The JAK2V617F tyrosine kinase mutation in ... - Wiley Online Library

research paper

The JAK2V617F tyrosine kinase mutation in myelofibrosis with myeloid metaplasia: lineage specificity and clinical correlates

Ayalew Tefferi,1 Terra L. Lasho,1 Susan M. Schwager,1 David P. Steensma,1 Ruben A. Mesa,1 Chin-Yang Li,1 Martha Wadleigh2 and D. Gary Gilliland2 1

Mayo Clinic and Mayo Clinic College of Medicine, Rochester, MN, and 2Department of

Medical Oncology, Dana Farber Cancer Institute, 1 Harvard Medical School, Boston, MA, USA

Received 12 July 2005; accepted for publication 24 August 2005

Summary An association between an activating JAK2 mutation (JAK2V617F) and BCR/ ABL-negative myeloproliferative disorders was recently reported in multiple simultaneous publications. In the current study, mutation analysis for JAK2V617F was performed in peripheral blood mononuclear cells (PBMC) from 157 patients with myelofibrosis with myeloid metaplasia (MMM) including 117 with agnogenic (AMM), 22 with postpolycythaemic (PPMM), and 18 with post-thrombocythaemic (PTMM) myeloid metaplasia. The detection rate for JAK2V617F was significantly higher in PPMM (91%; homozygous in 18%) compared with either AMM (45Æ3%; homozygous in 2Æ6%) or PTMM (38Æ9%; homozygous in 11Æ1%). Concomitant analysis in granulocytes (n ¼ 57) and CD34+ cells (n ¼ 25) disclosed a higher incidence of homozygous JAK2V617F mutation but the overall mutation rate was similar to that obtained from PBMC. JAK2V617F was not detected in DNA derived from T cells (n ¼ 19). In AMM, the presence of JAK2V617F was associated with an older age at diagnosis and a history of thrombosis or pruritus. Multivariate analysis identified only age and the Dupriez prognostic score as independent prognostic factors; JAK2V617F had no prognostic significance. In conclusion, JAK2V617F is a myeloid lineage-specific event, its incidence in MMM is significantly higher with an antecedent history of polycythaemia vera (PV), and its presence in AMM does not affect prognosis but is associated with PV-characteristic clinical features.

Correspondence: Ayalew Tefferi, MD, Mayo Clinic, 200 First Street SW, Rochester MN 55905, USA. E-mail: [email protected]

Keywords: Myelofibrosis, JAK2, mutation, prognosis, clonality, polycythae2 mia vera.

Myelofibrosis with myeloid metaplasia (MMM) presents either de novo (agnogenic myeloid metaplasia; AMM) or in the setting of both polycythaemia vera (postpolycythaemic myeloid metaplasia; PPMM) and essential thrombocythaemia (post-thrombocythaemic myeloid metaplasia; PTMM) (Cervantes et al, 2002; Passamonti et al, 2003). These three MMM variants display a similar clinical phenotype that includes progressive anaemia, marked splenomegaly, a leucoerythroblastic blood smear, bone marrow fibrosis and osteosclerosis, and extramedullary haematopoiesis (Mesa et al, 2000; Tefferi, 2000). Median survival in MMM ranges from 3Æ5 to 10 years (Rupoli et al, 1994; Dupriez et al, 1996; Cervantes et al, 1997; Okamura et al, 2001) and is affected by a defined set of prognostic factors including haemoglobin level, constitutional symptoms, circulating blasts, leucocyte count, and cytogenetics (Rupoli et al, 1994; Cervantes et al, 1997; Okamura et al, doi:10.1111/j.1365-2141.2005.05776.x

2001). Current drug therapy has not altered the natural history of MMM in terms of either survival or risk of leukaemic transformation, which occurs in 8–23% of the patients (Tefferi, 2003). Allogeneic haematopoietic stem cell transplantation has resulted in durable remission in a select group of patients (Deeg et al, 2003). However, the majority of affected patients are not suitable candidates for this particular treatment modality. MMM is a clonal stem cell disease with myeloid as well as lymphoid lineage involvement, as are polycythaemia vera (PV) and essential thrombocythaemia (ET) (Reeder et al, 2003). However, the primary oncogenic event(s) in all three disorders have been elusive. Most recently, five independent studies have described an association between an activating somatic mutation in the gene encoding the cytoplasmic JAK2 tyrosine kinase (JAK2V617F) and BCR/ABL-negative myeloproliferative disease

ª 2005 Blackwell Publishing Ltd, British Journal of Haematology, 131, 320–328

Lineage Specificity, Clinical Correlates of JAK2V617F in MMM (MPD) (Baxter et al, 2005; James et al, 2005; Kralovics et al, 2005; Levine et al, 2005; Zhao et al, 2005). The number of study patients with PV, ET, and MMM among the first four studies [the fifth study (Zhao et al, 2005) did not involve patients with MMM] ranged from 45 to 164, 21 to 115, and 7 to 46, respectively. The respective detection rates for JAK2V617F, in granulocyte cell fractions, was highly variable among these studies and ranged from 65% to 97%, 32% to 57%, and 43% to 50%, with the highest figures applying to PV and the lowest to ET in each instance. Clinical correlative studies were hampered by the small sample size in MMM but revealed significant associations between the presence of a mutant allele and female gender in PV (Levine et al, 2005), a longer disease duration in PV/ET patients with homozygous JAK2 mutation (Kralovics et al, 2005; Levine et al, 2005) and, in one study, a higher incidence of myelofibrosis, haemorrhage, and thrombosis in patients with the mutation as opposed to those without (Kralovics et al, 2005). Thus, there is a need for analysis of larger number of patients with MMM to evaluate mutational frequency and prognostic relevance. In this report, a larger patient population with MMM (n ¼ 157) evaluated at a single centre, were genotyped for the presence the mutant JAK2 allele. The relatively large sample size made it possible to perform valid clinical as well as laboratory correlative studies. It also enabled a more accurate estimate of mutant allele prevalence not only in MMM but also in its subcategories of AMM, PPMM, and PTMM. Finally, the current study also provides new and comparative information, in MMM, on the occurrence (or lack of) of JAK2V617F in circulating CD34+ cells as well as in PBMC and T cells.

(Yoon et al, 1999). Bone marrow histology in all cases was reviewed by Mayo Clinic haematopathologists and re-reviewed by one of the authors. Diagnoses were assigned according to standard World Health Organisation (WHO) criteria (Vardiman et al, 2001) and categorisation into AMM, PPMM, and PTMM required complete bone marrow and historical documentation that was confirmed by the senior author and two of the other authors. The circulating CD34+ cells were quantified using the direct immunofluorescence flow cytometry according to previously described methods (Siena et al, 1991). For cytogenetic studies, both direct technique and un-stimulated 24-h culture methods were used to harvest 20 metaphases, whenever possible, in all bone marrow specimens (Dewald et al, 1985).

Study sample preparation Double density gradient centrifugation (Histopaque-1077TM layered over Histopaque-1119TM; Sigma Diagnostics, St Louis, MO, USA) was used to separate out the granulocyte and peripheral blood mononuclear cell layers (PBMC) from each sample. The mononuclear cell layer was further fractionated by magnetic-activated cell sorting (MACS; Miltenyi Biotech, Auburn, CA, USA) using antibodies that were specific to myeloid progenitor cells (CD34+) as well as T (CD3+) lymphocytes (Reeder et al, 2003; Tefferi et al, 2004). More than 98% cell purity of the different cell fractions was confirmed by flow cytometry using the FACScalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA).

JAK2 mutation analysis Methods Patients The current study was approved by the Institutional Review Board (IRB) of the Mayo Clinic, and Minnesota and The Health Insurance Portability and Accountability Act (HIPAA) guidelines regarding access to medical records were followed. The study population included a consecutive cohort of patients in whom peripheral blood samples were available for extraction of genomic DNA: study samples were donated by patients to the Mayo Clinic myeloproliferative disease cell bank under an IRB-approved protocol. The particular protocol allowed for the steady procurement of blood mononuclear cells (MNC) at all times whereas neutrophil preparations required certain times of the day for blood draw as well as onsite availability of a specifically trained technician. For this reason, as well as cell utilisation by other competing neutrophil-based projects, mutation screening in neutrophils was restricted to a fraction of the study patients. Bone marrow biopsy slides prepared from paraffin-embedded blocks were stained with haematoxylin and eosin and each marrow was examined and subsequently graded for reticulin fibrosis (grades 1–4) according to previously published criteria

Genomic DNA was extracted using QIAamp Blood Mini Kit (Qiagen, Valencia, CA, USA). Genomic DNA was amplified by polymerase chain reaction (PCR), and successful amplification was confirmed by electrophoresis on an ethidium bromideimpregnated 1Æ5% agarose gel. Each 50 ll PCR reaction contained approximately 25 ng of DNA template, 5 ll 10X Roche Buffer (final concentration of MgCl2: 1Æ5 mmol/l), 3 1Æ5 U Taq polymerase (Roche, Indianapolis, IN, USA), 0Æ8 mmol/l dNTPs (Roche), and 20 pmol/l each of sense and antisense primers (5¢-TGCTGAAAGTAGGAGAAAGTGCAT3¢ and 5¢-TCCTACAGTGTTTTCAGTTTCAA-3¢, respectively). PCR cycling parameters were: one cycle of 94C for 2 min; 35 cycles of 94C for 30 s, 52C for 40 s and 72C for 40 s; followed by one cycle of 72C for 2 min. PCR products were cleaned with QIAquick PCR purification Kit (Qiagen, Valencia, CA, USA). Fluorescent dye chemistry sequencing was performed using the same primers used for amplification, on an ABI PRISM 3700 DNA Analyzer (Applied Biosystems, Foster City, CA, USA). Sequencher 4Æ2 (Gene Codes Corporation, Ann Arbor, MI, USA) and GenBank accession number NM_004972 (JAK2 mRNA) and the corresponding region from the NC_000009 chromosome 9 contig were used for sequence analysis.


321

A. Tefferi et al

Statistical analysis Descriptive and statistically analysed data were obtained from the entire cohort of patients (n ¼ 157) both at diagnosis and at the time of study entry (i.e. time of mutation screening for JAK2V617F). All study patients were considered to estimate the prevalence of both heterozygous and homozygous JAK2V617F. However, only AMM patients (n ¼ 117) were considered for further statistical analysis to explore clinical and laboratory associations with the mutant allele. In this regard, because of the small number of AMM patients with homozygous abnormalities, patients with either heterozygous or homozygous defects were combined into one group in order to be compared with patients with wildtype allele. Comparison between categorical variables was performed by chi-squared statistics. Comparison between categorical and continuous variables was performed by either the Mann– Whitney U-test or Kruskal–Wallis test. Survival was calculated by Kaplan-Meier plots taking the interval from the date of diagnosis to death or last contact and checked for significance by both the Logrank and Breslow–Gehan test. For all 41 patients who that had died at the time of writing, the ‘date of last contact’ was the date of death. In patients that were alive, every attempt was made to update follow-up information, by means of a questionnaire/telephone call sent to both patients and their primary doctors, and the ‘date of last contact’ reflected this time point and not the last time they were seen at the Mayo Clinic. A Cox proportional hazards regression analysis was used to assess the prognostic relevance, to survival, of key prognostic variables as well as the presence of JAK2V617F. In order to avoid the problem with multiple tests of significance, our multivariate model included only those variables that are currently considered to have independent prognostic value in MMM (‡1% circulating blasts, haemoglobin level of < 10 g/dl, leucocyte count of either > 30 or < 4 · 109/l) (Dupriez et al, 1996; Cervantes et al, 1998). All data were analysed by using SAS software (SAS, Inc., Cary, NC, USA) and statistical significance was set at the level of P ¼ 0Æ05.

Results Patient characteristics at diagnosis and at time of mutation screening for JAK2V617F The study population was 157 consecutive patients with MMM evaluated at the Mayo Clinic. Clinical and laboratory characteristics at diagnosis are presented in Table I. Four additional patients were excluded from the study because of inadequate information regarding the history of antecedent PV or ET. Among the entire cohort of 157 patients (median age at diagnosis 58Æ9 years; range 18Æ4–80Æ8 years), 55 were females and 102 males. The fact that Mayo Clinic serves primarily as a tertiary referral centre for these diseases might partly account 322

for the preponderance of younger patients and males in the current study. Because of a previously reported increased prevalence of MPD in persons of Jewish ancestry (Najean et al, 1998), we considered the possibility of ethnic differences in mutational frequency and therefore collected information on ethnic origin in 119 patients in whom such information was available: 14 patients stated they were of Jewish ancestry and 105 did not. Table I also provides clinical and laboratory details for each one of the three MMM variants; AMM (n ¼ 117), PPMM (n ¼ 22), and PTMM (n ¼ 18). Table II provides additional laboratory information obtained at time of JAK2V617F mutation screening. The median time from initial diagnosis to mutation analysis was 11Æ8 months (range 0–317 months).

Clinical course The median duration of follow-up from initial diagnosis was 37Æ2 months (range 0–357Æ8 months) and from the time of JAK2V617F mutation screening was 12Æ3 months (0– 58Æ5 months). During this time, 41 patients have died and documented causes of death included non-leukaemic progression of disease (n ¼ 4), infection (n ¼ 4), acute myeloid leukaemia (n ¼ 4), cardiac failure (n ¼ 3), pulmonary embolus (n ¼ 1), renal failure (n ¼ 1), and endometrial carcinoma (n ¼ 1). The exact cause of death was not well documented in the remaining cases. Among all 157 patients, a history of moderate or severe bleeding, thrombosis, and pruritus were documented in five (all five in AMM), nine (five in AMM), 12 (seven in AMM) patients, respectively. No thrombo-haemorrhagic complications occurred in patients with PTMM. Changes in disease parameters from the time of diagnosis are presented in Table II. Only four patients underwent therapeutic splenectomy during the study period. A total of 10 leukaemic transformations were detected; eight of them in patients with AMM. Median survival was nearly identical in MMM (n ¼ 157) and AMM (n ¼ 117) at 11Æ5 years.

Prevalence and lineage specificity of JAK2V617F Mutation screening was performed on genomic DNA of PBMC from all 157 patients with MMM and JAK2V617F was detected in 80 (51%) patients (heterozygous in 45% and homozygous in 5Æ7%). There was a significant difference in mutation detection rates among AMM, PPMM, and PTMM (Table III; P ¼ 0Æ0002). PPMM displayed the highest prevalence of JAK2V617F with an overall detection rate of 91% (homozygous in 18%) whereas the incidence figures were not statistically different between AMM and PTMM (Table III). Table III also displays prevalence of the specific mutation in different myeloid cell fractions as well as T lymphocytes. Consistent with previous observations (Baxter et al, 2005; James et al, 2005), JAK2V617F was not detected in T lymphocytes.


Demographics Male/female Age at diagnosis, years (median; range) Interval diagnosis to JAK2 Screening, months (median; range) Laboratory studies at diagnosis Haemoglobin, g/dl (median; range) Proportion of patients with haemoglobin level < 10 g/dl Proportion of patients with red cell transfusion dependency Leucocyte count, ·109/l (median; range) Dupriez prognostic risk score (Dupriez et al, 1996) (low/intermediate/high score) Platelet count, ·109/l (median; range) Blood CD34 count ·109/l (median; range) Lactate dehydrogenase, U/l (median and range) (normal range 94–257) Physical examination/marrow findings at diagnosis of MMM Palpable spleen size in centimeters (median and range) Cytogenetic profile Normal Abnormal 20 q13 qChromosome 9 abnormalities Other abnormalities Reticulin fibrosis Grade 1 Grade 2 Grade 3 Grade 4

Clinical features

76/41 58Æ3 (18Æ5–80Æ8) 11Æ8 (0–317) 10Æ8 (5Æ8–14Æ9) (n ¼ 104) 33% (n ¼ 104) 12% (n ¼ 117) 7Æ7 (1Æ6–67) (n ¼ 101) 57/39/8 (n ¼ 104) 301 (18–2400) (n ¼ 103) 0Æ0401 (0–2Æ418) (n ¼ 35) 440 (113–2032) (n ¼ 53)

5 (0–30) (n ¼ 79) (n ¼ 86) 45 41 10 5 6 20 (n ¼ 99) 12 42 33 12

10Æ8 (5Æ8–15) (n ¼ 138) 33% (n ¼ 138) 12% (n ¼ 157) 8Æ8 (1Æ6–67) (n ¼ 135) 77/49/12 (n ¼ 138) 300 (18–2400) (n ¼ 137) 0Æ0458 (0–3Æ784) (n ¼ 50) 461 (113–2032) (n ¼ 75)

6 (0–30) (n ¼ 107) (n ¼ 115) 59 56 12 7 9 28 (n ¼ 135) 19 57 43 16

AMM (n ¼ 117)

102/55 58Æ9 (18Æ5–80Æ8) 11Æ8 (0–317)

MMM (n ¼ 157)

ª 2005 Blackwell Publishing Ltd, British Journal of Haematology, 131, 320–328 12 (3–30) (n ¼ 15) (n ¼ 16) 8 8 1 0 2 5 (n ¼ 19) 6 6 5 2

258 (97–531) (n ¼ 19) 0Æ0571 (0–0Æ1983) (n ¼ 11) 470 (232–1418) (n ¼ 13)

13 (4–28Æ9) (n ¼ 19) 14/5/0 (n ¼ 19)

11Æ4 (7Æ1–15) (n ¼ 19) 26% (n ¼ 19) 5% (n ¼ 22)

14/8 58Æ7 (44Æ6–72Æ3) 6 (0–270)

PPMM (n ¼ 22)

10 (0–25) (n ¼ 13) (n ¼ 13) 6 7 1 2 1 3 (n ¼ 17) 1 9 5 2

303 (26–618) (n ¼ 15) 0Æ1653 (0Æ0395–3Æ785) (n ¼ 4) 600 (236–1776) (n ¼ 9)

14Æ6 (2Æ9–57Æ5) (n ¼ 15) 6/5/4 (n ¼ 15)

10 (6Æ4–13Æ6) (n ¼ 15) 47% (n ¼ 15) 11% (n ¼ 18)

12/6 63Æ8 (39Æ2–77Æ6) 20 (0–141Æ7)

PTMM (n ¼ 18)

Table I. Clinical and laboratory features at diagnosis for 157 patients with myelofibrosis with myeloid metaplasia (MMM) including 117 patients with agnogenic (AMM), 22 with postpolycythaemic (PPMM) and 18 with post-thrombocythaemia (PTMM) myeloid metaplasia.

Lineage Specificity, Clinical Correlates of JAK2V617F in MMM

323

A. Tefferi et al Table II. Additional clinical and laboratory features at time of blood sample collection for JAK2V617F screening for 157 patients with myelofibrosis with myeloid metaplasia (MMM) including 117 patients with agnogenic (AMM), 22 with postpolycythaemic (PPMM) and 18 with post-thrombocythaemia (PTMM) myeloid metaplasia. Clinical and laboratory features Physical examination features at time of blood sample collection for JAK2 analysis Palpable spleen size, cm (median and range) Laboratory features at time of blood sample collection for JAK2 analysis Haemoglobin, g/dl (median; range) Leucocyte count, · 109/l (median; range) Platelet count, · 109/l (median; range) Proportion with circulating blasts Blood CD34 count, · 109/l (median; range) Lactate dehydrogenase, U/l (median and range) (normal range 94–257)

MMM (n ¼ 157)

AMM (n ¼ 117)

PPMM (n ¼ 22)

PTMM (n ¼ 18)

8 (0–33) (n ¼ 137)

7 (0–33) (n ¼ 104)

11Æ5 (0–31) (n ¼ 18)

6 (0–25) (n ¼ 15)

10Æ3 (5–16Æ1) 8Æ9 (1–84Æ4) 251 (5–1007) 54% 0Æ0527 (0–5Æ9864) (n ¼ 136) 500 (420–2435) (n ¼ 149)

10Æ2 (1Æ3–15Æ2) 8Æ3 (1–84Æ4) 242 (18–1007) 55% 0Æ0515 (0–2Æ605) (n ¼ 99) 492 (420–2435) (n ¼ 110)

12Æ2 (8Æ4–16Æ1) 13Æ6 (4Æ9–68Æ3) 280Æ5 (5–635) 36% 0Æ0529 (0–1Æ116) (n ¼ 21) 540 (111–1418) (n ¼ 22)

9Æ1 (6Æ6–14Æ5) 8Æ3 (1Æ7–66Æ2) 285 (42–872) 67% 0Æ0742 (0–5Æ986) (n ¼ 16) 607 (130–1856) (n ¼ 17)

Table III. Prevalence of JAK2V617F in 157 patients with myelofibrosis with myeloid metaplasia (MMM) including 117 patients with agnogenic (AMM), 22 with postpolycythaemic (PPMM) and 18 with post-thrombocythaemia (PTMM) myeloid metaplasia. Specific cell fractions

MMM

AMM

PPMM

PTMM

Peripheral blood mononuclear cells Homozygous Heterozygous Total with mutation Neutrophils Homozygous Heterozygous Total with mutation Myeloid progenitor cells (CD34+) Homozygous Heterozygous Total with mutation T lymphocytes (CD3+)

n* ¼ 157 9 (5Æ7) 71 (45Æ2) 80 (51) n ¼ 57 10 (17Æ5) 16 (28Æ1) 26 (45Æ6) n ¼ 25 3 (12) 13 (52) 19 (76) n ¼ 20 All negative

n ¼ 117 3 (2Æ6) 50 (42Æ7) 53 (45Æ3) n ¼ 44 6 (13Æ6) 13 (29Æ5) 19 (43Æ2) n ¼ 17 2 (11Æ8) 8 (47Æ1) 10 (58Æ8) n ¼ 9 All negative

n ¼ 22 4 (18) 16 (72Æ7) 20 (91) n¼6 4 (66Æ7) 2 (33Æ3) 6 (100) n¼5 1 (20) 4 (80) 5 (100) n ¼ 8 All negative

n ¼ 18 2 (11Æ1) 5 (27Æ8) 7 (38Æ9) n¼7 0 (0) 1 (14Æ3) 1 (14Æ3) n¼3 0 (0) 1 (33Æ3) 1 (33Æ3) n ¼ 3 All negative

*n represents all patients with and without JAK2V617F mutation. The values in parenthesis are percentages.

The overall detection rates for the mutant allele were largely similar among PBMC, neutrophils, and CD34+ cells (Table III). However, the incidence of homozygous mutations appeared to be higher in myeloid-enriched cell fractions (Table III). Consistent with this observation, among the 71 patients who were heterozygous for JAK2V617F in their PBMC, 22 had concomitant analysis in their neutrophils and 12 in their CD34+ cell fractions that showed a homozygous defect in six and one patients, respectively. However, in patients with wild-type JAK2 genotype, in DNA derived from PBMC, the results from neutrophil mutation analysis were always concordant. In other words, the likelihood of detecting either a heterozygous or homozygous JAK2V617F mutation was the same whether the analysis involved either PBMC or neutrophils. Finally, there was a 100% concordance of the results 324

between neutrophils and CD34+ cells in terms of the likelihood of detecting both heterozygous and homozygous JAK2V617F mutation.

Clinical and laboratory correlative studies The entire cohort of MMM patients (n ¼ 157) were considered for correlative studies of epidemiological features and did not identify statistically significant associations between the presence of JAK2V617F mutation in PBMC (either homozygous or heterozygous) and gender (P ¼ 0Æ32), age at diagnosis (0Æ09), or ethnic origin (P ¼ 0Æ66). The results were the same when patients with homozygous mutant JAK2 were analysed separately. Additional statistical analyses were restricted to patients with AMM (n ¼ 117) because of the aforementioned


Lineage Specificity, Clinical Correlates of JAK2V617F in MMM significant difference in the incidence of the mutation among MMM subcategories as well as the relatively small sample sizes for both PPMM (n ¼ 22) and PTMM (n ¼ 18). In this population, the presence of either a heterozygous or homozygous JAK2V617F mutation did not correlate with gender, ethnic origin, the interval from diagnosis to JAK2V617F screening, overall disease duration, spleen size, haemoglobin level, incidence of a hemoglobin level