JAK2V617F occurs in approximately 50% of patients with essential thrombocythemia (ET). Qualitative studies of mutation analysis have previously reported an association between JAK2V617F and advanced age, higher hemoglobin level, higher leukocyte count, and lower platelet count. A possible association with thrombotic complication has also been considered.
Allele-specific, quantitative polymerase chain reaction (PCR) analysis for JAK2V617F was performed in 176 patients with ET using genomic DNA from archived bone marrow, which was collected within 1 year (n = 72 patients), between 1 and 5 years (n = 64 patients), or after 5 years (n = 40 patients) of diagnosis.
JAK2V617F was detected in 96 patients (55%), in whom mutant allele burden ranged from 1% to 100% (median, 6.3%). Neither mutational frequency (P = .37) nor mutant allele burden (P = .62) was affected by the timing of bone marrow sample collection. The presence of JAK2V617F was found to be significantly associated with higher hemoglobin level (P < .0001), lower platelet count (P = .001), higher leukocyte count (P = .008), increased incidence of venous thrombosis occurring after diagnosis (P = .02), and older age at diagnosis (P = .03). All but age retained significance in multivariable analysis. In mutation-positive patients (n = 96 patients), JAK2V617F allele burden clustered between 1% and 22% in 94 cases, in whom it correlated directly and significantly with platelet and leukocyte counts, palpable splenomegaly at diagnosis, and venous thrombosis occurring after diagnosis. The latter 2 associations remained significant with the inclusion of the remaining 2 outlier cases with 100% mutant allele burden; in addition, an association with male gender became evident.
The reported JAK2V617F mutational frequencies in essential thrombocythemia (ET) range from 23%1 to 75%2 and the presence of the mutation has been significantly associated with older age at diagnosis,3–5 higher hemoglobin level,3–8 higher leukocyte count,3, 4, 8 and lower platelet count.4–6 In addition, the few cases transforming from ET to polycythemia vera (PV) were nearly all JAK2V617F-positive.3, 4 In contrast, the presence of JAK2V617F did not appear to influence either survival3 or disease transformation rates into either acute myeloid leukemia (AML) or myelofibrosis.3, 4 Furthermore, in what to our knowledge is the largest relevant study in ET, the presence of JAK2V617F was associated with venous but not arterial thrombotic events,4 whereas a general correlation with thrombosis risk was not apparent in 3 other reports with more than 100 study patients.3, 6, 8 In the latter 3 reports as well as in other smaller studies5, 7 that demonstrated a significant association between JAK2V617F and thrombosis, a distinction between venous and arterial thrombosis was not made.
Interstudy differences in diagnostic criteria applied, source of DNA/RNA used for mutation analysis, and/or screening assay sensitivity might underlie the discrepancy in reported JAK2V617F incidence figures in ET. Furthermore, the impact of interpatient variability in JAK2V617F allele burden,2 on either the previously outlined genotype-phenotype associations or clinical outcome, has not been previously addressed. In the current study we employed an allele-specific, quantitative real-time polymerase chain reaction assay (AS-qPCR),9 on genomic DNA obtained from archived bone marrow of 176 consecutive patients with World Health Organization (WHO)-defined ET10 to examine clinical correlates of both the presence (as opposed to the absence) of JAK2V617F and mutant allele burden. We also aimed to clarify the details of the possible association between JAK2V617F and thrombosis by separately considering ‘arterial’ vs ‘venous’ thrombosis and ‘at diagnosis’ vs ‘postdiagnosis’ thrombosis.
MATERIALS AND METHODS
With permission from the Mayo Clinic Institutional Review Board, the institutional database for ET was reviewed to select patients in whom archived bone marrow, collected either at the time of diagnosis or during the chronic phase of the disease, was available for JAK2V617F mutation analysis. The database included all ‘ET’ patients seen at the Mayo Clinic in the last 30 years. All cases were rereviewed for both clinical information and bone marrow histopathology, regardless of when they were diagnosed, to ensure fulfillment of the WHO criteria for ET.10
Previously published methodology was used for JAK2V617F AS-qPCR.9 Quantitative PCR amplification and detection were performed on an ABI Prism 7900 HT Analyzer (Applied Biosystems, Foster City, Calif). Mutant allele burden was reported as the percentage of total JAK2 represented by JAK2V617F (ie, JAK2V617F/ JAK2V617F + JAK2 wildtype).2, 9 This JAK2V617F percentage was obtained from a standard curve for JAK2V617F/total JAK2 against ΔCt (CtJAK2V617F-CtJAK2WT), which was constructed using various proportions of genomic DNA from the cell lines K562 (homozygous for wildtype JAK2) and HEL (homozygous for JAK2V617F). Two μL of each standard mixture (25 ng/μL) and each sample (25 ng/μL) were aliquoted into separate wells of the 96-well plate. Standards and samples were run in triplicate and the mean ΔCt, for each standard mixture and sample, was used to plot the JAK2V617F/JAK2 total percentage.
Statistical procedures utilized were conventional and all data were analyzed using StatView software (SAS Institute Inc., Cary, NC). All P-values were 2-tailed and statistical significance was set at the level of P < .05. Comparison between categorical variables was performed by chi-square statistics. Linear regression analysis was used for testing associations between continuous variables. Comparison between categorical and continuous variables was performed by either the Mann-Whitney U-test or Kruskal-Wallis test. The association of variables selected from univariate analysis was explored using logistic regression models. Overall survival functions were estimated by Kaplan-Meier plots taking the interval from the date of diagnosis to death or last contact. The log-rank test was used to test the homogeneity of survival curves over different groups. The Cox proportional hazards model was utilized to determine the impact of various clinical and laboratory variables on survival.
The study population was comprised of 176 consecutive patients with ET (median age, 57 years; 110 women), in whom archived bone marrow, which was collected during the chronic phase of the disease, was available for JAK2V617F qPCR analysis of genomic DNA. The bone marrow samples used for mutation analysis were collected either within 1 year of diagnosis (n = 72 patients) or at a variable time from diagnosis: between 1 and 5 years (n = 64 patients) or after 5 years (n = 40 patients) of diagnosis. At the time of this writing, 34 patients had died and the median follow-up for the patients who were alive was 59 months (range, 0–287 months). During this period, disease transformation into AML (n = 3 patients), myelofibrosis (n = 3 patients), or PV (n = 1 patient) occurred in 7 patients. A total of 70 patients (40%) experienced major thrombosis, either at diagnosis (40 patients; 23%) or during follow-up (45 patients; 26%). Overall treatment included no platelet-lowering drug in 47 patients, only anagrelide or interferon (IFN)-α in 18 patients, hydroxyurea as a single agent in 49 patients, hydroxyurea and anagrelide in 45 patients, and other alkylating agents in 17 patients. In addition, 128 patients received aspirin therapy.
JAK2 Analysis Correlates
Overall, JAK2V617F was detected in 96 of the 176 study patients (55%). In JAK2V617F-positive patients, the median mutant allele burden ranged from 1% to 100% (median, 6.3%). However, JAK2V617F allele burden in 94 of the 96 patients (98%) was tightly clustered between 1% and 22% (median, 6%), whereas the remaining 2 patients both displayed 100% mutant allele burden in bone marrow samples collected within 1 year of diagnosis. The clinical phenotype in these latter 2 patients was typical for ET, in terms of both bone marrow histology and other clinical and laboratory features, and, at the time of last follow‒up, both patients had been followed for 3 to 9 years without the development of PV, myelofibrosis, or thrombotic complications noted.
Neither mutational frequency (P = .37) nor allele burden distribution among JAK2V617F-positive cases (P = .62) was affected by the timing of test sample accrual (Table 1). The presence of JAK2V617F was significantly associated with higher hemoglobin level, lower platelet count, higher leukocyte count, venous thrombosis occurring during follow-up, and older age at diagnosis (Table 2). These associations, except the 1 with age, remained significant during multivariable analysis (P-values of 0.004, 0.04, 0.04, 0.05, and 0.09, respectively). Of note, there was no significant association noted between the presence of JAK2V617F and either arterial thrombosis, occurring either at diagnosis or postdiagnosis, or venous thrombosis at diagnosis (Table 2). The association between JAK2V617F and postdiagnosis venous vascular events did not maintain its significance in a multivariable model that incorporated history of venous thrombosis at time of diagnosis.
Table 1. Mutational Frequency and JAK2V617F Allele Burden in Relation to Time of Sample Accrual in Essential Thrombocythemia
Time of sample collection
JAK2V617F-positive cases (%)
Median JAK2V617F allele burden in mutation-positive cases (range)*
JAK2V617F allele burden refers to the proportion of the mutant allele in relation to total JAK2 (ie, JAK2V617F/JAK2V617F + wildtype JAK2).
At any time during the chronic phase of the disease (n = 176)
Within 1 y of diagnosis (n = 72)
Between 1 and 5 y of diagnosis (n = 64)
5 y after diagnosis (n = 40)
P = .37
P = .62
Table 2. Comparison of JAK2V617F-Positive and JAK2V617F-Negative Cases of Essential Thrombocythemia
All patients (n = 176)
JAK2V617F-positive cases (n = 96)
JAK2V617F-negativecases (n = 80)
AML indicates acute myeloid leukemia; MMM, myelofibrosis with myeloid metaplasia; PV, polycythemia vera.
Note: Mutation analysis was performed by allele-specific, quantitative real-time polymerase chain reaction assay (AS-qPCR).
Median age at diagnosis, y (range)
Median hemoglobin level at diagnosis, g/dL (range)
Median leukocyte count at diagnosis, ×109/L (range)
Median platelet count at diagnosis ×109/L (range)
Palpable splenomegaly at diagnosis (%)
Microvascular symptoms at diagnosis (%)
Major thrombosis at diagnosis (%)
Major thrombosis during follow-up (%)
Major thrombosis at or after diagnosis (%)
Arterial thrombosis at or after diagnosis (%)
Arterial thrombosis at diagnosis (%)
Arterial thrombosis during follow-up (%)
Venous thrombosis at or after diagnosis (%)
Venous thrombosis at diagnosis (%)
Venous thrombosis during follow-up (%)
Cases that developed AML/MMM/PV
Median survival in months (Kaplan-Meier estimate)
Next, we focused on JAK2V617F-positive patients (n = 96) and our first analysis excluded the 2 aforementioned outliers with 100% mutant allele burden. In the remaining 94 patients, using linear regression analysis, we found a significant and direct correlation between mutant allele burden and both platelet (P = .04) (Fig. 1A) and leukocyte (P = .05) (Fig. 1B) counts, but not age (P = .94) (Fig. 1C) or hemoglobin level (P = .36) (Fig. 1D). In addition, mutant allele burden was significantly higher in patients presenting with palpable splenomegaly (P = .006) as well as those experiencing venous thrombosis after diagnosis (P = .04). Conversely, mutant allele burden did not appear to be significantly correlated with gender (P = .13), the presence of microvascular symptoms (P = .25), arterial thrombosis either at diagnosis (P = .54) or during follow-up (P = .21), venous thrombosis at diagnosis (P = .26), or major hemorrhage (P = .6). Furthermore, these findings remained valid during comparison of mutation-positive patient categories with an allele burden cutoff level at either 6% (the median value) or 10% (ie, 22 patients with ≥10% mutant allele burden were compared with 72 with lower allele burden). However, when the 2 outlier cases with 100% mutant allele burden were included in the analysis (ie, n = 96 patients rather than 94 patients), a new significant (P = .05) association between higher JAK2V617F allele burden and male gender became evident, and, of the aforementioned associations, only those with splenomegaly (P = .01) and venous thrombosis (P = .05) remained significant.
Survival analysis by both Kaplan-Meier statistics and Cox proportional hazards model analysis did not disclose a significant difference between mutation-positive and mutation-negative patients (P = .1 and .11, respectively). It is interesting to note that in mutation-positive patients (n = 96), JAK2V617F allele burden did not significantly affect survival when all 96 patients were considered, but a significant association with inferior survival became apparent when the aforementioned 2 outlier cases with 100% mutant allele burden were excluded from the analysis. The number of patients with documented disease transformation into PV, myelofibrosis, or AML was too small to allow valid statistical analysis in that regard (Table 1). However, qPCR analysis for JAK2V617F was performed in a separate group of 10 ET patients in whom bone marrow samples were collected at time of disease transformation into either AML or myelofibrosis and mutational frequency in these 10 patients was similar to the 176 patients in whom test samples were collected during the chronic phase of the disease (55% vs 50%; P = .78). In addition, in JAK2V617F-positive cases the samples collected at time of disease transformation had significantly higher mutant allele burden (P = .03).
JAK2V617F is now considered a valuable clonal marker in the evaluation of polycythemia,11 thrombocytosis,2 and ‘idiopathic’ abdominal vein thrombosis.12 The original studies regarding JAK2V617F applied mostly PCR sequencing (assay sensitivity of 5–20%) for mutation screening. The use of AS-PCR (assay sensitivity of <5%) in 2 of these reports increased the likelihood of detecting the mutation: in PV for example from the 70% to 80% range to over 95%.9, 13, 14 However, a substantially better concordance rate between the 2 methods of mutation detection was subsequently reported by another group of investigators15 and mutational frequency in our previous ET study (49%), in which we used PCR sequencing,3 was not significantly different than what is reported in the current study (55%). Regardless, AS-qPCR assay is currently the preferred mutation screening test because of its ability to quantify allele burden, a feature with potential utility for monitoring treatment response.16
In the current study, using AS-qPCR (assay sensitivity of 0.5%) to detect JAK2V617F, we confirmed our previous observations that were based on mutation analysis by PCR sequencing; the presence of JAK2V617F in ET was found to be significantly associated with older age at diagnosis and higher levels of both hemoglobin and leukocyte count3 (Table 1). Several other studies have also found similar associations in this regard.3–8 In the current, but not our previous3 study, the presence of the mutation was also associated with a lower platelet count. The association between the presence of JAK2V617F and a lower platelet count in ET has also been previously noted.4–6 In the current study, mutant allele burden in JAK2V617F-positive ET was not found to be directly associated with either age or hemoglobin level, whereas the direction of its association with platelet count was opposite to what was expected. These observations, if confirmed by others, suggest a certain degree of dichotomy in the phenotypic effects of the presence vs allele load of JAK2V617F in ET. However, it is important to recognize, as is evident from Figure 1, that the correlations between JAK2V617F allele burden and clinical features in ET are not as tight as might be implied by the results of the statistical analysis for significance.
Another finding from the current study was the association of both the presence of JAK2V617F and its allele burden with venous thrombosis occurring after diagnosis. The specificity of the association between JAK2V617F and venous, but not arterial, thrombosis was previously noted by Campbell et al4 and provides a potential explanation for the otherwise discrepant reports on the relation of the specific mutation with ‘thrombosis’ in general.4–8 Because of recent reports of a significant association between leukocytosis and thrombosis in ET8, 17 as well as in PV,18 we performed a multivariable analysis that included both JAK2V617F mutational status and leukocytosis as covariates and confirmed the independent effect of the presence of the mutation on venous thrombosis risk. This was not totally unexpected because the aforementioned prothrombotic effect of leukocytosis in ET was restricted to arterial events.8 Conversely, leukocytosis might be the link in some of the reported association between the presence of the mutation and ‘thrombosis’ if the latter was primarily comprised of arterial events. Regardless, the association between either the presence of JAK2V617F or its allele burden with postdiagnosis venous thrombosis did not remain significant in a multivariable analysis that incorporated history of venous thrombosis as a covariate, a finding that might undermine the overall clinical relevance of the particular observation.
Other JAK2V617F allele burden associations in mutation-positive patients included a direct correlation with the presence of palpable splenomegaly, male gender, and, when survival analysis excluded the 2 outlier cases (each of whom displayed 100% mutant allele burden), inferior survival. However, we consider these findings preliminary and subject to confirmation by other studies. The potential association of splenomegaly with JAK2V617F allele burden was recently reported in PV, where mutant allele burden was also directly correlated with thrombosis, hemoglobin level, leukocyte count, and presence of pruritus.19 Patients with high JAK2V617F allele burden in the latter study were also reported to display lower platelet count as well as symptomatic disease. Some of these findings were noted earlier during comparison of homozygous vs heterozygous JAK2V617F in PV assessed by PCR sequencing.20
Finally, it is unlikely that the current findings would be any different had all study samples been collected at the time of diagnosis because JAK2V617F allele burden as well as overall clonal load in ET remain stable over many years.21 Furthermore, the substantially lower levels of mutant allele burden in the current study, compared with those reported in other studies,2, 9 most likely is due to the fact that our study utilized whole bone marrow, which is a mixture of myeloid, lymphoid, and other stromal cells, as opposed to granulocyte-enriched samples. We have previously shown that the mutation might be absent in nonmyeloid cells as well as affect a lesser fraction of myeloid progenitors, as opposed to granulocytes.22