Philadelphia negative Myeloproliferative Neoplasms (MPNs) that are associated with Janus kinase 2 ( JAK2 ) mutation include Polycythemia Vera (PV), Essential Thrombocythemia (ET) and Primary Myelofibrosis (PMF). The discovery of JAK2 V617F mutation in MPNs has facilitated the diagnosis of MPNs and led to a better understanding of their pathophysiology [1-3]. The majority of PV patients are JAK2 mutated, while 50-60% of ET and PMF are JAK2 mutated [2,4-6]. Other driver mutations described in MPN include Myeloproliferative Leukemia (MPL) and calreticulin ( CALR ). MPL is mutated in 5-8% of JAK2 wild type ET and PMF [7,8]. About 30-40% of ET and PMF have neither JAK2 nor MPL mutation (double-negative). In December 2013, two groups simultaneously reported the presence of calreticulin mutation ( CALR ) in ET and PMF [9,10]. The incidence of CALR mutation is approximately 70% in JAK2 wild type ET and PMF, but rare in PV. CALR mutation is almost mutually exclusive with JAK2 mutation with rare concomitant mutations (<1%) reported in a number of studies [11-14]. More than 50 different CALR mutations are reported, but 80% of CALR mutated patients have one of two mutation variants: type 1 (52-bp deletion) or type 2 (5-bpinsertion) [15].

      In Caucasians, CALR mutated ET is associated with a specific phenotype with younger age, male sex, higher platelet count (Plt), lower Hemoglobin (Hb) and a lower incidence of thrombotic events when compared to JAK2 mutated ET [9,10,16]. Whether the lower thrombotic risk is conferred by the presence of the CALR mutation or due to the mere absence of the JAK2 mutation is unclear and the prognostic effect of CALR on overall survival is conflicting [9,17]. In this retrospective study, we aim to describe the clinicopathologic features of ET in different genotypes in an Asian population and also determine the prognostic significance of JAK2 and CALR mutation status on thrombotic events and overall survival.

Patients and samples

      We retrospectively identified 331 consecutive patients with ET at National University Hospital of Singapore (NUH), Tan Tock Seng Hospital (TTSH), and Singapore General Hospital (SGH)between 1990-2015. Diagnosis of ET was in accordance with the British Committee of Society of Hematology (BCSH) guideline [18]. Test samples were archival genomic DNA samples in the Molecular Diagnostic Centre of NUH. Out of these 331 cases, 126 cases were JAK2 -wild type and were subjected to CALR mutation testing. CALR mutation was detected by conventional Sanger Sequencing method on the CALR exon 9 gene by our Molecular Diagnostic Centre. This would result in three genotypes as follow: JAK2 -mutated is JAK2 / CALR , CALR -mutated is JAK2 / CALR , and CALR -wild type is JAK2 / CALR . Baseline characteristics and other clinical information were extracted from our Computerized Patient Support System. This study was approved by the respective Institutional Review Boards.

Statistical Analysis

      Baseline demographic and clinical characteristics were summarized by mean and Standard Deviation (SD) for continuous variables with approximately normal distribution, and frequency and percentage for categorical variables. Features were compared across three genotypes ( JAK2 mutated, CALR mutated and both JAK2 and CALR unmutated (double-negative) using one-way ANOVA for continuous variables and Fisher’s exact test for categorical variables. Pair wise comparison ( JAK2 mutated versus CALR unmutated, JAK2 mutated versus CALR mutated and CALR mutated versus CALR unmutated) was subsequently conducted using the independent two-sample t-test and the Fisher’s exact test with Bonferroni correction. Baseline characteristics were also compared between JAK2 mutated and JAK2 wildtype (irrespective of CALR mutation status)by the independent two-sample t test and the Fisher’s exact test for continuous and categorical variables respectively.

      We adopted a competing risks approach where thrombotic event, peripheral blast and death were considered as competing events. The proportional sub-distribution hazards regression [19] was used to study the association between different genotypes and clinical outcomes of thrombotic event and peripheral blast. The corresponding cumulative incidence curves were plotted and compared using the Gray’s test [20]. To look at the independent predictors of thrombotic event, we further performed a multivariable analysis by adjusting for age, White Blood Cell (WBC) and plts.

      The Overall Survival (OS) across the three genotypes were plotted using the Kaplan-Meier survival curves and compared by the Log-rank test. The association between each genotype and overall survival was studied by the univariate Cox regression analysis.

      All statistical analyses assumed a two-sided test with a significance level of 5%, and were performed using the statistical software Stata SE 14 (StataCorp LP, College Station, Texas, USA) and the statistical package cmprsk in R (www.r-project.org).

Prevalence of JAK2 and CALR mutation

      JAK2 mutation was found in 61.9% of ET patients while CALR mutation was found in 12.1% (or 31.7% among JAK2 wild type ET). The MPL mutation was only performed in 9 out of 331 cases, and no positive result was observed (qual resultado positive?). Type 1 and Type 2 CALR mutations constituted 83% of CALR mutations with each having a similar frequency of 41.5%. The remaining CALR mutations consisted of deletions and insertions of various lengths, or a combination of both (Table 1). There was no significant difference in gender, age, Hb, Plt, WBC, Lactate Dehydrogenase (LDH), thrombotic events, and incidence of raised peripheral blast of >5% (PB5) between type 1 and type 2 mutations.

Demographic and clinical characteristics

      Demographic and baseline clinical characteristics as well as the comparison between different genotypes are shown in Table 2. The mean age of the cohort was 61 years (SD=16) and 78.5% of all patients were Chinese ethnicity. Gender distribution was balanced in the overall group but male gender was more common in CALR -mutated ET (male: female ratio =1.5) while female sex was more common in CALR unmutated ET (male: female ratio =0.54). CALR -mutated group were significantly younger than JAK2 -mutated group (mean±SD: 56±16 and 65±15 years respectively, p=0.005).

      The mean presenting WBC count was 11.7 x 109 /L (SD=5.9 x 109 ), Hb level was 13.3g/dL (SD=2.0), and Plt count was 804.7 x 109 /L (SD=339.3x 109 ). 4 patients had low Hb which were unrelated to ET (one patient had severe B12 deficiency, one had severe iron deficiency and two patients had active gastrointestinal bleeding).

      JAK2 -mutated group had a significantly higher Hb than JAK2 - wild type group (mean±SD: 13.5±2.0g/dL and 12.9±1.9g/dL respectively, p=0.002). There was no statistical difference in Hb levels between CALR -mutated and CALR -ET.

      JAK2 -mutated ET had a significantly higher presenting WBC count compared to JAK2 -wild type ET (p=0.002) with no statistical difference between CALR -mutated and CALR - unmutated ET.

      CALR -mutated group has a significantly higher platelet count at presentation compared to Jak2-mutated or CALR -groups (p<0.001). The mean Plt count for JAK2 -mutated, CALR -mutated and CALR - unmutated ET were 784, 1066, and 734 x 109 /L respectively. LDH level at diagnosis in double negative group was significantly lower when compared to JAK2 -mutatedor CALRmutated ET. Overall, 5.7% of ET patients had splenomegaly at diagnosis (8.2%, 2.8%, and 2.5% for JAK2 -mutated, CALR -mutated, and CALR - unmutated ET respectively). Although JAK2 - mutated ET appeared to have the higher rate of splenomegaly but this was not statistically significant.

Incidence of thrombotic events

      A total of 78 thrombotic events were documented from the time of diagnosis until the last follow-up. 27 (8.2%) patients presented with a thrombotic event at diagnosis (24 JAK2 -mutated, one CALR -mutated and two CALR -wild type). JAK2 -mutated ET had a significantly higher incidence of thrombotic event sat presentation when compared to Jak2-wild type ET, p=0.002 (Table 2). Similarly, JAK2 -mutated ET had a higher cumulative incidence of thrombosis at 10 years compared with CALR -mutated (crude sub-distribution hazard ratio (SHR) 5.59, p=0.011) or CALR -wild type ET (SHR 6.55, p=0.002) (Table 3). There was no difference in thrombotic event at diagnosis (p=1.0) or cumulative incidence of thrombosis at 10 years (SHS 1.17, p=0.856) between CALR mutated and CALR -wild type ET patients (Table 2&3). The 10-year estimated cumulative incidence of thrombotic events for JAK2 -mutated, CALR -mutated and CALR -wild type ET were 34%, 3% and 6% respectively (Figure 1).

      In multivariate analysis, after adjusting for age, WBC and platelet count, JAK2 mutation status remained an independent predictor of subsequent thrombotic events (adjusted SHR = 5.98, 95% CI: 1.84-19.44, p=0.003) (Table 4).

Increase in peripheral blast

      The overall incidence of PB5% (from the time of diagnosis until the last follow-up) was 1.5% (5patients). The PB5% incidence was 7.3% (n=3), 1.0% (n=2) and 0% for CALR -mutated, JAK2 - mutated and CALR -unmutated groups respectively (p=0.020, Fisher’s exact test). This incidence was significantly higher in CALR -mutated ET. With competing risk analysis, the crude SHR for CALR -mutated group compared to JAK2 -mutated group was 10.2 (95%CI: 0.86-121.11, p=0.066). The estimated 10 years cumulative incidence of PB5% were 12%, 1%, and 0% for CALR mutated, Jak2-mutated and CALR -wild type groups respectively (p=0.013) (Figure 2).

Total weekly dose of hydroxyurea

      The total weekly dose of hydroxyurea needed to control the platelet count below 600 x 109 /L was calculated. The mean±SD (g) for CALR -mutated, JAK2 -mutated, and CALR -wild type ET were 6.02 ± 2.95, 4.05 ± 2.14, and 4.37 ± 1.82 respectively, p<0.001 (Table 2). CALR -mutated ET patients required a significantly higher total weekly dose of hydroxyurea. Pair wise comparison (with Bonferroni correction) confirmed that CALR mutated ET patients required a significantly higher dose of hydroxurea when compared to JAK2 mutated ET(p<0.001) or CALR unmutated ET patients (p=0.019) (Table 1).

Overall survival for ET

      Median follow-up for the overall cohort was 4.3 years ( JAK2 - mutated 4.3 years, CALR -mutated 5.1 years, and CALR -wild type 3.4 years). The estimated 10-year OS for CALR -mutated, JAK2 - mutated and CALR -wild type ET were 89%, 76%, and 84% respectively (p=0.217) (Figure 3).

      The main purpose of this study was to clarify the thrombotic risk of CALR mutated ET as compared to CALR wild type ET. It was often reported in the initial studies that CALR mutated ET has lower thrombotic risk compared to Jak2 mutated ET, however the comparison with CALR wild type ET was mostly not reported [9,10,14]. We also seek to ascertain the prevalence of difference genotypes of ET in the Asian population and their clinical phenotypes.

      The prevalence of JAK2 mutation on a large cohort of ET patients was first reported in year 2005 [21]. The prevalence was reported to be around 53% in this large European series, while a North America group reported a prevalence of 50% (n=605) [22]. In the Asian context, Lin and co-workers reported a prevalence of 58.4% in cohort of 428 Chinese patients [23]. In a separate group, Taiwanese investigators reported a prevalence of 63.9% [16]. We have also reported a higher prevalence (61.9%) in this predominantly Chinese population. These could suggest that Chinese population has a relatively higher prevalence of JAK2 -mutated ET compared to European or North American populations. On the other hand, the Japanese group reported the incidence that was quite similar to Western report [13].

      CALR mutation was first discovered almost concurrently by two groups of investigators. Nangalia et al. reported an incidence of 82% in the JAK2 -wild type ET [10], and Klampfl et al. reported a rate of 67% with the latter having a larger cohort [9]. A North America group has also reported an incidence close to 67% in JAK2 - wild type ET [17]. On the other hand, the Chinese group reported an incidence of 54.5% within JAK2 -wt ET [23]. However, the Taiwanese group reported a slightly higher incidence of 62.3% in their 147 case of JAK2 non-mutated ET. The Japanese group reported a much lower incidence (39.6%) [13] while our cohort showed the lowest incidence (31.7%). It appeared that the prevalence of Jak2 mutation and CALR mutation among ET patients in the Far East Asia, in particular Chinese population are quite different from the Western population. The differences in prevalence of JAK2 and CALR mutations in our study may reflect our use of BCSH diagnostic criteria while the other studies used WHO 2008 to diagnose ET. Differences in CALR assay sensitivity may also play a role.

      Even though the gender distribution in our cohort was quite equal (male to female ratio of 1.0), a striking gender discrepancy between different mutation groups were observed. Males appeared to be more prevalent in the CALR -mutated group (male 60% and female 40%), while females appeared to be more prevalent in the CALR -wild type group (male 34.9% and female 65.1%), p=0.009. This finding concurred with a meta-analysis that looked into gender distribution between CALR -mutated and JAK2 -mutated group [24]. CALR -mutated MPN are often younger compared to other genotypes. Rumi and co-workers from Italy reported a median age of 45 and 50 years for CALRmutated and JAK2 -mutated ET respectively (p=0.001) [25]. We found a similar pattern with our cohort. The mean age for JAK2 - mutated, CALR -mutated and CALR - unmutated ET were 65, 56 and 53 years respectively (Table 2). It is important to note that there was no significant difference between CALR -mutated and CALR -wild type groups. This would suggest that JAK2 -mutated ET was associated with older age compared to Jak2-wild type group.

      CALR -mutated ET has been shown repeatedly to have lower Hb compared to JAK2 -mutated ET in a number of studies [16,25-29]. However in our cohort, we could only demonstrate that JAK2 -mutated ET had a significantly higher Hb at diagnosis when compared to Jak2- wt ET (p=0.002) or CALR - wild type ET (p=0.01), but not when compared to CALR -mutated ET. The difference between CALR -mutated and CALR - wild type ET was also not significant. Jak2- mutated ET was almost always associated with elevated WBC count, while CALR - mutated ET was often associated with high platelet count [23,25-29]. We found the same association, but in addition, we have also showed that CALR -mutated ET required higher dose of Hydroxyurea (Hydroxycarbamide) to control the platelet count below 600.

      JAK2 -mutated myeloproliferative neoplasm is often associated with an increased thrombotic risk [30-33]. The overall incidence of thrombosis in JAK2 -mutated ET was reported to be around 21%-41% compared to CALR -mutated ET (10-31%). [9,17,28]. The prevalence of JAK2 mutation was found to be as high as 40% among patients with splanchnic vein thrombosis without clinical manifestation of MPN [34]. The prevalence of CALR mutation was reported to be 0.7% to 2.4% in patients with splanchnic vein thrombosis [35-39]. Since the discovery of CALR mutation in MPNs, it has been widely reported that the incidence of thrombotic in this group was significantly lower compared to JAK2 -mutated group, and many have concluded that CALR mutation is associated with lower risk of thrombotic event [17,25,29]. However, none of these previous studies compared the incidence of thrombosis between CALR -mutated and CALR -wild type ET. We have demonstrated that the incidence of thrombotic event at diagnosis or the cumulative incidence of subsequent thrombotic events was clearly highest among JAK2 -mutated ET compared to CALR -mutated or CALR -wild type ET. However, the incidence was almost identical between CALR mutated and CALR -wild type ET. It was mere presence of Jak2 mutation that conferred an increased risk of thrombosis irrespective CALR mutation status. (see Figure 1, Table 2 & 3).

      In the multivariable analysis that included, age, JAK2 mutation status, WBC, and platelet count as covariates, only JAK2 mutation status stood out as an independent predictor for subsequent thrombotic event. Though there were study groups reported an association of leukocytesis with increased risk of thrombosis [30,40,41], we were not able to demonstrate that in our cohort. Palandri and co-workers further demonstrated that WBC of more than 11.0 x 109 /L was significantly associated with thrombotic events, which we have failed to demonstrate. One of the reasons for this difference may be that we included JAK2 mutation status in the multivariable model, and it is well recognized that JAK2 mutation is associated with higher WBC count [42,43].

      It is well reported that platelet count in ET does not correlate well with thrombotic events [44-46]. In fact, extreme thrombocytosis is a risk factor for bleeding due to acquired von Willebrand disorder. Similarly, one group reported that a platelet count >1000 x 106 /micro L was associated with a significantly decreased risk for arterial thrombosis (HR 0.42; 95% CI 0.22- 0.78) [30]. With the discovery of CALR mutation in myeloproliferative neoplasm, which is often associated with high platelet count but with lower risk of thrombosis could explain this paradox [17,25,29]. This raises the question of the role cytoreduction therapy in preventing thrombotic events. An anti-platelet therapy may be adequate as long as the platelet count is not in a range of increased risk of bleeding, i.e more than 1000 x 109 /L.

      The association of CALR mutated ET with higher risk of myelofibrotic progression and leukemic transformation is inconclusive. The Italian group reported no significant increase in leukemic or myelofibrotic transformation in CALR mutated ET [25], so did Tefferi and co-workers [17]. The latter reported an incidence of leukemic transformation of 5% and 8.4% for JAK2 mutated and CALR mutated ET respectively. A Belgian cohort reported an increase risk of progression to myelofibrosis in CALR mutated compared to JAK2 mutated ET [47]. In another separate report, the investigators found the type 1 CALR mutation was associated with myelofibrotic progression in ET [48]. We were unable to fully determine the incidence of myelofibrotic or leukemic transformation accurately as most patients declined a repeat bone marrow examination at the point of suspected disease progression and were on palliative management. However, we did observe that CALR mutated ET had a higher incidence of increased peripheral blasts on long-term follow-up. Estimated 10 years cumulative incidence of PB5% were 12%, 1% and 0% for CALR -mutated, JAK2 mutated and CALR -wild type ET respectively, p=0.013) suggesting an increased risk of myelofibrotic or leukemic transformation in CALR mutated ET. Further confirmation is required from a larger cohort with well-documented blast or myelofibrotic transformation.

      After two initial reports showing superior survival in CALRmutated ET [9,47], most groups could not demonstrate the superior survival of CALR mutated ET [16,17,25-27,29]. In our cohort the estimated 10-year OS for CALR mutated, JAK2 mutated and double-negative groups were 89%, 76%, and 84% respectively (p=0.217).

      More than 80% of the mutations were reported to be due to the two commonest mutations, type 1 and 2 [9,49,50] with type 1 being the most common (45-63.3%). We reported a similar incidence for both the type 1 and 2 – 41.5%. In 114 cases of CALR mutated ET, Tefferi and co-workers demonstrated that type 2 mutation had a significantly higher platelet count compared to type 1 [49]. Riera’s group with a smaller cohort (n=60) demonstrated that type 1 mutation was associated with male gender while type 2 mutation was associated with younger age [50]. In our cohort, we were unable to demonstrate any of these differences between type 1 and type 2. There was also no difference in term of other clinical parameters or thrombotic event. We have detected several complex mutations of various lengths (deletions and/or insertions). In almost all mutants, CALR proteins possess an altered C-terminus with a longer peptide stretch caused by a disrupted reading frame due to these frame-shift mutations.

      The main limitation of this study was the missing MPL mutation status. It was only performed in 9 out of 331 cases. The possible reasons being the test was introduced not long before the CALR mutation testing and not many physician were aware of it or feel the need to perform it. Nevertheless, the reported incidence of MPL mutated ET was less than 5% [51,52], and that would not have significant impact on the analysis.

      We reconfirmed some of the unique clinical features of CALR mutated ET reported in the literature. In addition, we have also demonstrated that the prevalence of JAK2 mutation and CALR mutation in Asians with ET differs from studies related to Western population. The presence of JAK2 mutation increases the risk of thrombosis, regardless of CALR mutation status. Although the presence of CALR mutation was associated with a higher platelet count, this did not appear to have any prognostic significance for cumulative incidence of thrombosis or overall survival when the JAK2 status is taken into account. Moving forward, we hope to better characterize the incidence of MPL in relation to Jak2 and CALR mutation status. The test has been incorporated as a triple test for MPN cases in particular ET and MF cases since year 2015.