HMR Mutations Drive Poor Prognosis in Myelofibrosis Patients with Lower JAK2V617F Allele Burden but Not in Those with Higher Allele Burden: Results of a Multicenter Study

Abstract

JAK2V617F is the most common driver mutation in myelofibrosis (MF), followed by CALR and MPL alterations. Clinical variables included in the International Prognostic Scoring System (IPSS), Dynamic IPSS (DIPSS), and DIPSS plus, as well as JAK2V617F allele burden are all well-established prognostic factors influencing the survival of MF patients. In addition, ASXL1, EZH2, SRSF2, IDH and U2AF1 mutations are classified as high-molecular risk (HMR) mutations in this context because of their close association with disease progression and leukemic transformation. However, the relationship between JAK2V617F allele burden and HMR mutations remains unclear. To address this, we recruited 122 MF patients (discovery cohort) diagnosed at the National Taiwan University Hospital (Taipei, Taiwan) who had available mutation profile determined by next-generation sequencing. For external validation, 82 MF patients from the Christie Hospital (Manchester, U.K.) and 51 MF patients from the Cleveland Clinic (Ohio, U.S.A.) and Vilnius University Hospital Santaros Klinikos (Vilnius, Lithuania) were enrolled. The median age of the discovery cohort was 61 years. JAK2V617F was the most common driver mutation (64.8%), followed by CALR (18%) and MPL mutations (9%). ASXL1 mutation was the most common (37%) genetic alteration other than driver mutations, followed by TET2 (16%) and EZH2 mutations (12%). Overall, 55 patients (45%) had at least one HMR mutation ( ASXL1, EZH2, SRSF2, IDH and U2AF1). With a median follow-up of 28.2 months, the median overall survival (OS) of the discovery cohort was not reached (NR) and was well risk-stratified by current prognostication systems, including IPSS, DIPSS plus, MIPSS70, and GIPSS. Notably, patients with at least one HMR mutation had significantly shorter leukemia-free survival (LFS) and OS than those without (both p<0.001). Next, we divided the 79 JAK2-mutated patients into lower- and higher-allele burden groups ( JAK2low and JAK2high) using the median (75%), as the cutoff. Concurring with previous reports ( Tefferi et al, Leukemia 2008; and Guglielmelli et al, Blood 2009), JAK2low patients had significantly inferior OS than JAK2high patients (median, 53 months vs NR, p=0.037). Among all JAK2-mutated patients, those with HMR mutations had a significantly shorter OS than those without (50 months vs NR, p<0.001). When considering HMR mutations and JAK2V617F allele burden together, the presence of HMR alterations conferred worse LFS and OS in JAK2low but not in JAK2high patients (HMR vs without HMR among JAK2low patients, p<0.001 and p<0.001; and JAK2high, p=0.822 and p=0.699, respectively). By and large, patients with concurrent HMR mutations and lower JAK2V617F allele burden had a distinctively inferior OS than those with other constellations of genomic lesions ( p<0.001, Figure 1A). This group of patients also had the shortest survival among the total cohort including both JAK2-mutated and wild patients ( p<0.001). The discriminating power of integrating HMR/ JAK2 allele burden for LFS and OS remained valid in the UK cohort (HMR vs without HMR: JAK2low, p=0.013 and p=0.012; and JAK2high, p=0.337 and p=0.600, respectively) and the US/Lithuania cohort (HMR vs without HMR: JAK2low, p=0.016 and p=0.013; and JAK2high, p=0.643 and p=0.446, respectively). Co-occurrence tests in the discovery and validation cohorts revealed that there was no shared mutational co-occurrence pattern among the 3 cohorts, excluding the potential confounding influence of specific concurrent mutations. More strikingly, time-dependent ROC curve analyses suggested that integrating JAK2 allele burden with HMR mutations could be complementary to current risk stratification systems in predicting LFS and OS of MF patients (Figure 1B) across 3 cohorts. Finally, HMR/ JAK2low was identified as an independent adverse prognostic factor by multivariable analysis, irrespective of age, risk stratification systems, and other mutations. In summary, we observe that HMR mutations accompanied by a lower mutant JAK2V617F allele burden identified a distinct MF population with significantly reduced survival and high risk of leukemic transformation, with independent prognostic relevance extrapolated to and validated in two independent cohorts. Experimental studies are ongoing to ascertain the underlying mechanistic basis and resolve the clonal architecture of this distinct MF genomic grouping.