The CBL gene (11q23.3) encodes an E3 ubiquitin ligase that negatively regulates signaling by promoting degradation of activated tyrosine kinase receptors. In addition, CBL is an adaptor protein that positively regulates signal transduction.1 Individuals with germline heterozygous CBL mutations present with a clinically variable condition that can resemble Noonan syndrome and will be further referred as the ‘CBL syndrome.’ 2These patients are at increased risk of developing juvenile myelo-monocytic leukemia (JMML), an aggressive myelodysplastic and myeloproliferative neoplasm of early childhood characterized by clonal macrophage/monocyte proliferation.3, 4 JMML cells are hypersensitive to granulocyte–macrophage colony-stimulating factor consecutively to the activation of the RAS-MAPK signaling pathway through the mutation of the following genes: PTPN11, NRAS, KRAS, NF1 or CBL. CBL-associated JMML can follow an aggressive clinical course or resolve without treatment.4, 5, 6 The current diagnostic criteria for JMML are based on an international consensus that recently incorporated NF1, RAS and PTPN11 mutational status and monosomy 7.7 CBL mutations have since been discovered and are also screened for in the workup of patients with suspected JMML.8, 9 CBL mutations reported so far are missense mutations or splice site variants clustered in the linker region, and at or near the zinc-coordinating amino acids of the RING finger domain.1 They all result in the expression of a CBL protein with defective E3 ligase activity that constitutively activates key RAS effector pathways.4 Importantly, in virtually all patients described so far, the wild-type CBL allele is lost in leukemic cells and replaced with the mutant allele by acquired uniparental isodisomy of the 11q23 chromosomal region leading to loss of heterozygosity (LOH) of the mutated CBL.4, 3, 6 Hence, LOH of the mutated CBL in hematopoietic cells is usually considered as a signature of JMML in patients with ‘CBL syndrome.’ However, to date, consequences induced by LOH of CBL mutant proteins remain partly unknown.
CBL mutations were screened in JMML patients of the French cohort (n=102) and in patients referred to our lab with a borderline JMML phenotype. The diagnosis of JMML was based on consensus criteria as described above and included centralized cytomorphological review of bone marrow and blood as well as a comprehensive genetic testing by bidirectional Sanger sequencing of tumoral DNA.5, 7 CBL mutations were identified in a total of 12 children. The germline origin of mutations was tested on constitutional DNA (fibroblasts and/or nails). 11q23 LOH was explored on tumor cells DNA paired with germline DNA by PCR analysis of microsatellite markers covering the 11q23 region and/or genome-wide single-nucleotide polymorphism (SNP)-array analysis (GeneChip Human SNP-Array 6.0, Affimetrix, Santa Clara, CA, USA).5 In vitro growth of myeloid progenitors from bone marrow was found positive in all CBL mutated patients who were tested (Table 1).10
A CBL mutation was identified in 9 out of 102 (9%) patients with JMML fulfilling consensus criteria.7 Although the overall M/F sex ratio of our cohort was 1.94, showing a large predominance of males in accordance with previous reports,8 it was inverted to 0.5 in patients with CBL-mutated JMML (Table 1). In six out of nine patients, the mutation was homozygous in JMML cells. All six patients had a germline CBL mutation. SNP array analysis confirmed the presence of an 11q23 LOH encompassing CBL, in line with the classical model described for CBL-driven JMML. However, in the three remaining patients, the JMML clone harbored a heterozygous CBL mutation. The absence of 11q23 LOH was confirmed in these patients by examination of data obtained by SNP array hybridization and/or microsatellites analyses, which is more sensitive. Of course, the presence of a minor subclone with LOH, which is not fully expanded at the time of JMML diagnosis but subsequently selected for during tumor evolution, is still possible. However, repeated analysis 4 months after diagnosis did not evidence the emergence of such a clone in patient 6 (Figure 1c). Interestingly, two out of three patients had a mutation affecting exon 7 splice (Figure 1; Table 1). One of these patients had a ‘CBL syndrome’ but in the two others, the CBL mutation was somatically acquired and selected for in JMML cells. Noteworthy, another somatic mutation activating RAS pathway was present in the JMML cells of these two patients (that is, PTPN11 and LOH encompassing NF1), suggesting that CBL mutation may represent a secondary event in these cases. In one patient, this could be confirmed by subsequent analysis of a second tumor showing wild-type CBL but the presence of the NF1 LOH. Of note, the presence of the NF1 loss in rhabdomyosarcoma cells of this patient suggests that NF1 loss occurred very early in development and might be considered as a mosaic.
Intriguingly, LOH was also evidenced in hematopoietic cells of three additional patients with ‘CBL syndrome’ and a borderline hematological phenotype that did not meet consensus criteria for JMML, that is, isolated splenomegaly, Sweet syndrome with multi-organ invasion by granulocytes and hemophagocytic syndrome (Table 1; Figure 1).
The first patient (P10) was referred initially for persistent splenomegaly and xanthogranuloma. At the age of 14 years, she developed a cerebral hemorrhage with a moyamoya arteriopathy on magnetic resonance imaging. A heterozygous c.1258C>G (p.R420G) mutation was found in germline DNA. LOH with homozygous CBL mutation was identified in blood cells, which was unexpected considering she had no peripheral blood anomalies but a mild monocytosis (1.3 × 109/l). Retrospective study showed that LOH was already present at the age of 6 years.
In the second patient (P11), the hematopoietic disorder consisted in a dramatic neutrophil proliferation. This young girl had unexplained growth delay and dysmorphic syndrome since birth. At the age of 5 years, she developed acute febrile neutrophilic dermatosis with multivisceral failure due to neutrophil invasion. White blood cell count showed increased circulating neutrophils and monocytes up to 25 × 109/l and 11 × 109/l, respectively. Bone marrow aspiration revealed a granular hyperplasia. A CBL c.1253T>C (p.F418S) homozygous mutation was identified in peripheral blood cells and in neutrophils infiltrating a cutaneous lesion. An inherited heterozygous mutation was found in her fibroblasts. Treatment with corticosteroid and mercaptopurine improved her condition and she remains healthy 2 years after with normal white blood cell, although LOH is still detected in her hematopoietic cells.
The third patient (P12) is a young boy who presented with a severe hemophagocytic syndrome secondary to an Epstein–Barr virus infection. Whole-exome sequencing of this patient and his two parents revealed the presence of a de novo c.1141T>G (p.C381G) CBL mutation, which was confirmed by Sanger sequencing. Here again, the CBL mutation was heterozygous in fibroblasts, with LOH in hematopoietic cells. Notably, alike the other patients, LOH of CBL persisted on several years although peripheral blood counts returned to normal.
These data show that, unlike what was previously thought, CBL mutation is not invariably associated with LOH in JMML. Notably, two patients with heterozygous mutations in JMML had mutations of the same splice acceptor site in intron 7 (Table 1). The only heterozygous CBL mutation reported so far in a JMML patient was a deletion.6 This is consistent with previous observations in other myeloid malignancies showing that 80% of the deletion mutations that arise from CBL splicing mutations are heterozygous versus less than 20% of the missense mutations.1 This finding raises an issue of whether the proteins containing deletions are more alike to induce transformation than those with point mutations, possibly via a dominant negative effect. Another surprising finding is that, in contrast with homozygous mutations, heterozygous CBL mutations were somatically acquired and associated with the presence of another RAS-activating lesion in two patients, suggesting that the CBL mutation can also represent a secondary event in JMML. Noteworthy, these two patients underwent a more aggressive course than other CBL-related JMML of our cohort.
On the other hand, LOH of mutated CBL has been evidenced in some of our patients with ‘CBL syndrome’ and non-tumoral hematological phenotypes. Clonal hematopoiesis observed in these patients reflects an overgrowth of the hematopoietic clone harboring LOH of the mutated CBL, in line with experimental data in mice showing that mutant Cbl proteins confer growth advantage when the normal Cbl is lost.11 More unexpected is the observation that long-term persistence of these mutated clones can be hematologically and clinically silent. Indeed, LOH was found in one patient with virtually no hematological abnormalities and persisted in other patients several years after regression of hematological alterations. Interestingly, long-term persistence of LOH has been also reported in some patients with bona fide JMML who continue to show homozygous CBL mutations in their peripheral blood despite having improved blood counts.4 Altogether, these observations suggest that CBL mutation is not always sufficient to drive and/or support leukemogenesis even in the absence of the normal CBL allele and despite persistence of a clonal hematopoiesis over years. This may be due to a lower oncogenic potential of some CBL mutations. However, mutations found in patients 10 and 11 have been reported in adult myeloid malignancies.12, 13 Alternatively, a specific background and/or additional oncogenic lesions may be needed in addition to CBL mutation to drive leukemogenesis. Interestingly, this is reminiscent of what has been reported for some patients with acquired RAS mutations and JMML with spontaneous remisson or non-malignant lymphoproliferative disease.14
In conclusion, LOH of the mutated CBL allele can be absent in children with bona fide JMML and CBL mutations. Conversely, although the occurrence of a mild transient myeloproliferation in the neonatal period cannot be absolutely ruled out in our patients, our data show that homozygous CBL mutations can be associated with non-malignant clonal hematological phenotypes different from JMML. Hence, LOH of the mutated CBL allele cannot be considered as a signature of JMML. Besides raising some questions about CBL-driven leukemogenesis, this is an important issue for patients’ management.
Our observations also highlight the variety of immuno-hematological phenotypes induced by CBL mutations. Besides JMML, a heterogenous spectrum of atypical immuno-hematological presentations can be observed in patients with ‘CBL syndrome’, ranging from very mild hematological symptoms to more aggressive presentations. The importance of Cbl in hematopoiesis has been demonstrated in knockout mice that show hyper responsiveness to hematopoietic growth factors, expansion of the progenitor and stem cell pool, and mild myeloproliferative features.11 The incidental finding of CBL mutations with LOH in disorders involving monocyte/macrophages, neutrophils and/or the immune system underlines the multilineage impact of CBL mutations on hematopoiesis.
SNP arrays were performed on the Plateforme de Génomique du GHU Nord and analyzed with the help of Steven Gazal. We thank Dr Franck Bourdeault (Institut Curie, Paris, France) for analysis of the rhabdomyosarcoma sample.
About this article
Changes in the World Health Organization 2016 classification of myeloid neoplasms everyone should know
Current Opinion in Hematology (2017)