Severe congenital neutropenia and chronic neutrophilic leukemia: an intriguing molecular connection unveiled by oncogenic mutations in CSF3R

Acquired mutations in the colony-stimulating factor 3 receptor gene (CSF3R), truncating the cytosolic region of the CSF3R protein, were discovered almost two decades ago in severe congenital neutropenia (SCN) patients receiving CSF3 treatment to alleviate neutropenia. These CSF3R mutations are thought to drive clonal expansion by overriding CSF3 hypo-responsiveness of hematopoietic stem and progenitor cells (HSPCs) and are associated with leukemic progression in SCN patients. Furthermore, malignant transformation in one SCN patient coincided with acquisition of an additional auto-activating CSF3R mutation, supporting the hypothesis that perturbed CSF3 signaling contributes to leukemic transformation of SCN. While acquisition of CSF3R mutations had so far mainly been observed in SCN patients receiving CSF3 therapy, Maxson and colleagues1 discovered that both truncating and auto-activating CSF3R mutations are frequently present in chronic neutrophilic leukemia (CNL) and atypical chronic myeloid leukemia (aCML). Furthermore, this study provided preliminary evidence for clinical utility of tyrosine kinase inhibitors (e.g. dasatinib or ruxolitinib) to eradicate CSF3R mutant clones. Here, we discuss the biological and clinical significance of these new findings in the light of the unanticipated and intriguing connection between SCN and aCML/CNL, diseases that are respectively characterized by a severe paucity or an excess of neutrophils.

CSF3R mutations in severe congenital neutropenia

Severe congenital neutropenia is a genetically heterogeneous disorder with a variable inheritance.2 Mutations in ELANE, the gene encoding neutrophil elastase, are the most frequent cause of autosomal dominant and sporadic SCN.2 CSF3 (also known as granulocyte colony-stimulating factor, G-CSF) is successfully used to treat SCN, alleviating the severe neutropenia and reducing infection-related mortality in over 90% of the patients.3 A major complication in SCN patients is the high risk of developing myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML).4,5 This leukemic transformation correlates with the acquisition of mutations in CSF3R.6–8 These are mostly nonsense mutations, causing the truncation of approximately 100 amino acids of the cytoplasmic domain of the receptor.6–8 Activation of CSF3R truncation mutants with CSF3 induces a hyperproliferative response of myeloid progenitors in vivo and results in sustained STAT5 activation, an essential step in the clonal expansion of HSPCs expressing these mutant receptors.9–11 Furthermore, activation of truncated CSF3Rs results in excessive production of reactive oxygen species (ROS).12 The time between the first detection of CSF3R mutations and the diagnosis of MDS/AML varies greatly, i.e. between several months to up to more than ten years, indicating that additional aberrations play a role in leukemic transformation. Whole exome sequencing on serial hematopoietic samples of an SCN patient who progressed to MDS/AML 17 years after the start of G-CSF treatment identified 12 non-synonymous somatic mutations.13 Two of these mutations, in LLGL2 and ZC3H18, were acquired in a CSF3R mutant clone in the early SCN phase, 15 years before overt leukemia. This clone, carrying the delta 715 CSF3R truncating mutation (CSF3R-d715), expanded in time, whereas clones harboring functionally similar but structurally distinct CSF3R mutations (d717, d725, d730) became extinct. These latter clones did not harbor additional mutations that were found in the leukemic clone, which may explain why they disappeared over time. Additionally, a new mutation (CSF3R-T595I) in the extracellular domain of CSF3R on the already affected CSF3R-d715 allele was present in the AML phase, resulting in autonomous proliferation of myeloid progenitors.13 These findings suggest that leukemic progression of SCN is a multistep process in which clones with mutations arising early in the SCN phase (CSF3R-d715, LLGL2, ZC3H18) selectively expand and acquire additional mutations (e.g. in SUZ12, ASXL1, RUNX1, CSF3R-T595I) at later stages of leukemia development. The question as to whether CSF3R mutations cause leukemic progression of SCN, e.g. by evoking DNA damage as a result of elevated ROS production, or merely drive clonal expansion of the affected HSPCs still has to be resolved.14

CSF3R mutations in chronic neutrophilic leukemia and atypical chronic myeloid leukemia

Chronic neutrophilic leukemia and aCML are myeloproliferative diseases with an unfavorable prognosis, i.e. a median survival between 14 and 21 months after initial diagnosis.15,16 Until recently, the molecular pathogenesis of these diseases was unknown and the diagnosis based on morphological aspects, clinical criteria and exclusion of known genetic entities like the Philadelphia translocation indicative of CML, or JAK2 mutations indicative of myeloproliferative neoplasms (MPN).17,18 Maxson and colleagues sequenced 1862 genes, representing all kinases, phosphatases, cytokine receptors and selected adapter proteins and identified CSF3R mutations in 16 of 27 patients with CNL or aCML.1 The most prevalent mutation was a missense mutation in CSF3R, annotated as T618I, replacing a threonine by an isoleucine in the membrane proximal extracellular region of the CSF3R protein. This is the same mutation as the auto-activating CSF3R-T595I mutation found in SCN/AML; the difference in nomenclature is due to inclusion of the signal peptide of 23 amino acids in the amino acid numbering by Maxson et al.1,13 Significantly, 5 CNL patients also had CSF3R truncation mutations which were present on the same allele as the CSF3R-T618I mutation, identical to what was reported in SCN/AML.13 However, the prevalence of auto-activating membrane proximal mutations (14 of 27) exceeded the number of distal truncating mutations (6 of 27) in CNL and aCML, while in SCN the opposite is observed (R Beekman and IP Touw, 2013, unpublished data). There is some controversy about the occurrence of CSF3R mutations in aCML when strict WHO-defined criteria are applied, and on the same basis, the frequency of auto-activating CSF3R mutations in CNL is estimated to exceed 80%.19

Why do SCN/AML and CNL share identical CSF3R mutations?

CSF3R mutations are rare or absent in MDS, chronic myelomonocytic leukemia (CMML), primary myelofibrosis (PMF) and de novo AML,19–23 which makes the frequent involvement of these mutations in the pathogenesis of SCN/AML and CNL exceptional. Although it can easily be seen why isolated CSF3R auto-activating mutations lead to the uncontrolled expansion of myeloid progenitors in CNL and SCN/AML, it remains puzzling how the combination of CSF3R auto-activating and truncating mutations unite the molecular pathogenesis of these conditions. Specifically, it is unclear why one-third of the CNL patients also had acquired CSF3R truncation mutations. Multiple studies have shown that perturbed JAK/STAT signaling promotes the clonal expansion of pre-leukemic CSF3R mutant clones in SCN patients,9,11,24,25 which is in line with the fact that clones with CSF3R truncation mutations in SCN patients usually appear before presentation of overt leukemia. Similar information on the timing of acquisition of these mutations in CNL is missing, because pre-neoplastic blood or bone marrow samples from these patients are usually not available. A scenario of a CSF3-driven clonal expansion phase cannot immediately be envisaged in CNL, unless patients would first go through a stage of CSF3 hypo-responsiveness and bone marrow failure from which HSPCs escape through the acquisition of CSF3R truncating mutations. Although there is still no experimental evidence for this, the example of mutations in JAK2 and NRAS in MDS, MPN and the MDS/MPN overlap syndrome provides an analogy of how mutations affecting cytokine signaling may contribute to the disease transition from a myelosuppressive state towards a myeloproliferative neoplasm.26 Another implication of the clonal evolution model is that CSF3R mutations in CNL may be preceded by other defects. However, this remains to be determined, as the candidate approach adopted to identify mutations in CNL and aCML1 by Maxson and colleagues did not identify the full spectrum of abnormalities. Until now, mutations in SETBP1, encoding a nuclear protein that interacts with Hoxa9 and Hoxa10 promoters and enhances the self-renewal of HSPCs in mice27 were found to be most frequently co-expressed with CSF3R mutations in CNL.19 Recurrent SETBP1 mutations were also detected in secondary AML but not in de novo AML,28,29 suggesting that they may be driving the transition from a pre-leukemic towards a leukemic state. Notably, SETBP1 mutations were reported to elevate the ability of SETBP1 to immortalize HSPCs,28 but whether and how they co-operate with CSF3R mutations is unknown. Still, the possibility remains that other abnormalities, e.g. in epigenetic regulators, precede the acquisition of CSF3R mutations, as has been demonstrated in MPN with JAK2-V617F mutations.30 These mutations might provide yet another missing link between SCN/AML, CNL and aCML.

Clinical implications

The discovery of CSF3R mutations in CNL and aCML was part of an integrated approach involving the use of small molecule tyrosine kinase inhibitors and interfering RNAs.1 The results clearly indicate that the CSF3R auto-activating and truncating mutations affect distinct signaling pathways which can be targeted by different inhibitors. The JAK1/JAK2 inhibitor ruxolitinib effectively reduced proliferative signaling from the auto-activating CSF3R-T618I mutant in an in vitro model and a preliminary study showed the potential of ruxolitinib to reduce neutrophil levels in a CNL patient. Whether ruxolitinib, either as a single treatment or combined with, for example, hydroxyurea will lead to durable and meaningful responses in CNL needs to be addressed in future clinical studies. The reported finding that dasatinib selectively inhibits clones expressing truncated CSF3Rs merits further studies to analyze the combined effect of ruxolitinib and dasatinib on CNL and aCML clones carrying both types of CSF3R mutations. Furthermore, the effectiveness of dasatinib on clones carrying the truncated CSF3R may not only be of clinical significance for CNL and aCML, but also for SCN. Future studies will explore how dasatinib and other inhibitors affect G-CSF signaling from wild type and truncated CSF3Rs. This will reveal how inhibitors can be employed in vivo to selectively eradicate clones harboring truncated CSF3Rs in SCN patients, to reduce the risk of malignant transformation.

Footnotes

• Ivo Touw is Professor at the Department of Hematology at the Erasmus University Medical Center (Erasmus MC), Rotterdam, The Netherlands. His main research interests are the mechanisms involved in the pathogenesis of bone marrow failure syndromes, and their transformation to myelodysplasia and acute myeloid leukemia. Renée Beekman did her PhD training at the Erasmus MC. She currently works as a post doc researcher at the August Pi i Sunyer Biomedical Research Institute (IDIBAPS) in Barcelona, Spain. She has received fellowships from the EU (Marie Curie) and The Netherlands Organisation for Scientific Research (Rubicon). Her main interests are the epigenetic and genetic mechanisms underlying hematologic malignancies.
• Financial and other disclosures provided by the author using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are available with the full text of this paper at www.haematologica.org.