FMS-Like Tyrosine Kinase 3 in Normal Hematopoiesis and Acute Myeloid Leukemia


  • Bertrand W. Parcells,

    1. Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories, Mattel Children's Hospital, Jonsson Comprehensive Cancer Center, Los Angeles, California, USA
    Search for more papers by this author
  • Alan K. Ikeda,

    1. Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories, Mattel Children's Hospital, Jonsson Comprehensive Cancer Center, Los Angeles, California, USA
    Search for more papers by this author
  • Tiffany Simms-Waldrip,

    1. Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories, Mattel Children's Hospital, Jonsson Comprehensive Cancer Center, Los Angeles, California, USA
    Search for more papers by this author
  • Theodore B. Moore,

    1. Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories, Mattel Children's Hospital, Jonsson Comprehensive Cancer Center, Los Angeles, California, USA
    Search for more papers by this author
  • Kathleen M. Sakamoto M.D., Ph.D.

    Corresponding author
    1. Division of Hematology/Oncology, Department of Pediatrics, Gwynne Hazen Cherry Memorial Laboratories, Mattel Children's Hospital, Jonsson Comprehensive Cancer Center, Los Angeles, California, USA
    2. Department of Pathology and Laboratory Medicine, David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, USA
    3. Division of Biology, California Institute of Technology, Pasadena, California, USA
    • A2-412 MDCC CHS, Division of Hematology-Oncology, Mattel Children's Hospital, David Geffen School of Medicine, UCLA, 10833 Le Conte Avenue, Los Angeles, California 90095-1752, USA. Telephone: 310-794-7007; Fax: 310-206-8089
    Search for more papers by this author


Ligand-mediated activation of the FMS-like tyrosine kinase 3 (FLT3) receptor is important for normal proliferation of primitive hematopoietic cells. However, activating mutations in FLT3 induce ligand-independent downstream signaling that promotes oncogenesis through pathways involved in proliferation, differentiation, and survival. FLT3 mutations are identified as the most frequent genetic abnormality in acute myeloid leukemia and are also observed in other leukemias. Multiple small-molecule inhibitors are under development to target aberrant FLT3 activity that confers a poor prognosis in patients.


The hematopoietic system is organized into a strict hierarchy that sustains a steady-state production of more than one million blood cells per second [1]. All hematopoietic lineages propagate from the pluripotent hematopoietic stem cells (HSCs) that exclusively possess the capacity for self-renewal to maintain the life-long balance of blood lineage reconstitution. The unique expression profiles of hematopoietic cells at various stages of commitment allow cytokines and growth factors to direct the differentiation cascade until development terminates at mature lineage-specific progeny [24]. The hematopoietic system has evolved to offset the cellular processes of differentiation and proliferation to limit opportunities for oncogenic mutations. However, in spite of the molecular and systemic checks and balances, malignancies arise in the hematopoietic system as various forms of leukemia.

Acute myeloid leukemia (AML) is a heterogeneous malignant disorder of myeloid precursor cells with clinical features of increased blasts in the blood and bone marrow [5]. Mutations leading to the constitutive activation of the FMS-like tyrosine kinase 3 (FLT3) receptor, also known as fetal liver kinase 2 and human stem cell kinase 1, have been identified as the most frequent genetic disorder in AML and also confer a poor clinical prognosis. This review examines the function of normal FLT3 signaling and the aberrant downstream signaling caused by FLT3 mutations that evade normal control mechanisms. Subsequently, the clinical implications associated with this oncogenic pathway and the potential for targeted therapy are considered.

FLT3 in normal Hematopoiesis

FLT3 Ligand

Several cytokines in the bone marrow act through stem cell-specific or progenitor cell-specific receptors to regulate the capacity of immature hematopoietic cells to potentiate downstream multilineage expansion. The FLT3 ligand (FL) regulates early hematopoiesis by stimulating the FLT3 signal transduction pathway. FL is a type I transmembrane protein belonging to a small family of cytokines including stem cell factor and macrophage colony-stimulating factor (CSF) [68]. FL mRNA is expressed in most hematopoietic and nonhematopoietic tissues, yet detection of the FL protein, as a membrane-bound and soluble isoform, is limited to T lymphocytes and stromal fibroblasts of the bone marrow microenvironment [6, 7]. Expression of the FL in cells that similarly express the FLT3 receptor suggests autocrine and paracrine signaling mechanisms for FLT3 response [9]. FL is constitutively expressed in these cells during steady-state hematopoiesis, but it is maintained at low levels in the serum [10]. Normal regulation limits FL release from intracellular stores to prevent hyperstimulation of primitive hematopoietic cells. Recent studies suggest that one pathway for leukemogenesis may arise from a disruption of intracellular FL retention [11].

FLT3 Receptor

FL associates with the transmembrane receptor FLT3, a member of the type III receptor tyrosine kinase (RTK) subfamily that includes c-KIT, c-FMS, and platelet-derived growth factor (PDGF) α/β [1215]. These cytokine receptors share close homology and mutually contribute to normal differentiation and proliferation of primitive hematopoietic cells. The human flt3 gene is over 1,000 kilobases in length and is composed of 24 exons located on chromosome 13 (13q12) [1618]. The flt3 gene encodes a 993-amino acid protein that is observed as a major 140-kDa band and a minor 160-kDa band because of N-linked glycosylation, and a 130-kDa band when unglycosy-lated and not membrane bound [8, 14, 1921]. The FLT3 receptor is characterized by an extracellular domain consisting of five immunoglobulin-like domains, one transmembrane region, a cytoplasmic juxtamembrane (JM) domain, and two cytoplasmic kinase domains linked by a kinase-insert domain [17] (Fig. 1).

FLT3 Receptor Expression

The narrow range of cells expressing the FLT3 receptor primarily determines the specificity of FL signaling. FLT3 expression in the bone marrow is restricted to CD34+ cells and the subset of dendritic precursor cells. FLT3 expression is correlated with “short-term” reconstituting HSCs, the Lin+Sca-1+c-Kit+ FLT3+ compartment [22], and there is no definitive evidence for its expression in “long-term” HSCs. FLT3 is expressed at high levels in early cell populations with lymphoid and myeloid differentiation potential [23, 24]. These progenitor cells possess granulocyte, monocyte, B cell, and T cell developmental potential, but in contrast to more primitive stem/precursor cells, the population cannot reconstitute erythroid and megakaryocytic lineages [22]. FLT3 is also observed in the placenta, gonads, and brain, although its function in these tissues is unknown [5].

Synergy of FL with Other Cytokines

Experimental evidence indicates that FL signals synergistically with other growth factors to promote proliferation in early progenitor cells of the myeloid and lymphoid lineages. FL-mediated signaling is highly dependent on related growth factors such as interleukin 3 (IL-3), granulocyte colony-stimulating factor (G-CSF), colony-stimulating factor-1 (CSF-1), and granulocyte macrophage colony-stimulating factor (CM-CSF), because FL inefficiently promotes proliferation as a single cytokine in vitro [5, 2528]. However, FL is a potent inducer of hematopoietic progenitor cell expansion when acting in conjunction with other hematopoietic growth factors and interleukins. For example, FL exhibits the greatest proliferative effect on primitive cells and more committed myeloid progenitor when combined with the KIT ligand and IL-3 [29]. FL combines with myeloid growth factors to enhance growth of primitive long-term culture-initiating cells, as well as committed colony-forming cells [5]. In vivo analysis of FL function further supports its vital role in maintenance and proliferation during early hematopoiesis. Targeted disruption of FL in mice reduced myeloid progenitor cells, whereas injection of FL resulted in transient HSC expansion and significant downstream expansion evidenced by bone marrow hyperplasia, splenomegaly, hepatomegaly, and enlarged lymph nodes [30].

Normal FL signaling is also associated with proliferation of the lymphoid lineage. In fact, targeted disruption of FL in mice led to a particular reduction of lymphoid precursor cells within the general reduction of primitive hematopoietic cells. FL−/−caused impairment of the immune system by reducing numbers of pro-B cells, dendritic cells (DCs), and natural killer cells [31]. Ligand-stimulated FLT3 induces development and expansion of dendritic cells [32, 33], and the number of cells expressing DC markers significantly increased from exogenous FL expression compared with the administration of related growth factors [34]. In fact, expression of FL has been used therapeutically to promote regression of solid and hematopoietic tumors in mouse models by enhancing immune response [3538]. Experiments in which FL was co-expressed with IL-7 and IL-11 resulted in the clonal expansion and differentiation of B-cell progenitors. Co-expressoin of FL with granulocyte-macrophage CSF or IL-4 promoted DC differentiation in vivo and in vitro [34].

FLT3 Receptor Signaling

FL binds to the FLT3 receptor to induce formation of a homodimer in the plasma membrane. The dimer couples cytoplasmic domains thereby enabling transphosphorylation of specific tyrosine residues, likely Tyr-589 and Tyr-591, on the JM domain [39]. The resulting conformational change exposes phosphoryl acceptor sites in the tyrosine kinase domain to initiate autophosphorylation. The downstream signaling cascade involves phosphorylation and activation of multiple cytoplasmic effector molecules in pathways involved with apoptosis, proliferation, and differentiation. The signal is transduced from the FLT3 receptor to the p85 subunit of phosphatidylinositol 3-kinase, a regulatory protein, and to the adaptor protein growth factor receptor-bound protein 2 [40, 41]. The FLT3 receptor signals to multiple other effectors of metabolism and of proliferation, including phospholipase Cγ1 and a GTPase-activating protein of the Ras proliferation pathway [40, 41]. Normal FLT3 signaling also activates extracellular-signal regulated kinase 1/2 but leads to only weak phosphorylation of STAT5, which is a key target during deregulated signaling [42]. Recent studies also implicated Src homology 2 (SH2)-domain-containing inositol phosphatase (Shp) as a direct effector of activated FLT3 [43]. The FLT3 signal-transduction pathways are not conclusively mapped, yet current understanding of downstream pathways provides molecular support for its vital role in proliferation and apoptosis of primitive hematopoietic cells.

Despite the observed molecular effects of the FLT3 pathway, knockout mice have relatively stable hematopoiesis [44]. Although these results indicate the presence of functional redundancies between hematopoietic growth factors, they do not detract from the role of FLT3 in differentiation and proliferation of early progenitor cells of the myeloid and lymphoid lineages.

FLT3 in Leukemia

Self-renewal and growth factor-independent proliferation are two important traits for oncogenesis. The involvement of FLT3 in proliferation of highly undifferentiated hematopoietic cells suggests the oncogenic potential of this signaling pathway. Clinical and experimental evidence both indicate that FLT3 is a proto-oncogene with the capacity to enhance survival and proliferation of leukemia blast cells. Wild type FLT3 is expressed in a wide range of hematopoietic malignancies, including acute lymphoid leukemia and mixed lineage leukemia; most notably, it is expressed in 70%–100% of AML [18, 45, 46]. The aberrantly activated FLT3 pathway is observed in about 29.9% of patients with AML [47]. Two types of mutations have been attributed to the deregulated FLT3 receptor.

FLT3 Internal Tandem Duplication Mutation

An in-frame internal tandem duplication (ITD) mutation in the JM domain of the FLT3 receptor correlates with the highest frequency of FLT3 related AML cases. Clinical studies identify the FLT3-ITD mutation in 17%–26% of AML cases [48, 49]. The ITD was first identified when screening primary leukemia specimens for FLT3 expression, through reverse transcriptase-polymerase chain reaction (PCR), produced amplification products that were longer than expected in multiple patients [50]. DNA sequencing revealed that the insertional mutation occurs within exon 14 and varies in length from 3 to >400 base pairs [51]. The cause of the duplication event is unknown, although it has been proposed that the mutation results from a general failure in a slippage repair or a mismatch repair mechanism during DNA replication [52, 53]. FLT3-ITD has been identified in all French American British subtypes, a categorization of malignant cells based on genotype and morphology. It appears at the highest frequency in the M3 (promyelocytic) subtype and at the lowest frequency in the M2 (myeloblast with differentiation) subtype [50, 5459].

The ITD mutation causes constitutive activation of the FLT3 receptor. The JM region of many RTKs acts through various mechanisms to regulate the kinase domain. The crystal structure of the normal FLT3 receptor offers direct insight into the mechanism used by the JM domain to regulate catalytic activity of the kinase domain and suggests how the ITD disrupts this mechanism. FLT3 activity occurs when normal phosphorylation of specific tyrosine residues prevents the JM domain from folding properly to induce autoinhibition [39]. The structural analysis of the EphR2 receptor, of a related RTK subfamily, also suggests that the mechanism of induced phosphorylation occurs by disrupting the normal steric hindrance of the JM domain [60]. Moreover, this repressive sequence is conserved in the JM domains of c-KIT and PDFGβ [6163]. The ITD insertion generally occurs near the JM hinge region and may offset the JM orientation thereby causing a “leaky” autoinhibition of catalytic activity.

There is no correlation between the size of the insertional mutation and the degree of autophosphorylation [47]. Some experiments indicated that the ITD mutation caused maximal activity of the receptor because exogenous FL expression failed to enhance the level of signaling, whereas other experiments observed enhanced pathway signaling through FL expression [6466]. Such conflicting results may be reconciled by further analysis into the ratio of wild type to mutant receptors on the plasma membrane. This ratio may affect the degree of autophosphorylation because the FLT3-ITD receptor can homodimerize with mutant receptors or heterodimerize with wild-type receptors independent of the ligand.

Single-Amino Acid Mutations in FLT3

Missense point mutations in the kinase domain can also confer constitutive activation of the FLT3 receptor. The most common activating point mutation is the substitution of tyrosine for aspartic acid at position 835 within the activation loop of the kinase domain. Point mutations at other positions, such as 836 or 841, have also been associated with FL-independent activation [67, 68]. The FLT3-D835 mutation is observed in 7% of AML cases and is easily identified because it results in the loss of an EcoRV restriction site [47, 54, 69]. The normal aspartate residue is integral for regulating the activation loop, as evidenced by its conservation across several RTK subfamilies. Furthermore, point mutations at this site in related receptors, such as c-KIT, MET, and RET, are associated with deregulated signaling [7072]. The mechanism for activation likely parallels that of related proteins whereby the substitution stabilizes the “open” ATP-binding conformation of the activation loop [72].

FLT3 Mutant Signaling

Parallels between the FLT3 receptor and related RTKs suggest that activating mutations disrupt an autoinhibitory mechanism that is critical for maintaining the inactive conformation of the kinase domain. The mutant receptor activates effector proteins to mediate proliferation and survival and blocks differentiation through ligand-independent autophosphorylation. Identifying molecules of the mutant signaling pathway offers insight into its role in the pathogenesis of leukemia (Fig. 2).

Aberrant stimulation of proteins that confer a proliferative advantage was first identified in mutant-FLT3 myeloid leukemia cells. In vitro, the 32D mouse myeloid cell line harboring the ITD-mutation showed factor-independent proliferation and resistance to radiation-induced apoptosis [73, 74]. Both STAT5 and Ras pathways were shown to be activated. Constitutive phosphorylation of mitogen-activated protein (MAP) kinase has been identified in FLT3-ITD transfected Ba/F3 lymphoid cell line and likely participates in the Ras pathway, causing factor-independent growth and survival [73]. Evidence that inhibition of MAP kinase suppressed colony formation further supports the necessity of this pathway, although other studies indicate that its upregulation alone is insufficient to induce proliferation. Increased expression of the protein kinase B (AKT) has likewise been attributed to promoting survival and proliferation. Although expression levels of MAP kinase and AKT increased weakly in mutant FLT3, their distinction is supported by evidence of their transient phosphorylation following normal FLT3 activation. Brandts et al. recently confirmed prior evidence by Scheijen et al. [75] that AKT inhibits apoptotic signaling by phosphorylating Forkhead family member Foxo3α in FLT3-ITD Ba/F3 cells [76]. The inactivation of Foxo3α prevents transcription of its proapoptotic target genes in the Bcl2 family. A similar pathway involving the Foxo family members has been observed in Bcr-abl-mediated transformation [77].

Constitutively phosphorylated STAT5 has been observed in primary AML blasts and in immortalized cell lines Ba/F3 and 32D that express the mutated FLT3 receptor [64, 73, 74, 78, 79]. The phosphorylation of STAT5 contrasts most clearly with the wild-type signaling pathway because normal FL-activated of FLT3 does not cause significant STAT5 activation. This evidence suggests that mutant FLT3 receptors recruit unique downstream phosphorylation targets to induce the abnormal phenotype [47]. It remains unclear whether mutant FLT3 receptors promote leukemogenesis by increasing quantitative signaling of normal effector proteins or by causing qualitative differences in downstream targets.

STAT5 is known to activate PIM-1, and RNA microarray data identified a consistent increase in PIM-1 expression in FLT3-ITD cells [80]. Quantitative PCR further implicated PIM-1 as a target of aberrant STAT5 expression by correlating a 10-fold decrease in its expression with FLT3 inhibition. PIM-1 is normally induced by a number of FL-related cytokines, including G-CSF and IL-3, to increase cell mitogenesis and survival. The effect of PIM-1 activity occurs through phosphorylation of a diversity of substrates including Cdc25A, which is key to cell cycle progression, and Bad, which blocks proapoptotic signaling. PIM-1 has also been reported to cooperate within antiapoptotic pathways in chronic myeloid leukemia (CML) cell lines.

FLT3 and Differentiation

FLT3 activity may also regulate differentiation pathways. Differentiation is tightly associated with the loss of proliferative capacity and increased propensity toward apoptosis. Normal myeloid progenitor cells proceed along a differentiation gradient to arrive at a terminally differentiated cell type with a limited lifespan. Blocking differentiation has been identified as an important mechanism for cancer progression. The preferential expression of FLT3 in primitive hematopoietic cells suggests its role in regulating differentiation. FL activation of 32D cells transfected with wild-type FLT3 mitigated their progression toward differentiated neutrophils but could not induce a complete block [81]. However, 32D cells transfected with FLT3-ITD showed a complete block to morphologic differentiation, demonstrated at the molecular level by inhibition of prodifferentiation and myeloid proteins. Further investigation identified the inhibition of CCAAT/enhancer-binding protein α (C/EBPα) as a critical pathway to escape differentiation [82]. Yet the study also found primary cells from one AML patient to reveal high levels of C/EBPα, which indicates that other contributing mutations may be more dominant inhibitors of differentiation. The capacity of FLT3 to block differentiation is additionally questioned by evidence that mice transplanted with FLT3 mutant cells fail to develop AML blasts phenotypes due in part to differentiation [81].

However, RGS2 mRNA, a regulator of G-protein signaling, was observed to be significantly repressed in the majority of primary AML bone marrow samples that harbor the FLT3-ITD mutation compared with AML samples lacking the mutation [83]. Several myeloid cell lines have been shown to induce RGS2 during granulocytic differentiation, and RGS2 overexpression in 32D cells expressing FLT3-ITD overcomes the block to differentiation. The study suggests that FLT3-ITD signaling represses differentiation to some degree in AML cases, depending on the cellular context.

FLT3 Phosphatases

The constitutive activation of FLT3 may be further promoted by suppression of protein-tyrosine phosphatases (PTPs). The SH2 domain-containing PTPs SHP-1 and SHP-2 are commonly involved in regulating cytokine- and growth factor-mediated signaling pathways [84]. Phosphatases are critical to balanced RTK activity, and deviations in activity have been reported in many leukemias and lymphomas. Chen et al. showed that SHP-1 was suppressed threefold in FLT3-ITD-transformed TF-1 cells and increased SHP-1 expression correlated with inhibition of FLT3 phosphorylation in primary AML cells harboring the ITD mutation [84]. SHP-1 loss of function led to factor-independent growth. Increased phosphatase activity normally occurs in response to kinase activation, yet the observed SHP-1 suppression common to RTK-mediated malignancies suggests the importance of overcoming PTP activity to shift the phosphorylation balance toward a proliferative advantage. SHP-1 also regulates signaling of related hematopoietic receptors, and its suppression by FLT3 may induce synergistic effects by sensitizing cells to multiple growth factors.

FLT3 and In Vivo Models of Leukemia

At the molecular level, FLT3 appears to recruit many proteins involved in proliferation, self-renewal, and anti-apoptosis. The expression profiles of mutant FLT3 cell lines support its association with leukemogenesis by identifying a number of genes involved with growth and survival. However, the diversity of protein function and the redundancies in cellular processes prevent clear correlations between proteins and their function.

FLT3 mutations are insufficient to induce leukemia blasts in vivo [81] despite the diversity of avenues that the FLT3 receptor mutations promotes deregulation. FLT3-ITD transduction into primary murine hematopoietic progenitors induced a myeloproliferative phenotype but could not confer cells with the necessary arrested differentiation that is associated with AML. The Knudson Two Hit Model is therefore supported by evidence that FLT3-ITD can only promote AML when expressed concurrently with other selected mutations. Kelly et al. transduced FLT-ITD into bone marrow cells of PML/RARα transgenic mice and observed complete penetrance and rapid progression of acute promyelocytic leukemia (APL) compared with the long latency period for leukemogenesis in transgenic mice without FLT3 mutation [85]. The association of the PML/ RARα mutation with block to differentiation suggests that although mutant FLT3 may effect differentiation, transformation depends on a more potent mechanism. The block to differentiation may be a critical limitation for leukemogenesis through aberrant FLT3 signaling.

Nevertheless, these studies are critical to implicating the mutated FLT3 receptor in leukemogenesis. As a single mutation, FLT3 induces a myeloproliferative phenotype, and primary AML cells expressing the FLT3-ITD receptor engraft into NOD/SCID mice with greater efficiency that those lacking the mutation [86].

Clinical Implications of FLT3 Mutations in AML

Substitution mutations in the tyrosine kinase domain (TKD) and insertional mutations in the JM domain of FLT3 represent the most frequent genetic abnormality associated with AML. Both mutations lead to constitutively active FLT3, although studies indicate that each mutation presents unique clinical features.

The FLT3-ITD mutation is recognized at diagnosis by leukocytosis, with the mean white blood cell (WBC) count typically elevated to levels significantly higher than that of patients without the mutation [87]. The incidence of FLT3-ITD mutations correlates with the age of AML patients. Its frequency in pediatric studies ranges from 5%–16% [55, 58, 8890], whereas FLT3-ITD occurs in 23%–35% of adult AML cases [48, 49, 51, 68, 69, 9194]. Three studies on the pediatric subgroup indicate that FLT3-ITD appears at a higher frequency in children over 10 years, which corresponds to the adult frequency, whereas studies on the adult AML population show no increase in FLT3 mutation frequency with increased patient age. The incidence of FLT3-ITD mutations is also observed at various frequencies that correlate with other genetic factors. The mutation appears less frequently in patients with poor-risk cytogenetics and with core-binding factor translocations [t(8;21) and (INV16)] [95].

Multiple studies indicate that FLT3-ITD mutations confer a poor prognosis in AML patients less than 60 years of age [48, 49, 51, 68, 69, 9194]. The prognostic impact of FLT3-ITD likely fluctuates among patients because the mutation is often identified in conjunction with other contributing genetic abnormalities associated with the highly heterogeneous disease. For example, the incidence of FLT3-ITD mutations in patients with moderate and good-risk cytogenetics correlates with a significantly reduced clinical outcome, whereas the prognosis of APL patients harboring FLT3-ITD appears minimally affected [47]. However, a recent multivariate analysis of 250 adult AML patients indicated that FLT3-ITD is an independent marker of poor prognosis [96], a correlation that is also attributed to the MLL gene [97] and the ETS-related gene [98] in AML. The data suggest that despite the substantial diversity of molecular abnormalities documented in AML, certain mutations, such as FLT3-ITD, exert substantial influence on tumor progression. Moreover, all but 1 of 16 studies on adult AML found a significant decrease in overall survival, disease-free survival, and event-free survival [47]. Two studies that examined 1,000 patients 4 years after receiving comparable therapies demonstrated that FLT3-ITD expression correlated with decreased overall survival [48, 69]. Another study, examining 224 AML patients with normal cytogenetics and a median follow-up of 34 months, found that patients expressing the FLT3-ITD mutation had a significantly lower overall survival [92]. Two other studies observed no effect of the FLT3-ITD mutation on overall survival, although these results may be explained by the relatively early follow-ups of 11.1 and 12.2 months [51, 68].

Multiple studies used PCR to correlate high ratios of mutant to wild-type FLT3 with decreased overall survival [48, 68]. The mutant to wild-type ratio often increased at relapse and a retrospective analysis of minimal residual disease (MRD) status for 11 AML patients indicated an increased tendency for relapse in those with FLT3-ITD positive MRD status [99]. This provides evidence that FLT3 expression levels significantly affect prognosis and relapse. This suggests that the pathway exerts a strong influence on leukemia progression. However, the heterogeneity in AML blast gene expression and the notable fraction of AML patients that lose the FLT3-ITD mutation in relapse indicate that multiple factors determine disease progression and that FLT3-ITD should not be exclusively used as a marker for MRD.

In pediatric cases, the ITD mutation provides a more negative and independent prediction of prognosis. Data from six combined pediatric studies, which examined 461 pediatric AML patients, identified a 19% survival rate for cases associated with FLT-ITD compared with a 58% survival rate for patients with wild-type FLT3 receptors [50, 100]. One of the studies reported a remission induction rate of 40% and an event-free survival of only 7%, compared with 74% and 44%, respectively, in patients without the mutation [50]. Unlike adult AML findings, pediatric studies also found a significant decrease in complete remission of AML harboring the FLT3-ITD mutation [58, 88].

The clinical pathogenesis of FLT3-TKD expressing AML is less understood than the FLT3-ITD due to its lower frequency in patients, which limits the number of statistically significant conclusions [55, 68, 88, 101103]. Conflicting clinical results have also prevented substantial inferences on the effects of FLT3-TKD. For example, one clinical study of 201 AML patients found no significant correlation of increased mean WBC counts or age-dependent frequency of FLT3-TKDs but associated FLT3-TKD with a shorter disease-free survival [69]. However, a larger clinical study of 979 AML patients less than 60 years of age found that FLT3-TKD was associated with elevated mean WBC counts without observing correlation with disease-free survival [68]. A recent meta-analysis of four major studies of FLT3 in AML patients indicated that TKD mutations confer a significant decrease in overall survival and are comparable to ITD mutations in lowering disease-free survival [87]. Although this analysis provides interesting insights, the heterogeneity and the abstracted nature of the patient data requires cautious interpretation. Genetic analysis of FLT3-TKD samples indicated that levels of FLT3 expression are greater than in either FLT3-ITD or wild-type FLT3 [104]. Studies have also identified a higher frequency of FLT3-TKD mutations in patients with MLL duplications or a double-stranded break in the MLL gene. Future clinical studies must examine larger AML populations to achieve significant insight into potential effects of FLT3-TKD mutations.

FLT3 Inhibitors

The fundamental principle of small-molecule therapy is to inhibit specific RTKs involved with tumorigenesis while minimizing the inhibition of normal cellular signaling to limit toxicity. The recognition of small-molecule inhibitors as therapy for deregulated RTKs emerged in the wake of the clinical success of imatinib mesylate in CML. CML progresses almost exclusively through the constitutively active bcr-abl fusion protein tyrosine receptor. Its robust response to imatinib provided proof of principle for the potential efficacy of targeted therapy [105]. Preliminary clinical data from therapeutic targeting of the FLT3 mutations in AML suggests that small-molecule inhibitors may effectively complement traditional therapy regimens to treat a broader range of malignancies.

FLT3 is an appropriate candidate for targeted therapy because it is expressed in many hematopoietic malignancies, it is the most frequent molecular abnormality in AML, it confers a poor prognosis, and its signaling cascade has been implicated in multiple tumorigenic pathways. Currently, there are several small-molecule therapies at various stages of development that target mutant forms of the FLT3 receptor (Table 1). All of the FLT3 inhibitors being studied are heterocyclic compounds with a purine ring-like subunit that competitively inhibits ATP binding to FLT3 [106]. Data on related receptors suggests that the molecules inhibit activity by entering into the ATP binding pocket through an induced fit or lock-and-key manner, although the specific mechanism may vary between inhibitors. A brief summary of the molecules is given below.

The compounds AG1295 and AG1296 were the first to be identified as inhibitors of cells harboring the FLT3-ITD mutation through a similar screening process used to identify imatinib [107, 108]. Despite the in vitro efficacy of these molecules on primary AML samples, their low solubility prevented clinical translation [109].

SU5416 was the first indolinone compound to be identified as a potential inhibitor of FLT3 activity. It has also been shown to inhibit vascular endothelial growth factor 2 (VEGF2) and c-KIT [110114]. SU5416 caused cell cycle arrest and apoptosis in cell lines expressing activated FLT3 at IC50 100–300 nM [114, 115]. In a phase I clinical trial, 55 AML patients with relapse or who were older than 70 years received treatment in 4-week cycles of twice-weekly intravenous infusion [116]. Four patients showed a clinically measurable response whereas three others showed partial response. In a second phase I trial, 43 patients with refractory AML received the treatment twice-weekly [117]. Seven patients demonstrated partial response, and one patient went into AML remission. The clinical efficacy of the compound may be limited by its high protein-bound state within the plasma [118].

The indolinone derivative PKC412 (4′-N-benzoylstaurosporine) was originally identified as an inhibitor of protein kinase C [119]. PKC412 was recognized as a therapeutic agent for aberrantly activated FLT3 when findings indicated that it inhibited growth and survival of Ba/F3 cells transfected with FLT3-ITD at IC50 of 10 nM [120]. At higher concentrations, the inhibitory effect of PKC412 extended to RTK relatives PDGFα, PDGFβ, c-KIT, and VEGF2 [120]. Its clinical efficacy as a therapy against FLT3 mutations was examined in phase II trials wherein 20 relapsed or refractory AML patients received 75 mg orally three times daily [121]. The drug showed measurable activity, evidenced by a 50% reduction in peripheral blast count in 14 patients, a greater than 2-log reduction in peripheral blast count in seven of these patients, and a 50% reduction in bone marrow blast counts in six patients. PKC412 was generally well tolerated.

The compound SU11248 is another indolinone derivative shown to inhibit FLT3, as well as VEGF2, PDGFβ, and fibroblast growth factor 1 (FGF1) [122, 123]. SU11248 inhibited proliferation and survival of MV4–11 and FLT3-ITD transfected 32D cells at IC50 of 10–50 nM [123]. In a phase I trial, 29 AML patients orally received escalating doses of the drug [124]. One-third of patients exhibited a significant decrease in FLT3 phosphorylation, although a minority of patients experienced gastrointestinal or cardiac symptoms of toxicity.

MLN518 (CT53518) is a piperazinyl quinazoline that inhibits growth of FLT3-ITD-transformed cells in vitro and in vivo [125]. It has also been observed to inhibit wild-type FLT3, PDGFβ, and c-KIT [126128]. In vitro, MLN518 inhibited proliferation and survival of FLT3-ITD transfected Ba/F3 cells at IC50 of 10–30 nM [125]. A phase I study of patients with relapsed or refractory AML showed that two of six patients had >50% reduction in the number of bone marrow blasts [129].

The indolocarbazole derivative CEP-701 was originally identified as an inhibitor of TrkA tyrosine kinase [130]. Its utility in FLT3 therapy was suggested by its potent inhibition of growth in FLT3-ITD transfected Ba/F3 cells at IC50 of 5 nM and downstream suppression of STAT5 and MAP kinase activity in mouse models [78, 131]. Data from phase I and II clinical trials indicated that 5 of 14 patients with relapsed or refractory AML harboring a FLT3 mutation showed measurable clinical response to the dose of 60 mg administered orally twice a day [132]. Most responses were observed by reduced peripheral blast counts; however, one patient showed a 95% reduction of bone marrow blasts before relapse 3 months later. Minimal toxicity was observed.

Komeno et al. recently identified Ki23819 as a potent inhibitor of mutant FLT3. The compound suppressed proliferation of MV4–11 cells at IC50 of 1 nM [133]. These preliminary in vitro results suggest promising translation into clinical inhibition of mutant FLT3, but the compound requires further study in mouse models. To summarize, most clinical studies with FLT3 inhibitors have demonstrated a moderate response, more evident in the peripheral blood than in bone marrow.


The development of small-molecule inhibitors is widely encouraged within the medical community because it offers an approach to treat cancer with relatively modest toxicity. Yet the targeting specificity that bestows a clinical advantage to this therapy may also be a critical flaw. The sensitivity of the compounds to the FLT3 structure confers not only specificity but also susceptibility to minor structural changes that are induced by mutation. The high mutational load that most tumors support during progression indicates a high probability of drug-resistant subclones. A mathematical model that closely recapitulates clinical findings from CML predicts that most leukemias will contain cells resistant to multiple molecular inhibitors at the time of diagnosis [134]. Clinical studies on CML have identified mutations in the bcr-abl receptor that confer resistance to imatinib therapy [135137]. Such findings on the frontrunner of targeted therapy suggest that similar obstacles will arise in AML therapy because it closely resembles the CML blast crisis phase when imatinib resistance most frequently develops [138].

Acquired resistance to small-molecule inhibition has been identified in AML blasts harboring constitutively active FLT3 receptors [139, 140]. Many compounds show variable capacities to inhibit FLT3 depending on the type of activating mutation. The compound MLN518 showed a broad range in its efficacy against eight different activating mutations at residue 835 in Ba/F3 cells [139]. Such evidence suggests that some small-molecule inhibitors are highly affected by single mutations and that this sensitivity may preclude their clinical application. Bagrintseva et al. report that in Ba/F3 cells the FLT3-D835Y variant retained sensitivity to SU5614 whereas the FLT3-D835H variant required a 100-fold increase in drug concentration [141]. Another study identified Asn-676, Ala-627, and Phe-691 as mutations in the activation loop that confer varying degrees of resistance [142]. These mutations recovered sensitivity to PKC412, SUS5614, and a CEP-701-like compound at higher concentrations in vitro, yet increased drug concentrations may not translate into a successful method for circumventing resistance in clinical AML. Many FLT3 inhibitors were originally identified as inhibitors of other RTKs and became directed specifically against FLT3 because the receptor showed inhibition at the lowest concentration threshold. If mutations can raise the concentration threshold to a level at which other RTKs respond, then normal cells may experience significant cytotoxicity, and clinical side effects may become severe. Evidence of acquired resistance does not invalidate the clinical potential of targeted therapy but suggests that curative treatment must extend beyond a single agent.

Data from in vitro and in vivo experiments offers promising strategies for combination therapy. The synergistic effects of SU11657 and all-trans retinoic acid induced rapid regression in an APL mouse model [143]. Combination therapy of rapamycin (Rap) and PKC412 inhibited proliferation of Ba/F3 cells expressing FLT3-ITD by more than 90%, whereas either monotherapy resulted in 60% inhibition at best [144]. The study also identified a PKC412-resistant cell line and showed that combination therapy with PKC412 and Rap could still synergistically inhibit proliferation. The result provides encouraging support for potential uses of combination therapy to overcome resistance conferred by activation loop mutations. A study by Levis et al. indicated that CEP-701 administered simultaneously or immediately following traditional chemotherapy synergistically caused cytotoxicity in Ba/F3 cells harboring the FLT3-ITD mutation [145]. The treatment regimen examined in this study is currently being applied to a clinical trial. The diversity of potential combination therapies and the significance of sequence in drug administration indicates that substantial optimization studies must be conducted before treatment regimens are designed for specific genetic abnormalities.

Many of the aforementioned compounds may translate into effective treatments for cancer, yet it is possible that they will only provide a selective force for highly resistant clones. One particular point mutation in the FLT3 receptor, Gly-697, poses a significant challenge because its structure confers resistance to multiple compounds, including PKC412, SU5614, and a molecule similar to CEP-701 [142]. If one mutation can overcome the dominant inhibitory mechanism of these therapies, then the cocktail strategy for combating resistance may prove ineffective. Cases of CML drug resistance also occur through overexpression of the bcr-abl. A study by Weisberg et al. identified a similar mechanism related to AML whereby Ba/F3 cells developed resistance to PKC412 by overexpressing the FLT3-ITD mutation [120]. Similar to the CML blast crisis, AML blast cells that harbor FLT3 mutations are likely to have additional mutations that may drive tumor progression while showing no response to FLT3 inhibition. Ideally, the response rate for FLT3 inhibition will compare to the 30% response to imatinib during CML blast crisis. Identification of new molecules will be key to preventing resistance in the future. Combining compounds that inhibit various proteins within the signaling pathway may also improve the efficacy of treatment.

Table Table 1.. Small-molecule tyrosine-kinase inhibitors
original image
Figure Figure 1..

Structure of the FLT3 receptor. Abbreviations: FLT, FMS-like tyrosine kinase 3; ITD, internal tandem duplication.

Figure Figure 2..

Signal transduction pathways downstream of FMS-like tyrosine kinase-3 receptor activation. Abbreviations: C/EBPα, CCAAT/ enhancer-binding protein α; ERK, extracellular-signal regulated kinase; MAP, mitogen-activated protein; PI-3 kinase, phosphatidylinositol 3-kinase.


The authors indicate no potential conflicts of interest.


B.W.P. is supported by the Skirball Foundation (awarded to T.B.M.). K.M.S. is a Scholar of the Lymphoma and Leukemia Society and is supported by the NIH (CA108545, HL 75826, RHL083077A), American Cancer Society (RSG-99-081-04-LIB), Department of Defense (CM050077), and the Diamond-Blackfan Anemia Foundation. Both T.B.M. and K.M.S. are funded by the UCLA Jonsson Comprehensive Cancer Center.