A Large Proportion of Patients With a Diagnosis of Essential Thrombocythemia Do Not Have a Clonal Disorder and May Be at Lower Risk of Thrombotic Complications

ESSENTIAL thrombocythemia (ET) is characterized by a sustained thrombocytosis with a tendency to thrombosis and hemorrhage. It is predominantly a disease of the elderly, with a median age at presentation of 60 to 70 years, although since the introduction of routine assessment of platelet counts, more patients are being diagnosed at an earlier age.1 The disorder is usually considered to be a clonal disease arising in a multipotent stem cell and to be related to the other clonal myeloproliferative disorders, such as polycythemia vera (PV) and chronic myeloid leukemia (CML). The criteria proposed by the Polycythemia Vera Study Group (PVSG) remain the gold standard for establishing a diagnosis of ET, but these are largely guidelines for the exclusion of other conditions associated with a thrombocytosis.2 PV is excluded either by a hemoglobin of less than 13 g/dL or a normal red blood cell mass in the presence of normal iron stores; myelofibrosis is excluded if collagen fibrosis in the trephine biopsy is less than 33% with absence of leucoerythroblastic features on the blood film; and CML is excluded by lack of cytogenetic or molecular evidence of the bcr/ablrearrangement. A small but significant proportion of patients, 3% to 4%, ultimately transform to an acute leukemic phase, which is in accord with the assumption that ET is a clonal disorder of a hemopoietic stem cell, although this is frequently related to the therapy administered, including radiophosphorous and alkylating agents.1 A recent report has also suggested that the ribonucleotide reductase inhibitor hydroxyurea may cause secondary leukemia in patients with ET.3 

The major diagnostic difficulty encountered is in discriminating ET from an occult cause of a reactive and persistent thrombocytosis (RT). A number of different approaches have been used to try and identify diagnostic features that are specific for ET and that enable exclusion from RT. A variety of platelet defects have been documented, but they are not consistently found in all ET patients and are not sufficiently specific to be of diagnostic use.4 Clonogenic assays have also attracted attention, because in some patients with ET and other myeloproliferative disorders, a proportion of hematopoietic progenitors show growth factor independence. However, there are problems of interpretation with these assays, because a wide variety of culture conditions have been used and spontaneous megakaryocyte colonies have also been found in patients with RT.5 No characteristic karyotypic abnormalities have been associated with ET at diagnosis, although a recent report has shown an increased incidence of abnormalities involving 17p in patients treated with hydroxyurea.3 

A number of groups have shown myeloid clonality using X-linked polymorphic markers in female ET patients. The earliest reports used isoenzymes of the protein glucose-6-phosphate dehydrogenase (G6PD), but the number of patients studied was small due to the low incidence of heterozygosity.6 Later studies used differential DNA methylation patterns of active and inactive X-chromosomes of genes, such as phosphoglycerate kinase (PGK), hypoxanthine phosphoribosyl transferase (HPRT), and the human androgen receptor (HUMARA).7-10 The most recent studies have used expression of transcripts in RNA of three polymorphic genes, iduronate-2-sulphatase (IDS), palmitoylated membrane protein p55, and G6PD.9 

These results must be interpreted with appropriate reference to both the individuals’ constitutive X-chromosome inactivation pattern (XCIP) and their age. Approximately 20% to 25% of hematologically normal females have a constitutively imbalanced or skewed XCIP with greater than 75% expression of one allele.11 An imbalanced pattern therefore can only be interpreted as clonal if control tissue from the same individual shows a balanced or significantly different pattern. T lymphocytes are thought to be the most suitable control for disorders of myeloid cell origin, because they originate from the same pluripotent stem cell, and in younger hematologically normal individuals, they have the same XCIP.12 However, studies have also shown that there is an increased incidence of skewing in myeloid cells of the elderly,13,14 which may not be reflected in T-lymphocyte XCIPs.10,11 In our own study, for example, an extremely skewed pattern with greater than 90% expression of one allele was found in the neutrophils of 33% and T lymphocytes of 9% of hematologically normal females more than 75 years of age, but in only 3% of younger individuals.11 In the few studies of ET patients in which T-lymphocyte XCIPs have been compared with neutrophil patterns, the results have been conflicting. Tsukamoto et al15 showed monoclonal XCIPs in neutrophils of all 13 patients in their study, suggesting that ET is almost invariably a clonal disorder, although we would only consider 6 of these patients to be assessable because of advanced age or constitutive skewing of the T cells. In a study of 46 patients by El-Kassar et al,9 the majority seemed to have monoclonal neutrophils (61%), but 14 patients had polyclonal patterns in both neutrophils and T lymphocytes. Ten of these 14 patients could be studied using RNA polymorphisms. Platelet polyclonality was confirmed in 7 patients, and monoclonal platelets were found in 3 patients, these latter results suggesting that ET could sometimes be restricted to the megakaryocytic line. Such lineage restriction is in marked contrast to other clonal myeloproliferative disorders.16 

In the current study, we have sought to determine the frequency of polyclonal myelopoiesis in female patients with a diagnosis of ET fulfilling the PVSG criteria and to determine in such patients the incidence of megakaryocyte/platelet-restricted clonality. In addition, we have analyzed the precursor/progenitor cell fractions of the bone marrow from 3 patients to ascertain whether the clonal status of these populations differs from the neutrophils in the peripheral blood. Finally, we have compared the clinical features of those patients with definite clonal and polyclonal hemopoiesis to determine the clinical relevance of clonality studies.

Peripheral blood samples were collected from 46 females with a diagnosis of ET according to the PVSG criteria.2 Bone marrow samples were collected from 3 of these patients. Table1 summarizes details of the patients’ age at diagnosis and time of XCIP analysis, platelet count at diagnosis, occurrence of hepatosplenomegaly, history of thrombosis and hemorrhage, and the number of months of follow-up. Mean age of the patients at diagnosis was 53 years (range, 10 to 89 years), mean age of the patients at the time of testing was 58 years (range, 11 to 89 years), and follow-up was 73 months (range, 6 to 260 months). Thrombotic events were recorded in 46%, and hemorrhagic symptoms were recorded in 13%. None of the patients had a family history of a myeloproliferative disorder, and in the 2 females less than 15 years of age who presented, the blood counts of siblings and parents were normal.

Peripheral blood samples were collected from 13 females with an RT, mean platelet counts 801 × 10⁹/L (range, 606 to 1,063 × 10⁹/L). Younger patients were specifically selected to minimize the possibility of age-related XCIP skewing, and the mean age was 39 years (range, 16 to 56 years). The patients had diverse causes of their RT: 3 disseminated malignancy, 2 autoimmune disorders, 2 acute myeloid leukemia in remission, 2 pneumonia, 2 postsplenectomy, 1 pancreatitis, and 1 gastrointestinal bleeding.

Peripheral blood collected into EDTA was centrifuged at 180gfor 15 minutes, then the upper two thirds of the plasma was harvested as platelet-rich plasma and washed twice at 2,000g with calcium and magnesium-free phosphate-buffered saline (PBS) containing 10 mmol/L EDTA. The remaining sample was resuspended to the original volume with PBS and sedimented with 10% dextran (final concentration, 1%) to reduce red blood cell contamination, and then the supernatant was spun through Ficoll (Ficoll Hypaque; Pharmacia Biotech, Uppsala, Sweden). T lymphocytes were purified from the mononuclear layer using anti-CD3–coated magnetic beads (Dynal, New Ferry, UK). Additional red blood cells were removed from the neutrophil fraction by hypotonic lysis. Purity of the CD3⁺ cells was determined by morphological examination and neutrophil and platelet fractions on a whole blood cell analyzer (Sysmex, Milton Keynes, UK).

Bone marrow samples were collected into minimum essential medium (MEM; GIBCO-BRL, Paisley, UK) containing 20 U/mL heparin and 1 mg/mL EDTA and incubated with a 1% solution of Red-Out, a glycophorin antibody (Robbins Scientific, Sunnyvale, CA), at room temperature for 10 minutes. Mononuclear cells were prepared by Ficoll density gradient centrifugation, washed twice with PBS containing 1% albumin and 1 mg/mL EDTA, and then incubated with anti-CD34 multisort beads (Miltenyi Biotech, Camberly, UK) at 4°C for 30 minutes. CD34⁺ cells were selected on a “Variomacs” column (Miltenyi Biotech), and the purity was checked by fluorescence-activated cell sorter (FACS) and alkaline phosphatase–anti-alkaline phosphatase (APAAP) analysis.17 The anti-CD34–coated beads were removed with a release reagent (Miltenyi Biotech), and the cells were washed then further separated into CD33⁻ and CD33⁺subpopulations by adding an anti-CD33-fluorescein isothiocyanate (FITC)–conjugated antibody (Coulter My-9; Luton, Bedfordshire, UK) followed by anti-FITC–conjugated magnetic beads (Miltenyi Biotech). In 1 patient, megakaryocytes were selected from the CD34⁻ population using an anti-CD41-FITC–conjugated antibody (Dako A/S, Glostrup, Denmark) and FITC-conjugated magnetic beads.

DNA was prepared from separated cells using detergent lysis18 and RNA with Trizol (GIBCO, Grand Island, NY).

Samples were screened for heterozygosity of the PGK and HUMARA genes, and clonal analysis was performed as described previously19(Harrison et al, manuscript in press). Samples were also screened for heterozygosity in the IDS, p55, and G6PD genes using mismatched primers. Details of the primers used, their location, fragment sizes, and restriction enzyme digests are given in Table 2. Approximately 1 μg DNA was added to a reaction mix containing 1× Taq polymerase buffer (50 mmol/L KCl, 10 mmol/L Tris-HCl, pH 9.0, 0.1% Triton X-100), 2.25 mmol/L (IDS and p55) or 3 mmol/L (G6PD) MgCl₂, 0.2 mmol/L each dNTP (all from Promega, Madison, WI), and 10 pmol each primer, in a total volume of 19 μL. The mix was heated to 95°C for 5 minutes, held at 85°C while 1 μL containing 0.5 U Taq polymerase was added, and then 35 cycles of 95°C for 30 seconds, 66°C (IDS) or 68°C (p55 and G6PD) for 30 seconds, and 72°C for 30 seconds were performed, followed by a final extension at 72°C for 5 minutes. Polymerase chain reaction (PCR) products were digested with the appropriate restriction enzyme for at least 4 hours, and then electrophoresed through 3% agarose in 1× Tris-Borate-EDTA (TBE) and visualized by ethidium bromide staining under UV light.

Approximately 1 μg RNA was incubated with 250 ng oligo (dT)₁₅ in a total volume of 12 μL at 65°C for 5 minutes, and then cooled to room temperature for 10 minutes. Reaction mix was added to give a final concentration of 1×Taq polymerase buffer, 5.25 mmol/L MgCl₂, 1 mmol/L each dNTP, 20 U RNAse inhibitor, and 5 U avian myeloblastosis virus (AMV) reverse transcriptase (all from Promega), and the mix was incubated at 42°C for 60 minutes followed by 95°C for 5 minutes. Complementary DNA (4 μL) was used in a PCR reaction as described above but with the following modifications: the primers used were all located in exons and differed from the DNA primers for IDS and p55 analysis (Table 2); final MgCl₂ concentrations were 2.25 mmol/L (IDS and p55) or 3 mmol/L (G6PD); 1 pmol ³²P end-labeled primer (IDS/UR, p55/DR, or G6PD/U) was added with the Taq polymerase; 25 cycles of amplification were performed; and annealing temperatures were 66°C (IDS), 64°C (p55), or 68°C (G6PD). PCR products (10 μL for IDS and G6PD and 5 μL for p55) were incubated with the appropriate restriction enzyme (Table 2) for at least 4 hours, and then electrophoresed through a nondenaturing polyacrylamide gel (10% for IDS and G6PD, 6% for p55; 10 × 8 cm, cross-linker ratio 19:1, 1× TBE). The gel was dried and exposed to Hyperfilm-MP (Amersham Life Science, Little Chalfont, UK). Autoradiographic signals were quantified by scanning densitometry (Hoefer Instruments, San Francisco, CA) and reported as the percentage of expression of the lower allele. Each sample was analyzed in duplicate, and results were expressed as a mean of the two values. Results were considered to be significantly different if there was greater than 20% difference between the two values, the limit of our technical variation as determined in previous studies19(Harrison et al, manuscript in press).

XCIP results were obtained on neutrophils and T cells from a total of 46 ET patients. The mean purity of neutrophil preparations was 91% (range, 75% to 99%), and the T-cell fraction contained an average of 97% lymphocytes (range, 82% to 100%). DNA was analyzed in 43 patients using either HUMARA (n = 38), PGK (n = 3), or both (n = 2). RNA from neutrophils, T cells, and platelets was analyzed in 21 patients using either IDS (n = 13), p55 (n = 4), or G6PD (n = 4). Results were also obtained on neutrophils and T cells from 13 RT patients using DNA (HUMARA, n = 10; PGK and HUMARA, n = 3) and from 9 of these patients using IDS-RNA analysis. Myeloid progenitor/precursor populations purified from the bone marrow were analyzed in three ET patients, and purified megakaryocytes from 1 of these patients were analyzed.

We have previously shown that comparable XCIP results can be obtained using DNA analysis and RNA analysis of the IDS gene across the complete spectrum of ratios.20 In the present study, DNA results were compared with RNA results using IDS (n = 20), p55 (n = 3), and G6PD (n = 4) on neutrophils and T cells from 18 ET and 9 RT patients. Overall, the results showed a good correlation for both groups of patients (r = .95; Fig1). However, in 3 ET patients with skewed XCIPs in myeloid cells by both DNA and RNA analysis, the T-cell patterns seemed to be just balanced using DNA (25%:75%, 27%:73%, and 27%:73%), but skewed using RNA (0%:100%, 0%:100%, and 9%:91%, respectively). Morphological examination of cytospins from the cell preparations showed that the CD3⁺ fraction was heavily contaminated with platelets, thereby invalidating the RNA result, and they have been excluded from Fig 1.

XCIPs were determined from neutrophils and T cells of 13 females with RT. In 4 patients, both the neutrophil and T-cell XCIPs were imbalanced, suggesting constitutive skewing, and these patients were thus unevaluable. In 9 patients, the DNA XCIP results from the neutrophils and T cells were all balanced. In 7 of these 9 patients, the results were confirmed using RNA analysis including platelet XCIPs, which did not differ from the neutrophil pattern. As expected, therefore, all of the evaluable patients with RT had polyclonal hematopoiesis.

On the basis of the neutrophil and T-cell XCIP results obtained, the 46 ET patients have been divided into three groups (Table3).

This group contains 23 patients (50%) in whom clonality analyses were uninterpretable, either because of constitutive skewing of the hematopoietic cells (ie, >75% expression of 1 allele), or because, although the neutrophils and not the T cells were skewed, the patients were over 65 years of age at the time of XCIP analysis. Seventeen patients (1 through 17, Table 3) had a skewed XCIP in both neutrophils and T cells and less than 20% difference between the two values for each patient (Fig 2A). Six patients (18 through 23, Table 3) had apparent monoclonal neutrophils, but were less than 65 years of age. Platelet RNA from 12 patients in this group was also analyzed, and the XCIPs were the same as the neutrophil XCIPs, ie, skewed, with less than 20% difference between the neutrophil and platelet values for each patient. Altogether, 16 of the patients in this group were greater than 65 years of age when XCIP analysis was performed.

This group of 10 patients (24 through 33, Table 3), all less than 65 years of age, had monoclonal myelopoiesis with a skewed neutrophil XCIP, a balanced T-cell XCIP, and greater than 20% difference between the two results for each patient (Fig 2B). This represents 22% of the total patients studied and 43% of those with interpretable results. DNA analysis was performed on all patients and was confirmed with RNA analysis in 3 patients. Eight of these 10 patients had an extremely skewed myeloid XCIP with greater than 90% expression of one allele. Platelet RNA was analyzed in 3 patients, and in each case, the XCIP was again skewed in agreement with the neutrophil results.

The neutrophil and T-cell XCIPs of 12 patients (34 through 45, Table 3) were both balanced and not significantly different from each other, indicative of polyclonal myelopoiesis (Fig 2C). Results from all these patients were obtained using DNA analysis and were confirmed by RNA analysis in six patients. In patient no. 46 (Table 3), the neutrophil and T-cell XCIPs had similar values, whether determined by DNA or RNA. The DNA results were marginally skewed with neutrophil and T-cell values of 81%:19% and 79%:21%, respectively, but the patient has been assigned to group C (rather than A), because the values for the RNA analysis were clearly balanced at 59%:41% and 66%:34%, respectively. These 13 patients represent 28% of the total and 56% of the evaluable patients. Five of these patients were older than 65 years of age at the time of XCIP analysis. Platelet RNA was studied in 7 patients, and XCIPs were found to be balanced in agreement with the neutrophil RNA data—this includes patient no. 46, justifying their inclusion in group C. In another patient who was not informative for an RNA polymorphism (no. 42, Table 3), purified megakaryocytes were prepared from a bone marrow aspirate, and the DNA XCIP was balanced and similar to the neutrophil and T-cell pattern (Fig3A and Table4).

Myeloid progenitors were purified from bone marrow of 3 ET patients and XCIPs assessed using HUMARA analysis. In 1 patient with a balanced T-cell pattern and skewed neutrophil pattern (no. 24), the XCIPs from bone marrow cells, including the CD34⁺CD33⁻and CD34⁺CD33⁺ fractions, were also skewed (Fig3 and Table 4). In 2 patients with balanced neutrophil and T-cell XCIPs (no. 35 and 42), the CD34⁺CD33⁻ and CD34⁺CD33⁺ subpopulations also had balanced XCIPs.

The clinical records of patients with monoclonal (group B, patients no. 24 through 33, Tables 1 and 3) and polyclonal disease (group C, patients no. 34 through 46) were compared to determine whether there were any differences. Parameters examined were age and platelet count at diagnosis, length of follow-up, evidence of hepatomegaly or splenomegaly, and occurrence of thrombotic or hemorrhagic complications (Table5). Statistical analysis showed no difference in age, platelet count, length of follow-up (Mann-U-Whitney), incidence of hepatosplenomegaly, or hemorrhagic complications (Fisher’s exact test). However, thrombotic complications were more common in the monoclonal patients, and this difference was statistically significant (P = .039, Fisher’s exact test). Thrombotic events were documented in 6 of the 10 monoclonal patients, and comprised 1 cerebrovascular accident, 3 transient ischemic episodes, 2 incidences of erythromelalgia, 2 deep venous thromboses, 1 pulmonary embolus, and 1 splenic infarct. Two of these patients had more than one event, and of these, one had multiple events, even when she had a normal platelet count. In the remaining 4 patients, the thrombotic episodes occurred at presentation. Two of the 13 polyclonal patients also had thromboses preceding the diagnosis of ET; 1 had recurrent miscarriages caused by placental infarction, and the other had a deep venous thrombosis after surgery for peripheral vascular disease. Interestingly, both of these patients had additional risk factors for thrombosis. The patient who had recurrent miscarriages was heterozygous for the factor V Leiden mutation and had anticardiolipin antibodies. The deep venous thrombosis of the other patient occurred immediately postoperatively.

None of the patients in this series has transformed to myelodysplasia or acute myeloid leukemia (AML), although 2 of the 10 patients in group B with clonal myelopoiesis have progressed to myelofibrosis compared with none of the 13 patients in group C with polyclonal disease.

This study illustrates the value of using DNA and RNA XCIPs to determine clonality in hematological disorders, but also shows some of the shortcomings of this technology. The results from only half of the 46 female patients were considered evaluable, and, because X chromosome inactivation only occurs in females, the diagnostic potential of XCIP analysis is thus limited to approximately one quarter of all patients with ET. Non-evaluability arises from either constitutive skewing of the hematopoietic cells, which is shown by analysis of the T cells, or from advanced patient age when acquired skewing in the myeloid cells may occur. We have chosen an arbitrary age cut-off, based on our previous studies11 of patients less than 65 years of age, to be able to diagnose clonality/oligoclonality of the myeloid cells. This does not mean that all analyses in patients greater than 65 years of age are unrewarding, because a polyclonal result is still interpretable.

ET has generally been considered to be a malignant disease, and some patients (group B) clearly have monoclonal neutrophils and platelets. More surprising is the high incidence of polyclonal neutrophils in this series—over half of the interpretable cases. The 13 cases designated as having polyclonal hematopoiesis (group C) were initially studied with DNA analysis. There is a possible concern that, in a malignant disease, there could be abnormal DNA methylation leading to changes in the pattern of methylation-sensitive enzyme digestion. In a small proportion of cases of AML, for instance, we have observed spurious XCIPs due to aberrant methylation at the HUMARA locus when digested with Hha I.19 However, it is unlikely that minor methylation changes could lead to expression of an inactivated X-chromosome allele and thus confound analysis of clonality at the RNA level, and it was important, therefore, that the XCIPs of neutrophils and T cells could also be examined using RNA in 7 patients. In 6 patients, the RNA results were fully concordant with the DNA results, and in the 1 patient in whom there was some discrepancy between the DNA and RNA results, it was the RNA result that indicated polyclonality (as opposed to an uninterpretable DNA result).

El-Kassar et al9 have suggested that in some cases of ET with apparent polyclonal hemopoiesis, there is restriction of the clonal population to cells of the megakaryocytic lineage. However, we could find no evidence for lineage restriction in the 8 patients in whom we could study platelets or megakaryocytes; the XCIP results of neutrophils, T cells, and platelets or megakaryocytes for each individual were all within 20% of each other and balanced. Conversely, the platelet XCIPs of 15 patients with a skewed neutrophil XCIP were all also skewed. We conclude that ET caused by a clonal disorder restricted to the megakaryocytic-platelet lineage is rare.

Asimakopoulos et al21 have reported that the bone marrow is clonal in PV with 20q−, but that only neutrophils lacking the 20q− may be released into the peripheral blood, although they are still clonal. We therefore studied bone marrow cells in 2 patients (no. 35 and 42, Table 4) with polyclonal peripheral blood neutrophils. Not only were whole marrow cells examined, but also early progenitor/precursor cell populations (CD34⁺CD33⁻ and CD34⁺CD33⁺ cells). The results in all populations were concordant with the neutrophil results, suggesting that such selection for release of only normal neutrophils into the blood does not account for the neutrophil polyclonality in ET. Furthermore, even if neutrophil selection occurred, if a clonal disorder accounted for the thrombocytosis that led to the diagnosis of ET, then the platelets would still be skewed. Neither of these patients was informative for any of the RNA polymorphisms, but polyclonality of the megakaryocytes was shown in one of them (patient no. 42, Fig 3A).

In patients with PV, at least during the early stages of the disease, the bone marrow appears to be clonal, but the progenitor cell compartment still contains a significant population of normal polyclonal cells.22 We therefore took advantage of the opportunity to study the XCIPs of different marrow fractions in 1 patient (no. 24) with definite clonal neutrophils and platelets. In this patient, the early cell populations were as skewed as the neutrophils, indicating that the majority of cells in these populations were also involved in the malignant clone (Fig 3B). It should be acknowledged that XCIP analysis is not a sensitive technique for detecting a small population of polyclonal cells in a monoclonal background. The patient analyzed was 12 years old at the time, 2 years after diagnosis, and had received no specific treatment. XCIP analysis should be performed on primitive myeloid cells from more patients in this category, before definitive conclusions can be drawn with regard to the absence of a small proportion of polyclonal precursors.

Other studies have also shown polyclonal hematopoiesis in patients diagnosed with ET,9 although not at such a high frequency as shown here. In some studies, the frequency of monoclonality has probably been overestimated because of failure to take into account constitutive skewing and acquired skewing associated with old age.11 It is also possible that there is a difference in the diagnostic criteria used to diagnose ET in different centers, although all of our patients had thrombocytosis lasting more than 6 months and fulfilled the PVSG criteria.2 Furthermore, it should be noted that the median follow-up in this series as a whole was 61 months (range, 6 to 260 months), and in those with polyclonal neutrophils (group C), it was 32 months (range, 6 to 116 months). In no case had an occult reactive condition declared itself during this period.

These results therefore suggest that ET, as diagnosed by the standard clinical and laboratory criteria, is a heterogeneous disorder in terms of clonality. The pathogenesis of sustained polyclonal thrombocytosis in the absence of an obvious systemic inflammatory or malignant disease is unknown. Two mutations have recently been described in the thrombopoietin gene in cases of “familial ET,” and it is unlikely that these patients would have clonal hematopoiesis.23,24Similar polymorphisms or acquired mutations might account for some cases of apparent ET with polyclonal myelopoiesis, and investigation of this possibility will be worthwhile. Congenital mutations in the thrombopoietin receptor c-mpl could also lead to polyclonal thrombocytosis, comparable with the erythropoietin receptor mutations described in cases of primary familial and congenital polycythemia.25 However, to date, no such mutations have been detected in patients with ET.26,27 Retroviral particles have been identified within platelets of patients with documented ET.28,29 It is possible that this could result in a polyclonal proliferation of platelet precursors, and this also warrants further investigation.

The important clinical issue is whether it is valuable to distinguish, whenever possible, monoclonal from polyclonal disease. No patient in either group has transformed to myelodysplasia or AML, although 2 of the 10 patients with monoclonal disease have developed myelofibrosis, compared with none of those with polyclonal disease. These numbers are too small to draw valid conclusions, but the issue is an important one, because the risks of developing myelofibrosis or acute leukemia could modify which therapies were thought to be most appropriate in any particular patient.

The patients with monoclonal disease could not be discriminated from those with polyclonal disease in terms of their platelet count or age at diagnosis, extent of follow-up, the presence of splenomegaly, or the occurrence of hemorrhagic complications. However, there was a statistically significant difference in the incidence of thrombosis associated with monoclonal disease, with thrombotic events occurring in 6 of 10 patients compared with only 2 of 13 with polyclonal disease (P < .05). The incidence of thrombosis has been shown to be influenced by age and elevated platelet count,30 but our two groups of patients do not differ in these respects, nor is the duration of follow-up different. The observation of increase in incidence of thrombotic complications in patients with monoclonal disease is thus important, and if confirmed in larger studies, might influence patient management. We therefore believe that further larger series of patients with a diagnosis of ET should be analyzed for clonality so that the natural history, in particular the incidence of thrombotic complications, of the different forms of the disease can be determined.