Essential thrombocythemia: a review of the clinical features, diagnostic challenges, and treatment modalities in the era of molecular discovery

Essential thrombocythemia (ET) is a chronic myeloproliferative neoplasm that is associated with diminished quality of life, thrombohemorrhagic complications, and transformation to myelofibrosis (MF) and acute leukemia (AML). The important recent discoveries of driver mutations, including the calreticulin gene in addition to JAK2 and MPL, have led to a greater understanding of disease pathogenesis and set the stage for the advent of more sophisticated prognostic, diagnostic, and therapeutic strategies. In this paper we summarize recent studies describing the molecular basis of ET. We review the prognostic importance of establishing a 'true' ET diagnosis, as well as risk factors for the development of adverse outcomes including thrombosis, AML (2% risk at 15 years), and MF (9% risk at 15 years). Finally, we discuss the decision to initiate treatment and assess the quality of evidence supporting the use of established, available therapies as well as novel treatments. Special situations, such as pregnancy, familial ET, and extreme thrombocytosis will also be discussed.

Keywords: Essential thrombocythemia; janus kinase 2; calreticulin; thrombopoietin; prefibrotic myelofibrosis

Essential thrombocythemia (ET) is a chronic myeloid neoplasm characterized by symptoms from small or large vessel occlusion and hemorrhage, along with characteristic molecular markers and histologic features, and a later risk for transformation to myelofibrosis (MF) or leukemia (AML). The estimated annual incidence of ET in the United States is 2.5 cases per 100,000 population, whereas the prevalence is estimated to be ∼24 cases per 100,000 population [[1]]. It is well established that women are more affected than men (with an approximate ratio of 2:1) [[3]]. Further, while the median age at diagnosis is 60 years, a significant proportion of patients (estimated 20%) are diagnosed at younger than 40 years of age [[4]]. Among the classical Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs), along with polycythemia vera (PV), and primary myelofibrosis (PMF), ET was the last to be described when, in 1934, two Austrian pathologists reported a case of 'hemorrhagic thrombocythemia' in a patient with extreme thrombocytosis and recurrent mucocutaneous bleeding (Figure 1) [[6]]. Over the next few decades numerous similar cases were described in patients with bone marrow megakaryocytic hyperplasia [[7]] making ET a relatively well-known clinical entity by the mid-twentieth century. Subsequent attempts to diagnose and characterize the disease, however, were limited by the lack of defining molecular features. The Polycythemia Vera Study Group (PVSG) created diagnostic criteria for ET placing an emphasis on excluding other causes of thrombocytosis and other mimic myeloproliferative diseases [[9]].

Graph: Figure 1. Timeline of important historical events in ET.

The past decade has witnessed a dramatic increase in our understanding of ET, in large part due to the strides made in the realm of molecular pathogenesis (Figure 1). The discovery of JAK2 V617F in 2005 [[10]], followed by MPL mutations in 2006 [[14]] and CALR mutations in 2013 [[16]] has contributed to a more nuanced understanding of disease pathogenesis and clinical heterogeneity, and has facilitated diagnostic capability. What was once a diagnosis of exclusion is now a well-defined clinical entity with a clear molecular basis and a unique epidemiology, including predominance of women often diagnosed in the 4th decade. ET is often considered the most indolent of the MPNs, and compared to PV and MF, has a more favorable long term prognosis, with an expected survival of 19.8 years (compared to 13.5 years in PV and 5.9 years in PMF). Similarly, cumulative incidence of blast transformation is lower in ET (3.8%) than in PV (6.8%) and PMF (14.2%). ET has also been shown to have a lower incidence of fibrotic transformations than in PV (9.2% vs 21% in one large Italian cohort), however this relationship has not proven true across cohorts [[18]]. Despite indolence for many patients, it has also become clear in the last decade that ET can carry a pronounced symptom burden even in patients with traditionally lower risk disease. In this review, we will discuss the diagnosis, prognosis, and treatment of ET, including in special situations as pregnancy, familial disease, and extreme thrombocytosis.

Prior to the JAK2 discovery in 2005, studies of chromosomal rearrangements suggested that the myeloproliferative neoplasms were likely clonal; however, the mechanism by which a single multipotent stem cell came to overwhelm the bone marrow and peripheral blood was unclear [[19]]. Further, progenitor cells in patients with PV and ET formed colonies in the absence of erythropoietin and were hypersensitive to certain growth factors [[22]], but the molecular underpinnings of these findings had not yet been identified.

The discovery of three major somatic mutations in the past decade had a major impact on the understanding and diagnosis of ET. JAK2 V617F (janus kinase 2, located on chromosome 9p24) was reported first in 2005 and is also the most frequent mutation, observed in ∼50–60% of patients with ET [[10]]. Mutations involving MPL (myeloproliferative leukemia virus oncogene, located on chromosome 1p34), also known as the thrombopoietin (TPO) receptor, are observed in ∼5% of patients and were discovered shortly thereafter [[14]]. JAK2 and MPL (namely MPL W515L/K) mutations both lead to aberrant cytokine-independent signaling via constitutive activation of the JAK-STAT pathway. Mutations in these tyrosine kinases lead to myeloproliferation that is independent of or hypersensitive to growth factor effects, especially TPO [[26]].

The more recent discovery of mutations involving CALR (calreticulin, located on chromosome 19p13.2), detected in ∼20–25% of patients with ET (who lack JAK2/MPL mutations), has provided a novel window into disease pathogenesis [[16]]. Unlike JAK2 and MPL W515L/K, CALR mutations occur almost exclusively in patients with thrombocytosis (ET, PMF, or refractory anemia with ring sideroblasts and marked thrombocytosis), suggesting a strong relationship between the mutation and platelet production [[17]]. A recent series of studies shed light on this association [[27]]. The studies suggest that CALR acts as a chaperone that assists in protein folding, including the TPO receptor (MPL) inside the endoplasmic reticulum. In cell line models, mutant CALR leads to exportation of MPL to the cell surface, leading to inappropriate activation of JAK-STAT (independent of TPO) and the production of mutant megakaryocytes. Further, mutated CALR requires MPL to have this effect, as MPL knockout mice with mutated CALR did not mount excessive thrombocytosis [[27]]. These studies provide valuable insights into the molecular pathogenesis of CALR-mutant MPNs, including the basis for TPO-independent megakaryocytopoiesis.

JAK2, MPL W515L/K, and CALR mutations are typically mutually exclusive in patients with ET, and the mutational profile influences phenotype. Patients with JAK2-mutated ET tend to be older, have higher hemoglobin levels and white blood cell counts, and lower platelet counts and serum EPO levels than patients without the mutation [[30]]. Such patients are also more likely to develop thrombosis [[32], [34]], undergo polycythemic transformation, and respond to hydroxyurea [[36]], suggesting that JAK2-mutated ET patients have a phenotype similar to PV. In contrast, CALR-mutated patients have higher platelet count, lower hemoglobin, and lower leukocyte count than JAK2- and MPL-mutated patients [[37]]. Unlike ET at large, CALR tends to affect more men than women [[37]]. There is a lower risk of thrombosis as well. Finally, recent studies have identified CALR variants that may be associated with different clinical phenotypes. Type 1 (52-bp deletion) and type 2 (5-bp insertion) variants are the most frequent, while type 1-like and type 2-like mutations (that share structural similarities to the type 1 and type 2 CALR variants based on the α-helix content of the mutant C-terminus) have also been identified. Type 1-like mutations may confer greater risk of myelofibrotic transformation, while type 2-like variants are associated with a more indolent clinical course [[39]]. The progression to MF may be related to impaired calcium binding activity that is seen with type 1 mutations [[39]].

While more than 90% of patients with ET will carry one of the three classical mutations, the molecular stimulus of the remaining 10% of patients (termed 'triple negative') remains unknown and is an area of active study. Using whole-exome and next-generation sequencing, investigators have identified novel somatic and germline mutations, including atypical/non-canonical MPL and JAK2 mutations, in triple-negative disease [[40]]. The clinical implications of these mutations are not yet clear. However, such studies also suggest that thrombocytosis may be nonmalignant in some 'triple-negative' patients.

As in MF, additional studies have looked at the impact of non-driving mutations in ET. Such mutations are identified in lower frequencies in ET (and PV), compared to MF. In one study, a median of 6.5 mutations per patient was identified in ET patients (compared to 13 in MF patients); the most common mutations included DNMT3A, TET2, and ASXL1 [[44]]. In another cohort of 181 ET patients, 46% were found to have somatic mutations, including TET2 (13%), ASXL1 (11%), DNMT3A (6%), SF3B1 (5%), CEBPA (4%), and TP53, SH2B3, EZH2, and CSF3R (each 2%) mutations [[45]]. The true impact on prognosis is not yet clear, and use of next-generation sequencing in ET is not routine, or recommended yet. Interestingly, a recent study shows that molecular information may aid in prognostication in patients with ET [[46]]. The authors analyzed 316 patients with ET (183) and PV (133). Through next-generation sequencing of bone marrow or whole blood DNA, they determined that 53% of the ET patients harbored 1 or more sequence variants/mutations other than the three driver mutations (JAK2, CALR, and MPL). SH2B3, SF3B1, U2AF1, TP53, IDH2, and EZH2 were found in 15% of the patients with ET and were labeled 'adverse variants/mutations' as they were associated with inferior survival (median 9 years vs 22 years in patients without the mutations) [[46]].

The prevalence of abnormal cytogenetics at initial diagnosis of ET is less than 10% [[47]], with the most common cytogenetic abnormalities being total/partial trisomy 1q, trisomy 8, and trisomy 9, del(13q) and del(20q). Studies have suggested that cytogenetic anomalies do arise at the time of transformation to acute leukemia [[47]], but not to myelofibrosis [[47]]. In one study, baseline cytogenetic abnormalities did not predict increased transformation to acute leukemia or myelofibrosis, nor did they have a significant impact on overall or leukemia-free survival [[47], [50]]. More data is needed to further delineate the contribution of cytogenetics to transformation and outcomes.

The diagnosis of ET is based on the World Health Organization (WHO) 2016 criteria, which combine laboratory data with morphological features and molecular-genetic findings [[51]]. Like the previous two editions of WHO criteria (in 2001 and 2008), the current WHO guidelines outline four criteria for diagnosing ET: an elevated platelet count (> 450 × 109/L), the presence of characteristic bone marrow (BM) histology, the absence of defining features of any of the other MPNs or myeloid neoplasms, and the presence characteristic clonal genetic information.

In addition to the inclusion of CALR, the 2016 WHO criteria place a stronger emphasis on BM histology/morphology, in particular, in distinguishing 'true' ET from prefibrotic/early PMF (prePMF). These entities are distinguished by certain histologic characteristics, with prePMF demonstrating atypical megakaryocytic morphology and grade 1 reticulin fibrosis (or less) while true ET lacks these features [[51]]. The clinical implications of this distinction are primarily prognostic. PrePMF patients are at increased risk of progressing to overt MF and/or AML over time [[52]], and have increased mortality (Table 1) [[52]]. Gisslinger et al. compared the diagnostic capabilities of the 2016 WHO to the British Committee of Standards in Haematology (BCSH), in which bone marrow morphology plays a minimal role. They noted that the BCSH identified a heterogeneous group of patients as having ET, many of whom actually had prePMF based on bone marrow. Without meticulous use of bone marrow histology, appropriate clinical management may be delayed [[56]]. It should be noted that the distinction on histological grounds can be challenging. From a clinical perspective, those with prePMF are more likely to have splenomegaly, leukocytosis, anemia, and an increased LDH [[52]].

Table 1. Clinical, histologic, and molecular features of 'true ET' versus prefibrotic PMF [44].

Thrombotic events are common complications of ET, and can include both arterial and venous vascular beds, leading to stroke, transient ischemic attack, deep vein thrombosis/pulmonary embolism, hepatic, portal, splenic and mesenteric vein thrombus, and coronary artery ischemia [[3]]. Small vessel occlusion can also occur, often mediated by platelet hypersensitivity, leading to headache, transient visual disturbance, digital ischemia, and erythromelalgia as examples [[57]]. Identifying risk factors for thrombotic complications is essential as treatment is largely based on reducing risk for vascular events, and a few predictive models have been proposed. The International Prognostic Score of thrombosis in Essential Thrombocythemia (IPSET-thrombosis) model (Table 2) identified age >60, thrombosis history, cardiovascular risk factors, and JAK2 status as determinants of thrombosis in a prognostic model that was able to divide patients into low (1.03% per year), intermediate (2.35% per year), and high risk (3.56% per year) of thrombosis [[58]]. The 10-year thrombosis-free survival of the low, intermediate, and high-risk groups was 89%, 84%, and 69%, respectively. The data-set to develop this score was recently re-analyzed, and a revised IPSET-thrombosis model was validated which includes an additional risk group ('very low') in addition to the original three groups. However, this revised score needs confirmation in prospective studies and as such has not yet been adopted clinically [[59]]. As discussed above, the mutational profile has significant bearing on the predisposition to thrombosis. Patients with JAK2 mutations are more likely to develop thrombotic events compared to those with CALR mutations.

Table 2. The IPSET-Thrombosis score for predicting yearly risk of thrombosis.

Transformation to AML and/or post-ET myelofibrosis (post-ET MF) are rare, but significant causes of mortality in ET. In early cohort studies, risk of transformation to acute leukemia was reported at ∼2–3% at 10 years and 5% at 15 years [[60]], but results were highly variable, likely because the distinction between true ET and pre-PMF had not yet been made [[61]]. Thus, more recent studies may be more accurate reflections of the risk of transformation in true ET. When ET is confirmed based on bone marrow morphology, the risk of leukemic transformation is estimated at 0.7% at 10 years and 2.1% at 15 years [[52]]. Multiple risk factors have been identified, including anemia [[61]], platelet count ≥1,000,000 [[52]], increased age [[60]], leukocytosis [[62]], previous thrombosis [[52]], and reticulin grading and bone marrow cellularity [[52], [62]].

Myelofibrosis is a more common complication of ET, heralded by a change in the MPN symptom burden, development of anemia, progressive splenomegaly, and increasing marrow fibrosis [[37], [64]]. The diagnosis of post-ET MF is based on the International Working Group for Myelofibrosis Research and Treatment (IWG-MRT) consensus criteria, which requires both documentation of a previous diagnosis of ET (as defined by WHO criteria), and demonstration of bone marrow fibrosis (grade ≥2 on a 3-point scale or ≥3 on a 4-point scale) [[64]]. In addition, patients must satisfy two of the following additional criteria: anemia and a ≥ 2mg/mL decrease from baseline hemoglobin level, a leukoerythroblastic peripheral blood picture, increasing splenomegaly, increased LDH, or the development of more than 1 constitutional symptom (weight loss, night sweats, or unexplained fever) [[64], [67]]. Risk factors for MF transformation include age [[52]], anemia [[52], [60], [62]], and features of hypercellularity and increased reticulin on bone marrow [[45], [62], [68]]. Similar to leukemic transformation, the risk of post-ET MF increases with time from ET diagnosis and is cumulative, but more significantly so. One of the largest longitudinal cohort studies reported the cumulative probability of myeloid disease transformation to be 3.8% at 10 years, 19.9% at 20 years, and 28.9% at 30 years [[69]]. The median time to development of myelofibrosis in this study was 12.4 years. Ultimately, distinguishing between true ET and prePMF by accurate morphologic diagnosis may have the most significant impact.

Mutation subtype may play a role in risk of transformation to AML or post-ET MF, but more data are needed. For AML, while one study showed that JAK2-mutated patients tended to have a higher risk of transforming to AML than CALR-mutated patients [[39]], these differences did not prove to be statistically significant. In contrast, the majority of studies demonstrate that JAK2, MPL, and CALR mutational status does not appear to impact leukemic transformation or leukemia-free survival (LFS) in ET [[61], [70]] In post-ET MF, the role of driver mutations is also unclear. The most cited studies to date using the revised WHO criteria have not shown that JAK2 mutations predict transformation to MF [[36], [61]], and one large multicenter study even showed that the presence of the JAK2 mutation significantly decreased the rate of MF progression among ET patients [[52]]. Similarly, the data on the predictive or prognostic significance of MPL and CALR mutations in post-ET MF are mixed, with studies showing negligible effect of mutational status [[37]], a trend towards increased transformation in patients carrying a driver mutation but without reaching statistical significance [[38]], and a higher risk of myelofibrotic transformation among patients with type 1-like CALR but not with type 2-like CALR or JAK2 mutations [[39], [70]]. More studies are needed that delineate type 1-like and type 2-like CALR mutations, as this the former prove to be a more significant risk factor for post-ET MF. Next-generation sequencing has identified non-driver mutations, including ASXL1 [[71]] that may be associated with post-ET MF, however more data are needed to verify these relationships.

ET is generally an indolent disease, and the majority of patients' life expectancy is not significantly different from the general population [[72]]. The 15-year survival is ∼75%, with risk factors for inferior survival identified as hemoglobin level below normal, age ≥60, leukocyte count ≥15 × 109/L, smoking, diabetes, thrombosis [[61]], thrombocythemia, and male sex [[72]]. The International Prognostic Score for ET (IPSET) is a prognostic model that was developed to predict both survival and risk of thrombosis at diagnosis, and may be used to set expectations [[74]]. Patients are allocated points based on age ≥60 (2 points), white blood cell count ≥11 × 109/L (1 point), and history of thrombosis (1 point). Patients identified as intermediate risk (total score 1–2) had a median survival of 24.5 years, while patients identified as high risk (total score 3–4) had a median survival of 13.8 years.

The primary goals of therapy in patients with ET are to reduce the risk of thrombosis or hemorrhagic complications. Minimizing the risk of transformation to acute leukemia or post-ET myelofibrosis and controlling systemic symptoms are also important goals [[75]]. While IPSET-thrombosis can provide important prognostic information for the development of thrombosis, the mere presence of the three major predictors of complications (age older than 60 years, previous thrombosis or major bleeding, and platelet count ≥1500 × 109/L) can be used as incentive to classify patients into low- or high-risk categories, thus providing incentive to treat [[3], [69], [75]]. Notably, extreme thrombocytosis (platelets greater than 1500 × 109/L) is an indication for cytoreductive therapy as it can be associated with acquired von Willebrand disease and bleeding complications [[78]]. In such patients with extreme thrombocytosis, measuring the ristocetin activity and von Willebrand antigen is important.

Importantly, symptom burden and the effect on quality of life remain central aspects of the patient experience in ET and should be considered when determining whether and when to initiate treatment. There is substantial evidence that, although often considered to have a more indolent course, ET leads to diminished quality of life, even in those considered to be traditionally lower risk populations [[79]] with patients reporting being moderately to highly symptomatic in ∼40% of cases [[80]]. Common symptoms in ET include fatigue (most prevalent), early satiety, inactivity, concentration issues, and abdominal discomfort [[81]].

Use of aspirin in ET has been typically extrapolated from a higher quality study including patients with PV [[83]]. In fact, a recent study challenges universal aspirin use in ET [[84]]. Among 433 low-risk ET patients, 271 were CALR-mutated, while 162 carried the JAK2 V617F mutation. With over 2215 person-years of follow-up time, 25 arterial or venous thrombotic events were noted, including 14 in anti-platelet treated patients, and 11 in those under observation alone (p = .7). Among CALR-mutated patients, the incidence of thrombosis was 9.7 vs. 6.9 per 1000 person years, in anti-platelet users, versus observed patients (p = .6). In those with JAK2 mutations, the thrombosis rate was higher while under observation, compared to anti-platelet users (21.1 vs 11.6 per 1000 person years; p = .3). In this group, there was a higher rate of venous thrombosis in observed patients, vs. aspirin users (15 vs. 2.9 per 1000 patient years, p = .045). When looking at interactions, there was an increased risk for thrombosis in those with JAK2 mutations and CV risks factors, but this risk was not offset by aspirin use.

Seventeen bleeding events were noted, including 13 in those treated with aspirin. While JAK2 617F did not influence bleeding, CALR-mutated patients had a higher rate of major bleeding on antiplatelet therapy, independent of the platelet count, compared to observed patients (12.9 vs. 1.9 events per 1000 person-years; p = .03). Interestingly, there was an interaction between JAK2 V617F, bleeding, and marked thrombocytosis, but not aspirin use. In summary, this study, representing a large series of low-risk ET patients, suggests that low-dose aspirin increased bleeding risk without reducing thrombosis risk in low risk CALR-mutated patients [[84]]. Therefore, aspirin use should not be universally administered, and rather, risk and mutational status should be considered, along with other risk factors for bleeding.

Unlike in PV, there is randomized trial data to support the use of hydroxyurea, in higher-risk ET patients. In 1995, Cortelazzo et al. reported a significantly lower thrombosis rate when comparing 114 patients randomized to HU versus no myelosuppressive therapy in higher-risk patients [[85]]. Subsequently, in a randomized study comparing 809 patients taking low dose aspirin randomized to HU or anagrelide, the former associated with a lower composite endpoint of arterial thrombosis, venous thrombosis, or death from thrombosis or hemorrhage (p = .03) [[86]]. These findings were observed despite similar platelet count control. In particular, anagrelide associated with a higher rate of arterial thrombosis, drug discontinuation, serious hemorrhage, and myelofibrotic transformation; anagrelide was superior to hydroxyurea in reducing VTE risk.

Such studies serve as a basis for the ELN's recommendation for HU as a front-line cytoreductive, for those ET patients with an indication for therapy [[75]]. However, a recent randomized study compared HU to anagrelide in patients with 'true ET' [[87]]. When comparing 259 high-risk, untreated ET patients, there were neither significant differences in the rates of major or minor arterial/venous thrombosis, nor significant differences in major/minor bleeding, or discontinuation rates between the two groups. Further, there were no transformations to MF, MDS, or AML. This study, using WHO-criteria to identify 'true' ET patients, suggested non-inferiority when comparing anagrelide and HU, as a front-line treatment [[87]]. A recent retrospective study suggested that the mutational profile had no impact on efficacy, though CALR-mutated ET patients appeared more likely to develop anemia, when treated with anagrelide [[88]].

The side effect profiles of anagrelide and HU do differ. Anagrelide is most commonly associated with cardiac complaints (often palpitations), while hydroxyurea is more often associated with hematologic events (cytopenias) and dermatologic side effects. Both medications are associated with gastrointestinal complaints (including diarrhea, nausea, and abdominal pain), and neurologic disorders (including headache, dizziness, and vertigo). In clinical trials, anagrelide was associated with higher rates of cardiovascular adverse events, while HU-treated patients had more leukopenia and minor infections [[87]].

Although ELN consensus recommendations also include interferon (IFN) as a front-line option, in typical hematology practice, this class of medications is less frequently used. While efficacy has been consistently demonstrated, tolerability has been a concern. However, there is somewhat of a renewed interest in use of pegylated-IFN (peg-IFN) in ET, which may have a more favorable tolerability profile. As such, important clinical trials conducted by an international MPN consortium are ongoing, to help define a role for either front-line or salvage treatment (clinical trial identifiers: NCT01259856, NCT01259817). Previously, a phase 2 trial, including 40 ET (23% newly diagnosed) patients (PV patients also included), with a median of 42 months follow-up, reported complete hematological (CHR) and molecular (CMR) responses in 77% and 17% of ET patients, respectively [[89]]. The investigators observed that the presence of additional non-driving mutations (TET2, DNMT3A, and ASXL1) impacted outcomes, since more patients without CMR (56% vs. 30% in those with CMR) had these mutations. Also, when paired samples were available, clonal evolution could be demonstrated in some patients that failed to achieve CMR [[89]]. With regard to tolerability, including PV patients, 27 (32%) patients discontinued from the study, with 17 due to causes directly related to peg-IFN. The most common causes for discontinuation included fatigue (5), depression (4), CVA, infection, dyspnea (2), financial/non-compliance, (2) retinal toxicity, MVA, death (2), neuropathy, neutropenia, nephritis, and non-response [[89]].

With the discovery of CALR mutations in ET, a subsequent study looked at outcomes in peg-IFN treated ET patients with this mutational profile. Among 31 patients with CALR mutations (49% type 1 and 32% type 2 mutations), 16% and 55% received peg-IFN as 1st or 2nd line therapies, and 67% (intention to treat) and 42% achieved CHR, and an overall MR (CMR plus PMR) [[90]]. The authors compared this MR rate to a cohort of CALR-ET patients treated with HU or aspirin, in whom the CALR burden did not change. This study also demonstrated that those ET patients with additional non-driving mutations had poorer molecular responses compared to those with CALR mutations alone. Twelve patients discontinued peg-IFN, including 6 with reasons directly related to the drug; 51% had AE's typical of peg-IFN, including mood changes, asthenia, myalgia, and headache [[90]].

Finally, real-world use of peg-IFN in ET has been reported. In a retrospective study of 118 patients, including 46 with ET, 63% and 15% had CR and PR, respectively [[91]]. There were no molecular analyses, and MR rates were not reported. Including PV patients, 17% discontinued therapy; only 3% had grade 3 adverse events, including myalgia, thyroiditis, and stomatitis. Fatigue was also common, reported in 20% of patients (3% with G3 fatigue). Another real-world series reported a 95% CR rate in ET, with a reduction in JAK2 allele burden from 24% to 10%; discontinuation rates were not explicitly reported [[92]]. Results from studies of novel therapies are included in Table 3.

Table 3. Summary of selected clinical trials involving novel therapies (imetelstat and ruxolitinib) in essential thrombocythemia [93-88].

Hemorrhage is a well-known complication of ET, affecting an estimated 5–10% of patients [[30], [95]]. Gastrointestinal and urogenital bleeding are most common, but intracranial bleeding has been reported as well [[30]]. Patients at greatest risk of bleeding events include those with significant thrombocytosis, acquired von Willebrand disease, previous bleeding events, and concomitant use of aspirin [[30], [97]]. However, more recent studies suggest that major bleeding associated with ET may actually be more specific to pre-PMF. In a 2012 study of bleeding complications among 891 patients with true ET and 180 patients with pre-PMF, the pre-PMF was a significant independent risk factor for bleeding (hazard ratio 1.74), along with leukocytosis, previous hemorrhage, and aspirin. Major bleeding occurred in 6% of patients with ET and 12% of patients with pre-PMF [[30]].

Management of bleeding in ET has not been studied in prospective trials, so there is little data to guide clinical practice. Management primarily involves supportive care. Anticoagulation and anti-platelet agents should be stopped. If acquired von Willebrand's disease is identified on laboratory studies, cytoreductive therapy should be considered, as it has been shown to resolve MPN-associated acquired von Willebrand's disease and normalize platelet counts [[93]]. Despomesssin and tranexamic acid may also be used in cases of severe acute bleeding, while platelet transfusions, von Willebrand factor-containing concentrates, or other factor concentrates should be reserved for more severe or life-threatening bleeding [[93]].

Randomized trials of platelet apheresis in ET have not been conducted. Historically, the removal of platelets by apheresis has been generally reserved for patients with extreme degrees of thrombocytosis leading to serious thrombotic or hemorrhagic events, such as severe or life-threatening organ dysfunction (particularly neurologic or pulmonary) or acute bleeding for whom rapid platelet reduction is needed [[99]]. The Myeloproliferative Neoplasms Working Group (MPN-WG) advocates for the use of acute platelet apheresis in such emergency settings to achieve a rapid reduction in platelet count, noting that cytoreductive therapy, preferably with hydroxyurea, should be initiated shortly after platelet removal due to the transient effect of apheresis [[100]]. It should be noted that platelet apheresis is unlikely to be of benefit in pregnant patients with ET who are deemed to be at increased risk of first-trimester spontaneous abortions [[101]]. Asymptomatic patients are also not likely to benefit.

Given a second peak incidence in the 4th decade, hematologists are challenged with the management of ET during pregnancy. Data from scarce studies of pregnant MPN patients accordingly include more patients with ET than PV or MF. Such retrospective studies have reported: an increased risk for fetal loss (∼3.5 fold, 1st trimester), and other pregnancy complications (IUGR, abruption, pre-eclampsia), possibly mediated by placental thrombosis; that miscarriage history may increase risk for future thrombosis and that rates of maternal thrombosis and hemorrhage were ≤5%; and that the JAK2 mutation has been inconsistently implicated as a risk factor for pregnancy complications with no clear impact from parity, or CBC parameters [[52],[[102]].

More recently, a UK study reported on outcomes when pregnant patients with MPN were followed prospectively [[107]]. Among 58 women, 47 (81%) had ET. Prior to pregnancy, 50% of women had been previously on aspirin, and during pregnancy, 88% were given aspirin, and 38% were prescribed prophylactic low molecular weight heparin. Further, 31% were on cytoreduction (78% interferon). The incidence of miscarriage was 1.7%, and the incidence of maternal complications was 9% with pre-eclampsia; 9% with post-partum hemorrhage; 3% post-partum hematoma (no reported relationship to aspirin or anticoagulation), and no maternal death or thrombosis. Nearly half of the deliveries were induced, and 45% of women underwent C-section; 85% of deliveries occurred after 37 weeks gestation. With regard to fetal complications, the perinatal mortality rate was 17 per 1000 live and stillbirths, and 22% were below the 10th percentile in weight, while 13% required admission to the neonatal care unit [[107]].

With a more complete understanding of the mutational profile, a recent study looked at correlations between driving mutations and pregnancy outcomes in 155 pregnancies in 94 ET patients [[108]]. Of these 94 patients, 62.8% had JAK2 V617F, 20.2% had CALR mutations, 2.1% had MPL mutations, and 14.9% were 'triple-negative'. Seventy-two (47.4%) of pregnancies were complicated by fetal loss (30%), 18 (11.8%) by maternal complications, and 13 (8.6%) by IUGR. The authors did not identify any association between mutational status and fetal loss, as a whole, but JAK2 V617F associated with late pregnancy loss (p = .027; 9.4% of JAK2 V617F, versus 0% in CALR, MPL, and triple-negative pregnancies). In a multivariable analysis, pregnancy prior to 2007 correlated with worse outcome, and there was a trend between CALR and a lower rate of complications (p = .06).

The European LeukemiaNet consensus recommendations outline treatment approaches during pregnancy, noting that it is not clear than any therapy has a specific impact on outcomes. Such strategies include aspirin, and VTE prophylaxis post-partum (or possibly during pregnancy in those with prior thrombosis), and consideration of interferon for high-risk pregnancies, including patients with extreme thrombocytosis [[75]]. A consensus from the Austrian-German Society of Hematology offers similar guidance [[109]]. While recombinant interferon has been recommended, less data are available regarding use of peg-IFN. A recent observational series of 10 pregnancies in 8 women between 2013 and 2015 reported on use of this agent [[110]]. Six patients were high risk (4 with extreme thrombocytosis, 2 with thrombosis); 6 women had 9 pregnancies, with 5 miscarriages and 1 termination. Peg-IFN at a median dose of 270 mcg/month was initiated prior to conception; all but one was treated with aspirin, and 5 were on low molecular weight heparin. Neither grade 3/4 abnormality was reported nor were there discontinuations; platelet counts were significantly reduced and 9 live births were noted (90%); no major bleeding or thrombosis was reported. The authors acknowledge a small sample size, as well as possible confounding by LMWH, but suggest that the safety and tolerance of peg-IFN would be a safe alternative to recombinant IFN in women who need such an approach during pregnancy.

While ET and the other myeloproliferative neoplasms occur sporadically in most instances, familial clustering of both ET and the MPNs has been reported. Predisposition haplotypes, including 46/1 and predisposition alleles have been identified which confer increased risk of developing not only MPN but also JAK2 V617F clonal hematopoiesis [[111]]. In terms of mutations, JAK2 V617F [[112]], CALR [[114]], and atypical MPL [[115]] mutations have all been described in ET as somatically acquired events in familial disorders [[112], [114]]. The differences between familial and sporadic ET in terms of presentation and clinical phenotype are minimal. In fact, patients with familial ET tend to follow similar patterns as patients with sporadic ET. Patients with familial CALR-mutated ET have lower rates of thrombosis and higher platelet counts than those with JAK2 mutations [[17], [37]]. Survival between patients with familial and sporadic MPNs is also similar [[116]].

We recommend that physicians interview all patients with ET to determine whether there is a family history of thrombocytosis or MPN. In familial trees with at least two cases of MPN, some experts recommend obtaining a complete blood count in all apparently healthy relatives in order to identify early asymptomatic phenotypes [[76]]. Regarding treatment, patients with familial and sporadic ET should be managed similarly.

The JAK2 V617F discovery has reinvigorated translational and clinical investigation of the MPNs, including ET. In the past decade, the hematologist's understanding of disease pathogenesis, recognition of clinical heterogeneity, and diagnostic capabilities have improved with a more nuanced understanding of the mutational profile, which includes MPL W515L/K and CALR mutations. The mutational profile influences prognosis, particularly with regard to risk for vascular complications. Perhaps the strongest influence on prognosis comes from a precise distinction of 'true ET' from prefibrotic MF around the time of diagnosis. Further, therapeutic decision-making is influenced by mutational status, and some frequently administered medications, including anti-platelet therapies, may offer more harm than benefit to certain patients. Despite these advances, hematologists would still benefit from a more precise understanding of an individual's thrombotic risk, since often-used predictors are generic, including age and thrombosis history. Another key question is whether or not patients should be treated early on in the disease process, rather than waiting for progression to a higher risk status. Early use of interferon is advocated by some; perhaps results from a phase 3 randomized study will better define the role of this agent. It is clear that though often indolent, ET patients can experience a significant symptom burden from their disease. Though symptom-oriented therapies have been impactful (and approved for use) in PV and MF, use is still investigational in ET, and the role is yet to be defined. Despite its status as a myeloid neoplasm, ET is not yet represented by United States-based guidelines, although this is needed [[117]]. Hematologists/oncologists would certainly benefit from such guidance, regarding management of typical and special situations. Thankfully, such guidelines are imminent, via the NCCN.

Disclosure forms provided by the authors are available with the full text of this article online at http://dx.doi.org/10.1080/10428194.2017.1312371