ALBERT

Description: Background. Pediatric patients with newly diagnosed acute lymphoblastic leukemia (ALL) are at high risk for developing bacterial infections, particularly during the induction treatment phase. Infections are the primary cause of treatment-related mortality during the induction phase, and also lead to prolonged hospitalization, as well as delays and dose modifications in planned chemotherapy. On DFCI ALL Consortium Protocol 05-001 (2005-2011), 26.6% of 794 enrolled patients (pts) experienced at least one infectious complication during induction. In the subsequent study, DFCI Protocol11-001, we studied whether the use of prophylactic fluoroquinolones during induction would decrease the incidence of bacterial infections. Patients and methods.Between 2012-2015, 229 pts with ALL (aged 1-21 years) were enrolled on Protocol 11-001 at 9 participating sites. Induction therapy, regardless of risk group, included vincristine, methylprednisolone, doxorubicin, low-dose methotrexate and pegylated L-asparaginase. Afebrile pts were started on fluoroquinolone prophylaxis at the time of initiation of therapy and continued until count recovery at the end of induction. Pts were switched to broad-spectrum antibiotics (eg, cefepime) for fever or documented infection. Pts with fever at presentation were started on broad spectrum antibiotics rather than fluoroquinolone, and either remained on broad-spectrum antibiotics or were switched to fluoroquinolone prophylaxis until count recovery per treating clinician. Antifungal prophylaxis was not required. All episodes of microbiologically documented bacterial infection, microbiologically and/or radiographically documented fungal infection, and Clostridium difficile (C. diff) enterocolitis were prospectively collected. Using a 1-sample binomial test, rates of infections on Protocol 11-001 were compared to those from the predecessor study, DFCI Protocol 05-001, which included nearly identical induction chemotherapy but did not include guidelines regarding antibiotic prophylaxis or duration during induction. Results. Of the 229 pts, 89% had B-ALL and 11% T-ALL. Median age was 5.1 yrs (range 1.0-20.9). Eighty-six afebrile pts (37.5%) were administered upfront antibiotic prophylaxis and 141 (61.6%) had fever at diagnosis and received broad-spectrum antibiotics; two afebrile patients did not receive antibiotic prophylaxis for unknown reasons. Of the 86 pts who began prophylaxis, 37 (43%) subsequently developed fever. Toxicity data was available for 222 pts. Thirty-eight episodes of infection occurred in 29 patients. Age, presenting white blood cell count and immunophenotype were not associated with the development of infection. The proportion of pts experiencing an infection on Protocol 11-001 (13.1%) was significantly lower than on Protocol 05-001 (26.6%, p

Description: T cells can be genetically modified to target tumor antigens through the expression of a chimeric antigen receptor (CAR). CAR T cells targeting the CD19 antigen is a novel therapeutic approach for patients with relapsed B cell acute lymphoblastic leukemia (B-ALL). We have previously demonstrated that CAR T cells have a significant clinical benefit in adult patients with relapsed B-ALL. The primary objective of this study (NCT01860937) is to extend the use and test the safety of CD19 specific CAR T cells in a multicenter trial for children and young adults with relapsed CD19+ B-ALL. To date, 24 patients with very high risk (VHR) or relapsed B-ALL have been enrolled on protocol with a median age of 12 years (range 2-20 years) at time of T cell collection. We have treated 9 patients with relapsed B-ALL with a median age 15 years (range 3-22 years) using patient derived T cells expressing a CD19 specific CAR (19-28z). Patients received a dose of 1-3 x 10^6 CAR T cells/kg and complete response (complete remission or complete remission with incomplete count recovery) occurred in 5/9 (55%) patients. Significantly, correlations with response included lower disease burden (as assessed by bone marrow cellularity; p20 fold), Flt-3L (〉55 fold), IL-5 (〉15 fold), IL-6 (〉100 fold), and IL-10 (〉15 fold) were demonstrated in patients with CRS. Monitoring of bone marrow demonstrated peak 19-28z CAR T cell detection within 1-2 weeks following infusion with gradual contracture over 1-2 months. These early results demonstrate the feasibility and significant clinical impact of this approach in patients with relapsed B-ALL. In an effort to more rapidly generate statistically relevant data, demonstrate the "exportability" of this technology between academic institutions, and offer this therapeutic option to a broader number of pediatric patients with chemo-refractory B-ALL we have expanded this trial to include a collaborating institution. The objective of our trial is not to provide an intent-to-treat cohort, but rather demonstrate the tolerability of this technology in patients with relapsed B-ALL. Furthermore, patients meeting disease eligibility were not pre-screened for lymphocyte function prior to collection and/or treatment. Subsequent cohorts of patients will receive 19-28z CAR T cells and will be evaluated for toxicity, persistence of CAR T cells, and for anti-leukemic efficacy. Disclosures Curran: Juno Therapeutics: Consultancy. Off Label Use: CAR T cells for B-ALL. Riviere:Juno Therapeutics: Other: Co-founder, stockholder and consultant. Prockop:Atara Biotherapeutics: Other: I have no financial disclosures, but Atara Biotherapeutics has exercised a licensing agreement with Memorial Sloan Kettering Cancer Center and MSKCC and some investigators at MSKCC have a financial interest in Atara.. Park:Actinium Pharmaceuticals, Inc.: Research Funding; Juno Therapeutics: Consultancy. O'Reilly:Atara Biotherapeutics: Research Funding. Sadelain:Juno Therapeutics: Other: Co-founder, stockholder and consultant. Brentjens:Juno Therapeutics: Other: Co-founder, stockholder and consultant.

Description: Background: Risk-adapted treatment strategies have contributed to excellent outcomes in pediatric acute lymphoblastic leukemia (ALL); however, treatment-associated acute and long-term toxicities persist. Therapy-associated toxicities of note in pediatric ALL are related to a treatment backbone that relies heavily on corticosteroids (prednisone and dexamethasone) and asparaginase (ASP). The most frequently observed toxicities include, but are not limited to, serious infection, pancreatitis, thrombosis and bony morbidities including osteonecrosis (ON) and fracture. Previous studies suggest that children of racial and ethnic minorities are at higher risk for treatment-associated toxicities. We assessed the incidence of treatment-related toxicities in Hispanic and non-Hispanic patients undergoing treatment for pediatric ALL. Patients and Methods: Retrospective cohort study investigating the incidence of treatment-related toxicities including infection, allergy to ASP, pancreatitis, thrombosis and bony morbidities in Hispanic and non-Hispanic children and adolescents with newly diagnosed ALL undergoing therapy on Dana-Farber Cancer Institute (DFCI) ALL Consortium Protocol 05-001. The ethnicity of each patient was designated at the time of study enrollment by research coordinators. Descriptive statistics were calculated, mean +/- SD for continuous variables and frequency and percentages for categorical variables. Toxicity rates were based on number of patients. Comparison between groups was done by Chi-square test or FisherÕs exact test and p -value 〈 0.05 was considered significant. Results: Between 2005 and 2011, 794 children and adolescents (ages 1 - 18 years) were enrolled on Protocol 05-001, 730 of whom were evaluable for this investigation: 150 Hispanic (18%), 580 non-Hispanic (73%). Sixty-four patients did not have ethnicity documented. There was no significant difference in disease-risk group, age or gender between the two groups. Weight was significantly higher in Hispanic patients (31.9 ± 24.4 kg in Hispanic and 26.9 ± 18.7 kg in non-Hispanic, p = 0.021). There was no significant difference in the incidence of ASP-related toxicities (allergy, pancreatitis, thrombosis) between Hispanic and non-Hispanic patients. There was no significant difference in the overall incidence of infection between the two groups (42% in Hispanic and 50% in non-Hispanic, p = 0.081). Non-Hispanic patients had significantly higher rates of opportunistic infection (Pneumocystis pneumonia) than Hispanic patients (0.7% in Hispanic and 4% in non-Hispanic, p = 0.041). A similar difference in the incidence of bacteremia between the two groups approached, but did not reach statistical significance (p = 0.052). The overall incidence of fracture in all patients was 14.5% and was significantly higher in non-Hispanic patients (6% in Hispanic and 16.7% in non-Hispanic, p 〈 0.001). The overall incidence of ON was 8.9% and was significantly higher in non-Hispanic patients (3.3% in Hispanic and 10.3% in non-Hispanic, p = 0.007). (Table 1) Conclusion: The incidence of opportunistic infections and bony morbidities was significantly higher in non-Hispanic patients undergoing treatment for pediatric ALL on the DFCI ALL Consortium Protocol 05-001. The risk for, and impact of therapy-related toxicities varies by a patientÕs treatment tolerance, perhaps as a function of age and gender or as a result of disease biology or genetic polymorphisms affecting drug metabolism. Additionally, non-biologic factors such as medication adherence and nutritional status may also contribute to toxicity incidence in patients undergoing treatment for pediatric ALL. Prospective studies to further investigate our findings are warranted. Table. All Patients, (N = 730) Non-Hispanic, (N = 580) Hispanic, (N = 150) p -value Overall Infection 353/730 (48.4%) 290/580 (50%) 63/150 (42%) 0.081 Bacteremia 289/730 (39.6%) 240/580 (41.4%) 49/150 (9.3%) 0.052 Opportunistic Infection 24/730 (3.3%) 23/580 (4%) 1/150 (0.7%) 0.041 Fracture 106/730 (14.5%) 97/580 (16.7%) 9/150 (6%)

Description: Background. Recurrent chromosomal rearrangements carry prognostic significance in pediatric B-lineage acute lymphoblastic leukemia (B-ALL). Recent genome-wide analyses identified a diverse spectrum of chromosomal rearrangements resulting in novel chimeric fusions associated with poor prognosis when treated with conventional chemotherapy. These fusions are observed more frequently in NCI High-Risk (HR) B-ALL compared with NCI Standard Risk (SR) patients. They often activate ABL and JAK-STAT signaling pathways and have demonstrated sensitivity to the relevant tyrosine kinase inhibitors (TKIs) in in vitro assays and ex vivomodels. The objective of this study was to determine the frequency of NCI HR B-ALL patients enrolled on DFCI ALL Consortium Protocol 05-001 with a kinase-activating fusion that would be amenable to TKI therapy and to describe their associated clinical characteristics and outcomes. Methods. Between 2005-2011, 219 NCI HR, Philadelphia chromosome (Ph)-negative, B-ALL patients were enrolled on DFCI ALL Consortium Protocol 05-001, 105 of whom had sufficient material to undergo kinase fusion testing by validated multiplex reverse transcription polymerase chain reaction (RT-PCR) assays. A total of 35 kinase fusions of ABL-class (ABL1, ABL2, PDGFRB, CSF1R), JAK2 and CRLF2 rearrangements were examined. IGH@-CRLF2 and EPOR rearrangements were not assessed. Fusion products were predicted by NCBI BLAST algorithms, confirmed by singleplex PCR and Sanger sequencing and aligned using CLC Main Workbench Version 7.6.1. IKZF1 deletion (del) status had previously been assessed by multiplex ligation-dependent probe amplification (MLPA). Fisher's exact test and the Wilcoxon rank sum test were used to compare patient characteristics to those with and without any identified fusion for categorical and continuous variables respectively. Event-free survival (EFS) and overall survival (OS) were estimated with the Kaplan-Meier method and compared using a log rank test. Univariate and multivariable Cox proportional hazards models of EFS were constructed. Results. Among 105 NCI HR, Ph-negative, B-ALL patients, 16 (15%) were found to harbor an ABL-class fusion (ETV6-ABL1: n=1; FOXP1-ABL1: n=1; SFPQ-ABL1: n=1; ZC3HAV1-ABL2: n=1) or a fusion activating the JAK-STAT pathway (P2RY8-CRLF2: n=8; PAX5-JAK2: n=4). Sixty-nine percent of patients with an identified fusion (Fusion +) had a concomitant IKZF1 del (n=11). Features associated with fusion-positivity were age of 10 years or older (p=0.003), male sex (p=0.03), Hispanic ethnicity (p=0.01) and IKZF1 del (p=0.0005) (Table 1). Fifty percent of Fusion+ patients experienced an event (induction death (n=1); induction failure (n=1); or relapse (n=6)) compared to 24% of patients without a fusion. The 5-year EFS and OS were 48% (95% CI 22-70%) and 68% (95% CI 39-85%) for Fusion+ patients compared to 78% (95% CI 67-85%) and 88% (95% CI 79-93%) for those without fusions (Figure 1). In univariate analysis, fusion-positivity (HR: 2.66, p=0.02) and IKZF1 del (HR: 3.21; p=0.0018) were each significantly associated with inferior EFS, while age and presenting leukocyte count were not. In multivariable analysis, IKZF1 del, but not fusion-positivity, retained statistical significance (HR: 2.64, p=0.02). Conclusion. Fifteen percent of NCI HR, Ph-negative, B-ALL patients enrolled on DFCI ALL Consortium 05-001 were found to have a kinase-activating fusion. Fusion+ patients frequently harbored concomitant IKZF1 deletion and had an inferior outcome. Future studies should focus on developing clinical strategies to rapidly identify these patients at diagnosis and to test whether the addition of the relevant TKIs to their treatment will improve their outcome. Disclosures Asselin: Jazz Pharmaceuticals: Consultancy, Speakers Bureau; Sigma Tau Pharamceuticals: Consultancy. Loh:Bristol Myers Squibb: Membership on an entity's Board of Directors or advisory committees; Abbvie: Research Funding.

Description: Background: E.coli L-asparaginase (L-ASP) is an important component of treatment for childhood acute lymphoblastic leukemia (ALL), but the optimal preparation and dosing remain to be determined. Pegaspargase (SS-PEG) is a pegylated L-ASP formulation commonly used in frontline therapy. Calaspargase pegol (SC-PEG) is a novel formulation that uses the same ASP enzyme and PEG moiety as SS-PEG but a different linker molecule that is more hydrolytically stable, leading to a longer half-life. On Dana-Farber Cancer Institute (DFCI) ALL Consortium protocols, patients (pts) typically receive a single dose SS-PEG during induction, and then 15 doses every 2-weeks (wks) during post-induction in order to maintain therapeutic serum asparaginase activity (SAA), defined as ≥ 0.1 IU/mL, for 30 consecutive wks. We hypothesized that SC-PEG could be administered less frequently than SS-PEG during post-induction therapy with a similar SAA and toxicity profile. Methods: Between 2012-2015, pts aged 1-21 years with newly diagnosed ALL or lymphoblastic lymphoma (LL) were eligible to enroll on DFCI ALL Consortium Protocol 11-001. Pts were randomized at study entry to receive either SS-PEG (N=120) or SC-PEG (N=119), each given intravenously (IV) at a dose of 2500 IU/m2. Both groups received a single dose during multi-agent remission induction. Post-induction, pts assigned to SS-PEG received 15 doses every 2-wks and those assigned to SC-PEG received 10 doses every 3-wks along with other risk-stratified chemotherapy. Serum samples were obtained 4, 11, 18 and 25 days after the induction dose to determine SAA and prior to each post-induction dose (2 wks after each SS-PEG and 3 wks after each SC-PEG dose) to determine nadir SAA (NSAA) by a validated biochemical assay. Pts were switched to Erwinia asparaginase for Grade 2 or higher allergy or for silent inactivation (defined as 2 consecutive non-detectable NSAA). Asparaginase was permanently discontinued for pancreatitis and held for thrombosis (but re-started once the clot improved). End-induction minimal residual disease (MRD) was assessed in ALL pts by IGH/TCRPCR assay, with low MRD defined as 〈 0.001. Results: 239 eligible pts were enrolled (230 ALL and 9 LL). There were no significant differences in presenting characteristics between randomized arms. SAA during induction and NSAA during post-induction are displayed in Figure 1. SAA was similar for the two preparations at 4, 11 and 18 days after the induction dose, with SAA ≥ 0.1 IU/mL in ≥ 95% of pts at these time points on both arms. 25 days after the induction dose, SAA was higher with SC-PEG (median 0.298 IU/mL vs 0.056 for SS-PEG), with significantly more pts on SC-PEG arm with SAA ≥ 0.1 IU/mL (88% vs 15%, p

Description: BACKGROUND: Bacillus species, including B. cereus and other non-anthracis species, are spore-forming, gram-positive rods, found ubiquitously in the environment and cause bacteremia and CNS infection in children with cancer. Case reports have demonstrated that Bacillus spp. infections carry a high morbidity, with rates of CNS involvement of 30%, and mortality as high as 40%. These studies indicate the preponderance of cases occur in children receiving induction chemotherapy for acute lymphoblastic leukemia (ALL). There have been no large, recent studies describing Bacillus spp. infection in children with an underlying oncologic condition. METHODS: We performed a retrospective medical record review of pediatric cancer patients who received care at the Dana-Farber/Boston Children's Cancer and Blood Disorders Center and developed Bacillus spp. infection between January 1st, 2005 and December 31st 2014. Given that the majority of these infections were found to occur in children with underlying hematologic malignancy, we provide a more detailed description of cases with these underlying conditions. RESULTS: Twenty-six children developed Bacillus bacteremia during the study period. Of the 26 patients, 21 (81%) had acute leukemia (18 ALL; 3 AML), 3 had neuroblastoma, one had Ewing's sarcoma and one had an optic pathway glioma. The mean age of patients with hematologic malignancies at time of Bacillus spp. infection was 7.3 years [range: 1-17]. Sixteen (76%) children were neutropenic at the time of infection. Of the children with ALL, 11 (61%) developed infection during the remission induction phase of treatment, 3 during post-induction/pre-maintenance treatment phases and 3 after relapse (1 during stem cell transplant, 1 at a year post-transplant and 1 during re-induction). Of the 26 patients, 5 (19%) developed CNS complications, all during remission induction for ALL, and the overall mortality two-weeks following Bacillus infection was 19%. DISCUSSION: Bacillus spp. continues to cause serious infections in children with cancer, especially in those with underlying hematologic malignancies. While the mortality at our center is lower than that reported in the literature, the rate of death is still higher than what is typically seen with other organisms and CNS involvement remains common. Further study in this area will include identification of risk factors for development of this infection and recommendations regarding use of prophylactic antibiotics. Disclosures No relevant conflicts of interest to declare.

Description: To expedite the translation of biologic discoveries into novel therapeutics, there is a pressing need for panels of in vivo models that capture the molecular complexity of human disease. While traditional cell lines and genetically engineered mouse models are useful tools, they are insufficient to assess the broad diversity of human tumors within a context that recapitulates in situ biology. Patient-derived xenografts (PDXs), generated by transplanting primary human tumor cells into immune-deficient NOD.Cg-Prkdcscid/Il2rgtm1Wjl/SzJ (NSG) mice, surmount some of the limitations of these traditional platforms and have been increasingly utilized as tools for preclinical investigation. However, the infrastructure required to generate, bank, and characterize PDX models limits their availability to only a few investigators. To address this issue, we established a repository of PDX models of leukemia and lymphoma, which we have named the Public Repository of Xenografts (PRoXe). At the time of this writing, PRoXe contains 213 independent lines that have been passaged through mice once (P0), 123 of which have been repassaged in a second generation (P1) or further repassaged. The repository encompasses AML, B- and T-ALL, and B- and T-cell non-Hodgkin lymphoma (NHL) across a range of cytogenetic- and molecularly-defined subtypes (Table 1). PRoXe is extensively annotated with patient-level information, including demographics, phase of treatment, prior therapies, tumor immunophenotye, cytogenetics, and molecular diagnostics. PDX lines available for distribution are characterized by immunophenotyping, whole transcriptome sequencing (RNAseq), and targeted exon sequencing of ~300 genes. To confirm fidelity of engrafted tumors to their corresponding clinical samples, lymphomas were morphologically assessed in P0 mice by H&E and, when pathologic adjudication was required, by immunohistochemistry. Xenografted leukemias were compared to their original tumors immunophenotypically. Unsupervised hierarchical clustering was performed on 132 of these lines based on transcriptome sequencing data and demonstrated 94% concordance between classification of the PDX lines by RNA expression and by the annotated clinical-pathologic diagnoses. Discordant cases highlighted unusual variants, such as B-ALL with aberrant expression of myeloid markers and a follicular lymphoma that underwent blastic transformation in the mouse. Multiple lines have been luciferized and confirmed to home to bone marrow, spleen, and liver. Existing lines from PRoXe have already been shared with more than ten academic laboratories and multiple industrial partners. All of the data referenced here are freely available through a customized web-based search application at http://proxe.org, and lines can be requested for in vitro or in vivo experiments. We are actively expanding the size of PRoXe to allow for large pre-clinical studies that are powered to detect differences across genetically defined subsets. Thus, we are happy to host additional lines from outside investigators on PRoXe and thereby expand the availability of these valuable reagents. Finally, we have made the source code for PRoXe (in R Shiny) open-access, so that other investigators can establish their own portals. Table 1. WHO diagnostic entities encompassed within PRoXe at P1 or later, or P0 or later for B-ALLs. WHO Classification - number of lines per diagnostic entity AML, Other Myeloid, and Ambiguous Lineage [n=32] ALL [n=107] AML - recurrent gene mutations 6 B-ALL - NOS 44 AML - MDS-related changes 5 B-ALL - MLL-rearranged 11 AML - NOS 4 B-ALL - BCR-ABL 10 AML - MLLT3-MLL 2 B-ALL - hyperdiploidy 9 Acute myelomonocytic leukemia 1 B-ALL - TEL-AML1 8 Acute monocytic leukemia 1 B-ALL - E2A-PBX1 3 AML unable to classify 2 B-ALL unable to classify 1 Blastic plasmacytoid dendritic cell neoplasm 8 T-ALL 21 Mixed phenotype, MLL rearranged 1 B/myeloid acute leukemia 1 Myelodysplastic syndrome 1 Mature B cell neoplasms[n=11] Mature T and NK cell neoplasms [n=4] DBLCL - NOS 4 Angioimmunoblastic T-cell lymphoma 1 Mantle cell lymphoma 3 Adult T-cell leukemia/lymphoma 1 Extranodal marginal zone lymphoma 1 Extranodal NK/T-cell lymphoma 1 B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and BL 3 SŽzary syndrome 1 Disclosures Konopleva: Novartis: Research Funding; AbbVie: Research Funding; Stemline: Research Funding; Calithera: Research Funding; Threshold: Research Funding. Etchin:Karyopharm: Research Funding. Lane:Stemline Therapeutics, Inc.: Research Funding. Stone:Abbvie: Consultancy; Novartis: Research Funding; Celator: Consultancy; Amgen: Consultancy; Celgene: Consultancy; Agios: Consultancy; Sunesis: Consultancy, Other: DSMB for clinical trial; Merck: Consultancy; Karyopharm: Consultancy; Roche/Genetech: Consultancy; Pfizer: Consultancy; AROG: Consultancy; Juno: Consultancy.

Description: Introduction: Treatment of relapsed childhood acute lymphoblastic leukemia (ALL) remains a significant clinical challenge with a need for novel, more effective therapies. Pre-clinical studies have demonstrated that mTOR inhibitors have single agent activity in ALL and are synergistic with other chemotherapy agents typically used to treat ALL, including vincristine, corticosteroids, doxorubicin and L-asparaginase. Methods: Dana-Farber Cancer Institute Protocol 11-237 tested the combination of everolimus (given orally on days 1-32) with a standard reinduction regimen of prednisone (40 mg/m2/daily orally on days 4-32), weekly vincristine (1.5 mg/m2/dose, max 2 mg, on days 4,11,18 and 25), PEG-asparaginase (2,500 U/m2/dose IV on days 5 and 18) and doxorubicin (30 mg/m2/dose on days 4 and 5) with dexrazoxane pretreatment (300 mg/m2/dose on days 4 and 5). Intrathecal cytarabine was administered on day 1 and intrathecal methotrexate, cytarabine and hydrocortisone (IT-MAH) was administered on days 18 and 32; patients (pts) with cerebrospinal fluid blasts at study entry received additional IT-MAH on days 11 and 25. Pts with first bone marrow relapse ALL occurring greater than 18 months after continuous remission were eligible for participation. The primary aim was to identify a maximum tolerated dose of everolimus in combination with this backbone utilizing a classic "3 + 3" dose escalation design. Three dose levels (DL) of everolimus were tested (2-, 3-, and 5-mg/m2/day). 5 mg/m2/day (DL3) was the highest dose tested because it was the recommended phase 2 dose (RP2D) of single-agent everolimus in children with solid tumors. Secondary aims included assessment of second complete remission (CR2) rate and levels of end-reinduction minimal residual disease (MRD) assessed by allele-specific oligonucleotide PCR. On-target activity of everolimus was assessed 6 hours after the first dose of everolimus by examining phosphorylated S6 (pS6) levels in peripheral blood blasts by flow cytometry. Results: Thirteen pts with first relapse B-ALL were enrolled. (Table 1) Median age at enrollment was 10 yrs (range: 2-22 yrs). Median duration of first CR was 3.7 yrs (range 1.5-7.7 yrs). One pt treated at DL 3 stopped study drug early due to Grade 4 infection (meningitis due to Rothia mucilaginosa) and was inevaluable for dose limiting toxicity (DLT) assessment. Table 2 displays the most common Grade 2 or higher adverse events (AE's) attributable to everolimus. There were no Grade 5 events and no episodes of interstitial pneumonitis. No DLTs were observed at DL1 or DL2. One of 6 evaluable pts treated at DL3 experienced a DLT (Grade 4 hyperbilirubinemia). 12 of the 13 pts achieved CR (92%). Of the 12 pts who achieved CR, 9 (75%) had low end-reinduction MRD (²0.001). (Table 1) Of the 7 pts with evaluable samples (2 at DL1, 1 at DL2 and 4 at DL3), 6 (86%) had decrease in pS6 in peripheral blasts (〉20% decrease in mean fluorescence intensity), consistent with on-target mTOR inhibition. Only 1 pt (treated at DL1) did not have evidence of pS6 inhibition; this pt achieved CR2 with low MRD. Discussion: Everolimus given in combination with a standard 4-drug reinduction was well tolerated at all dose levels tested in pediatric pts with first relapse ALL. CR2 rates were comparable to published re-induction regimens, with a promising proportion of pts with low end-reinduction MRD. An expansion cohort is currently enrolling to further assess toxicity of this combination, particularly infectious risks, in order to define the RP2D. Table 1. Patients and Response to Therapy Pt # Dose Level Age (yrs) Duration of CR1 (yrs) DLT? CR? MRD 001 1 10 7.7 No Yes High 002 1 20 3.5 No Yes Low 003 1 8 3.1 No Yes Low 004 2 2 1.5 No Yes High 005 2 22 5.1 No No N/A 006 2 5 3.7 No Yes Low 007 3 12 5.7 No Yes Low 008 3 10 3.0 N/A Yes Low 009 3 11 4.3 No Yes Low 010 3 21 4.8 Yes Yes High 011 3 7 3.4 No Yes Low 012 3 13 5.0 No Yes Low 013 3 17 3.0 No Yes Low Table 2. Grade 2 or higher Adverse Events Attributable to Everolimus (Possible, Probable, Definite) at all dose levels; N=13 patients Toxicity Grade 2 Grade 3 Grade 4 Grade 5 Bacteremia 0 3 0 0 Bacterial meningitis 0 0 1 0 Invasive fungal disease 0 1 0 0 Febrile Neutropenia 0 7 0 0 Mucositis 2 2 0 0 Anorexia 3 1 0 0 AST increase 1 5 2 0 ALT increase 1 5 2 0 Bilirubin increase 0 2 1 0 Hypophosphatemia 2 5 1 0 Disclosures Pauly: Seattle Genetics, Inc.: Research Funding. Stegmaier:Novartis Pharmaceuticals: Consultancy.

Description: The tendency of mitochondria to undergo or resist BCL2-controlled apoptosis (so-called mitochondrial priming) is a powerful predictor of response to cytotoxic chemotherapy. Fully exploiting this finding will require unraveling the molecular genetics underlying phenotypic variability in mitochondrial priming. We analyzed pre-treatment T-ALL clinical specimens from a cohort of 47 patients (enriched for treatment failure, but with sufficient controls) treated on the COG AALL0434 or DFCI 05001 clinical trials using BH3 profiling analysis to assess mitochondrial apoptotic priming. We found that there was a strong association between resistance to mitochondrial apoptosis and a poor response to induction chemotherapy (P = 0.008). Furthermore, mitochondrial apoptosis resistance predicted significantly inferior event-free survival (65% vs. 91% at 5 years; P = 0.0376). To define the molecular determinants of this mitochondrial apoptosis resistance, we performed targeted exon sequencing and array CGH copy number analysis. This revealed that loss-of-function mutations in the polycomb repressive complex 2 (PRC2) core subunits (EZH2, EED or SUZ12) were associated with mitochondrial apoptosis resistance (P = 0.007) in clinical specimens. PRC2 is a chromatin modifying complex best known for its role in transcriptional repression, which functions as a tumor suppressor in T-ALL, but whether PRC2 regulates mitochondrial apoptosis is unknown. Using shRNA knockdown in human T-ALL cells, we found that depletion of PRC2 subunits in T-ALL cells induced mitochondrial apoptosis resistance, as assessed by BH3 profiling analysis (P 〈 0.001). PRC2 inactivation also induced resistance to chemotherapy-induced apoptosis (P 〈 0.0001), and increased T-ALL fitness following treatment with the antileukemic drug vincristine (P = 0.0001). Apoptosis resistance upon inactivation of EZH2 (a PRC2 catalytic subunit) was reversed by transduction of wild-type EZH2, but not by an EZH2 point mutant with impaired methyltransferase activity, indicating that this effect is mediated by the enzymatic activity of PRC2. In normal mouse thymocytes, heterozygous deletion of the PRC2 subunits Ezh2 or Eed was sufficient to induce apoptosis resistance in non-transformed double-negative T-cell progenitors (P 〈 0.010), indicating that apoptosis resistance can arise prior to oncogenic transformation. The best-known regulators of mitochondrial apoptosis are BCL2-family genes, but RNA-seq analysis of shRNA knockdown of the PRC2 subunits in a T-ALL cell line revealed that PRC2 did not regulate expression of any of the known BCL2 family members. Instead, PRC2 loss led to upregulation of TRAP1, a mitochondrially localized chaperone of the HSP90 family. TRAP1 upregulation was necessary for induction of apoptosis resistance following PRC2 inactivation, because shRNA knockdown of TRAP1 in the human CCRF-CEM cell line completely blocked induction of apoptosis resistance following PRC2 inactivation (P 〈 0.0001). Moreover, pharmacologic TRAP1 inhibition synergized with the antileukemic drugs dexamethasone and doxorubicin (combination index = 0.37 and 0.42, respectively). To define how PRC2 regulates TRAP1, we performed ChIP-seq analysis, which revealed that TRAP1 regulation by PRC2 is indirect. Combined ChIP-seq and RNA-seq analysis revealed a number of direct targets of PRC2, all of which were tested for their ability to upregulate TRAP1 and induce apoptosis resistance. This showed that the LIM domain transcription factor CRIP2 is a direct target of PRC2 that is necessary and sufficient for regulation of TRAP1, and for induction of apoptosis resistance downstream of PRC2 inactivation. To confirm the relevance of our findings, we used the EZH2 inhibitor GSK126 to inhibit enzymatic activity of PRC2, which revealed that EZH2 normally represses CRIP2 and TRAP1 expression in primary patient-derived xenografts. Finally, we found that increased TRAP1 expression correlates with treatment failure in T-ALL clinical specimens (P = 0.028). Taken together, our findings support a model in which loss of PRC2 induces transcriptional upregulation of its direct target CRIP2, which subsequently activates expression of TRAP1, leading to resistance to chemotherapy-induced mitochondrial apoptosis. Disclosures Aries: Pfizer: Employment. Teachey:Amgen: Consultancy; La Roche: Consultancy. Letai:AstraZeneca: Consultancy, Other: Lab research report; AbbVie: Consultancy, Other: Lab research report; Flash Therapeutics: Equity Ownership; Novartis: Consultancy, Other: Lab research report; Vivid Biosciences: Equity Ownership.