Mutation status of essential thrombocythemia and primary myelofibrosis defines clinical outcome

For the majority of patients with essential thrombocythemia (ET) and primary myelofibrosis (PMF), clinical diagnosis can be aided by mutational findings in JAK2, CALR or MPL. The JAK2 V617F mutation is found in about half of the cases and has been included as a major criterion for diagnosis in the WHO 2008 classification.1 A minor proportion has instead been reported to harbor mutations in the MPL gene. The most frequently reported mutations, W515L and W515K, have been found in up to 5% of ET and 10% of PMF patients.2 In remaining cases, mutations in exon 9 of the CALR gene have filled a substantial part of the molecular diagnostic gap.43 Although there is a diagnostic value of mutations in the three genes JAK2, CALR and MPL, the impact of the mutation type correlated to the clinical outcome for patients has not been fully elucidated. Reports indicate that the outcome varies with different mutations. The clinical course of patients with CALR-mutated disease seems to be more indolent with a lower risk of thrombosis in ET and better overall survival (OS) in PMF compared to JAK2 mutated patients.5 It has been suggested that myeloproliferative neoplasms (MPNs) might be diagnosed based on mutation type to better match prognosis, rather than the classical phenotypical division.6 Moreover, a small group of ET and PMF patients are triple-negative (TN), i.e. lacking any of these mutations. This indicates that mutations in other genes might be important for the development of MPN. If so, these mutations could be useful as diagnostic or prognostic markers. In this study, we searched for additional mutations in the TN group and compared the findings and clinical data to patients with mutations in JAK2, CALR or MPL, to study if gene mutations could diversify diagnosis and serve as prognostic markers. The type of mutation was found to have prognostic implications, and TN ET and PMF are uncommon but indicate inferior survival.

91% of all patients diagnosed with ET or PMF between 2008–2013 in Western Sweden were analyzed for mutations in JAK2, CALR and MPL (Figure 1). Among ET patients, 12% were TN while 19% were TN in the PMF group. The heterogeneous nature and lack of recurrent mutations led us to critically review the diagnosis of all TN patients to make sure the initial diagnosis was correct. Two ET patients with TN disease were assigned as PMF instead. Four patients (two ET and two PMF) were excluded because they could not be classified as MPN according to the current diagnostic criteria.

Figure 1.Study design and distribution of mutations. Patients diagnosed with ET or PMF between 2008 and 2013 in Western Sweden. All patients were initially analyzed for the JAK2 V617F mutation. Since mutations in JAK2, CALR and MPL have been reported to be mutually exclusive in MPN, patients without JAK2 mutations were further analyzed for mutations in CALR exon 9 and, if negative, also for the MPL W515K/L mutations as described in Online Supplementary Methods. After a critical review of the diagnosis of TN patients and the findings of additional MPL mutations, patients were regrouped and excluded if they could not be classified as MPN.

To search for other mutations, a sequencing panel including genes associated with myeloid malignancies was used. All TN patients and controls in the ET group harboring JAK2, CALR or MPL mutations were analyzed (Online Supplementary Methods). Interestingly, the number of patients harboring MPL mutations doubled after extended analysis. One ET case had a S505N mutation. This mutation was originally associated with familial thrombocythemia7, but has since been identified as also being a somatic mutation.8 The other mutations affected the W515 codon, but led to alternative amino acid substitutions. One ET and one PMF patient harbored a W515A mutation, one ET patient had a W515R mutation and one had a W515R mutation in combination with a Q516E mutation. Both the W515R and W515A mutations have been reported as rare, pathogenic mutations.9 In the current patient cohort, however, these were as frequent as the W515L mutation. This finding warrants more extensive screening for MPL mutations in a diagnostic setting, e.g. with a sequencing panel or high-resolution melting technique.

The discovery of additional MPL mutations in combination with a review of the diagnosis led to a regrouping of the patients which decreased the TN ET to 6% (Figure 1). Thus, it is relatively uncommon with ET without mutations in JAK2, CALR or MPL.

To investigate if the type of mutation could serve as a prognostic factor, correlation to clinical outcome was performed for the four groups (JAK2, CALR, MPL and TN) after regrouping (Online Supplementary Table S1 and S2). This revealed fewer vascular complications (P=0.035) and a milder clinical course, in spite of significantly higher platelet levels (P=0.0002) in ET patients with CALR mutations than in patients in the other mutation groups. ET cases with JAK2 V617F mutation had features similar to polycythemia vera (PV), but importantly without meeting the diagnostic criteria for this disease. These patients had significantly higher hemoglobin and hematocrit values and lower EPO values (P=0.0007, P=0.0081 and P<0.0001 respectively) compared to non JAK2-mutated patients. This strengthens the idea that the phenotypical division in PV and ET should be substituted for a classification based on the type of mutation status to better match clinical prognosis.6 ET patients harboring a MPL mutation had significantly lower hemoglobin and hematocrit values and consequently higher EPO values (P=0.0002, P=0.0004 and P=0.0094 respectively). These patients also had a significantly higher proportion of transformation to secondary MF and acute myeloid leukemia (AML, P=0.031) compared to the other mutation groups. Two out of seven ET patients with a MPL mutation had transformed (both to AML), in contrast to two out of all the remaining 118 patients. PMF patients with TN disease also had a significantly higher proportion of transformation to AML (P=0.024) compared to the other groups of mutations. Similar findings were found in the PMF patient group as in the ET group, but no other differences regarding either clinical data or vascular complications were significant in this smaller group of patients.

Survival analysis of the different mutation groups was performed on all patients. Both ET and PMF patients showed a significant difference in OS based on mutation (Figure 2). Patients with a CALR-mutated disease had a superior OS rate. Among the ET patients, there was a significantly inferior OS rate for patients with MPL-mutated disease. Only two patients were still alive at the end of the study. Both surviving patients had the W515L mutation, which previously has not been associated with inferior survival.8 However, since the median age of patients with MPL mutations was significantly higher than in the other groups, survival was also analyzed in patients aged 70 and over. This still showed a pattern of inferior survival of patients with MPL mutated ET, although no longer statistically significant. No significant difference in survival was seen among ET patients in the JAK2 or TN groups compared to the rest of the patients (P=0.8537 and P=0.1759 respectively). In the PMF group TN patients had a significantly shorter survival time. This indicates a different clinical course with worse prognosis for this subgroup.

Figure 2.Survival correlated to gene mutations. Survival curves of patients according to mutation profile. (A) Survival of all ET patients (n=125) and PMF patients (n=41) based on mutation group (JAK2, CALR, MPL or TN). (B) Survival of ET patients with CALR mutation (n=28) compared to ET patients without CALR mutation (n=97). (C) Survival of ET patients with MPL mutation (n=7) compared to ET patients without MPL mutation (n=118). (D) Survival of ET patients with MPL mutation aged 70 and above. Patients with MPL mutation (n=7) compared to patients without MPL mutation (n=49). (E) Survival of TN PMF patients (n=7) compared to PMF patients with mutation in JAK2, CALR or MPL (n=34). (F) Survival of ET patients with SRSF2 mutation (n=3) compared to ET patients without SRSF2 mutation (n=48). (G) ASXL1 mutated ET (n=7) compared to ET patients without ASXL1 mutation (n=44).

In addition to mutations in JAK2, CALR and MPL, several other genes have been found to be mutated in MPN, e.g. TET2, DNMT3A and ASXL1.10 In the current study, mutations were found in 35 additional genes (Figure 3 and Online Supplementary Table S3). All but one patient with MPL mutated ET had additional mutations, while in 39% of ET cases with JAK2 mutation (n=7) and 28% with CALR mutation (n=5) no additional mutations were found. This was also the case in three ET and one PMF patient with TN disease. TN patients could have mutations affecting the same signaling pathways as do mutations in JAK2, CALR or MPL, although not included in our gene panel. A more extended search for additional mutations, e.g. with exome sequencing, could be warranted to unravel the molecular mechanism behind their disease.

Figure 3.Mutations in additional genes. Mutations detected by TruSight myeloid sequencing panel in TN patients (ET and PMF) and in ET patients with mutation in JAK2, CALR or MPL. On average two and four additional genes were mutated in ET and PMF, respectively (range 0–11). No clear differences were seen between the mutation groups, neither regarding the number (P=0.10) nor the type of identified mutations.

TET2 was the most frequently mutated gene, mutated in one third of analyzed cases. The large number of different mutations (n=19) makes it difficult to use TET2 as a molecular marker, and TET2 mutations have also been reported in other hematological malignancies.11 Moreover, TET2 mutations, in addition to DNMT3A and ASXL1 mutations, have been reported in apparently healthy individuals (i.e. clonal hematopoiesis) especially in the elderly.12 Herein, mutations in DNMT3A were found in 14% of analyzed samples and ASXL1 in 12% of analyzed samples. When correlating these mutations to survival in the analyzed ET cohort (n=51), patients with ASXL1 mutation were found to have worse OS (Figure 2). This has recently been reported for PMF, but no significant difference for survival in ET was seen.13 In addition, mutation in SRSF2 significantly correlated to inferior OS in ET patients in the current study. Three different mutations affecting the mutational hotspot P95 were found in three ET and three PMF patients. All patients with SRSF2 mutated ET were deceased (median survival 35 months). Only one PMF patient harboring SFSR2 mutation was still alive. Notably, however, the MPL and TN groups harbored most of the mutations in ASXL1 and SRSF2. No significant difference in survival was seen for patients with other mutations. However, survival analyses for mutations in additional genes were performed only on a subset of cases and need to be confirmed in larger patient cohorts.

In conclusion we show that mutations in JAK2, CALR and MPL in ET and PMF also have prognostic implications besides diagnostic value. ET and PMF without mutations in JAK2, CALR and MPL seem to be a rare circumstance, and thus an extended analysis for mutation in these genes is the first priority in diagnosis. Additional gene screening gave limited extra prognostic information, with the exception of SRSF2 and ASXL1. A more extended genetic analysis of TN ET and PMF patients will reveal the validity of these entities.