Internal and Emergency Medicine

, Volume 7, Issue 6, pp 509–515

Polycythemia vera: gender-related phenotypic differences


    • Institute of Internal Medicine and Geriatrics, Haemostasis Research CenterCatholic University School of Medicine
  • Leonardo Di Gennaro
    • Institute of Internal Medicine and Geriatrics, Haemostasis Research CenterCatholic University School of Medicine
  • Maria Anna Nicolazzi
    • Institute of Internal Medicine and Geriatrics, Haemostasis Research CenterCatholic University School of Medicine
  • Igor Giarretta
    • Institute of Internal Medicine and Geriatrics, Haemostasis Research CenterCatholic University School of Medicine
  • RosaMaria Marfisi
    • Laboratory of Clinical Epidemiology of Cardiovascular Disease
  • Roberto Marchioli
    • Laboratory of Clinical Epidemiology of Cardiovascular Disease

DOI: 10.1007/s11739-011-0634-3

Cite this article as:
Landolfi, R., Di Gennaro, L., Nicolazzi, M.A. et al. Intern Emerg Med (2012) 7: 509. doi:10.1007/s11739-011-0634-3


In polycythemia vera, gender has recently been shown to influence the JAK2V617F allele burden, but its effect on the disease phenotype is unknown. This issue was investigated using the database of the European Collaboration on Low-dose Aspirin in Polycythemia Vera (ECLAP) Study. The ECLAP Study recruited 1,638 polycythemic subjects and followed for 2.7 ± 1.3 years. At study entry, men, compared to women, had a higher prevalence of myocardial infarction (11.3 vs. 5.8%; P < 0.0001) and peripheral arterial disease (6.1 vs. 2.9%; P < 0.05) while a history of venous thrombosis was more common in women (11.4 vs. 7.9%, P = 0.016). Among 234 venous thrombosis, there were 39 splanchnic vein thromboses (33 extra-hepatic portal vein thromboses and 6 Budd-Chiari syndromes). Most of these events occurred as an early disease presentation in young female subjects. Women, compared to men, had higher platelet counts (average value 430 ± 213 vs. 375 ± 201 × 109/L; P < 0.0001) and lower hematocrits (0.46 ± 0.06 vs. 0.48 ± 0.06 l/l; P < 0.0001). Cholesterol plasma level, available in 995 subjects (61%), was lower in male patients (180.8 ± 43.1vs. 196 ± 46.6 mg/dl; P < 0.0001). During follow-up there were 205 major thromboses confirming an high incidence of myocardial infarction in men although not statistically significant (1.2 vs. 0.6 cases per 100 person-years; P > 0.05). These data show several gender-related differences both in the thrombotic diathesis and in the prevalence of vascular risk factors of PV patients.


Myeloproliferative neoplasmsPolycythemia veraThrombosisGender


Polycythemia vera (PV) is a myeloproliferative neoplasm characterized by a thrombophilic state that manifests with microcirculatory disturbances and arterial and venous thromboses [1].

In the past three decades, the pathogenesis of thrombosis was investigated by experimental and epidemiological studies that extensively explored the relationship between clinical manifestations, haemostatic abnormalities and several disease-related variables including blood count and JAK2 status [2, 3]. While the pathogenesis of thrombophilia still remains largely elusive, novel information has been acquired as to the epidemiology of vascular events, on the role of low-dose aspirin in the prevention of these events, as well as on the significance of disease-related parameters, namely leukocyte count in influencing the thrombotic risk [4, 5]. As to common risk factors, only smoking is found to influence the vascular risk while the role of gender, diabetes, blood cholesterol and hypertension is still debated. This lack of knowledge derives from the small size of most clinical studies performed in this disease.

Thus, the aim of this study was to explore the large database of the European Collaboration of Low-dose Aspirin in Polycythemia Vera (ECLAP) for further investigating these issues with particular attention to the role of gender. This variable has been shown to influence the clinical features of a number of diseases including cardiovascular disorders, solid neoplasms, and hematological malignancies [610]. Some previous reports indicate that in myeloproliferative neoplasms, gender may influence the incidence of thrombotic manifestations such as erythromelalgia and splanchnic vein thrombosis that are most peculiar to these disorders [11, 12]. Finally, the interest in exploring this issue was further raised by the recent report of an influence of gender on JAK2V617F allele burden in myeloproliferative neoplasms [13].

Patients and methods

We reviewed the electronic medical records of 1,638 PV patients collected within the European Collaboration on Low-Dose Aspirin in Polycythemia Vera (ECLAP) trial. The Italian coordinating center received all data from several European hematological centers, and created the largest centralized data base existing on PV [14]. In all these patients, the diagnosis of PV was made on the basis of history, physical examination and laboratory testing using previously described criteria [14].

Treatment strategies had to comply with the recommendation of maintaining the hematocrit below 45% and platelet count less than 400 × 109/L.

Data regarding clinical outcomes, treatments, and laboratory values were recorded at baseline and during the follow-up visits at 12, 24, 36, 48, and 60 months.

Information was collected on major vascular events that consisted of arterial thromboses and venous thromboembolism (VTE), and of minor occlusive events, such as erythromelalgia and superficial thrombophlebitis of the extremities. Arterial thromboses included cardiovascular death, stroke, transient ischemic attack (TIA), acute myocardial infarction (AMI) and peripheral arterial thrombosis (PAT) of the legs and abdominal vessels, while VTE included pulmonary embolism and deep venous thrombosis of the limb and splanchnic veins, according to the definition of vascular events previously described [14]. Deaths from congestive heart failure and intracranial bleeding were not included in cardiovascular deaths.

Finally, cardiovascular risk factors such as smoking, hypertension, total blood cholesterolemia and diabetes were considered. Total cholesterol was measured at study entry, and expressed in milligrams per deciliter of blood (mg/dl). Cholesterol levels were classified by different levels according to the National Cholesterol Education Program (NCEP) guidelines for detection of high cholesterol [15]. Spleen size was derived from ultrasound scan measurements, and spleen longitudinal diameter greater than 12 cm was taken as index of splenomegaly.

Statistical analysis

Data were expressed as mean ± SD. The Shapiro–Wilk test was applied to check the normality of the variables. Differences in the mean of variables were tested using both parametric and nonparametric tests depending on the distribution of the variables. Differences in proportions were ascertained by χ2 tests. Data were explored using patients’ characteristics at baseline and crude incidence rates were calculated as number of main vascular events × 100 person-years. Some covariates (age at recruitment, time from diagnosis to enrollment, thrombotic events prior to recruitment, history of hypertension, intermittent claudication, or erythromelalgia) were determined only at baseline, whereas others were updated during follow-up (smoke, diabetes, total blood cholesterol, splenomegaly, immature cells, hematocrit, platelets and leukocyte count, therapeutical interventions such as antiplatelets and anticoagulants).

All P values are two-tailed (<0.05). Analyses were performed using the SAS statistical package (SAS Institute Inc, Cary, NC).


The ECLAP population consisted of a total of 1,638 patients (42.5% females). The mean duration of the disease at study entry was 5.0 ± 5.0 years, while the duration of the follow-up was 2.7 ± 1.3 years.

Patients’ characteristics at recruitment

The main patient characteristics at recruitment according to gender are summarized in Table 1. At study entry, male and female subjects had comparable lengths of disease (4.9 ± 5.1 years), but the men were significantly younger (64 ± 12.8 vs. 68 ± 12.1 years; P < 0.0001), had higher hematocrit values (0.48 ± 0.063 vs. 0.46 ± 0.063 l/l; P < 0.0001), lower platelet counts (375 ± 200.8 vs. 430 ± 212.9 × 109/L; P < 0.0001), and more frequently had splenomegaly (47 vs. 42%; P < 0.05). Moreover, men showed a higher prevalence of intermittent claudication as well as peripheral arterial thrombosis of lower extremities, possibly as a consequence of a higher prevalence of a smoking habit (16.9 vs. 7.2%; P < 0.0001). Also, a history of microcirculatory disturbances such as erythromelalgia was more common in women (6.5 vs. 4.4%; P < 0.05).
Table 1

Baseline characteristics of 1,638 PV patients according to gender

Clinical and laboratory parameters


696 (42.5%)


942 (57.5%)

Age at diagnosis of PV#

62.6 (12.9)

58.8 (13.2)

Age at recruitment#

67.7 (12.1)

63.7 (12.8)

Prior thrombosis

255 (36.6)

332 (35.2)

 Arterial thrombosis

190 (27.3)

279 (29.6)

 Venous thrombosis*

79 (11.4)

74 (7.9)


40 (5.8)

106 (11.3)


127 (18.3)

157 (16.7)


37 (5.3)

52 (5.5)

Prior bleeding

59 (8.5)

74 (7.9)


45 (6.5)

41 (4.4)

Intermittent claudication*

20 (2.9)

57 (6.1)


50 (7.2)

159 (16.9)


314 (45.1)

333 (35.4)

Diabetes mellitus

43 (6.2)

74 (7.9)


291 (41.8)

441 (46.8)

Immature cells

33 (4.7)

51 (5.4)

Packed cell volume (l/l)#

0.461 (0.063)

0.479 (0.063)


252 (42.0)

225 (27.1)


189 (31.5)

283 (34.1)


159 (26.5)

323 (38.9)

WBC count (×109/L)

10.8 (8.7)

10.9 (8.6)


429 (63.3)

561 (61.1)


149 (22.0)

216 (23.5)


100 (14.8)

141 (15.4)

Platelet count (×109/L)#

430.2 (212.9)

375.1 (200.8)


173 (25.2)

365 (39.1)


258 (37.5)

285 (30.6)


257 (37.4)

283 (30.3)


 Phlebotomy alone#

155 (22.3)

310 (32.9)


448 (64.4)

521 (55.3)


24 (3.5)

20 (2.1)


368 (52.9)

425 (45.1)


28 (4.0)

33 (3.5)


3 (0.4)

2 (0.2)


38 (5.5)

68 (7.2)


23 (3.3)

41 (4.4)


404 (58.1)

551 (58.5)


46 (6.6)

64 (6.8)

For continuous variables, values are mean (SD)

For categorical variables, percentage values are given in parenthesis

AMI acute myocardial infarction, P phosphorus, PAT peripheral arterial thrombosis, TIA transitory ischemic attack, WBC white blood cells

P < 0.05, P < 0.0001

The most prevalent cardiovascular risk factors at study entry were arterial hypertension (39.5%) that was more frequent in women (45.1 vs. 35.4%; P < 0.0001) and smoking (12.8%), while hypercholesterolemia was reported in only 3.5% of these patients. On the contrary, hypocholesterolemia was frequently observed and among 995 PV patients in whom cholesterol values were available, men (57.6%) had a significantly lower cholesterol mean level (181 ± 43 vs. 196 ± 47 mg/dl; P < 0.0001).

As reported in Table 1, a history of previous thrombosis was present in 587 PV patients (35.8%) with a significantly higher prevalence of myocardial infarction in men (11.3 vs. 5.8%; P < 0.0001) and of venous thrombosis in women (11.4 vs 7.9%; P < 0.05).

Among 234 venous thromboses, 39 (16.7%) were splanchnic thromboses (SVT) and consisted of 33 extra-hepatic portal thromboses and 6 Budd-Chiari syndromes. In 16 patients, SVT was an early and presenting vascular event, while in 15 other cases these events preceded PV diagnosis by 1–5 years. SVT occurred after diagnosis in only six patients, and in only one other case after another vascular event. Most typically, SVT occurred in young female subjects, and 5 out of 6 Budd Chiari syndromes were observed in women.

SVT caused chronic severe portal hypertension with ascites in 40% of 37 subjects and two female patients with previous Budd-Chiari syndrome had undergone an orthotopic liver transplantation successfully.

Finally, as reported in Table 1, the use of antiplatelet and anticoagulant therapy was similar in male and female patients, phlebotomy was more frequently used in men (33 vs. 22%; P < 0.0001), while chemotherapy (mostly hydroxyurea) was more commonly administered to women (64 vs. 55%; P = 0.0002).

Clinical and laboratory data during follow-up

At the time of study closure, the mean follow-up was 2.7 ± 1.3 years (4,393 person-years).

Follow-up controls were made at 12, 24, 36, 48 and 60 months and confirmed higher values of hematocrits (0.46 ± 0.04 vs. 0.44 ± 0.04 l/l; P < 0.0001) and lower values of platelet counts (349 ± 163 vs. 399 ± 162 × 109/L; P < 0.0001) in male patients (Table 2). No significant gender-related differences were found in other laboratory hematological parameters.
Table 2

Packed cell volume and platelet count at baseline and follow-up in PV patients according to gender




Packed cell volume (l/l)


0.48 (0.06)

0.46 (0.06)

 12 months#

0.46 (0.04)

0.44 (0.04)

 24 months*

0.46 (0.04)

0.45 (0.04)

 36 months

0.45 (0.06)

0.44 (0.06)

 48 months

0.44 (0.05)

0.44 (0.05)

 60 months



Platelet count (×109/L)


375 (201)

430 (213)

 12 months#

349 (163)

399 (162)

 24 months#

359 (224)

380 (183)

 36 months

362 (243)

371 (187)

 48 months

384 (307)

388 (154)

 60 months



Values are mean (SD)

NA not available

P < 0.05, P < 0.0001

Moreover, no statistically significant differences were observed in crude incidence rate of overall major and venous thrombosis, although arterial thromboses, particularly myocardial infarction, were more frequently observed in male than in female subjects (Fig. 1). During follow-up there were 164 deaths (10% of all patients), 72 cardiovascular and 90 nonvascular with no significant differences between genders. Other characteristics of ECLAP population had been described in detail elsewhere [5, 14].
Fig. 1

Incidence rate of the main outcome events during follow-up (×100 person-years). The number inside each column refers to the percentage of patients with event. In all cases P > 0.05


Estimating the vascular risk of the individual patient is critical for defining the optimal treatment strategy in polycythemia vera. Vascular risk stratification generally relies on age and past thrombotic history, while the role of leukocyte count remains debated [15]. The weight attributed to other vascular risk factors such as smoking, diabetes, high cholesterol is quite variable, due to the lack of clinical studies adequately sized to assess the relative contribution of these variables. Also, gender has an uncertain influence on thrombophilia in PV patients, while it is known to affect the incidence of vascular disease in the general population, and to have a significant impact on the phenotype of some hematological and nonhematological neoplasms [610].

This study first investigated such gender-related aspects, and provided evidence of an influence on the disease phenotype. Our data were gathered from the ECLAP prospective observational study that recruited 1,638 patients with a variable history of disease [14]. While the disease duration was similar in both genders, women were significantly older at recruitment, which reflected the fact that the disease had been diagnosed at an older age, in accordance with previous reports [16]. Whether the disease develops or manifests later in women is uncertain. A role may be played by several factors that include the relative iron deficiency in women, hormonal influences, as well as the lower hematocrit normal values of female subjects. The latter could help the blood hyperviscosity in women to escape medical attention and investigation.

Data gathered at recruitment showed that women had more frequently received chemotherapy, likely because they were older and had a higher platelet count. This finding may partially be a consequence of the reduced iron stores in female subjects, although we cannot provide data to support this hypothesis.

In our analysis at study entry, a history of thrombotic manifestations was also different in the two genders with microcirculatory disturbances and venous thromboses being more frequent in women, while arterial thrombotic events and particularly myocardial infarction were more frequent in men.

Although the difference in the history of thrombosis registered at study entry could reflect the differences in well-recognized risk factors such as smoking or the different treatment strategy (as reported in Table 1), our analysis seems to suggest a role of phenotype in PV thrombophilia.

Indeed, splanchnic vein thrombosis is often observed in women as an early vascular complication, according to previous observational studies [17, 18]. Moreover, our analysis confirms that Budd-Chiari syndrome also typically occurs in female patients.

The current analysis provides a prevalence of SVT in a large cohort of well-characterized polycythemic patients in which SVT constitutes 4 and 17% of all the vascular thromboses and all deep venous thrombosis, respectively. Previous studies provide the prevalence of SVT in much more limited and heterogeneous patient populations also including essential thrombocythemia and primary myelofibrosis [1921]. This study also confirms the association between Budd Chiari syndrome and female gender, and between SVT and younger age. Furthermore, SVT is the presenting complication of PV, and this complication preceded the disease recognition. In fact, in 43% of the patients, SVT is the presenting complication of PV and in almost the same percentage; PV is diagnosed years later as the abdominal vascular event. These data are similar to those reported in the study of SVT in essential thrombocythemia by Tefferi et al. [20].

Recently, a prospective analysis of 165 consecutive PV patients shows gender differences in the JAK2 mutation allele burden that was significantly higher in male patients with a lower prevalence of homozygous JAK2 clones, and less clonal expansion in women [13]. In our particular data setting, the more frequent use of cytoreductive agents observed in female patients seems to reduce the load of JAK2mutated hematopoiesis, and has a negative effect on JAK2 activation; however, the retrospective design of our study only can suggest an hypothetical association, and the data reported in literature are still conflicting [22].

Unfortunately, in our study, we have no data about JAK2mutant allele burden, so we can only hypothesize that the different JAK2V617F allele burden may play a role in some of the observed differences between genders.

Moreover, although the role of JAK2V617F allele burden in the development of vascular events in PV patients accounts for conflicting results [23, 24], some studies suggest its contribution to the hypercoagulable state, and probably, to thrombotic risk [25]. In a recent study, Barbui et al. [26] analyze the impact of JAK2 mutation allele values on the incidence of thrombosis in 415 PV patients, and show a progressive increase of arterial vascular events according to the amount of allele burden. On the other hand, a relatively low JAK2 mutation allele burden has also been reported for patients with splanchnic vein thromboses [27].

Moreover, this study has intrinsic limitations and pitfalls due to retrospective analysis. Therefore, we cannot know the weight of prothrombotic conditions such as the cytokines pattern and thrombophilic factors in the different prevalence of thrombosis in our population. Especially in female subjects, the levels of protein C, protein S and homocysteine might be helpful in explaining the increased number of SVT, and to contribute to the major difference among men and women in the arterial thrombotic events [28].

Finally, among the various risk factors, there is a surprisingly high prevalence of smoking habit among male PV subjects, and this should call attention to the need for more effective life style interventions in these high risk subjects. Diabetes is relatively uncommon, while hypertension affects about 45% of female and 35% of male subjects. In our analyses, we also focused our attention on blood cholesterol that was available in 995 of 1,638 subjects.

Hypocholesterolemias present in a significant percentage of our PV patients in accordance with what is reported in previous studies [29, 30], and cholesterol mean levels are significantly higher in women.

In conclusion, the ECLAP observational study allowed us to focus several gender differences in the PV phenotype of a large PV population, and these differences seem particularly interesting due to the recent observation of a different JAK2mutant allele burden in male and females PV patients.


The contribution of TizianoBarbui, Raimondo De Cristofaro and Luca Miele to data interpretation and discussion is gratefully acknowledged.

Conflict of interest


Copyright information

© SIMI 2011