sign in sign up
<<
Home , Acute myeloid leukemia, Myeloid leukemia

(2014) 28, 1804–1810 & 2014 Macmillan Publishers Limited All rights reserved 0887-6924/14 www.nature.com/leu

ORIGINAL ARTICLE The number of prognostically detrimental and in primary myelofibrosis: an international study of 797 patients

P Guglielmelli1,16, TL Lasho2, G Rotunno1, J Score3, C Mannarelli1, A Pancrazzi1, F Biamonte1, A Pardanani2, K Zoi4, A Reiter5, A Duncombe6, T Fanelli1, D Pietra7, E Rumi7, C Finke2, N Gangat2, RP Ketterling8, RA Knudson8, CA Hanson9, A Bosi1, A Pereira10, R Manfredini11, F Cervantes12, G Barosi13, M Cazzola14, NCP Cross15, AM Vannucchi1,16 and A Tefferi2,16

We recently defined a high-molecular risk category (HMR) in primary myelofibrosis (PMF), based on the presence of at least one of the five ‘prognostically detrimental’ mutated (ASXL1, EZH2, SRSF2 and IDH1/2). Herein, we evaluate the additional prognostic value of the ‘number’ of mutated genes. A total of 797 patients were recruited from Europe (n ¼ 537) and the Mayo Clinic (n ¼ 260). In the European cohort, 167 (31%) patients were HMR: 127 (23.6%) had one and 40 (7.4%) had two or more mutated genes. The presence of two or more mutations predicted the worst survival: median 2.6 years (hazard ratio (HR) 3.8, 95% confidence interval (CI) 2.6–5.7) vs 7.0 years (HR 1.9, 95% CI 1.4–2.6) for one vs 12.3 years for no mutations. The results were validated in the Mayo cohort and prognostic significance in both cohorts was independent of International Prognostic Scoring System (IPSS; HR 2.4, 95% CI 1.6–3.6) and dynamic IPSS (DIPSS)-plus (HR 1.9, 95% CI 1.2–3.1), respectively. Two or more mutations were also associated with shortened leukemia-free survival (HR 6.2, 95% CI 3.5–10.7), also Mayo validated. Calreticulin mutations favorably affected survival, independently of both number of mutations and IPSS/DIPSS-plus. We conclude that the ‘number’ of prognostically detrimental mutations provides added value in the combined molecular and clinical prognostication of PMF.

Leukemia (2014) 28, 1804–1810; doi:10.1038/leu.2014.76

INTRODUCTION powerful predictive value of selected cytogenetic abnormalities, 5 Primary myelofibrosis (PMF), one of the the DIPSS-plus score was developed, which includes the presence negative chronic myeloproliferative neoplasms, is characterized by of unfavorable plus transfusion dependence and significantly reduced overall survival (OS) and increased risk of . acute (AML).1 Median survival is estimated at We recently discovered that a restricted set of mutated genes, 6.5 years, but marked heterogeneity exists with survival ranging namely ASXL1, EZH2, SRSF2 and IDH1/ IDH2, allow to identify from less than 2 to over 10 years.2 Identification of patients at patients with significantly shorter OS and increased risk of AL, 6 high risk of death has immediate therapeutic relevance as stem independent of IPSS- and DIPSS-plus risk stratification. Patients cell transplantation is the only intervention that, to date, has harboring any one of these mutated genes were considered to produced evidence for potential cure or prolongation of survival; constitute a ‘high-molecular risk’ (HMR) category with a hazard unfortunately, stem cell transplantation is associated with ratio (HR) for shortened OS of 2.29 (95% confidence interval (CI) considerable mortality and morbidity, which warrants careful 1.65–3.19) and a HR for AML of 2.96 (95% CI 1.85–4.76) compared patient selection. This prompted the development of different with the ‘low-molecular risk’ category (that is, patients unmutated scoring systems over the years; those currently adopted have for the five genes). The prognostic impact on OS of the HMR been developed by the International Working Group for category has been confirmed also in the prospectively collected Myeloproliferative Neoplasms Research and Treatment. They cohort of patients receiving the Janus kinase 1 (JAK)1 and JAK2 include the International Prognostic Scoring System (IPSS),3 used inhibitor ruxolitinib within the phase III COMFORT-II study.7 at diagnosis and employing five hematologic and clinical variables Recently, mutations in the calreticulin (CALR),8 expressed (age, , , peripheral blasts and by 50–80% of patients with JAK2 and MPL unmutated essential constitutional symptoms) to stratify patients in a low, thrombocythemia and PMF,9,10 have been shown to bear a intermediate-1, intermediate-2 and high-risk category, and the favorable impact on survival10 and, in case of essential dynamic IPSS (DIPSS)4 for patients evaluated after diagnosis that thrombocythemia, on the risk of thrombosis.11,12 In particular, in relies upon the same five variables. Finally, having recognized the a cohort of 254 PMF patients studied at Mayo Clinic, CALR

1Laboratorio Congiunto MMPC, Department of Experimental and Clinical Medicine, University of Florence, Florence, Italy; 2Division of , Department of Internal Medicine, Mayo Clinic, Rochester, MN, USA; 3Wessex Reg Genetics Lab, Salisbury, UK; 4Haematology Research Laboratory, Biomedical Research Foundation, Academy of Athens, Athens, Greece; 5III. Medizinische Klinik, Universita¨tsmedizin Mannheim, Mannheim, Germany; 6Department of Haematology, Southampton University Hospital, Southampton, UK; 7Department of Hematology , Fondazione IRCCS Policlinico San Matteo, University of Pavia, Pavia, Italy; 8Division of , Department of Laboratory Medicine, Mayo Clinic, Rochester, MN, USA; 9Division of , Department of Laboratory Medicine, Mayo Clinic, Rochester, MN, USA; 10Hemotherapy & Hemostasis, Hospital Clı´nic, Barcelona, Spain; 11Centre for Regenerative Medicine ‘Stefano Ferrari’, University of Modena and Reggio Emilia, Modena, Italy; 12Hematology Department, Hospital Clinic, IDIBAPS, University of Barcelona, Barcelona, Spain; 13Center for the Study of Myelofibrosis, IRCCS Policlinico S.Matteo Foundation, Pavia, Italy; 14Department of Hematology Oncology, IRCCS Policlinico San Matteo Foundation & University of Pavia, Pavia, Italy and 15Faculty of Medicine, University of Southampton, Southampton, UK. Correspondence: Dr AM Vannucchi, Laboratorio Congiunto MMPC, Department of Experimental and Clinical Medicine, University of Florence, Florence 50134, Italy or Dr A Tefferi, Division of Hematology, Department of Medicine, Mayo Clinic, Rochester, MN 55905, USA. E-mail: amvannucchi@unifi.it or [email protected] 16These authors contributed equally to this work and share correspondence. Received 11 January 2014; accepted 14 February 2014; accepted article preview online 19 February 2014; advance online publication, 7 March 2014 Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1805 mutations showed a favorable impact on survival that was sixty-seven patients (31.1%) presented at least one of the prognos- independent of DIPSS-plus score and contributed with ASXL1 tically detrimental mutated genes (that is, ASXL1, EZH2, SRSF2 and mutations to delineate prognostically different mutational IDH1/2) and were accordingly classified as HMR; 127 (23.6%) patients categories of patients.13 In a subsequent study including 570 had one mutation and 40 (7.4%) had two or more mutations, PMF patients, it was found that survival was the longest in accounting for 76.0% and 24.0%, respectively, of the HMR cohort. CALR þ ASXL1 À and shortest in CALR–ASXL1 þ patients, whereas CALR þ ASXL1 þ and CALR–ASXL1 À patients had similar survival and were grouped together in an intermediate-risk category. The Validation cohort. In the Mayo Clinic cohort, the median prognostic model based on CALR/ASXL1 mutational status was follow-up was 4.3 years (range, 0.13–20); 188 patients died independent of IPSS and DIPSS-plus, and in multivariable analysis (72%) and 31 (12%) had progressed to AML (Table 1). According that included all eight DIPSS-plus risk factors, CALR–ASXL1 þ to the DIPSS-plus criteria, 13, 19, 35 and 33% were low, mutational status was identified as the most significant risk intermediate-1-, intermediate-2 and high risk, and their respective factor for survival.14 survival was 15, 8.2, 5.1 and 2.3 years (Po0.0001). One hundred The aim of the current study was to investigate the effect of the fifteen patients (44.2%) presented at least one of the prognos- ‘number’ of prognostically detrimental mutated genes, comprising tically detrimental mutated genes and were accordingly classified the HMR category (ASXL1, EZH2, SRSF2 or IDH1/2), on overall and as HMR; 92 patients (35.3%) had one mutation and 23 (8.8%) had leukemia-free survival (LFS), in two large cohorts of patients with two or more mutations, accounting for 80% and 20%, respectively, PMF: a ‘test cohort’ from Europe, including 537 patients who were of the HMR cohort. genotyped at diagnosis and IPSS-stratified, and a ‘validation cohort’ from Mayo Clinic, including 260 patients who were analyzed after diagnosis, DIPSS-plus stratified. The impact of CALR mutations on the overall results was also evaluated, in light of the recently highlighted prognostic stratification model based on Table 1. Clinical and laboratory features of the 797 patients with CALR and ASXL1 mutational status.14 primary myelofibrosis from the test (European) and validation (Mayo Clinic) cohort

Variables European cohort Mayo Clinic cohort SUBJECTS AND METHODS (at time of (at time of referral) The current study was approved by the institutional review boards of the diagnosis) (n ¼ 260) different European institutions and the Mayo Clinic, Rochester, MN, USA. (n ¼ 537) All patients provided informed written consent for study sample collection as well as permission for its use in research. Inclusion to the current study Age in years; median (range) 62 (15–90) 64 (32–87) Males (%) 332 (62) 167 (64) required availability of archived peripheral blood/ or bone Hemoglobin, g/dl; median (range) 11.5 (4.4–16.5) 10.4 (5.8–16.1) marrow sample collected at the time of diagnosis or within one year Leukocytes, Â 109/l; median (range) 9 (1.4–106) 9.5 (1.0–219) (European cohort) or first referral (Mayo cohort); a total of 537 samples , Â 109/l; median (range) 328 (7–3279) 228 (11–2466) were available in the European cohort, contributed by 10 different Constitutional symptoms; n (%) 176 (32.8) 91 (35) institutions, and 260 samples in the Mayo cohort. The diagnoses of PMF Circulating blasts X1%; n (%) 85 (15.8) 144 (55) and leukemic transformation were according to WHO criteria.15 Cytogenetic categories; n (%) ‘N’ evaluable ¼ 272 ‘N’ evaluable ¼ 257 Unfavorable karyotype designation and IPSS or DIPSS-plus risk Abnormal 76 (27.9%) 97 (37.7%) Unfavorable karyotypec 20 (7.3%) 26 (10.1%) categorization were as previously described.3,16,17 Mutation analysis was performed in DNA from whole blood, purified IPSSa risk group; n (%) or samples using real-time quantitative-PCR, Low 196 (36) NA high resolution melting and direct sequencing, as more appropriate for the Intermediate-1 166 (31) NA different genes analyzed; details of the procedure have been reported Intermediate-2 95 (18) NA High 80 (15) NA previously.6 Mutations in exon 9 of CALR were assessed by bidirectional 13 sequencing. DIPSS-plusb risk group; n (%) All statistical analyses considered clinical and laboratory parameters Low NA 34 (13) obtained at the time of diagnosis (European cohort) or first referral (Mayo Intermediate-1 NA 49 (19) cohort). Differences in the distribution of continuous variables between Intermediate-2 NA 91 (35) categories were analyzed by either Mann–Whitney (for comparison of two High NA 86 (33) groups) or Kruskal–Wallis (comparison of three or more groups) test. JAK2V617F; n (%) 331 (61.6) 145 (56) 2 Patient groups with nominal variables were compared using w -test. MPLW515; n (%) 30 (5.6) 19 (7.3) OS analysis was considered from the date of diagnosis (European cohort) CALR mutated; n/evaluable (%) 70/394 (18) 63/260 (24) or first referral (Mayo cohort) to date of death (uncensored) or last contact ASXL1 mutated; n (%) 122 (22.7) 81/256 (32) (censored). Date of leukemic transformation replaced date of death, as the EZH2 mutated; n (%) 29 (5.4) 15/257 (5.8) IDH1/2 mutated; n (%) 12 (2.2) 16 (6.2) uncensored variable, for estimating LFS. OS and LFS curves were prepared SRSF2 mutated; n (%) 45 (8.4) 32/259 (12.4) by the Kaplan–Meier method and compared using the log-rank test. Cox HMR; n (%) 167 (31.1) 115 (44.2) proportional hazard regression model was used for multivariable analysis. Progression to leukemia; n (%) 89 (16.6) 31 (12.3) P-values o0.05 were considered significant. The SPSS software was used Death; n (%) 195 (36.3) 188 (72.3) for all calculations of the European cohort and the Stat View (SAS Institute, Cary, NC, USA) statistical package was used for the Mayo cohort.11 Abbreviations: DIPSS-plus, dynamic International Prognostic Scoring System-plus; HMR, high-molecular risk category; IPSS, International Prognostic Scoring System; NA, not applicable. aIPSS uses five independent predictors of inferior survival: age 465 years, hemoglobin o10 g/dl, RESULTS leukocytes 425 Â 109/l, circulating blasts X1%, constitutional symptoms. Characteristics of patients in the two cohorts The presence of 0, 1, 2 and X3 adverse factors defines low, intermediate-1, Test cohort. In the European cohort, median follow-up was 4.1 intermediate-2 and high-risk disease. bDIPSS-plus uses three more years (range, 0.3–29.2); 195 patients died (36.3%) and 89 (16.6%) variables in addition to IPSS’ ones to predict inferior survival: red cell 9 had progressed to AML (Table 1). According to the IPSS criteria, 36, transfusion dependency, count o100 Â 10 /l and unfavorable X 31, 18 and 15% were low, intermediate-1-, intermediate-2 and karyotype. The presence of 0, 1, ‘2 or 3’ and 4 adverse factors defines low, intermediate-1, intermediate-2 and high-risk disease. cUnfavorable karyo- high risk, and their respective survival was 22.8, 10.5, 5.6 and 2.9 type: complex karyotype or sole or two abnormalities that include þ 8, years (Po0.0001). An unfavorable karyotype was found in 7.3% -7/7q-, i(17q), inv(3), -5/5q-, 12p- or 11q23 rearrangement. of the 272 patients with cytogenetic information. One hundred

& 2014 Macmillan Publishers Limited Leukemia (2014) 1804 – 1810 Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1806

Figure 1. Survival data of 537 patients with primary myelofibrosis from the European cohort stratified by the number (zero, one, two or more) of prognostically detrimental mutations (ASXL1, EZH2, SRSF2 or IDH1/2 (a)). The impact of the number of prognostically detrimental mutations on survival of patients stratified by IPSS category is shown in b. for the lower (low plus intermediate-1) risk categories and in c for the higher (intermediate-2 plus high) risk categories.

Effect of the number of prognostically detrimental mutated genes one mutation (HR 2.7; 95% CI, 1.4–5.2; P ¼ 0.004), whereas the on survival difference was borderline in the higher-risk category (HR 1.6; 95% CI Test cohort. In the test cohort, the OS of the 167 HMR patients was 0.9–2.9, P ¼ 0.059). significantly reduced (5.9 versus 12.3 years; Po0.0001) compared with low-molecular risk patients (that is, those with no prognos- Validation cohort. The seminal findings in the European cohort tically detrimental mutated gene; n ¼ 370); in a multivariate analysis were validated in the Mayo series. Kaplan–Meier estimates of including IPSS as covariate the HR was 1.7 (95% CI 1.3–2.3; survival showed that patients having either two or more or one Po0.0001) for HMR versus low-molecular risk category. mutations had significantly shorter survival (2.4 years; HR 3.4, 95% We then considered separately patients who had no mutations CI, 2.1-5.6 and 3.0 years; HR 1.8, 95% CI, 1.3–2.5, respectively) (low-molecular risk) versus those who presented with one or X2 compared with zero mutations (median survival, 6.6 years; mutated genes, in order to determine the prognostic impact of the Figure 2a). Having two or more mutations was significantly worse ‘number’ of mutations. In univariate analysis, the presence of either than one mutation (HR 1.7, 95% CI, 1.0–2.8; P ¼ 0.04). When two or more (median survival: 2.6 years; HR 3.8, 95% CI 2.6–5.7) and adjusted for DIPSS-plus, the presence of two or more mutations one mutated gene (median survival: 7.0 years; HR 1.8, 95% CI 1.3–2.5) retained its significance (HR 2.1; 95% CI, 1.3–3.6). In DIPSS-plus low was significantly more detrimental for OS compared with no and intermediate-1 risk categories, median survival was signifi- mutations (median survival: 12.2 years; Po0.0001) (Figure 1a). cantly affected by having two or more mutations compared with Having two or more mutated genes corresponded to a HR of 2.0 either zero (HR 23.8, 95% CI, 5.8–98.0; Po0.0001) or one mutations (95% CI 1.3–3.1) compared with one mutated gene (Po0.001). IPSS- (HR 29.1, 95% CI, 2.9–290.7; P ¼ 0.004; Figure 2b); in the DIPSS-plus adjusted multivariable analysis confirmed the independent prog- intermediate-2 and high-risk categories, median survival was 3.1 nostic relevance of two or more mutations for OS (HR 2.4, 95% CI years (HR 1.9; 95% CI, 1.1–3.2), 2.6 years (HR 1.5, 95% CI 1.0–2.1) 1.6–3.6; Po0.0001), which was evident in both the lower (IPSS low and 4.1 years for patients with two or more, one or zero mutations, and intermediate-1; median survival was 3.2, 11.3 and 20.2 years for respectively (P ¼ 0.02; Figure 2c), however having two or more patients with two or more, one or zero mutations, respectively; mutations was not significantly different from one. Po0.0001) (Figure 1b) and higher (IPSS intermediate-2 and high risk; median survival was 2.2, 4.4 and 5.7 years for patients with two Effect of the number of prognostically detrimental mutated genes or more, one or zero mutations, respectively; P ¼ 0.02) risk on progression to leukemia categories (Figure 1c). In the lower IPSS risk categories, two or Test cohort. The LFS was significantly affected by the presence of more mutations had a significant impact on survival compared with at least one prognostically detrimental mutated gene in the HMR

Leukemia (2014) 1804 – 1810 & 2014 Macmillan Publishers Limited Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1807

Figure 2. Survival data on 260 patients with primary myelofibrosis of the Mayo cohort stratified by the number (zero, one, two or more) of prognostically detrimental mutations (ASXL1, EZH2, SRSF2 or IDH1/2 (a)). The impact of the number of prognostically detrimental mutations on survival of patients stratified by DIPSS-plus category is shown in b. for the lower (low plus intermediate-1) risk categories and in c for the higher (intermediate-2 plus high) risk categories. category compared with no mutated gene (HR 3.3; 95% CI 2.1–5.0; observed in both the lower (Po0.0001; Figure 4b) and higher Po0.0001). In univariate analysis, having either two or more or one (P ¼ 0.0013; Figure 4c) DIPSS-plus risk categories. prognostically detrimental mutated genes correlated with signifi- Interaction of prognostically detrimental mutations and CALR cantly shortened LFS (6.6 years; HR 6.2, 95% CI 3.5–10.7 for two or mutations on survival more and 11.1 years; HR 2.6, 95% CI 1.6–4.1 for one) compared with CALR mutations were found in 70 of 394 (18%) evaluated PMF no mutations (26.7 years; Po0.0001; Figure 3a). The difference patients of the test cohort and 63 (24%) of 260 patients of the between two or more and one mutated gene was statistically 10,13 significant (HR 2.3; 95% CI 1.3–4.1, P ¼ 0.004). Multivariate analysis validation cohort. Confirming previous data, presence of CALR showed that two or more mutated genes (HR 4.1; 95% CI 2.3–7.4) mutation was associated with significantly better outcome. In the and IPSS high-risk category (HR 5.3; 95% CI 2.7–10.3) independently European cohort, the median survival of CALR mutated patients predicted for shorter LFS. The negative impact of harboring two or was 22.8 years as compared with 9.6 years for CALR unmutated more mutated genes compared with one or no mutated gene on (P ¼ 0.001; HR 0.43 (95% CI, 0.25–0.72). In the Mayo cohort, the the time to progression to leukemia was maintained in the lower median survival of CALR mutated PMF patients was 8.5 years (median LFS was 8.2 years, 14.4 years and not reached for two or as compared with 4.0 years for CALR unmutated (HR 0.4 95% CI 0.3–0.6). Multivariable analysis was then performed to analyze more, one or zero mutated genes, respectively; Po0.0001; Figure 3b) and higher (median LFS was 2.2, 6.4 and 13.0 years for the interaction of the number of prognostically detrimental mutations (that is, one or two or greater mutated genes), the twoormore,oneorzeromutatedgenes,respectively;Po0.0001; Figure 3c) IPSS risk categories. Having two or more mutated genes IPSS (European cohort) and DIPSS-plus (Mayo cohort) risk was significantly worse than one mutation in the higher IPSS risk categories and the CALR mutational status in the two cohorts of category (HR 2.5; 95% CI 1.1–5.8, P ¼ 0.02) and borderline in the PMF patients. In both cohorts, all the variables considered were lower (HR 1.8; 95% CI 0.8–4.4; P ¼ 0.061). independently associated with survival, as detailed in Table 2 that reports the corresponding HR values. Validation cohort. The adverse prognostic impact of two or more mutations for LFS was validated also in the Mayo cohort (Figure 4). DISCUSSION Although in the entire series median LFS was not reached, Current prognostication in PMF is based on the IPSS and/or DIPSS- patients presenting two or more prognostically detrimental plus scores that combine a number of hematological, clinical and mutated genes had significantly shorter LFS compared with cytogenetic variables to stratify patients in four risk categories having one mutation and no mutations (Po0.0001; Figure 4a). The associated with significantly different OS. In addition, higher DIPSS negative impact of the number of mutated genes on LFS was risk categories have been shown to predict progression to AML,18

& 2014 Macmillan Publishers Limited Leukemia (2014) 1804 – 1810 Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1808

Figure 3. Leukemia-free survival of 537 patients with primary myelofibrosis from the European cohort stratified by the number (zero, one, two or more) of prognostically detrimental mutations (ASXL1, EZH2, SRSF2 or IDH1/2 (a)). The impact of the number of prognostically detrimental mutations on leukemia-free survival of patients stratified by IPSS category is shown in (b) for the lower-(low plus intermediate-1) risk categories and in c for the higher-(intermediate-2 plus high) risk categories.

and a revised DIPSS-plus based model showed better resolution mutated genes had significantly reduced OS and LFS compared of patient categories at different risk of AML progression.19 not only with unmutated patients but also to those presenting The presence of selected karyotypic abnormalities (‘unfavorable one mutation, were fully validated in the Mayo cohort. The karyotype’) is prognostically detrimental, yet these involve no prognostic impact of two or more mutated genes was IPSS/DIPSS more than 7% of the patients. As a whole, these scoring systems, independent and provided information in both the lower and although accurate and friendly, cannot account for the intrinsic higher-risk categories. Why two is worse than one cannot be variability of the disease that reflects its complex mutational mechanistically explained at present time, but we argue that it landscape. This is well exemplified by the demonstration that PMF could simply reflect the mutation heterogeneity of PMF resulting patients harboring the novel CALR mutations constitute a well- from clonal instability. It is of interest that a similar negative defined low-risk category when compared with JAK2V617F/MPL impact on survival and risk of transformation into AML was mutated and triple negative patients13 and by the inferior survival recently reported in a study that analyzed 105 -associated of patients who are CALR unmutated but harbor mutations in genes by next generation sequencing in a cohort of 197 patients ASXL1 (CALR À /ASXL1 þ ). As in de novo AML and myelodysplastic with the three classic myeloproliferative neoplasms.20 In patients syndromes, it is very likely that incorporation of somatic mutations with myelodysplastic syndromes, an association of the number of into the prognostic assessment of patients with PMF might result mutations and time to AML was also found.21 Therefore, patients in refinement of prognostic scores. with multiple mutations represent a ‘very high-risk’ category with We recently demonstrated that patients with PMF harboring reduced survival and increased risk of leukemic transformation. mutations in any one of five prognostically detrimental genes These findings may have clinical relevance allowing to identify a (ASXL1, EZH2, SRSF2 and IDH1/2) constitute a IPSS/DIPSS-plus category of patients, not recognized by the clinical scores, who independent HMR category with shorter survival and greater risk may be the best candidate for early intervention with stem cell of AML compared with the unmutated, low risk counterpart. transplantation due to their expected reduced survival and the Observations from the current study that involved almost 800 PMF risk of leukemia. This molecular-based approach has advantages patients from two cohorts, a European test cohort (n ¼ 537) and a also over cytogenetics, as the proportion of patients falling in the Mayo validation cohort (n ¼ 260), strengthen the value of HMR category is higher than those characterized by an unfavor- molecular-based prognostication in PMF by showing that the able karyotype (25–30% versus 5–7%), the methodology has ‘number’ of prognostically detrimental mutated genes adds shorter turnaround and the cost is even lower. As we focused on a further to the original molecular-based stratification. Results in restricted number of genes as those five included in the the learning cohort, showing that patients harboring two or more prognostically detrimental category, this approach is feasible in the

Leukemia (2014) 1804 – 1810 & 2014 Macmillan Publishers Limited Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1809

Figure 4. Leukemia-free survival of 260 patients with primary myelofibrosis from the Mayo cohort stratified by the number (zero, one, two or more) of prognostically detrimental mutations (ASXL1, EZH2, SRSF2 or IDH1/2 (a)). The impact of the number of prognostically detrimental mutations on leukemia-free survival of patients stratified by DIPSS-plus category is shown in b for the lower-(low plus intermediate-1) risk categories and in c for the higher-(intermediate-2 plus high) risk categories.

Table 2. Multivariable analysis assessing the interactions of the number of prognostically detrimental mutations, the IPSS and DIPSS-plus risk categories and CALR mutational status in the two cohort of PMF patients

Variable European (test) cohort Mayo Clinic (validation) cohort

HR 95% CI P-value HR 95% CI P-value

CALR mutation 0.51 0.29–0.90 0.02 0.45 0.30–0.69 0.002 Prognostically detrimental mutations One mutation 1.57 1.04–2.47 0.04 1.43 1.03–1.98 0.03 X2 Mutations 3.17 1.80–5.57 o0.0001 2.06 1.24–3.41 0.005 IPSS DIPSS-plus Intermediate-1 3.45 1.89–6.27 o0.0001 2.47 1.23–4.94 0.01 Intermediate-2 7.71 4.06–14.64 o0.0001 5.24 2.67–10.30 o0.0001 High 14.06 7.30–27.08 o0.0001 9.32 4.68–18.58 o0.0001 Abbreviations: CALR, calreticulin; CI, confidence interval; DIPSS, dynamic International Prognostic Scoring System; HR, hazard ratio; IPSS, International Prognostic Scoring System; PMF, primary myelofibrosis. clinical practice, unlike wider analysis based on cancer-associated Malattie Mieloproliferative’ (#1005); for a description of the AGIMM project, genes. We believe that recent findings concerning the prognostic see at (http://www.progettoagimm.it/). The study was partially supported by impact of mutations in PMF warrants a revision of the current Ministero della Universita` e Ricerca (MIUR; FIRB project #RBAP11CZLK and PRIN prognostication scores by incorporating mutational analysis for 2010NYKNS7 to AMV). It was also supported by the Bando ricerca finalizzata e giovani refined risk prediction and more targeted therapeutic approaches. ricercatori, Ministero della Salute GR-2011-02352109 to PG. JS and NCPC were supported by Leukemia and Lymphoma Research. Studies at the Mayo Clinic were supported by the Mayo Clinic Brian and Phyllis Harvey and Janet Yulman Charitable CONFLICT OF INTEREST Foundation for Myelofibrosis Tissue Bank and Clinical Database of Molecular and The authors declare no conflict of interest. Biological Abnormalities.

ACKNOWLEDGEMENTS REFERENCES This study was supported by a special grant from Associazione Italiana per la 1 Vannucchi AM, Guglielmelli P, Tefferi A. Advances in understanding and Ricerca sul Cancro-‘AIRC 5 per Mille’- to AGIMM, ‘AIRC-Gruppo Italiano management of myeloproliferative neoplasms. CA Cancer J Clin 2009; 59: 171–191.

& 2014 Macmillan Publishers Limited Leukemia (2014) 1804 – 1810 Number of mutations and prognosis in myelofibrosis P Guglielmelli et al 1810 2 Cervantes F, Dupriez B, Passamonti F, Vannucchi AM, Morra E, Reilly JT et al. substantially different clinical course and outcomes. Blood 2013; e-pub ahead of Improving survival trends in primary myelofibrosis: an International Study. J Clin print 23 December 2013; doi:10.1182/blood-2013-11-539098. Oncol 2012; 30: 2981–2987. 13 Tefferi A, Lasho TL, Finke CM, Knudson RA, Ketterling R, Hanson CH et al. 3 Cervantes F, Dupriez B, Pereira A, Passamonti F, Reilly JT, Morra E et al. New CALR vs JAK2 vs MPL mutated or triple-negative myelofibrosis: clinical, prognostic scoring system for primary myelofibrosis based on a study of the cytogenetic and molecular comparisons. Leukemia 2014; 28: 1472–1477. International Working Group for myelofibrosis research and treatment. Blood 14 Tefferi A, Guglielmelli P, Lasho TL, Rotunno G, Finke C, Mannarelli C et al. 2009; 113: 2895–2901. CALR and ASXL1 mutations-based molecular prognostication in primary 4 Passamonti F, Cervantes F, Vannucchi AM, Morra E, Rumi E, Pereira A et al. myelofibrosis: an international study of 570 patients. Leukemia 2014; 28: A dynamic prognostic model to predict survival in primary myelofibrosis: a study 1494–1500. by the IWG-MRT (International Working Group for Myeloproliferative Neoplasms 15 Vardiman JW, Thiele J, Arber DA, Brunning RD, Borowitz MJ, Porwit A et al. Research and Treatment). Blood 2010; 115: 1703–1708. The 2008 revision of the World Health Organization (WHO) classification of 5 Gangat N, Caramazza D, Vaidya R, George G, Begna K, Schwager S et al. myeloid neoplasms and acute leukemia: rationale and important changes. DIPSS Plus: a refined dynamic international prognostic scoring system for primary Blood 2009; 114: 937–951. myelofibrosis that incorporates prognostic information from karyotype, platelet 16 Gangat N, Caramazza D, Vaidya R, George G, Begna K, Schwager S et al. count, and transfusion status. J Clin Oncol 2011; 29: 392–397. DIPSS plus: a refined Dynamic International Prognostic Scoring System 6 Vannucchi AM, Lasho TL, Guglielmelli P, Biamonte F, Pardanani A, Pereira A et al. for primary myelofibrosis that incorporates prognostic information from Mutations and prognosis in primary myelofibrosis. Leukemia 2013; 27: 1861–1869. karyotype, platelet count, and transfusion status. JClinOncol2011; 29: 7 Guglielmelli P, Biamonte F, Rotunno G, Artusi V, Artuso L, Bernardis I et al. Impact of 392–397. mutational status on outcomes in myelofibrosis patients treated with ruxolitinib in 17 Caramazza D, Begna KH, Gangat N, Vaidya R, Siragusa S, Van Dyke DL et al. the COMFORT-II Study. Blood 2014; e-pub ahead of print 23 January 2014; Refined cytogenetic-risk categorization for overall and leukemia-free survival in doi:10.1182/blood-2013-11-536557. primary myelofibrosis: a single center study of 433 patients. Leukemia 2011; 25: 8 Guglielmelli P, Nangalia J, Green AR, Vannucchi AM. CALR mutations in 82–88. myeloproliferative neoplasms: Hidden behind the reticulum. Am J Hematol 2014; 18 Passamonti F, Cervantes F, Vannucchi AM, Morra E, Rumi E, Cazzola M et al. e-pub ahead of print 24 January 2014; doi:10.1002/ajh.23678. Dynamic International Prognostic Scoring System (DIPSS) predicts progression 9 Nangalia J, Massie CE, Baxter EJ, Nice FL, Gundem G, Wedge DC et al. to in primary myelofibrosis. Blood 2010; 116: Somatic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2. 2857–2858. N Engl J Med 2013; 369: 2391–2405. 19 Tefferi A, Pardanani A, Gangat N, Begna KH, Hanson CA, Van Dyke DL et al. 10 Klampfl T, Gisslinger H, Harutyunyan AS, Nivarthi H, Rumi E, Milosevic JD et al. Leukemia risk models in primary myelofibrosis: an International Working Group Somatic mutations of calreticulin in myeloproliferative neoplasms. N Engl J Med study. Leukemia 2012; 26: 1439–1441. 2013; 369: 2379–2390. 20 Lundberg P, Karow A, Nienhold R, Looser R, Hao-Shen H, Nissen I et al. 11 Rotunno G, Mannarelli C, Guglielmelli P, Pacilli A, Pancrazzi A, Pieri L et al. Impact Clonal evolution and clinical correlates of somatic mutations in myeloproliferative of calreticulin mutations on clinical and hematological phenotype and outcome neoplasms. Blood 2014; e-pub ahead of print 29 January 2014; doi:10.1182/ in essential thrombocythemia. Blood 2013; e-pub ahead of print 26 December blood-2013-11-537167. 2013; doi:10.1182/blood-2013-11-538983. 21 Papaemmanuil E, Gerstung M, Malcovati L, Tauro S, Gundem G, Van Loo P et al. 12 Rumi E, Pietra D, Ferretti V, Klampfl T, Harutyunyan AS, Milosevic JD et al. JAK2 or Clinical and biological implications of driver mutations in myelodysplastic CALR mutation status defines subtypes of essential thrombocythemia with syndromes. Blood 2013; 122: 3616–3627.

Leukemia (2014) 1804 – 1810 & 2014 Macmillan Publishers Limited