External validation and clinical evaluation of the International Prognostic Score of Thrombosis for Essential Thrombocythemia (IPSET-thrombosis) in a large cohort of Chinese patients

Objectives: In patients with essential thrombocythemia (ET), vascular complications contribute to both morbidity and mortality. To better predict the occurrence of thrombotic events, an International Prognostic Score of thrombosis for ET (IPSET‐thrombosis) was recently developed. We hereby presented an external validation and analysis of this model in a large Cohort of Chinese Patients. Methods: We retrospectively evaluated the characteristics and risk factors for thrombosis in 970 Chinese patients with ET and estimated the clinical implications of the IPSET‐thrombosis model. Results: The median follow‐up was 49 months (range, 0–360). Chinese ET patients had similar clinical characteristics as Caucasian patients. Similar to the IPSET‐thrombosis study, our multivariate analysis revealed age >60 (HR = 1.949), previous thrombosis (HR = 2.484), JAK2V617F mutation (HR = 1.719), and cardiovascular risk factors (HR = 1.877) as independent risk factors for thrombosis. We confirmed that the above risk factors in IPSET‐thrombosis, when compared with traditional risk factors (e.g., age ≥60 and previous thrombotic events), were more predictive of thrombotic events (C‐index 0.714 vs. 0.647). Classification by IPSET‐thrombosis risk groups revealed different cumulative thrombosis‐free survival (P < 0.001). For treatment, patients in the intermediate‐ and high‐risk group derived clinical benefit from cytoreductive agents (P < 0.05), but those in the low‐risk group did not (P = 0.446). The lower risk of thrombosis on cytoreductive therapy was related to decrease in leukocyte count during the disease course. Conclusions: We validate the reproducibility of IPSET‐thrombosis in Chinese ET patients and provide key clinical implications.

thrombosis; risk factors; treatment; primary thrombocythemia

Essential thrombocythemia (ET) is a type of myeloproliferative neoplasms characterized by increased megakaryopoiesis, sustained thrombocytosis, increased risk of vascular events (thrombosis and hemorrhage), and increased risk of transformation to myelofibrosis or acute leukemia [1] , [2] . Among various myeloproliferative neoplasms, previous studies have demonstrated differences between Asian and Caucasian patients. For example, in polycythemia vera (PV), Chinese subjects exhibit higher percentage of JAK2 exon 12 mutation, whereas in primary myelofibrosis, Chinese subjects have younger age, more anemia, and longer survival [3] , [4] . In ET, several studies have evaluated risk factors for thrombotic events in Caucasian patients; however, such data from Chinese subjects are still lacking [2] , [5] , [6] , [7] .

Traditionally, advanced age ≥60 and history of thrombosis are regarded as predictive risk factors for future thrombosis [1] . Recently, a new International Prognostic Score of thrombosis for ET (IPSET‐thrombosis) was proposed by the International Working Group for Myeloproliferative Neoplasms Research and Treatment (IWG‐MRT) [6] . In this model, age >60 (1 point), history of thrombosis (2 points), cardiovascular risk factors (1 point), and JAK2V617F mutation (2 points) are used to divide patients into three risk categories (low risk = 0–1 points, intermediate risk = 2 points, and high risk ≥3 points). Compared with the traditional model, IPSET‐thrombosis outlined a clear definition of ‘intermediate‐risk group’ and provided a well‐defined weight to the general cardiovascular risk factors and JAK2V617F mutation [6] . Despite the advantages, this model has not been validated among Asian patients. We hereby present, to our best knowledge, the first report to evaluate the prognostic value and clinical implication of the IPSET‐thrombosis model in a large cohort of Chinese ET patients.

The study was approved by hospital‐based ethics committees, according to guidelines of the Helsinki Declaration of 1975. After hospital‐based ethics committee approval, 970 consecutive patients with an initial diagnosis of ET from March 1982 to April 2012 were identified from the Institute of Hematology and Blood Disease Hospital registry. Their clinical data were carefully reassessed, and all of them met the 2008 WHO criteria [8] . Patients were excluded if they had a positive family history of ET or thrombocytosis. Post‐ET myelofibrosis was diagnosed per the IWG‐MRT criteria [9] . Major bleeding was defined as symptomatic hemorrhage in vital organs, sudden decrease in hemoglobin >20 g/L, or over hemorrhage requiring transfusion. Major thrombotic events were defined as ischemic stroke, cerebral transient ischemic attacks, acute myocardial infarction, or thrombosis in vital visceral organs or peripheral vessels [6] . JAK2V617F mutation and endogenous erythroid colony (EEC) were investigated as described previously [10] , [11] . Blood cell counts at the time of complication were defined as the last available value before a thrombotic event. In the group who did not have thrombotic events, the last available value before the median time for thrombotic events in the other group was recorded.

Treatment strategies were determined by individual physicians' best clinical judgment. In general, antiplatelet agents (e.g., low‐dose aspirin, clopidogrel, dipyridamole) were given in patients with age <60, platelet count <1000 × 109/L, and lacking history of thrombosis, whereas cytoreductive agents (e.g., busulfan and meisoindigo in early years, hydroxyurea and interferon‐alpha in recent years) were added in those with age ≥60, platelet count >1000 × 109/L, or previous history of thrombosis.

Data were analyzed using the software spss 16.0 (SPSS Inc., Chicago, IL, USA). Categorical variables were compared by Pearson's chi‐squared test, Chi‐squared test with Yates' continuity correction or Fisher's exact test. The t‐test for independent samples or Mann–Whitney U‐test was used for quantitative values. The receiver operating characteristic (ROC) analysis was used to find a cutoff value of age at which thrombosis would occur. Life‐table method was used to calculate the 5‐, 10‐, and 15‐yr cumulative rate of thrombotic events. Thrombosis‐free survival was estimated by Kaplan–Meier method, and log‐rank test was used to compare thrombosis‐free survival data. Cox proportional hazards regression was employed to carry out univariate survival analysis, and factors that were significant were forced to multivariate survival analysis. A two‐tailed P ≤ 0.05 was considered to be statistically significant. The C‐index was calculated based on ROC analysis using prognostic index. Figures were made by spss 16.0 or graphpad prism 5 software. Factors analyzed were selected as previously described, including sex, age >60 yr, history of thrombosis, JAK2V617F mutation status, cardiovascular risk factors (i.e., active smoking, hypertension, diabetes, and hyperlipidemia), platelet count, hemoglobin level, and white blood cell count at diagnosis [6] . Symptoms related to thrombocytosis (i.e., microcirculatory, neurological and gastrointestinal symptoms, and fatigue) at diagnosis or during follow‐up were also evaluated.

The median age at diagnosis was 52 yr (range, 18–87), with male/female ratio 0.7. The median follow‐up was 49 months (range, 0–360). Fifty‐five (5.7%) patients were lost during follow‐up. Clinical and laboratory features were summarized in Table [NaN] . Four hundred and ninety‐nine (51.4%) patients were found to have thrombocytosis incidentally. JAK2V617F mutation was investigated in 746 patients (76.9%) among whom 380 (50.9%) were found to be positive. Compared to patients with wild‐type JAK2, V617F mutation‐positive patients had significantly older age (P < 0.001), higher rate of organomegaly (P = 0.017), higher white blood cell count (P < 0.001), higher granulocyte count (P < 0.001), higher hemoglobin levels (P < 0.001), lower platelet count (P < 0.001), and higher frequency of positive EEC (P = 0.009). During follow‐up, 45 (4.6%) patients progressed to post‐ET myelofibrosis, whereas 10 (1.0%) transformed to acute myeloid leukemia, including 6 cases that evolving from post‐ET myelofibrosis. One patient transformed to multiple myeloma after 8.4 yr of follow‐up.

Clinical characteristics at diagnosis in 970 patients with essential thrombocythemia

1 EEC, endogenous erythroid colony; F, female; HB, hemoglobin; M, male; PLT, platelet; WBC, white blood cell.

During follow‐up, 111 major thrombotic events were observed, including 91 arterial (82.0%) and 20 venous (18.0%) thrombosis. Ninety‐nine (10.2%) patients experienced at least one major thrombotic event during follow‐up. Detailed information of thrombotic events was summarized in Table [NaN] . The 5‐, 10‐, and 15‐yr cumulative rates of thrombotic events were 9.0%, 22.0%, and 35.0%, respectively. The median age at the time of thrombosis was 61 yr (range, 26–88). ROC analysis revealed the cutoff value of age at which thrombosis would occur was 60.5.

Major thrombotic events during follow‐up in 970 patients with essential thrombocythemia

Among all patients, 43 (4.4%) experienced at least one major hemorrhagic event. Among the 45 total major hemorrhagic events, gastrointestinal (n = 24; 53.3%) and intracranial bleeding (n = 9; 20.0%) were among the most common events. Subgroup analysis of the gastrointestinal bleeding revealed that 4 patients had concomitant peptic ulcers, and another 4 had esophageal varices at the time of bleeding. At the onset of major hemorrhage, median platelet count was 1111 × 109/L (range, 481–3958). Among 43 patients who suffered from major hemorrhagic events, 30 (69.8%) were under antiplatelet treatment. Patients with antiplatelet agents had a higher rate of major hemorrhagic events (30/532, 5.6%) than those without (13/438, 3.0%) (odds ratio, 1.954; 95% confidence interval, 1.006–3.793; P = 0.044).

In univariate analysis, factors significant for thrombosis included gender (P = 0.037), age >60 yr (P < 0.001), JAK2V617F mutation (P = 0.012), cardiovascular risk factors (P < 0.001), history of thrombosis (P < 0.001), and hemoglobin level (P = 0.017). Some of these risk factors overlapped. For example, higher rate of cardiovascular risk factors was observed in male than female patients (35.1% vs. 22.0%, P < 0.001) and in older (>60) than younger (≤60) patients (42.5% v. 21.9%, P < 0.001). Furthermore, higher hemoglobin level was found associated with a positive JAK2V617F mutation (P < 0.001).

In Cox multivariate analysis, 4 independent risk factors were identified, including age >60 (HR = 1.949), previous thrombosis (HR = 2.484), JAK2V617F mutation (HR = 1.719), and cardiovascular risk factors (HR = 1.877). These were in concordance with the IPSET‐thrombosis model. Detailed information on risk factors was shown in Table [NaN] . The C‐index was 0.714 (95% confidence interval, 0.655–0.772), which was higher than that of the conventional risk stratification (0.647; 95% confidence interval, 0.577–0.717).

Risk factors for thrombotic events in multivariate analysis in 970 patients with essential thrombocythemia

2 95% CI, 95% confidence interval; CVF, cardiovascular risk factors; HR, hazard ratio; ND, not done.

A total of 746 patients had all the available factors for risk‐group stratification according to the IPSET‐thrombosis model. The number of low‐, intermediate‐, and high‐risk patients were 307 (41.2%), 218 (29.2%), and 221 (29.6%), respectively. As shown in Fig. [NaN] , the 3 groups exhibited significantly different thrombosis‐free survival (P < 0.001). Table [NaN] displayed the cumulative thrombosis‐free survival rate at 5, 10, and 15 yr after diagnosis. Among the evaluable 746 patients, 77 (10.3%) experienced one or more thrombotic events during follow‐up. When these 77 patients were classified using the conventional risk factors (age and history of thrombosis), 33 (42.9%) would be considered as low risk, whereas 44 (57.1%) would be high risk. However, by the new IPSET‐thrombosis classification, only 11 (14.3%) were low risk, whereas 23 (29.9%) were intermediate risk, and 43 (55.8%) were high risk.

Cumulative thrombosis‐free survival after diagnosis in risk groups according to International Prognostic Score of thrombosis for ET (IPSET)‐thrombosis in 746 patients with essential thrombocythemia

3 SE, standard error; TFS, thrombosis‐free survival.

For therapy, 746 patients available for stratification according to IPSET‐thrombosis were analyzed. Two hundred patients (26.8%) did not receive any therapy, 128 (17.2%) received antiplatelet agents alone, 127 (17.0%) received cytoreductive therapy alone, and 291 (39.0%) were given both antiplatelet and cytoreductive therapy. We first evaluated the use of cytoreductive therapy in each of IPSET‐thrombosis risk groups. In 327 patients that did not receive any antiplatelet therapy, 46 (14.1%) experienced major thrombotic events. Compared to patients with no treatment, those with cytoreductive treatment alone had a significantly lower rate of thrombosis in the intermediate‐ and high‐risk groups (P < 0.05), but not in the low‐risk group (P = 0.446; Table [NaN] ).

Cytoreductive treatment in risk groups according to International Prognostic Score of thrombosis for ET (IPSET)‐thrombosis in 746 patients with essential thrombocythemia

Compared with the conventional stratification, three major changes were made in the new IPSET‐thrombosis model: older patients (>60) without other risk factors moved to the low‐risk group; an intermediate‐risk group was added; younger patients (≤60) with mutated JAK2 and cardiovascular risk factors but without history of thrombosis moved to the high‐risk group. We evaluated whether changes in these subgroups had clinical significance. The rate of thrombosis in older patients in the low‐risk group (3/39, 7.7%) was similar to that in younger ones in the same group (8/268, 3.0%; P = 0.309), but much lower than that in older patients in the intermediate‐ and high‐risk groups (34/155, 21.9%; P = 0.043). In the low‐risk group, older patients given cytoreductive agents alone (1/8, 12.5%) had a similar rate of thrombosis to those without treatment (1/9, 11.1%; P = 1.000), while in the intermediate‐ and high‐risk groups, older patients using cytoreductive agents alone (2/20, 10.0%) had fewer thrombotic events than no treatment (19/53, 35.8%; P = 0.030). We then focused on the treatment strategies in the intermediate‐risk group. Treatment strategies and the rate of thrombosis in each treatment group are displayed in Fig. [NaN] . In this group, patients with cytoreductive monotherapy had a lower rate of thrombosis than no treatment (2.8% vs. 19.0%; P = 0.049). The rate of thrombosis in patients with combination treatment was lower than no treatment (5.9% vs. 19.0%; P = 0.011) or antiplatelet agents alone (5.9% vs. 21.7%; P = 0.046). However, no difference was found between patients receiving combination treatment and cytoreductive therapy alone (5.9% vs. 2.8%; P = 0.765). The use of cytoreductive therapy in younger patients (≤60) with mutated JAK2 and cardiovascular risk factors but without previous thrombosis in the high‐risk group was not analyzed, because only 5 patients used cytoreductive agents as monotherapy in this subgroup.

Finally, we evaluated whether changes in blood cell counts or JAK2V617F mutation status after cytoreductive treatment affected the risk of thrombotic events. The median time from diagnosis to the onset of thrombosis was 50 months (range, 2–232). At the time of thrombotic events, cytoreductive treatment decreased the leukocyte count to a median level of 7.8 × 109/L (range, 1.1–26.3), reduced the level of hemoglobin to 133 g/L (range, 51–164), and platelet count to 633 × 109/L (range, 48–2760). We then compared blood cell counts at the time of complication between patients with and without thrombotic events. Leukocyte count was significantly higher in patients with thrombotic events (median, 11.0 × 109/L) than those without (median, 7.5 × 109/L; P < 0.001), while no difference was found in the level of hemoglobin (median, 134 vs. 132 g/L; P = 0.286) or platelet count (median, 653 vs. 629 × 109/L; P = 0.228) between two groups. When specific to each risk group, the difference in leukocyte count between patients with and without thrombotic events was significant in the intermediate‐risk group (median, 10.0 vs. 8.0 × 109/L; P = 0.009) and in the high‐risk group (median, 11.3 vs. 8.6 × 109/L; P = 0.001), but not in the low‐risk group (median, 8.9 vs. 6.6 × 109/L; P = 0.059). In patients with mutated JAK2 and with cytoreductive treatment, JAK2V617F mutation status was reexamined in 59 patients, and 17 (28.8%) became negative. There was no difference in the rate of thrombotic events between patients with persistent positive JAK2V617F (5/42, 11.9%) and those among whom JAK2V617F turned to be negative (2/17, 11.8%; P = 1.000).

Based on comparative analysis, Chinese ET patients have similar clinical characteristics as Caucasian ET patients. In our study, age and gender distributions were similar to those reported in previous studies [6] , [7] , [12] . Patients with acquired JAK2V617F mutation represented 50.9% of tested patients, in consistent with the rate in Caucasian patients (50–60%) [11] . Similar to previous report, our JAK2+ patients also had significantly older age, higher level of hemoglobin, lower platelet count, and higher rate of positive EEC [13] . EEC may be useful in diagnosing ET, especially in those with wild‐type JAK2. Mustjoki et al. [10] and Campbell et al. [13] reported the rates of positive EEC in JAK2V617F‐negative ET patients to be 20% and 33%, respectively. These salient features may indicate a subtype that mimics PV or masked PV in WHO‐defined ET [13] , [14] . The rate of major thrombotic events during follow‐up was 10.2%, comparable to 8–31% reported in previous studies of Caucasian patients [2] , [5] , [7] , [15] , [16] , [17] , [18] , [19] , [20] . Similar to these studies, arterial thrombotic events (82.0%) were more common than venous thrombotic events (18.0%). Transformation to myelofibrosis (4.6%) or leukemia (1.0%) was extremely rare, which was also similar to previous report [7] , [16] , [21] . Therefore, preventing thrombotic events becomes one of the most important goals in the treatment of ET patients.

Historically, advanced age over 60 and history of thrombosis were the only two well‐accepted risk factors in ET [1] . In our study, ROC analysis of thrombotic events revealed that the age at which thrombosis would occur was 60.5, supporting the practice to use 60 as the common cutoff of age. More recently, JAK2V617F mutation and allele burden are becoming better recognized as risk factors in both Asian and Caucasian subjects with nearly doubling of odds ratio for thrombosis in mutated individuals [22] , [23] , [24] , [25] , [26] , [27] , [28] , [29] , [30] . In a cohort of 224 patients with ET, JAK2V617F mutation increased the relative risk of overall thrombosis by 45% [26] . Carobbio et al. [24] reported that the risk of thrombosis increased progressively with increasing amount of mutated JAK2 allele burden, and those with allele burden >50% had the highest rate of vascular complications. Various explanations have been proposed for the above‐observed phenomenon, including increased activation of platelets and leukocytes, and platelet P‐selectin expression [31] . In addition to JAK2 mutation, cardiovascular risk factors have also been identified as a risk factor in ET [22] , [32] , [33] . Other proposed risk factors, such as leukocytosis or thrombocytosis, have not been consistently shown to contribute to the pathogenesis of overall thrombosis [5] , [7] , [17] , [34] , [35] , [36] .

Through multivariate analysis, we identified the same risk factors for thrombosis in Chinese ET patients as the IPSET‐thrombosis model, including age >60, previous thrombosis, JAK2V617F mutation, and cardiovascular risk factors. The concordance is likely explained by similar study designs, such as large sample size, balanced patient features, long follow‐up period, and WHO‐definition of ET. In contrast to the IPSET‐thrombosis report, we discovered that age >60 had a higher HR than JAK2V617F mutation in Chinese subjects.

In Chinese ET patients, we also confirmed that the new risk classification according to the IPSET‐thrombosis model was more predictive of thrombosis than the traditional risk model. The time distribution of vascular events in the 3 new risk groups demonstrated significantly different cumulative thrombosis‐free survival after diagnosis. The C‐index of the risk factors identified in our study (0.714) was higher than that of the conventional risk factors (0.647). Among all 77 patients who experienced thrombosis in 746 evaluable patients, only 14.3% were truly low risk per the IPSET‐thrombosis model (as JAK2V617F mutation and cardiovascular risk factors would push them into a higher risk group), while as many as 42.9% would be low risk according to the traditional risk factors. Thus, the IPSET‐thrombosis risk classification may better identify patients in need of early treatment in ET patients.

In terms of treatment, our study provides evidence for using cytoreductive therapy in Chinese ET patients based on the new risk classification. Traditionally, cytoreductive therapy was only offered in patients with age >60, platelet count >1000 × 109/L, or previous history of thrombosis [37] . More recently, some experts have proposed treatment strategies based on the new IPSET‐thrombosis risk classification discussed previously [6] , [38] , [39] . However, there has been little direct evidence on whether cytoreductive therapy should be used in the intermediate‐risk group and in older patients (>60) without other risk factors in the low‐risk group. In our study, patients receiving cytoreductive agents alone derived clinical benefit (i.e., lower rate of thrombosis) in both high‐risk and intermediate‐risk groups, but not low‐risk group. This finding applies even to older patients (>60). Cytoreductive agents benefited older patients (>60) in the high‐risk and intermediate‐risk groups, but not in the low‐risk group. Furthermore, in the low‐risk group, the rate of thrombosis in older patients (>60; 7.7%) was not significantly different from that in younger ones (≤60; 3.0%). Due to the small sample size, definitive conclusion could not be drawn, but it appears that age alone should not be a sole determinant for cytoreductive treatment. As for patients in the intermediate‐risk group, both cytoreductive therapy alone (2.8%) and combination therapy with antiplatelet plus cytoreductive agents (5.9%) lead to fewer thrombotic events than antiplatelet therapy alone (21.7%) or no treatment (19.0%). Of note, patients with antiplatelet therapy alone had a slightly higher rate of thrombosis than those without treatment. It seemed that antiplatelet therapy failed in preventing thrombotic events in the intermediate‐risk group, but it could also be due to several limitations of our study, including the retrospective design and small sample size in patients with antiplatelet therapy alone in this group. Even so, we believe that patients in the intermediate‐risk group should receive cytoreductive therapy. We then explored possible reasons for lower incidence of thrombosis on cytoreductive therapy. Decrease in the leukocyte count during disease course would be one of reasons, because it reduced the risk of thrombotic complication, and the effect was significant in the intermediate‐ and high‐risk groups. However, molecular response induced by cytoreductive therapy did not reduce the risk of thrombosis. Other reasons might include decrease in the level of vascular endothelial growth factor or increased apoptosis of megakaryocytes, but the pathogenesis still remains uncertain [40] , [41] .

In conclusion, we present the initial external validation and clinical evaluation of the IPSET‐thrombosis model in a large cohort of Chinese ET patients with similar disease characteristics and clinical course as their Western counterparts. We find that the new model more reliably predicts thrombotic events than the conventional model and should be used to guide cytoreductive therapeutic decisions, especially in the intermediate‐risk group. Future prospective studies are needed to confirm the value of the new model in daily clinical practice for treating ET patients.

The authors would like to thank Dr Ang Li (Massachusetts General Hospital, Harvard University, USA) for critical review of the manuscript. All of the authors declare no conflict of interest. This study was supported by 863 projects of Ministry of Science and Technology of China (2012AA02A211 and 2011ZX09302‐007‐04), Ministry of Health (201202017), National Natural Science Foundation of China (81270595), Young Fund of Peking Union Medical College (2013), and Tianjin Municipal Science and Technology Commission (11JCZDJC18600).

Graph: Cumulative thrombosis‐free survival in three risk groups according to International Prognostic Score of thrombosis for ET ( IPSET )‐thrombosis in 746 patients with essential thrombocythemia.

Graph: Treatment strategies in the intermediate‐risk group according to International Prognostic Score of thrombosis for ET ( IPSET )‐thrombosis. * P  < 0.05; # Number of patients with thrombosis/total number of patients in each group.