An Evolutionary Perspective on Chronic Myelomonocytic Leukemia

SPOTLIGHT REVIEW An evolutionary perspective on chronic myelomonocytic leukemia

R Itzykson1,2,3 and E Solary1,2,3

Chronic myelomonocytic leukemia (CMML) shares with other myeloid diseases a number of somatic gene mutations. These mutations can now be integrated within the framework of evolution theory to address the mechanisms of the disease. Several evidences indicate that the disease emerges in adult hematopoietic stem cells (HSC) through the age-dependent accumulation of DNA damage, leading stochastically to a driver mutation that confers a competitive advantage to the cell. A mutation in TET2 gene could be one of these driver mutations provoking the emergence of clonality. After a long latency, secondary lesions, such as mutations in the SRSF2 gene, contribute to progression to full-blown malignancy, with abnormal differentiation. Additional mutations accumulate and branching arising mostly through mitotic recombination generates clonal heterogeneity. Modifications in the microenvironment probably affect this clonal dynamics, whereas epigenetic alterations, such as hypermethylation of the TIF1g gene promoter, may generate phenotypic diversification of otherwise clonal populations. The preserved although deregulated myeloid differentiation that characterizes CMML, with granulomonocyte expansion and various cytopenias, may depend on early clonal dominance in the hematopietic cell hierarchy. Progression to acute myeloid leukemia observed in 25–30% of the patients may arise from the massive expansion of a clone with novel genetic lesions, providing a high fitness to previously minor subclones when in chronic phase of the disease. This review discusses the various models of disease emergence and progression and how this recent knowledge could drive rational therapeutic strategies.

Leukemia (2013) 27, 1441–1450; doi:10.1038/leu.2013.100 Keywords: chronic myelomonocytic leukemia; evolution; clone architecture; gene mutations; epigenetics

INTRODUCTION immunophenotypic and functional observations, now require 16,17 The WHO classification of myeloid malignancies describes chronic integration with the concepts of evolution theory. myelomonocytic leukemia (CMML) as the most frequent myelo- In CMML, the first results of whole-exome sequencing indicate dysplastic syndrome/myeloproliferative neoplasm (MDS/MPN).1 that the number of non-synonymous mutations in coding regions 12,18,19 As MPN and MDS, this disease is characterized by a clonal varies between 5 and 20 per patient, (Merlevede et al., (or sometimes oligoclonal) hematopoiesis with abnormal myeloid unpublished results). Cytogenetic abnormalities have been found 20 21 differentiation. In CMML, myeloid differentiation is either in 30%, and copy-neutral acquired uniparental disomies in 50%, 13 dysplastic, leading to cytopenias, or exaggerated, leading to of cases. Yet, the precise extent of non-exonic genetic, as well as 22–24 myeloproliferation, or both. All these myeloid neoplasms share a epigenetic abnormalities remains unknown. In the present variable risk of transformation to acute myeloid leukemia (AML). review, we integrate recent findings on the biology of CMML They also share somatic gene mutations that target the epigenetic within the conceptual framework of evolution theory to address control of transcription, such as histone modification (ASXL1, EZH2, the mechanisms of disease initiation, progression, phenotype UTX), DNA methylation (TET2, DNTM3A), or both (IDH1, IDH2), and specification and transformation, and discuss how these recent the cohesin complex components STAG2, SMC3, SMC1A and insights could support the design of rational therapeutic strategies. SPOTLIGHT RAD21), cytokine signaling (JAK2, MPL, LNK, CBL, RAS and genes of the Notch2 pathway) and pre-mRNA splicing (SRSF2, SF3B1, U2AF1, ZRSR2). These mutations and the current understanding of their HEMATOPOIETIC STEM CELLS (HSC) AS CELLS-OF-ORIGIN AND molecular consequences have been extensively reviewed in the LIC FOR CMML recent years.2–6 There prevalence, phenotypic association and There are two main evidences that adult HSC are the normal cells, prognostic significance are summarized in Table 1. in which initial genetic lesions have arisen in CMML. First, gene Starting with Peter Nowell,7 cancer has long been envisioned as mutations are present in sorted HSC, as defined by their CD34 þ / an evolutionary, Darwinian process driven by the sequential CD38 À /CD90 þ phenotype.25 Similar findings have been made acquisition of genetic lesions followed by clonal expansion and with chromosomal lesions in MDS,26 and with JAK2 and TET2 selection.8 The advent of high throughput genetic analyses,9 mutations in MPN.27 Though the limited SCID-repopulating single-cell assays,10 and xenotransplantation assays,11 has activity of CMML has hampered the functional validation of revealed the genetic heterogeneity of some myeloid neoplasms, these phenotypically-defined HSC,28 they contain most of the with frequent oligoclonality,12,13 and the functional heterogeneity serial replating potential of CD34 þ cells in vitro.25 of leukemic cells, including those defined as leukemia-initiating Second, a fraction of T cells belongs to the leukemic clone, cells (LIC).14,15 These new findings that reconcile genetic, which thus has lympho-myeloid potential. Again, a similar

1Department of Hematology, Inserm UMR1009, Institut Gustave Roussy, Villejuif, France; 2IFR54, Institut Gustave Roussy, Villejuif, France and 3Faculty of Medicine, University Paris- Sud 11, Le Kremlin-Biceˆtre, France. Correspondence: Professor E Solary, Department of Hematology, Inserm UMR1009, Institut Gustave Roussy, 114, Rue Edouard Vaillant, 94805 Villejuif, France. E-mail: [email protected] Received 13 January 2013; revised 29 March 2013; accepted 29 March 2013; accepted article preview online 5 April 2013; advance online publication, 3 May 2013 Evolution and CMML clone architecture R Itzykson and E Solary 1442 Table 1. List of main genes whose recurrent mutation has been observed in CMML cells

Gene Main function(s) of the protein Phenotypic correlations Prognostic relevance Mutated (%)

TET2 Epigenetics (methylcytosine dioxygenase 2) Thrombocytopenia 40–60 SRSF2 Pre-mRNA splicing factor Thrombocytopenia Poor, univariate 30–50 ASXL1 Epigenetics (PRC2-mediated H3K27me3) Anemia, EMDa Poor, multivariate B40 RUNX1 Transcription factor Thrombocytopenia Poor, univariate B15 NRAS Signal transduction protein (small GTPase) Hyperleukocytosis; EMD Poor, univariate B10 CBL Signal transduction protein (E3 ligase) Hyperleukocytosis; EMD Poor, univariate B10 JAK2 Signal transduction protein (tyrosine kinase) Hyperleukocytosis 5–10 KRAS Signal transduction protein (small GTPase) Hyperleukocytosis; EMD 5–10 ZRSF2 Pre-mRNA splicing factor 5–10 IDH2 Oxido-reductase (isocitrate deshydrogenase) Poor, univariate 5–10 SF3B1 Pre-mRNA splicing factor Ring sideroblasts Poor, univariate 5–10 SETBP1 Protein that binds SET, involved in DNA replication AML transformation 5–10 UTX (KDM6A) Epigenetics (H3K27 demethylase) 5–10 EZH2 Histone methylase (PRC2 complex) o5 FLT3 Signal transduction protein (tyrosine kinase) o5 DNMT3A DNA (cytosine 5-) methyltransferase 3 alpha o5 U2AF1 Pre-mRNA splicing factor o5 NOTCH2b Signal transduction protein o5 NPM1 Ribosome biogenesis, centrosome duplication and others AML transformation 1 IDH1 Oxido-reductase (isocitrate deshydrogenase) 1 Cohesin genesc Regulation of gene transcription 1 TP53 Tumor suppressor protein AML transformation 1

SPOTLIGHT aEMD: extramedullary disease. bSomatic heterozygous mutations in multiple Notch pathway genes including NCSTN, APH1, MAML1 and NOTCH2. cCohesin complex components STAG2, SMC3, SMC1A and RAD21.

observation was made in MPN, in which mutations in JAK229 and independently,43 and private mutations may not necessarily be in TET230 were identiﬁed also in some lymphoid cells. This ‘passengers’, they can be used as probes or ‘molecular clocks’ to suggests that a normal self-renewal hierarchy, with HSC serving as measure the mutational rate in HSC, which has been estimated to LIC, is maintained in CMML hematopoiesis, which is different from be 0.1 mutation per HSC and per year.13 Such a rate is in keeping de novo31 or secondary32 AML, in which progenitors are candidate with both the number of single nucleotide variant in CMML and LIC. Though this notion needs strengthening in CMML, it raises other myeloid neoplasms, and the estimated number and division issues regarding the contribution of HSC niches,33 compartmental rate of normal human HSC,44 further indicating that stochastic heterogeneity34 and physiological aging,35 to the emergence of consequences of aging, rather than genetic instability, may be at the disease. the origin of these diseases.

INITIATION OF CMML: EARLY GENETIC LESIONS CMML INITIATION: EMERGENCE OF CLONALITY Accumulation of DNA damage depends on cell replication and At some point, a lesion conferring a competitive advantage to the metabolic activity and is therefore limited in quiescent and HSC can occur. Such ‘competitive advantage’ is best expressed in hypoxic HSC.36 DNA damage that nevertheless occurs in a HSC can terms of ‘fitness’, a concept brought from evolution theory to the trigger the p5337,38 and p1639 checkpoints required for the cancer field8,16 and measuring the proliferation rate of a maintenance of self-renewal as well as survival. If DNA damage homogeneous population compared with that of a reference accumulates, HSC are either driven to exhaustion through population. The fitness of a preleukemic or leukemic clone is increased differentiation,37 or outcompeted by other HSC relative to that of other clones and of residual normal HSC. Fitness because of impaired self-renewal.40 It is unlikely that a defect in is also context-dependent, for example, is affected by changes in those checkpoints has a role in CMML initiation as mutations in the microenvironment or by therapy. The fitness variations of a TP53 and related genes are very infrequent,25 somatic mutations given clone are referred to as the ‘fitness landscape’ of the clone. in other key checkpoint and DNA repair genes have not been In most cancers, driver mutations are defined based on inter- reported, and progression to CMML has not been observed in patient recurrence and predicted functional relevance. In myeloid patients with congenital DNA repair defects. malignancies, the availability of competitive transplantation assays Therefore, early genetic lesions may be generated by DNA in mice provides an opportunity to estimate fitness experimen- damage accumulating in aging HSC41 despite functional repair tally. Driver mutations may be those that confer a selective and checkpoints machineries. Critical telomere attrition markers advantage to the tumor cell (that is, fitness 41) as opposed to have not been detected in CMML,42 in which the single nucleotide passenger mutations, which are neutral or slightly detrimental. variant pattern is dominated by transitions (Merlevede et al., In addition to the fitness landscape of the mutations, the unpublished results), as in MDS,12,18,19 and in AML.13 Normal HSCs kinetics of clonal expansion is influenced by the size of the target accumulate these transitions with age through replication errors, cell population and the mutation rate (Figure 1).16 Current deamination of methylcytosines and oxidative damage.13 mathematical models of chronic myeloid neoplasms assume Insertions/deletions generating CMML may be fueled by defaults that preleukemic clones compete with normal HSC for the same in non-homologous end joining, the only mechanism available to niches, with a constant number of stem cells.45 The recent finding quiescent cells to repair double strand breaks.38 of such a ‘competition’ process, neutral in terms of HSC pool size, Most of the mutations that arise randomly in HSC may not during normal hematopoiesis,40 and the direct visualization of affect HSC self-renewal and differentiation potential, thus are likely leukemic cells competing with normal HSC for the same niches,46 to be ‘passengers’. Though mutations may not always occur supports this model. Though this hypothesis may apply to CMML,

Leukemia (2013) 1441 – 1450 & 2013 Macmillan Publishers Limited Evolution and CMML clone architecture R Itzykson and E Solary 1443 35 Microenvironment the ‘adaptive oncogenesis’ model, which postulates that some Distinct mutations may prove advantageous only when competing against suclone normal cells functionally impaired by aging. It remains Mutation Therapy rate undetermined whether TET2 mutations confer an advantage to HSC when competing with older but not younger normal HSC, but this model could account for the rapid accumulation of copy- number alterations after the age of 50 in healthy subjects.55 þ Residual HSC The heterogeneous clonal burden of TET2 mutations in CD34 Driver Passenger cells, depending on the chronic myeloid neoplasm, is another mutation mutations parameter that could affect disease progression. TET2 mutations are present in the majority of HSC in CMML25 and primary Figure 1. Determinants of fitness of a leukemic clone: The 61 emergence and clonal expansion of a preleukemic or leukemic myelofibrosis and in a minority of HSC in essential thrombocytemia, polycythemia vera52 and possibly MDS.25 These clone, generated by a driver mutation in a target cell, for example, a 62 HSC, is determined by the fitness of this first mutation (red dot), the differences may reflect (1) the stochasticity of clonal expansion, 34 mutation rate, the size of the target cell population (in shaded gray, (2) the heterogeneity of normal HSC self-renewal potential, (3) a w/o shown nucleus), the influence of passenger mutations (green ‘density dependence’ phenomenon observed when the mutated and yellow dots), the competition with residual normal cells, and clone behaves as a ‘social cheater’ relying on non-mutant cells for with distinct sublones, the microenvironment (cells shown with a resources,63 for example, a mutation that would confer a self- nucleus) and therapy. renewal advantage but enhance the cell dependence on paracrine signaling (Figure 2). Finally, one of the likeliest explanations for these differences in clonal expansion trajectories could be the role where the proportion of phenotypic HSC is similar to age-matched of additional mutations. controls,25 it warrants careful experimental validation. Several lines of evidence suggest that mutations in TET2 are initial driver lesions in some CMML. These include the high frequency of PROGRESSION TO CMML: THE ROLE OF ADDITIONAL these mutations, which can be detected in up to 60% of CMML MUTATIONS patients,30,47,48 the competitive advantage of murine and human The finding that TET2 and ASXL1 mutations both confer selective HSC invalidated for TET2,49–51 and the results of single-cell clonal advantages as single lesions, and that their association in CMML is tracking experiments indicating that a TET2 mutation, when present, less frequent than expected by chance25 comforts the notion that is often the earliest recurrent genetic event in CMML,25 as well as in these two lesions are independent drivers of CMML. Then, the MPN52 and in AML.53 Finally, TET2-mutated clones can be detected two-step ‘linear’ model proposed earlier suggests that a in a small fraction of older subjects with clonal, but non-leukemic, combination of mutations is required for the progression from hematopoiesis.54 Actually, SNP arrays can detect in a significant clonal hematopoieisis to CMML (Figure 3). The occurrence of a proportion of healthy subjects (0.8%) a somatic mosaicism (50% secondary mutation could be favored by the higher mutation rate acquired uniparental disomies, 35% deletion, 15% gain) that may observed following the initial mutation, suggesting genetic predispose to a broad spectrum of hematological malignancies, instability, for example, JAK2 mutations activate DNA damage including chronic myeloid neoplasms.55 checkpoints64 and NRAS mutations promote oxidative damage.65 These observations suggest a two-step model, where initial In another scenario called epistasis, a neutral or even slightly lesions, such as mutations in TET2, drive the emergence of detrimental mutation hitchhikes on the initial driver lesion, then clonality, and secondary lesions contribute to progression to full- becomes advantageous.66 One example is the cooperation of Ras blown malignancy, with abnormal differentiation. In exceptional and Myc oncogenes in the mouse. The specific prognostic impact cases, the initial event may be germinal, for example, germline of gene mutations combinations in AML67 and in MDS,68 argues RUNX1 mutations were reported to evolve to CMML by acquiring for a role of epistasis in myeloid malignancies. CBL mutations.56 Other gene mutations, such as the frequent In CMML, an epistatic relation could exist between SRSF2 and mutations in the ASXL1 gene (B40% of cases), could also trigger TET2 mutations. A mutation in at least one splice gene is observed CMML. Mechanisms of ASXL1-induced leukemogenesis could in more than 60% of CMML and, in most cases, mutations in 25 involve the increased expression of HOXA genes, which are diverse splice genes are mutually exclusive. In CMML and in SPOTLIGHT critical regulators of HSC self-renewal.57 MDS, mutations in the splice gene SRSF2 are frequently combined Mathematical models predict a long latency in chronic myeloid with mutations in TET2.25,47,68 In normal hematopoietic cells, neoplasms.45 If the high fitness value of initial driver mutations could splice-gene mutations induce a proliferation defect in vitro and a account for the rapid onset of clonality and clinical manifestations in competitive disadvantage in vivo.18 A TET2/SRSF2 synthetic AML,13 the long latency observed in CMML and other chronic viability, with SRSF2 mutation being neutral or detrimental in a myeloid neoplasms could be due to the small fitness advantage of TET2-wild-type, and becoming advantageous in a TET2-mutated TET2 and other initial driver mutations in these diseases, that is, these cell, could explain why the mutational order is biased,66 with TET2 mutations may not deregulate HSC proliferation. Such a limited mutations often preceding SRSF2 mutations.25 competitive advantage (fitness as low as 0.4%) has been described An epistatic relation could also exist between mutations in for bona fide oncogenic mutations involved in solid tumors, leading signaling genes (CBL, NRAS, KRAS and JAK2) and other genetic to long times (410 years) to fixation.58 Mutations with a small fitness lesions, especially TET2 mutations, in CMML.25 In some situations, value may not always reach fixation and could disappear isolated JAK2 mutations may be detrimental to HSC. Similarly, spontaneously, a phenomenon that has been exceptionally noted although they may confer a small, but significant competitive after diagnosis but could occur at the preclinical stage.59 Of note, advantage to murine HSC,69 KRAS and NRAS mutations may also these observations raise concerns regarding the above-mentioned trigger oncogenic senescence, which would be bypassed by capability of competitive repopulation assays to discriminate driver additional mutations, for example, in RUNX170 or ASXL1.57 from passenger mutations. The long latency between the initial driver mutation appearance and disease expression could explain why TET2 mutations are not CLONAL HETEROGENEITY AND ARCHITECTURE found in pediatric myeloid neoplasms, including juvenile myelo- In contrast with Peter Nowell’s model,7 in which each additional monocytic leukemias (JMML).60 Another explanation is provided by mutation induced a selective sweep removing its ancestor

& 2013 Macmillan Publishers Limited Leukemia (2013) 1441 – 1450 Evolution and CMML clone architecture R Itzykson and E Solary 1444 Clonal Clonal extinction Dominance Sub-clones

abcdef Figure 2. Emergence of clones and subclones: mutations with a small fitness value may not reach fixation, leading to clonal extinction (a). For clonal dominance to occur, an intrinsic fitness advantage of the driver mutation (b), or a favorable competition with impaired surrounding cells because of aging (so-called ‘adaptive oncogenesis’) (c), or the occurrence of an additional mutation that would be neutral or even slightly

SPOTLIGHT detrimental by itself (in the so-called ‘epistasis’ process) (d) is needed. The clone can also remain a subclone in the targeted population, due to the heterogeneity of the targeted population (e), or the dependence of the mutated cells on a paracrine effect (arrow) provided by residual normal cells in the same population (f).

Myelomonocytic hyperplasia eg. N/KRAS or CBL

GM-CSF Granulomonocytic hypersensitivity skewing Ring sideroblasts e.g. Early clonal dominance SF3B1 e.g. : TET2 ± SRSF2

e.g. Thrombocytopenia RUNX1

« Myelodysplastic » « Myeloproliferative » CMML CMML Figure 3. Early clonal dominance in CMML. A combination of initiating mutations (represented here as red and green dots, for example, mutations in TET2 and in SRSF2) induces early clonal dominance at the HSC level. These genetic alterations promote the myelomonocytic hyperplasia that deﬁnes the disease, possibly at the expense of the erythroid and megakaryocytic compartments. The thrombocytopenia can be further affected by mutations in RUNX1. The addition of mutations in signaling genes (NRAS, KRAS, CBL) will promote the expansion of myelomonocytic cells that deﬁnes the myeloproliferative form of the disease.

clone, the genetic heterogeneity within tumors is now extensively cohesin complex13 could promote mitotic recombination, recognized.8,16,17 This heterogeneity can be explored by different together with increased HSC proliferation. methods with variable sensitivity71,72 (see also the legend of Lastly, ‘clonal interference’ is an infrequent process leading to Figure 5). clonal branching that occurs when two driver lesions with similar In most CMML cases (Figure 4), the clone architecture is mostly ﬁtness arise independently.73 We have observed such a case with linear, as in de novo MDS12 and in AML.13,53 Split architecture with a competition between an NRAS and a KRAS-mutated clone, but several branching arising from the same ancestor can however, be failed to identify any mosaicism of concomitant splice-gene observed, resulting mostly from LOH at mutated loci, for example, mutations.25 at the TET2 or at the CBL loci,25 and possibly at many other loci without any ﬁtness advantage. LOH was also described in ALL where it is due to illegitimate recombination.14 In CMML, LOH EVOLUTIONARY MECHANISMS SUSTAINING GENETIC mostly involves mitotic recombination,25 generating two daughter HETEROGENEITY cells, one homozygous for the mutation, the other one reverting Several mechanisms could explain why additional mutations fail to to wild-type status. Mutations in JAK2, RAS65 and members of the eradicate their ancestor clone. First, these additional mutations

Leukemia (2013) 1441 – 1450 & 2013 Macmillan Publishers Limited Evolution and CMML clone architecture R Itzykson and E Solary 1445 ‘commensalism’ could exist between subclones, through paracrine signaling,76 or direct cell–cell interactions.77 Last, minor subclones could demonstrate increased adaptability in the context of a changing environment, for example, hypoxia or therapeutic challenge.78 Leukemias with a minor fraction of LIC spontaneously accumulate a high degree of genetic heterogeneity, with some minor subclones having reduced ﬁtness, probably because the infrequent replication of LIC alleviates selection pressures. This last model explains how genetic and functional heterogeneities are both robustly maintained during oncogenesis, each LOH mechanism reinforcing the other.

NICHE CONTRIBUTIONS CI MR MR Normal HSCs are tightly regulated by microenvironmental cues, including cell–cell interactions, soluble molecules and anatomical and physical constraints in the so-called HSC niche. HSC may traffic between an endosteal, presumably hypoxic niche, where Figure 4. Clonal architecture in CMML. In most cases, clonal they interact with immature osteoblasts, and a perivascular, architecture is characterized by the linear accumulation of muta- possibly normoxic niche. The limited ability to xenotransplant tions (left panel). Rare cases of clonal interference (CI) lead to CMML in immunodeficient mice could indicate a role for the detection of subclones with distinct mutations (middle panel). microenvironment in sustaining leukemogenesis.28 In silico Branching can also occur by loss of heterozygosity (LOH) or mitotic modeling suggests that increasing the frequency, at which HSC recombination (MR) of a mutated locus, the latter generating two traffic between different niches can promote tumor growth79 and daughter cells, one without the mutation and the other one with two mutated alleles (right panel). Each triangle represents a gene favor the coexistence of competing subclones with different allele. Each color indicates a distinct gene. requirements. Leukemic cells can induce microenvironmental changes impairing the functionality of residual normal HSC.46 Conversely, gene deletion targeted to a microenvironmental cell 33 80 Long-term selection – Self-renewal type in the mouse could lead to MDS, or to MPN. Short-term selection - Proliferation Evolution theory predicts three ways, in which microenvironment modifications could affect clonal dynamics. First, altered Stem cells niches could reduce the size of the HSC pool, thereby promoting ~0.01% the fixation of neutral mutations. However, bone marrow failure syndromes do not progress to CMML. Second, altered niches Progenitors could contribute to genetic instability of HSC by modulating ~1% oxygen tension, and promoting oxidative DNA damage such as 8- oxoguanines. Finally, alterations in environmental cues could modulate the fitness landscapes of mutations; in particular Precursors microenvironment aging promotes benign and preleukemic 99% clones.81 This combination of mechanisms may be at work when disease initiation is triggered by chronic inflammation,82 which induces Figure 5. Competition between mutated and normal cells in the oxidative damage and triggers HSC proliferation.83 Older HSC are hematopoietic hierarchy:in CMML as in other chronic myeloid particularly sensitive to global elevation in levels of reactive malignancies, myeloid differentiation is preserved but deregulated. oxygen species.84 Inflammatory cytokines like TNFa could also A first level of competition between normal and mutated cells modulate fitness landscapes of some mutations.85 Future work occurs in the HSC compartment and may rely on self-renewal 86 potential. Competition in more mature compartments may rely on should focus on microenvironment alterations in CMML, SPOTLIGHT proliferation potential and cytokine responsiveness. Each triangle because these modifications, which are mostly non-clonal, can represents a gene allele. Each color indicates a distinct gene. Single- represent an Achilles heel for cancer cells. cell assays are needed to explore heterogeneity at the stem cell level. NGS techniques still detect part of the clonal heterogeneity, whereas Sanger sequencing only detects mutations in precursors, EPIGENETIC HETEROGENEITY thus does not explore clonal heterogeneity in this disease. Epigenetic marks, including DNA methylation, histone post- translational modifications, nucleosomal positioning and non- coding RNAs, are altered in cancer cells and contribute to could have a relatively low fitness value compared with early oncogenesis. In CMML, a third of patients harbor aberrant driver lesions. However, this explanation does not fit with their hypermethylation of the TIF1g gene, which is involved in non- short fixation time compared with initial lesions.58 Second, each canonical TGF-b signaling and in transcription elongation, and new oncogenic lesion could further perturb cellular signaling and haplo-insufficiency of Tif1g in mouse myeloid lineage generates a transcription networks, which exhibit limited flexibility.74 As CMML-like neoplasm.23 Though knowledge on the acquisition of mutations accumulate, the ‘network compatibility’ of additional epigenetic alterations is still beyond that of genetic lesions, it is lesions may narrow until the dominant clone faces an evolutionary likely that similar processes of clonal selection are at work with bottleneck.67 The accumulation of numerous so-called ‘passenger’ these aberrations. On the basis of this hypothesis, methylation mutations could generate a molecular ‘noise’ leading to marks have been used as molecular clocks in an attempt to trace phenotypic alterations,74 (for example, the accumulation of the history of cancer clones.87 altered proteins may trigger a heat-shock protein response), but Two non-mutually exclusive theories could explain the acquisi- these noise-driven phenotypic transitions could remain under tion of epigenetic alterations in CMML cells. In a ‘deterministic’ tight control in multicellular organisms.75 Third, a cooperative model, epigenomic changes are induced by genetic mutations, as

& 2013 Macmillan Publishers Limited Leukemia (2013) 1441 – 1450 Evolution and CMML clone architecture R Itzykson and E Solary 1446 genes encoding components of the epigenetic machinery, early phases of hematopoiesis. Importantly, the finding that the including TET2 and ASXL1, are frequently mutated in CMML.2 characteristic phenotype of CMML emerges at an earlier stage Even mutations in the JAK2 kinase could affect histone than JAK2 and SF3B1 phenotypes is concordant with the clonal methylation.88 In accordance with this model, a characteristic expansion at the level of the most immature CD34 þ /CD38 À methylation pattern has been described in TET2-mutated CMML.24 cells.25 These aberrant epigenetic marks can be tracked to leukemic Fourth, the granulomonocytic hyperplasia that is characteristic immature progenitors, suggesting that they do not simply reflect of CMML has long been thought to result from hypersensitivity to abnormal differentiation.89 In the other, ‘stochastic’ model, GM-CSF, which is actually restricted to the B35% of patients epigenetic changes are randomly acquired, biological ‘noise’ harboring mutations affecting signaling genes (N/KRAS, CBL, playing the same role as genetic instability for genetic JAK2).98 In many cases, these mutations are acquired during the mutations. Epigenetic fluctuations leading to phenotypic course of the disease,25,99 indicating that another mechanism may diversification of otherwise clonal populations have indeed been account for the granulomonocytic expansion. Granulomacro- observed in vitro, notably in hematopoietic cells during phagic progenitors are quantitatively and qualitatively normal, differentiation,90 and in cancer cells.91 In addition, there is but multipotent progenitors and common myeloid progenitors evidence that epigenetic alterations accumulate with age, are expanded,25 as in MDS,89 and these progenitors are skewed including in HSC.92 Altogether, from an evolutionary perspective, towards granulomonocytic differentiation at the expense of the epigenetic lesions could behave as genetic lesions during CMML erythroid lineage. initiation and progression. Such differentiation skewing, which had been suspected also in PMF27,100 could involve biased lineage decision or lineage amplification. Solving this issue will require the use of single-cell ABNORMAL DIFFERENTIATION IN CMML: THE ROLE OF tracking experiments.101,102 If our model holds true, the rare CLONAL HETEROGENEITY patients with benign TET2-mutated clonal hematopoiesis54 are CMML, together with other chronic myeloid malignancies, are expected to result from late clonal amplification, after lineage characterized by a preserved although deregulated myeloid commitment. The level of clonal expansion of TET2-mutated

SPOTLIGHT differentiation, with various phenotypic features related to clones in immature versus mature progenitor compartments dysplasia and proliferation of the granulomonocytic, erythroid might depend on the presence of additional mutations, and on and megakaryocytic lineages. Beyond the long-term clonal the self-renewal capacity of the HSC origin.34 Lineage skewing competition within the HSC compartment, a second layer of could be affected by the self-renewal potential of HSC, by the competition between normal cells and pathological subclones cellular microenvironment,103 and by paracrine loops between occurs during myeloid differentiation. Competition in the HSC differentiated stem cells.104 How these lineage-biased HSC subsets compartment may rely on self-renewal potential, whereas contribute to leukemogenesis remains an exciting avenue for competition in more mature compartments relies on proliferation future research.37 potential and cytokine responsiveness (Figure 5). This explains the In the absence of TET2 mutations, the granulomonocytic distinct size of the mutated clone in the bulk of the tumor and in skewing of immature progenitors that characterizes CMML may the HSC compartment. Mutations can be dominant in differen- be caused by other oncogenes, for example, a marked differentia- tiated leukemic cells, yet only be present in a fraction of HSC.25,71 tion bias was noticed in ASXL1-mutated cases (Itzykson et al. The differential competition between normal and mutated cells unpublished observations), and a similar differentiation bias could in the HSC versus differentiating compartments has been be induced by rare translocations involving the RET oncogene.105 exemplified by MPN-associated JAK2 mutations,93–95 which undergo massive expansion in differentiating cells, when homodimeric receptors such as granulocyte colony-stimulating JMML SPECIFICITIES factorG-CSF and EPO receptors become expressed.92 The Three main characteristics separate JMML, which is a rare and mechanism driving TET2-mutated cell expansion with aggressive myeloid malignancy of infants or young children, from differentiation is less clear, but TET2 mutations could promote CMML, which is basically a disease of aging.106,107 First, the expansion of JAK2-mutated clones during erythroid hypersensitivity of myeloid progenitor cells to GM-CSF in colony differentiation.95 The genetic characteristics of leukemic cells can forming assays (CFU-GM) is one of the hallmarks of JMML, while shape the phenotype of chronic myeloid neoplasms. being limited to a fraction of CMML, mainly those with mutations First, a direct link can exist between the function of a mutated that affect cell signaling components. Second, a molecular gene and the phenotype, for example, mutations in SF3B1 gene alteration of the RAS signaling pathway, which is often unique cause erythroid dysplasia with ringed sideroblasts, a phenotype in the whole exome, and sometimes appears on a background of that emerges at a late stage of differentiation. The percentage of specific congenital syndrome such as neurofibromatosis or ringed sideroblasts simply mirrors the size of the SF3B1-mutated Noonan syndrome, is identified in 85–90% of JMML patients. clone in erythroid precursors.19,96 Conversely, mutations affecting epigenetic, splice or transcription Second, the copy number of the mutation can drive the factor genes are very rare in JMML. Last, allogeneic stem cell phenotype, for example, erythrocytosis correlates with the transplantation is usually the best therapeutic option to propose expansion of homozygous JAK2V617F-mutated subclones,93 with to these young patients, even though some children with RAS or increased sensitivity to EPO of homozygous subclones explaining CBL mutations may experience spontaneous resolution. their ability to outcompete other clones during myeloid differentiation.93–95 Additional mechanisms may account for the rare cases of minor and stable JAK2V617F homozygous subclones in AML PROGRESSION essential thrombocytemia.93,97 Twenty to 30% of CMML cases eventually transform to AML. The Third, the disease phenotype could depend on the level of rare CMML cases harboring DNTM3A108 and NPM1 mutations clonal dominance in the hematopietic cell hierarchy. In particular, resemble ‘smoldering’ acute leukemias.25 In the other cases, the early dominance of TET2 in the CD34 þ /CD38- compartment, but mechanisms of transformation to AML remain unknown. The not late dominance (in the CD34 þ /CD38 þ compartment and dominant view is that transformation arises from the massive below), could lead to the specific monocytosis of CMML expansion of a clone with novel, highly fit, genetic lesions, thereby (Figure 3).25 MDS or atypical MDS/MPN could evolve into CMML increasing clonal heterogeneity, as described in transformed through the extension of mutated TET2 dominance in the very MDS.12 On the basis of observations made in JAK2-mutated

Leukemia (2013) 1441 – 1450 & 2013 Macmillan Publishers Limited Evolution and CMML clone architecture R Itzykson and E Solary 1447 50 searched when mast cells infiltrate the skin or the bone marrow, and fusion genes involving PDGFR (platelet-derived growth factor receptor, a or b)ofFGFR1 (fibroblast growth factor receptor 1) 40 genes have to be explored when an eosinophilia is associated to monocytosis. In these three situations, refining the diagnosis will lead to the therapeutic use of a tyrosine kinase inhibitor. 30 Otherwise, the treatment of CMML is not fully codified. Allogeneic stem cell transplantation, which is currently the only curative option, is deferred in most cases because of patients age 20 and unclear prognosis.113 The frequent relapses suggest the presence of subclones escaping the immune effects of allogeneic graft. Non curative options mainly include cytoreductive 10 chemotherapy like hydroxyurea to mitigate myeloproliferation, erythropoieisis-stimulating agents to cope with anemia, and Cumulative incidence of AML (%) hypomethylating agents that can do both. Cell genotyping and 0 next-generation sequencing confirmed that none of these therapeutic approaches can eradicate the clone.25 These 0122436 4860 therapies sometimes select minor subclones, for example, we Months observed the selection of an NRAS-mutated subclone upon Figure 6. Cumulative incidence of AML transformation in CMML: hydroxyurea. We noticed also that recombinant erythropoietin could increase the proportion of wild-type CD34 þ /CD38 À cells observed cumulative incidence of WHO-defined AML transforma- 25 tion in the Groupe Francophone des Myelodysplasies cohort of 312 through still hypothetic mechanisms. CMML patients,110 considering death as a competing risk. Two Owing to the limited number of clinical trials specifically mechanisms could possibly promote acute blastic transformation: dedicated to CMML and the clonal heterogeneity of the disease, the first (green line) would be the rapid expansion of existing clones, the mutations or epigenetic changes that affect the therapeutic for example, when ASXL1 gene is mutated, whereas the second (red response are still poorly identified. In a phase II clinical trial, testing line) would be the later occurrence of additional genetic events, for the efficacy of the demethylating agent decitabine in a small example, mutations in NPM1 or in DNMT3A. cohort of patients with an aggressive form of the disease, TIF1g promoter methylation did not predict decitabine efficacy.114 Nevertheless, the re-expression of TIF1g gene in peripheral MPN, the dominant clone at transformation may arise from a blood monocytes could be used as a biomarker of decitabine minor subclone of the chronic phase of disease.109 The nature of efficacy.25 Interestingly also, the response rate was higher in genetic lesions involved, and the role of microenvironment patients with TET2 mutations, without reaching significance in this changes in the transformation are a matter of future research. A small size phase II trial. Ongoing studies will indicate if TET2 natural evolution of high-fit clones, for instance those with ASXL1 mutations (when they are not associated with a mutation in mutations, leading to AML without any novel genetic lesion ASXL1), are predictors of a better response to demethylating cannot be excluded. Both mechanisms likely contribute to agents, as described in MDS, and whether a specific pattern of progression, explaining the inflection point in the cumulative DNA methylation will predict hypomethylating agent efficacy. incidence of AML transformation in CMML (Figure 6).25,111 Tailoring new rational therapies aiming at eradicating the pathological clone is the ultimate goal, allowing residual normal HSC to expand. As suggested above, assessing the clonal status of PROGNOSTIC IMPACT OF MOLECULAR ALTERATIONS HSC, which is distinct from that of differentiated cells, will be Several prognostic scores have been proposed in CMML, using essential to design and evaluate these new therapeutic strategies, diverse combinations of clinical, biological and cytogenetic and iterative snapshots of clonal architecture in the early phase of features, but none gained widespread acceptance for various therapy may allow extrapolating long-term clonal evolution and reasons. A very recently described CMML-specific prognostic adapting the therapeutic approach. Indeed, a treatment that will scoring system included FAB and WHO subtypes, cytogenetic, and correct the proliferative advantage of a subclone during SPOTLIGHT red blood cell transfusion dependency, with or without lacticode- differentiation will not necessarily alter its fitness in the stem cell shydrogenase level.112 By integrating molecular lesions with compartment. clinical variables, we also proposed an efficient score that, for Ideally, identification of an oncogenic event, to which the the first time, includes a molecular parameter, that is, mutations in leukemia-initiating cells are addicted is needed. This event will be ASXL1 gene. Among all the genes that can be mutated in CMML, an oncoprotein that will need to be degraded or inhibited, or a ASXL1 is the only one whose mutation has a negative prognostic loss-of-function that will be targeted indirectly if synthetic lethality impact on AML-free and overall survival in multivariate analyses. can be triggered. Major sources of resistance can be anticipated if The proposed score attributes 3 points for WBC 415 Â 109/l, and such a targeted therapy is used alone, including the induction or 2 points for four age465 years, anemia, platelet count selection of minor subclones with sometimes normal or even o100 Â 109/l and ASXL1 status, respectively. This score allows detrimental fitness before therapy.77 Therefore, a non-specific categorizing patients into low (0–4 points), intermediate (5–7 cytoreduction will probably be needed first in an attempt to points; 34%) and high (8–12 points; 24%) risk groups, with median eliminate these minor subclones before the onset of the targeted overall survival not reached, 38 and 14 months, respectively.113 therapy.77,115 Alternating strategies would be to seek ‘limited’ clonal eradication, allowing the persistence of clonal benign hematopoi- THERAPEUTIC CONSEQUENCES OF MOLECULAR ANALYSES IN esis,54 or restore normal differentiation without clonal eradication, CMML as obtained with MEK inhibitors116 and hypomethylating The search for molecular alterations can be used first to refine the agents.25,89,117,118 The combination of drugs selecting opposite diagnosis in patients with monocytosis. BCR-ABL rearrangement that phenotypes, with alternative rather than concomitant identifies chronic myeloid leukemia must be systematically explored, administration to generate a vicious circle of futile adaptation, a mutation in KIT gene that characterizes systemic mastocytosis is could also mitigate the disease progression.

& 2013 Macmillan Publishers Limited Leukemia (2013) 1441 – 1450 Evolution and CMML clone architecture R Itzykson and E Solary 1448 CONCLUSION 18 Yoshida K, Sanada M, Shiraishi Y, Nowak D, Nagata Y, Yamamoto R et al. The genetic heterogeneity of CMML, both in each patient and Frequent pathway mutations of splicing machinery in myelodysplasia. Nature between patients, indicates that the disease can be generated 2011; 478: 64–69. through a wide array of evolutionary trajectories. Current 19 Papaemmanuil E, Cazzola M, Boultwood J, Malcovati L, Vyas P, Bowen D et al. approaches suggest that a preferred order of mutation accumula- Somatic SF3B1 mutation in myelodysplasia with ring sideroblasts. N Engl J Med tion is TET2 (or IDH1 and IDH2)orASXL1 first, SRSF2 or an 2011; 365: 1384–1395. 20 Such E, Cervera J, Costa D, Sole F, Vallespi T, Luno E et al. Cytogenetic risk alternative splice gene second, then, in 35–40% of cases, a stratification in chronic myelomonocytic leukemia. Haematologica 2011; 96: signaling gene mutation inducing GM-CSF hypersensitivity and 375–383. myeloproliferation. The respective frequency of each gene 21 Dunbar AJ, Gondek LP, O’Keefe CL, Makishima H, Rataul MS, Szpurka H et al. mutation could possibly bias this evaluation. Nevertheless, early 250 K single nucleotide polymorphism array karyotyping identifies acquired clonal dominance, linear accumulation of mutations with limited uniparental disomy and homozygous mutations, including novel missense branching, and competitive advantage to the more mutated cells substitutions of c-Cbl, in myeloid malignancies. Cancer Res 2008; 68: during differentiation are the main characteristics of the CMML 10349–10357. clonal architecture. Exploiting the evolutionary dynamics of clonal 22 Tessema M, Langer F, Dingemann J, Ganser A, Kreipe H, Lehmann U. heterogeneity to treat cancer is certainly a daunting task. After all, Aberrant methylation and impaired expression of the p15(INK4b) cell cycle regulatory gene in chronic myelomonocytic leukemia (CMML). Leukemia 2003; modern antiretroviral combinations allow control, rather than 17: 910–918. eradication of HIV, a virus counting only nine proteins. 23 Aucagne R, Droin N, Paggetti J, Lagrange B, Largeot A, Hammann A et al. Transcription intermediary factor 1gamma is a tumor suppressor in mouse and human chronic myelomonocytic leukemia. J Clin Invest 2011; 121: 2361–2370. CONFLICT OF INTEREST 24 Perez C, Martinez-Calle N, Martin-Subero JI, Segura V, Delabesse E, Fernandez- Mercado M et al. TET2 mutations are associated with specific 5-methylcytosine The authors declare no conflict of interest. and 5-hydroxymethylcytosine profiles in patients with chronic myelomonocytic leukemia. PLoS One 2012; 7: e31605. 25 Itzykson R, Kosmider O, Renneville A, Morabito M, Preudhomme C, Berthon C et al. Clonal architecture of chronic myelomonocytic leukemias. Blood 2013; 121: SPOTLIGHT ACKNOWLEDGEMENTS 2186–2198. We are indebted to William Vainchenker and Olivier Bernard for insightful 26 Nilsson L, Astrand-Grundstrom I, Arvidsson I, Jacobsson B, Hellstrom-Lindberg E, discussions. We apologize to several other authors contributing to the field of Hast R et al. Isolation and characterization of hematopoietic progenitor/stem myeloid neoplasms or clonal evolution of cancer whose work could not be cited due cells in 5q-deleted myelodysplastic syndromes: evidence for involvement at the to space restrictions. This research is supported by grants from the Ligue Nationale hematopoietic stem cell level. Blood 2000; 96: 2012–2021. Contre le Cancer (Label, ES), and the French National Cancer Institute (PHRC 2011 to 27 James C, Mazurier F, Dupont S, Chaligne R, Lamrissi-Garcia I, Tulliez M et al. The ES and support to RI). UMR1009 is supported by the ARC Foundation and the Re´gion hematopoietic stem cell compartment of JAK2V617F-positive myeloproliferative Ile-de-France. disorders is a reflection of disease heterogeneity. Blood 2008; 112: 2429–2438. 28 Thanopoulou E, Cashman J, Kakagianne T, Eaves A, Zoumbos N, Eaves C. Engraftment of NOD/SCID-beta2 microglobulin null mice with multilineage neoplastic cells from patients with myelodysplastic syndrome. Blood 2004; 103: REFERENCES 4285–4293. 1 Orazi A, Germing U. The myelodysplastic/myeloproliferative neoplasms: myelo- 29 Delhommeau F, Dupont S, Tonetti C, Masse A, Godin I, Le Couedic JP et al. proliferative diseases with dysplastic features. Leukemia 2008; 22: 1308–1319. Evidence that the JAK2 G1849T (V617F) mutation occurs in a lymphomyeloid 2 Shih AH, Abdel-Wahab O, Patel JP, Levine RL. The role of mutations in epigenetic progenitor in polycythemia vera and idiopathic myelofibrosis. Blood 2007; 109: regulators in myeloid malignancies. Nat Rev Cancer 2012; 12: 599–612. 71–77. 3 Tefferi A. Novel mutations and their functional and clinical relevance in mye- 30 Smith AE, Mohamedali AM, Kulasekararaj A, Lim Z, Gaken J, Lea NC et al. loproliferative neoplasms: JAK2, MPL, TET2, ASXL1, CBL, IDH and IKZF1. Leukemia Next-generation sequencing of the TET2 gene in 355 MDS and CMML patients 2010; 24: 1128–1138. reveals low-abundance mutant clones with early origins, but indicates no defi- 4 Vainchenker W, Delhommeau F, Constantinescu SN, Bernard OA. New mutations nite prognostic value. Blood 2010; 116: 3923–3932. and pathogenesis of myeloproliferative neoplasms. Blood 2011; 118: 1723–1735. 31 Goardon N, Marchi E, Atzberger A, Quek L, Schuh A, Soneji S et al. Coexistence of 5 Itzykson R, Droin N, Solary E. Current insights in the cellular and molecular LMPP-like and GMP-like leukemia stem cells in acute myeloid leukemia. Cancer biology of chronic myelomonocytic leukemia. Int J Hematol Oncol 2012; 1: Cell 2011; 19: 138–152. 147–158. 32 Jamieson CH, Ailles LE, Dylla SJ, Muijtjens M, Jones C, Zehnder JL et al. Granu- 6 Abdel-Wahab O, Dey A. The ASXL-BAP1 axis: new factors in myelopoiesis, cancer locyte-macrophage progenitors as candidate leukemic stem cells in blast-crisis and epigenetics. Leukemia 2013; 27: 10–15. CML. N Engl J Med 2004; 351: 657–667. 7 Nowell PC. The clonal evolution of tumor cell populations. Science 1976; 194: 33 Raaijmakers MH, Mukherjee S, Guo S, Zhang S, Kobayashi T, Schoonmaker JA 23–28. et al. Bone progenitor dysfunction induces myelodysplasia and secondary leu- 8 Greaves M, Maley CC. Clonal evolution in cancer. Nature 2012; 481: 306–313. kaemia. Nature 2010; 464: 852–857. 9 Yates LR, Campbell PJ. Evolution of the cancer genome. Nat Rev Genet 2012; 13: 34 Muller-Sieburg CE, Sieburg HB, Bernitz JM, Cattarossi G. Stem cell heterogeneity: 795–806. implications for aging and regenerative medicine. Blood 2012; 119: 3900–3907. 10 Bendall SC, Nolan GP. From single cells to deep phenotypes in cancer. Nat 35 Henry CJ, Marusyk A, DeGregori J. Aging-associated changes in hematopoiesis Biotechnol 2012; 30: 639–647. and leukemogenesis: what’s the connection? Aging (Albany NY) 2011; 3: 11 Dick JE. Stem cell concepts renew cancer research. Blood 2008; 112: 4793–4807. 643–656. 12 Walter MJ, Shen D, Ding L, Shao J, Koboldt DC, Chen K et al. Clonal architecture 36 Takubo K, Goda N, Yamada W, Iriuchishima H, Ikeda E, Kubota Y et al. Regulation of secondary acute myeloid leukemia. N Engl J Med 2012; 366: 1090–1098. of the HIF-1alpha level is essential for hematopoietic stem cells. Cell Stem Cell 13 Welch JS, Ley TJ, Link DC, Miller CA, Larson DE, Koboldt DC et al. The origin and 2010; 7: 391–402. evolution of mutations in acute myeloid leukemia. Cell 2012; 150: 264–278. 37 Wang J, Sun Q, Morita Y, Jiang H, Gross A, Lechel A et al. A differentiation 14 Anderson K, Lutz C, van Delft FW, Bateman CM, Guo Y, Colman SM et al. Genetic checkpoint limits hematopoietic stem cell self-renewal in response to DNA variegation of clonal architecture and propagating cells in leukaemia. Nature damage. Cell 2012; 148: 1001–1014. 2011; 469: 356–361. 38 Milyavsky M, Gan OI, Trottier M, Komosa M, Tabach O, Notta F et al. A distinctive 15 Clappier E, Gerby B, Sigaux F, Delord M, Touzri F, Hernandez L et al. Clonal DNA damage response in human hematopoietic stem cells reveals an apoptosis- selection in xenografted human T cell acute lymphoblastic leukemia recapitu- independent role for p53 in self-renewal. Cell Stem Cell 2010; 7: 186–197. lates gain of malignancy at relapse. J Exp Med 2011; 208: 653–661. 39 Janzen V, Forkert R, Fleming HE, Saito Y, Waring MT, Dombkowski DM et al. 16 Turner NC, Reis-Filho JS. Genetic heterogeneity and cancer drug resistance. Stem-cell ageing modified by the cyclin-dependent kinase inhibitor p16INK4a. Lancet Oncol 2012; 13: e178–e185. Nature 2006; 443: 421–426. 17 Marusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity: a looking glass for 40 Bondar T, Medzhitov R. p53-mediated hematopoietic stem and progenitor cell cancer? Nat Rev Cancer 2012; 12: 323–334. competition. Cell Stem Cell 2010; 6: 309–322.

Leukemia (2013) 1441 – 1450 & 2013 Macmillan Publishers Limited Evolution and CMML clone architecture R Itzykson and E Solary 1449 41 Rube CE, Fricke A, Widmann TA, Furst T, Madry H, Pfreundschuh M et al. 65 Sallmyr A, Fan J, Rassool FV. Genomic instability in myeloid malignancies: Accumulation of DNA damage in hematopoietic stem and progenitor cells increased reactive oxygen species (ROS), DNA double strand breaks (DSBs) and during human aging. PLoS One 2011; 6: e17487. error-prone repair. Cancer Lett 2008; 270: 1–9. 42 Calado RT, Young NS. Telomere maintenance and human bone marrow failure. 66 Ashworth A, Lord CJ, Reis-Filho JS. Genetic interactions in cancer progression Blood 2008; 111: 4446–4455. and treatment. Cell 2011; 145: 30–38. 43 Stephens PJ, Greenman CD, Fu B, Yang F, Bignell GR, Mudie LJ et al. Massive 67 Patel JP, Gonen M, Figueroa ME, Fernandez H, Sun Z, Racevskis J et al. Prognostic genomic rearrangement acquired in a single catastrophic event during cancer relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med development. Cell 2011; 144: 27–40. 2012; 366: 1079–1089. 44 Catlin SN, Busque L, Gale RE, Guttorp P, Abkowitz JL. The replication rate of 68 Damm F, Kosmider O, Gelsi-Boyer V, Renneville A, Carbuccia N, Hidalgo-Curtis C human hematopoietic stem cells in vivo. Blood 2011; 117: 4460–4466. et al. Mutations affecting mRNA splicing define distinct clinical phenotypes and 45 Vickers MA. JAK2 617V4F positive polycythemia rubra vera maintained by correlate with patient outcome in myelodysplastic syndromes. Blood 2012; 119: approximately 18 stochastic stem-cell divisions per year, explaining age of onset 3211–3218. by a single rate-limiting mutation. Blood 2007; 110: 1675–1680. 69 Li Q, Haigis KM, McDaniel A, Harding-Theobald E, Kogan SC, Akagi K et al. 46 Colmone A, Amorim M, Pontier AL, Wang S, Jablonski E, Sipkins DA. Leukemic Hematopoiesis and leukemogenesis in mice expressing oncogenic NrasG12D cells create bone marrow niches that disrupt the behavior of normal hemato- from the endogenous locus. Blood 2011; 117: 2022–2032. poietic progenitor cells. Science 2008; 322: 1861–1865. 70 Motoda L, Osato M, Yamashita N, Jacob B, Chen LQ, Yanagida M et al. Runx1 47 Meggendorfer M, Roller A, Haferlach T, Eder C, Dicker F, Grossmann V et al. SRSF2 protects hematopoietic stem/progenitor cells from oncogenic insult. Stem Cells mutations in 275 cases with chronic myelomonocytic leukemia (CMML). Blood 2007; 25: 2976–2986. 2012; 120: 3080–3088. 71 Hou Y, Song L, Zhu P, Zhang B, Tao Y, Xu X et al. Single-cell exome sequencing 48 Jankowska AM, Makishima H, Tiu RV, Szpurka H, Huang Y, Traina F et al. Muta- and monoclonal evolution of a JAK2-negative myeloproliferative neoplasm. Cell tional spectrum analysis of chronic myelomonocytic leukemia includes genes 2012; 148: 873–885. associated with epigenetic regulation: UTX, EZH2, and DNMT3A. Blood 2011; 72 Navin N, Kendall J, Troge J, Andrews P, Rodgers L, McIndoo J et al. Tumour 118: 3932–3941. evolution inferred by single-cell sequencing. Nature 2011; 472: 90–94. 49 Pronier E, Almire C, Mokrani H, Vasanthakumar A, Simon A, da Costa Reis Monte 73 Lasho TL, Tefferi A, Finke C, Pardanani A. Clonal hierarchy and allelic mutation Mor B et al. Inhibition of TET2-mediated conversion of 5-methylcytosine to 5- segregation in a myelofibrosis patient with two distinct LNK mutations. Leukemia hydroxymethylcytosine disturbs erythroid and granulomonocytic differentiation 2011; 25: 1056–1058. of human hematopoietic progenitors. Blood 2011; 118: 2551–2555. 74 Rozenblatt-Rosen O, Deo RC, Padi M, Adelmant G, Calderwood MA, Rolland T 50 Quivoron C, Couronne L, Della Valle V, Lopez CK, Plo I, Wagner-Ballon O et al. et al. Interpreting cancer genomes using systematic host network perturbations TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse by tumour virus proteins. Nature 2012; 487: 491–495. and is a recurrent event during human lymphomagenesis. Cancer Cell 2011; 20: 75 Arias AM, Hayward P. Filtering transcriptional noise during development: con- 25–38. cepts and mechanisms. Nat Rev Genet 2006; 7: 34–44. 51 Moran-Crusio K, Reavie L, Shih A, Abdel-Wahab O, Ndiaye-Lobry D, Lobry C et al. 76 Inda MM, Bonavia R, Mukasa A, Narita Y, Sah DW, Vandenberg S et al. Tumor Tet2 loss leads to increased hematopoietic stem cell self-renewal and myeloid heterogeneity is an active process maintained by a mutant EGFR-induced transformation. Cancer Cell 2011; 20: 11–24. cytokine circuit in glioblastoma. Genes Dev 2010; 24: 1731–1745. 52 Delhommeau F, Dupont S, Della Valle V, James C, Trannoy S, Masse A et al. 77 Wu M, Pastor-Pareja JC, Xu T. Interaction between Ras(V12) and scribbled clones Mutation in TET2 in myeloid cancers. N Engl J Med 2009; 360: 2289–2301. induces tumour growth and invasion. Nature 2010; 463: 545–548. 53 Jan M, Snyder TM, Corces-Zimmerman MR, Vyas P, Weissman IL, Quake SR et al. 78 Sottoriva A, Vermeulen L, Tavare S. Modeling evolutionary dynamics of epige- Clonal evolution of preleukemic hematopoietic stem cells precedes human netic mutations in hierarchically organized tumors. PLoS Comput Biol 2011; 7: acute myeloid leukemia. Sci Transl Med 2012; 4: 149ra118. e1001132. 54 Busque L, Patel JP, Figueroa ME, Vasanthakumar A, Provost S, Hamilou Z et al. 79 Anderson AR, Weaver AM, Cummings PT, Quaranta V. Tumor morphology and Recurrent somatic TET2 mutations in normal elderly individuals with clonal phenotypic evolution driven by selective pressure from the microenvironment. hematopoiesis. Nat Genet 2012; 44: 1179–1181. Cell 2006; 127: 905–915. 55 Laurie CC, Laurie CA, Rice K, Doheny KF, Zelnick LR, McHugh CP et al. Detectable 80 Rupec RA, Jundt F, Rebholz B, Eckelt B, Weindl G, Herzinger T et al. Stroma- clonal mosaicism from birth to old age and its relationship to cancer. Nat Genet mediated dysregulation of myelopoiesis in mice lacking I kappa B alpha. 2012; 44: 642–650. Immunity 2005; 22: 479–491. 56 Shiba N, Hasegawa D, Park MJ, Murata C, Sato-Otsubo A, Ogawa C et al. CBL 81 Vas V, Wandhoff C, Dorr K, Niebel A, Geiger H. Contribution of an aged micro- mutation in chronic myelomonocytic leukemia secondary to familial platelet environment to aging-associated myeloproliferative disease. PLoS One 2012; 7: disorder with propensity to develop acute myeloid leukemia (FPD/AML). Blood e31523. 2012; 119: 2612–2614. 82 Kristinsson SY, Bjorkholm M, Hultcrantz M, Derolf AR, Landgren O, Goldin LR. 57 Abdel-Wahab O, Adli M, LaFave LM, Gao J, Hricik T, Shih AH et al. ASXL1 Chronic immune stimulation might act as a trigger for the development of acute

mutations promote myeloid transformation through loss of PRC2-mediated myeloid leukemia or myelodysplastic syndromes. J Clin Oncol 2011; 29: SPOTLIGHT gene repression. Cancer Cell 2012; 22: 180–193. 2897–2903. 58 Bozic I, Antal T, Ohtsuki H, Carter H, Kim D, Chen S et al. Accumulation of driver 83 Takizawa H, Regoes RR, Boddupalli CS, Bonhoeffer S, Manz MG. Dynamic varia- and passenger mutations during tumor progression. Proc Natl Acad Sci USA 2010; tion in cycling of hematopoietic stem cells in steady state and inflammation. 107: 18545–18550. J Exp Med 2011; 208: 273–284. 59 Muller-Schmah C, Solari L, Weis R, Pfeifer D, Scheibenbogen C, Trepel M et al. 84 Kohler A, Schmithorst V, Filippi MD, Ryan MA, Daria D, Gunzer M et al. Altered Immune response as a possible mechanism of long-lasting disease control in cellular dynamics and endosteal location of aged early hematopoietic pro- spontaneous remission of MLL/AF9-positive acute myeloid leukemia. Ann genitor cells revealed by time-lapse intravital imaging in long bones. Blood 2009; Hematol 2012; 91: 27–32. 114: 290–298. 60 Muramatsu H, Makishima H, Jankowska AM, Cazzolli H, O’Keefe C, Yoshida N 85 Fleischman AG, Aichberger KJ, Luty SB, Bumm TG, Petersen CL, Doratotaj S et al. et al. Mutations of an E3 ubiquitin ligase c-Cbl but not TET2 mutations are TNFalpha facilitates clonal expansion of JAK2V617F positive cells in myelopro- pathogenic in juvenile myelomonocytic leukemia. Blood 115: 1969–75. liferative neoplasms. Blood 2011; 118: 6392–6398. 61 Stein BL, Williams DM, Rogers O, Isaacs MA, Spivak JL, Moliterno AR. Disease 86 Marcondes AM, Mhyre AJ, Stirewalt DL, Kim SH, Dinarello CA, Deeg HJ. burden at the progenitor level is a feature of primary myelofibrosis: a multi- Dysregulation of IL-32 in myelodysplastic syndrome and chronic myelomono- variable analysis of 164 JAK2 V617F-positive myeloproliferative neoplasm cytic leukemia modulates apoptosis and impairs NK function. Proc Natl Acad Sci patients. Exp Hematol 2011; 39: 95–101. USA 2008; 105: 2865–2870. 62 Dingli D, Pacheco JM. Stochastic dynamics and the evolution of mutations in 87 Siegmund KD, Marjoram P, Woo YJ, Tavare S, Shibata D. Inferring clonal stem cells. BMC Biol 2011; 9: 41. expansion and cancer stem cell dynamics from DNA methylation patterns in 63 Velicer GJ, Kroos L, Lenski RE. Developmental cheating in the social bacterium colorectal cancers. Proc Natl Acad Sci USA 2009; 106: 4828–4833. Myxococcus xanthus. Nature 2000; 404: 598–601. 88 Liu F, Zhao X, Perna F, Wang L, Koppikar P, Abdel-Wahab O et al. JAK2V617F- 64 Nakatake M, Monte-Mor B, Debili N, Casadevall N, Ribrag V, Solary E et al. mediated phosphorylation of PRMT5 downregulates its methyltransferase JAK2(V617F) negatively regulates p53 stabilization by enhancing MDM2 activity and promotes myeloproliferation. Cancer Cell 2011; 19: 283–294. via La expression in myeloproliferative neoplasms. Oncogene 2012; 31: 89 Will B, Zhou L, Vogler TO, Ben-Neriah S, Schinke C, Tamari R et al. Stem and 1323–1333. progenitor cells in myelodysplastic syndromes show aberrant stage-specific

& 2013 Macmillan Publishers Limited Leukemia (2013) 1441 – 1450 Evolution and CMML clone architecture R Itzykson and E Solary 1450 expansion and harbor genetic and epigenetic alterations. Blood 2012; 120: 104 Reynaud D, Pietras E, Barry-Holson K, Mir A, Binnewies M, Jeanne M et al. IL-6 2076–2086. controls leukemic multipotent progenitor cell fate and contributes to chronic 90 Pina C, Fugazza C, Tipping AJ, Brown J, Soneji S, Teles J et al. Inferring rules of myelogenous leukemia development. Cancer Cell 2011; 20: 661–673. lineage commitment in haematopoiesis. Nat Cell Biol 2012; 14: 287–294. 105 Ballerini P, Struski S, Cresson C, Prade N, Toujani S, Deswarte C et al. RET fusion 91 Gupta PB, Fillmore CM, Jiang G, Shapira SD, Tao K, Kuperwasser C et al. Sto- genes are associated with chronic myelomonocytic leukemia and enhance chastic state transitions give rise to phenotypic equilibrium in populations of monocytic differentiation. Leukemia 2012; 26: 2384–2389. cancer cells. Cell 2011; 146: 633–644. 106 Emanuel PD. Juvenile myelomonocytic leukemia and chronic myelomonocytic 92 Chambers SM, Shaw CA, Gatza C, Fisk CJ, Donehower LA, Goodell MA. Aging leukemia. Leukemia 2008; 22: 1335–1342. hematopoietic stem cells decline in function and exhibit epigenetic dysregula- 107 Loh ML. Recent advances in the pathogenesis and treatment of juvenile mye- tion. PLoS Biol 2007; 5: e201. lomonocytic leukaemia. Br J Haematol 2011; 152: 677–687. 93 Dupont S, Masse A, James C, Teyssandier I, Lecluse Y, Larbret F et al. The JAK2 108 Abdel-Wahab O, Pardanani A, Rampal R, Lasho TL, Levine RL, Tefferi A. DNMT3A 617V4F mutation triggers erythropoietin hypersensitivity and terminal ery- mutational analysis in primary myelofibrosis, chronic myelomonocytic leukemia throid amplification in primary cells from patients with polycythemia vera. Blood and advanced phases of myeloproliferative neoplasms. Leukemia 2011; 25: 2007; 110: 1013–1021. 1219–1220. 94 Anand S, Stedham F, Beer P, Gudgin E, Ortmann CA, Bench A et al. Effects of the 109 Beer PA, Delhommeau F, LeCouedic JP, Dawson MA, Chen E, Bareford D et al. JAK2 mutation on the hematopoietic stem and progenitor compartment in Two routes to leukemic transformation after a JAK2 mutation-positive myelo- human myeloproliferative neoplasms. Blood 2011; 118: 177–181. proliferative neoplasm. Blood 2010; 115: 2891–2900. 95 Swierczek SI, Yoon D, Bellanne-Chantelot C, Kim SJ, Saint-Martin C, Delhommeau F 110 Itzykson R, Kosmider O, Renneville A., Gelsi-Boyer V, Meggendorfer M, Morabito et al. Extent of hematopoietic involvement by TET2 mutations in JAK2V(6)(1)(7)F M et al. A prognostic score including gene mutations in chronic myelomonocytic polycythemia vera. Haematologica 2011; 96: 775–778. leukemia. J Clin Oncol 2013, in press. 96 Visconte V, Rogers HJ, Singh J, Barnard J, Bupathi M, Traina F et al. SF3B1 111 Gelsi-Boyer V, Trouplin V, Roquain J, Adelaide J, Carbuccia N, Esterni B et al. haploinsufficiency leads to formation of ring sideroblasts in myelodysplastic ASXL1 mutation is associated with poor prognosis and acute transformation in syndromes. Blood 2012; 120: 3173–3186. chronic myelomonocytic leukaemia. Br J Haematol 2010; 151: 365–375. 97 Chen E, Beer PA, Godfrey AL, Ortmann CA, Li J, Costa-Pereira AP et al. Distinct 112 Such E, Germing U, Malcovati L, Cervera J, Kuendgen A, Della Porta MG et al. clinical phenotypes associated with JAK2V617F reflect differential STAT1 sig- Development and validation of a prognostic scoring system for patients with naling. Cancer Cell 2010; 18: 524–535. chronic myelomonocytic leukemia. Blood 2013; 121: 3005–3015. 98 Itzykson R, Kosmider O, Renneville A, Morabito M, Buet D, Preudhomme C et al. 113 Cheng H, Kirtani VG, Gergis U. Current status of allogeneic HST for chronic SPOTLIGHT Two distinct mechanisms contribute to granulomonocytic hyperplasia in chronic myelomonocytic leukemia. Bone Marrow Transplant 2012; 47: 535–541. myelomonocytic leukemias (CMML). ASH Ann Meet Abstr 2012; 120:309. 114 Braun T, Itzykson R, Renneville A, de Renzis B, Dreyfus F, Laribi K et al. Molecular 99 Ricci C, Fermo E, Corti S, Molteni M, Faricciotti A, Cortelezzi A et al. RAS mutations predictors of response to decitabine in advanced chronic myelomonocytic contribute to evolution of chronic myelomonocytic leukemia to the proliferative leukemia: a phase 2 trial. Blood 2011; 118: 3824–3831. variant. Clin Cancer Res 2010; 16: 2246–2256. 115 Ding L, Ley TJ, Larson DE, Miller CA, Koboldt DC, Welch JS et al. Clonal evolution 100 Jamieson CH, Gotlib J, Durocher JA, Chao MP, Mariappan MR, Lay M et al. The in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. JAK2 V617F mutation occurs in hematopoietic stem cells in polycythemia vera Nature 2012; 481: 506–510. and predisposes toward erythroid differentiation. Proc Natl Acad Sci USA 2006; 116 Lyubynska N, Gorman MF, Lauchle JO, Hong WX, Akutagawa JK, Shannon K et al. 103: 6224–6229. A MEK inhibitor abrogates myeloproliferative disease in Kras mutant mice. Sci 101 Sarrazin S, Mossadegh-Keller N, Fukao T, Aziz A, Mourcin F, Vanhille L et al. MafB Transl Med 2011; 3: 76ra27. restricts M-CSF-dependent myeloid commitment divisions of hematopoietic 117 Fenaux P, Mufti GJ, Hellstrom-Lindberg E, Santini V, Finelli C, Giagounidis A et al. stem cells. Cell 2009; 138: 300–313. Efficacy of azacitidine compared with that of conventional care regimens in the 102 Rieger MA, Hoppe PS, Smejkal BM, Eitelhuber AC, Schroeder T. Hematopoietic treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, cytokines can instruct lineage choice. Science 2009; 325: 217–218. phase III study. Lancet Oncol 2009; 10: 223–232. 103 Wei J, Wunderlich M, Fox C, Alvarez S, Cigudosa JC, Wilhelm JS et al. Micro- 118 Ji H, Ehrlich LI, Seita J, Murakami P, Doi A, Lindau P et al. Comprehensive environment determines lineage fate in a human model of MLL-AF9 leukemia. methylome map of lineage commitment from haematopoietic progenitors. Cancer Cell 2008; 13: 483–495. Nature 2010; 467: 338–342.

Leukemia (2013) 1441 – 1450 & 2013 Macmillan Publishers Limited