Lasho, Terra L.

Abstract Abstract 1756 Background: We hypothesized that surrogate markers of host immune response to clonal myeloproliferation predict survival in myeloid malignancies. Because of immediate applicability to current practice, we chose plasma immunoglobulin free light chain (FLC) concentration (B-cell activation marker) as the biomarker of interest. Patients and Methods: Two independent cohorts of patients with primary myelofibrosis (PMF) or myelodysplastic syndromes (MDS) were studied. Both groups were fully annotated for karyotype. Kappa (κ) and lambda (λ) FLC were measured by a quantitative nephelometric assay. Patients with monoclonal FLC were excluded. Results: Above upper normal-limit values for κ or λ FLC was documented in 33% of 240 study patients with PMF and 46% of 74 patients with MDS. Multivariable analysis revealed significant associations between increased FLC and elevated creatinine, as well as advanced age in PMF (p=0.0004) and hemoglobin <10 g/dL in MDS (p=0.005). Increased FLC predicted shortened survival in both PMF and MDS, independent of age, creatinine, and other conventional risk factors. Receiver-operating characteristic analysis-based cutoff levels for κ plus λ total FLC delineated risk groups with highly significant differences in overall survival (Figure); International Prognostic Scoring System-adjusted HR (95% CI) in PMF was 1.9 (1.3-2.7) and in MDS 6.3 (2.7-16.6). No correlations were seen with leukemia-free survival, karyotype or JAK2, MPL or IDH mutations. Conclusions: Elevated plasma FLC concentration predicts inferior survival in both PMF and MDS. Its lack of correlation with leukemia-free survival and tumor-specific genetic markers suggests a primarily host-driven biological phenomenon that might be applicable to other malignant and non-malignant conditions. Disclosures: No relevant conflicts of interest to declare.

Abstract Background : Refractory anemia with ring sideroblasts and thrombocytosis (RARS-T) is a provisional entity in the MDS/MPN overlap syndromes, with diagnostic features of RARS, along with a platelet count > 450 x 10(9)/L and large atypical megakaryocytes. Mutations involving SF3B1 (~80%), JAK2 (~60%), TET2 (~25%) and ASXL1 (~15%) have been reported. The frequency of thrombotic events in RARS-T is thought to be similar to that of essential thrombocytosis (ET) (3.6 vs 3.9/100 patient years; Hematologica 2012; 1036-41). However, unlike in ET, it remains unclear as to whether thrombotic events in RARS-T impact overall (OS) or thrombosis-free (TFS) survival. Methods : 82 patients with WHO-defined RARS-T were included in the study. All patients had bone marrow (BM) biopsies, iron stains for detection of BM ring sideroblasts (RS) and cytogenetics performed at diagnosis. Mutational analysis for JAK2, CALR, MPL, SF3B1, SRSF2 and ASXL1 was carried out on BM DNA obtained at the time of diagnosis. Details of type of thrombotic event and cardiovascular (CV) risk factors including; hypertension, diabetes, smoking and dyslipidemia were obtained by careful review of the medical record. OS was calculated from time of initial diagnosis to the time of last follow-up or death. TFS was calculated from the time of diagnosis of RARS-T to development of thrombosis in uncensored patients and date of last follow up or death in patients censored for thrombosis. Conventional statistics were utilized for all analyses. Results: Among the 82 study patients, 46 (56%) were males and median age was 72 years (range, 48-93). At a median follow-up of 26.5 months; 48 (59%) deaths and 2 leukemic transformations were documented. Median OS was 44 months. Karyotype was diploid in 61 (74%) and analyzed mutational frequencies were; SF3B1 67% (28/42), JAK2 V617F 40% (25/61), ASXL1 21% (9/41), SRSF2 6% (2/31), MPL 1% (2/41) and CALR 0, respectively. Cardiovascular risk factors were present in 52 (63%). a) Thrombotic events and their impact on thrombosis free survival: Eight (10%) patients had a thrombotic event prior to or at the time of diagnosis of RARS-T (venous-8, arterial-0). Nine (11%) patients developed subsequent thrombotic events (venous-7, arterial-2), with no fatalities. In univariate analysis, lower hemoglobin level (p=0.008), lower BM RS % (p=0.04), history of thrombosis (p=0.02), and absence of SF3B1 mutations (p=0.017) were associated with an inferior TFS. Age (p=0.07), JAK2 mutation status (p=0.56) and CV risk factors (p=0.95) did not have prognostic impact. Given the association between SF3B1 mutations and BM RS, in a multivariable model that included these two as covariates, only the absence of SF3B1 mutations retained prognostic significance (p=0.02). Absence of SF3B1 mutation remained significant when history of thrombosis or lower hemoglobin level were introduced into the multivariable model. b) Risk factors for overall survival: In univariate analysis, hemoglobin <10 gm/dl (p=0.007), presence of circulating blasts (p=0.002) and presence of CV risk factors (p=0.04) were associated with an inferior OS. Neither history of thrombosis (p=0.59) nor SF3B1 (p=0.12) mutations affected OS. In multivariable analysis, hemoglobin <10 gm/dl (p=0.03) [HR 1.9; 95% CI 1.1-3.5] and circulating blasts (p=0.002) [HR 1.3; 95% CI 1.1-1.6] remained significant. Conclusions: In contrast to what is observed in ET, thrombosis history does not appear to affect either overall or thrombosis-free survival in RARS-T. The current study suggests a lower thrombosis risk in SF3B1 -mutated patients with RARS-T, compared to their SF3B1 wild-type counterparts. Disclosures No relevant conflicts of interest to declare.

Abstract Introduction: CMML is a clinically heterogeneous myeloid neoplasm hallmarked by the coexistence of dysplastic and proliferative patho-clinical features, which can include cytopenias, constitutional symptoms, splenomegaly, and leukocytosis. However, according to the FAB classification schema, CMML may be differentiated in dysplastic and proliferative subgroups only by the presence of leukocytosis in the latter (WBC ≥13 x 109/L). We hypothesize that incorporation of other clinically discriminating features may yield a more informative CMML stratification system. To address this, we propose three distinct CMML categories and explore their clinical relevance leveraging our existing international CMML database (Padron E et al. Blood Cancer J. 20151). Method: 1622 WHO-defined CMML cases diagnosed between 1973 and 2014 were collected from eight large cancer centres that include up to 80 discrete data elements as previously described1. Cases were placed into three clinically distinct groups and the Pearson Chi-Square test and the Kruskal-Wallis test were applied respectively to compare categorical and continuous characteristics. The Kaplan-Meier (KM) method was used to estimate median OS and the log rank test was used to compare survival curves.Cox models whereapplied to obtain univariate and adjusted hazard ratios. Identification of optimal cut-off values for continuous variable was supported by graphical inspection of martingale residuals from the null Cox model. Statistical analyses were done in SPSS v23 and R v.3.3.0. Results: We propose three categories to delineate clinically distinct CMML subtypes: (1) Myelodysplastic (MD)-CMML: WBC≤10 x 109/L, PB-immature myeloid cells (IMC) = 0%, no splenomegaly (2) MD/MP-CMML: WBC 10-20 x 109/L or WBC ≤10 x 109/L but PB-IMC>0% and/or splenomegaly (3) Myeloproliferative (MP)-CMML: WBC >20 x 109/L. A recursive partitioning approach was used to identify the WBC cut points, with splenomegaly and IMC added to more accurately depict the MPN aspect of CMML. Numbers of patients included in the MD-, MD/MP-, and MP-CMML subcategories were 319 (19.7%), 789 (48.6%) and 514 (31.7%), respectively (Table 1). According to the FAB criteria, the MD/MP group included 521 (66%) MD- (WBC ≤13 x 109/L) and 268 (34%) MP-CMML (WBC>13 x 109/L) patients suggesting that the proposed classification clinically reclassifies FAB-defined CMML. Within the MD/MP group, 344 patients (21.2%) had a WBC <10 x 109/L but with IMC>0% and/or splenomegaly. Comparison of overall survival (OS) among proposed groups demonstrated that this classification schema was capable of discriminating the CMML natural history (Figure 1). In comparison to MD/MP-CMML, the unadjusted OS Hazard Ratio (HR) was 0.60 (95% CI 0.49-0.73) for MD-CMML and 1.57 (95% CI 1.36-1.81) for MP-CMML (p<0.001 for both). This difference was retained after adjusting for BM-Blasts (< vs ≥5%), IPSS or CPSS cytogenetics (High-risk vsInterm/low-risk), and elevated levels of LDH (HR was 0.73 with 95% CI 0.57-0.93, p=0.011 for MD-CMML, and 1.33 with 95% CI 1.12-1.58, p=0.001 for MP-CMML, respectively). We next explored whether each proposed group may have distinct variables that uniquely govern its prognosis. Peripheral blood blasts >5%, RBC- and PLT-transfusion-dependence predicted poor OS only in the MD and MD/MP-subcategories (p<0.005), but had no impact in the MP-CMML. Male gender and elevated LDH were only significantly associated with prognosis in the MD/MP group (p<0.05 and p<0.005, respectively) while absolute lymphocyte count >2.5 x 109/L and absolute monocyte count >10 x109/L were uniquely predictive for shorter survival in the MP-CMML subcategory (p<0.005). Notably, we additionally identified gene mutations that uniquely predicted prognosis in each proposed group (Table 2). Conclusions: We demonstrate that our proposed 3-group clinical classification schema is capable of independently stratifying prognosis. Further, our analysis identified clinical and genetic variables that uniquely govern each group's prognosis, suggesting independent clinical behaviour. Further investigations are warranted to validate these groups. Disclosures Jabbour: ARIAD: Consultancy, Research Funding; Pfizer: Consultancy, Research Funding; Novartis: Research Funding; BMS: Consultancy. Fenaux:Celgene, Janssen,Novartis, Astex, Teva: Honoraria, Research Funding. Kantarjian:Bristol-Myers Squibb: Research Funding; ARIAD: Research Funding; Amgen: Research Funding; Pfizer Inc: Research Funding; Delta-Fly Pharma: Research Funding; Novartis: Research Funding. Komrokji:Celgene: Membership on an entity's Board of Directors or advisory committees, Research Funding; Novartis: Consultancy, Speakers Bureau.

Abstract Abstract 4009 Trisomy 8 is the most common among sole cytogenetic abnormalities in both acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS). In the very first paper published on isocitrate dehydrogenase (IDH) mutations in AML, 13 of the 16 IDH1 mutations detected were associated with normal karyotype, 2 with trisomy 8 and one with trisomy 13. Trisomy 8 was also recurrent in patients with IDH1/IDH2-mutated post-MDS AML. In the current study we examined the prevalence and disease distribution of IDH1 and IDH2 mutations in a large (n=157) group of patients with hematologic malignancies and isolated trisomy 8. The Mayo Clinic cytogenetic database allowed identification of 157 patients with isolated trisomy 8. Archived bone marrow cell pellets were used to extract DNA for IDH1 and IDH2 mutation analysis. Eighteen IDH mutations were identified: 15 IDH2 (14 R140Q and one R140W) and 3 IDH1 (2 R132C and one R132G). Seventeen of the 18 IDH mutations occurred in myeloid malignancies whereas one (IDH2R140W) occurred in a patient with angioimmunoblastic lymphoma who was not previously exposed to chemotherapy or radiotherapy, and in whom the IDH2 mutation disappeared after effective lymphoma chemotherapy. Among the 17 IDH-mutated myeloid malignancies, disease-specific IDH1/IDH2 mutational frequencies were as follows: 27% (3/11) for post-MDS AML, 25% (3/12) for therapy-related MDS/AML, 15% (8/54) for de novo MDS, 13% (2/15) for de novo AML and 3% (1/32) for myeloproliferative neoplasm (MPN). In contrast, IDH mutational frequencies were significantly lower among 64 additional patients with AML or MDS without isolated trisomy 8: 7% in de novo AML (n=28), 0% in de novo MDS (n=21), 0% in post-MDS AML (n=11) and 0% in therapy-related MDS/AML (n=4). In the 54 patients with trisomy 8-associated de novo MDS, prognosis was similar between IDH mutated (n=8; median survival 14 months) and unmutated (n=46; median survival 16 months) cases (p=0.7). The majority of IDH-mutated cases with de novo MDS belonged to high risk MDS disease category. However, 3 of the 8 IDH-mutated patients with de novo MDS and 2 of the 3 with therapy-related MDS did not display excess bone marrow blasts. The current study suggests a possible association between IDH mutations and trisomy 8 in AML and MDS but not in MPN or MDS/MPN. The fact that the mere presence of trisomy 8 did not result in a more than expected incidence of IDH mutations in MPN or MDS/MPN makes it unlikely that such an association would be secondary to trisomy 8-associated genetic or biologic changes. Instead, it is possible that IDH mutations, which have been shown to cluster with high-risk disease in both MPN and MDS associated with 5q-, promote a selective advantage for the survival of the clone that harbors trisomy 8. Consistent with this contention, we were able to demonstrate in one of our patients with relapsed AML, the presence of IDHR132C both at initial AML diagnosis and time of relapse, whereas the trisomy 8 abnormality was seen only at the time of relapse. Regardless, the presence of molecular heterogeneity among patients with trisomy 8-associated AML or MDS might explain the controversial prognostic influence of the specific cytogenetic abnormality. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 4112 We report the characterization of BMS-911543, a potent and functionally selective small molecule inhibitor of the Janus kinase family (JAK) member, JAK2. BMS-911543 is a reversible inhibitor of JAK2 with a biochemical IC50 of 0.001 μ M and Ki of 0.48 nM. It has over 74- and 350-fold selectivity against the other JAK family members, JAK3 and JAK1, respectively. Further, examination of > 450 other kinases did not reveal significant inhibitory activity for this JAK2 inhibitor. Functionally, BMS-911543 displayed potent anti-proliferative and pharmacodynamic (PD) effects in mutated JAK2-expressing cell lines dependent upon JAK2-STAT signaling and had little activity in cell types dependent upon other pathways such as JAK1 and JAK3. BMS-911543 was evaluated in colony growth assays using primary progenitor cells isolated from patients with JAK2V617F-positive myeloproliferative disease (MPD) and resulted in an increased anti-proliferative response in MPD cells as compared with those from healthy volunteers. Similar to these in vitro observations, BMS-911543 was also highly active in in vivo models of JAK2-pSTAT signaling in multiple species (mouse, rat, dog and monkey) with sustained pathway suppression being observed after a single oral dose. Additionally, BMS-911543 was evaluated for effects in a JAK2V617F-expressing SET-2 xenograft model system and displayed a minimally effective dose of <2 mg/kg on pSTAT5 pathway suppression, which lasted up to 8 hours. BMS-911543 was also compared to pan-JAK inhibitors in a mouse model of immunosuppression. At low dose levels active in JAK2-dependent PD models, no effects were observed on antigen-induced IgG and IgM production whereas a pan-JAK family inhibitor showed pronounced effects at all dose levels tested. The mechanistic selectivity of BMS-911543 to pan-JAK family inhibitors was extended through comparative analysis of these inhibitors in whole genome gene expression profiling experiments performed in sensitive cell types. In this comparison, BMS-911543 modulated a distinct subset of transcriptional changes as compared to pan-JAK inhibitors, thereby defining a minimal set of transcriptional changes underlying the pharmacologic effects of JAK2 inhibition. Collectively these results define the mechanistic basis for a differential therapeutic index between selective JAK2 and pan-JAK family inhibition pre-clinically and suggest a therapeutic rationale for the further characterization of BMS-911543 in patients with MPD and in other disorders characterized by constitutively active JAK2 signaling. Disclosures: Purandare: Bristol-Myers Squibb: Employment. McDevitt:Bristol-Myers Squibb: Employment. Gottardis:Bristol-Myers Squibb: Employment. You:Bristol-Myers Squibb: Employment. Lombardo:Bristol_Myers Squibb: Employment. Penhallow:Bristol-Myers Squibb: Employment. Vuppugalla:Bristol-Myers Squibb: Employment. Trainor:Bristol-Myers Squibb: Employment. Lorenzi:Bristol-Myers Squibb: Employment.

Abstract Abstract 460 Background: CYT387 is a potent JAK-1/2 inhibitor that suppresses the in vitro growth of cells harboring JAK2V617F (Leukemia 2009;23:1441) and was effective in a murine model of myeloproliferative neoplasms (MPN) (Blood 2010;115:5232). Aims/Methods: To assess the safety, tolerability, and pharmacokinetic behavior of CYT387 in a Phase I dose-escalation study in patients with high- or intermediate-risk primary myelofibrosis (PMF) and post-PV or post-essential thrombocythemia (ET) myelofibrosis. The secondary objective was evaluation of preliminary efficacy. CYT387 was administered orally once daily in 28-day cycles. Once dose-limiting toxicity (DLT) was identified, a dose-confirmation cohort initiated treatment at the maximum tolerated dose (MTD) or lower. Result: Thirty six subjects (median age 64 years) have been enrolled (targeted accrual 120); 18 each in the dose escalation and dose confirmation phases. Twenty-three subjects had PMF, 8 post-PV MF, and 5 post-ET MF; 81% were JAK2V617F-positive. Median palpable spleen size was 18 cm and 20 subjects (56%) were red cell transfusion-dependent at study entry. Prior treatment included JAK inhibitors (9 and 1 subjects with INCB018424 and TG101348, respectively) and pomalidomide in 9 patients. The median treatment duration to date is 15 weeks (range 4–38). Dose-linear plasma exposures were observed up to 300 mg/day, with mean elimination T1/2 at steady state ranging from 3.9 to 5 hours across doses. Toxicity: All 36 subjects were evaluable for toxicity. At 400 mg/day, 2 of 6 subjects experienced DLT (1 each with asymptomatic grade 3 hyperlipasemia and grade 3 headache that were reversible upon holding drug); consequently, the MTD was declared at 300 mg/day. In the dose-confirmation phase, subjects were started at one of 2 dose levels that were deemed clinically effective: 150 mg/day (n=15) and 300 mg/day (n=3). Thirty-five subjects are currently on active therapy: 100 mg/day (n=2), 150 mg/day (n=20), 300 mg/day (n=10), and 400 mg/day (n=3). CYT387 was well tolerated. No grade 4 non-hematological toxicities were observed. Grade 3 non-hematologic adverse events were infrequent and included increased transaminases (n=2), increased alkaline phosphatase (n=2), headache/head pressure (n=2), increased lipase (n=1), and QTc prolongation (n=1). Thirteen (36%) subjects experienced “first-dose effect” characterized by grade 1 lightheadedness and hypotension; this phenomenon was self-limited and generally resolved within 3–4 hours with rare recurrence. Grade 3/4 thrombocytopenia was seen in 8 (22%) subjects, and treatment-emergent grade 3 anemia was seen in 1 subject only (3%). Treatment-emergent grade 3/4 neutropenia was not observed. Efficacy: Thirty two of 36 subjects who completed at least 1 cycle were eligible for response assessment: Anemia: Twenty two subjects were evaluable for anemia response (baseline Hgb <10 g/dL or red cell transfusion-dependent). Of these, 9 subjects (41%) achieved the threshold of response for “Clinical Improvement (CI)” per the International Working Group for MPN Research and Treatment (IWG-MRT) criteria, including 2 of 4 subjects who were previously treated with INCB018424. An additional 5 subjects experienced a >50% reduction in transfusion requirement, thus increasing the total anemia response rate to 63%. Splenomegaly: Thirty of 32 evaluable subjects had splenomegaly at baseline: median 20 cm; range 10–32 cm. Twenty nine subjects (97%) had some degree of spleen size reduction (median 9 cm; range 2–18 cm): 11 (37%) patients have achieved a minimum 50% decrease in palpable spleen size, thus qualifying them for a CI, including 3 of 8 subjects (38%) who were previously treated with INCB018424. Constitutional symptoms: The proportion of patients with the following symptoms at baseline, are: fatigue (97%), pruritus (22%), night sweats (38%), cough (13%), bone pain (28%), and fever (16%). At last follow up, improvement (complete resolution) in these symptoms was reported by 68% (16%), 86% (57%), 83% (75%), 75% (50%), 78% (44%), and 100% (100%), respectively. Conclusion: CYT387 is first-in-class of the JAK inhibitors with a significant response rate in anemia in myelofibrosis patients. The drug also shows substantial activity in reducing spleen size and controlling constitutional symptoms. CYT387 is well tolerated, and treatment responses have been seen both at (300 mg/day) and below (150 mg/day) the MTD. Disclosures: Pardanani: Cytopia Inc.: Research Funding. Off Label Use: Clinical trial data for CYT387 use in Myelofibrosis. Fida:YM Biosciences Australia: Employment, Equity Ownership. Burns:YM Biosciences Australia: Employment, Equity Ownership. Smith: YM Biosciences Australia: Employment, Equity Ownership.

Abstract Abstract 4105 Background: JAK2 mutations are found in approximately 99% of patients with polycythemia vera (PV) and 60% of those with essential thrombocythemia (ET) or primary myelofibrosis (PMF). It is currently assumed that other mutations relevant to JAK signaling contribute to the pathogenesis of JAK2 mutation-negative myeloproliferative neoplasms (MPN). The same might hold true for some cases of “idiopathic erythrocytosis” associated with subnormal serum erythropoietin level (sEpo). LNK is a plasma membrane-bound adaptor protein whose function includes inhibition of wild-type and mutant JAK2 phosphorylation. LNK exon 2 mutations were recently described in two patients with JAK2V617F-negative ET or PMF. Both mutations involved the LNK pleckstrin homology (PH) domain; one was a 5 base-pair deletion and missense mutation leading to a premature stop codon (603_607delGCGCT; 613C>G) and the other a missense mutation (622G>C; glutamic acid to glutamine substitution; E208Q). Objectives: i) To estimate the prevalence of LNK mutations in chronic and blast phase MPN ii) To determine if LNK mutations are mutually exclusive of other MPN-associated mutations iii) To test the hypothesis that LNK mutations might contribute to the pathogenesis of JAK2 mutation-negative PV or otherwise unexplained erythrocytosis Methods: LNK mutation analysis was performed on bone marrow or blood cells using modified primers for amplifying the SH2 and PH domains, using previously published methods (Oh et al. Blood First Edition Paper, prepublished online April 19, 2010; DOI 10.1182/blood-2010-02-270108). Results: LNK mutation studies were performed in 172 patients; paired chronic-blast phase samples were analyzed in 26 cases. Diagnoses in the 172 study patients were as follows: 78 chronic-phase MPN (39 JAK2V617F-positive PV, 25 PMF and 14 ET) enriched for TET2, IDH, JAK2V617F, or MPL mutations, 61 blast-phase MPN (41 blast-phase PMF, 11 blast-phase PV and 9 blast-phase ET), 25 JAK2V617F-negative PV and 8 “unexplained erythrocytosis” associated with subnormal sEpo and negative for Epo receptor (EPOR) mutations. Ten novel heterozygous LNK mutations, all but one affecting exon 2 in the PH domain, were identified: 6 missense mutations involving codons 215, 220, 223, 229 and 234, one nonsense mutation involving codon 208 (622G>T leading to glutamic acid to a stop codon substitution; E208X), one synonymous mutation involving codon 208, and 2 deletion mutations involving exon 2 (685-691_delGGCCCCG) or exon 5 (955_delA). LNK mutations were most frequent in blast-phase MPN, occurring in 6 (9.8%) of the 61 informative patients; chronic-phase sample analysis in 4 of these revealed the same mutation in only one case. Mutant LNK was detected in chronic-phase samples only in 2 additional patients with blast-phase MPN. Among the 8 LNK-mutated blast-phase MPN cases, 7 had blast-phase PMF. The clinical phenotype in the remaining 2 LNK-mutated patients was that of an isolated erythrocytosis without overt features of PV. JAK2V617F was documented in 3 and IDHR140Q in 1 LNK-mutated patients. Conclusions: LNK mutations i) target an exon 2 ‘hot spot' in the PH domain spanning residues E208-D234, ii) might be most prevalent in blast-phase PMF iii) are not mutually exclusive of JAK2 or IDH mutations and iv) might be part of the “missing link” in the pathogenesis of JAK2 mutation-negative “idiopathic” erythrocytosis. Disclosures: No relevant conflicts of interest to declare.

Abstract Background : Anemia is one of the most prominent symptoms in primary myelofibrosis (PMF) and is often associated with inferior quality of life and survival. Current drugs, including JAK inhibitors, are suboptimal in the treatment of PMF-associated anemia and better information on its pathogenesis is critical for the development of more effective drugs. In the current study of JAK2/CALR/MPL annotated patients with PMF, we examined its correlation with both "driver" and "non-driver mutations", as well as cytogenetic abnormalities, in order to gain better insight into its pathogenesis. Methods: Study patients were selected based on availability of "driver" mutation information. PMF diagnosis was according to World Health Organization criteria (Blood. 2009;114:937). Previously published methods were used for CALR, JAK2 and MPL mutation analyses and determination of CALR variants (Blood. 2014;124:2465). Considering their relatively high mutational frequency in PMF, subsets of patients were also screened for ASXL1, spliceosome component (SF3B1, U2AF1, SRSF2, ZRSR2) and TET2 mutations. Cytogenetic analysis and reporting was done according to the International System for Human Cytogenetic Nomenclature. Statistical analyses considered clinical and laboratory parameters obtained at time of first referral at the Mayo Clinic. Results : Analysis was conducted on 722 patients (median age 64 years; 64% males). DIPSS-plus risk distribution was 14% low, 17% intermediate-1, 37% intermediate-2 and 33% high. All patients were annotated for JAK2/CALR/MPL mutations as well as CALR variants; 477 harbored JAK2, 139 CALR and 41 MPL mutations; 65 patients were triple-negative. The 139 CALR -mutated patients included type 1/type 1-like (n =115) and type 2/type 2-like (n =24). Non-driver mutations screened included ASXL1 (n =480; mutated 38%), SRSF2 (n =474; mutated 14%), U2AF1 (n =457; mutated 16%), SF3B1 (n =328; mutated 8%), ZRSR2 (n =180; mutated 11%) and TET2 (n =180; mutated 18%). Karyotype was normal in 60%, favorable in 28% and unfavorable in 12%. Anemia was defined as being absent (normal sex-adjusted hemoglobin level; n =110; 15%), mild (hemoglobin level of ≥10 g/dL but below sex-adjusted normal value; n =263; 36%), moderate (hemoglobin level below 10 g/dL but not transfusion-dependent; n =108; 15%) and severe (transfusion-dependent anemia; n =241; 33%). Presence of at least mild anemia was associated with abnormal karyotype (p=0.006) with no difference between favorable and unfavorable abnormalities, U2AF1 (p=0.002), TET2 (p=0.02) and ASXL1 (p=0.04) mutations; other significant associations included male sex and older age. Presence of moderate to severe anemia was associated with U2AF1 (p<0.0001), SRSF2 (p=0.007) and driver mutations other than CALR type 1/type 1-like (p<0.0001). Presence of severe anemia was associated with U2AF1 (p<0.0001), SRSF2 (p=0.003) and non-CALR driver mutations (17% incidence in both types of CALR variants vs 51% in triple negative, 35% JAK2, 39% MPL mutated cases; p<0.0001). An association with older age but not gender was also noted for both moderate to severe and severe anemia (p<0.0001). During multivariable analysis, independent associations with moderate to severe anemia were confirmed for U2AF1 (p<0.0001), SRSF2 (p=0.007) and age (p=0.0001), but not driver mutation profile (p=0.30). A similar analysis for severe anemia also identified U2AF1, SRSF2 and age as being significantly relevant. Conclusions : The current study identifies older age and the spliceosomal mutations U2AF1 and SRSF2 as having strong and independent association with moderate to severe anemia in PMF. Targeting the spliceosome machinery or its mutant components offers a potential approach in the treatment of PMF-associated anemia. Disclosures Pardanani: Stemline: Research Funding.

Abstract Background The 2016 revised World Health Organization (WHO) classification system distinguishes prefibrotic (pre-PMF) from overtly fibrotic (overt PMF) primary myelofibrosis, based on the respective absence or presence of grade ≥2 bone marrow (BM) reticulin fibrosis (Blood. 2016;127:2391). The main objective of the current study was to determine the prognostic relevance of this distinction and compare clinical and genetic features between the two entities. Methods Study patients fulfilled the 2016 WHO criteria for the diagnosis of pre-PMF or overt PMF (Blood. 2016;127:2391). The degree of BM reticulin fibrosis was based on "real life" BM reports from Mayo Clinic hematopathologists and often in accordance with the European consensus scoring system (Haematologica 2005;90:1128). Targeted next generation sequencing was used to screen for prognostically-relevant mutations (Blood 2015 126:354). Statistical analyses considered clinical and laboratory parameters obtained at time of BM examination that was graded for reticulin fibrosis. Results Patient characteristics: Analysis was conducted on 467 patients (median age 64 years; 62% males), including 63 with pre-PMF and 404 with overt PMF. 302 (65%) patients harbored JAK2 mutations, 90 (19%) CALR, 24 (5%) MPL and 51 (11%) were "triple-negative"; among the 90 CALR-mutated cases, 74 (82%) were classified as "type 1/type 1-like". Dynamic international prognostic scoring system (DIPSS)-plus risk distribution (JCO 2011;29:392) was 30% high, 38% intermediate-2, 16% intermediate-1 and 15% low; 29% displayed red cell transfusion-dependency, 29% constitutional symptoms, 70% palpable splenomegaly and 37% abnormal karyotype, including 13% with unfavorable karyotype. 308 patients were screened for ASXL1 mutations with 39% mutated, 305 for SRSF2 mutations with 16% mutated, 291 for U2AF1 with 18% mutated, 264 for SF3B1 with 6% mutated, 240 for EZH2 with 5% mutated and 118 for IDH1/2 with 7% mutated. BM reticulin fibrosis was reported as grade '0' in 5 (1%) patients, grade '1' in 58 (12%), grade '2' in 169 (36%) and grade '3' in 235 (50%). After a median follow up of 3 years, 273 (59%) deaths and 42 (9%) leukemic transformations were documented. Clinical, cytogenetic and molecular comparisons between pre-PMF vs overt PMF: Significant differences included higher hemoglobin level (p<0.0001), higher platelet count (p<0.0001), higher incidence of thrombocytosis (p<0.0001), lower circulating blast percentage (p=0.0002), lower incidence of constitutional symptoms (p=0.003), lower incidence of palpable splenomegaly and lower DIPSS-plus risk distribution (p=0.0004), in pre-PMF, compared to overt PMF. There were significantly more triple-negative patients with pre-PMF (22% vs 9% for overt PMF; p=0.02) but otherwise no significant difference in the distribution of ASXL1, SRSF2, U2AF1, SF3B1, EZH2 or IDH1/2 mutations or incidence of unfavorable karyotype. Survival analysis: In univariate analysis, all 8 variables included in DIPSS-plus were significantly associated with shortened survival (p<0.001 in all instances). Significant risk factors in univariate analysis also included overt PMF vs pre-PMF (p=0.002; HR 2.0, 95% CI 1.3-3.1), absence of CALR type 1/type 1-like (p<0.0001) and presence of ASXL1 (p<0.0001) or SRSF2 (p<0.0001) mutations. The significant difference in survival between pre-PMF and overt PMF was independent of DIPSS-plus (p=0.03), driver mutational status (p=0.001), ASXL1 (p=0.008) and SRSF2 (p=0.02) mutations; no significant difference in leukemia-free survival was noted (p=0.25). The survival difference between pre-PMF and overt PMF was most evident in high/intermediate-2 DIPSS-plus risk groups (Figure). Of note, we found no difference in survival when patients with grade '0' were compared to those with grade '1' reticulin fibrosis (p=0.96) and when those with grade '2' were compared with grade '3' (p=0.08). Conclusion The current study demonstrates the risk-adjusted prognostic relevance of the WHO distinction between pre-PMF and overt PMF. The two clinic-pathologic entities were otherwise similar in their spectrum of prognostically-relevant mutations or cytogenetic abnormalities, although pre-PMF clustered with thrombocythemic and triple-negative phenotype and overt PMF with adverse clinical features. Figure Figure. Disclosures No relevant conflicts of interest to declare.

Abstract Background: In a previously published phase-1/2 study, momelotinib (a JAK 1/2 inhibitor) induced 53% anemia, 39% spleen and >50% constitutional symptoms response rates in patients with myelofibrosis (MF) (Pardanani A, et al. Leukemia. 2013;27:1322). The particular study included 100 patients from the Mayo Clinic who experienced 43% spleen and 44% anemia response rates (Pardanani et al. Leukemia. 2015;29:741); CALR and ASXL1 mutations independently predicted spleen but not anemia response. In the current study, we examined the impact of momelotinib therapy on DIPSS-plus- and molecular risk-adjusted survival. Methods: The current study represents sponsor-independent analysis of long-term survival among 100 consecutive Mayo Clinic patients who received momelotinib therapy as part of a larger (n =166) phase-1/2 study (NCT00935987). Safety and efficacy data from NCT00935987, as well as protocol design and drug dosing and schedule, have already been formally reported and not reiterated here (Pardanani A, et al. Leukemia. 2013;27:1322; Pardanani A, et al. Leukemia. 2015;29:741; Abdelrahman RA, et al. BJH. 2015;169:77). All statistical analyses considered clinical and laboratory parameters obtained at the time of entry to study. Survival analysis was considered from the date of study entry to date of death or last contact. Cox proportional hazard regression model was used for multivariable analysis of survival. Samples from time of study entry were used for mutation screening. Driver mutations were classified into two prognostically-relevant groups: CALR type 1/type 1-like (favorable) vs all others including CALR type 2/type 2-like, JAK2, MPL and triple-negative (unfavorable). Results: 100 patients with MF (median age 66 years; 58% males) were treated at the Mayo Clinic between 11/20/09 and 11/10/10; 64 patients had primary MF, 22 post-polycythemia vera MF and 14 post-essential thrombocythemia MF. Baseline data: DIPSS-plus risk distribution was 63% high, 36% intermediate-2 and 1% intermediate-1 (JCO 2011;29:392); for the purposes of the current analysis, the single patient with intermediate-1 risk disease was included in the intermediate-2 category. 54 (54%) patients displayed abnormal karyotype. 73 (73%) patients harbored JAK2, 16 (16%) CALR, 7 MPL and 4 "triple-negative". Among the 16 CALR -mutated cases, 13 were classified as "type 1/type 1-like". 94 patients were screened for ASXL1 mutations; 41 (44%) were mutated. Follow-up data was updated in July 2015, providing a minimum follow-up of over 4.5 years since the last patient registration on the protocol. At a median follow-up of 3.2 years, 67 (67%) deaths and 13 (13%) leukemic transformations were documented. Median survival from the time of study entry was 3.2 years and was affected by driver mutation prognostic groups (median survival "not reached" for "favorable" vs 3 years for "unfavorable"; p=0.01), ASXL1 mutational status (median survival 3.9 years in unmutated vs 2.5 years in mutated; p=0.02) and DIPSS-plus risk stratification (median survival 2.6 years in high-risk vs 4 years in intermediate-2 risk; p=0.07). In order to assess the impact of momelotinib on long-term outcome, we compared survival of the study cohort (n =100) with that of 442 consecutive Mayo Clinic patients with high or intermediate-2 risk primary MF (median age 67 years; 66% males). There was no significant difference between this control group of PMF patients versus the momelotinib patient cohort in distribution of age (p=0.56), sex (p=0.12), JAK2/CALR/MPL mutational status (p=0.41) or ASXL1 mutational status (p=0.9); however, the momelotinib cohort had a higher fraction of patients with DIPSS-plus high-risk disease (63% vs 47%; p=0.004). Overall survival was similar between the momelotinib cohort and the control group of PMF patients who were not treated with momelotinib (Figure 1; p=0.99); the difference in survival remained not significant during multivariable analysis that included risk stratification by DIPSS-plus (p=0.32), "favorable" vs "unfavorable" driver mutational status (p=0.78) or ASXL1 mutational status (Figure 2; p=0.61). Conclusions: In the current long-term study, we could not demonstrate a survival impact from momelotinib therapy in myelofibrosis. Although adjusted for clinical and molecular risk status, our study is retrospective and cannot be relied upon to discount marginal survival effect. Disclosures Al-Kali: Novartis: Research Funding. Pardanani:Stemline: Research Funding.