Truncation mutations of the receptor cytoplasmic domain for colony-stimulating factor 3 (CSF3R) are frequently seen in severe congenital neutropenia, whereas activating missense mutations affecting the extracellular domain (exon 14) have been described in hereditary neutrophilia and chronic neutrophilic leukemia (CNL). In order to clarify mutational frequency, specificity and phenotypic associations, we sequenced CSF3R exons 14–17 in 54 clinically suspected cases of CNL (n=35) or atypical chronic myeloid leukemia (aCML; n=19). Central review of these cases confirmed WHO-defined CNL in 12 patients, monoclonal gammopathy (MG)-associated CNL in 5 and WHO-defined aCML in 9. A total of 14 CSF3R mutations were detected in 13 patients, including 10 with CSF3RT618I (exon 14 mutation, sometimes annotated as CSF3R T595I). CSF3RT618I occurred exclusively in WHO-defined CNL with a mutational frequency of 83% (10 of 12 cases). CSF3R mutations were not seen in aCML or MG-associated CNL. CSF3RT618I was also absent among 170 patients with primary myelofibrosis (PMF; n=76) or chronic myelomonocytic leukemia (CMML; n=94). SETBP1 mutational frequencies in WHO-defined CNL, aCML, CMML and PMF were 33, 0, 7 and 3%, respectively. Four CSF3RT618I-mutated cases co-expressed SETBP1 mutations. We conclude that CSF3RT618I is a highly sensitive and specific molecular marker for CNL and should be incorporated into current diagnostic criteria.
Colony-stimulating factor 3 receptor gene (CSF3R), mapping to chromosome 1p34.3, encodes the trans-membrane receptor for granulocyte colony-stimulating factor (G-CSF; CSF3), which provides the proliferative and survival signal for granulocytes and also contributes to their differentiation and function.1 CSF3R harbors 17 exons and its protein 813 amino acids. The cytoplasmic domain of CSF3R is functionally assigned to proliferation (proximal region) and differentiation/regulation of proliferation (distal region).2
Nonsense somatic mutations affecting the cytoplasmic domain of CSF3R and leading to carboxyl-truncation have been described in ∼40% of patients with severe congenital neutropenia, where they are acquired and occur in conjunction with other inherited mutations (for example, ELANE and HAX1).3 Such mutations appear to be stem cell-derived,3 associated with but not essential for severe congenital neutropenia -associated acute myeloid leukemia,4, 5 promote STAT5-mediated clonal advantage in mouse progenitor cells6 and co-operate with other oncogenes to induce acute myeloid leukemia.5, 7 Severe congenital neutropenia -associated CSF3R mutations occasionally affect the extracellular domain of the receptor.4, 8
A germline CSF3R mutation (C-to-A substitution at nucleotide 2088; T617N) was recently described in autosomal dominant hereditary neutrophilia associated with splenomegaly and increased circulating CD34-positive myeloid progenitors;9 functional studies suggested enhanced receptor dimerization and signaling that was abrogated by JAK2 inhibition and induction of neutrophilia and splenomegaly in mice.9 Similar but acquired extracellular domain mutations are infrequently reported in acute myeloid leukemia.8, 10 Most recently, Maxson et al.11 made the seminal observation regarding the association between CSF3R mutations and chronic neutrophilic leukemia (CNL). The current study was undertaken to determine the frequency, location and specificity of CSF3R mutations in CNL and the closely related atypical chronic myeloid leukemia (aCML).
Patients and methods
Patients and samples
The current study was approved by the Mayo Clinic institutional review board. Patients were primarily identified through search of hematopathology databases for a diagnosis of ‘CNL’ or ‘aCML’. Inclusion to the current study required availability of archived bone marrow or peripheral blood granulocytes for DNA extraction, as well as bone marrow morphology and cytogenetic information at the time of first referral to the Mayo Clinic. The diagnoses of CNL, aCML, chronic myelomonocytic leukemia (CMML) and primary myelofibrosis (PMF) were confirmed by World Health Organization (WHO) criteria.12 CMML and PMF patients were selected from databases of patients previously annotated for other mutations.13, 14 Patient information was updated through review of patient histories and correspondence, social security death index or a telephone call to the patient or their local physician.
For CSF3R mutation analysis, exons 14–17 were amplified for all clinically suspected cases of CNL or aCML using standard PCR conditions. Primers for CSF3R were as follows: 14 forward: 5′-CCACGGAGGCAGCTTTAC-3′, 14 reverse: 5′-AAATCAGCATCCTTTGGGTG-3′; 15 forward: 5′-TGACTTTGAATCCCCTGGTC-3′, 15 reverse: 5′-TGAGGTTCCCTGTGGGTG-3′; 16 forward: 5′-AAAATGGAAAGATCGGAGGG-3′, 16 reverse: 5′-CTTGGCTTCAGAAGGTGTCC-3′, and 17 forward: 5′-CTGTCACTTCCGGCAACAT-3′, 17 reverse: 5′-TGGCCCAAAGACACAGTCGT-3′. Following amplification, the PCR products were sequenced via standard capillary electrophoresis by Applied Biosystems 3730 series DNA Analyzers (Carlsbad, CA, USA), and results were analyzed using Sequencher software (Gene Codes Inc., Ann Arbor, MI, USA). Patients with CMML and PMF were screened for exon 14 mutations only.
PCR and Sanger sequencing was used for SETBP1 mutation screening in PMF, CNL and aCML patients (forward primer 5′-ATGCACCCACTTTCAACACA-3′ and reverse primer 5′-AAAAGGCACCTTTGTCATGG-3′ to generate sequence for the amino-acid region 825–1013). For the CMML cohort, we used the ViiA7 quantitative RT-PCR platform (qPCR) and MeltDoctor high-resolution melting assay (Life Technologies, Grand Island, NY, USA) using forward primer 5′-GCGAGATTGGCTCCCTAAAG-3′ and reverse primer 5′-CCAGGGAGCAGAAATCAAAA-3′ to generate sequence for the amino-acid region 860–1000. Targeted cases were validated using Sanger sequencing to confirm the presence of a mutation.
All statistical analyses considered clinical and laboratory parameters obtained at the time of first referral, which coincided in most instances with the time of bone marrow /granulocyte collection. Differences in the distribution of continuous variables between categories were analyzed by either Mann–Whitney (for comparison of two groups) or Kruskal–Wallis (comparison of three or more groups) test. Patient groups with nominal variables were compared by χ2 test. Overall survival was calculated from the date of first referral to the date of death (uncensored) or last contact (censored). Survival curves were prepared by the Kaplan–Meier method and compared by the log-rank test. The Stat View (SAS Institute, Cary, NC, USA) statistical package was used for all calculations.
CSF3R mutation screening included exons 14 through 17 for all 54 clinically suspected cases of ‘CNL’ (n=35) or ‘aCML (n=19). An additional 170 patients with CMML (n=94) or PMF (n=76) were screened for mutations involving CSF3R exon 14 only. In addition, SETBP1 mutation screening was performed in all the cases. All study patients underwent bone marrow biopsy with cytogenetic assessment and presence of BCR–ABL1 was excluded in every case of suspected CNL by FISH and/or PCR analysis.
Confirmation of diagnosis according to WHO criteria
Central review of clinicopathological data for the 54 clinically suspected cases of CNL or aCML, in order to identify those who strictly met WHO criteria for such diagnoses, was performed before the results of mutation analysis became available. Accordingly, five broad patient groups could be defined: (i) CNL-suspected cases meeting WHO criteria for CNL and without concomitant monoclonal gammopathy (MG) or lymphoid neoplasm (WHO-defined CNL; n=12); (ii) CNL-suspected cases meeting WHO criteria for CNL but with associated MG or lymphoid neoplasm (MG-associated CNL; n=6); (iii) CNL-suspected cases not meeting WHO criteria for CNL (n=17); (iv) aCML-suspected cases meeting WHO criteria for aCML (n=9); and (v) aCML-suspected cases not meeting WHO criteria for aCML (n=10).
Among the six patients with MG-associated CNL, five had monoclonal gammopathy with uncertain significance (MGUS) and one a low-grade lymphoproliferative neoplasm. One patient each with MGUS subsequently developed overt multiple myeloma and smoldering multiple myeloma. Of the 17 patients with unconfirmed CNL, 2 had MGUS (12%) and 1 a lymphoproliferative neoplasm (6%). The reasons for the 17 cases of ‘CNL’ not meeting WHO criteria included the presence of concurrent infection/inflammatory condition (n=4), circulating immature cells >10% (n=6), circulating blasts 1% (n=2), leukocyte count <25 × 109/l (n=2), dysplastic changes (n=1), or alternative hematological diagnosis (CMML or polycythemia vera; n=2). Relevant clinical data for the 35 clinically suspected cases of CNL are summarized in Table 1.
Among the 19 patients with clinically suspected aCML, central review disclosed 9 cases of WHO-defined aCML. The other 10 were confirmed to have BCR–ABL1-positive CML (n=2), CMML (n=3), myelodysplastic syndrome/myeloproliferative neoplasm-unclassified (MDS/MPN-U; n=2) PMF (n=1), systemic mastocytosis associated with MDS/MPN-U (n=1), and MPN-U (n=1). The 170 cases of CMML or PMF were all confirmed to meet WHO diagnostic criteria.
CSF3R and SETBP1 mutation results
A total of 14 CSF3R mutations were identified in 13 patients, all of whom belonged to the group with either WHO-defined CNL (n=12) or unconfirmed CNL (n=1). The overall CSF3R mutational frequency was 100% in WHO-defined CNL vs 0% in MG-associated CNL vs 6% in CNL-suspected cases not meeting WHO criteria for CNL vs 0% in WHO-defined aCML vs 0% in aCML-suspected cases not meeting WHO criteria for aCML (P<0.0001). CSF3RT618I was the most frequent CSF3R mutation, occurring in 10 (all with WHO-defined CNL) of the 13 patients with CSF3R mutations; the remaining 3 patients harbored CSF3RI598I (WHO-defined CNL) or CSF3RM696T (one case each with WHO-defined CNL and unconfirmed CNL).
CSF3RT618I occurred exclusively in WHO-defined CNL with mutational frequency of 83% and was absent in WHO-defined aCML, MG-associated CNL or all other cases of unconfirmed CNL or aCML. Furthermore, similar exon 14 mutations were absent in another 170 patients with CMML or PMF. One WHO-defined CNL patient with CSF3RT618I (extracellular domain mutation) also harbored another CSF3R exon 17 mutation (2341_2342insC; cytoplasmic domain), which introduces stop codon 7 residues from the insertion site.
Among all 34 cases with clinically suspected CNL, six harbored SETBP1 mutations (G870D/S=4; G872R=1; and D868N=1); 4 of the 6 patients, all with WHO-defined CNL, also harbored CSF3RT618I mutations. Overall, SETBP1 mutational frequencies in WHO-defined CNL, aCML, CMML and PMF were 33, 0, 7 and 3%, respectively. JAK2V617F screening among all clinically suspected CNL patients disclosed only three positive cases (one had MG-associated CNL and two had unconfirmed CNL). Among the 10 patients with unconfirmed aCML, one patient whose diagnosis was revised to CMML displayed SETBP1 mutation (D868N) and two with revised diagnoses of CMML or PMF harbored JAK2V617F.
We found no significant correlation between the three CNL groups (WHO-defined vs MG-associated vs unconfirmed) and either age, gender or leukocyte count (P>0.05). Similarly, there was no significant correlation between presence of CSF3R mutation and either age, gender or leukocyte count (P>0.05). After a median follow-up of 18 months (range 1–146), 23 deaths (66%) were recorded in the overall cohort of clinically suspected CNL. The median survival of patients with MG-associated CNL was significantly (P<0.05) longer than the other two CNL subgroups (Figure 1). Considering all 35 cases of clinically suspected CNL, the presence or absence of CSF3R mutation did not affect survival (Figure 2; P=0.83), whereas there was a trend for shortened survival among SETBP1-mutated patients (Figure 3; P=0.1).
The current study confirms the seminal observation by Maxson et al.11 regarding the association between CNL and CSF3R mutations and clarifies a number of clinically relevant issues including mutational frequency, location and specificity in strictly WHO-defined CNL. We show an apparently invariable association between these mutations and WHO-defined CNL and a high degree of specificity. The most frequent CSF3R mutation we encountered was CSF3RT618I (also annotated as CSF3R T595I); we did not have access to germline DNA to confirm the somatic nature of the other genetic variations including CSF3RM696T and I598I. These observations warrant the inclusion of CSF3R mutation analysis in the clinical diagnosis of CNL and formal incorporation into WHO diagnostic criteria. Additional screening for SETBP1 mutations might be helpful considering the possibility that co-expression of CSF3R and SETBP1 mutations might be detrimental to survival, although this particular observation requires validation in a larger cohort of WHO-defined CNL.
Our observation regarding the absence of CSF3R mutations in patients with MG-associated CNL and their apparently better survival suggests a different etiopathogenesis for the particular entity, including the possibility of a causal relationship between the MG and neutrophilia. However, one of these patients harbored JAK2V617F, suggesting an underlying myeloid malignancy in at least some of the patients with MG-associated CNL. The other major point from our study was the critical need for careful morphologic assessment in making genotype–phenotype associations.15 For example, we did not detect either CSF3R or SETBP1 mutations in strictly WHO-defined aCML, whereas others have reported 24% mutational frequency for SETBP1 mutations in ‘aCML’.16 Instead, in our patient cohort, SETBP1 mutations were most frequent in WHO-defined CNL (33%). Indeed, incorporation of molecular characteristics such as CSF3R and SETBP1 mutations into WHO diagnostic criteria may help further clarify distinction between these disorders.
In contrast to severe congenital neutropenia, the overwhelming majority of our cases with CSF3R-mutated CNL were represented by a missense exon 14 mutation (CSF3RT618I) that is akin to the aforementioned germline CSF3RT617N mutation associated with hereditary neutrophilia and shown to activate CSF3R signaling by possible enhancement of receptor dimerization.9 We suspect the same mechanism might be at play in CSF3RT618I-mutated CNL and thus amenable to treatment with JAK inhibitors.17, 18
About this article
A Pardanani and A Tefferi designed the study, contributed patient samples, analyzed the data, and wrote the paper. J Tyner contributed to data analysis and writing of the paper. MA Elliott contributed patient samples. CA Hanson reviewed bone marrow histology. RA Knudson and RP Ketterling reviewed the cytogenetics data. TL Lasho and R Laborde performed the molecular studies.