Chapter 137:  Essential Thrombocythemia

Epidemiology

ET is an uncommon disorder; estimates of its age- and gender-adjusted incidence range from 0.6 to 2.5 cases per 100,000 persons per year.^15–17 Among younger people (ages 30–50), the disease appears to be more common in women, but this gender imbalance is not as clearly found in other age groups, and case ascertainment bias due to the symptom of recurrent fetal loss unique to young women may confound such analyses. No clear environmental risk factor has been identified.^17,18 The incidence of ET (and also of PV and MMM) appears to be increased among Ashkenazi Jews, and the median age at diagnosis for all three CMPD is approximately 60 years.¹⁹ Familial clustering of ET has been described but is exquisitely rare; some familial cases are associated with mutations in the thrombopoietin (TPO) gene that result in increased TPO production (a megakaryocyte growth factor) with consequent megakaryocyte hyperstimulation.^20–23 Mutations in TPO and its receptor, c-Mpl, have been sought but not yet found in the more common nonfamilial cases of ET.²⁴

Diagnosis

The chief clinical challenge with respect to ET diagnosis is differentiating genuine, autonomous, illegitimate thrombocytosis from the myriad causes of “secondary” or “reactive” thrombocytosis (RT). Platelets, like fibrinogen, act as an acute-phase reactant; this nonspecific procoagulant response may have evolutionary significance by promoting organism survival in the setting of serious stress. As a consequence, in more than 80% of unselected cases of thrombocytosis, the elevated platelet count is simply a reaction to the presence of a nonmyeloid disorder.^25,26 The thrombocytosis associated with inflammatory, infectious, and malignant conditions is thought to be due to the action of megakaryocyte-stimulatory cytokines such as interleukin-6 (IL-6), whereas the peculiar thrombocytosis seen occasionally in iron-deficient states remains poorly understood and may be mediated by erythropoietin (EPO) cross-stimulation of precursor cells committed to platelet production.^27,28

Often RT is obvious, but in some instances the distinction between ET and RT may be murky since several potential causes of RT are occult. The degree of thrombocytosis is of little help here, as reactive thrombocytosis can result in platelet counts above 2,000,000μ L.²⁶ Platelet morphology and platelet functional differences are also not reliable for separating ET from RT. More recently, serum TPO levels have also been found to be of little use in distinguishing ET from either RT or other non-ET causes of primary thrombocytosis.^29,30

There is an increased risk of thrombosis and bleeding associated with ET, but the risk appears to be much less with RT, so the ET-RT distinction is quite important.²⁵ Furthermore, even when RT appears to have been excluded, yet another diagnostic challenge remains: It is important to distinguish ET from other myeloid disorders that may have a very different prognosis and require different therapy.

If the presence of RT is not obvious, serum ferritin and C-reactive protein (CRP) may be diagnostically useful.²⁷ Elevated CRP levels suggest RT and may prompt a more thorough search for an occult inflammatory stimulus. Of course, an elevated CRP does not strictly rule out ET, for a patient may potentially have ET and a comorbid inflammatory condition. Ferritin levels likewise must be interpreted with caution; although a low value is consistent with iron deficiency and a high value suggests RT, neither definitively excludes ET.

Surgical hyposplenism is usually obvious from a patient's history, but functional hyposplenism due to amyloidosis, celiac sprue, or another cause may not be so blatant.³¹ Therefore, examination of a blood smear searching specifically for Howell-Jolly bodies should be done during the initial evaluation of each patient with chronic thrombocytosis.

Once these steps have been taken to try to rule out RT, ET increases in likelihood, but the diagnosis of ET can be rendered only when peripheral blood and bone marrow findings do not suggest another chronic myeloid disorder.³² ET look-alike disorders include some cases of MDS, CML, and the other two non-ET CMPD (ie, PV and MMM).

Most cases of MDS have fairly striking erythroid dysplasia and/or morphologic changes in the granulocytic lineage, features that are not consistent with ET. If dysplasia is limited to cells of the megakaryocyte lineage, however, the MDS-ET distinction may be more difficult because atypical megakaryocytes are often associated with ET. The most difficult distinction between MDS and ET arises in the setting of the so-called 5q-syndrome, which is often associated with thrombocytosis, minimal evidence of dysplasia in the erythroid and granulocytic lineages, a clinically indolent course, and an increased number of small, hypolobated megakaryocytes.^33,34 In such cases, cytogenetics should be definitive.

There are some individuals who consider Philadelphia chromosome-positive ET to be a unique diagnostic entity, but most clinicians are not convinced and would consider anyone with a chronic myeloid disorder and bcr-abl positivity to have CML.^35–38 If conventional cytogenetic analysis does not reveal a Philadelphia chromosome, it is prudent to obtain peripheral blood or bone marrow fluorescent in situ hybridization (FISH) studies at least once to exclude the possibility of karyotypically occult CML.

The distinction between ET and PV may be subtle. An outright elevation of hematocrit and a substantial increase in hematocrit from a measured baseline value are generally not associated with ET. As discussed in the section on PV, endogenous (ie, spontaneous, autonomous, erythropoietin-independent) erythroid colony growth, low serum erythropoietin level, and aberrant expression of the polycythemia rubra vera (PRV)-1 gene are associated with PV, but these findings are not specific to that disease and do not rule out ET.^39,40 In the past, red cell mass (RCM) measurement was commonly used to distinguish PV from ET in equivocal cases, but it is now widely recognized that hematocrit correlates quite closely with true RCM, so RCM analysis is being ordered less often.^32,41–43 For some cases that are initially challenging, a distinction between PV or ET may become more clear over time.

Distinguishing ET from MMM is usually not difficult, but there can be exceptions. The heavy marrow fibrosis characteristic of MMM is an atypical finding in ET, although mild (grade 1 or 2) reticulin fibrosis can be found in approximately 15% of ET cases.⁴⁴ Massive splenomegaly and a leukoerythroblastic peripheral blood picture are also much more typical of MMM than ET. However, the cellular phase of MMM, characterized by marrow hypercellularity with florid atypical megakaryocytic hyperplasia but not necessarily heavy fibrosis, can mimic ET in presentation.⁴⁵ Abnormal megakaryocyte clusters on the bone marrow biopsy (Figure 137-2)-megakaryocytes are solitary creatures and do not normally cluster-are typical of ET, and the cellularity is generally not as dense in ET as in cellular phase MMM.^44,46,47 Clonal karyotypic abnormalities are more typical of MMM and other myeloid disorders than they are of ET. Although patients with ET can have cytogenetic abnormalities at diagnosis, fewer than 10% of patients have them.^36,48 As with PV and ET, sometimes a distinction between MMM and ET is not possible initially, and an expectant, observant approach is appropriate.⁴⁹

Pathogenesis

As mentioned above, clonal myeloproliferation involving the megakaryocytic lineage and sometimes other myeloid lineages (even in cases where the white count and hematocrit are normal) is demonstrable in the majority of female patients with ET via X chromosome-linked DNA or gene product analysis.^13,50,51 Surprisingly, a portion of ET patients were recently shown to have polyclonal hematopoiesis, and these patients appear to have less thrombosis risk than those with “monoclonal” ET.^12–14 It is important to keep in mind that clonality is not equal to malignancy; for example, apparently monoclonal X chromosome inactivation patterns (XCIP) can be demonstrated in the hematopoietic cells of some healthy elderly controls.^52,53 The prevalence of this finding depends on what degree of skewed X inactivation (eg, 75% of identical X-chromosomes inactivated vs 90%) is accepted as consistent with monoclonality.⁴⁸

In ET, megakaryocyte proliferation and platelet production are apparently autonomous, and normal regulatory pathways are apparently defective. The mechanism of this out-of-control, prodigious proliferation remains obscure in most cases. Megakaryocytes from ET patients are not inhibited by antibodies against key growth-and-differentiation cytokines, including interleukin-3 (IL-3), IL-6, granulocyte-monocyte colony-stimulating factor (GM-CSF), and TPO.^54,55 TPO and its receptor, c-Mpl, are thought to comprise the major feedback loop controlling megakaryocyte growth and development, but TPO and c-Mpl dynamics in ET and related disorders appear to be complex.^24,48,56 TPO levels are not reproducibly different among normal marrow, ET, and RT cases, while c-Mpl expression is often (but not always) markedly downregulated in ET (Figure 137-3). However, c-Mpl expression is also downregulated in other CMPD, so its diagnostic utility for ET is limited.^57–61 The specific cytokines driving megakaryocyte proliferation in ET have not been worked out in detail, but megakaryocyte progenitors from ET patients may display unexplained hypersensitivity to both IL-3 and TPO.^62,63

Clinical Manifestations

ET is often an incidental finding, a simple sequela of the fact that hemograms are ordered for a great diversity of clinical indications. At least half of patients with ET are asymptomatic at presentation, and with appropriate therapy many can remain asymptomatic throughout the course of their illness.^44,64,65 At presentation, vasomotor symptoms are found in about 25% to 50% of ET patients, major thrombosis in roughly 15%, and major hemorrhage in approximately 5%. In contrast to other CMPD, where splenomegaly is very common, less than 25% of ET patients have palpable splenomegaly at the time of diagnosis.⁴⁴

Vasomotor disturbances (eg, headaches, lightheadedness, visual symptoms including blurring and scotomata, palpitations, chest pain, erythromelalgia, and distal paresthesias) are troublesome but not generally life threatening.^44,64,65 The proximate cause of such symptoms remains poorly defined; speculation has focused on abnormal platelet-endothelium interactions in the microvasculature, which can be associated with transient thrombotic occlusion. Various platelet products such as thromboxane A2 are vasoactive, and some of these probably also play a role. Erythromelalgia is the most dramatic vasomotor symptom, characterized by erythema, warmth, and pain in distal extremities; this symptom is rare and not specific for ET (Figure 137-4).⁶⁶ The presence of vasomotor microvessel disturbances does not clearly predict hemorrhage or thrombosis in large vessels.

There are several potentially life-threatening complications of ET: large-vessel thrombosis (both arterial and venous), hemorrhage, and transformation of the disease into either a fibrotic phase resembling MMM or acute myeloid leukemia (AML).^44,64,65

Arterial thrombosis can lead to cerebrovascular events, cardiovascular ischemia, organ infarction, and digital gangrene. Venous thrombosis in ET occurs both in sites common to other thrombotic diatheses (pulmonary embolism and lower extremity deep venous thrombosis) but also in more unusual sites (eg, cerebral sinus thrombosis, retinal vein thrombosis, and hepatic and portal vein thrombosis).⁶⁴

Mucocutaneous bleeding (epistaxis, gingival bleeding, ecchymoses, and petechiae) can be a major nuisance and is the most common hemorrhagic problem in ET.⁶⁷ Because epistaxis and easy bruising are very common in the general public, it can be difficult to assess the contribution of ET to these symptoms in afflicted patients. If careful control of the platelet count controls the mucocutaneous symptoms, it is reasonable to form a post hoc link. Serious hemorrhage in ET is most common in the gastrointestinal tract; the widespread use of aspirin (ASA) to reduce vasomotor symptoms and prevent thrombosis may exacerbate this tendency.⁶⁸ Hemorrhage also occurs in the central nervous system (CNS) and the retina, but such events are, fortunately, uncommon. Paradoxically, patients with extreme thrombocytosis may be at special risk for bleeding, in part related to the development of an acquired von Willebrand factor (vWF) deficiency related to platelet adsorption of large multimers of vWF, a phenomenon that can be seen with thrombocytosis of any cause.^69,70

Fibrotic and leukemic evolution of ET are rare events (< 5% of patients) during the first 10 years after diagnosis.^44,64,65 High-quality long-term studies evaluating the incidence of these problems in the second and third decades of disease are not yet available.

Prognosis and Therapy

When considering therapy for ET, it is important to keep in mind two facts: (1) ET is generally an indolent disorder, with a life expectancy in the first decade of the disease quite close to that of an age- and gender-matched control population, and (2) no treatment to date has been shown to influence overall survival.^44,71 Therefore, primum non nocere is a rule that should not be forgotten in the treatment of patients carrying the diagnosis of ET.

Vasomotor disturbances in ET can often be relieved with low dose ASA; 81 mg per day is usually enough.^44,64,65,72 When ASA does not alleviate these symptoms, it is reasonable to add a platelet-lowering agent (Table 137-1).

There are only a few randomized trials assessing the most appropriate therapy for diminishing the thrombotic risk in patients with ET. While the overall lifetime incidence of thrombosis in ET is reported to be on the order of 20% to 40%, major and significant thrombotic events occur in only 10% to 20% of patients.^44,64,65 Therapy should be guided by an individualized assessment of thrombotic risk. Two clinical parameters appear to be quite important in making this judgment: a history of thrombosis and age above 60 years.^{44,65,68,73–75} Based on the presence or absence of these two risk factors, patients with ET have been grouped into low-risk, high-risk, and indeterminate-risk groups (Table 137-2).⁷⁶ This classification will need to be refined as more data on thrombotic risk become available, but it is useful to guide management at present. It is likely that other factors also significantly affect the intensity of the clotting diathesis in ET patients, such as cigarette smoking, male gender, and the coexistence of plasma hypercoagulability due to abnormalities like the Factor V “Leiden” mutation, but these have not yet been subject to multivariate analysis.^77–79 More recently, monoclonal hematopoiesis in ET has also been associated with a higher thrombotic risk than polyclonal hematopoiesis.¹²

The use of ASA (81 to 325 mg/day) to decrease thrombotic risk in all classes of patients with ET seems reasonable, especially if there are other compelling indications for its use (eg, coexisting cardiovascular disease).⁸⁰ Data are lacking on the prophylactic utility of other anticoagulants, such as coumadin and heparin, in higher-risk groups. The more common approach for high-risk ET patients is to try to lower the platelet count into the normal range. One randomized trial demonstrated a 20% absolute risk reduction (24% to 3.6%) in thrombotic events with the use of hydroxyurea in ET patients in a high-risk group.⁸¹ Other cytoreductive agents have not yet been evaluated in a randomized fashion. It is not clear precisely how low the goal platelet count should be in the high-risk group, but < 400,000/μ L seems to be a reasonable target and is supported by retrospective data.^82,83 For low-risk and indeterminate-risk patients, it is not clear that platelet-lowering agents are of any benefit.

Table 137-1 summarizes information regarding currently used platelet-lowering agents in CMPD. Hydroxyurea should be the first choice in ET because of the high-quality evidence supporting its use.⁸¹ It is reasonable to substitute anagrelide or pipobroman (where available) for patients intolerant to hydroxyurea.^84,85 Interferon-α (IFNα ) is another alternative but its well-known toxicity limits its use, especially since many high-risk patients are elderly and may have an especially difficult time tolerating the necessary doses.⁸⁶ Radiophosphorus (³²P) and alkylating agents like chlorambucil may still be useful “backup” regimens but are clearly leukemogenic in PV, so caution is indicated. The leukemogenicity of hydroxyurea is controversial; the risk, if there is any, appears to be very small (this controversy is discussed in more detail below, under PV).^87,88

Major bleeding (ie, enough blood loss to drop hemoglobin level or cause bleeding in a critical organ, like the CNS) occurs in less than 10% of ET patients.^44,64,65 Extreme thrombocytosis (eg, > 1 million/μL) appears to be a risk factor for bleeding, in part because of the acquired von Willebrand factor deficiency described above.⁶⁹ Therapy designed to lower the platelet count and indirectly raise vWF levels may be indicated in the presence of a substantial reduction in large vWF multimers.⁷⁰ In an emergency situation, platelet apheresis is the fastest way to lower the platelet count, but its utility has not been proven conclusively.

Aggressive approaches like bone marrow transplantation should only be considered in exceptional cases.

Special Considerations: Pregnancy

ET is associated with first trimester spontaneous abortions in at least one-third of patients who become pregnant, and recurrent fetal loss occasionally is the presenting symptom of ET.⁸⁹ There does not appear to be a postpartum thrombosis risk.⁹⁰ There is no clear association between the increased risk of spontaneous abortion and either the degree of thrombocytosis or the concurrent use of aspirin, nor is there any clear benefit from prophylactic platelet apheresis.⁸⁹ High-risk pregnant women with ET (ie, women with previous thrombosis—pregnancy in women over 60 is, at least at present, quite rare) require cytoreductive therapy just like other high-risk patients. There is anecdotal evidence of the safety of IFNα in pregnancy, but no controlled data in ET.⁸⁶ Hydroxyurea and anagrelide do cross the placenta, may be teratogenic, and are therefore contraindicated in pregnancy. The teratogenicity of hydroxyurea, however, does not appear to be severe enough to justify elective abortion in cases of inadvertent early fetal exposure.⁹¹