Advanced age and a history of thrombosis were well-established risk factors for thrombosis in essential thrombocythemia (ET) and polycythemia vera (PV); cytoreductive therapy was indicated in their presence. Recent studies have suggested leukocytosis as an additional risk factor; however, such an association would be treatment-relevant in the context of low-risk disease.
In this retrospective study, a Cox proportional hazards model was used to determine the impact of various clinical and laboratory variables, including leukocytosis, on thrombosis-free survival (TFS). Arterial-specific or venous-specific TFS curves for different leukocyte count-defined risk groups were constructed by the Kaplan-Meier method and compared using the log-rank test.
A total of 407 low-risk patients (254 with ET and 153 with PV) were considered. After a respective median follow-up of 104 months and 130 months, respectively, 47 (19%) patients with ET and 41 (27%) with PV experienced a total of 55 (41 arterial and 14 venous) and 46 (22 arterial and 24 venous) thrombotic events, respectively. Leukocytosis at the time of diagnosis, defined by a cutoff level of either 15 or 9.4 × 109/L, did not appear to be predictive of either arterial or venous thrombosis during follow-up; similar results were obtained when analysis was restricted to patients with platelet counts of <1000 × 109/L. Instead, advanced age was found to be significantly associated with arterial thrombosis in patients with PV and higher hemoglobin level with venous thrombosis in patients with ET.
The 2 well-established risk factors for thrombosis in both essential thrombocythemia (ET) and polycythemia vera (PV) are advanced age (≥60 years) and a prior thrombotic event.1-5 Whether the presence of cardiovascular risk factors confers additional risk is open to discussion.6-8 What is no longer controversial is the lack of prognostic relevance attached to platelet count.2, 9 Instead, leukocytosis has recently been implicated as an adverse prognostic feature, in both ET and PV, in terms of both thrombosis and other disease complications. For example, in PV, a leukocyte count of ≥15 × 109/L has been associated with inferior survival,2 leukemic2 or fibrotic10 transformation, venous thrombosis during follow-up,2 and myocardial infarction.11 In ET, a similar degree of leukocytosis was associated with inferior survival12 and thrombosis at the time of diagnosis (both arterial and venous),13 whereas a leukocyte count >8.7 × 109/L was associated with arterial thrombosis at diagnosis13 and total thrombotic events during follow-up.5
The results from more recent studies, regarding the relation between thrombosis and leukocytosis in ET, have been inconsistent. For example, in a Mayo Clinic study, multivariate analysis that included conventional risk factors as covariates did not find a leukocyte count of either ≥15 × 109/L or >8.7 × 109/L to be an independent risk factor for either arterial or venous thrombosis during follow-up.13 However, another more recent Italian study reported an independent association between leukocytosis at the time of diagnosis and arterial thrombosis (especially myocardial infarction) during follow-up.14 Furthermore, the detrimental effect of leukocytosis was most apparent in low-risk patients. The latter observation is particularly noteworthy because of its potential to affect treatment decision. In the current study, we sought to further clarify the relation between leukocytosis at the time of diagnosis and the subsequent occurrence of either arterial or venous thrombosis in low-risk patients with ET or PV. The study also examined other predictors of thrombosis under similar circumstances.
MATERIALS AND METHODS
We reviewed the medical records of all patients diagnosed with PV and ET at the Mayo Clinic between 1956 and 2005 with subsequent follow-up information last updated in March 2008. The study protocol, which included waiver of informed consent, was approved by the Mayo Clinic Institutional Review Board. The study population was selected based on clinical and laboratory information, including bone marrow histology, at the time of initial diagnosis. The World Health Organization criteria were used for diagnosis.15 It is important to underscore the finding that the current study included only low-risk patients (age <60 years and no history of major thrombosis at or before the time of diagnosis). Major arterial thromboses were defined as transient ischemic attacks, thrombotic cerebrovascular accidents, angina pectoris, myocardial infarction, and peripheral arterial thromboembolism. Major venous thromboses were defined as deep venous thrombosis of the peripheral vasculature, pulmonary embolism, and abdominal large vein thrombosis. Cardiovascular risk factors considered in this study included hypertension, tobacco use, diabetes mellitus, and hyperlipidemia. A subgroup of patients for whom archived bone marrow or peripheral blood cells were available were screened for JAK2V617F mutation according to previously published methods.16
The Cox proportional hazards model was used to determine the impact of various clinical and laboratory variables on thrombosis-free survival (TFS). A leukocyte count cutoff level of 15 × 109/L was used for PV,2, 11 whereas cutoff levels of both 15 × 109/L and 9.4 × 109/L were used for ET,3, 5, 12-14, 17 based or previously published information. Arterial-specific or venous-specific TFS curves were constructed by the Kaplan-Meier method, taking the interval from the date of diagnosis to a thrombotic event or last contact. The log-rank test was used to test the homogeneity of TFS curves over the different leukocyte count-defined risk groups. Data analysis was performed using Stat View (SAS Institute, Inc, Cary, NC). All analyses were based on laboratory parameters obtained at the time of diagnosis. All P values were 2-tailed, and statistical significance was set at the level of P < .05.
Of the 407 study patients, 153 had PV (median age, 48 years [range, 11-59 years]; 43% female) and 254 had ET (median age, 42 years [range, 5-59]; 71% female). Patient characteristics at the time of diagnosis and the incidences of arterial and venous thrombotic events, major bleeding episodes, and leukemic or fibrotic disease transformation during follow-up are outlined in Table 1. At least 1 cardiovascular risk factor was present in 51% of patients with PV and 46% of patients with ET. A total of 46 thrombotic events (22 arterial and 24 venous) were recorded in 41 (27%) patients with PV during a median follow-up of 130 months (range, 2‒562 months). The corresponding figures in patients with ET were 55 total thrombotic events (41 arterial and 14 venous) in 47 (19%) patients at a median follow-up of 104 months (range, 0.25-424 months). The incidences of other disease complications are listed in Table 1.
Table 1. Patient Characteristics at Diagnosis and the Incidences of Arterial and Venous Thrombotic Events, Major Bleeding Episodes, and Leukemic or Fibrotic Disease Transformation During Follow-Up Among 153 Low-Risk Patients With Polycythemia Vera and 254 Low-Risk Patients with Essential Thrombocythemia
Polycythemia Vera (n = 153)
Essential Thrombocythemia (n = 254)
M indicates male; F, female.
Major hemorrhagic complication included a symptomatic bleeding episode that affected the retina, the central nervous system, or the retroperitoneum; any other site of bleeding that resulted in hemodynamic instability; the need for transfusion of ≥2 units of red blood cells; or a decrease in hemoglobin level by at least 2 g/dL from baseline.
As is evident from Table 2, using the Cox proportional hazard model, leukocyte count when considered as either a continuous or categoric variable (using cutoff levels of 15 × 109/L for PV and either15 × 109/L or 9.4 × 109/L for ET)7, 14 was not found to be significantly associated with either arterial or venous thrombosis. TFS curves are shown in Figure 1.
Table 2. Cox Proportional Analysis of Time to Arterial and Venous Thrombosis in Low-Risk Patients With Polycythemia Vera and Essential Thrombocythemia
We performed a subgroup analysis of 197 ET patients with platelet counts of <1000 × 109/L and found no association between thrombosis and leukocyte count using cutoff levels of both 15 × 109/L or 9.4 × 109/L (P = .21 and P = .31, respectively). Similarly, in a subgroup analysis of 137 PV patients with platelet counts of <1000 × 109/L, we found no association between thrombosis and leukocyte count using cutoff level of 15 × 109/L (P = .17). This particular subanalysis was performed because of the observations from a recent ET study from Italy, which suggested that the association between leukocytosis and thrombosis was more apparent in the absence of extreme thrombocytosis.17 We further analyzed 60 ET and 66 PV patients with platelet counts <1000 × 109/L and an absence of cardiovascular risk factors and found that leukocyte count, using cutoff levels of 15 × 109/L or 9.4 × 109/L for ET and 15 × 109/L for PV, was not significantly associated with thrombosis (P = .54, P = .47, and P = .26, respectively).
With regard to other disease parameters obtained at time of diagnosis, only advanced age was found to be significantly associated with arterial thrombosis in PV patients and higher hemoglobin level with venous thrombosis in patients with ET. Among 107 ET patients tested for the JAK2V617F mutation, 47 patients were positive. On univariate analysis, the presence of the JAK2V617F mutation was significantly associated with arterial thrombosis in ET (P = .049) but significance was lost during multivariate analysis that included age as a covariate.
The primary objective of the current study was to clarify whether leukocytosis at the time of diagnosis, at previously defined leukocyte cutoff levels (15 or 9.4 × 109/L),2, 3, 11, 14, 17 is predictive of thrombosis in low-risk patients with ET or PV. We focused on low-risk patients because of the potential impact of such an association on current practice that dictates cytoreductive therapy for high-risk but not for low-risk patients.18 The results from the current study, derived from a large study population of 407 patients, revealed that leukocytosis at diagnosis did not predict either arterial or venous thrombosis during follow-up (Fig. 1). Because a recent study17 suggested that the association between leukocytosis and thrombosis in ET might be affected by the presence or absence of extreme thrombocytosis, we separately analyzed patients with a platelet count of <1000 × 109/L and again found no correlation between leukocytosis and thrombosis; the same results were obtained when analysis was restricted to patients without either extreme thrombocytosis or cardiovascular risk factors (data not shown). These latter analyses helped refute the possibility that the results from the current study might have been influenced by a confounding effect from treatment introduction before a vascular event; most physicians do not prescribe cytoreductive therapy to low-risk PV or ET patients in the absence of both extreme thrombocytosis and cardiovascular risk factors.
The current study should not be directly compared with other studies dealing with the same subject matter because of interstudy differences in the constitution of the study population as well as methods of analysis. For example, a PV study by Landolfi et al11 compared patients with leukocyte counts >15 × 109/L with those with leukocyte counts ≤10 × 109/L when demonstrating a higher risk of myocardial infarction in the former group. Similarly, in a more recent study by Carobbio et al,17 the authors found significant differences in total thromboses when patients with leukocyte counts of <8 × 109/L were compared with those whose leukocyte counts were >11 × 109/L. Furthermore, the leukocyte count that was considered in the former study11 was obtained at time of database registry and not necessarily at time of diagnosis. It should also be noted that previous studies in ET or PV included all risk categories and leukocytosis was found to be correlated with either thrombosis at diagnosis13 or total thrombotic events during follow-up.2, 5 At the same time, a most recent study of 605 patients with ET did not find leukocytosis as an independent risk factor for thrombosis.19 Taken together, these observations underscore the need for a properly designed, prospective study if one wishes to treat leukocytosis in ET or PV patients as a formal risk factor for thrombosis and modify current practice accordingly.
The secondary objective of the current study was to identify other predictors of thrombosis in the same study population. In general, our observations regarding other parameters of significance for thrombosis in patients with PV and ET are somewhat similar to those from previous reports. For example, we have repeatedly shown the lack of a significant and independent association between the presence of JAK2V617F and thrombosis in patients with ET.20 Others have reported either similar or contrary findings.14, 21-24 However, in most of the latter instances, it is unclear whether the reported associations would have been sustained during a multivariate analysis that included covariates that are known to be significantly associated with both JAK2V617F and thrombosis, including advanced age and a higher hemoglobin level.22, 24, 25 For example, in what to our knowledge is the largest, to date, reported study in ET patients (n = 776),22 the presence of JAK2V617F was associated with venous but not arterial thrombosis. In contrast, the current study revealed a significant and independent effect of increased hemoglobin, but not JAK2V617F, on venous thrombosis in patients with ET. Finally, our observation regarding the association between arterial thrombosis and advanced age in patients with PV is in keeping with that of many other previous studies that have consistently demonstrated a strong association between advanced age and thrombosis in patients with PV,11 as well as in those with ET.3 Regardless, this particular observation is not expected to impact current practice because advanced age is already considered to be a treatment-relevant risk factor in both ET and PV.18 Finally, at least in PV, the presence of the recently described TET2 mutation does not appear to influence the risk of thrombosis, but additional studies are necessary to clarify the clinical relevance of mutant TET2 in PV, ET, and other myeloid neoplasms.26, 27