Molecular mechanisms associated with leukemic transformation of MPL-mutant myeloproliferative neoplasms

Abstract

Somatic activating mutations in MPL, the thrombopoietin receptor, occur in the myeloproliferative neoplasms, although virtually nothing is known about their role in evolution to acute myeloid leukemia. In this study, the MPL T487A mutation, identified in de novo acute myeloid leukemia, was not detected in 172 patients with a myeloproliferative neoplasm. In patients with a prior MPL W515L-mutant myeloproliferative neoplasm, leukemic transformation was accompanied by MPL-mutant leukemic blasts, was seen in the absence of prior cytoreductive therapy and often involved loss of wild-type MPL by mitotic recombination. Moreover, clonal analysis of progenitor colonies at the time of leukemic transformation revealed the presence of multiple genetically distinct but phylogenetically-related clones bearing different TP53 mutations, implying a mutator-phenotype and indicating that leukemic transformation may be preceded by the parallel expansion of diverse hematopoietic clones.

Introduction

Acquired mutations in MPL, encoding the thrombopoietin receptor, are found in the myeloproliferative neoplasms (MPN) essential thrombocythemia (ET) and primary myelofibrosis (PMF).1–5 Exon 10 alterations affect the juxtamembrane (W515L/K/A/R) or transmembrane (S505N) domains, resulting in ligand-independent receptor activation.6,7 An exon 9 T487A mutation, reported in a single case of de novo acute myeloid leukemia (AML), produced an ET-like phenotype in a mouse model8 although its prevalence in human MPN is unknown.

Virtually nothing is known about molecular events associated with disease progression in MPL-mutant MPN. Although mutations in TET2 and MPL may coexist,9 their clonal relationship has not been reported. A mutant allele burden exceeding 50% occurs in patients with MPL-mutant PMF or rarely ET1–5 and often reflects duplication of the mutant MPL allele by mitotic recombination.10–12 AML following a JAK2 V617F-positive MPN commonly lacks the JAK2 mutation,13–15 and although MPL mutations have been observed in unfractionated post-MPN AML bone marrow samples,2 the MPL status of the prior MPN and of purified blast cells was not established.

We have studied the role of MPL mutations in early and leukemic phase MPN, focusing on the prevalence of mutations in exon 9, the role of MPL and additional mutations in leukemic transformation and mechanisms by which the wild-type MPL allele is lost.

Design and Methods

Screening for MPL exon 9 mutations was performed on a cohort of 172 patients attending a single MPN clinic in Cambridge, UK. Three patients who developed AML following an MPL-mutant MPN were identified on an ad hoc basis from clinics in Cambridge, Ulm and Florence. Patients were diagnosed with ET, post-ET myelofibrosis or PMF according to published criteria.16,17 A diagnosis of AML transformation required 20% blasts or more in blood and/or bone marrow. Local Research Ethics Committee approval was obtained and studies were carried out in accordance with the principals of the Declaration of Helsinki. Cell fractionation and progenitor colony assays were performed as described.15 Leukemic blasts, purified by CD34-immunomagnetic selection, were 90% or more pure by morphological criteria. Mutations in MPL (exons 9 and 10), N/KRAS (codons 12, 13 and 61), CEBPA (exon 1), RUNX1 (all coding exons), GATA2 (exon 4), NPM (exon 12), WT1 (exons 7 and 9), TP53 (exons 4 – 8), CBL (exons 8 and 9), IDH1 (exon 2), IHD2 (exon 4) and TET2 (all coding exons) were assessed by direct sequencing. MPL copy number was assessed by real-time PCR using control regions on 13q and 9p.

Results and Discussion

Expression of the AML-associated MPL exon 9 T487A allele in mouse bone marrow cells produced an ET-like disease in vivo that was indistinguishable from a similar mouse model of MPL W515L,8 an allele associated with human ET and PMF. Of note, JAK2 V617F mutations have been observed in occasional patients with de novo AML, indicating that MPN-associated mutations may be seen in de novo acute leukemia, and/or occasional patients may present in blastic phase of a previously undiagnosed MPN.15 To ascertain whether the MPL T487A allele or other changes in the MPL extracellular-juxtamembrane domain are associated with chronic phase ET or with PMF, MPL exon 9 was sequenced in granulocyte DNA from 172 patients (Table 1). No mutations were detected. These data indicate that MPL exon 9 mutations occur rarely, if at all, in human MPN.

Progression to acute leukemia is observed in a proportion of patients with a JAK2-mutant, MPL-mutant or mutation negative MPN, and in ET the presence or absence of an MPL mutation does not appear to modulate this risk.4 To investigate the role of MPL mutations in leukemic transformation, we studied 3 patients with AML following an MPL W515L-positive MPN (Table 2). All 3 patients were negative for the JAK2 V617F mutation. In patients 1 and 2, bone marrow studies performed at AML progression showed granulocytic hyperplasia, dysplastic megakaryocytes, reticulin fibrosis of 3 or more (graded on a 0–4 scale) and clusters of CD34 cells, in keeping with the AML subtype ‘acute panmyelosis with fibrosis’18 (Figure 1A). In patient 3, AML was diagnosed by more than 95% blast cells in the peripheral blood. Patients 1 and 3 had received hydroxycarbamide but patient 2 had not received cytoreductive therapy. As patients with a JAK2 V617F-positive MPN may develop leukemia that lacks the JAK2 mutation,13–15 leukemia MPL mutation status was determined using purified blasts, free from contamination by the preceding MPN. In patient 1, initially diagnosed with ET, leukemic blasts were heterozygous for the MPL W515L mutation, whereas in patients 2 and 3, with preceding PMF and post-ET myelofibrosis, respectively, only the mutant MPL allele was detected (Figure 1B). The absence of wild-type allele in patients 2 and 3 might reflect acquisition of a second mutation, deletion of the wild-type allele or mitotic recombination.

Table 1.Cohort of patients with a myeloproliferative neoplasm screened for mutations in MPL exon 9.

Figure 1.Acute myeloid leukemia following an MPL-mutant myelo-proliferative neoplasm may be heterozygous or homozygous for the MPL W515L mutation, with homozygosity arising by mitotic recombination. (A) Bone marrow trephine biopsies from patients 1 and 2, acquired at time of leukemic transformation, showing prominent dysplastic megakaryocytes, reticulin fibrosis and clusters of CD34⁺ cells (examples circled in red). All images original magnification x400. (B) Sequencing of MPL exon 10 in T cells, granulocytes from the MPN phase of disease (MPN Grans) and purified leukemic blasts (AML blasts), showing W515L-heterozygous AML in patient 1 and W515L-homozygous AML in patients 2 and 3. (C) Real-time PCR MPL copy number assay on three W515L-homozygous patient samples (leukemic blasts from patients 2 and 3; granulocytes from patient 4) with loss of heterozygosity at telomeric end of the MPL locus (rs498166 or rs499163) and close to the 1p telomere (rs7537577 or rs1870509): the presence of two copies of MPL in all 3 cases establishes mitotic recombination as the mechanism by which the wild-type MPL allele is lost. Analysis of 10 normal individuals and 3 cell lines (NB4, HT3 and LAN1) known to harbor a single copy of the MPL locus19 are shown as controls.

To distinguish between these possibilities, we studied W515L-homozygous leukemic blasts from patients 2 and 3 together with granulocytes from a W515L-positive JAK2 V617F-negative PMF patient (patient 4) with a mutant allele proportion of over 0.9. In all 3 patients, informative SNPs (genotyped by direct sequencing) at both the telomeric end of the MPL locus and 39Mb distal (close to the 1p telomere) showed loss of heterozygosity in leukemic blasts (patients 2 and 3) or granulocytes (patient 4) (data not shown), excluding acquisition of a second MPL mutation. To distinguish deletion of the wild-type allele from mitotic recombination, MPL copy number was assessed by real-time PCR, which demonstrated two copies of MPL in all cases (Figure 1C). These findings demonstrate that homozygosity for an acquired MPL W515L mutation had arisen by mitotic recombination in these 3 patients, confirming previous studies in which acquired uniparental disomy affecting the MPL locus had been detected by SNP array technology.10–12 Together these findings mirror the situation with other signaling pathway mutations, such as JAK2 V617F and FLT3-ITD, where mitotic recombination results in duplication of the mutant allele, implying a selective advantage is conferred by either increased mutant gene dosage or loss of the wild-type allele.

Leukemic blasts were screened for mutations in N/KRAS, CEBPA, RUNX1, GATA2, NPM, WT1, TP53, CBL, IDH1/2 and TET2. Mutations were identified in TP53 and TET2 in patient 1 but no additional lesions were found in patients 2 and 3. In patient 1, the TET2 mutation was predominant in bone marrow cells obtained three years prior to leukemic transformation, whereas the MPL mutation was present at a relatively low level (Figure 2A). Sequencing is not highly quantitative, but the magnitude of the observed difference suggests that the TET2 mutation preceded acquisition of the MPL mutation. Erythroid and granulocyte-macrophage colonies (n=41 and n=21, respectively), confirmed by cytological analysis, all harbored both MPL and TET2 mutations, demonstrating that the mutations arose within the same clone. In addition, 13 of 62 colonies harbored mutations in TP53. Remarkably, a total of four different TP53 mutations were identified, all of which are recurrent, functionally-significant cancer-associated alleles.20 Progression to acute leukemia was associated with loss of wild-type TP53 in one subclone (Figure 2A).

Table 2.Clinical and laboratory features of MPL W515L-positive patients at presentation and at time of progression to acute myeloid leukemia.

Figure 2.Leukemic progression associated with the proliferation of divergent, TP53-mutant clones. (A) Analysis of sequential samples from patient 1, demonstrating acquisition of a TET2 mutation prior to a mutation in MPL, the proliferation of erythroid and granulocyte-macrophage colonies harboring different heterozygous mutations in TP53, and loss of wild-type TP53 in leukemic blasts. (B) Model of disease progression in patient 1, characterized by the parallel expansion of multiple genetically distinct but phylogenetically-related clones bearing heterozygous TP53 mutations, with loss of wild-type TP53 in one of these sub-clones associated with progression to acute leukemia. BM: bone marrow cells, MPN: myeloproliferative neoplasm, AML: acute myeloid leukemia.

Detection of TP53 mutations within erythroid and granulocyte-macrophage colonies indicates that terminal differentiation may proceed in the presence of mutant p53 where the wild-type allele is retained. Furthermore, the presence of multiple p53-mutant clones, all involving C:G-to-T:A transitions (Figure 2A), implies a mutator-phenotype prior to the development of an AML-associated differentiation block. One possible mechanism invokes a mutagenic effect of mutant MPL, TET2 or other unidentified genes, as reported in models of oncogenic ERBB2 and BCR-ABL1 which resulted in a bias towards transversion or transition mutations, respectively.21,22 In this patient, no other acquired synonymous or non-synonymous mutations were identified in 11.5Kb of DNA sequence from leukemic blasts, although the mutation prevalence in solid tumors (one mutation every 10 – 10 bases) suggests that a genome-wide approach would be necessary to elucidate the true mutation frequency. In addition, it is possible that alterations of MPL, TET2 or unidentified gene(s) within the parental clone impart a strong selective pressure for the acquisition of TP53 mutations. Alternatively, diverse clones may arise following exposure to an exogenous agent. Hydroxycarbamide (received by this patient) has been linked to abnormalities of 17p (which harbors the TP53 locus),23,24 although a specific mutation signature has not been reported.

Taken together, our data demonstrate the parallel expansion of genetically distinct but phylogenetically-related clones prior to leukemic transformation (Figure 2B). Of note, clonal diversity (assessed by loss of heterozygosity and TP53/CDKN2A mutation) in patients with Barrett’s esophagus has been associated with an increased risk of progression to adenocarcinoma,25 suggesting that in this disease expansion of competing clones may also presage progression to a fully malignant phenotype.

In conclusion, this study used paired MPN/AML samples to demonstrate that progression to AML is part of the natural history of MPL W515L-associated disease, may occur in the absence of prior cytoreductive therapy and may involve loss of the wild-type MPL allele by mitotic recombination. Moreover, studies of progenitor colonies revealed the expansion of divergent but phylogenetically-related clones during progression from MPL-mutant MPN to AML.

Acknowledgments

we would like to thank Drs Penny Wright and Wendy Erber for helpful comments on bone marrow histology and the Addenbrooke’s Haematology Disorders Sample Bank for processing and managing patient samples.

Footnotes

• Funding: work in the authors’ laboratories is supported by the UK Medical Research Council, Leukaemia and Lymphoma Research, the Kay Kendal Leukaemia Fund, the NIHR Cambridge Biomedical Research Centre, the Leukemia and Lymphoma Society of America, and the 2008 MIUR Project funded to AMV.
• Authorship and Disclosures The information provided by the authors about contributions from persons listed as authors and in acknowledgments is available with the full text of this paper at www.haematologica.org.
• Financial and other disclosures provided by the authors using the ICMJE (www.icmje.org) Uniform Format for Disclosure of Competing Interests are also available at www.haematologica.org.

• Received June 21, 2010.
• Revision received July 27, 2010.
• Accepted August 24, 2010.