Activated haemostasis and JAK2V617F
In the era of a better molecular understanding of MPDs, much attention has focused on whether the presence of JAK2V617F has any bearing on thrombotic risk. Several hypothetical mechanisms exist. It is now widely recognised that the JAK2 phenotype of ET results in a higher haemoglobin and higher white cell count, factors which both increase the risk of thrombosis through concepts of rheology and platelet activation. However, there may be more specific actions. It has recently been observed that JAK2V617F may modify red cell adhesion molecules and promote increased adhesiveness favouring thrombosis (Wautier et al, 2007). The direct effect of the JAK2-activating mutation may be involved in impaired expression of cMPL signal transduction for TPO-induced platelet priming, leading to chronic platelet hyper-responsiveness (Kubota et al, 2004). JAK2 may also affect platelet activation by modifying cMPL cell surface localisation and stability (Royer et al, 2005). Abnormal expression of bcl-x, an inhibitor of apoptosis, and overexpression of certain genes (CD177 and transcription factor NF-E2) are also thought to be due to activation of the JAK/STAT pathway, a consequence of the JAK2 mutation (Kralovics et al, 2005b) and may indirectly affect coagulation.
Recently, Robertson et al (2007) reported significantly elevated soluble p- selectin levels in MPD patients with the JAK2 mutation compared to wild type. P-selectin mediates platelet binding to leucocytes, which induces highly procoagulant microparticles and platelet-leucocyte aggregates that promote thrombosis. Similar results have been observed using a whole blood flow cytometric method (as opposed to enzyme-linked immunosorbent assay, ELISA), which revealed significantly higher p-selectin expression both at baseline (P = 0·006) and following arachidonic acid activation (P = 0·03) for JAK2V617F positive ET cases versus the wild type (Arellano-Rodrigo et al, 2006). A separate study, with a slightly larger patient group, evaluated platelet p-selectin (CD62p) also by flow cytometry and reported that whilst they observed an over-expression of p-selectin in platelets from ET patients, they did not find any difference in CD62p expression between the JAK2 mutation carriers compared to the JAK2 wild type subjects (Falanga et al, 2007).
However, this group did identify a number of other haemostatic variables that appeared to be associated with the JAK2V617F mutation. One of these was platelet membrane-bound tissue factor (TF). Compared to JAK2 wild type cases, JAK2V617F positive cases had significantly (P < 0·01) higher levels of TF-positive platelets by flow cytometric analysis. This was associated with a concomitant increase in the total TF antigen content in washed platelets, as measured by ELISA. The increased platelet surface TF in the JAK2 mutation carriers remained significant even after multivariate analysis taking age, sex and treatment into account. The exact origin of platelet expressed tissue factor is unresolved and may be from contact with leucocytes or microparticles or may have been formed in the platelets themselves (Siddiqui et al, 2002; Camera et al, 2003). Nevertheless, the overriding point is that the JAK2V617F mutation appears to have an increased presence of aberrant TF, a definitive feature of a prothrombotic phenotype.
Falanga et al (2007) also confirmed previous findings of increased levels of circulating platelet/leucocyte aggregates in ET patients, but in addition demonstrated that these aggregates are significantly greater in JAK2V617F mutation carriers compared to the wild-type cases of ET. The investigators also observed significantly increased surface expression of neutrophil CD14 and leucocyte alkaline phosphatase (LAP) in JAK2V617Fversus wild type cases. A Similar study in 49 ET patients also found higher levels of platelet–neutrophil and platelet monocyte complexes in JAK2 mutation carriers compared to wild type counterparts, but in this instance the difference did not reach statistical significance (Arellano-Rodrigo et al, 2006). Similar parameters have been studied in patients with primary myelofibrosis (PMF), where although platelet-leucocyte aggregates were increased regardless of JAK2 phenotype, there was no difference regarding JAK2 status (Alvarez-Larran et al, 2008). However, these investigators did show that compared to the JAK2 wild type, patients with the JAK2V617F mutation had significantly higher monocyte and neutrophil CD11b expression, both at baseline and following activation. Here, as is observed in many MPD studies, mutation positive patients also had higher leucocyte counts, but in this study the analysis failed to demonstrate that the higher CD11b was independent of the leucocyte count. The authors speculated that the increased CD11b expression would enhance neutrophil adhesion to megakaryocytes, and facilitate emperipolesis, a phenomenon often implicated in myelofibrotic development.
Other markers of haemostatic activation, including prothrombin fragment 1 + 2 (F1 + 2) and thrombin-antithrombin (TAT) complexes have previously been shown to be increased in MPD compared to controls (Falanga et al, 2000). Recently investigators have addressed these (and other markers) in relation to JAK2 status. Robertson et al (2007) did not find any association of hypercoagulation markers TAT and F1 + 2 or D-dimer with JAK2 status, or even the presence of MPD, but their control group were all patients with uncontrolled hypertension and at increased risk of vascular events. Conversely, in 75 ET patients, F1 + 2, TAT, D-dimer, plasminogen activator inhibitor (PAI-1), tissue plasminogen activator (t-PA), thrombomodulin (TM) and neutrophil elastase were analysed (Falanga et al, 2007). All haemostatic markers were higher in the ET patients, with only TAT complex levels not reaching significance. However, with regards to the JAK2 mutation, only soluble TM levels were significantly higher when compared to wild type ET patients (P = 0·020). In PMF, D-dimer, F1 + 2 and TM have been investigated and high levels of the latter two measures have been reported in association with the disease, but in a subgroup of JAK2V617F positive patients, only F1 + 2 levels were observed to be significantly higher compared to wild type PMF patients (Alvarez-Larran et al, 2008).
Despite the variable conclusions with regards to which disordered haemostatic parameter is associated with JAK2V617F, this should not overshadow the overwhelming evidence supporting the concept of increased neutrophil and platelet activation in ET. Further, there is growing evidence to support the suggestion that the JAK2V617F phenotype displays a greater degree of spontaneous platelet and leucocyte activation compared to the wild type. Large prospective studies are needed to address if the JAK2V617F mutation is indeed related to increased activation of platelets and leucocytes and that these parameters play a vital role in the pathophysiology of thrombosis in MPD.
JAK2V617F status and thrombotic risk – clinical evidence
Fourteen published studies have examined the interaction between JAK2V617F mutation status and thrombotic risk in patients with MPD (Table II). The largest prospective study to address the issue followed 776 patients with ET enrolled into the MRC PT1 trial, and two smaller studies (Campbell et al, 2005). 414 patients were V617F-positive (53%), and these were found to have a significantly higher risk of venous thromboembolism (VTE) in the year prior to diagnosis (11 vs. 2 events, P = 0·04) and a non-significant trend towards VTE after trial entry (12 vs. 4 events, P = 0·06) compared to V617F-negative patients. There was no difference in the rate of arterial events. V617F-positive patients were also significantly older than V617F-negative ones, and had significantly higher haemoglobin concentrations and neutrophil counts. Analysis of response to treatment showed that amongst V617F-positive patients, those randomised to receive anagrelide had a significantly higher arterial thrombotic rate than those who received hydroxycarbamide (19 vs. 5 events, P = 0·03), but this difference was not seen in V617F-negative patients. With respect to VTE, there was no apparent interplay between mutation status, treatment, and thrombotic risk.
Table II. Studies addressing the association between JAK2V617F and thrombosis.
|Study||n‡||Patient group||%JAK2 V617F+|| Median f/up (months)||Association between JAK2 V671F and thrombosis on univariate analysis?||Details||Potential confounding factors associated with JAK2 V617F status||Independent association between JAK2 V617F and thrombosis on multivariate analysis?||Comments|
|Kralovics et al, 2005a||244||128 PV, 93 ET, 23 IMF||48|| ||Yes||Higher overall rate of thrombosis in V617F+ group (26% vs. 15%)*||Age, disease duration||ND||No separate analysis for ET subgroup|
|Wolanskyj et al, 2005||150||ET||49||137||No||No difference in rate of thrombosis at any time (45% vs. 38%) or after diagnosis (33% vs. 29%).||Age Hb, WCC||No||Not powered to detect small differences in thrombotic rate (eg. 5% and 15%)|
|Campbell et al, 2005||776||ET||53|| ||Yes||Higher rate of VTE prior to diagnosis in V617F+ group (11 vs. 2 events)*. Trend towards higher rate of VTE after diagnosis (12 vs. 4 events). No difference in arterial events.||Age, Hb, neutrophil count||ND||Patients mostly derived from MRC PT-1 trial|
|Cheung et al, 2006|| 60||ET||48|| ||Yes||Higher overall rate of thrombosis in V617F+ group (62% vs. 26%)†.||Hb||ND|| |
|Tefferi et al, 2006|| 63||PV||92|| 33||No||No difference in overall rate of thrombosis.|| ||No|| |
|Heller et al, 2006|| 50||ET||48|| 79||Yes||Higher overall rate of thrombosis in V617F+ group (11 vs. 1 events)†||Age, Hb||Yes||Relatively young cohort (78%≤ 60 years). Majority of thrombotic events occurred prior to or at diagnosis of ET|
|Finazzi et al, 2007||179||ET||58||∼66||Yes||Higher overall rate of thrombosis in V617F+ group (33% vs. 17%)*||Hb, WCC||ND||77 further patients had PV, in whom the thrombotic rate was 43% (not significantly different to V617F+ ET group)|
|Ohyashiki et al, 2007c|| 49||ET||63|| ||Yes||Higher overall rate of thrombosis in V617F+ group (29% vs. 6%)||Hb, WCC||ND||Also analysed 34 PV patients: no association between homo- or heterozygosity for JAK2 V617F and thrombotic risk|
|Carobbio et al, 2007||277||ET|| 55|| ||No||No difference in postdiagnosis rate of thrombosis (hazard ratio = 1·4, 95% CI 0·7 to 3·0)||Hb, WCC||No|| |
|Kittur et al, 2007||176||ET|| 55|| 59||Yes||Higher rate of VTE after diagnosis in V617F+ group (11% vs. 3%)*. No difference in prediagnosis VTE (11% vs. 6%), or arterial events at any time (38% vs. 29%).||Age, Hb, WCC||Yes||Significance on multivariable analysis was lost if history of prior thrombosis was included.|
|Alvarez-Larran et al, 2007||103||ET < 40 years old|| 43||120||No||No difference in postdiagnosis rate of thrombosis (9 vs. 13 events).||Hb||No||Major thrombotic events only included|
|Vannucchi et al, 2007b||173||PV||100|| 24||Yes||Patients with a V617F-allele burden >75% had a higher risk of thrombosis at diagnosis and during follow-up combined (P = 0·004), and at diagnosis alone (P = 0·003), compared with those whose allele burden was <25%.||Haematocrit, WCC||Yes|| |
|Vannucchi et al, 2007a||962||323 PV, 639 ET|| 73|| 60||Yes||Increased rate of arterial and venous thrombosis in V617F-homozgygous ET patients (n = 14) compared with all other ET patients, at diagnosis and during follow-up*.||Age, Hct||Yes||See text|
|Hsiao et al, 2007||53||ET|| 66|| 35||Yes||Higher overall rate of thrombosis in V617F+ group (43% vs. 11%)*||Hb, WCC||ND||Those with thrombosis were significantly older and had significantly higher WCC than those without|
Vannucchi et al (2007a) retrospectively analysed 962 patients with chronic MPD (323 PV, 639 ET), subdivided into three groups with respect to their JAK2V617F status: wild-type, heterozygous (percentage of mutant allele in granulocytic cells <50%), and homozygous (mutant allele >50%). All the patients with PV were V617F-positive (68% heterozygous), as were 60% of the ET patients (of these, 96% were heterozygous). Over a median follow-up of 5·0 years, a significant excess of thrombotic events was noted in the homozygous patients with ET compared with their heterozygous and wild-type counterparts, and indeed compared with the PV patients. The significant difference in overall thrombotic rates was observed both at the time of diagnosis, largely due to a marked excess of venous events in the homozygous group, and during follow-up, when both arterial and venous events were more common in the homozygous ET group.
Compared with wild-type patients, JAK2V617F homozygosity was independently associated with the occurrence of cardiovascular events in a multivariate analysis that incorporated established risk factors for thrombosis in ET (age, history of prior thrombosis, leucocytosis, and sex), though it should be emphasised that this group only represented 2·2% of the total patient cohort. More generally, the heterozygous and homozygous ET patients, when considered together, did have a significantly higher incidence of thrombosis at diagnosis (but not during follow up), compared with their wild-type counterparts. They were, however, also older and exhibited a higher haematocrit, than the wild-type patients, which may have confounded this comparison (Vannucchi et al, 2007a). In a retrospective study of 176 patients with ET, Kittur et al (2007) also observed an increased rate of venous thrombotic events in the 96 patients (55%) who were positive for the JAK2V617F mutation (11% vs. 3%, P = 0·02). This association was, however, confined to venous events after diagnosis. Again, a positive correlation between JAK2V617F mutation status and haemoglobin, age and white cell count was noted.
Six further studies suggested the possibility of an association between JAK2 mutation status and overall thrombotic risk – i.e. combined venous and arterial events, both before and after diagnosis (Kralovics et al, 2005a; Zhao et al, 2005; Cheung et al, 2006; Heller et al, 2006; Finazzi et al, 2007; Hsiao et al, 2007; Ohyashiki et al, 2007b; Vannucchi et al, 2007b). Kralovics et al (2005a) evaluated 244 patients with MPD (128 with PV, 93 with ET, and 23 with myelofibrosis), of whom 117 (48%) had the JAK2V617F mutation. The rate of thrombotic complications was significantly higher in this group than the wild-type group (26% vs. 15%, P = 0·03), even though a significantly greater proportion of the V617F-positive group had received cytoreductive therapy. However, they also noted that patients with the mutation had significantly longer disease duration than those without, and were significantly older at the time of sampling. Both of these factors could have contributed to the increased thrombotic rate in this patient group, and so the effect of the JAK2 mutation alone could not be isolated.
Indeed, potentially confounding factors were found in all eight of the studies in which an association between JAK2 mutation status and thrombosis was observed. In seven studies, the JAK2V617F-positive subgroup had significantly higher haemoglobin concentrations at diagnosis; in five, a significantly higher white cell count or neutrophil count was noted at diagnosis; and in four, patients with the mutation were found to be significantly older than those without. The association between JAK2V617F mutation status and leucocyte count is especially noteworthy, since a number of groups have found that a relative leucocytosis (white cell count above the population median) at diagnosis is a strong, independent predictor of thrombotic complications, especially in otherwise ‘low-risk’ patients (Wolanskyj et al, 2006; Carobbio et al, 2007).
By contrast, a number of studies have failed to demonstrate any relationship between JAK2 mutation status and vascular outcome. Wolanskyj et al (2005) followed up 150 patients with ET: 73 (49%) were V617F-positive, and these did not show a higher incidence of thrombosis (combined arterial and venous) than their V617F-negative counterparts, despite a significantly higher haemoglobin concentration, leucocyte count and median age. The former group had an excess of deaths during the median follow-up of 11·4 years (21 vs. 11 deaths, P = 0·03), with a non-significant trend to inferior survival, but this was accounted for by the significantly higher median age. The authors concede that the study was not sufficiently powered to detect relatively small differences in complication rate (e.g. between 5% and 15%). Antonioli et al (2005) similarly found that, of 130 patients with ET, the V617F-positive group (n = 74) did not appear to have a significantly higher risk of thrombosis than the V617F-negative group (17 vs. 13 events). Again, venous and arterial events were grouped together.
Retrospective analysis of the large Italian ET cohort supports these findings (Carobbio et al, 2007). 129 thrombotic events – arterial or venous – were recorded prior to or at diagnosis in 439 patients. During follow-up, a further 78 events were diagnosed in 67 patients. Multivariate analysis demonstrated that ‘standard risk factors’ (age ≥ 60 years, previous thrombosis), as well as leucocytosis, were independently associated with thrombotic events. The JAK2V617F mutation status was evaluated in 277 patients (38 vascular events) but, on multivariate analysis, V617F-positivity was not found to independently predict thrombosis during follow-up (hazard ratio = 1·4, 95% confidence interval 0·7 to 3·0).
Overall, there is conflicting evidence for the importance of the JAK2V617F mutation status in predicting the risk of thrombosis in patients with MPD. Nine studies have noted an association between mutation status and thrombotic risk (albeit confined to venous events in two of these); however, all were hindered by the presence of potentially confounding factors. In the three series where multivariate analysis was performed (Heller et al, 2006; Kittur et al, 2007; Vannucchi et al, 2007a), the association did persist after adjusting for these factors, but there were caveats in each. In Vannucchi et al (2007a), the independent association was only with rare homozygous-mutant cases, in Kittur et al (2007) the association was lost if prior thrombosis was incorporated into the model, and in Heller et al (2006), the majority of the thrombotic events occurred prior to or at diagnosis. Four studies refuted the association between JAK2V617F mutation status and thrombotic risk (Wolanskyj et al, 2005; Tefferi et al, 2006; Alvarez-Larran et al, 2007; Carobbio et al, 2007), though these tended to include fewer patients, and analysed arterial and venous events together. Consequently, they may not have detected small differences in rates of vascular events between the V617F-positive and the V617F-negative groups, particularly differences confined to either arterial or venous subtypes. There are no comprehensive studies evaluating the influence on thrombotic risk of co-existing inherited common thrombophilic states in patients with the JAK2V617F mutation.