient by themselves to transform murine HSPCs (e.g. ETV6–RUNX1) or do so only after long latency periods (e.g. MLL–AF4). This observation suggests that in some cases, the ability to accurately recapitulate certain human leukemias will likely require a better understanding and quantification of the contribution of typically heterogenic mutations seen in patient samples or isolation of relevant cell of origin for distinct leukemia subtypes.
      The cell of origin used for the model leukemias that have been developed is also an important consideration for the accurate duplication of disease phenotypes across species. Although some earlier models using genetically manipulated embryonic stem cells from transgenic mice have been found to generate leukemias, most fusions in pediatric leukemias are typically tumor restricted along with recurrent somatic mutations. The continual refinement of cell surface markers for murine HSPCs has meant that many of the later models have specifically targeted these cells for transformation. Despite this more precise targeting, even relatively pure HSPC populations remain heterogeneous with a range of differentiation potential [
      • Dykstra B
      • Kent D
      • Bowie M
      • et al.
      Long-term propagation of distinct hematopoietic differentiation programs in vivo.
      ] whose impact on leukemia development (e.g. lineage phenotype) remains undefined. In the case of particularly potent oncogenes such as KMT2A fusions, the cell of origin (or even the promoter driving the fusion [
      • Metzler M
      • Forster A
      • Pannell R
      • et al.
      A conditional model of MLL–AF4 B-cell tumourigenesis using invertor technology.
      ]) may have a negligible impact on leukemia development. This may not be the case, however, for other oncogenic fusions that have been used to model pediatric leukemias, including AML1–ETO [
      • de Guzman CG
      • Warren AJ
      • Zhang Z
      • et al.
      Hematopoietic stem cell expansion and distinct myeloid developmental abnormalities in a murine model of the AML1–ETO translocation.
      ], E2A–PBX1 [
      • Duque-Afonso J
      • Feng J
      • Scherer F
      • et al.
      Comparative genomics reveals multistep pathogenesis of E2A-PBX1 acute lymphoblastic leukemia.
      ], and, more recently, models looking at higher risk groups such as acute megakaryoblastic leukemia (AMKL) [
      • Dang J
      • Nance S
      • Ma J
      • et al.
      AMKL chimeric transcription factors are potent inducers of leukemia.
      ].
      The creation of murine models of pediatric leukemia has been particular informative for performing functional studies to define the essential genetic requirements for leukemia development from normal progenitor cells. The ability to perform these experiments iteratively, under a range of different conditions of genetic perturbation, provides significant flexibility to investigate specific genes. Although similar experiments can be performed in individual leukemia cell lines, in this case the leukemia is already maintained by a specific collection of genetic lesions that may or may not be relevant for a given patient subgroup, limiting the scope of experiments. At the same time, because mouse models are typically designed to address the role of specific proteins or mutations, testing complex combinations of the 4–10 driver mutations found in adult AMLs [
      • Metzeler KH
      • Herold T
      • Rothenberg-Thurley M
      • et al.
      Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia.
      ] is not feasible (Table 1).
      Table 1Human and mouse models of leukemia
      SpeciesYearDriver usedPhenotype of modelNotesReference
      Mouse2018GATA2‐HOXA9, MN1–FLI, NIPBLHOXB, (CBFA2T3–GLIS2)AMKLCBFA2T3–GLIS2 fusion alone is insufficient to generate leukemia
      • Dang J
      • Nance S
      • Ma J
      • et al.
      AMKL chimeric transcription factors are potent inducers of leukemia.
      2017ETV6–RUNX1B-Cell precursor ALLLoss of function in KDM family genes potentially relevant for disease penetrance
      • Rodriguez-Hernandez G
      • Hauer J
      • Martin-Lorenzo A
      • et al.
      Infection exposure promotes ETV6–RUNX1 precursor b-cell leukemia via impaired H3K4 demethylases.
      2015E2A–PBX1B-Cell precursor ALLPre‐BCR signaling/JAK kinases are potential therapeutic targets
      • Duque-Afonso J
      • Feng J
      • Scherer F
      • et al.
      Comparative genomics reveals multistep pathogenesis of E2A-PBX1 acute lymphoblastic leukemia.
      2013MLL–AF6AMLSelective sensitivity to Dot1l inhibitor (EPZ0004777)
      • Deshpande AJ
      • Chen L
      • Fazio M
      • et al.
      Leukemic transformation by the MLL–AF6 fusion oncogene requires the H3K79 methyltransferase Dot1l.
      2009ETV6–RUNX1Increased HSPC frequency but no leukemiaChemical mutagenesis required for leukemia development
      • Bernardin F
      • Yang Y
      • Cleaves R
      • et al.
      TEL-AML1, expressed from t(12;21) in human acute lymphocytic leukemia, induces acute leukemia in mice.
      2009OTT–MALAMKLNotch signaling (RBPJ) and cytokine (MPL) signaling are required for AMKL generation
      • Mercher T
      • Raffel GD
      • Moore SA
      • et al.
      The OTT–MAL fusion oncogene activates RBPJ-mediated transcription and induces acute megakaryoblastic leukemia in a knockin mouse model.
      2008MLL–AF4B‐ALL/AMLH3K79 methylation patterns correlate with gene expression patterns and are conserved between human and mouse samples
      • Krivtsov AV
      • Feng Z
      • Lemieux ME
      • et al.
      H3K79 methylation profiles define murine and human MLL–AF4 leukemias.
      2006MLL–AF4B‐ALLExpression of MLL–AF4 restricted to lymphoid cells, drives B‐ALL development
      • Metzler M
      • Forster A
      • Pannell R
      • et al.
      A conditional model of MLL–AF4 B-cell tumourigenesis using invertor technology.
      2006MLL–AF4B‐ALL/AMLBoth MLL–AF4/MLL–AF9 transform BM cells, but MLL–AF9 has a shorter latency and favors AML vs. BALL/MPAL for MLL–AF4
      • Chen W
      • Li Q
      • Hudson WA
      • Kumar A
      • Kirchhof N
      • Kersey JH
      A murine Mll-AF4 knock-in model results in lymphoid and myeloid deregulation and hematologic malignancy.
      2005AML1–ETOAMLDemonstrated that AML1‐ETO fusions collaborate with FLT3 internal tandem duplications to generate AML
      • Schessl C
      • Rawat VP
      • Cusan M
      • et al.
      The AML1–ETO fusion gene and the FLT3 length mutation collaborate in inducing acute leukemia in mice.
      2003MLL–ENLAMLRapid highly penetrant leukemias result from Cre–loxP-mediated reciprocal chromosomal translocations
      • Forster A
      • Pannell R
      • Drynan LF
      • et al.
      Engineering de novo reciprocal chromosomal translocations associated with Mll to replicate primary events of human cancer.
      2003NUP98–HOXD13Myeloproliferation/AMLFull transformation of BM cells into AML was dependent on co‐expression of MEIS1
      • Pineault N
      • Buske C
      • Feuring-Buske M
      • et al.
      Induction of acute myeloid leukemia in mice by the human leukemia-specific fusion gene NUP98–HOXD13 in concert with Meis1.
      2002ETV6–RUNX1B‐ALL/T‐ALLDevelopment of leukemia is potentiated by loss of p16/p19
      • Bernardin F
      • Yang Y
      • Cleaves R
      • et al.
      TEL-AML1, expressed from t(12;21) in human acute lymphocytic leukemia, induces acute leukemia in mice.
      2002AML1–ETOAMLAltered HSC compartment size but without leukemia of disseminated disease
      • de Guzman CG
      • Warren AJ
      • Zhang Z
      • et al.
      Hematopoietic stem cell expansion and distinct myeloid developmental abnormalities in a murine model of the AML1–ETO translocation.
      2001NUP98–HOXA9AMLInduces a polyclonal AML and defined a role for MEIS1 cofactor in accelerating the disease
      • Kroon E
      • Thorsteinsdottir U
      • Mayotte N
      • Nakamura T
      • Sauvageau G
      NUP98–HOXA9 expression in hemopoietic stem cells induces chronic and acute myeloid leukemias in mice.
      1996MLL–AF9AMLFirst mouse model of AML generated through Cre‐loxP mediated fusion in ES cells

      • Corral J
      • Lavenir I
      • Impey H
      • et al.
      An Mll–AF9 fusion gene made by homologous recombination causes acute leukemia in chimeric mice: a method to create fusion oncogenes.
      Human2018MLL–AF9AMLCRISPR‐based editing can be used in human hematopoietic stem and

      progenitor cells to generate leukemias
      • Schneidawind C
      • Jeong J
      • Schneidawind D
      • et al.
      MLL leukemia induction by t(9;11) chromosomal translocation in human hematopoietic stem cells using genome editing.
      2017RBM15–MKL1HPSCs with AMKL‐like expressionTwo HPSC cell lines expressing RBM15–MKL1 oncogenic fusion seen exhibit some gene expression patterns similar to that of AMKL
      • Ayllon V
      • Vogel-Gonzalez M
      • Gonzalez-Pozas F
      • et al.
      New hPSC-based human models to study pediatric acute megakaryoblastic leukemia harboring the fusion oncogene RBM15-MKL1.
      2017MLL–AF9AML/B‐ALLRET highlighted as a therapeutic target and sequencing data showing the KMT2A–MLLT3 fusion alone is sufficient to generate leukemias
      • Barabe F
      • Gil L
      • Celton M
      • et al.
      Modeling human MLL-AF9 translocated acute myeloid leukemia from single donors reveals RET as a potential therapeutic target.
      2015MLL–AF9/ENLALL/AML/MPALTALEN induced gene editing of the endogenous MLL gene in CD34 cells
      • Buechele C
      • Breese EH
      • Schneidawind D
      • et al.
      MLL leukemia induction by genome editing of human CD34+ hematopoietic cells.
      2014NUP98–HOXD13AMLForced expression of MN1 is insufficient to generate leukemias in the absence of HOX gene fusion
      • Imren S
      • Heuser M
      • Gasparetto M
      • et al.
      Modeling de novo leukemogenesis from human cord blood with MN1 and NUP98HOXD13.
      2012MLL–AF4Block in HSC commitmentEnforced expression of MLL–AF4 in hESCs impairs hematopoietic development and is insufficient to transform cells
      • Bueno C
      • Montes R
      • Melen GJ
      • et al.
      A human ESC model for MLL–AF4 leukemic fusion gene reveals an impaired early hematopoietic-endothelial specification.
      2012MLL–AF9AMLIntrinsic properties of the cell of origin affect the efficiency of transformation in human model leukemias
      • Horton SJ
      • Jaques J
      • Woolthuis C
      • et al.
      MLL-AF9-mediated immortalization of human hematopoietic cells along different lineages changes during ontogeny.
      2012MLL–AF10HSPCs with MLL–AF10 and activated K‐ras develop AMLMLL‐AF10 alone is insufficient to generate AML in humanized mice
      • Moriya K
      • Suzuki M
      • Watanabe Y
      • et al.
      Development of a multi-step leukemogenesis model of MLL-rearranged leukemia using humanized mice.
      2011MLL–AF4HSPCs with increased proliferation and clonogenic potentialMLL‐AF4 enhances proliferation but is insufficient to generate a leukemia in human cells
      • Montes R
      • Ayllon V
      • Gutierrez-Aranda I
      • et al.
      Enforced expression of MLL–AF4 fusion in cord blood CD34+ cells enhances the hematopoietic repopulating cell function and clonogenic potential but is not sufficient to initiate leukemia.
      2010BCR–ABLCMLBMI1 collaborates with BCR–ABL in leukemia development; lymphoid phenotype in vivo with myeloid/lymphoid cells established in culture
      • Rizo A
      • Horton SJ
      • Olthof S
      • et al.
      BMI1 collaborates with BCR–ABL in leukemic transformation of human CD34+ cells.
      2008MLL–AF9AML/ALL/MPALPhenotype of transformed CB cells is impacted by growth factors, microenvironment, and mouse strain used
      • Wei J
      • Wunderlich M
      • Fox C
      • et al.
      Microenvironment determines lineage fate in a human model of MLL-AF9 leukemia.
      2007MLL–AF9/ENLAML/B‐ALLFirst human model of AML developed using CB cells
      • Barabe F
      • Kennedy JA
      • Hope KJ
      • Dick JE
      Modeling the initiation and progression of human acute leukemia in mice.

      Human models of pediatric leukemia

      One of the most significant constraints on the functional study of pediatric leukemias, aside from their rarity, is the fact that there is typically very little material available for studies. To overcome this challenge, immunocompromised (NOD-SCID, NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ)) mice have been used to expand these limited tumor cells through xenografting of primary patient cells. In the case of both ALL [
      • Lock RB
      • Liem N
      • Farnsworth ML
      • et al.
      The nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model of childhood acute lymphoblastic leukemia reveals intrinsic differences in biologic characteristics at diagnosis and relapse.
      ,
      • Baersch G
      • Mollers T
      • Hotte A
      • et al.
      Good engraftment of B-cell precursor ALL in NOD-SCID mice.
      ] and AML [
      • Krivtsov AV
      • Wang X
      • Farnoud NR
      • et al.
      Patient derived xenograft (PDX) models recapitulate the genomic-driver composition of acute leukemia samples.
      ], a leukemia similar to the patient sample can be expanded. These PDXs have differing success rates (60–75% [
      • Xujun W
      • Farnoud NR
      • Hadler M
      • et al.
      Patient derived xenograft (PDX) models faithfully recapitulate the genetic composition of primary AML.
      ,
      • Wang K
      • Sanchez-Martin M
      • Wang X
      • et al.
      Patient-derived xenotransplants can recapitulate the genetic driver landscape of acute leukemias.
      ,
      • Vick B
      • Rothenberg M
      • Sandhofer N
      • et al.
      An advanced preclinical mouse model for acute myeloid leukemia using patients' cells of various genetic subgroups and in vivo bioluminescence imaging.
      ]) based on the permissibility of the recipient mice [
      • Goyama S
      • Wunderlich M
      • Mulloy JC
      Xenograft models for normal and malignant stem cells.
      ,
      • Byrne AT
      • Alferez DG
      • Amant F
      • et al.
      Interrogating open issues in cancer precision medicine with patient-derived xenografts.
      ], the specific leukemia sample used, and the presence of cytokines and growth factors that can influence engraftment success [
      • Wunderlich M
      • Chou FS
      • Link KA
      • et al.
      AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF.
      ]. Such PDX models present multiple unique advantages, both in fundamental research and in a clinical setting, including the ability to preserve stromal components in the context of orthotopic engraftment and the ability to be used in in vivo functional studies including screening of drugs capable of targeting the patient's tumor [
      • Byrne AT
      • Alferez DG
      • Amant F
      • et al.
      Interrogating open issues in cancer precision medicine with patient-derived xenografts.
      ,
      • Pompili L
      • Porru M
      • Caruso C
      • Biroccio A
      • Leonetti C
      Patient-derived xenografts: a relevant preclinical model for drug development.
      ,
      • Weroha SJ
      • Becker MA
      • Enderica-Gonzalez S
      • et al.
      Tumorgrafts as in vivo surrogates for women with ovarian cancer.
      ]. Given the potential value of PDX samples, it is not surprising that numerous academic and commercial repositories have been established, including ProXe [
      • Townsend EC
      • Murakami MA
      • Christodoulou A
      • et al.
      The public repository of xenografts enables discovery and randomized phase II-like trials in mice.
      ], HuBase (CrownBio), EurOPDX [
      • Hidalgo M
      • Amant F
      • Biankin AV
      • et al.
      Patient-derived xenograft models: an emerging platform for translational cancer research.
      ], and PROPEL (St Jude Children's hospital), along with meta-search services such as PDXFinder [
      • Conte N
      • Mason J
      • Halmagyi C
      • et al.
      PDX finder: a portal for patient-derived tumor xenograft model discovery.
      ] (covering almost 2,000 available PDXs in eight repositories). These repositories, some of which include pediatric leukemia PDXs, allow researchers to easily obtain viable tumor samples, many of which have also undergone significant initial molecular or genomic characterization. These collections of PDX tumors, searchable by tumor type, specific mutations present, and drug dosing information, give researchers unprecedented power in selecting relevant tumor samples for functional tests.
      Although PDXs represent a powerful resource, questions remain as to the extent to which selective conditions within the mouse alter the phenotype of the original tumor and how stable the tumor genotype remains. For example, recent large-scale studies have reported a high frequency of copy number alternations within four cell passages, along with expansion of minor clones within the initial tumor [
      • Ben-David U
      • Ha G
      • Tseng YY
      • et al.
      Patient-derived xenografts undergo mouse-specific tumor evolution.
      ]. And although in AML PDXs, a majority of driver mutations remain concordant with the primary tumor (even when serial transplantations are performed [
      • Vick B
      • Rothenberg M
      • Sandhofer N
      • et al.
      An advanced preclinical mouse model for acute myeloid leukemia using patients' cells of various genetic subgroups and in vivo bioluminescence imaging.
      ]), some somatic variants present at a low frequency in the primary tumors (e.g., variant allele frequency: 10–30%) exhibit large changes (>two fold) in PDXs [
      • Wang K
      • Sanchez-Martin M
      • Wang X
      • et al.
      Patient-derived xenotransplants can recapitulate the genetic driver landscape of acute leukemias.
      ]. Despite these observations, PDXs clearly provide a complementary approach to functional studies of patient tumors, and it remains to be seen how such alterations relate to similar changes seen during tumor evolution, driven by the selective pressure of patient treatment.
      Patient-derived xenograft models such as those described above have been useful in removing some experimental constraints; however, the use of xenografts still presents a limitation with respect to accurately reproducing the human bone marrow niche and immune environment, which can play a critical role in tumor development and progression [
      • Lamble AJ
      • Lind EF.
      Targeting the immune microenvironment in acute myeloid leukemia: a focus on T cell immunity.
      ,
      • Kumar B
      • Garcia M
      • Weng L
      • et al.
      Acute myeloid leukemia transforms the bone marrow niche into a leukemia-permissive microenvironment through exosome secretion.
      ,
      • Lane SW
      • Scadden DT
      • Gilliland DG
      The leukemic stem cell niche: current concepts and therapeutic opportunities.
      ]. Moreover, although the engraftment of human peripheral blood mononuclear cells (PBMCs) or reactive T cells can provide some immune activity against the human tumor, the lack of human antigen presenting cells (APCs) and development of host-versus-graft responses limit the value of this approach. This limitation has been addressed by the development of a humanized mouse model, where sublethally irradiated immunodeficient mice received transplanted fetal thymic tissue (FTHY) and CD34+ fetal liver cells (FLCs) [
      • Brodeur GM
      • Nichols KE
      • Plon SE
      • Schiffman JD
      • Malkin D
      Pediatric cancer predisposition and surveillance: an overview, and a tribute to Alfred G.
      ]. Such humanized mice, when challenged by KMT2A–MLLT3-bearing leukemia cells, exhibited an immune response similar to that seen in patients while eliciting a more representative immunological response to the disease [
      • Iorio F
      • Knijnenburg TA
      • Vis DJ
      • et al.
      A landscape of pharmacogenomic interactions in cancer.
      ].
      Given the species differences between murine and human cells and the potential genetic differences induced through the generation of human PDXs, the ability to generate models of pediatric leukemia starting from normal human HSPCs would represent an attractive solution. Interestingly, similar genetic approaches using human hematopoietic progenitor cells have been slow to develop because of the technical challenges involved in achieving the levels of transformation seen in mouse progenitor cells. Although the differences responsible have not been completely defined, studies looking at different HSPC populations indicate that HSPC transformational competency decreases in cells with age (cord blood vs. bone marrow) [
      • Horton SJ
      • Jaques J
      • Woolthuis C
      • et al.
      MLL-AF9-mediated immortalization of human hematopoietic cells along different lineages changes during ontogeny.
      ] and location (bone marrow vs. peripheral blood) [
      • Horton SJ
      • Jaques J
      • Woolthuis C
      • et al.
      MLL-AF9-mediated immortalization of human hematopoietic cells along different lineages changes during ontogeny.
      ]. This observation also agrees with age-dependent changes in the competence of specific progenitor cells to be transformed in the case of pediatric leukemias that are extremely rare in adults (e.g. AMKL [
      • Gruber TA
      • Downing JR.
      The biology of pediatric acute megakaryoblastic leukemia.
      ]).
      One of the first successful demonstrations of a human model leukemia was performed using potent KMT2A fusions such as MLLT3 [
      • Barabe F
      • Kennedy JA
      • Hope KJ
      • Dick JE
      Modeling the initiation and progression of human acute leukemia in mice.
      ]. The derived leukemias phenotypically resembled patient samples and were transplantable but, unlike primary AMLs, were also able to be maintained in in vitro culture for extended periods. Subsequent detailed genomics studies on similar single donor-derived leukemias also confirmed the generally held hypothesis that the fusion gene itself was sufficient to generate the leukemia and that no recurrent secondary mutations were required [
      • Barabe F
      • Gil L
      • Celton M
      • et al.
      Modeling human MLL-AF9 translocated acute myeloid leukemia from single donors reveals RET as a potential therapeutic target.
      ]. Other human leukemia model fusions have not only illustrated the importance of the fusion partner in the leukemia [
      • Buechele C
      • Breese EH
      • Schneidawind D
      • et al.
      MLL leukemia induction by genome editing of human CD34+ hematopoietic cells.
      ,
      • Bueno C
      • Montes R
      • Melen GJ
      • et al.
      A human ESC model for MLL–AF4 leukemic fusion gene reveals an impaired early hematopoietic-endothelial specification.
      ], but have also clearly highlighted differences between murine and human models with the same oncogene [
      • Lin S
      • Luo RT
      • Ptasinska A
      • et al.
      Instructive role of MLL-fusion proteins revealed by a model of t(4;11) pro-B acute lymphoblastic leukemia.
      ], underlining the importance of human model systems. As with the murine models, many of the oncogenic fusions used in initial human models involved the KMT2A gene, which has strong oncogenic potential. Although other fusions have been tested in human models, such as the NUP98–HOXD13 fusion gene [
      • Imren S
      • Heuser M
      • Gasparetto M
      • et al.
      Modeling de novo leukemogenesis from human cord blood with MN1 and NUP98HOXD13.
      ] for AML and the RBM15–MKL1 fusion gene [
      • Ayllon V
      • Vogel-Gonzalez M
      • Gonzalez-Pozas F
      • et al.
      New hPSC-based human models to study pediatric acute megakaryoblastic leukemia harboring the fusion oncogene RBM15-MKL1.
      ] in the case of AMKL, these models have exhibited partial or weak leukemia penetrance or dependence on other genetic alterations to generate a leukemia. Given the fact that human HSPCs seem to be more resistant to transformation compared with their murine equivalent, it remains unclear how straightforward the development of other human models will be, even given the technical advances demonstrated through direct genome editing. Nevertheless, the existing models of pediatric leukemia indicate that with optimized constructs and culture conditions, it is possible to generate a human model of the disease that, importantly, closely recapitulates its behavior and phenotype.
      With the development of both murine and human models of pediatric leukemia, there has been a great interest in using these models for functional studies to uncover the genetic underpinnings of the disease. Although space constraints preclude including an exhaustive list of these studies, a representative sample is presented in Table 2. Overall, these efforts have provided highly valuable insight into the molecular mechanisms involved in specific models of leukemia and have resulted in the advancement of several specific compounds to clinical trials [
      • Stein EM
      • Garcia-Manero G
      • Rizzieri DA
      • et al.
      The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia.
      ,
      • Berthon C
      • Raffoux E
      • Thomas X
      • et al.
      Bromodomain inhibitor OTX015 in patients with acute leukaemia: a dose-escalation, phase 1 study.
      ,
      • Shukla N
      • Wetmore C
      • Brien MM
      • et al.
      Final report of phase 1 study of the DOT1L inhibitor, pinometostat (EPZ-5676), in children with relapsed or refractory MLL-r acute leukemia.
      ]. At the same time, given the variety of screening approaches applied to the models (including siRNA and shRNA knockdown, or CRISPR-based screens), it is perhaps not surprising that the consistency of essential targets identified is relatively low. In fact, this diversity recapitulates the range of inhibitory responses seen in adult primary AML samples when challenged with a diverse collection of small molecules [
      • Baccelli I
      • Krosl J
      • Boucher G
      • et al.
      A novel approach for the identification of efficient combination therapies in primary human acute myeloid leukemia specimens.
      ]. If the underlying heterogeneity of somatic and germline variants in AML patients does explain the diversity of response seen, it suggests that although models of pediatric leukemia will be useful for dissecting specific pathways, their ability to identify “universal” therapeutic targets may be inherently limited. Moreover, the model leukemias used for such functional studies are heavily biased toward the subset in which strong oncogenic drivers have been validated. As a result, their ability to uncover relevant therapeutic targets for the range of other pediatric subgroups (that are perhaps less frequent but have clinical outcomes that are just as poor) remains unclear. These challenges notwithstanding, the ability to reproducibly generate a range of experimentally tractable human model leukemias, from multiple HSPC sources that could potentially be genetically manipulated beforehand [
      • Bak RO
      • Dever DP
      • Porteus MH
      CRISPR/Cas9 genome editing in human hematopoietic stem cells.
      ], will provide an essential resource that may provide more flexibility than the generation of PDXs in which the engraftment is variable.
      Table 2Screens performed using human or mouse leukemia models
      SpeciesYearFormatMethodPhenotype of cells screenedScreen hitsReference
      Mouse2018In vitro/in vivoCRISPRAMLDCPS
      • Yamauchi T
      • Masuda T
      • Canver MC
      • et al.
      Genome-wide CRISPR-Cas9 screen identifies leukemia-specific dependence on a pre-mRNA metabolic pathway regulated by DCPS.
      2017In vivoCRISPRMLL–AF9, HOXA9, Meis1B4galt1
      • Mercier F
      • Shi J
      • Sykes DB
      • et al.
      A genome-wide, in vivo, dropout CRISPR screen in acute myeloid leukemia identifies an essential role for beta-galactosylation in leukemic cell homing.
      2015In vitro/in vivoshRNAB‐ALLPhf6
      • Meacham CE
      • Lawton LN
      • Soto-Feliciano YM
      • et al.
      A genome-scale in vivo loss-of-function screen identifies Phf6 as a lineage-specific regulator of leukemia cell growth.
      2014In vitroshRNAMLL–AF9, HOXA9, Meis1Jmjd1c
      • Sroczynska P
      • Cruickshank VA
      • Bukowski JP
      • et al.
      shRNA screening identifies JMJD1C as being required for leukemia maintenance.
      2013In vivoSmall moleculeMLL–AF9Lovastatin
      • Hartwell KA
      • Miller PG
      • Mukherjee S
      • et al.
      Niche-based screening identifies small-molecule inhibitors of leukemia stem cells.
      2013In vivoshRNAMLL–AF9Itgb3
      • Miller PG
      • Al-Shahrour F
      • Hartwell KA
      • et al.
      In vivo RNAi screening identifies a leukemia-specific dependence on integrin beta 3 signaling.
      2011In vitro/in vivoSmall moleculeMLL–AF9/ENLBrd3/4
      • Dawson MA
      • Prinjha RK
      • Dittmann A
      • et al.
      Inhibition of BET recruitment to chromatin as an effective treatment for MLL-fusion leukaemia.
      2011In vitroshRNAMLL–AF9Brd4
      • Zuber J
      • Shi J
      • Wang E
      • et al.
      RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia.
      Human2018In vitroshRNAMLL–AF9CHD4
      • Heshmati Y
      • Turkoz G
      • Harisankar A
      • et al.
      The chromatin-remodeling factor CHD4 is required for maintenance of childhood acute myeloid leukemia.
      2017In vitroshRNAMLL–AF9ZEB2
      • Li H
      • Mar BG
      • Zhang H
      • et al.
      The EMT regulator ZEB2 is a novel dependency of human and murine acute myeloid leukemia.
      2017In vitroSmall moleculeAMLDiverse
      • Baccelli I
      • Krosl J
      • Boucher G
      • et al.
      A novel approach for the identification of efficient combination therapies in primary human acute myeloid leukemia specimens.
      2017In vitro/in vivoshRNAAMLRET
      • Barabe F
      • Gil L
      • Celton M
      • et al.
      Modeling human MLL-AF9 translocated acute myeloid leukemia from single donors reveals RET as a potential therapeutic target.
      2017In vitroCRISPRAMLPREX1
      • Wang T
      • Yu H
      • Hughes NW
      • et al.
      Gene essentiality profiling reveals gene networks and synthetic lethal interactions with oncogenic Ras.
      2016In vitroCRISPRFive leukemia cell linesDOT1L, BCL2, Men1
      • Tzelepis K
      • Koike-Yusa H
      • De Braekeleer E
      • et al.
      A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia.
      2015In vitroRNAiAMLBNIPL1, ROCK1, RPS13, STK3, SNX27, WDHD1
      • Wermke M
      • Camgoz A
      • Paszkowski-Rogacz M
      • et al.
      RNAi profiling of primary human AML cells identifies ROCK1 as a therapeutic target and nominates fasudil as an antileukemic drug.
      2012In vitroshRNAAMLWEE1
      • Porter CC
      • Kim J
      • Fosmire S
      • et al.
      Integrated genomic analyses identify WEE1 as a critical mediator of cell fate and a novel therapeutic target in acute myeloid leukemia.
      2012In vitroshRNAAMLGSK‐3α
      • Banerji V
      • Frumm SM
      • Ross KN
      • et al.
      The intersection of genetic and chemical genomic screens identifies GSK-3alpha as a target in human acute myeloid leukemia.

      Conclusions

      Given the limits of either the static information from patient cohorts, or murine models that, in some cases, do not recapitulate all aspects of the human disease, the need to create new models that more closely reflect human cancer biology is clear. Such tools will be essential in understanding the stepwise requirements for leukemogenic transformation that will underpin the next generation of rationally designed, and genotype-specific, treatments. Many technical challenges remain, however, in identifying the genetic requirements of individual fusions to transform human progenitor cells to the correct phenotype. For this, the availability of both patient-derived xenograft models and engineered human model leukemias provides a range of biological tools to explore the functional dependencies. Whether PDX models or engineered leukemias are “superior” remains something of an open question, and so for the moment, it is most useful to view them simply as complementary systems, each well suited to solving different problems related to our incomplete understanding of leukemia biology and novel therapeutic avenues.

      Acknowledgements

      This work was supported through grant funding from the Canadian Cancer Society Research Institute Grant IMP-17 (BTW, SC, FB), the Cole Foundation (BTW, SC), La Fondation du Centre de Cancérologie Charles Bruneau and La Fondation CHU Sainte-Justine (SC).
      We acknowledge members of the Wilhelm and Cellot labs for critical comments on the manuscript during revision.

      References

        • Appelbaum FR
        • Gundacker H
        • Head DR
        • et al.
        Age and acute myeloid leukemia.
        Blood. 2006; 107: 3481-3485
        • Estey E
        • Dohner H
        Acute myeloid leukaemia.
        Lancet. 2006; 368: 1894-1907
        • Juliusson G
        • Antunovic P
        • Derolf A
        • et al.
        Age and acute myeloid leukemia: real world data on decision to treat and outcomes from the Swedish Acute Leukemia Registry.
        Blood. 2009; 113: 4179-4187
        • Phekoo KJ
        • Richards MA
        • Moller H
        • Schey SA
        • South Thames Haematology Specialist Committee
        The incidence and outcome of myeloid malignancies in 2,112 adult patients in southeast England.
        Haematologica. 2006; 91: 1400-1404
        • Steliarova-Foucher E
        • Colombet M
        • Ries LAG
        • et al.
        International incidence of childhood cancer, 2001–10: a population-based registry study.
        Lancet Oncol. 2017; 18: 719-731
        • Siegel DA
        • King J
        • Tai E
        • Buchanan N
        • Ajani UA
        • Li J
        Cancer incidence rates and trends among children and adolescents in the United States, 2001–2009.
        Pediatrics. 2014; 134: e945-e955
        • Belson M
        • Kingsley B
        • Holmes A
        Risk factors for acute leukemia in children: a review.
        Environ Health Perspect. 2006; 115: 138-145
        • TJ Ley
        • Miller C
        • et al.
        • Cancer Genome Atlas Research Network
        Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia.
        N Engl J Med. 2013; 368: 2059-2074
        • Welch JS
        • Ley TJ
        • Link DC
        • et al.
        The origin and evolution of mutations in acute myeloid leukemia.
        Cell. 2012; 150: 264-278
        • Bolouri H
        • Farrar JE
        • Triche Jr, T
        • et al.
        The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions.
        Nat Med. 2018; 24: 103-112
        • Liu YF
        • Wang BY
        • Zhang WN
        • et al.
        Genomic profiling of adult and pediatric B-cell acute lymphoblastic leukemia.
        EBioMedicine. 2016; 8: 173-183
        • Mussai FJ
        • Yap C
        • Mitchell C
        • Kearns P
        Challenges of clinical trial design for targeted agents against pediatric leukemias.
        Front Oncol. 2014; 4: 374
        • National Cancer Institute (NCI)
        SEER cancer statistics review.
        Author, Bethesda, MD2017
        • Barretina J
        • Caponigro G
        • Stransky N
        • et al.
        The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity.
        Nature. 2012; 483: 603-607
        • Yang J
        • Li A
        • Li Y
        • Guo X
        • Wang M
        A novel approach for drug response prediction in cancer cell lines via network representation learning.
        Bioinformatics. 2019; 35: 1527-1535
        • Zhang F
        • Wang M
        • Xi J
        • Yang J
        • Li A
        A novel heterogeneous network-based method for drug response prediction in cancer cell lines.
        Sci Rep. 2018; 8: 3355
        • Challen GA
        • Boles N
        • Lin KK
        • Goodell MA
        Mouse hematopoietic stem cell identification and analysis.
        Cytometry A. 2009; 75: 14-24
        • Dykstra B
        • Ramunas J
        • Kent D
        • et al.
        High-resolution video monitoring of hematopoietic stem cells cultured in single-cell arrays identifies new features of self-renewal.
        Proc Natl Acad Sci USA. 2006; 103: 8185-8190
        • Okada S
        • Nakauchi H
        • Nagayoshi K
        • Nishikawa S
        • Miura Y
        • Suda T
        In vivo and in vitro stem cell function of c-kit- and Sca-1-positive murine hematopoietic cells.
        Blood. 1992; 80: 3044-3050
        • Dang J
        • Nance S
        • Ma J
        • et al.
        AMKL chimeric transcription factors are potent inducers of leukemia.
        Leukemia. 2017; 31: 2228-2234
        • Bijl J
        • Sauvageau M
        • Thompson A
        • Sauvageau G
        High incidence of proviral integrations in the Hoxa locus in a new model of E2a-PBX1-induced B-cell leukemia.
        Genes Dev. 2005; 19: 224-233
        • Forster A
        • Pannell R
        • Drynan LF
        • et al.
        Engineering de novo reciprocal chromosomal translocations associated with Mll to replicate primary events of human cancer.
        Cancer Cell. 2003; 3: 449-458
        • Corral J
        • Lavenir I
        • Impey H
        • et al.
        An Mll–AF9 fusion gene made by homologous recombination causes acute leukemia in chimeric mice: a method to create fusion oncogenes.
        Cell. 1996; 85: 853-861
        • Barabe F
        • Kennedy JA
        • Hope KJ
        • Dick JE
        Modeling the initiation and progression of human acute leukemia in mice.
        Science. 2007; 316: 600-604
        • TJ Hudson
        • Anderson W
        • et al.
        • International Cancer Genome Consortium
        International network of cancer genome projects.
        Nature. 2010; 464: 993-998
        • Stratton MR
        • Campbell PJ
        • Futreal PA
        The cancer genome.
        Nature. 2009; 458: 719-724
        • Hoadley KA
        • Yau C
        • Hinoue T
        • et al.
        Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer.
        Cell. 2018; 173 (291–304.e296)
        • Alexandrov LB
        • Nik-Zainal S
        • Wedge DC
        • et al.
        Signatures of mutational processes in human cancer.
        Nature. 2013; 500: 415-421
        • Shen H
        • Li J
        • Zhang J
        • et al.
        Comprehensive characterization of human genome variation by high coverage whole-genome sequencing of forty four Caucasians.
        PLoS One. 2013; 8: e59494
        • Genomes Project C
        • Abecasis GR
        • Altshuler D
        • et al.
        A map of human genome variation from population-scale sequencing.
        Nature. 2010; 467: 1061-1073
        • Sudmant PH
        • Rausch T
        • Gardner EJ
        • et al.
        An integrated map of structural variation in 2,504 human genomes.
        Nature. 2015; 526: 75-81
        • Grobner SN
        • Worst BC
        • Weischenfeldt J
        • et al.
        The landscape of genomic alterations across childhood cancers.
        Nature. 2018; 555: 321-327
        • Schwartz JR
        • Ma J
        • Lamprecht T
        • et al.
        The genomic landscape of pediatric myelodysplastic syndromes.
        Nat Commun. 2017; 8: 1557
        • Hsu CH
        • Nguyen C
        • Yan C
        • et al.
        Transcriptome profiling of pediatric core binding factor AML.
        PLoS One. 2015; 10e0138782
        • Andersson AK
        • Ma J
        • Wang J
        • et al.
        The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias.
        Nat Genet. 2015; 47: 330-337
        • Jaffe JD
        • Wang Y
        • Chan HM
        • et al.
        Global chromatin profiling reveals NSD2 mutations in pediatric acute lymphoblastic leukemia.
        Nat Genet. 2013; 45: 1386-1391
        • Roberts KG
        • Morin RD
        • Zhang J
        • et al.
        Genetic alterations activating kinase and cytokine receptor signaling in high-risk acute lymphoblastic leukemia.
        Cancer Cell. 2012; 22: 153-166
        • Ma X
        • Liu Y
        • Liu Y
        • et al.
        Pan-cancer genome and transcriptome analyses of 1,699 paediatric leukaemias and solid tumours.
        Nature. 2018; 555: 371-376
        • Brunner AM
        • Graubert TA.
        Genomics in childhood acute myeloid leukemia comes of age.
        Nat Med. 2018; 24: 7-9
        • Zelent A
        • Greaves M
        • Enver T
        Role of the TEL–AML1 fusion gene in the molecular pathogenesis of childhood acute lymphoblastic leukaemia.
        Oncogene. 2004; 23: 4275-4283
        • Mori H
        • Colman SM
        • Xiao Z
        • et al.
        Chromosome translocations and covert leukemic clones are generated during normal fetal development.
        Proc Natl Acad Sci USA. 2002; 99: 8242-8247
        • Bernt KM
        • Hunger SP.
        Current concepts in pediatric Philadelphia chromosome-positive acute lymphoblastic leukemia.
        Front Oncol. 2014; 4: 54
        • Reshmi SC
        • Harvey RC
        • Roberts KG
        • et al.
        Targetable kinase gene fusions in high-risk B-ALL: a study from the Children's Oncology Group.
        Blood. 2017; 129: 3352-3361
        • Chen B
        • Jiang L
        • Zhong ML
        • et al.
        Identification of fusion genes and characterization of transcriptome features in T-cell acute lymphoblastic leukemia.
        Proc Natl Acad Sci USA. 2018; 115: 373-378
        • Papenhausen P
        • Kelly CA
        • Zhang Z
        • Tepperberg J
        • Burnside RD
        • Schwartz S
        Multidisciplinary analysis of pediatric T-ALL: 9q34 gene fusions.
        Cancer Genet. 2019; 231–232: 1-13
        • Downing JR
        • Wilson RK
        • Zhang J
        • et al.
        The Pediatric Cancer Genome Project.
        Nat Genet. 2012; 44: 619-622
        • Churchman ML
        • Qian M
        • Te Kronnie G
        • et al.
        Germline genetic IKZF1 variation and predisposition to childhood acute lymphoblastic leukemia.
        Cancer Cell. 2018; 33 (937–948 e938)
        • Qian M
        • Cao X
        • Devidas M
        • et al.
        TP53 germline variations influence the predisposition and prognosis of B-cell acute lymphoblastic leukemia in children.
        J Clin Oncol. 2018; 36: 591-599
        • Phillips CL
        • Gerbing R
        • Alonzo T
        • et al.
        MDM2 polymorphism increases susceptibility to childhood acute myeloid leukemia: a report from the Children's Oncology Group.
        Pediatr Blood Cancer. 2010; 55: 248-253
        • Wei X
        • Calvo-Vidal MN
        • Chen S
        • et al.
        Germline lysine-specific demethylase 1 (LSD1/KDM1A) mutations confer susceptibility to multiple myeloma.
        Cancer Res. 2018; 78: 2747-2759
        • Zhang J
        • Walsh MF
        • Wu G
        • et al.
        Germline mutations in predisposition genes in pediatric cancer.
        N Engl J Med. 2015; 373: 2336-2346
        • Brodeur GM
        • Nichols KE
        • Plon SE
        • Schiffman JD
        • Malkin D
        Pediatric cancer predisposition and surveillance: an overview, and a tribute to Alfred G.
        Knudson Jr. Clin Cancer Res. 2017; 23: e1-e5
        • Iorio F
        • Knijnenburg TA
        • Vis DJ
        • et al.
        A landscape of pharmacogenomic interactions in cancer.
        Cell. 2016; 166: 740-754
        • Klijn C
        • Durinck S
        • Stawiski EW
        • et al.
        A comprehensive transcriptional portrait of human cancer cell lines.
        Nat Biotechnol. 2015; 33: 306-312
        • Garnett MJ
        • Edelman EJ
        • Heidorn SJ
        • et al.
        Systematic identification of genomic markers of drug sensitivity in cancer cells.
        Nature. 2012; 483: 570-575
        • Drexler HG
        • Quentmeier H
        • MacLeod RA
        Malignant hematopoietic cell lines: in vitro models for the study of MLL gene alterations.
        Leukemia. 2004; 18: 227-232
        • Munker R
        • Nordberg ML
        • Veillon D
        • et al.
        Characterization of a new myeloid leukemia cell line with normal cytogenetics (CG-SH).
        Leuk Res. 2009; 33: 1405-1408
        • Pabst C
        • Krosl J
        • Fares I
        • et al.
        Identification of small molecules that support human leukemia stem cell activity ex vivo.
        Nat Methods. 2014; 11: 436-442
        • Ben-David U
        • Siranosian B
        • Ha G
        • et al.
        Genetic and transcriptional evolution alters cancer cell line drug response.
        Nature. 2018; 560: 325-330
        • Moy C
        • Oleykowski CA
        • Plant R
        • et al.
        High chromosome number in hematological cancer cell lines is a negative predictor of response to the inhibition of Aurora B and C by GSK1070916.
        J Transl Med. 2011; 9: 110
        • Chang YT
        • Hernandez D
        • Alonso S
        • et al.
        Role of CYP3A4 in bone marrow microenvironment-mediated protection of FLT3/ITD AML from tyrosine kinase inhibitors.
        Blood Adv. 2019; 3: 908-916
        • Heshmati Y
        • Turkoz G
        • Harisankar A
        • et al.
        The chromatin-remodeling factor CHD4 is required for maintenance of childhood acute myeloid leukemia.
        Haematologica. 2018; 103: 1169-1181
        • Li H
        • Mar BG
        • Zhang H
        • et al.
        The EMT regulator ZEB2 is a novel dependency of human and murine acute myeloid leukemia.
        Blood. 2017; 129: 497-508
        • Porter CC
        • Kim J
        • Fosmire S
        • et al.
        Integrated genomic analyses identify WEE1 as a critical mediator of cell fate and a novel therapeutic target in acute myeloid leukemia.
        Leukemia. 2012; 26: 1266-1276
        • Wang T
        • Yu H
        • Hughes NW
        • et al.
        Gene essentiality profiling reveals gene networks and synthetic lethal interactions with oncogenic Ras.
        Cell. 2017; 168 (890–903.e815)
        • Tzelepis K
        • Koike-Yusa H
        • De Braekeleer E
        • et al.
        A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia.
        Cell Rep. 2016; 17: 1193-1205
        • Daley GQ
        • Van Etten RA
        • Baltimore D
        Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome.
        Science. 1990; 247: 824-830
        • Kelliher MA
        • McLaughlin J
        • Witte ON
        • Rosenberg N
        Induction of a chronic myelogenous leukemia-like syndrome in mice with v-abl and BCR/ABL.
        Proc Natl Acad Sci USA. 1990; 87: 6649-6653
        • Winters AC
        • Bernt KM.
        MLL-rearranged leukemias—an update on science and clinical approaches.
        Front Pediatr. 2017; 5: 4
        • Meyer C
        • Burmeister T
        • Groger D
        • et al.
        The MLL recombinome of acute leukemias in 2017.
        Leukemia. 2018; 32: 273-284
        • Dykstra B
        • Kent D
        • Bowie M
        • et al.
        Long-term propagation of distinct hematopoietic differentiation programs in vivo.
        Cell Stem Cell. 2007; 1: 218-229
        • Metzler M
        • Forster A
        • Pannell R
        • et al.
        A conditional model of MLL–AF4 B-cell tumourigenesis using invertor technology.
        Oncogene. 2006; 25: 3093-3103
        • de Guzman CG
        • Warren AJ
        • Zhang Z
        • et al.
        Hematopoietic stem cell expansion and distinct myeloid developmental abnormalities in a murine model of the AML1–ETO translocation.
        Mol Cell Biol. 2002; 22: 5506-5517
        • Duque-Afonso J
        • Feng J
        • Scherer F
        • et al.
        Comparative genomics reveals multistep pathogenesis of E2A-PBX1 acute lymphoblastic leukemia.
        J Clin Invest. 2015; 125: 3667-3680
        • Metzeler KH
        • Herold T
        • Rothenberg-Thurley M
        • et al.
        Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia.
        Blood. 2016; 128: 686-698
        • Lock RB
        • Liem N
        • Farnsworth ML
        • et al.
        The nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model of childhood acute lymphoblastic leukemia reveals intrinsic differences in biologic characteristics at diagnosis and relapse.
        Blood. 2002; 99: 4100-4108
        • Baersch G
        • Mollers T
        • Hotte A
        • et al.
        Good engraftment of B-cell precursor ALL in NOD-SCID mice.
        Klin Padiatr. 1997; 209: 178-185
        • Krivtsov AV
        • Wang X
        • Farnoud NR
        • et al.
        Patient derived xenograft (PDX) models recapitulate the genomic-driver composition of acute leukemia samples.
        Blood. 2014; 124: 286
        • Xujun W
        • Farnoud NR
        • Hadler M
        • et al.
        Patient derived xenograft (PDX) models faithfully recapitulate the genetic composition of primary AML.
        Blood. 2013; 122: 1328
        • Wang K
        • Sanchez-Martin M
        • Wang X
        • et al.
        Patient-derived xenotransplants can recapitulate the genetic driver landscape of acute leukemias.
        Leukemia. 2017; 31: 151-158
        • Vick B
        • Rothenberg M
        • Sandhofer N
        • et al.
        An advanced preclinical mouse model for acute myeloid leukemia using patients' cells of various genetic subgroups and in vivo bioluminescence imaging.
        PLoS One. 2015; 10e0120925
        • Goyama S
        • Wunderlich M
        • Mulloy JC
        Xenograft models for normal and malignant stem cells.
        Blood. 2015; 125: 2630-2640
        • Byrne AT
        • Alferez DG
        • Amant F
        • et al.
        Interrogating open issues in cancer precision medicine with patient-derived xenografts.
        Nat Rev Cancer. 2017; 17: 254-268
        • Wunderlich M
        • Chou FS
        • Link KA
        • et al.
        AML xenograft efficiency is significantly improved in NOD/SCID-IL2RG mice constitutively expressing human SCF.
        GM-CSF and IL-3. Leukemia. 2010; 24: 1785-1788
        • Pompili L
        • Porru M
        • Caruso C
        • Biroccio A
        • Leonetti C
        Patient-derived xenografts: a relevant preclinical model for drug development.
        J Exp Clin Cancer Res. 2016; 35: 189
        • Weroha SJ
        • Becker MA
        • Enderica-Gonzalez S
        • et al.
        Tumorgrafts as in vivo surrogates for women with ovarian cancer.
        Clin Cancer Res. 2014; 20: 1288-1297
        • Townsend EC
        • Murakami MA
        • Christodoulou A
        • et al.
        The public repository of xenografts enables discovery and randomized phase II-like trials in mice.
        Cancer Cell. 2016; 29: 574-586
        • Hidalgo M
        • Amant F
        • Biankin AV
        • et al.
        Patient-derived xenograft models: an emerging platform for translational cancer research.
        Cancer Discov. 2014; 4: 998-1013
        • Conte N
        • Mason J
        • Halmagyi C
        • et al.
        PDX finder: a portal for patient-derived tumor xenograft model discovery.
        Nucleic Acids Res. 2019; 47: D1073-D1079
        • Ben-David U
        • Ha G
        • Tseng YY
        • et al.
        Patient-derived xenografts undergo mouse-specific tumor evolution.
        Nat Genet. 2017; 49: 1567-1575
        • Lamble AJ
        • Lind EF.
        Targeting the immune microenvironment in acute myeloid leukemia: a focus on T cell immunity.
        Front Oncol. 2018; 8: 213
        • Kumar B
        • Garcia M
        • Weng L
        • et al.
        Acute myeloid leukemia transforms the bone marrow niche into a leukemia-permissive microenvironment through exosome secretion.
        Leukemia. 2018; 32: 575-587
        • Lane SW
        • Scadden DT
        • Gilliland DG
        The leukemic stem cell niche: current concepts and therapeutic opportunities.
        Blood. 2009; 114: 1150-1157
        • Horton SJ
        • Jaques J
        • Woolthuis C
        • et al.
        MLL-AF9-mediated immortalization of human hematopoietic cells along different lineages changes during ontogeny.
        Leukemia. 2013; 27: 1116-1126
        • Gruber TA
        • Downing JR.
        The biology of pediatric acute megakaryoblastic leukemia.
        Blood. 2015; 126: 943-949
        • Barabe F
        • Gil L
        • Celton M
        • et al.
        Modeling human MLL-AF9 translocated acute myeloid leukemia from single donors reveals RET as a potential therapeutic target.
        Leukemia. 2017; 31: 1166-1176
        • Buechele C
        • Breese EH
        • Schneidawind D
        • et al.
        MLL leukemia induction by genome editing of human CD34+ hematopoietic cells.
        Blood. 2015; 126: 1683-1694
        • Bueno C
        • Montes R
        • Melen GJ
        • et al.
        A human ESC model for MLL–AF4 leukemic fusion gene reveals an impaired early hematopoietic-endothelial specification.
        Cell Res. 2012; 22: 986-1002
        • Lin S
        • Luo RT
        • Ptasinska A
        • et al.
        Instructive role of MLL-fusion proteins revealed by a model of t(4;11) pro-B acute lymphoblastic leukemia.
        Cancer Cell. 2016; 30: 737-749
        • Imren S
        • Heuser M
        • Gasparetto M
        • et al.
        Modeling de novo leukemogenesis from human cord blood with MN1 and NUP98HOXD13.
        Blood. 2014; 124: 3608-3612
        • Ayllon V
        • Vogel-Gonzalez M
        • Gonzalez-Pozas F
        • et al.
        New hPSC-based human models to study pediatric acute megakaryoblastic leukemia harboring the fusion oncogene RBM15-MKL1.
        Stem Cell Res. 2017; 19: 1-5
        • Stein EM
        • Garcia-Manero G
        • Rizzieri DA
        • et al.
        The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia.
        Blood. 2018; 131: 2661-2669
        • Berthon C
        • Raffoux E
        • Thomas X
        • et al.
        Bromodomain inhibitor OTX015 in patients with acute leukaemia: a dose-escalation, phase 1 study.
        Lancet Haematol. 2016; 3: e186-e195
        • Shukla N
        • Wetmore C
        • Brien MM
        • et al.
        Final report of phase 1 study of the DOT1L inhibitor, pinometostat (EPZ-5676), in children with relapsed or refractory MLL-r acute leukemia.
        Blood. 2016; 128: 2780
        • Baccelli I
        • Krosl J
        • Boucher G
        • et al.
        A novel approach for the identification of efficient combination therapies in primary human acute myeloid leukemia specimens.
        Blood Cancer J. 2017; 7: e529
        • Bak RO
        • Dever DP
        • Porteus MH
        CRISPR/Cas9 genome editing in human hematopoietic stem cells.
        Nat Protoc. 2018; 13: 358-376
        • Rodriguez-Hernandez G
        • Hauer J
        • Martin-Lorenzo A
        • et al.
        Infection exposure promotes ETV6–RUNX1 precursor b-cell leukemia via impaired H3K4 demethylases.
        Cancer Res. 2017; 77: 4365-4377
        • Deshpande AJ
        • Chen L
        • Fazio M
        • et al.
        Leukemic transformation by the MLL–AF6 fusion oncogene requires the H3K79 methyltransferase Dot1l.
        Blood. 2013; 121: 2533-2541
        • Bernardin F
        • Yang Y
        • Cleaves R
        • et al.
        TEL-AML1, expressed from t(12;21) in human acute lymphocytic leukemia, induces acute leukemia in mice.
        Cancer Res. 2002; 62: 3904-3908
        • Mercher T
        • Raffel GD
        • Moore SA
        • et al.
        The OTT–MAL fusion oncogene activates RBPJ-mediated transcription and induces acute megakaryoblastic leukemia in a knockin mouse model.
        J Clin Invest. 2009; 119: 852-864
        • Krivtsov AV
        • Feng Z
        • Lemieux ME
        • et al.
        H3K79 methylation profiles define murine and human MLL–AF4 leukemias.
        Cancer Cell. 2008; 14: 355-368
        • Chen W
        • Li Q
        • Hudson WA
        • Kumar A
        • Kirchhof N
        • Kersey JH
        A murine Mll-AF4 knock-in model results in lymphoid and myeloid deregulation and hematologic malignancy.
        Blood. 2006; 108: 669-677
        • Schessl C
        • Rawat VP
        • Cusan M
        • et al.
        The AML1–ETO fusion gene and the FLT3 length mutation collaborate in inducing acute leukemia in mice.
        J Clin Invest. 2005; 115: 2159-2168
        • Pineault N
        • Buske C
        • Feuring-Buske M
        • et al.
        Induction of acute myeloid leukemia in mice by the human leukemia-specific fusion gene NUP98–HOXD13 in concert with Meis1.
        Blood. 2003; 101: 4529-4538
        • Kroon E
        • Thorsteinsdottir U
        • Mayotte N
        • Nakamura T
        • Sauvageau G
        NUP98–HOXA9 expression in hemopoietic stem cells induces chronic and acute myeloid leukemias in mice.
        EMBO J. 2001; 20: 350-361
        • Schneidawind C
        • Jeong J
        • Schneidawind D
        • et al.
        MLL leukemia induction by t(9;11) chromosomal translocation in human hematopoietic stem cells using genome editing.
        Blood Adv. 2018; 2: 832-845
        • Moriya K
        • Suzuki M
        • Watanabe Y
        • et al.
        Development of a multi-step leukemogenesis model of MLL-rearranged leukemia using humanized mice.
        PLoS One. 2012; 7: e37892
        • Montes R
        • Ayllon V
        • Gutierrez-Aranda I
        • et al.
        Enforced expression of MLL–AF4 fusion in cord blood CD34+ cells enhances the hematopoietic repopulating cell function and clonogenic potential but is not sufficient to initiate leukemia.
        Blood. 2011; 117: 4746-4758
        • Rizo A
        • Horton SJ
        • Olthof S
        • et al.
        BMI1 collaborates with BCR–ABL in leukemic transformation of human CD34+ cells.
        Blood. 2010; 116: 4621-4630
        • Wei J
        • Wunderlich M
        • Fox C
        • et al.
        Microenvironment determines lineage fate in a human model of MLL-AF9 leukemia.
        Cancer Cell. 2008; 13: 483-495
        • Yamauchi T
        • Masuda T
        • Canver MC
        • et al.
        Genome-wide CRISPR-Cas9 screen identifies leukemia-specific dependence on a pre-mRNA metabolic pathway regulated by DCPS.
        Cancer Cell. 2018; 33 (386–400.e385)
        • Mercier F
        • Shi J
        • Sykes DB
        • et al.
        A genome-wide, in vivo, dropout CRISPR screen in acute myeloid leukemia identifies an essential role for beta-galactosylation in leukemic cell homing.
        Blood. 2017; 130: 2493
        • Meacham CE
        • Lawton LN
        • Soto-Feliciano YM
        • et al.
        A genome-scale in vivo loss-of-function screen identifies Phf6 as a lineage-specific regulator of leukemia cell growth.
        Genes Dev. 2015; 29: 483-488
        • Sroczynska P
        • Cruickshank VA
        • Bukowski JP
        • et al.
        shRNA screening identifies JMJD1C as being required for leukemia maintenance.
        Blood. 2014; 123: 1870-1882
        • Hartwell KA
        • Miller PG
        • Mukherjee S
        • et al.
        Niche-based screening identifies small-molecule inhibitors of leukemia stem cells.
        Nat Chem Biol. 2013; 9: 840-848
        • Miller PG
        • Al-Shahrour F
        • Hartwell KA
        • et al.
        In vivo RNAi screening identifies a leukemia-specific dependence on integrin beta 3 signaling.
        Cancer Cell. 2013; 24: 45-58
        • Dawson MA
        • Prinjha RK
        • Dittmann A
        • et al.
        Inhibition of BET recruitment to chromatin as an effective treatment for MLL-fusion leukaemia.
        Nature. 2011; 478: 529-533
        • Zuber J
        • Shi J
        • Wang E
        • et al.
        RNAi screen identifies Brd4 as a therapeutic target in acute myeloid leukaemia.
        Nature. 2011; 478: 524-528
        • Wermke M
        • Camgoz A
        • Paszkowski-Rogacz M
        • et al.
        RNAi profiling of primary human AML cells identifies ROCK1 as a therapeutic target and nominates fasudil as an antileukemic drug.
        Blood. 2015; 125: 3760-3768
        • Banerji V
        • Frumm SM
        • Ross KN
        • et al.
        The intersection of genetic and chemical genomic screens identifies GSK-3alpha as a target in human acute myeloid leukemia.
        J Clin Invest. 2012; 122: 935-947