Advances and challenges in the management of essential thrombocythemia

General background and pathophysiology

Essential thrombocythemia (ET) belongs to a family of related disorders characterized by an uncontrolled cell growth, named myeloproliferative neoplasms (MPNs), also including polycythemia vera (PV) and primary myelofibrosis (PMF). In the World Health Organization (WHO) classification, the word ‘neoplasm’ was introduced instead of ‘disorder’ in order to underline the clonal character of the diseases. ET by far has the best prognosis of the three, with expected life duration close to normal.

The phenotypes of the three related neoplasms differ considerably, even if there are similarities. In PV, the main characteristic is increased red cell mass and in ET increased platelet levels, whereas anemia is the main feature in myelofibrosis (MF). Thrombocytosis is a prerequisite for the diagnosis of ET, but may accompany both of the others, although more seldom. Transitions may occur: ET may develop into PV or MF, and PV may develop into MF. One school of thought believes in a pathophysiological continuum, starting from ET and ending in MF. However, this development is seen in a small minority of patients. Somewhat more common is the transition from ET directly to MF, but even this is a rare event, making it unlikely that the continuum theory may explain the basic pathophysiology.

The fact that made the continuum theory attractive was the detection of the V617F mutation in the tyrosine pseudokinase region of the JAK2 gene, being a gain-of-function mutation, resulting in uncontrolled cellular growth in the hematopoietic compartment. It is found in >95% of PV patients, but only in 50–60% of ET and MF patients. Therefore, it is still an open question which role the mutation has for the development of disease in ET and MF. The presence of a JAK2V617F mutation indicates MPNs, but does not differentiate between them. The allele burden of the mutated JAK2 gene is much lower in ET than in PV, and homozygous mutated cells are hardly ever found in ET but are common in PV.

An important addition to our knowledge was the finding of a new mutation, the CALR mutation, in 2013. It is present in about 20% of patients with ET and PMF but very rarely in PV, and in ET it is (with very few exceptions) not present in JAK2- or MPL-positive patients. A small percentage of ET patients have other mutations, especially MPL mutations, but there is still no reliable singular molecular marker for the disease. The picture is becoming increasingly complex, with some patients having several mutations. Mutations of TET 2, ASXL1, IDH1/2, EZH2 and other genes have been found, although these are more frequent in PMF.

New classification of true ET

An important change in the definition of ET has been introduced in recent years, separating true ET from early MF by means of bone marrow morphology (Figure 1) [Thiele et al. 1999; Thiele and Kvasnicka, 2003, 2006; Tefferi et al. 2007].

Figure 1.

WHO 2008 criteria for the diagnosis of ET.

CML, chronic myelogenous leukemia; ET, essential thrombocythemia; MDS, myelodysplastic syndromes; PMF, primary myelofibrosis; PV, polycythemia vera; WHO, World Health Organization.

Previous classifications have allowed a considerable degree of bone marrow fibrosis and morphologic features more resembling MF, which has produced heterogeneity in patient material in research in the field. Studies using the new WHO classification have recently shown the usefulness of this distinction: true ET is characterized by lower white blood cell (WBC) counts, lower hemoglobin (Hb) levels (normal), lower lactate dehydrogenase (LDH) levels in plasma and, importantly, a better prognosis, which is close to normal, as shown in a large retrospective study (n = 1104) with reclassification of the bone marrow biopsy from the time of diagnosis. Substantial differences were shown after 15 years of follow up with regard to survival rates (59% and 80%, respectively) and leukemic transformation rates (11.7% and 2.1%, respectively) [Barbui et al. 2012b]. A prospective study with 7 years follow up did a similar re-evaluation of the bone marrow at diagnosis and showed that transformation to overt MF was rare in the true ET group but common among the patients with early MF [Ejerblad et al. 2013]. A recent prognostic model for WHO-classified ET indicates that expected survival from diagnosis is 13.7 years for high risk patients, 24.5 years for an intermediate group and >25 years for low risk patients [Passamonti et al. 2012].

The distinction between true ET and early MF so far has no great consequences for the pharmacological treatment, since this is directed by risk stratification for thrombohemorrhagic events, but it is of course important in communications with the patients and may soon be very important for treatment decisions with the new drugs under development. It is also important for making patient cohorts more homogenous in studies, thereby making studies comparable. One group of hematopathologists have found it difficult to reach diagnostic consensus using the new classification [Brousseau et al. 2010; Wilkins et al. 2008] and remain skeptical about its usefulness. However, even these critical authors have confirmed the negative impact on prognosis of bone marrow reticulin at diagnosis, thereby indirectly supporting the importance of a distinction [Campbell et al. 2009]. A recent study found good concordance in establishing the diagnosis (kappa index 0.67) and a high specificity of histopathological changes (98.5%), whereas the sensitivity was rather low (54%) [Alvarez-Larran, 2014a]. Several studies now support both feasibility and usefulness of the WHO ET and early PMF definitions, and it is to be expected that prospective studies from now on will use it [Gianelli et al. 2006, 2008, 2013, 2014a, 2014b; Thiele et al. 2011; Pozdnyakova et al. 2014]. A recent paper summarizes the discussion of this issue [Barbui et al. 2013].

Diagnosis

The discussion above of the need to distinguish true ET from early PMF shows that a current challenge is to have pathologists trained in the new morphologic diagnostic criteria. It also underlines the fact that a bone marrow biopsy is essential for a correct diagnosis. It seems likely that the 2008 WHO criteria (Figure 1) will be including the CALR mutation in point 4, but this does not change the criteria.

A problem which has been the focus of several studies lately is the distinction between ET and PV in patients with thrombocythemia and a ‘masked’ polycythemia with Hb and hematocrit (Hct) below the diagnostic cutoff for PV [Barbui et al. 2014b]. This was shown to be much more common in JAK2+ patients [Barbui et al. 2014a]. In the absence of previous thrombosis, such patients under the age of 60 would be classified as low risk ET and possibly would be undertreated both with regard to antiplatelet therapy and phlebotomies. A discussion about an amendment to Hb and Hct cutoff limits for PV diagnosis is ongoing. Figure 2 suggests a practical guide for the workup of thrombocythemia.

Figure 2.

Suggested algorithm for diagnostic work-up of sustained thrombocytosis. Some labs perform all three mutation analyses upfront.

BMB, bone marrow biopsy; ET, essential thrombocythemia; MPN, myeloproliferative neoplasm; PMF, primary myelofibrosis; WHO, World Health Organization.

The role of mutations for management

The detection of driver mutations in ET and other MPNs has greatly increased our understanding of the pathophysiology of the diseases. However, the situation is complex and the role of the various mutations for the initiation and progression of disease remains unclear. This area is outside the scope of this review. From a clinical point of view, it is interesting that different mutations cause differences in phenotypic expression and the question is whether this also can be the basis of management decisions.

The JAK2V617F mutation

In 2005, a breakthrough in MPN research was made when a mutation in the pseudokinase domain of the JAK2 gene was reported by four different groups [Baxter et al. 2005; James et al. 2005; Kralovics et al. 2005; Levine et al. 2005]. Whereas around 95% of PV patients carry the mutation, around 60% of ET and PMF patients are JAK2 mutation positive. Interestingly, JAK2+ ET patients showed a slightly different phenotypic pattern than wild type patients, presenting with higher Hb and WBC levels, lower serum erythropoietin (S-Epo) and lower platelets [Campbell et al. 2005]. The differences were statistically significant at a group level, but clinically small: Hb 145 versus 135 g/l, WBC 10.6 versus 9.3, platelets 902 versus 1030 × 10^{/l. This study did not use the WHO diagnostic criteria and as a result included a rather heterogeneous population where a large proportion had bone marrow fibrosis grade 2 or higher. No difference in fibrosis grade could be seen in} JAK2-mutated versus wild type. In a large retrospective study in 867 ET patients with a 20 year retrospective follow up, the difference in thrombosis frequency between JAK2+ and wild type did not reach significance (p = 0.054) [Carobbio et al. 2009]. Three meta-analyses were published in 2009. All three found an increased risk of thrombosis with an odds ratio (OR) around 1.8–1.9 [Ziakas, 2008; Dahabreh et al. 2009; Lussana et al. 2009].

A recent study investigated the importance of the development of the allele burden in PV and ET patients, and it was shown that progression of the allele burden as well as a high stable level was significantly correlated to development of MF [Alvarez-Larran et al. 2014b].

Even if there are conflicting results in various studies, there seems to be an increased risk for thrombosis in JAK2+ ET patients. What this means for risk profiling and management is as yet not clear, but is discussed below.

The CALR mutation

The most recent addition to the mutational pattern in MPNs is the detection of a driver mutation of CALR, a highly conserved multifunctional endoplasmic reticulum protein with partly unknown functions. The mutation is found in ET and PMF, but almost exclusively in JAK2- and MPL-negative patients [Klampfl et al. 2013; Nangalia et al. 2013]. The frequency is around 20% in both ET and PMF, which means that about 85% of ET patients can now be diagnosed with a molecular marker. This will be of importance for diagnosis, and inclusion of CALR status in the WHO classification for ET has been suggested [Tefferi et al. 2014b], even if none of the mutations can differentiate between PMF and ET. Interestingly, the CALR mutation in ET produces a phenotype profile with clear differences from JAK2+patients. Compared with JAK2+, CALR+ patients are younger, more commonly male, have higher platelet levels, lower leukocyte and Hb levels, and lower thrombosis rate [Gangat et al. 2014; Rotunno et al. 2014; Rumi et al. 2014; Tefferi et al. 2014c]. The difference in thrombotic rate is quite marked: the 10-year cumulative incidence being 5.1% and 14.5%, respectively [Rotunno et al. 2014], and correspondingly, the 15-year rate 10.5% and 25.1% [Rumi et al. 2014]. With regard to survival and transformation to PMF, results are at present conflicting [Tefferi et al. 2014a,c].

Other risk factors and new prognostic risk score models

ET treatment is based on risk stratification with the clinical goal to reduce the frequency of thrombosis and hemorrhage. These are so far the only complications of the disease that have been shown unequivocally to be modified by treatment. The classical factors in the stratification are age >60, previous thrombosis and platelets >1500 × 10^{/l (for bleeding). New risk factors of potential value have been suggested: WBC count,} JAK2 mutation status and cardiovascular (CV) risk factors.

A number of studies have investigated whether the WBC count is correlated to incidence of thrombosis, and results have been contradictory both with regard to whether there is any correlation at all and what level of WBC count that would mark a risk level. An early study showed a correlation with thrombosis before but not after diagnosis [Gangat et al. 2007]; the same group later confirmed the lack of correlation with thrombosis during follow up [Gangat et al. 2009]. Several other studies have indicated that elevated WBC level is a risk factor for thrombosis [Carobbio et al. 2008, 2011; Passamonti et al. 2012]. However, in one of these studies only low risk ET showed this correlation and, in another, a correlation was only found for arterial thrombosis, and various studies have defined different cutoff levels of WBC counts for the association. Finally, in the most ambitious attempt to define a new risk score model for ET, the IPSET-thrombosis, WBC count did not come out as a factor in the multivariate analysis [Barbui et al. 2012a; Passamonti et al. 2012]. Other risk score models include this variable, however. The reasonable conclusion may be that there probably is a weak correlation, however not dominating.

CV risk factors have been included in multivariate analysis and have been found to correlate with a higher incidence of thrombosis in several studies [Barbui et al. 2012a; Passamonti et al. 2012]. They are therefore included in some risk score models, but not all.

Several risk score models have been published recently that all seem to differ reasonably well between patient groups with high, medium and low risk. However, they do not use the same variables, but different sets of criteria hold up in multivariate analysis in different studies [Passamonti et al. 2012; Tefferi and Barbui, 2013; Fu et al. 2014; Montanaro et al. 2014]. Results are sometimes quite confusing. In a large recent study, 8 studies with >300 patients each (range 311–1220) were compared with regard to the result of multivariate analysis. Variables like WBC >11 × 10^{/l and} JAK2 positivity/allele burden were very inconsistent as predictors, whereas age, previous thrombosis and CV risk factors were more consistent [Montanaro et al. 2014].

Hemostasis and cell interaction

Relatively few groups have previously worked in this area compared with other risk factors for thrombosis, but it has been known for a long time that MPN patients have important dysfunctions in the hemostatic system. Platelets as well as leukocytes are activated, and their interaction disturbed [Falanga et al. 2005]. Furthermore, recent studies have shown high levels of membrane bound as well as plasma soluble P-, E- and L-selectins, indicating that activated platelets and endothelial cells may promote thrombus formation [Karakantza et al. 2004]. JAK2V617F+ endothelial cells have been found to contribute to clot formation in mice [Etheridge et al. 2014]. A high level of activated protein C resistance was recently found in ET patients with previous thrombosis [Brinkman et al. 2005]. Furthermore, an elevated level of circulating microparticles has been found in ET patients [Trappenburg et al. 2009]. All these findings indicate that the whole intravascular milieu of MPN patients presents a hypercoagulation state that needs to be further explored. On top of these ET-related disturbances, there are inherited factors with special importance in ET that are independent risk factors for thrombosis such as thrombophilic single nucleotide polymorphisms (SNPs) in genes of Factor VII [Buxhofer-Ausch et al. 2014].

As patients on aspirin still get thrombosis, it has been challenged whether the thromboxane inhibition produced by low dose aspirin is sufficient, and data supporting this view have been repor-ted [Dragani et al. 2010; Pascale et al. 2012]. Furthermore, adenosine diphosphate (ADP) receptor inhibition has been suggested as an additional therapeutic effort apart from thromboxane inhibition [Panova-Noeva et al. 2013].

This whole area of hemostasis and cell interaction is currently undergoing a vigorous expansion and may provide important information on thrombotic risk management in MPN.

Symptom burden in ET

The attitude to patient complaints of fatigue and other symptoms in ET has previously been one of skepticism among many doctors. Many patients probably have been told that ‘you don’t get tired by having increased platelets’. And it is true that about 50% of newly diagnosed ET patients are asymptomatic. However, recent studies have shown that ET patients have a significant symptom burden with an effect on quality of life (QoL). In an international effort, a symptom assessment tool specific to the MPN population was developed [Emanuel et al. 2012] and validated [Scherber et al. 2011]. It is now available in many languages to be used both for research purposes and as a clinical tool in the individual patient. The most commons symptoms are constitutional like fatigue, night sweats and weight loss, as well as resulting from microvascular complications like headache, dizziness and erythromelalgia [Scherber et al. 2011]. Additionally, about 20% of ET patients have experienced thrombosis before or at diagnosis.

The use of risk stratification and risk score models

As shown above a number of new risk score models have been presented over the past few years. It seems important to realize that a risk score model may be helpful in further defining a patient’s risk, but equally important that such a model is only a first step towards an instrument for treatment decisions. In a recent study, the usefulness of the IPSET-thrombosis score model was tested against the traditional risk factors age and previous thrombosis, and was not found to add any further, useful information [Angona et al. 2014]. Ideally, randomized treatment studies proving the usefulness of a treatment decision should be used. However, with several factors indicating a lower thrombotic risk and as many indicating the opposite, it is very unlikely that there will be randomized studies large enough to validate a specific model. Therefore, a pragmatic approach seems warranted. The classical risk stratification based on age and previous thrombosis indicating high risk status is still useful, as well as platelets >1500 × 10^{/l indicating high risk for hemorrhage. The importance of reducing CV risk factors can hardly be questioned and this should therefore always be evaluated in the individual patient. Whether, for example, a low risk ET patient (age <60, no previous thrombosis) displaying one or all of} JAK2+, WBC >15 and CV risk factors should be treated as a high risk with aspirin and cytoreduction cannot for the moment be answered by risk score models, but is up to the judgment of the treating physician. Likewise, there are not sufficient data in the literature to say that a CALR+ patient with a high risk criterion can be given no treatment, but the information may be used in special situations.

Standard treatment of WHO diagnosed (‘true’) ET

The main clinical goal of treatment is still to reduce thrombohemorrhagic complications and the means are anti-aggregatory treatment, treatment of CV risk factors and cytoreduction with the target platelet level of <400 × 10^{/l. There is no hard evidence that a target of 400 × 10/l gives better thrombosis protection than 600 × 10/l, which previously was a common target. However, since the diagnostic cutoff was lowered from 600 to 450 × 10/l in the WHO classification, normalization of platelet counts has become the standard target. Other possible goals such as prolongation of survival or reduction of transformations are still not obtainable, which is why the classical risk stratification is still valid.}

Aspirin

The use of aspirin [acetylsalicylic acid (ASA)] in ET is not based on prospective randomized studies, but instead is built on extrapolation from a large PV study showing a significant reduction in thrombotic events in ASA-treated patients [Landolfi et al. 2004]. There is general agreement that high risk ET patients should be treated with low dose aspirin, and, as mentioned above, some data support a higher dose. With regard to low risk ET, there is a discrepancy between recommendations and general practice. Most doctors give ASA to practically all ET patients, whereas the European Leukemia Net (ELN) consensus recommendations advise a more restrictive approach: ASA treatment only for patients with microvascular symptoms and those with CV risk factors [Barbui et al. 2011]. In a retrospective study, effects on thrombosis and hemorrhage were compared with ASA treatment in the general population and in low risk ET; no reduction in thrombosis incidence could be detected though there was an increased bleeding incidence, 1.26 versus 0.6 events per 100 patient-years [Baigent et al. 2009].

A somewhat restrictive practice seems advisable in low risk ET. A recent retrospective study investigated the effect of ASA treatment in a cohort of high risk ET patients on cytoreductive therapy and found no significant reduction of thrombosis in the ASA-treated group, except in elderly patients [Alvarez-Larran et al. 2013]. The dilemma for the clinician then is that, on the one hand, there is a lack of solid data showing which patients benefit from antiplatelet therapy and, on the other hand, maybe the doses used are too low to give optimal protection.

Antiplatelet therapy should be used with caution in patients with platelet counts >1000 × 10^{/l due to the presence, in some patients, of secondary von Willebrand disease with increased bleeding risk [}Budde and van Genderen, 1997; van Genderen et al. 1997]. In patients with very high platelet levels and previous hemorrhages, the use of antiplatelet therapy should be avoided, except if there is a very high thrombosis risk, i.e. in the presence of recurrent thrombosis.

Interferon

Interferon (IFN) inhibits the bone marrow fibroblast progenitors, suppresses the proliferation of hematopoietic progenitors and inhibits the action of platelet-derived growth factor, transforming growth factor-β and other cytokines which may be involved in the development of MF [Kiladjian et al. 2008]. Theoretically a beneficial effect on fibrosis therefore could be suspected, but no convincing such data have been presented [Silver et al. 2011]. Although not licensed for these diseases, IFN is widely used in MPN. In a study including 39 ET patients, a complete hematological response rate of 76% was found at a dose of 90 µg/week of pegylated (PEG) IFNα2a with good tolerability (12% discontinuation) [Quintas-Cardama et al. 2009]. In a 42-month follow up of this study, the hematologic complete response (CR) rate was the same and a complete molecular response was found in 17% of the ET patients. [Quintas-Cardama et al. 2013] Similar results were obtained in another study of 36 high risk ET patients [Langer et al. 2005]. However, a higher dropout rate (23/42) was observed in a study with PEG-IFNα2b [Samuelsson et al. 2006]. A follow up of 19 patients who had been on long-term (median 299 months) IFN therapy showed a reduction in JAK2 allele expression from 24% to 10% (median) [Stauffer Larsen et al. 2013]. A complete response rate (ELN response criteria) of 63% and a dropout rate of 17% was seen in a retrospective follow-up study (median duration 17 months) of PEG-IFNα2a treatment [Gowin et al. 2012]. An interesting observation was made in two CALR+ ET patients who had long-lasting (60 and 18 months) clinical remission and reduction of mutation allele burden after treatment and cessation of PEG-IFNα2a [Cassinat et al. 2014].

On the whole, randomized studies with IFN in ET are lacking. Most studies are small, many are retrospective follow-up studies and attempts to engage drug companies in randomized comparisons of IFN with other drugs for a long time have failed (finally, one such study has been launched, results pending). However, there is enough evidence that IFN is effectively inducing hematological remission in ET to give it a prominent place in guidelines for ET treatment. There is also an interesting discussion ongoing as to whether early treatment with IFN gives a chance of deep effects on disease progression [Hasselbalch, 2011; Silver et al. 2013]. However, IFN is not registered for use in MPN anywhere, which strongly inhibits its use in many countries, since it is either not allowed or not reimbursed. In the Nordic area, off label use of IFN registered for other diseases is up to the judgment of the treating physician and is reimbursed, and therefore IFN has been used off label for many years.

The experience shows that PEG IFN is preferable to short-acting, that the initial dose should be low in order to avoid side effects, that remission sometimes comes rather late, and that remission may be durable for a long time after cessation.

Hydroxicarbamide

Hydroxycarbamide (HC), also known as hydroxyurea (HU), is a urea derivative with longstanding use in myeloproliferative disorders, although it is not specifically registered for this in many countries. It is given orally and is usually well tolerated. The most common side effects are nausea and mucocutaneous lesions, especially leg ulcers. The mucocutaneous lesions may vary in severity and character from small, hard and painful ulcers to severe vasculitis, dermatomyositis, actinic keratosis or squamous cell changes [Ravandi-Kashani et al. 1999; Ruzzon et al. 2006]. A cessation rate of about 10% is usually reported [Harrison et al. 2005; Randi et al. 2005], mostly due to leg ulcers. Criteria of resistance and intolerance for HU have been published from an ELN consensus procedure [Barosi et al. 2010].

Through its action on DNA formation, HU produces macrocytosis and dysplasia can be seen in the bone marrow [Thiele et al. 2005]. This is usually disregarded as clinically irrelevant. However, HU has been suspected of increasing the risk for leukemia transformation and has been shown to be mutagenic in vitro [Osterman Golkar et al. 2013]. The issue is controversial, since various studies either give support for or minimize the leukemic risk [West, 1987; Weinfeld et al. 1994; Najean et al. 1998; Passamonti et al. 2003; Chim et al. 2005; Finazzi et al. 2005]. In the largest prospective follow-up study of cytoreductive therapy in high risk ET (EXELS), a higher event rate for transformation to acute leukemia was seen in HC-treated patients than in those on anagrelide [Birgegård et al. 2014]. The only published long-term prospective randomized study found a cumulative incidence of acute myeloid leukemia (AML)/myelodysplastic syndrome (MDS) in the HC treatment arm of 6.6%, 16.5%, and 24.2% at 10, 15 and 20 years, respectively [Kiladjian et al. 2011], whereas a large register-based, retrospective study found no support for the leukemogenic effect [Bjorkholm et al. 2011]. Current recommendations, accepting the concern, advocate caution in the use of HC to ‘younger patients’ [Barbui et al. 2011] without taking a clear stand in the debate. It is to be noted that, with the expected life duration of an ET patient, any treatment initiated before the age of 60 can be expected to last for at least 20 years. It is more widely accepted that the combination of HC with other cytostatic drugs increases leukemic risk.

The efficacy of HC in lowering platelets is well documented. With the standard starting dose of 0.5 g twice daily, the target platelet level of <400 × 10^{/l is usually reached within 6–8 weeks. The clinical efficiency in reducing thrombosis has only been tested against a nontreatment control arm in a randomized setting in one study [}Cortelazzo et al. 1995]. Two versus 14 patients had a thrombosis during the follow-up time in this study, and although quite limited in the size of the cohort, this study has made it seem unethical ever to repeat a study with a nontreatment arm in high risk ET. Two randomized studies have compared the efficacy of HC versus anagrelide. In the PT1 study, including 809 patients for an observation time of 39 months, HC was superior to anagrelide in reducing arterial thrombosis but inferior in reducing venous thrombosis [Harrison et al. 2005]. This result was not reproduced in the ANAHYDRET study, including 259 patients for an observation time of 36 months, where no significant differences were seen [Gisslinger et al. 2013]. The latter study was designed as a noninferiority study and therefore did not have the same statistical power as the former and also differed from PT1 in the selection criteria, recruiting only WHO-diagnosed true ET, previously untreated. In the largest prospective study in ET so far (EXELS), 3600 patients were observed for 5 years. The results are partly in line with the ANAHYDRET study, showing no difference in the event rate for total thrombosis, but also supporting the difference seen for arterial versus venous thrombosis seen in the PT1 study [Birgegård et al. 2014].

HC is recommended as first line therapy for older ET patients in most guidelines [Barbui et al. 2011], without a clear definition of ‘older’, and in some as first-line therapy for all patients [Harrison et al. 2010].

Anagrelide

Anagrelide (ANA) is an imidazoquinazoline, originally developed as an anticoagulation drug, which was shown to have a potent platelet reducing effect. Unlike the other cytoreductive drugs used in MPN, it is specific for the megakaryocyte line, not affecting red or white cell progenitor proliferation. It reduces platelet production by inhibiting megakaryocyte colony development, thus reducing megakaryocyte size, ploidy and maturation [Hong and Erusalimsky, 2002; Hong et al. 2006], possibly by repressing GATA-1 and FOG-1 expression [Ahluwalia et al. 2010]. Side effects are common during the first weeks of treatment, but usually subside. The most common side effects, tachycardia and headache, are probably due to phosphodiasterase III inhibiting properties of ANA. In the PT1 study, more patients developed MF in the ANA arm than the HC arm (4.2% versus 1.2%). This pattern was also seen in the EXELS study, where the event rates were 1.31 versus 0.32. It should be noted that none of these studies used the new WHO classification, meaning that patients with early MF according to the new definition were included. In true ET, transformation to MF is an even rarer event, regardless of therapy [Barbui et al. 2012b; Ejerblad et al. 2013]. Combination of ANA and aspirin induces an increased risk of hemorrhage [Harrison et al. 2005; Birgegård et al. 2014 and the label recommends caution in patients with previous bleedings. There is no concern for leukemogenesis with ANA.

The dropout rate in studies varies, but is usually around 20%.

The efficacy in platelet reduction is high, response rates varying between 76% and 94% in different studies [Petitt et al. 1997; Petrides et al. 1998; Mills et al. 1999; Laguna et al. 2000; Birgegård et al. 2004; Penninga et al. 2004; Steurer et al. 2004; Fruchtman et al. 2005]. The variation is largely due to varying definitions of response.

There is no placebo-controlled study with ANA against nontreatment investigating the thrombosis-reducing effect. The evidence for ANA rests on historical controls, showing a lower incidence of thrombotic events after start of ANA treatment [Steurer et al. 2004] and comparisons with HU [Harrison et al. 2005; Gisslinger et al. 2013; Birgegård et al. 2014].The results from the latter studies are discussed above (under the Hydroxycarbamide heading). It seems clear that ANA gives a protective effect against thrombosis which is almost similar to that of HC.

Other cytoreductive drugs

Traditionally, busulphan and radioactive phosphorus (P32) have been used in special situations in ET. Both carry an increased risk for leukemia and are therefore not recommended except in patients with ‘short’ expected life duration [Barbui et al. 2011]. In very elderly patients with intolerance or refractoriness to HC and ANA, busulphan is still a useful drug that may be used intermittently with good control of platelet levels. The effect of P32 is usually rather short-lived.

New drugs

ET patients have been included in some studies with JAK2 inhibitors. However, in the author’s opinion these drugs have little relevance in true ET. The severe symptoms that respond well to JAK2 inhibitors are extremely rare in this disease. Patients who develop splenomegaly and severe constitutional symptoms have generally already transformed to MF. There are rare patients with severe pruritus and/or splenomegaly without signs of transformation, but in ordinary ET patients, indication is lacking and the cost–benefit ratio is much too high. Patients with treatment failure on first- and second-line therapy may be considered for JAK2 inhibition, however, if cytoreductive therapy is judged to be important in the individual case. Histone deacetylase (HDAC) inhibitors and other experimental drugs tested in MF are still not tested in ET studies. An interesting new drug is a telomerase inhibitor, tested in phase I trials in ET with promising results, not yet published except as abstracts (Iancu-Rubin et al. 2014).

Treatment strategy

A treatment strategy for true ET is outlined in Figure 3. In the author’s opinion, there are insufficient data to support any of the published new risk score models as a basis for general treatment choices. The basis for therapy therefore still is the three traditional risk factors age >60, previous thrombosis and platelets >1500 × 10^{/l. The presence of any of these puts the patient in the high risk category, and cytoreduction plus low dose aspirin should be started if platelets are not >1000 × 10/l. In the latter case, aspirin should be started when cytoreduction has reduced platelets to <1000 × 10/l.}

Figure 3.

Suggested management of true ET. In patients with previous hemorrhage: avoid combination of aspirin and anagrelide.

*Caution with aspirin if platelets are>1000 × 10^/l.

CV, cardiovascular; ET, essential thrombocythemia; HC, hydroxycarbamide; HU, hydroxyurea; IFN, interferon; WHO, World Health Organization.

CV risk factors increase the basic thrombotic risk for any of the risk categories. There are therefore good arguments for naming an intermediate risk group with low risk profile plus CV risk factors. These patients should be treated with low dose aspirin, and those with badly controlled or heavy CV risk factors may be considered for cytoreductive therapy, especially if they are JAK2V617F+. For patients with short expected survival, busulphan is a possible second- or third-line therapy, and for younger patients with failure on first- and second-line therapy, JAK2 inhibition may be considered.