ALBERT

Description: Background: The mutational status of immunoglobulin VH genes (IgVH) is an important prognostic marker in chronic lymphocytic leukemia (CLL), but its determination remains unadapted to routine practice. Several reports have showed that ZAP-70, whose expression can be detected by flow cytometry, can be considered as a reliable surrogate marker of IgVH mutational status. We recently conducted a gene expression profiling study of 18 cases of CLL, which pointed out 2 other genes which might also discriminate CLL groups: the lipoprotein lipase (LPL) and the disintegrin and metalloprotease 29 (ADAM29) genes which were overexpressed preferentially in unmutated (UM) and mutated (MT) cases respectively. Methods: We analyzed frozen cells obtained at diagnosis for 127 CLL patients (87 Binet stage A, 40 stage B or C). LPL, ADAM29 and house keeping GAPDH gene expression were measured by real time quantitative polymerase chain reaction in 111 cases, and ZAP-70 protein by flow cytometry in 107 cases, both analyses being performed in 93 cases. LPL and ADAM29 were also quantified in peripheral blood mononuclear cells (PBMC, n=4) and purified B cells (n=3) of healthy individuals. We correlated the results with the previously determined IgVH mutational status and clinical outcome. Results: With a cut-off value determined to be 1 for the LPL/GAPDH copy number ratio, LPL expression had a positive predictive value (PPV) of 68% for UM cases and a negative predictive value (NPV) of 85% for MT patients. Alternatively ADAM29 expression (ADAM29/GAPDH 〉 3) had a PPV of 77% for MT cases and of 86% for UM cases. Combining LPL and ADAM29 RNA quantifications by a simple 1:1 ratio (L/A ratio; threshold=1) provided slightly better results than those obtained with ZAP-70 (positivity threshold = 20%) for PPV of UM status (90% vs 76%) and similar results for NPV of MT status (90% vs 91%). Simultaneaous usage of L/A ratio and ZAP-70 expression allowed an almost perfect (73/74) assessment of the IgVH status in 80% (74/93) of patients with concordant results (L/A+, ZAP-70+ or L/A-, ZAP-70-). Serial measurements of L/A ratio and ZAP-70 expression showed that both parameters can change over time (median follow-up 38 months, range 6–159) in a fraction of patients (5/25 tested). ADAM29 was not detected while LPL was present at very low levels or absent in PMBC or purified B cells from healthy donors. The IgVH mutational status, ZAP-70 and L/A ratio were predictive of event-free survival for the whole cohort and for stage A patients. However only the L/A ratio was significantly associated with a shorter survival (p=0.03) for stage B/C patients. Conclusions: Combination of LPL and ADAM29 mRNA quantification provides a good surrogate marker of the IgVH mutational status in CLL patients. Used in association with ZAP-70, it represents an reliable alternative to the sequencing of IgVH genes. In addition, it might constitute a more powerful prognostic marker than IgVH mutational status or ZAP-70.

Description: Abstract 1594 Background: Follicular Lymphoma (FL) is the most frequent low-grade NHL. Clinical course is heterogeneous, some patients (pts) presenting an indolent clinical course with overall survival (OS) over 15 y and others developing a more aggressive disease with shorter survival. The Follicular Lymphoma International Prognostic Index (FLIPI) or FLIPI-2 are commonly used to predict pts outcome, but fail to identify pts with a really poor prognosis. At diagnosis, few FL pts present with detectable leukemic phase (FL-LP) and this characteristic has been seldom described. Aim: The aim of the study was to describe the clinical features and outcome of FL-LP pts. Results: Among 499 pts diagnosed with FL according to the WHO criteria in the Centre Hospitalier Lyon Sud (transformation and grade 3b excluded) and treated between 01/1992 and 01/2012, 37 (7.4%) had characteristic FL-LP detected by cytological blood smears analysis with confirmation by flow cytometry (kappa/lambda clonality). Median age was 58 y and FLIPI score repartition was 4 pts in low, 16 in intermediate, and 17 in high risk groups. Splenomegaly was present in 23 pts, high tumour burden (GELF criteria) in 11, B symptoms in 8, and ECOG PS〉1 in 3. Seven pts had anaemia, 17 platelets UNL, and 11 LDH 〉UNL. The circulating lymphoma cells expressed the CD10 in 29/37 cases and surface Ig expression was detected in 31/35 cases, mainly IgM or IgG isotype. The median count of circulating lymphoma cells was 1.95.109/L, ranging from 0.6 to 129.109/L, 15 pts having count 〉4.109/L and 6 pts 〉10.109/L. Cytogenetic data were available for 21 pts, 20 carried the t(14;18), or its variant t(18;22), 16 of them having complex karyotype. Two pts were on watchful waiting for 24 and 82 m and 35 received a chemotherapy regimen at diagnosis including rituximab in 27 cases. Overall response rate was 83% (29/35) with 23 CR/CRu. Median progression-free survival (PFS) was 29 m. PFS and OS estimates were 37% and 86% at 5 y and 31% and 68% at 10 y, respectively. Splenomegaly (P=.035), high tumour burden (P=.017), lymphoma cells count greater than 4.109/L (P=.028), β2-m〉UNL (P=.036), and thrombocytopenia (4.109/L (HR 5.92; P=.000916) as independent prognostic factors. After progression, 12 pts received high-dose therapy (HDT) with HSCT as salvage with a long second PFS (68% at 10 y) compared to 8.3 m median PFS in first line. Pts not receiving HDT at salvage had a median second PFS of 27 m. To further evaluate the impact of FL-LP on pts outcome, a 1:3 matched analysis was performed. The FL-LP 37 pts were successfully matched with 111 newly diagnosed FL without FL-LP according to FLIPI score, age, treatment type (abstention vs chemotherapy, with or without rituximab) and treatment period (before or after 2000). In these 111 matched pts, 5- and 10-y OS was 97% and 91%, respectively. Considering all 148 pts, high FLIPI score, presence of FL-LP, and β2-m〉UNL were all significantly associated with a worse PFS. In a Cox regression model for PFS (120/148 pts with complete data), high FLIPI score (P=.0034; HR=2) and presence of FL-LP (P=.0085; HR=2.2) remained independently associated with shorter PFS. High FLIPI score and presence of FL-LP were also associated with a shorter OS. Interestingly pts with less than 4.109/L of circulating lymphoma cells had a similar PFS than those without. When circulating lymphoma cells 〉4.109/L as variable (see abstract figure) instead of FL-LP were tested in the Cox regression model for PFS including β2-m〉UNL and FLIPI score as variables, the most significant predictor for a shorter PFS was circulating lymphoma cells 〉4.109/L (P=.0004; HR=3.56) as compared to FLIPI score (P=.051; HR=1.6) and β2-m (P=.09; HR=1.52). Conclusion: The presence of circulating lymphoma cells in FL is a rare event and is associated with shorter PFS independently of FLIPI score and β2-m level. A validation of these findings on pts from the PRIMA study is on progress. However, this population is not homogenous and pts with circulating lymphoma cells 〉4.109/L have a poorer outcome. Although ∼1/3 of the pts experience long term PFS, these pts should be monitored carefully during and after first line treatment to consider HSCT as a therapeutic option to achieve a sustained response. Disclosures: No relevant conflicts of interest to declare.

Description: Introduction: Splenectomy was historically regarded as the gold standard for treatment in chronic adult immune thrombocytopenic purpura (ITP). However, the recent emergence of new drugs has deeply modified ITP management and splenectomy is no longer viewed as an unavoidable step in adult chronic ITP in many countries. The estimation of the risk over benefit of this potential curative treatment remains challenging both for patients and physicians. A retrospective Italian study focused on long-term outcome of patients splenectomized for ITP gave reassuring data concerning safety. A recent study from a large cohort of American veterans showed an increased risk of death due to septicemia, pulmonary embolism, coronary artery disease and cancer more than 10 years after splenectomy. We reported here the results of the first single center case-control study evaluating the long-term incidence of splenectomy complications with a minimum follow-up of 10 years. Methods: We retrospectively selected in a clinical computer database all primary ITP patients splenectomized more than 10 years ago in our unit. We matched 1 by 1 to non-splenectomized ITP patients based on date and age at ITP diagnosis and sex criteria. Clinical data were then completed from medical charts. All patients were interviewed by phone and a standardized questionnaire was used. Medical records from general practitioner or from Medical care center have been systematically obtained if necessary, especially for deceased patients. Comparison between groups were made using Fisher’s test for qualitative variables, Kaplan-Meier method to estimate incidence and Rank test for comparison of cumulative incidence, with p

Description: The use of tyrosine kinase inhibitor (TKI) therapies has dramatically changed the prognosis of chronic myeloid leukemia (CML) and modified the natural history of the disease. However, the generation of ABL-kinase domain mutations due to the genetic instability of leukemic cells continues to be very significant challenge, especially in advanced phases of the disease. Amongst these mutations, T315I is the most problematic as it reduces the binding of Imatinib, Dasatinib or Nilotinib to their target, leading to a total resistance to all three TKIs. Although the mechanism of the appearance of this mutation is not completely known, it is possible that it induces a signalling pathway not entirely similar to that induced by classical wild-type BCR-ABL. From the mechanistic point of view, the mechanism of resistance of T315I-mutated BCR-ABL to Imatinib is related to the substitution of Isoleucine to Threonine, which normally forms a hydrogen bond with Imatinib, this bond being absolutely required for binding of the drugs to the catalytic site. T315I mutation can be identified by direct sequencing, D-HPLC, allele specific PCR or double gradient gel electrophoresis of the PCR products amplified from the BCR-ABL kinase domain. However these techniques cannot identify directly the leukemic cells expressing T315I-mutated BCR-ABL which are known to co-exist with native BCR-ABL-expressing cells in bone marrow. Infrared microspectroscopy allows the identification of metabolic features of mammalian cells, based on their protein, lipid, amid and nucleic acid contents, generating a spectral signature. In this work we asked whether infrared microspectroscopy can be used to identify metabolic changes occurring in single cells bearing BCR-ABL T315I mutation. For this purpose we have used human (UT7) and murine embryonic stem cells (GS2) expressing either native (N) or T315I-mutated BCR-ABL. In the human UT7 cells expressing N-BCR-ABL, we were able to clearly distinguish at the single cell level, cells expressing BCR-ABL from wild type cells using principal component analysis. However, the presence of T315I mutation induced a clearly different signature, allowing a highly significant separation by the use of spectral signature. To confirm the specificity of this signature, we have used UT7 cells engineered to express either native or T315I-mutated BCR-ABL under the control of Doxycycline (DOX) -sensitive promoters. In this TET-OFF system, the addition of doxycycline to the culture medium inhibits BCR-ABL expression as monitored by Western blots, in approximately 6 days. Using this system, at day 0, UT7 cells expressing N-BCR-ABL and T315I mutated BCR-ABL were clearly distinguishable from parental UT7 cells. Upon inhibition of BCR-ABL expression by addition of Doxycycline, the spectral signature of N-BCR-ABL and T315I-mutated BCR-ABL cells became similar at day 2 and indistinguishable at day 4, but still distinguishable from wild type UT7 cells expressing DOX-inducible GFP. To confirm these results in a different cell context we have used a murine embryonic stem (ES) cell line (GS2) which was transduced with either N-BCR-ABL or T315I-mutated BCR-ABL-expressing lentiviral vectors. BCR-ABL-expressing individual clones grown were analyzed to analyze the spectral signature. In this cell line also, infrared microspectroscopy was able to distinguish ES cells expressing BCR-ABL T315I as compared to N-BCR-ABL. Thus, these results suggest that T315I mutation clearly induces metabolic changes different from N-BCR-ABL in leukemic cells, rendering them identifiable by the analysis of nucleic acid, lipid, amid and sugar contents. This new methodology can now be applied to the identification of primary CML leukemic cells harboring T315I at the single cell level and could be of interest for rapid identification of leukemic cells in a single step. This technique can also be used for rapid screening of novel compounds active against T315I mutation using microfluidic technologies leading to novel drug discoveries. Disclosures Turhan: Bristol Myers Squibb, Novartis: Honoraria, Research Funding.

Description: Allogeneic stem cell transplantation (AlloSCT) is a curative option for acute myeloid leukemia (AML) patients. In first complete remission (CR1), young patients (〈 60 years) with intermediate or unfavorable ELN disease risk are usually considered for AlloSCT. In contrast, because of the expected higher toxicity in older patients, the benefit of this treatment remains a matter of debate after 60 years, especially in the intermediate ELN risk group. In this multicenter analysis from the French Innovative Leukemia Organization (FILO), we investigated whether AlloSCT was beneficial for AML patients over 60 years old in CR1. Inclusion criteria were: patients between 60 and 70 years of age diagnosed with AML from 2007 to 2017; CR1 after intensive chemotherapy; ELN-2010 intermediate or unfavorable risk group. AlloSCT was evaluated as a time dependent variable in survival calculations and in a multivariate Cox model adjusted on age, ELN group, transplantation period and stratified by transplantation center. We also used a multistate model as follow: initial state for all patients was "No Allo - CR" with time 0 at the time of CR. From initial state, patients can transit to "Allo-CR", or move through 2 absorbing states (non-relapse death (No Allo-NRM) or relapse (No Allo-Relapse). Similarly, once transplanted (i.e. in the state "Allo-CR"), patients can move to "Allo-Relapse" or "Allo - NRM" when such events occurred. The model allows the dynamic prediction of probability for a patient to be in a specific state considering specific initial state and time. We analyzed 521 consecutive patients in 6 centers who matched inclusion criteria. Median age was 65 years (range: 60-70). ELN-2010 risk was intermediate and unfavorable in 376 (72%) and 145 (28%) patients, respectively. While all patients had a theoretical indication for AlloSCT in CR1, 199 (38%) were actually transplanted (129 (34%) and 70 (48%) in the intermediate and unfavorable risk group, respectively). In the whole cohort, AlloSCT as time-dependent variable significantly improved relapse-free survival ([RFS] at 5 years, No AlloSCT vs. AlloSCT: 14% vs. 47% p 〈 0.001) and overall survival ([OS] at 5 years, No AlloSCT vs. AlloSCT: 24% vs. 51% p 〈 0.001). In subgroup analysis based on ELN-2010 risk classification, AlloSCT significantly improved outcome of both ELN intermediate (No AlloSCT vs. AlloSCT: 5-y RFS: 16% vs. 50% p 〈 0.001; 5-y OS: 26% vs. 54% p 〈 0.001) and unfavorable (No AlloSCT vs. AlloSCT: 5-y RFS: 7% vs. 44% p 〈 0.001; 5-y OS: 17% vs. 46% p 〈 0.001) risk group patients (Figure A and B). By multivariate time-dependent Cox model, AlloSCT significantly decreased the risk of relapse (HR [95%CI]: 0.29 [0.20-0.41] p 〈 0.001) and increased the risk of NRM (HR [95%CI]: 2.61 [1.38-4.94] p = 0.003). This led to a significant advantage for AlloSCT in both RFS (HR [95%CI]: 0.48 [0.36-0.64] p 〈 0.001) and OS (HR [95%CI]: 0.60 [0.44-0.81] p = 0.001). This benefit was observed in both intermediate and unfavorable ELN-2010 risk groups, with lower risk of relapse (intermediate: HR [95%CI]: 0.30 [0.19-0.46] p 〈 0.001; unfavorable: HR [95%CI]: 0.37 [0.19-0.72] p = 0.004) and better OS (intermediate: HR [95%CI]: 0.67 [0.47-0.96] p = 0.028; unfavorable: HR [95%CI]: 0.51 [0.29-0.91] p = 0.022). Multistate model showed that 5 years after CR1, few patients were still alive in CR without AlloSCT (i.e. in the initial "No Allo-CR" state), whatever the ELN-2010 risk group (intermediate: 9%; unfavorable: 1% Figure C and D). Among patients who were transplanted, the probability for transiting to Allo-NRM state within 5 years post CR1 was 17% and 24% in intermediate and unfavorable ELN-2010 groups, respectively. Corresponding values for patients without AlloSCT transiting to No Allo-NRM state were 11% and 12%. Moreover, considering a landmark at 6 months after CR1, the multistate model showed that patients who received AlloSCT had lower probability of relapse at 5 years (22% and 33% in intermediate and unfavorable ELN-2010 groups, respectively) compared to those who did not (68% and 78% in intermediate and unfavorable groups, respectively). AlloSCT for CR1 AML patients over 60 years of age is routinely feasible and significantly improves outcome in both intermediate and unfavorable ELN-2010 risk groups. Less than 10% of patients are long term disease free survival without AlloSCT, even in intermediate risk group, supporting that AlloSCT remains the first curative option for these patients. Figure. Figure. Disclosures No relevant conflicts of interest to declare.

Description: Background: In B-cells somatic hypermutation (SHM) and class switch recombination (CSR) depend of AID expression, a typical event in germinal centers upon CD40 ligand (CD40L) stimulation. We recently showed, that in contrast to normal B-cells, AID is constitutively expressed in a majority of unmutated CLL patients (Oppezzo et al, Blood, 2003). Pax-5 gene, encodes for the B-specific activator protein (BSAP) an essential protein in B-cell development. BSAP has been shown to up-regulate AID gene expression, through binding to its promoter region, whereas inhibitor of differentiation Id-2 down-regulates it. To study the molecular mechanisms underlying AID regulation, we analyzed expression of BSAP and Id-2 transcripts in normal and CLL B-cells and correlated these results to expression of AID and CSR process. Results: B-cells from 6 healthy donors and 40 CLL patients were analyzed for Pax-5, Id-2, AID and CSR. Results show that CLL and normal B-cells not expressing AID exhibit diminished expression of BSAP and express a spliced variant of the Pax-5 gene (Pax-5 ΔEx8) displaying a deletion in the C-terminal-active domain. Upon CD40-L and IL-4 stimulation expression of AID transcripts and disappearance of the Pax-5 ΔEx8 transcripts was observed. Translation of complete Pax-5 and Pax-5/ΔEx8 isoforms and their binding to AID gene promoter have been demonstrated by Western Blot and EMSA assays, respectively. In contrast, CLL B-cells with constitutive AID expression and active CSR, constantly show increased levels of BSAP and absence of Pax-5ΔEx8 isoform. High expression of AID and BSAP is also correlated with a decrease of Id-2 transcripts. Conclusion: These results suggest that the presence of BSAP isoform deleted in exon 8 may interfere with the binding of the complete BSAP protein to AID promoter and down-regulate the expression of Pax-5a. Thus, we show for the first time a subtle regulation mechanism controlling the expression of AID protein through the differential splicing of a transcription factor as BSAP.

Description: The female donor/male recipient combination (FM) increases the risks of graft-versus-host disease (GVHD) and non-relapse mortality (NRM) after allogeneic stem cell transplantation (allo-SCT), with a possible deleterious impact on overall survival (OS) [Gahrton G. Best Pract Res Clin Haematol. 2007 Jun;20(2):219-29]. Most patients in these studies received T-cell replete transplants. As a consequence, and because antithymocyte globulin (ATG) reduces the risk of GVHD, the impact of FM in the specific setting of allo-SCT with ATG remains poorly defined. To explore the impact of FM in the ATG setting, we undertook a retrospective analysis of male adult patients transplanted with rabbit ATG at our center between 01/01/2000 and 12/31/2012. Overall survival and disease-free survival (DFS) were calculated using the Kaplan-Meier estimate, with comparisons by the log-rank test. Cumulative incidences (CI) were used for NRM, relapse (REL), and GVHD in a competing risks setting, NRM being a competing event for REL, and death for GVHD. The Gray test was used to compare CI curves. For multivariate analysis, the variables with p value 〈 0.1 were entered into a Fine-Gray model where the least significant variables were excluded in sequential fashion until all remaining factors were significant at the p=0.05 level. The following variables were considered: recipient age (≥ vs 〈 56 years, median age), female donor (FD) vs male donor (MD), matched related (MRD) vs matched unrelated (MUD) donor, peripheral blood (PB) vs bone marrow (BM) graft, GVHD prophylaxis with cyclosporine (CsA)+metho vs others, myeloablative (MAC) vs reduced-intensity conditioning (RIC) regimen, CMV-seropositive vs seronegative recipient, early (CR1 or PR1 or chronic phase or untreated) vs advanced disease. Steroid-refractory (SR) acute (a) GVHD was defined as aGVHD progressing after 3 days of treatment, or unchanged after 7 days, or in incomplete response after 14 days. MUD were matched at the allele level for HLA-A, B, C, DRB1, DQB1. Two hundred and twelve male patients were identified and included in the present study. The median age was 56 years (18-67). Diseases were AML (n=61), ALL (n=16), NHL (n=36), Hodgkin's disease (n=10), myeloma (n=37), MDS (n=26), aplastic anemia (n=7), CLL (n=11), CML (n=1), and MPS (n=7). Status at transplant were CR1 or PR1 or chronic phase (n=74), 〉 CR1 or PR1 (n=83), refractory (n=34), or untreated (n=21). Conditioning regimens were RIC (n=195) or MAC (n=17). Donors were MRD (n=110) or MUD (n=102). Eighty-eight patients were transplanted with a FD. Source of stem cells were PB (n=193), BM (n=18), missing data (n=1). Prophylaxis of GVHD consisted of CsA+metho for 84 patients. The median follow-up was 42 months (5-149). In univariate analysis, the 3-year OS and DFS in FD vs MD groups were 48% ± 6% vs 56% ± 5% (p=0.3) and 42% ± 5% vs 50% ± 5% (p=0.4), respectively. The 3-year NRM and REL in the same groups were 25% ± 5% vs 19.6% ± 4% (p=0.4) and 30% ± 5% vs 31% ± 4% (p=0.9), respectively. The CI of aGVHD II-IV, aGVHD III-IV, SR aGVHD, and extensive chronic GVHD in FD vs MD groups were 37.5% ± 5% vs 33% ± 4% (p=0.5), 22.7% ± 4% vs 17% ± 3% (p=0.3), 17.2% ± 4% vs 8.3% ± 3% (p=0.049), and 20% ± 4% vs 22.5% ± 4% (p=0.9), respectively. Other variables considered in univariate analysis for SR aGVHD were recipient age (p=0.9), recipient CMV status (p=0.7), disease status (p=0.9), RIC vs MAC (p=0.16), MRD vs MUD (p=0.09), PB vs BM graft (p=0.4), and GVHD prophylaxis (p=0.2). In multivariate analysis, the risk factors for SR aGVHD were FD (HR=3.1, 95%CI: 1.2-7.9, p=0.01) and MUD (HR=2.9, 95%CI: 1.1-7.1, p=0.02). The CI of SR aGVHD were 32.3% ± 9.7% in the FD + MUD group (n=29), 9.7% ± 3.8% in the MD + MUD group (n=73), 10.7% ± 4.1% in the FD + MRD group (n=59), and 6.5% ± 3.7% in the MD + MRD group (n=51); p=0.004. We conclude that FD was an independent risk factor for SR aGVHD after allo-SCT with rabbit ATG in male adult patients mostly transplanted with PB after RIC. There was no impact of FD on aGVHD II-IV or III-IV, chronic GVHD, NRM, REL or survival. Finally, the absence of impact of FD on NRM and survival should be interpreted with caution given the retrospective design of the study and because we cannot exclude a possible limiting effect of an insufficient number of patients. Disclosures: No relevant conflicts of interest to declare.

Description: Introduction: Asparaginase Erwinia chrysanthemi (Erwinia), an asparagine-specific enzyme indicated as a component of a multiagent chemotherapeutic regimen for the treatment of patients with acute lymphoblastic leukemia (ALL) who have developed hypersensitivity to E coli-derived asparaginase, is approved in the US for intramuscular and intravenous (IV) administration (Erwinaze USPI 2016). IV administration of proteins, such as asparaginase, may be associated with infusion reactions that may be mitigated by prolongation of the infusion time. Pharmacokinetic (PK) simulations based on a population PK (PPK) model were performed to evaluate PK of IV Erwinia infusions over 2 hours vs 1 hour. Methods: The serum asparaginase activity (SAA) levels from a phase 2, open label, single arm, multicenter, PK study following Erwinia treatment by IV infusion in ALL subjects (1-30 years old) with a previously documented hypersensitivity reaction (≥Grade 2) to pegaspargase were characterized by nonlinear mixed-effects modeling using NONMEM® software. The PPK model included data after 6 doses (1 cycle) of IV Erwinia administered over 1 hour at a 25,000 IU/m2 dose given on a Monday/Wednesday/Friday schedule. The PPK base model for IV Erwinia was identified by comparing different structural and error models. A stepwise approach with forward addition and backward elimination was used to identify important covariates. Visual predictive checks (VPC) were used to demonstrate PPK model predictability. Resampling of the original PK population was used to create a virtual patient population (n = 9999) to simulate average SAA levels and corresponding variability for both 1- and 2-hour IV infusion regimens, summarized by 95% prediction intervals. Results: The PK analysis set contained 24 subjects with 331 evaluable PK samples collected over 6 cycles of therapy. The majority of PK samples were collected during Cycle 1 (55%) with the majority of subjects being male (62.5%) and Caucasian (79%). The median (range) for age was 6.5 years (1-17) and for weight 21.4 kg (10.9 to 118.3). Trough SAA levels ≥0.1 IU/mL at 48 hours post dose 5 (primary endpoint) were achieved by 83% of the evaluable subjects in Cycle 1 (95% CI: 63%-95%). A 2-compartment model with inter-individual variability on clearance (CL) was found to best describe the PK of Erwinia. The final model for CL was: CL [mL/hr] = 123[mL/hr] + 17[mL/hr] * (weight [kg] / 21.4 kg) The central and peripheral volumes of distribution were 1.59 L and 0.154 L, respectively. As more PK samples were available for Cycle 1, separate proportional error models partitioned residual variability for Cycle 1 (36.6%) vs Cycles 2-6 (48.6%). The derived mean half-life estimate (CV%) was 7.51 hours (23.9%). The VPC demonstrated the bulk (〉90%) of the observed SAA falling within the 95% PPK model-based simulation prediction intervals. The PPK model-based SAA simulation results after 1- and 2-hour IV infusions are displayed in Table 1 and Figure 1. Despite different initial rates of rise, similar SAA vs time profiles were observed when comparing 1- vs 2-hour infusion durations (Figure 1), with a similar number of subjects achieving therapeutic trough levels ≥ 0.1 IU/mL (Table 1). Conclusions: The final PPK model demonstrated a good ability to predict SAA allowing estimation of elimination characteristics and the application of subsequent model-based simulations.The PK simulations for 1- vs 2-hour IV Erwinia showed similar mean 48-hour trough SAA levels, as well as similar proportions of subjects having therapeutic 48-hour trough SAA ≥ 0.1 IU/mL. Based on these PK results, duration of IV Erwinia could be increased from 1 to 2 hours, which may reduce symptoms of infusion reactions. Reducing infusion reactions may potentially enablemore patients to complete Erwinia treatment. Upon approval from FDA in 2016, IV infusion duration in the Erwinaze USPI was revised from 1 hour to 1-2 hours; this is also consistent with the updated Children's Oncology Group protocols. Support: Jazz Pharmaceuticals. Disclosures Zomorodi: Jazz Pharmaceuticals: Employment, Equity Ownership. Dumas:Quintiles: Employment; Jazz Pharmaceuticals: Consultancy. Berry:Quintiles: Employment; Jazz Pharmaceuticals: Consultancy. Johnston:Jazz Pharmaceuticals: Consultancy; Quintiles: Employment. Eller:Jazz Pharmaceuticals: Employment, Equity Ownership.

Description: The standard risk factors for acute graft-versus-host disease (aGVHD) after allogeneic stem cell transplantation (allo-SCT) from related or unrelated donors are well defined and include HLA mismatch or unrelated donor, older recipient age, and female donor for male recipient (FM). The steroid-refractory (SR) forms of aGVHD are important to consider because they often have a major deleterious impact on transplant outcome. Unfortunately, the specific risk factors for SR aGVHD are less clearly defined. To characterize these risk factors after allo-SCT from matched related or unrelated donors, we undertook a retrospective analysis of adult patients transplanted at our center between 01/01/2000 and 12/31/2012. Steroid-refractory aGVHD was defined as aGVHD progressing after 3 days of treatment, or unchanged after 7 days, or in incomplete response after 14 days. GVHD occurring after donor lymphocytes infusion were excluded. Overall survival (OS) was calculated using the Kaplan-Meier estimate. Cumulative incidences (CI) were used for SR aGVHD in a competing risk setting with death as a competing event. The Gray test was used to compare CI curves. For multivariate analysis, the variables with p value 〈 0.1 were entered into a Fine-Gray model and the least significant variables were excluded in sequential fashion until all remaining factors were significant at the p=0.05 level. The variables considered were recipient age (≥ vs 〈 50 years, median age), female vs male donor, FM vs other combinations, matched related (MRD) vs matched unrelated donor (MUD), peripheral blood (PB) vs bone marrow (BM) graft, GVHD prophylaxis with cyclosporine (CsA)+metho vs others, myeloablative (MAC) vs reduced-intensity conditioning (RIC) regimen, antithymocyte globulin (ATG) vs no ATG, number of CD34+ cells in the graft ≥ vs 〈 5.6 x 106/kg (median number), early (CR1, PR1, chronic phase, or untreated) vs advanced disease, CMV-seropositive vs seronegative recipient. Unrelated donors were matched at the allele level for HLA-A, B, C, DRB1, DQB1. Six hundred and thirty four patients were identified and included in the present study. The median age was 50 years (18-67). Diseases were AML (n=230), ALL (n=104), myeloma (n=80), NHL (n=74), Hodgkin's disease (n=18), MDS (n=47), CLL (n=29), CML (n=18), aplastic anemia (n=19), and MPS (n=15). Status at transplant were CR1 or PR1 or chronic phase (n=260), 〉 CR1 or PR1 (n=237), refractory (n=101), or untreated (n=36). Conditioning regimens were RIC (n=405) or MAC (n=229). Rabbit ATG was administered to 327 patients, of whom 298 received a RIC regimen. Donors were MRD (n=360) or MUD (n=274). Sources of stem cells were PB (n=452), BM (n=177), missing data (n=5). The prophylaxis of GVHD was CsA+metho for 339 patients. In the whole population, 71 patients presented a SR aGVHD at a median time of 29 days (8-137) after transplant, representing a CI of 11.2% ± 1.2%. Their OS at 1 year post-transplant was 27% ± 5%. In univariate analysis, the risk factors for SR aGVHD were MAC (p=0.02), MUD (p=0.02), no ATG (p=0.01), and a trend for FM (p=0.07). Other variables considered in univariate analysis were recipient age (p=0.6), female vs male donor (p=0.6), recipient CMV status (p=0.7), status at transplant (p=0.2), PB vs BM graft (p=0.9), number of CD34+ cells (p=0.4), and GVHD prophylaxis (p=0.6). In multivariate analysis, the risk factors for SR aGVHD were MUD (HR=2.5, 95%CI: 1.5-4.1, p=0.0003), FM (HR=2, 95%CI: 1.2-3.4, p=0.008), and no ATG (HR=2.1, 95%CI: 1.3-3.4, p=0.002). Patients were then divided into 3 groups. The CI of SR aGVHD was 2.7% ± 1.6% in the low-risk group (no MUD + no FM + ATG; n=112), 21.8% ± 3.4% in the high-risk group (MUD + FM +/- ATG, or MUD + no FM + no ATG; n=147), and 9.6% ± 1.5% in the intermediate-risk group (n=375); p=10-6. We conclude that MUD, FM, and no ATG were independent risk factors for SR aGVHD in adult patients after allo-SCT from MRD or MUD. These data support the prevalent roles of HLA and donor T cell alloreactivity in the pathogenesis of SR aGVHD and may help identify patients at high risk of SR aGVHD, candidates for experimental first-line therapies of aGVHD. Disclosures: No relevant conflicts of interest to declare.