Leukocytosis is a risk factor for thrombosis in essential thrombocythemia: interaction with treatment, standard risk factors, and Jak2 mutation status

Essential thrombocythemia (ET) is a Philadelphia-chromosome–negative myeloproliferative disorder (MPD) distinct from polycythemia vera and idiopathic myelofibrosis and characterized by persistent thrombocytosis, excessive proliferation of megakaryocytes in the bone marrow, normal erythrocytic mass, and the absence of prominent bone marrow fibrosis.¹  In spite of the insidious clinical onset, ET may be complicated by severe vascular complications and presents a variable potential to undergo progression to myelofibrosis or transformation into acute leukemia. The aim of cytoreductive therapy is to prevent thrombosis and hemorrhage without increasing the risk of hematologic transformation. To accomplish this goal, drugs are prescribed after a stratification approach dictated by the thrombotic risk.²  In a previous paper we showed that age over 60 years and prior thrombotic events were the 2 independent factors able to predict the occurrence of vascular complications in ET patients.³  This stratification is now widely accepted for treatment decisions in the single patient and in clinical trials.^4,5 

In the last few years, new information on the pathogenesis of thrombosis in MPD became available, including the role of leukocyte activation and interaction with platelets and endothelial cells^6–9  and a possible place of the recently identified JAK2 V617F mutation, found in nearly 50% of ET patients.¹⁰ 

Therefore, by using our own database, we analyzed the prognostic role of thrombosis for other potential determinants of individual risk such as white blood cell (WBC) values and JAK2 mutational status in a large population of ET patients.

All of the analyses were performed using the database of 439 patients with ET regularly followed at our Hematology Division (Ospedali Riuniti di Bergamo) from 1981 to 2006 (Table 1). ET was diagnosed according to Polycythemia Vera Study Group (PVSG) criteria.¹¹  One hundred eighty-nine patients (43%) were newly diagnosed whereas the others were referred to our division after being diagnosed with ET elsewhere. Only patients with confirmed diagnosis were included in the follow-up. The time elapsed from diagnosis to referral is reported in Table 1. Data regarding laboratory values, treatments, and clinical outcomes were collected at diagnosis and at least every 6 months during the follow-up in all patients. JAK2 mutation was assessed in 277 cases (63%). Informed consent was obtained for each subject.

Patients were classified as being at low or high risk for thrombosis according to standard risk factors (age ≥ 60 years and/or a previous major thrombotic event). Low-risk patients were followed with no cytoreductive therapy whereas high-risk patients were given hydroxyurea (HU) with the aim to reduce platelet counts to less than 600 × 10⁹/L.⁴  HU was also given to 10 (2%) patients at low thrombotic risk but with extreme thrombocytosis (platelet count > 1500 × 10⁹/L) to avoid hemorrhagic complications. Two hundred nine patients (47%) with clear indication to antiplatelet therapy (ie, previous arterial thrombosis or microvascular symptoms such as erythromelalgia or atypical cerebral or visual disturbances) received low-dose aspirin (100 mg/d).

Vascular events considered in the present analysis included ischemic stroke, cerebral transient ischemic attack (TIA), acute myocardial infarction (AMI), peripheral arterial thrombosis (PAT), and venous thromboembolism (VTE). Diagnostic procedures for establishing thrombosis included cerebral computed tomography (CT) or magnetic resonance imaging for stroke, characteristic neurologic symptoms for TIA, electrocardiography and/or increased cardiac enzymes for AMI, angiography for PAT, and ultrasonography of the arms or legs or pulmonary ventilation-perfusion scan or CT scan for VTE. The median duration of the follow-up was 6.2 years (0.1-21 years).

The study was conducted according to good clinical and laboratory practice rules and the principles of the Declaration of Helsinki, and the Ospedali Riuniti Bergamo Ethics Review Committee approved the protocol.

Peripheral-blood granulocytes, mononuclear cells, and platelets were prepared using HetaSept gradient (Stem Cell Technologies, Vancouver, BC, Canada) and DNA was extracted using pure gene DNA Isolation Kit (Genera Systems, Minneapolis, MN) in accordance with manufacturer's procedure. One hundred nanograms of patient's DNA was amplified by allele-specific polymerase chain reaction (PCR) using a common reverse primer JAK2 R (5′-CTG AAT AGT CCT ACA GTG TTT TCA GTT TCA -3′; 50 μM) and 2 forward primers (25 μM) named JAK2 F Mut (5′-AGC ATT TGG TTT TAA ATT ATG GAG TAT ATT-3′), specific for the mutant allele, and JAK2 F WT (5′-ATC TAT AGT CAT GCT GAA AGT AGG AGA AAG-3′), which amplifies the wild-type allele. PCR was performed at an annealing temperature of 59°C for 35 cycles. The result was a 203-bp product for mutant samples and a 364-bp product for mutant and wild-type alleles. The heterozygous or homozygous status of exon 12 of JAK2 was analyzed with a restriction enzyme–based assay: PCR product (460 bp) obtained from 100 ng DNA amplified with primers JAK2 ex12-F (5′-GGG TTT CCT CAG AAC GTT GA-3′) and JAK2 ex12-R (5′-TCA TTG CTT TCC TTT TTC ACA A-3′) with an annealing temperature of 57°C was digested with BsaXI enzyme for 4 hours at 37°C. The mutant homozygous allele was not digested; the mutant heterozygous allele was partially digested while the wild-type allele was completely digested into 241-bp, 189-bp, and 30-bp products.¹² 

Univariate analysis was performed to evaluate differences in proportions by the chi-square and Fisher exact tests. Differences in continuous variables were tested using the t test.

Two different multivariable approaches have been adopted by fitting various Cox proportional-hazards models. Firstly, a multivariable analysis using values measured at diagnosis was used to assess whether the level of exposition for a potential risk predictor captured at diagnosis could be found to be a statistically significant marker of increased probability of recurrence of thrombosis during follow-up. The multivariable model to assess this hypothesis has been fitted after adjusting for sex; standard risk factors for thrombosis (age ≥ 60 years and/or previous thrombosis); and median levels of white blood cell count, hemoglobin level, hematocrit, and platelet count registered at diagnosis. Secondly, the database was explored using Cox proportional-hazards model with time-varying covariates to evaluate risk, and with right censoring at the first thrombotic event or last date of follow-up. This analysis included variables considered “fixed” in time (ie, sex and standard risk factors) and time-varying variables: treatments (HU, antiplatelet drugs) and count of blood cells (median levels of hemoglobin, hematocrit, white blood cells, platelets). Continuous variables have been analyzed using median levels at baseline as cutoff values and maintaining the same cutoff values in the time-dependent analysis for consistency. Time-dependent variables were updated yearly, and the last available value before a thrombotic event was used to estimate the relative risk. Where appropriate, the substitution of the missing data for incomplete repeated measures was done with the last value carried forward. The latter analysis allowed assessing whether the level of exposition to a factor measured before a thrombotic recurrence was associated with the probability of having that event.

Analyses were performed with SAS 9.1 software (Cary, NC). All probability values are 2 tailed (<.05).

The characteristics of the 439 patients are shown in Table 1. One hundred twenty-nine episodes of vascular thrombosis were registered at diagnosis or in the previous history in 113 patients (26%). During the follow-up, 78 events were diagnosed in 67 patients (15%). The distribution of events is quoted in Table 1.

On univariate analysis, parameters at diagnosis associated with thrombosis registered in the follow-up were age 60 years or older and previous history of thrombosis (P = .02) and leukocyte count above the median level of 8.7 × 10⁹/L (P = .01). No significant association was found considering arterial and venous thrombosis separately, sex, median levels of hemoglobin, hematocrit, and platelet count.

Two statistical multivariable analyses were performed considering the variables at diagnosis and before the vascular events occurring during follow-up (Table 2). The multivariable “baseline” analysis taking into account sex; standard risk factors for thrombosis (age ≥ 60 years and/or previous thrombosis); and median levels of hemoglobin, hematocrit, platelets, and white blood cells showed a significant and independent association between standard risk factors as well as high levels of white blood cells with the occurrence of thrombosis in the follow-up (hazard ratio = 2.3 for both). Platelet count above the median level of 784 × 10⁹/L was not found to be an independent risk factor for subsequent vascular events.

On multivariable “time-dependent” analysis we took into account the last value of blood cell counts measured before vascular events. Treatment with HU decreased the leukocyte count to about 30%, from the baseline median level of 8.7 × 10⁹/L to 6.1 × 10⁹/L, and their statistical significance was lost (hazard ratio = 1.6, 95% confidence interval = 0.9 to 2.8, P = .06). In contrast, conventional risk factors (age ≥ 60 years and/or previous thrombosis) retained their independent significant value (hazard ratio = 1.8, 95% confidence interval = 1.1 to 3.0, P = .04). The antithrombotic efficacy of treatment was confirmed by the demonstration of an independent protective effect of both HU and low-dose aspirin (hazard ratio = 0.5 for both).

In order to evaluate the interaction between conventional risk categories and WBC levels, univariate analysis showed a different risk profile of white blood cell levels in low-risk and high-risk patients. The rate of thrombotic events was significantly higher in low-risk patients with high leukocyte counts compared with low-risk patients with low leukocyte counts (relative risk 2.7, 95% confidence interval 1.2 to 6.4). In contrast, leukocyte levels did not significantly predict the risk of thrombosis among high-risk ET patients (relative risk 1.6, 95% confidence interval 0.8 to 3.2). In a multivariable time-dependent model (Table 3), low-risk patients with high levels of leukocytes were associated with a significantly increased risk of thrombosis (hazard ratio = 3.1, 95% confidence interval = 1.4 to 7.1), similar to that observed in high-risk patients with a normal level of leukocytes (hazard ratio = 2.5, 95% confidence interval = 1.0 to 6.0). The presence of both of the risk factors was associated with the highest probability to develop a vascular event during the follow-up (hazard ratio = 5, 95% confidence interval = 2.1 to 11.9).

As far as the JAK2 mutation is concerned, in univariate analysis performed on 277 patients (38 vascular events valuable in the follow-up), leukocyte counts were found to be significantly higher in patients with the V167F mutation (90 of 151, or 60%) than in JAK2 wild-type patients (45 of 126, or 36%; P < .001). Moreover, V617F mutation status was associated with higher hemoglobin levels and lower platelet counts (P < .001 for both). In multivariable analysis, JAK2 mutation was not found to be a significant independent predictor of thrombosis occurring in the course of the disease (hazard ratio = 1.4, 95% confidence interval = 0.7 to 3.0).

This study shows that an increased number of WBCs is a relevant marker of thrombosis in patients with ET and that this risk factor is reduced by cytoreductive treatment with HU. Several pieces of evidence support these statements.

Firstly, a multivariable analysis demonstrates that patients with a baseline WBC count above the median had a hazard risk of developing thrombosis about twice that of patients with lower WBC counts. This finding is in keeping with a recent analysis by Wolanskyj et al¹³  who reported that age, previous thrombosis, and leukocytosis were independent risk factors for both major thrombosis and a poorer long-term survival when evaluated at diagnosis. Unlike our results, these investigators found that previous history of arterial but not venous thrombosis was associated with recurrence of thrombotic events.

Secondly, cytoreductive therapy lowered the number of WBCs during the follow-up and this was associated with a reduction of their thrombogenic effect.

Thirdly, the role of WBCs in predicting thrombosis appears to be more evident in untreated low-risk than in treated high-risk patients.

Several pathophysiologic mechanisms may explain a thrombogenic role of WBCs in MPDs. Our group and others showed that in these disorders neutrophils circulate in an activated state and are able to bind to platelets in a dynamic adhesive process, which reflects the activation of both platelets and leukocytes.^6–9  This process triggers the expression of tissue factors as well as endothelial activation and damage.¹⁴  In addition, leukocytosis may contribute to inflammatory processes in atherosclerotic plaques and in this way increase the probability of vascular events.^15–17  Interestingly, leukocytes are of major importance in the pathophysiology of other occlusive diseases, such as sickle-cell disease.¹⁸  In these patients, an increased baseline white cell count has been found to be an independent risk factor for acute chest syndrome and cerebral infarction, and quantitative and qualitative reductions in leukocytes during hydroxyurea treatment were correlated with a better disease outcome.

In all our analyses, platelet number did not show any correlation with the vascular events. While extreme thrombocytosis may correlate with the occurrence of hemorrhage due to an acquired von Willebrand disease,^19–21  the contribution of platelet number to major thrombotic events remains poorly established.²²  In a randomized clinical trial that included high-risk ET patients, HU treatment was found to be superior to anagrelide, a selective cytoreductive antiplatelet drug, in preventing arterial thrombosis.⁵  This supports the concept that the reduction of events achieved by HU was not due to platelet count normalization only, but, very likely, to the suppression of the 3 lineages of bone marrow hematopoiesis, including leukocytes.

The putative thrombogenic role of leukocytosis generates the hypothesis of a reassessment of the risk stratification of ET patients. Our data suggest that “true” low-risk patients, not requiring cytoreductive therapy, are those without conventional clinical risk factors (age < 60 years and no previous thrombosis) and a low WBC count. In fact, patients previously classified as “low-risk” but carrying leukocytosis present a thrombotic risk similar to the conventional “high-risk” patients. These patients should be strongly considered for a cytoreductive therapy if our retrospective data are confirmed in other prospective clinical studies.

The contribution of JAK2 mutation in the risk classification of ET patients is still a matter of debate. Some authors have reported an association between JAK2 V617F and thrombosis, ^23,24  but this was not confirmed in other studies.^25,26  Different patients' selections and study designs may have played a role in these discrepancies. In our analysis, leukocytosis was more frequent in patients carrying the JAK2 mutation, but a significant role of JAK2 in predicting thrombosis was not found. It should be noted, however, that the JAK2 mutation was evaluated in only 277 of the 439 cases included in the study (with only 38 vascular events valuable) and this could have reduced the statistical power of the analysis. Interestingly, JAK2-mutated ET patients were found to have a more pronounced activation of platelets and leukocytes^9,27  and a more effective response to HU treatment²³  than those without the mutation. The findings in the current study suggest that previous reports of an association between thrombosis and the presence of JAK2 V617F in ET might have been an indirect effect from leukocytosis, which is significantly associated with the presence of the mutation.

Contribution: A.C. performed research and analyzed data; G.F. performed research and wrote the manuscript; V.G. performed laboratory experiments; O.S. supervised the molecular analysis of JAK2 mutation; F.D. collected data; R.M. supervised the project and wrote the manuscript; G.B. supervised statistical analyses; A.R. performed research; and T.B. designed and supervised the research project, wrote the manuscript, and raised funds.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Tiziano Barbui, Divisione di Ematologia, Ospedali Riuniti, Largo Barozzi 1, 24100 Bergamo, Italy; e-mail: tbarbui@ospedaliriuniti.bergamo.it.

This work was supported in part by grants from Associazione Italiana per la Ricerca contro il Cancro (AIRC), European LeukemiaNet Sixth Framework Programme LSH-2002-2.2.0-3, Associazione Paolo Belli, and Associazione Italiana Lotta alla Leucemia (AIL), sezione Paolo Belli.