Selective inhibition of protein methyltransferases is a promising new approach to drug discovery. An attractive strategy towards this goal is the development of compounds that selectively inhibit ...binding of the cofactor, S-adenosylmethionine, within specific protein methyltransferases. Here we report the three-dimensional structure of the protein methyltransferase DOT1L bound to EPZ004777, the first S-adenosylmethionine-competitive inhibitor of a protein methyltransferase with in vivo efficacy. This structure and those of four new analogues reveal remodelling of the catalytic site. EPZ004777 and a brominated analogue, SGC0946, inhibit DOT1L in vitro and selectively kill mixed lineage leukaemia cells, in which DOT1L is aberrantly localized via interaction with an oncogenic MLL fusion protein. These data provide important new insight into mechanisms of cell-active S-adenosylmethionine-competitive protein methyltransferase inhibitors, and establish a foundation for the further development of drug-like inhibitors of DOT1L for cancer therapy.
In current library practice, trained human experts usually carry out document cataloguing and indexing based on a manual approach. With the explosive growth in the number of electronic documents ...available on the Internet and digital libraries, it is increasingly difficult for library practitioners to categorize both electronic documents and traditional library materials using just a manual approach. To improve the effectiveness and efficiency of document categorization at the library setting, more in-depth studies of using automatic document classification methods to categorize library items are required. Machine learning research has advanced rapidly in recent years. However, applying machine learning techniques to improve library practice is still a relatively unexplored area. This paper illustrates the design and development of a machine learning based automatic document classification system to alleviate the manual categorization problem encountered within the library setting. Two supervised machine learning algorithms have been tested. Our empirical tests show that supervised machine learning algorithms in general, and the k-nearest neighbours (KNN) algorithm in particular, can be used to develop an effective document classification system to enhance current library practice. Moreover, some concrete recommendations regarding how to practically apply the KNN algorithm to develop automatic document classification in a library setting are made. To our best knowledge, this is the first in-depth study of applying the KNN algorithm to automatic document classification based on the widely used LCC classification scheme adopted by many large libraries.
Introduction: New approaches to find and then drug pediatric acute myeloid leukemia (AML)-specific targets are clearly needed to help the nearly 35% of patients who still die from the disease. While ...RARA is a known druggable target in acute promyelocytic leukemia (APL), the utility of using retinoic acid agonists in non-APL AML has not proven consistently beneficial. Super enhancers (SEs), large regions of highly active chromatin, define cell state and cell identity by regulating oncogenes in many cancers. Recent enhancer profiling of 66 adult non-APL AML patient samples revealed SE-defined, prognostically relevant subgroups. An SE was detected at the retinoic acid receptor alpha (RARA) gene locus in 59% of the samples, which were sensitive to the second-generation retinoic acid agonist tamibarotene which has led to a phase II clinical trial. This study confirms that characterization of the chromatin-defined dependencies in specific cancers can pinpoint targets which can be drugged. We are delineating the transcriptional regulation of pediatric AML (pAML) by SE analysis, which has already elucidated deeper insights into pediatric leukemogenesis, as typified by strong RARA dependence in a majority of pAML.
Methods: Three AML cell lines and 19 pAML primary samples were enhancer profiled by H3K27Ac chromatin immunoprecipitation followed by massively parallel sequencing (ChIP-seq). SEs were detected and assigned to genes using the rank ordering of super-enhancers (ROSE) algorithm. Tamibarotene treatment of cell lines and patient samples were assessed for gene and protein expression changes and phenotypic differences. For in vivo assessment, 200,000 cells of a RARA SE+ pAML patient sample were injected into each NSGS mouse by tail-vein injection. One week after injection, tamibarotene (6mg/kg) or vehicle treatment was initiated by gavage (n=7 each arm). Peripheral blood monitoring of leukemia burden was determined flow cytometry.
Results: The primary pAML sample cohort encompassed the diverse pAML cytogenetic subtypes (Fig 1a), with an overrepresentation of KMT2A rearrangements (n=9, 47%). The number of unique enhancer regions was nearly saturated in the 19 samples. Median SE size was 3,780bp, much larger than the 511bp of typical enhancers. When SE regions across all samples were clustered together, a RARA SE was seen in two of the ten clusters (Fig 1b). Eleven of the 19 samples (58%) contained a RARA SE, crossing multiple cytogenetic subtypes (Fig 1c). Tamibarotene treatment of RARA SE+ pAML cell lines and patient samples suppressed proliferation and increased apoptosis (detected by annexin V+), with minimal effect in Kasumi, a pAML cell line without a RARA SE (Fig 2a). In the RARA SE+ cell lines and samples only, tamibarotene increased CD38 (a myeloid differentiation marker usually suppressed by ligand-unbound RARA) (Fig 2b). High RARA mRNA levels confirmed the SE assignments in both cell lines and patient samples (Fig 2c). Tamibarotene induced the transcription of DHRS3 (another RARA target gene used as a pharmacodynamic biomarker in the adult tamibarotene phase II trial) (Fig 2d). Tamibarotene suppressed colony formation ability in RARA SE+ cell lines. An ongoing RARA SE+ patient-derived pAML xenograft confirmed tamibarotene markedly suppressed disease progression (Fig 3), with vehicle mice requiring euthanasia 42 days after tail-vein injection for significant disease burden, while the tamibarotene treated mice continued to be well-appearing at the same timepoint.
Conclusion: We have profiled the enhancer landscapes of 19 primary pAML samples, the largest dataset of its kind, and seen a high frequency of a RARA SE in pAML. Tamibarotene has anti-proliferative, proapoptotic, and on-target pro-differentiation effects in RARA SE+ pAML in vitro and marked anti-leukemia activity in vivo. Given these positive findings, we are evaluating combinations of other AML-active agents with tamibarotene. Additionally, as there is a range of sensitivity to tamibarotene in RARA SE+ samples, we are also interrogating other SE-regulated genes interacting with RARA that may predict degree of response or resistance to tamibarotene. Our studies confirm that studying the transcriptional regulation of pAML samples through SE analysis can identify druggable targets and also lay the preclinical foundation for a biomarker-defined tamibarotene trial in pediatric AML.
Display omitted
Wei:NHI NHLBI Grant: Other: received funding . Lin:Syros Pharmaceuticals: Equity Ownership, Patents & Royalties.
Introduction: Hepatotoxicity is a challenging adverse event associated with acute lymphoblastic leukemia (ALL) therapy. Due to a lack of large, diverse, and well-characterized cohorts in pediatric ...ALL, factors predictive of therapy-related adverse events such as hepatotoxicity are not well understood. Here, we leveraged a data repository jointly developed between two large pediatric oncology treatment centers to assess the impact of host and disease-related factors on risk for liver dysfunction. Our aim was to better understand risk determinants of hepatotoxicity in patients risk-stratified to receive one of two potentially hepatotoxic regimens during their first interim maintenance (IM1) therapy phase.
Methods: All patients were treated for ALL between 2006-2014 at either Children's Hospital of Philadelphia (CHOP) or Texas Children's Hospital (TCH). We collected patient data by combining targeted manual abstraction and extensive semi-automated extraction of patient electronic medical record (EMR) data. Demographic information, disease-related data, and all laboratory values collected during treatment (along with age-related normal values) were obtained and stored in REDCap™ databases. To reduce cohort heterogeneity and better address our objective of understanding hepatotoxicity risk with standard ALL regimens, we removed patients who received non-standard ALL therapies. We separated the remaining patients into those who received more intensive IM1 therapy with high dose methotrexate (‘HDMTX’: 5grams/m2 intravenously over 24hrs q2 weeks x 4) vs. those who received less intensive therapy with Capizzi methotrexate (‘Capizzi’: dose escalation by 50 mg/m2 every 10 days x 5, starting at 100mg/m2). We restricted our analysis to liver-relevant labs obtained during this phase: AST (SGOT), ALT (SGPT), total bilirubin, conjugated bilirubin, and creatinine. Each was normalized as a ratio of the age-related upper limit of normal (ULN), and then each lab obtained throughout the phase was averaged across all patients within the two treatment groups. Multivariate mixed-effects linear regression with random effects was used to identify differences in each averaged lab value during IM1 related to the following covariates: race/ethnicity, sex, age, treating center, and overweight/obesity (yes/no). Significance cutoff: p <0.05
Results: Out of 923 ALL patients, 658 received IM1 as their third block of ALL therapy and were included for analysis. Mean age was 7.1 years, with those receiving HDMTX older than those receiving Capizzi, as expected (4.91 years vs. 10.03 years, p <0.0001). Non-Hispanic white patients comprised nearly 50% of patients, followed by Hispanic patients (35%), and the remaining Non-Hispanic black, Asian, and other each <10%. There was no difference between sex, treating center, or being overweight/obese between the Capizzi and HDMTX groups (Table 1). Both treatment intensity groups had mildly elevated mean transaminases (ALT, AST) during IM1 (Table 2), however the HDMTX group slightly higher transaminase averages than the Capizzi group (ALT: 2.72 vs. 2.42; AST: 1.70 vs. 1.22). For all patients, both mean bilirubin (total and conjugated) and mean creatinine were below the ULN. Analysis of all potential covariates identified that both race and age were significantly associated with higher ALT and AST, but only in the Capizzi group (Tables 3 and 4). Markedly, Hispanic patients treated with Capizzi were 2.5x more likely than non-Hispanic white patients to have higher ALT. There was no effect on bilirubin by any of the covariates.
Conclusions: This is the first comprehensive study of factors related to hepatotoxicity in a large and uniformly-treated cohort of pediatric patients of ALL. A wide distribution of liver function laboratory values was seen in IM1 for both Capizzi and HDMTX groups, with modest ALT and AST elevations noted but not bilirubin. Not surprisingly, HDMTX patients (who received both high-dose methotrexate and 6-mercaptopurine in IM1) did have slightly higher ALT/AST on average than those who received Capizzi. Race and/or age influence ALT and AST during IM1 for the Capizzi group only. Further studies are needed to delineate the temporal correlation of liver function changes in IM1 and, more broadly, all phases of therapy to predict toxicities, guide dose modifications, and help guide the incorporation of investigational agents into existing ALL regimens.
Display omitted
No relevant conflicts of interest to declare.
Introduction: Hepatotoxicity is a frequent and challenging adverse event in children with acute lymphoblastic leukemia (ALL), but patient factors that are predictive of hepatotoxicity are not well ...understood. We leveraged a data repository jointly developed by two pediatric oncology centers within the Leukemia Electronic Abstraction of Records LEARN Consortium to assess the landscape and determinants of liver dysfunction throughout ALL therapy in patients who were risk-stratified to receive either standard- or high-intensity treatment blocks.
Methods: The subjects were children ages 1-21 years who were treated for ALL between 2006-2014 at either Children's Hospital of Philadelphia or Texas Children's Hospital. Demographics, disease-related data, and every laboratory value collected during treatment were obtained by targeted manual abstraction and extensive semi-automated extraction of patient electronic medical record (EMR) data. To reduce cohort heterogeneity, we excluded patients who received non-standard ALL therapies. Patients were categorized as receiving either standard-intensity or high-intensity treatment for their first three blocks of therapy (Induction, Consolidation, Interim Maintenance 1 IM1) based on chemotherapeutic agents delivered in those blocks. Differences in laboratory value-determined hepatoxicity were then analyzed based on this categorization for all remaining phases of therapy. Hepatic lab values (AST SGOT, ALT SGPT, total bilirubin t. bili, and conjugated bilirubin c. bili) were first normalized to the age-based upper limit of normal (ULN), and the median value was then determined. A multivariate mixed-effects linear regression model with random effects was used to identify differences in the treatment group medians and the following covariates: age, race/ethnicity, sex, BMI, and ALL immunophenotype. Laboratory values were classified by the CTCAE v5.0 grading system, with grade ≥ 3 considered ‘elevated.‘
Results: 805 pediatric ALL patients were included in the analysis, representing 114,095 hepatic lab values (Table 1). Less than 10% of patients had elevated lab values at diagnosis. Throughout treatment, the majority of lab values fell within 1-2x ULN for age for both standard- and high-intensity treatment groups (Fig. 1a-d). The median hepatic lab values for the high-intensity group were slightly higher than the standard risk group across all treatment phases, and this difference was most consistently significant in Consolidation and Delayed Intensification. Among the four hepatic labs that were assessed, ALT had the most significant deviation above normal (up to 30x ULN, Fig 1a). Patients were more likely to have elevated transaminases during maintenance than prior to maintenance (Fig. 1e). Similarly, but to a lesser degree, patients were more likely to have elevated t. or c. bili during maintenance than prior to maintenance. Age, race, and BMI were correlated with elevated hepatic labs, with Hispanic and/or overweight patients more likely to have elevations in 3 or more phases of therapy (Table 2). However, no hepatic lab abnormalities were correlated with either overall or relapse-free survival.
Conclusions: This is the first comprehensive study of measures of hepatotoxicity in a large and uniformly-treated cohort of pediatric ALL. While significantly elevated hepatic labs are rare at diagnosis, they are common during ALL treatment and are seen more commonly in maintenance than in prior phases. Patients who are overweight and/or Hispanic are more likely to experience grade 3 or higher hepatoxicity. We observed no relationship between hepatotoxicity and relapse or survival. Further studies are ongoing to delineate the temporal correlation of liver function and chemotherapy dosing and administration.
Display omitted
No relevant conflicts of interest to declare.
▪
Background: Approximately 10-20% of children and young adults with B-cell acute lymphoblastic leukemia (B-ALL) will relapse. Therapies that can bridge patients (pts) to hematopoietic stem cell ...transplantation (HSCT) or chimeric antigen receptor (CAR)-T cell therapy are critical to improving outcomes. Inotuzumab ozogamicin (InO) is a CD22-targeted antibody-drug conjugate linked to calicheamicin that is FDA-approved for adults with R/R refractory B-ALL. Prospective data on the efficacy and safety of InO in pediatric pts is lacking.
Methods: This single arm phase 2 trial enrolled pts age 1-21 years with CD22-positive B-ALL in >2nd relapse, refractory to 2 prior induction regimens, any relapse after HSCT, or 1st relapse with Down syndrome (DS). The primary aim was rate of complete response (CR) or CR with incomplete count recovery (CRi) following cycle 1. Secondary aims included adverse events, incidence of sinusoidal obstruction syndrome (SOS), and level of minimal residual disease (MRD) measured by multiparameter flow cytometry in responders. Eligibility included ≥5% marrow blasts, direct bilirubin ≤1.5x upper limit of normal, and no prior SOS. Pts received one cycle of InO at the FDA-approved adult dose of 1.8mg/m2 (0.8mg/m2 on day 1, 0.5mg/m2 on days 8 and 15). Central nervous system (CNS) status dictated intrathecal therapy. Based on response at day 28, pts with at least stable disease (SD) could receive a 2nd cycle; those with CR/CRi received InO 0.5mg/m2 on days 1, 8, and 15 in cycle 2, while those without CR/CRi received the same fractionated dose as cycle 1. Pts with CR/CRi after 2 cycles were eligible to receive up to 6 total cycles. The trial used an admissible two-stage design (α=0.05, β=0.20) testing the null hypothesis of CR/CRi rate of 30% vs. alternative of 48%, requiring a minimum of 9/24 (response/enrolled) for stage 1 and 20/48 total. The trial was continuously monitored for dose-limiting toxicities (DLT) and SOS. Results herein are based on data cutoff of June 30, 2019.
Results: Forty-eight pts received InO and were evaluable for response/toxicity. Median age was 9 years (range 1-21). Three pts had DS. 67% were in >2nd relapse, 21% were in 1st relapse but refractory to reinduction, 23% had prior HSCT, 23% had prior CD19 CAR-T, and 29% had prior blinatumomab. Median marrow blasts were 81% (range 6-100), with 81% of pts having M3 marrow. 19 pts achieved CR and 9 achieved CRi after cycle 1 (CR/CRi rate: 58.3%, 95%CI 43.2-72.4%). Three pts had partial response (PR), 9 SD, and 8 progressive disease (PD); 2 pts (1 PR, 1 SD) had CR/CRi with cycle 2. Two pts with PD in cycle 1 had marrow CR (MRD 0.02% and <0.01% in 1 case each) but progression in the CNS. MRD was reported for 26 of the 28 pts with CR/CRi; of these, 17 (65.4%) had MRD <0.01% and 4 (15.4%) had MRD 0.01-0.099%. The most common adverse events (AE) in cycle 1 were febrile neutropenia (27%) and infection (18.8%). No treatment related toxic deaths were reported. DLTs were reported in 9 pts in cycle 1; the most common DLT was hematologic absolute neutrophils <500/µL or platelets <20,000/µL beyond 42 days from the start of cycle 1 (n = 7), but DLT stopping bounds were not crossed. One pt was not evaluable for hematologic DLT. Four (8.3%) pts had ALT elevation, 6 (12.5%) had AST elevation (maximum grade 3), and 1 had grade 3 bilirubin in cycle 1. No pt required dose modification for hepatic toxicity. Thirteen pts had subsequent HCST and 4 (30.7%) developed SOS. All SOS cases were grade 3 (CTCAE v 5.0) and treated with defibrotide; 3 resolved quickly while one was more severe but resolving at the time of death from other HSCT complications. There were no cases of SOS without subsequent HSCT. Limited samples showed that alteration in surface CD22 expression is a likely mechanism for non-response. Peripheral B cell aplasia was nearly universal in evaluated pts.
Conclusions: InO demonstrated a CR/CRi rate of 58% in these heavily pretreated children and young adults with R/R CD22-positive B-ALL. In responders, 65.4% achieved MRD <0.01%. Minimal hepatic toxicity was observed during InO therapy. SOS occurred in 30.7% of pts who underwent subsequent HSCT (8.3% of pts overall), and all but 1 case resolved quickly with supportive care and defibrotide. The most common DLT was prolonged marrow aplasia. Given the observed efficacy, InO will be incorporated into a randomized phase 3 trial for newly diagnosed pediatric pts with high-risk B-ALL.
O'Brien:Pfizer: Research Funding; Celgene: Research Funding; AbbVie: Research Funding; Amgen: Research Funding; BMS: Research Funding; BTG: Research Funding. Rheingold:Pfizer: Research Funding; Novartis: Consultancy. Borowitz:Beckman Coulter: Honoraria. Raetz:Pfizer: Research Funding. Gore:Amgen: Consultancy, Equity Ownership, Honoraria, Membership on an entity's Board of Directors or advisory committees, Other: travel expenses; Novartis: Consultancy, Other: Service on Data Safety Monitoring Committee; travel, accommodations, expenses; Roche/Genentech: Consultancy, Honoraria, Other: travel expenses; Anchiano: Equity Ownership, Other: spouse employment and company leadership; Blueprint Medicines: Equity Ownership; Celgene: Equity Ownership, Other: DSMC member; Clovis: Equity Ownership; Mirati: Equity Ownership; Sanofi Paris: Equity Ownership. Loh:Medisix Therapeutics, Inc.: Membership on an entity's Board of Directors or advisory committees.
Inotuzumab ozogamicin is FDA-approved for adults with relapsed B-ALL but is not approved for children. Its use in a pediatric population with relapsed B-ALL will be presented
Abstract
RATIONALE
Over 70% of supratentorial (ST) ependymoma are characterized by an oncogenic fusion between C11ORF95 and RELA. C11ORF95-RELA fusion is frequently the sole genetic driver detected ...in ST ependymoma, thus ranking this genomic event as a lead target for therapeutic investigation. RELA is a transcription factor (TF) central to mediating NF-kB pathway activation in processes such as inflammation, cellular metabolism, and chemotaxis. HYPOTHESIS: We posited that C11ORF95-RELA acts as an oncogenic TF that aberrantly shapes the tumor epigenome to drive aberrant transcription. Approach: To this end we developed an in utero electroporation (IUE) mouse model of ependymoma to express C11ORF95-RELA during embryonic development. Our IUE approach allowed us to develop C11ORF95-RELA driven tumor models and cell lines. We comprehensively characterized the epigenome and transcriptome of C11ORF95-RELA fusion driven mouse cells by H3K27ac ChIP-seq, ATAC-seq, and RNA-seq.
RESULTS
This data revealed that: 1) C11ORF95-RELA directly engages ‘open’ chromatin and is enriched at regions with known RELA TF binding sites as well as novel genomic loci/motifs, 2) C11ORF95-RELA preferentially binds to both H3K27ac (active) enhancers and promoters, and 3) Bound C11ORF95-RELA promoter loci are associated with increased transcription of genes shared with human ependymoma.
CONCLUSION
Our findings shed light on the transcriptional mechanisms of C11ORF95-RELA, and reveal downstream targets that may represent cancer dependency genes and molecular targets.
Cooperation between several epigenetic modulators defines MLL-rearranged leukemia as an epigenomic-driven cancer. Wild type MLL catalyzes trimethylation of lysine 4 on histone 3 from the methyl donor ...S-adenosylmethionine (SAM) at homeobox and other genes important for hematopoiesis, promoting their expression during development. However, in MLL-rearrangements, its methyltransferase domain is ubiquitously lost and replaced with >70 known fusion partners. Many of these fusion partners recruit DOT1L, the only known SAM-dependent lysine methyltransferase responsible for the methylation of lysine 79 of histone 3 (H3K79)—a mark associated with most actively transcribed genes. Therefore, the recruitment of DOT1L by MLL fusion partners to MLL-target genes leads to aberrant H3K79 hypermethylation at these loci, resulting in inappropriate gene expression and leukemogenesis. DOT1L as a therapeutic target in MLL has been genetically validated by several groups, leading to the development of SAM-competitive small molecule inhibitors of DOT1L. These inhibitors exhibit excellent biochemical activity and selectivity, yet have delayed cellular activity and needing relatively high doses, with viability effects requiring 7-10 days and EC50s for H3K79 methylation depletion of 1-3 μM in cell lines. In animal studies, this translates to a modest survival benefit while requiring high doses through continuous osmotic subcutaneous infusion. Further optimization of DOT1L inhibitors is therefore needed. To date, development of DOT1L inhibitors has been slow, perhaps related to inadequacy of discovery chemistry assay technologies. All biochemical assays are radioactivity-based and are not miniaturizeable; low-throughput and delayed cellular effects of DOT1L inhibition all hamper the discovery of improved inhibitors. Therefore a pressing need towards improved DOT1L inhibitor discovery is a robust, accessible, and rapid profiling platform.
Toward this goal, we synthesized both FITC- and biotin-tagged DOT1L probe ligands. We confirmed by structural studies that binding of the probes were similar to our previously published inhibitor, depleted H3K79 methylation, and had antiproliferative effects in MLL-rearranged cell lines. We then utilized the probes to devise two non-radioactive, orthogonal biochemical assays to competitively profile putative inhibitors: one employing bead-based, proxmity fluorescence technology and the second using fluorescence polarization technology. These assays are robust and adaptable to high-throughput screening. We also designed a miniaturizable high-content imaging, immunofluorescence-based assay to assess the effect of DOT1L inhibitors on H3K79 methylation, reporting cellular IC50s after just four days of treatment. These three assays were validated against three known DOT1L inhibitors of different potencies, accurately differentiating between the compounds. Together, these orthogonal assays define an accessible platform capability to discover and optimize DOT1L inhibitors.
Our platform rank-ordered a library of SAM derivatives that we synthesized, indicating that large substituents off the SAM base does not affect DOT1L binding. We also explored other features of the SAM core structure, identifying several chlorinated probes that had increased cellular potency (IC50 values ~10nM) relative to the initial compounds published, without losing specificity for DOT1L. The inhibitory effect on MLL-target gene expression correlated to the H3K79me2 decrease reported in high content assay, validating that our high-content assay accurately reports on downstream biology seen later in treatment. And as expected, the high-content potencies of our chlorinated DOT1L probes also correlated to increased anti-proliferative effect in MLL cells.
Overall, we utilized chemistry, biology, and chemical biology tools to develop this profiling platform capability for more rapid discovery and optimization of small molecule DOT1L inhibitors. These assays can additionally be used to screen for non-SAM competitive inhibitors in high-throughput fashion. Furthermore, the DOT1L inhibitors and probes synthesized here (available as open-source tools) are useful in deeper mechanistic studies of the DOT1L complex and its role in MLL.
Armstrong:Epizyme: Consultancy.