Sequential genotyping for phenotype-driver mutations in JAK2 (exon 14), CALR (exon 9), and MPL (exon 10) is recommended in patients with myeloproliferative neoplasms. Yet, atypical JAK2-and MPL-mutations were described in some triple-negative patients. Whether noncanonical and/or concomitant JAK2-and MPL-mutations exist in myelofibrosis (MF) regardless of phenotype-driver mutations is not yet elucidated. For this, next-generation sequencing (NGS) was performed using blood genomic DNA from 128 MF patients (primary MF, n = 93; post-ET-MF, n = 18; post-PV-MF, n = 17). While no atypical JAK2-or MPL-mutations were seen in 24 CALR-positive samples, two JAK2-mutations [c.3323A > G, p.N1108S; c.3188G > A, p.R1063H] were detected in two of the 21 (9.5%) triple-negative patients. Twelve of the 82 (14.6%) JAK2V617Fpositive cases had coexisting germline JAK2-mutations [JAK2R1063H, n = 6; JAK2R893T, n = 1; JAK2T525A, n = 1] or at least one somatic MPL-mutation [MPLY591D, n = 3; MPLW515 L, n = 2; MPLE335K, n = 1]. Overall, MPL-mutations always coexisted with JAK2V617F and/or other MPL-mutations. None of the JAK2V617F plus a second JAK2-mutation carried a TET2-mutation but all patients with JAK2V617F plus an MPL-mutation harbored a somatic TET2-mutation. Four genomic clusters could be identified in the JAK2V617F-positive cohort. Cluster-I (10%) (noncanonical JAK2 mutated (mut) + TET2 wildtype (wt) ) were younger and had less proliferative disease compared with cluster-IV (5%) (TET2 mut + MPL mut ). In conclusion, recurrent concomitant classical and/or noncanonical JAK2-and MPL-mutations could be detected by NGS in 15.7% of JAK2V617F-and MPLW515-positive MF patients with genotype-phenotype associations. Many of the germline and/or somatic mutations might act as "Significantly Mutated Genes" contributing to the pathogenesis and phenotypic heterogeneity. A cost-effective NGS-based approach might be an important step towards patient-tailored medicine. K E Y W O R D S germline mutations, JAK2 mutation, MPL mutation, myelofibrosis, noncanonical mutations, somatic mutations

Allogeneic HCT is curative in myelofibrosis (MF) with a survival (OS) advantage compared to non-JAK-inhibitor treatments in patients (pts) with Int-2/high-risk DIPSS (Kroeger et al, Blood 2015). On the other hand, data demonstrate a OS benefit of the JAK1/JAK2 inhibitor ruxolitinib (RUX) in intermediate (Int)-2/high-risk IPSS (Cervantes et al, Blood 2013; Verstovsek et al, Haematologica 2015). Yet, comparative data to the impact of RUX and HCT on OS are lacking. This was pursued retrospectively at the University of Leipzig in light of clinical-, cytogenetic-, phenotype-driver-, and somatic mutations-based prognostic variables. Patient and Methods: RUX became available in November, 2009. MF pts (n=137; median age 58y) seen afterwards were included with the exception of MPL + pts (small number). Characteristics are shown in table 1. RUX was given for spleenomegaly and/or constitutional symptoms [n=76 (55%)]. Irrespective of RUX, a donor search was initiated in int-2 and high-risk IPSS pts < 70 y and HCT performed if a suitable donor identified. Based on this biologic randomization, 50 (37%) received HCT, 58 (42%) continued with RUX (median exposure 19 months), and 29 (21%) pts had other options. The RUX- and HCT-groups were balanced in terms of IPSS, platelets, cytogenetic-, and epigenetic molecular-risks. Treatment was initiated within the first year (median 10 months), thus probability of OS 5 years after diagnosis was calculated (SPSS20 software). Cytogenetic-risk was conducted according to Caramazza et al. Leukemia 2011. Somatic mutations detected by NGS were used to calculate molecular-risk (Vannucchi et al. Leukemia 2013). Results: Median OS time was 6.8 y. The RUX-group was older (median age 64 y) with profound spleenomegaly and a higher JAK2 V617F allele burden compared to the HCT group (median age 55 y) (p<0.0005). The 56% OS at 5 years in the RUX group was comparable to that after HCT (60%; p=0.5). In the transplant setting, prior RUX-treatment (n=21; OS 65%) had no negative impact compared to up-front HCT (n=29; OS 55%) (p=0.9). Whilst IPSS low-/int-1-risk retained its survival impact (p=0.04), OS at 5 years was comparable for int-2 and high-risk. OS in the HCT and RUX groups for int-2 was 59% and 62% respectively (p=0.4) and for high-risk IPSS 54% and 50% respectively (p=0.2). Advanced age was associated with an inferior OS (p=0.002) and correlated with higher IPSS, transfusion-dependency and thrombocytopenia. In the HCT group, OS in pts <50 y was 91% vs. 49% in pts >50 y (p=0.001). In the RUX group, age <60 and not <50 y was OS-relevant [79% in pts <60y vs. 43% in those >60 y (p=0.02)]. No prognostic implication of anemia, WBC, peripheral blasts, and constitutional symptoms could be detected with both treatments. OS in JAK2 + pts was similar to CALR + pts in both therapy cohorts (p=0.3). Triple negativity retained its detrimental influence (p<0.0001) even after HCT, with frequent AML-typical mutations (transcription and/or signaling genes) and a leukemic transformation (LT) rate of 35%. Although spliceosome mutations were associated with anemia (p=0.004)/transfusion-dependency (p=0.006) and mutations in chromatin modifiers (ASXL1, EZH2) with splenomegaly(p=0.02),high-molecular epigenetic risk had no impact on OS or LT in the treatment groups. Unfavourable cytogenetics were associated with LT (p=0.04) but not OS in the two groups. Of the 24 pts with a LT, 15 (62%) received no RUX and no HCT. Conclusions: Our data imply that, besides allogeneic HCT, a prolonged JAK1/JAK2 inhibition could, at least partly, attenuate most of the prognostic detrimental parameters. Pre-treatment with ruxolitinib prior to HCT does not seem to have a negative impact on survival. It is the first time where outcome with a non-transplant procedure is shown to be comparable to that after HCT in MF, even in high-risk disease. This needs to be verified in prospective randomized trials to define the role of allogeneic HCT in the era of JAK1/JAK2 inhibition. Table 1. Patients characteristics ( n = 137 ) Variable Impact on OS Impact on LT Median age (years) 58 0.002 0.6 JAK2+/CALR+/Double negative (%) 63 / 20 / 17 <0.0005 0.04 Median number of MF-related mutations (range) 1 (0-5) 0.4 0.07 High-molecular risk profile (%)(ASXL1, EZH2, SRSF2, IDH1/2) 33.6 0.8 0.7 Mutations in transcription/signaling genes (%) 28 0.3 0.008 Unfavorable cytogenetics (%) 30 0.1 0.04 IPSS (Int-2 / high-risk) (%) 34 / 41 0.04 0.2 Disclosures Roskos: Novartis: Honoraria. Al-Ali:Celgene: Honoraria, Research Funding; Novartis: Consultancy, Honoraria, Research Funding.