Splanchnic vein thrombosis in myeloproliferative neoplasms: pathophysiology and molecular mechanisms of disease

Clinical factors associated with SVT in MPN

The proportion of SVT among all VTEs is disproportionately higher in MPN patients compared with the general population, as SVTs make up an estimated 7.5% of total VTE cases [De Stefano et al. 2008]. In comparison, the frequency of DVTs and PEs in the general population is 400 times greater than that of SVTs [Deitelzweig et al. 2011]. Interestingly, the risk factors associated with these unusual thromboses are different from the risk factors traditionally associated with all-cause arterial and venous thrombosis. Figure 1 compares the known clinical risk factors of SVTs with those of all-cause VTE.

OPEN IN VIEWER

Figure 1.

Venn diagram comparing cited clinical risk factors for SVTs and all-cause VTEs. Bold indicates the most validated risk factors.

JAK2 mutation in SVT in the absence of overt MPN

An estimated 40% of BCS patients and 28% of PVT patients also test positive for the JAK2 V617F mutation [Smalberg et al. 2012]. For this reason, in any patient presenting with an SVT, JAK2 testing is recommended as part of the initial work-up for MPN. This has led to the observation that a significant proportion of SVT patients (15–17%) test positive for JAK2 V617F, but do not otherwise meet WHO criteria for an MPN [Smalberg et al. 2012]. Indeed, one study estimated that testing for JAK2 can identify an additional 30% of MPN patients who do not otherwise have overt evidence of disease [De Stefano et al. 2007]. The most frequently cited reason for the absence of elevated blood counts is the presence of portal hypertension and splenomegaly that result from SVT and MPN, which can lead to sequestration of blood cells as well as hemodilution, thus ‘masking’ the clinical phenotype of MPN. Investigators have found that while ‘masked’ patients display lower hemoglobin levels, they also have increased plasma volumes secondary to splenomegaly or SVT formation [Lamy et al. 1997].

It is also possible that SVT patients with JAK2 V617F, but no overt clinical features of MPN, may represent a distinct subtype of MPN or an early phase of the disease. Up to 70% of patients presenting with SVT do not carry a prior diagnosis of MPN, and it has been estimated that SVTs represent the first thrombotic event in 7.5% of patients with PV and ET [De Stefano et al. 2008; Hoekstra et al. 2011]. SVTs may thus occur during an early stage of MPN, prior to the development of the clinical phenotype. For instance, one meta-analysis found that 21 out of 41 JAK2-positive patients without an MPN diagnosis at the time of SVT presentation subsequently developed an MPN during follow up [Dentali et al. 2009]. A more recent study looking at long-term outcomes of SVT patients also observed that 9 out of 13 JAK2-positive patients were found to have an MPN after 64 months [Colaizzo et al. 2013]. These findings suggest that in these patients, SVTs represent an early manifestation of their disease. In Colaizzo and colleagues’ study, six out of the eight SVT patients who were initially JAK2-negative were found to be JAK2-positive with clinical symptoms of an MPN by 21 months [Colaizzo et al. 2013]. Such observations suggest either that a disease process was already in existence before the occurrence of the JAK2 mutation, or that the initial JAK2 V617F allele burden had not reached a detectable level by conventional laboratory testing at the time of SVT diagnosis. The hypothesis that SVTs are early manifestations of MPN may partially account for the finding that MPN-associated SVT patients tend to be younger with lower JAK2 V617F allele burdens.

The hypothesis that SVT patients represent an earlier stage of MPN disease has important implications in evaluating the underlying molecular pathogenesis of MPNs. It has been hypothesized that blood cancers arise from stepwise genetic mutations that confer survival and growth advantages to hematopoietic stem cells, eventually resulting in clonal hematopoiesis and malignancy [Xie et al. 2014]. While early mutations may allow clonal expansion, cooperating mutations that occur later enable development of a hematopoietic cancer. JAK2-positive patients without clinical findings of an MPN may lack these cooperating mutations that result in blood cell proliferation. For instance, a significant percentage of MPN-associated mutations, including JAK2 V617F, have been found in elderly individuals without overt evidence of an MPN [Genovese et al. 2014; Xie et al. 2014]. These individuals, however, are at an increased risk of developing a subsequent hematologic malignancy and cardiovascular mortality [Jaiswal et al. 2014]. These data suggest that mutations such as JAK2 V617F can be initiating events that may ultimately lead to the development of overt malignancy. Similarly, in SVT patients who subsequently develop MPNs, it is possible that the development of thrombosis reflects pathology related to an initiating clone. The progression to overt MPN, including elevated cell counts, may only become apparent upon acquisition of subsequent mutations. These patients with positive JAK2 mutation status may thus be considered a ‘pre-malignant’ state, and as such represent a unique population for studying the clonal evolution of MPNs.

Prevalence of the JAK2 V617F in MPN-associated SVTs

There is a high prevalence of the JAK2 V617F mutation found in SVT, and the percentage is even higher when restricted to patients meeting MPN criteria with SVT [Patel et al. 2006]. An estimated 71–100% of patients with overt MPN and SVT test positive for JAK2 V617F, which is significantly higher than the prevalence of the JAK2 mutation in MPN patients with other VTEs [Austin and Lambert, 2008]. A systematic review of 24 studies and 3123 patients found that JAK2 V617F conferred a significantly increased risk of SVT formation, with an odds ratio of 54 [Dentali et al. 2009]. In addition, while the mean prevalence of the JAK2 mutation in SVTs was high (33%, similar to rates reported in the literature), the mean prevalence of JAK2 V617F in other VTEs was low (0.88–2.57%) [Dentali et al. 2009]. Despite these associations, one study observed that young females tended to have a lower JAK2 mutant allele burden, even though they are at higher risk for SVTs [Stein et al. 2010]. The findings of increased incidence of the JAK2 mutation but lower allele burden may suggest an ‘all or nothing’ effect of the JAK2 mutation, which is perhaps subsequently modified by additional clinical risk factors such as concomitant hypercoagulable disorders, as discussed in the previous section.

JAK2 V617F mutation as a molecular driver of SVT formation

The majority of SVTs (with the exception of BCS) are due to local perturbations in the venous system, such as cirrhosis or nearby malignancy. It is possible that the JAK2 mutation itself may have local effects on the splanchnic venous system. This venous system is unique in that blood flow velocity is much slower, leading to prolonged interactions between blood and sinusoidal endothelial cells [Aird, 2007a]. In vitro models have shown that PV red blood cells demonstrate increased adhesion to endothelial laminin, and this adhesiveness is particularly increased at low shear rates [Wautier and Wautier, 2013]. Furthermore, the portal venous system is exposed to gut-derived inputs, altering its immunogenicity such that vessels are more vulnerable to activated platelets and states of high viscosity such as increased cell count [Aird, 2007b]. These characteristics may put the splanchnic venous system at particular risk to perturbations caused by JAK2 V617F, which affect not only cell quantity, but cell quality.

The JAK2 mutation has a widespread effect on the hematologic system. JAK2-mutant megakaryocytes display hypersensitive signaling and increased mobility and aggregation [Hobbs et al. 2013]. Studies with JAK2 V617F knock-in mice have demonstrated that megakaryocytes carrying the JAK2 mutation show greater chemotaxis along migration assays, as well as increased pro-platelet formation in response to low-level thrombopoietin signaling [Hobbs et al. 2013]. JAK2-positive platelets also display increased in vitro thrombus formation when exposed to collagen-coated surfaces, although Lamrani and colleagues found decreased platelet reactivity when tested in a different mouse model [Hobbs et al. 2013; Lamrani et al. 2014]. However, accelerated platelet aggregation occurred when studied in vivo in artificially injured blood vessels [Lamrani et al. 2014]. Blood samples taken from JAK2-positive patients with MPN also show significantly higher levels of soluble P-selectin, a marker of platelet activation, compared with JAK2-negative patients with MPN [Robertson et al. 2007].

Leukocyte activation also contributes to the pro-coagulant milieu, as studies have shown that markers of polymorphonuclear activation correlate with markers of hypercoagulability in ET and PV patients [Falanga et al. 2005]. It is thus believed that activated leukocytes can cause secretion of pro-coagulant factors, leading to further platelet activation and aggregation. In JAK2-positive patients, the level of both leukocyte and platelet activation increases with higher JAK2 mutational burden, and an epidemiological study by Barbui and colleagues found that the highest incidence of thrombosis was seen in PMF patients with both leukocytosis and the JAK2 mutation [Arellano-Rodrigo et al. 2006; Barbui et al. 2013]. Red blood cells also display increased adhesion to the endothelium in PV patients [Wautier and Wautier, 2013]. This effect may be mediated by aberrations in JAK/STAT signaling, as JAK2 V617F can induce phosphorylation of glycoproteins critical for red blood cell adhesion [De Grandis et al. 2013]. Taken together, these findings suggest that the JAK2 mutation leads to distinct changes in hematopoietic cells that increase the likelihood of thrombus formation.

The propensity for SVT development in MPN is also potentially due to differential regulation of hemostasis in specific vascular beds. Endothelial cells are integrally involved in regulating thrombosis, and are capable of expressing both pro- and anticoagulant factors depending on both the environmental milieu and cell type [Rosenberg and Aird, 1999]. This is supported by observations that MPNs are associated with specific anatomical locations within the splanchnic venous system. In MPN patients with BCS, for example, thrombosis tends to occur in the large, hepatic veins, while BCS patients with other hypercoagulable disorders tend to have thrombosis involving the inferior vena cava [Valla, 2015]. Signaling pathways specific to the splanchnic venous system can then cause unique interactions with mutated blood cells. For instance, in PNH, a complement-mediated disorder that also presents with increased susceptibility to SVT, the presence of gut-derived antigens in the splanchnic veins may cause higher local activation of complement, with subsequent higher rates of hemolysis and thrombotic activation [van Bijnen et al. 2012].

Mutant endothelial cells may also contribute to SVT pathogenesis. Liver endothelial cells isolated from two BCS patients have been found to carry the JAK2 mutation [Sozer et al. 2009]. Further studies have identified the JAK2 mutation in spleen endothelial cells and circulating endothelial progenitor cells [Teofili et al. 2011; Rosti et al. 2013]. JAK2 V617F-mutated endothelial progenitor cells also displayed increased adherence to normal mononuclear cells, and were found to abnormally activate the JAK/STAT pathway [Teofili et al. 2011]. These findings provide a framework in which potentially aberrant endothelial cells in the splanchnic venous system interact with aberrant blood cells, eventually leading to the formation of SVT. The exact mechanism of how alterations in the endothelium lead to SVT is still unclear, and warrants further investigation. Given the unique environment of the splanchnic venous system, it is likely that the mechanisms of SVT formation differ from the mechanisms of arterial and DVT, accounting for the difference in risk factors.

CALR mutations

The CALR gene encodes for calreticulin, a multifunctional protein that regulates calcium signaling and protein folding in the endoplasmic reticulum and cytoplasm [Johnson et al. 2001]. Only recently has a mechanism for CALR-mutant MPN pathogenesis been proposed. A series of papers demonstrated that mutated calreticulin is able to induce MPN phenotypes in mice through interaction with the thrombopoietin receptor (MPL) [Araki et al. 2016; Chachoua et al. 2016; Elf et al. 2016; Marty et al. 2016]. Mutated calreticulin has been shown to bind preferentially to MPL, resulting in thrombopoietin-independent activation of MPL with subsequent stimulation of the JAK/STAT pathway [Araki et al. 2016; Chachoua et al. 2016; Elf et al. 2016; Marty et al. 2016]. The specificity of mutant calreticulin interaction with MPL accounts for the findings that CALR mutations are largely absent in PV, but found in 30% of all ET and PMF patients, and 70–80% of JAK2-negative ET or PMF [Klampfl et al. 2013; Nangalia et al. 2013]. Compared with JAK2, CALR-mutant patients tend to have increased platelet counts, lower hemoglobin and white cell counts, as well as decreased rates of thrombosis, with a more indolent disease course [Nangalia et al. 2013].

How or whether CALR mutations induce SVT susceptibility is less clear. Despite the relatively high frequency of CALR mutations in MPNs, recent studies have failed to find an increased incidence of CALR mutations in SVT patients. CALR mutations have been identified in 0–2% of SVT, with an estimated total prevalence of approximately 0.9% of cases [Castro et al. 2015; Colaizzo et al. 2015; Haslam and Langabeer, 2015; Iurlo et al. 2015; Marzac et al. 2015; Plompen et al. 2015; Roques et al. 2015; Turon et al. 2015]. Given its relatively low prevalence, testing for CALR is not generally recommended in patients with SVT [Colaizzo et al. 2015]. Furthermore, its rarity suggests that it is not a major contributor to SVT formation, a fact consistent with the observation that CALR-mutated patients exhibit decreased rates of thrombosis overall [Rumi et al. 2014].

MPL mutations

Mutations in the MPL gene are the third most commonly tested for mutation after JAK2 and CALR in MPN patients. MPL encodes for the thrombopoietin receptor, and mutations expectedly cause abnormalities in megakaryocyte and platelet production. The most common mutations result in an amino acid substitution (either lysine or leucine) at the 515 position (MPL W515), are found in approximately 5% of ET/PMF patients, and are mutually exclusive to JAK2 and CALR mutations [Pikman et al. 2006; Rumi et al. 2013]. MPL-mutant patients tend to be older, with lower cell counts compared with JAK2 and CALR-positive patients [Beer et al. 2008]. MPL mutations have not been shown to confer significant prognostic impact, and patients do not have differing risks of thrombosis or major hemorrhage [Beer et al. 2008]. However, similar to CALR, MPL W515 mutations are rare in SVT, with incidence rates even less than CALR [Bergamaschi et al. 2008; Kiladjian et al. 2008; Jadli et al. 2012; Smalberg et al. 2012; Colaizzo et al. 2015]. This may be related to the overall low prevalence of MPL in MPNs, but there is no evidence to suggest that MPL-positive patients have higher rates of SVTs.

The JAK2 46/1 haplotype

The JAK2 46/1 haplotype has recently been found to be associated with JAK2-positive MPNs, and it is believed that the JAK2 V617F mutation arises specifically on the 46/1 allele [Jones et al. 2009]. Its role in SVT formation is controversial, with some studies showing that the haplotype is an independent risk factor for SVT development in JAK2 V617F patients [Colaizzo et al. 2010a; Smalberg et al. 2011], and some studies showing only a weak association [Kouroupi et al. 2011]. A recent meta-analysis of 26 observational studies evaluated the risk of MPN with the JAK2 haplotype, and found that the haplotype was significantly increased in MPN and SVT patients. In JAK2-negative patients, however, the haplotype did not confer an additional risk of SVT [Li et al. 2014]. The increased risk of SVT with the 46/1 haplotype may be due to an interaction of the haplotype with the JAK2 V617F mutation, rather than the haplotype itself.

Additional mutations in MPN-associated SVTs

Approximately 10% of MPN patients are ‘triple-negative’ and do not harbor JAK2, CALR, or MPL mutations [Tefferi et al. 2014]. Additional mutations are found in varying frequencies in PV, PMF, and ET patients, although they tend not to be mutually exclusive with JAK2, CALR, MPL mutations, or with each other. These mutations include TET2, DNMT3A, EZH2 and ASXL1, and are involved in DNA methylation and chromatin structuring [Delhommeau et al. 2009; Grand et al. 2009; Ernst et al. 2010; Abdel-Wahab et al. 2011]. The specific role of these mutations in driving MPN disease formation is not fully understood. As discussed previously, SVT patients may be an ideal population for investigating the role of precursor genes and initiating mutations, as these patients may present at an early stage of MPN disease evolution. For instance, TET2 is a tumor suppressor gene which plays a key role in DNA methylation, and mutations in TET2 have been identified in up to 12% of MPN patients, including both JAK2-positive and JAK2-negative MPNs [Delhommeau et al. 2009]. In the hematopoietic stem cells of patients with both JAK2 and TET2 mutations, the TET2 mutation could often be found in the absence of the JAK2 mutation, suggesting the acquisition of TET2 before JAK2 [Delhommeau et al. 2009]. In surveying SVT patients, however, only a minority (3/23) of SVT patients tested positive for TET2, while all were JAK2 positive [Colaizzo et al. 2010b]. A larger study of 43 BCS patients identified 6 patients with a deleterious TET2 mutation; however, only 3 were JAK2-negative and none of these 3 patients developed an overt MPN [Westbrook et al. 2012]. Based on the studies by Colaizzo and colleagues and Westbrook and colleagues, the prevalence of TET2 in SVTs is roughly 14%, which is similar to the prevalence rate of TET2 in all MPNs [Delhommeau et al. 2009]. While TET2 is found more frequently than CALR and MPL in SVT, its frequent co-occurrence with JAK2 and lower prevalence rate suggests a limited role in SVT formation. Further investigations for mutations present in JAK2-negative SVT patients are limited. A small study of 25 cases of JAK2-negative BCS found no patients with IDH1 or IDH2 mutations [Jadli et al. 2012].

Table 2 summarizes the prevalence of the most common MPN-associated mutations in SVTs and all MPNs. In light of the existing literature, there seems to be a lack of evidence to suggest that these other mutations are associated with SVT formation in MPN, despite the finding that 14–20% of SVT patients with an MPN are JAK2-negative [Kiladjian et al. 2008; Smalberg et al. 2012]. The existence of a precursor mutation to JAK2 V617F critical to SVT pathogenesis has yet to be identified. Additional research examining the clinical and molecular differences between JAK2-positive and JAK2-negative patients with SVT is needed to further elucidate the driving mechanisms of SVT formation and MPN disease progression.

OPEN IN VIEWER

Table 2.

Prevalence of MPN-associated mutations in SVT and MPN patients. JAK2 makes up the highest proportion of patients, while CALR and MPL have very low prevalence rates in SVTs. TET2 is found at higher prevalence rates but is not mutually exclusive with JAK2.

	SVT patients	MPN patients
JAK2	33%	95% in PV; 60% in ET and PMF
CALR	0.9%	30% in ET and PMF
MPL	0.7%	5% in ET and PMF
Other
TET2	14%	12%

Sources: MPL- 0.7% in SVT, JAK2- 33% in SVT patients- Smalberg et al. [2012]; CALR- 0.9% in SVT patients- De Stefano et al. [2015]; MPL- 0.7% in SVT- Bergamaschi et al. [2008]; CALR- 0.9% in SVT patients- Colaizzo et al. [2015]; MPL- 0.7% in SVT- Kiladjian et al. [2008]; MPL- 0.7% in SVT, TET2- 14% in SVT- Jadli et al. [2012]; CALR- 0.9% in SVT patients- Turon et al. [2015]; TET2- 14% in SVT- Westbrook et al. [2012]; TET2- 14% in SVT- Colaizzo et al. [2010b]; 12% in MPN- Delhommeau et al. 2009; 5% in ET and PMF - Pikman et al. 2006; 95% in PV, 60% in ET/PMF- Campbell and Green 2006; 30% in ET and PMF- Klampf et al. 2013.

ET, essential thrombocythemia; MPN, myeloproliferative neoplasms; PMF, primary myelofibrosis; PV, polycythemia vera; SVT, splanchnic vein thrombosis.