The myeloproliferative neoplasms, PV, ET and PMF, share phenotypic features and molecular lesions, yet PMF distinguishes itself by its unfavorable natural history and rate of leukemic evolution. These distinctions may occur as a result of cooperating genomic lesions specific to PMF compared to PV or ET. We performed single nucleotide polymorphism array (SNP-A)-based karyotyping in 210 MPN patients and identified 20q11 deletions in 10% of PMF cases and in none of the PV or ET cases. The 20q11 deletion region spanned 1,662 KB and encompassed 37 genes, of which ASXL1 was included. To test whether ASXL1 contained lesions in the MPN cohort at large, we directly sequenced key regions of the ASXL1 gene in 65 PMF, 11 PV and 14 ET cases, as well as 7 controls from the SNP-array cohort. Genomic DNA from neutrophils and in select cases, purified CD34+ cells was used for both SNP-A and direct sequencing. Clinical parameters were correlated with genomic findings and the quantitative JAK2 V617F neutrophil allele burden
26/65 (40%) of PMF cases had abnormalities in ASXL1 (4 deletions, 22 mutations) whereas none of the 32 PV, ET or control cases had such lesions. The majority of ASXL1 sequence variations were nonsense lesions including the previously reported 1934dupG which comprised 30% of all of the mutations. The residual ASXL1 allele in all 20q11 deletion cases containing the ASXL1 gene was intact. In three PMF cases, more than one distinct ASXL1 mutation was identified, and cloning experiments on two of those cases indicated that the lesions were biallelic. Using banked samples, we observed the acquisition of an ASXL1 lesion over time, and established that ASXL1 lesions detected in 2 post ET-MF cases were also detected at low levels in the ET phase of the MPN.
ASXL1 deletions and mutations were prevalent in de novo PMF (37%), post PV-PMF (20%) post ET-PMF (62%) and in PMF/AML (33%). ASXL1 mutations did not associate with chemotherapy exposure as the prevalence of hydroxyurea use was similar in patients with and without mutations, and ASXL1 –mutation positive cases were present in patients who had never received any form of chemotherapy. There was no dependence upon JAK2 status as the presence of ASXL1 mutations were identified in JAK2 V617F-negative cases (9/26); JAK2 V617F-heterozygous cases (10/26); and JAK2 V617F-homozygous cases (7/26). Based on results of SNP-A, patients with ASXL1 mutations were equally as likely to have uniparental disomy (involving 9p or other regions) and loss/gain abnormalities (>1MB) compared to those without ASXL1 mutations. There were no differences in sex, age, or disease duration between PMF patients with and without ASXL1 mutations. In the ASXL1-mutant group, there was a trend toward a lower median white blood cell count (8 vs. 12.5 k/cu mm; p=0.3) and hemoglobin (9.7 vs. 11 g/dl; p=0.3) compared to ASXL1-wild-type patients. Furthermore, those PMF patients with ASXL1 mutations were significantly more likely to have received anemia-directed therapy (transfusion, erythropoietin, immunomodulating agents, steroids) compared to those without mutations (15/26 (58%) vs. 11/39 (23%); p=0.02). Post ET-MF patients comprised 31% (8/26) of ASXL1-mutant cases, compared to only 10% (4/39) ASXL1- wild-type cases (p=0.03). However, the presence of an ASXL1 mutation did not associate with an accelerated transition rate from ET to MF; among the 12 post ET-MF cases in the cohort, the median time of transition from ET to MF was 15.5 years in those with ASXL1 mutations compared to 7 years in those with ASXL1 wild-type status (p=0.02).
Disruption of the ASXL1 gene occurs in 40% of PMF cases. The association of ASXL1 lesions, due to either mutation or deletion, suggests that ASXL1 haplo-insufficiency is associated with a PMF phenotype in the context of other known and unknown lesions, and that disruption of ASXL1 function may directly contribute to the pathophysiology and clinical complications of primary and secondary myelofibrosis. These data support the concepts that cooperative lesions in addition to JAK2 V617F are critical in generating PMF, that PMF is molecularly more complex than either PV or ET, and that the transition of PV or ET to PMF is associated with the acquisition of genomic lesions, such as ASXL1, that are present in PMF at large.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.