Objective: We report an extension study of patients with essential thrombocythaemia (ET) in the Hungarian Myeloproliferative Neoplasm (HUMYPRON) Registry, which demonstrated that over 6 years anagrelide significantly decreased the number of patients experiencing minor arterial and minor venous thrombotic events (TEs) vs hydroxyurea+aspirin. Methods: Data on patients with ET were collected through completion of a questionnaire developed according to 2008 WHO diagnostic criteria and with regard to Landolfi, Tefferi and IPSET criteria for thrombotic risk. Data were entered into the registry from 14 haematological centres. TEs, secondary malignancies, disease progression and survival were compared between patients with ET treated with anagrelide (n = 116) and with hydroxyurea+aspirin (n = 121). Results: Patients were followed for (median) 10 years. A between‐group difference in the number of patients with TEs was observed (25.9% anagrelide vs 38.0% hydroxyurea+aspirin; P =.052). Minor arterial events were more frequently reported in the hydroxyurea+aspirin group (P <.001); there were marginally more reports of major arterial events in the anagrelide group (P =.049). TE prior to diagnosis was found to significantly influence TE incidence (P >.001). Progression‐free survival (P =.004) and survival (P =.001) were significantly increased for the anagrelide group vs hydroxyurea+aspirin. Conclusions: Anagrelide reduced TEs, and increased progression‐free and overall survival vs hydroxyurea+aspirin over (median) 10 years.
Keywords: anagrelide; essential thrombocythaemia; HUMYPRON registry; hydroxyurea + aspirin; progression; risk factors; thrombosis
This was a 10‐year study of patients with essential thrombocythaemia (ET) in the Hungarian Myeloproliferative Neoplasm (HUMYPRON) Registry treated with anagrelide vs hydroxyurea+aspirin.
Anagrelide reduced thrombotic events, and increased progression‐free and overall survival vs hydroxyurea+aspirin over (median) 10 years.
Anagrelide is often recommended only as second‐line treatment in ET; the present data indicate that anagrelide is a valid first‐line treatment option with potentially superior outcomes to hydroxyurea+aspirin in long‐term treatment.
Essential thrombocythaemia (ET) a Philadelphia chromosome‐negative chronic myeloproliferative neoplasm (PN‐MPN) is a clonal disorder arising from malignant transformation of hematopoietic stem cells.1
PN‐MPNs are characterised by mutually exclusive acquired somatic mutations in the Janus kinase 2 (JAK2) gene, calreticulin (CALR) gene and the myeloproliferative leukaemia virus oncogene (MPL), which are key drivers of the myeloproliferative phenotype.2‐6JAK2V617F is the most frequent JAK2 gene mutation, associated with 50%–60% of ET cases.4,5,7CALR and MPL mutations have been reported in 15%–32% and 3%–4% of ET cases, while 10%–20% of patients are triple‐negative having none of these mutations.4,5,7
In addition to the WHO diagnostic criteria for ET,1,7 patients with ET can present with clonal thrombocytosis, leukocytosis, microvascular symptoms, thrombotic and bleeding complications, and risks of myelofibrotic or leukaemic transformation.5,6 Reported frequencies of disease transformation from ET to myelofibrosis are 0.8%–4.9% at 10 years and 4%–11% at 15 years, and from ET to acute myeloid leukaemia (AML) 0.7%–3% at 10 years and 2.1%–5.3% at 15 years.8 For patients with ET, life expectancy is compromised when compared with an age‐ and gender‐matched population, with survivals of 33 and 20 years in patients younger and older than 60 years of age, respectively.9
In ET, the presence of JAK2V617F or increased allele burden has not been found to alter survival or leukaemic transformation,10,11 although it has been identified as an independent risk factor for thrombosis, linked to an increased risk of arterial thrombosis.11‐13 Arterial or venous thrombosis has been found to occur in 12% of patients with WHO‐defined ET. In accordance, the primary goal of ET therapy is to prevent thrombohaemorrhagic complications without increasing the risk of bleeding.5
In 2016, we reported a study of ET management using MPN data taken from a registry created by the Hungarian MPN Working Group (HUMYPRON GROUP).14 The HUMYPRON GROUP MPN registry includes patient data on epidemiology, diagnostics, therapeutics and disease complications,14,15 with data entered online enabling continuing update and searches for association analysis.
Our first report detailed outcomes of ET patients treated with anagrelide vs hydroxyurea+aspirin followed for (median) 6 years.14 Significantly fewer minor arterial and minor venous thrombotic events (TEs) were reported in anagrelide‐treated patients vs hydroxyurea+aspirin over 6 years (P < .001). Here, we report findings from the Hungarian MPN registry from outcome data of ET patients treated with anagrelide vs hydroxyurea+aspirin followed for a median of 10 years.
This is a retrospective analysis of real‐life data of patients with ET in the HUMYPRON Registry treated with anagrelide or with hydroxyurea+aspirin. It builds on the findings of analyses at 6 years, reporting data at 10 years. Study methods are fully reported elsewhere.14 In brief, a questionnaire based on the 2008 revision of the World Health Organization (WHO) diagnostic criteria16 was used to collect data on ET patients electronically. The questionnaire focused on complications, risk stratification and treatment. With regard to comedication with antiplatelet and anticoagulant therapy, generally patients on anagrelide were not receiving such arterial TE prophylaxes, and patients used anticoagulants only if they were already being taken for a previous major venous TE (these data were therefore neither requested on the questionnaire nor collected from the registry). Landolfi criteria,17 and Tefferi and International Prognostic Scoring for ET (IPSET) scoring systems5,18,19 were used to stratify thrombotic risk and risk‐adapted treatment characteristics. Survival was estimated using the IPSET‐survival score system.20,21
Clinicians at 14 haematological centres in Hungary entered patient data from completed questionnaires into the HUMYPRON Registry on all patients with ET who they had diagnosed using WHO criteria and treated. Data cut‐off was December 2018, with analysis done 5 years after the first analysis.14 The current analysis included ET patients who were in the first analysis, alive, and who remained in their treatment arm. Crossover patients and patients receiving both anagrelide and hydroxyurea+aspirin were excluded. Data were extracted from the registry on the characteristics and outcomes of ET patients treated with anagrelide or with hydroxyurea+aspirin.
Predefined criteria were used to assess TEs occurring during the study. Classification of minor and major arterial and major venous TEs was as follows22:
- major arterial thrombosis, including stroke, myocardial infarction, peripheral arterial disease and splanchnic arterial thrombosis;
- minor arterial events, including transient ischaemic attack, angina pectoris, unstable angina, generalised convulsions, erythromelalgia, ocular symptoms, other peripheral arterial microcirculatory disturbances and angina abdominalis (transient abdominal ischaemia);
- major venous thrombosis, including iliofemoral thrombosis, pulmonary embolism, splanchnic vein thrombosis, pelvic vein thrombosis and all deep vein thromboses.
Minor event diagnoses were based on the clinical judgment of the clinicians and on symptoms and medical notes of the patients; major events were established through the clinicians usual practice procedures.
Unlike in the 6‐year analysis,14 minor venous events were not reported in this update after a median of 10 years. Generally, such events appear of little interest within the published literature.
Continuous variables were examined for normal distribution (test of normality: Kolmogorov‐Smirnov with Lilliefors significance correction, type I error = 10%). Those with normally distributed data were compared between cohorts and between subgroups using the t test for independent samples (test for variance homogeneity: Levene test, type I error = 5%); otherwise, data were compared using the Mann‐Whitney U test, with exact probabilities. The Mann‐Whitney U test, with exact probabilities, was also used for the comparisons of ordinal variables. Categorical variables were compared by the Fisher's exact test or by the exact chi‐squared test. The log‐rank test was used for comparisons of time to event variables, as depicted by Kaplan‐Meier plots.
Cox regression analyses, using the forward method based on the likelihood ratio approach, were used to investigate the influence of the following covariates on the occurrence of death and the onset of disease progression: treatment (anagrelide vs hydroxyurea+aspirin), age at diagnosis (<60 years vs ≥60 years), gender (male vs female), JAK2‐positive status (no vs yes), TE prior to ET diagnosis (no vs yes) and Landolfi risk summary score.
Logistic regression analyses, using the forward method based on the likelihood ratio approach, were used to investigate the influence of the same aforementioned variables and observation time on the occurrence of TEs and secondary malignancies.
The type I error was not adjusted for multiple testing, and therefore, the results of inferential statistics are descriptive only. Statistical analysis was performed using the open‐source R statistical software package, version 3.1.2 (The R Foundation for Statistical Computing); statistical tests were interpreted at a 5% significance level.
As for the initial study,14 this extension study was approved by the ETT‐TUKEB and was performed in compliance with the Declaration of Helsinki. Similarly, there was no requirement for patient consent, as there is no risk of patient identification with data presented for patient groups.
In the extension study year of registry data collection, a total of 14 haematological centres provided patient data, yielding 237 evaluable patients with ET. Of these, 116 patients were treated with anagrelide and 121 were treated with hydroxyurea+aspirin. Patients were followed for a median of 10 years (range 1–29 years) (Table 1). Fewer ET patients were evaluated in this analysis compared to the first analysis 14 because not all patient data could be collected from two of the participating haematology centres and because patients switched treatments.
1 TableCharacteristics of patients with ET treated with anagrelide vs. hydroxyurea + aspirin
Anagrelide (N = 116) Hydroxyurea + Aspirin (N = 121) Total (N = 237) Gender, n (%) Male 34 (29.3) 43 (35.5) 77 (32.5) .333 Female 82 (70.7) 78 (64.5) 160 (67.5) Age at diagnosis, years, median 60 63 62 .026 Minimum‐maximum 25.0‐92.0 29.0‐89.0 25.0‐92.0 Mean ± SD 58.9 ± 15.1 62.8 ± 11.4 60.9 ± 13.4 Age at diagnosis categorised, years, n (%) <60 56 (48.3) 46 (38.0) 102 (43.0) .117 ≥60 60 (51.7) 75 (62.0) 135 (57.0) JAK2 positive, yes, n (%) 80 (69.0) 87 (71.9) 167 (70.5) .67 CALR positive, yes, n (%) 27 (23.3) 18 (14.9) 45 (19.0) .135 MPL positive, yes, n (%) 6 (5.2) 6 (5.0) 12 (5.1) >.999 Triple‐negative status, yes, n (%) 3 (2.6) 10 (8.3) 13 (5.5) .084 Time to death/length of follow‐up, years, median Minimum‐maximum 10 8 10 .011 Mean ± standard deviation (1‐29) (1‐29) (1‐29) 11.3 ± 5.4 9.7 ± 5.3 10.5 ± 5.4 Landolfi risk summary score, median 4 4 4 .157 Minimum‐maximum (0.0–10.0) (0.0–10.0) (0.0–10.0) Mean ± SD 4.2 (2.3) 4.7 (2.2) 4.5 (2.2) Landolfi risk category, n (%) Low risk 6 (5.2) 2 (1.7) 8 (3.4) .005 Intermediate risk 29 (25.0) 12 (9.9) 41 (17.3) High risk 50 (43.1) 65 (53.7) 115 (48.5) Very high risk 31 (26.7) 42 (34.7) 73 (30.8) Tefferi risk category, n (%) Very low risk 17 (14.7) 8 (6.6) 25 (10.5) .068 Low risk 32 (27.6) 30 (24.8) 62 (26.2) Intermediate risk 12 (10.3) 15 (12.4) 27 (11.4) High risk 55 (47.4) 68 (56.2) 123 (51.9) IPSET thrombosis risk category, n (%) Low risk 11 (9.5) 5 (4.1) 16 (6.8) .102 Intermediate risk 26 (22.4) 23 (19.0) 49 (20.7) High risk 79 (68.1) 93 (76.9) 172 (72.6) TE prior to diagnosis, yes, n (%) 25 (21.6) 38 (31.4) 63 (26.6) .106 TE after diagnosis, yes, n (%) 30 (25.9) 46 (38.0) 76 (32.1) .052 Arterial minor 8 (6.9) 36 (29.8) 44 (18.6) <.001 Arterial major 15 (12.9) 12 (9.9) 27 (9.9) .049 Venous major 11 (9.5) 8 (6.6) 19 (6.6) .102 Bleeding event, yes, n (%) 4 (3.4) 4 (3.3) 8 (3.4) >.999
- 3 Abbreviations: CALR, calreticulin; ET, essential thrombocythaemia; IPSET, International Prognostic Scoring for ET; JAK2, Janus kinase 2; MPL, myeloproliferative leukaemia virus oncogene; TE, thromboembolic event.
- 4 Bold value indicates statistical significant P <.005.
Both groups comprised a gender ratio of approximately 1 male:2 females, and most patients were JAK2V617F‐positive (Table 1). There was no significant difference between the treatment groups in gender, age at diagnosis categorised (<60 years and ≥60 years) and gene mutation positivity (yes vs no).
The Landolfi risk summary score was median 4.0 (total range 0.0–10.0) for both the anagrelide and hydroxyurea+aspirin groups (Table 1). There was a significant difference between the proportions of patients in the four Landolfi risk categories, with most patients considered as being at high thrombotic risk. The majority of patients were also in the high‐risk category when evaluated using Tefferi scoring system (defined by presence of thrombosis history or presence of JAK2/MPL mutation in an older patient5; and the IPSET scoring system also considers major cardiovascular risk factors.18,19
Prior to diagnosis, TEs were reported in statistically similar proportions of patients in each group (Table 1). After diagnosis, a marginal between‐group difference in the number of patients with TEs was observed (P = .052), which was reported in 25.9% of patients in the anagrelide group and 38.0% of patients in the hydroxyurea+aspirin group (Table 1). Of the patients with a TE postdiagnosis, there was a significant between‐group difference in minor arterial events (P < .001) and major arterial events (P = .049). Minor arterial events were more frequently reported in the hydroxyurea+aspirin group, while there were marginally more reports of major arterial events in the anagrelide group (a 4.5‐fold and a 1.25‐fold between‐group difference in numbers of patients with events, respectively). The numbers of patients experiencing major venous events were similar between the treatment groups. There were four haemorrhagic complications in each treatment group.
Comparisons between the cohorts of patients (total n = 76; data not shown) with a TE after diagnosis were found not to be significantly different (P > .05) in terms of patient characteristics, including gender, age at diagnosis ≥60 years, gene mutation status, Landolfi summary score or risk category, Tefferi risk category, IPSET risk category and time to death/length of follow‐up.
A significant difference between treatment groups in the continuous variables, median age at diagnosis (older in the hydroxyurea+aspirin group) and time to death/length of follow‐up (shorter in the hydroxyurea+aspirin group) (Table 1). Logistic regression analysis (multivariate model) for "TE after diagnosis = YES" in the total population (n = 237) indicated that neither of these findings influenced the occurrence of TE; TE prior to diagnosis only was found to significantly influence TE incidence (P < .001; Table 2).
2 TableInfluence of ET cohort characteristics on the occurrence of TE after diagnosis (total n = 237)
Variables Exp( EXP( Medication (anagrelide vs hydroxyurea + aspirin) .114 1.641 0.888, 3.031 Age at diagnosis (y) .298 1.016 0.986, 1.048 Gender (male vs. female) .771 0.908 0.473, 1.741 JAK2 positive (no vs. yes) .115 1.733 0.874, 3.437 TE prior to diagnosis (no vs. yes) <.001 6.130 2.897, 12.972 Landolfi risk categories, overall .508 — — Low risk (reference category) — — — Intermediate risk vs Low risk .541 2.021 0.211, 19.321 High risk vs Low risk .915 1.134 0.111, 11.580 Very high risk vs Low risk .874 0.891 0.070, 9.603 Time to death or length of follow‐up (years) .903 1.004 0.944, 1.067
1 Note
- 5 Logistic regression (multivariate model): the dependent variable is the number of patients with a TE after diagnosis ("yes") among the total population (n = 237). Covariates: treatment (anagrelide vs hydroxyurea + aspirin), age at diagnosis (years), gender (female vs male), JAK2 mutation positive (no vs yes), thrombotic event before the diagnosis (no vs yes), Landolfi risk categories (low = reference category, intermediate, high, very high) and time to death or end of observation (years).
- 6 Abbreviations: ET, essential thrombocythaemia; JAK2, Janus kinase 2; TE, thromboembolic event.
- 7 Bold value indicates statistical significant P <.005.
Secondary malignancies were reported in 11 (9.5%) of patients (all female) in the anagrelide group and 17 (14.0%) of patients (9 females and 8 males) in the hydroxyurea+aspirin group (n = 237; P = .854). These included urogenital (2 patients vs 4 patients, respectively), lung (1 vs 4 patients), skin (2 vs 3 patients), oropharyngeal‐laryngeal (1 vs 2 patients), breast (1 vs 2 patients), CNS (2 vs 0 patients), lymphoproliferative (1 vs 1 patients) and gastrointestinal (1 vs 1 patients) malignancies, which each occurred in 2–6 patients.
Logistic regression analysis (multivariate model) for "secondary malignancies = YES" (n = 237) in the total population (n = 237) indicated that the patient characteristics examined including treatment and gender did not influence the occurrence of secondary malignancies (Table 3).
3 TableInfluence of ET cohort characteristics on the occurrence of secondary malignancies
Variables Exp( EXP( Medication (anagrelide vs hydroxyurea + aspirin) .419 1.411 0.621, 3.254 Age at diagnosis (y) .891 0.997 0.956, 1.040 Gender (male vs female) .497 0.734 0.302, 1.788 JAK2 positive (no vs yes) .281 1.701 0.648, 4.468 TE prior to diagnosis (no vs yes) .381 0.634 0.228, 1.758 Landolfi risk categories, overall .504 – – Low risk (reference category) – – – Intermediate risk vs low risk .445 0.359 0.026, 4.954 High risk vs low risk .977 0.964 0.078, 11.854 Very high risk vs low risk .776 1.480 0.100, 21.966 Time to death or length of follow‐up (years) .501 0.970 0.888, 1.060
- 2 Note
- 8 Logistic regression (multivariate model): the dependent variable is the number of patients with secondary malignancies ("yes") in the total population (n = 237). Covariates: treatment (anagrelide vs hydroxyurea + aspirin), age at diagnosis (years), gender (female vs male), JAK2 mutation positive (no vs yes), thrombotic event before the diagnosis (no vs yes), Landolfi risk categories (low = reference category, intermediate, high, very high) and time to death or end of observation (years).
- 9 Abbreviations: ET, essential thrombocythaemia; JAK2, Janus kinase 2; TE, thromboembolic event.
Disease progression, defined as bone marrow evidence of AML, myelofibrosis or myelodysplastic syndrome (MDS), was reported in 10 (8.6%) of patients in the anagrelide group and 23 (19.0%) of patients in the hydroxyurea+aspirin group (n = 237; P = .024).
Time to progression/length of follow‐up was median 10.0 and 8.00 years (range for both 1.00‐29.00 years) in the anagrelide group and the hydroxyurea+aspirin group, respectively (n = 237; P = .004).
For the 33 patients with disease progression, time to progression/length of follow‐up was median 12.50 years (range 4.00‐19.00 years; mean 11.60 years, standard deviation [SD] 4.35) in the anagrelide group and median 7.00 years (range 2.00‐14.00 years; mean 7.70 years, standard deviation [SD] 3.36) in the hydroxyurea + aspirin group (P = .016).
Kaplan‐Meier survival analysis for time to progression/length of follow‐up is shown in Figure 1. Progression‐free survival (%) was significantly increased for anagrelide‐treated patients vs hydroxyurea+aspirin (log‐rank test P = .004). Separation of the curves occurred after approximately 5 years treatment; the low numbers of "at‐risk" patients at 27 years post‐treatment resulted in a sharp fall in progression‐free survival rates.
Cox regression analysis (multivariate model) for "time to progression/length of follow‐up, years" indicated that of the patient characteristics examined, treatment (anagrelide vs hydroxyurea+aspirin) influenced the time to progression/length of follow‐up (n = 233; P = .006; hazard ratio 3.091, 95% confidence interval 1.387, 6.886; Table 4).
4 TableInfluence of ET cohort characteristics on the time to progression/length of follow‐up and on the time to death/length of follow‐up (years)
Variables Exp(B): hazard ratio EXP(B): 95% CI Time to progression/length of follow‐up (years) Medication (anagrelide vs hydroxyurea + aspirin) .006 3.091 1.387, 6.88 Age at diagnosis (years) .134 1.031 0.991, 1.074 Gender (male vs female) .249 1.527 0.743, 3.136 JAK2 positive (no vs yes) .198 0.618 0.297, 1.285 TE prior to diagnosis (no vs yes) .055 0.327 0.104, 1.023 Landolfi risk categories, overall .083 – – Low risk (reference category) – – – Intermediate risk vs low risk .427 0.403 0.043, 3.786 High risk vs low risk .519 0.457 0.042, 4.944 Very high risk vs low risk .566 1.476 0. 038, 5.984 Time to death/length of follow‐up (years) Medication (anagrelide vs hydroxyurea + aspirin) .004 2.333 1.318, 4.131 Age at diagnosis (years) <.001 1.127 1.090, 1.165 Gender (male vs female) .002 2.387 1.395, 4.085 JAK2 positive (no vs yes) .548 0.842 0.481, 1.474 TE prior to diagnosis (no vs yes) .517 1.222 0.666, 2.241 Landolfi risk categories, overall .358 – – Low risk (reference category) – – – Intermediate risk vs low risk .075 0.103 0.008, 1.259 High risk vs low risk .124 0.162 0.016, 1.651 Very high risk vs low risk .142 1.164 0. 015, 1.836
- 10 Abbreviations: ET, essential thrombocythaemia; JAK2, Janus kinase 2; TE, thromboembolic event.
- 11 a Cox regression (multivariate model): the dependent variable is the time to progression/length of follow‐up (years) (n = 233).
- 12 b Logistic regression (multivariate model): the dependent variable is the number of patients with a TE after diagnosis ("yes") among the total population (n = 237). Covariates: treatment (anagrelide vs hydroxyurea + aspirin), age at diagnosis (years), gender (female vs male), JAK2 mutation positive (no vs yes), thrombotic event before the diagnosis (no vs yes) and Landolfi risk categories (low = reference category, intermediate, high, very high).
- 13 c Cox regression (multivariate model): the dependent variable is the time to death/length of follow‐up (years) (n = 237).
- 14 Bold value indicates statistical significant P <.005.
Of the patients with disease progression, disease transformation from ET to myelofibrosis occurred in all 10 patients in the anagrelide group and 21 of 23 patients in the hydroxyurea+aspirin group. Transformation direct to AML was found in two patients in the hydroxyurea+aspirin group only. Of the patients with myelofibrosis, 1 patient and 5 patients in the anagrelide and hydroxyurea+aspirin groups, respectively, transformed from myelofibrosis to AML. No significant between‐group difference was found in rates of transformation.
Death during the study was reported in 20 (17.2%) of patients in the anagrelide group and 41 (33.9%) of patients in the hydroxyurea+aspirin group (n = 237; P = 0.005). Causes of death included secondary malignancy (4 patients vs 9 patients, respectively), vascular event (3 vs 6 patients), heart failure (2 vs 5 patients), disease progression (0 vs 5 patients), COPD/respiratory failure (3 vs 1 patients), sepsis (0 vs 2 patients), ileus (0 vs 1 patients) and unknown (8 vs 12 patients).
Time to death/length of follow‐up was median 10.0 and 8.00 years (range for both 1.00‐29.00 years) in the anagrelide group and the hydroxyurea+aspirin group, respectively (n = 237; P = .011; Table 1).
For the 61 deaths, time to death/length of follow‐up was median 8.00 years (range 1.00‐27.00 years) for both treatment groups; anagrelide group: mean 9.40 years, SD 6.47; hydroxyurea+aspirin group mean 8.66 years, SD 5.49 (P = .510).
Kaplan‐Meier survival analysis for time to death/length of follow‐up is shown in Figure 2. Survival (%) was significantly increased for patients treated with anagrelide than with hydroxyurea+aspirin (log‐rank test P = .001), with separation of the curves seen after approximately 5‐years treatment.
Cox regression analysis (multivariate model) for "time to death/length of follow‐up, years" indicated that of the patient characteristics examined, treatment (anagrelide vs hydroxyurea+aspirin; P = .004), age at diagnosis (years; P < .001), and gender (P = .002) influenced the time to death/length of follow‐up (n = 237; Table 4).
In this evaluation of data from ET patients in the Hungarian MPN Registry, treatment with anagrelide was associated with a reduction in the number of patients experiencing TEs compared with hydroxyurea+aspirin over a median period of 10 years. This was primarily due to the significantly fewer minor arterial events reported in the anagrelide group vs the hydroxyurea+aspirin group (8 vs 36; P > .001).
The observed reduction in minor arterial TEs at median 10 years in patients treated with anagrelide vs hydroxyurea+aspirin was consistent with that found at patient evaluation after a median of 6 years.14 While there was no between‐group difference in the frequency of major arterial TEs at 6 years,14 there were marginally more reports of major arterial events in the anagrelide group at 10 years (15 vs 12; P = .049). The absence of aspirin co‐treatment in anagrelide‐treated patients may have partly been responsible for the high number of major arterial events and low occurrence of bleeding events. Anagrelide has shown efficacy in decreasing platelet counts, with significant reductions in the rates of major and minor thrombotic complications over 6 months in patients with ET.23 Our analysis suggests that anagrelide has continuous effect on reducing minor arterial TEs, while not increasing major venous TEs or bleeding events vs hydroxyurea+aspirin over a median of 10 years.
The Primary Thrombocythemia 1 (PT‐1) study showed that arterial thrombosis and serious haemorrhage developed in significantly more ET patients in the anagrelide+low‐dose aspirin group than in the hydroxyurea+low‐dose aspirin group over 39 months.24.The increased bleeding risk in the anagrelide+aspirin group24 may suggest synergism between the two agents on platelet function, via their inhibition of platelet phosphodiesterase activity 25 and inhibiting cyclooxygenase activity,26 respectively. In contrast, the ANAHYDRET study showed that there was no significant difference in the incidence of ET‐related major and minor arterial and venous thrombotic or haemorrhagic events between anagrelide and hydroxyurea over 36 months.22 Meta‐analysis of these two trials concluded that thrombosis occurred at similar frequency between patients treated with anagrelide or hydroxyurea and that major bleeding rates were lower in the hydroxyurea group.27
Our study has the advantage of data capture over a longer time period than these trials, firstly reported at 72 months14 and here at median 120 months. However, it is unclear whether over time the protective effect from major arterial TEs is reduced by anagrelide or is increased by hydroxyurea, or whether long‐term control of platelet function and thus thrombosis is influenced by other mechanisms.
Consistent with the results of our study, a history of thrombosis in ET is established as the strongest risk factor for such events.28 While age greater than 60 years and presence of JAK2V617F are among the additional risk factors for arterial thrombosis and male sex is known predictor of venous thrombosis,13 these factors were found not to influence the occurrence of TE after diagnosis in our study. Interestingly, risk stratification for recurrent thrombosis in ET evaluates age, and the presence/absence of thrombosis history and JAK2/MPL mutation with high risk defined by an age greater than 60 years or presence of thrombosis with JAK2/MPL mutation.5 Risk levels for thrombosis as assessed and scored using the Landolfi criteria showed that the Landolfi risk summary score was not a significant risk factor for TE after diagnosis at 10 years although it had been at 6 years.14
Secondary malignancies of various types were more frequently reported in patients treated with hydroxyurea+aspirin than with anagrelide, although the difference was not significant over the 10 year. A database study to assess long‐term complications of ET and its treatments confirmed low numbers of various cancer types.29
Disease progression in the hydroxyurea+aspirin group was observed to occur earlier and at twice the frequency than in the anagrelide group. Progression‐free survival was significantly increased for anagrelide‐treated patients vs hydroxyurea+aspirin, with a treatment difference showing after approximately 5 years. Treatment was confirmed to significantly influence the time to progression/length of follow‐up. Of patients with disease progression, transformation from ET to myelofibrosis was reported in twice as many patients in the hydroxyurea+aspirin group vs anagrelide.
In contrast, the PT‐1 study24 reported a significantly greater rate of transformation from ET to myelofibrosis in patients treated with anagrelide+low‐dose aspirin vs hydroxyurea+low‐dose aspirin. Other studies of anagrelide have reported no occurrence of disease progression to myelofibrosis, MDS or AML in patients with WHO‐ET.22,29,30 Hydroxyurea is a cytostatic drug that interferes with DNA metabolism, and considered potentially leukaemogenic.31,32 A study of hydroxyurea's leukaemogenic potential concluded that discrepancies between hydroxyurea exposure and risk of AML/MDS are likely to be associated with differences in patient characteristics including therapy, but could not totally rule out a leukaemogenic effect.33 While our cohorts were balanced for baseline characteristics, it does not exclude the influence of unreported factors such as other gene mutations on AML/MDS development.
Anagrelide was associated with half of the death rate and longer duration of survival vs the hydroxyurea+aspirin group, with a difference occurring after approximately 5‐year treatment. Anagrelide‐treated patients were younger at diagnosis, which influenced survival, but treatment appeared also to be implicated as a significant survival factor (Table 4). We speculate that hydroxyurea+aspirin may provide less protection than anagrelide in terms of survival. ET has been associated with reduced survival compared with the general population in studies with more than 10 years of follow‐up.34 Both age and gender are recognised as predictors of prognosis, with higher age at diagnosis and male gender predicting poorer relative survival.34,35
The strength of our study is that the HUMYPRON Registry provides a nationwide database of patients with MPN, with "real‐world" patient data providing the opportunity for extended follow‐up of patient outcomes. A limitation is that we were confined to work with the data reported on questionnaires and entered onto the database. As such, comedication with antiplatelet and antithrombotic drugs was not reported. Data were subject to clinicians' misreporting and misconceptions regarding treatment, complications and risk stratification. Treatment was according to clinical practice; therefore, confounding by indication cannot be disregarded whereby clinicians prescribed the study treatments to patients with different prognoses. Patient outcomes may have been affected by previously received ET treatments and comedications. Antithrombotic treatment, given as prophylaxis and as study treatment, may also have presented a confounding factor influencing patient outcomes. Patients receiving anagrelide were not treated with aspirin as the combination has been shown to significant increase the rate of serious haemorrhage.25 Bias may have also arisen through the greater difficulty in defining minor arterial TEs compared with major arterial TEs, and because treatment was not blinded.
Current ET treatment guidelines stratify patients according to thrombotic risk, recommending hydroxyurea+aspirin as first‐line therapy in patients in the intermediate‐ and high‐risk groups.5 Anagrelide is often advised only after failure of other drug options, based on observed arterial thrombosis, bleeding events and fibrotic progression in a controlled comparative study in a PVSG‐ET population.24 However, after a median 10 years treatment, our findings showed an association between anagrelide exposure and a reduced risk for minor arterial TEs in patients with WHO‐ET but the risk of major arterial TEs may marginally increase compared with hydroxyurea+aspirin. Importantly, anagrelide was associated with an increase in progression‐free and overall survival vs hydroxyurea+aspirin over (median) 10 years. Further comparison of anagrelide‐ and hydroxyurea‐treated WHO‐ET patients is warranted in larger populations in prospective studies and in observational studies or registries over an even longer time period than 10 years.
This work was supported by the participating hospitals. The preparation of this manuscript and statistical analyses were supported by AOP Orphan Pharmaceuticals AG. The authors would like to thank Mónika Horváth, Judit Halász, Dr Hamvas József, Dr Bodnár Mária, Dr Dömötör Mária, Dr Altay Elvira and Dr Palaczki Aranka for data entry; Dr Wolfgang Schimetta for statistical analyses; and Dr Susan Libretto for preparation of the manuscript and for editorial assistance.
The authors do not have financial and/or personal relationships with other people or organisations that could inappropriately influence this work.
All authors contributed to the organisation of the Hungarian MPN registry, participation in the Steering Committee, and data entry. LH was responsible for construction of the online database. PD, LH and ME carried out data analysis, and ME was responsible for data interpretation.
By Adam Kellner; Peter Dombi; Arpad Illes; Judit Demeter; Lajos Homor; Ibolya Ercsei; Zsofia Simon; Eva Karadi; Jozsef Herczeg; Viktoria Gy Korom; Zoltan Gasztonyi; Laszlo Szerafin; Miklos Udvardy and Miklos Egyed
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author