More than 60% of patients with primary myelofibrosis (PMF) are positive for the JAK2V617F mutation, 20–25% harbour a mutation in CALR, 4–8% show mutated MPL and <20% are ‘triple negative’, that is, without a mutation in the coding regions of JAK2, CALR and MPL.1, 2 Lack of any of the three ‘classic’ mutations has been shown to be prognostically adverse,1, 2 and the prognostic impact of additional mutations, such as mutations in ASXL1 is currently being explored.2 Mutations in the exon 9 of CALR in 80% of affected patients are either of type I or type II.3, 4, 5 The type I mutation in CALR exon 9 is a deletion of 52 bp at L367, the type II mutation in CALR exon 9 is an insertion of 5 bp at K385. In general, CALR-mutated patients are of younger age, display a higher platelet count and haemoglobin level but lower leukocyte counts, and have a favourable overall survival (OS).2 Tefferi et al.3 described 95 patients carrying a CALR mutation (76 patients with type I, 10 patients with type II mutation and 9 patients with different mutations in CALR). Interestingly, the favourable outcome of CALR-mutated patients as compared to patients with a mutation in JAK2 or triple negative patients might be confined to patients with a type I CALR mutation as presence of a type II CALR mutation was associated with significant worse survival.3, 6 The hazard ratio for OS was 2.5 with a 95% confidence interval of 1.1–5.4 favouring CALR type I mutations.3 Similar observations have been reported by an additional group.7 In our recent analysis, we showed that CALR-mutated patients who received allogeneic stem cell transplantation had an excellent OS.8 There, we had observed an OS of 82% at 4 years, which was significantly better than for JAK2V617F mutated or triple negative patients.8
Here, we analysed whether the type of the CALR mutation impacted outcome after reduced intensity conditioning (RIC) followed by allogeneic stem cell transplantation as has been described for the non-transplant setting.3, 6 The mutational status of CALR was detected by Next Generation Amplicon Sequencing using a Personal Genome Machine (PGM, Ion Torrent—Life Technologies GmbH/ Thermo Fisher, Darmstadt, Germany) as well as qualitative PCR to detect CALR type I mutations and digital droplet PCR (ddPCR, BioRad, Munich, Germany) for the detection of CALR type II mutations.9 Forty consecutively treated patients with myelofibrosis and CALR mutation were included. The type I mutation was found in 26 patients (65%), 7 patients (18%) each had the type II mutation or a different mutation. Different mutations were frameshifts at K374 (2 ×), K368, E380, D384fs and K386 (2 ×). Median age of the 13 female and 27 male patients at transplantation was 58 years (range 28–75 years). Myelofibrosis was of post-essential thrombocythaemia or post-polycythaemia vera origin in 11 patients (28%). Bone marrow fibrosis was scored as 1° in 2 patients (5%), 2° in 6 patients (15%) and 3° in 31 patients (78%). Normal karyotype was observed in 18 patients (45%). The Dynamic International Prognostic Scoring System (DIPSS) is a current model to predict OS in patients suffering primary myelofibrosis.10 It incorporates information on age above or below 65 years, on the presence or absence of constitutional symptoms, haemoglobin content of below or above 10 g/dL, WBC count of above or below 25 × 109 cells/L and the presence or absence of blasts in peripheral blood. Upon DIPSS scoring, patients are then stratified into four risk categories: low (median OS not reached in the original publication10), intermediate 1 (median OS 9.8 years), intermediate 2 (median OS 4.8 years) and high risk (median OS 2.3 years). In the cohort analysed here, DIPSS was low risk in 1 patient (3%), intermediate 1 in 8 patients (20%), intermediate 2 in 22 patients (55%) and high in 9 patients (23%). Cytogenetics, in PMF frequently difficult to obtain, were adverse in 5 patients (13%) when the following classification scheme set forth by the DIPSS plus was used, with adverse defined as complex karyotype, +8, -7/7q-, i(17q), -5/5q-, 12p-, inv(3) or 11q23 rearrangement.11 Conditioning for most of the patients consisted of IV busulfan (ten doses of 0.8 mg/kg body weight) and fludarabin (six doses of 30 mg/m2), combined with antithymocyteglobulin (ATG F, Neovii Biotech, Gräfelfing, Germany) at 3 × 10 mg/kg for matched-family donors, 3 × 20 mg/kg for matched-unrelated donors or 3 × 20–30 mg/kg for mismatched donors as recently described.12 All patients received PBSC with a median of 6.5 × 106 CD34 positive stem cells/kg BW (range 2.2–13 × 106). HLA-matched donors were available for 28 patients (70%). An unrelated donor was chosen for 32 patients (80%). Specifically, 8 patients (20%) had a matched-related donor present, 20 patients (50%) were grafted from a matched-unrelated donor and 12 patients (30%) received their transplant from a mismatched-unrelated donor. Post-grafting immunosuppression mainly consisted of cyclosporine and mycophenolic acid. Further characteristics are listed in Table 1, grouped according to CALR mutational status. Here, no striking difference was observed between the groups type I mutation, type II mutation and other mutations (Pearson χ2 test for categorical variables, Kruskal–Wallis test for continuous variables). Although the DIPSS ‘high’ category was only observed in the CALR type I mutated group, this difference was not statistically significant (P=0.233, Pearson χ2 test). Also, adverse cytogenetics did not differ statistically significant between the group analysed (P=0.262, Pearson χ2 test).
Cumulative incidences (CI) of relapse and of non-relapse mortality (NRM) were calculated in a competing risk model using the cmprsk package in R (https://cran.r-project.org/) as described earlier.13, 14 The cumulative relapse incidence was 7±5% (type I), 14±14% (type II) and 29±19% (other CALR mutations) at 3 years, and did not differ significantly (P=0.838, Gray test15). CI of NRM was 12±7% (type I), 0 (type II) and 0 (other CALR mutations) at 1 year, and did not differ significantly (P=0.303, Gray test15). Kaplan–Meier estimated OS at 3 years was 75±9% (type I mutation), 86±13% (type II mutation) and 71±17% (other mutation), and did not differ significantly (P=0.903, Log-rank test, Figure 1, Table 1) between the three groups analysed. Kaplan–Meier estimation and Log-rank testing was performed using SPSS version 24 (IBM, Armonk, NY, USA). For the entire cohort, CI of relapse at 3 years and NRM at 1 year were 13±6% and 5±4%, respectively, resulting in OS of 75±7% at 3 years.
Taken together, the negative prognostic impact of type II mutations in CALR that was observed in the non-transplant setting might be overcome by allogeneic haematopoietic stem cell transplantation. Further analyses with larger patient numbers transplanted will have to address this in more detail. Overall, RIC plus allogeneic HSCT for myelofibrosis with type I and type II CALR mutations as well as with other CALR mutations resulted in excellent OS independent of the disease status.