CBFA2T3-GLIS2 Fusion Research Articles

CBFA2T3 Is PPARA Sensitive and Attenuates Fasting-Induced Lipid Accumulation in Mouse Liver.

Peroxisome proliferator-activated receptor alpha (PPARA) is a ligand-activated transcription factor that is a key mediator of lipid metabolism and metabolic stress in the liver. Accumulating evidence shows that PPARA regulates the expression of various protein coding and non-coding genes that modulate metabolic stress in the liver. CBFA2/RUNX1 partner transcriptional co-repressor 3 (CBFA2T3) is a DNA-binding transcription factor that belongs to the myeloid translocation gene family. Many studies have shown that CBFA2T3 is associated with acute myeloid leukemia. Especially, CBFA2T3-GLIS2 fusion is a chimeric oncogene associated with a poor survival rate in pediatric acute megakaryocytic leukemia. A previous study identified that PPARA activation promoted Cbfa2t3 induction in liver and that Cbfa2t3 may have a modulatory role in metabolic stress. However, the effect of CBFA2T3 gene expression on metabolic stress is not understood. In this study, the PPARA ligand WY14643 activated Cbfa2t3 expression in mouse liver. Glucose tolerance test and insulin tolerance test data showed that insulin resistance is increased in Cbfa2t3-/- mice compared to Cbfa2t3+/+ mice. Hepatic CBFA2T3 modulates heat shock protein family A member 1b and carbonic anhydrase 5a expression. Histology analysis revealed lipid droplet and lipid accumulation in the liver of fasting Cbfa2t3-/- mice but not Cbfa2t3+/+ mice. The expression of lipid accumulation-related genes, such as Cd36, Cidea, and Fabp1, was increased in the liver of fasting Cbfa2t3-/- mice. Especially, basal expression levels of Cidea mRNA were elevated in the liver of Cbfa2t3-/- mice compared to Cbfa2t3+/+ mice. Much higher induction of Cidea mRNA was seen in the liver of Cbfa2t3-/- mice after WY14643 administration. These results indicate that hepatic CBFA2T3 is a PPARA-sensitive gene that may modulate metabolic stress in mouse liver.

Abstract P16: CBFA2T3-GLIS2 fusion leads to a distinct DNA methylation enhancer landscape in pediatric acute myeloid leukemia

Abstract P16: CBFA2T3-GLIS2 fusion leads to a distinct DNA methylation enhancer landscape in pediatric acute myeloid leukemia

Clinical Analysis of Pediatric Acute Megakaryocytic Leukemia With CBFA2T3-GLIS2 Fusion Gene.

CBFA2T3-GLIS2 is the most frequent chimeric oncogene identified to date in non-Down syndrome acute megakaryocytic leukemia (AMKL), which is associated with extremely poor clinical outcome. The presence of this fusion gene is associated with resistance to high-intensity chemotherapy, including hematopoietic stem cell transplantation (HSCT), and a high cumulative incidence of relapse frequency. The clinical features and clinical effects of China Children's Leukemia Group-acute myeloid leukemia (AML) 2015/2019 regimens and haploidentical HSCT (haplo-HSCT) for treatment of 6 children harboring the CBFA2T3-GLIS2 fusion gene between January 2019 and December 2021 were retrospectively analyzed. The 6 patients included 4 boys and 2 girls with a median disease-onset age of 19.5 months (range: 6-67mo) who were diagnosed with AMKL. Flow cytometry demonstrated CD41a, CD42b, and CD56 expression and lack of HLA-DR expression in all 6 patients. All the children were negative for common leukemia fusion genes by reverse transcription polymerase chain reaction, but positive for the CBFA2T3-GLIS2 fusion gene by next-generation sequencing and RNA sequencing. All patients received chemotherapy according to China Children's Leukemia Group-AML 2015/2019 regimens, and 4 achieved complete remission. Four children underwent haplo-HSCT with posttransplant cyclophosphamide-based conditioning; 3 had minimal residual disease negative (minimal residual disease <0.1%) confirmed by flow cytometry at the end of the follow-up, with the remaining patient experiencing relapse at 12 months after transplantation. Transcriptome RNA sequencing is required for the detection of the CBFA2T3-GLIS2 fusion gene and for proper risk-based allocation of pediatric patients with AML in future clinical strategies. Haplo-HSCT with posttransplant cyclophosphamide-based conditioning may improve survival in children with AMKL harboring the fusion gene.

CD276 (B7-H3) Is an Immunotherapeutic Target in Acute Myeloid Leukemia with Preclinical Efficacy of Vobramitamab Duocarmazine, an Investigational CD276 Antibody-Drug Conjugate

CD276 (B7-H3) is an immune checkpoint inhibitor expressed in a majority of solid tumors and hematologic malignancies, however it has not been well-described in acute myeloid leukemia (AML). A variety of immunotherapies targeting CD276 including CAR T cells and antibody-drug conjugates (ADCs) are being explored in solid tumors, however their role in leukemia has not been well elucidated. We interrogated the transcriptome and associated clinical and laboratory data from a large cohort of pediatric and adult AML patients to define the prevalence, frequency of co-occurring genetic alterations and outcomes associated with CD276 expression. Lastly, we evaluated preclinical in vitro efficacy in AML of vobramitamab duocarmazine (vobra duo), an investigational ADC with a humanized B7-H3 mAb conjugated via a cleavable linker to the prodrug seco-DUocarmycin hydroxyBenzamide Azaindole. Ribo-depleted RNA sequencing data from 2,241 patients with AML (Pediatric ages 0-29 years N=1,412, from recent Phase 3 COG clinical trials; Adult ages 18-88 years BEAT AML N=451, SWOG N=199, TCGA-LAML N=179) with available cytogenetic, molecular, and clinical data were utilized. Conventional karyotyping and mutational analysis including NGS were used to determine additional cytomolecular aberrations. Response was measured by 5-year overall survival (OS), event-free survival (EFS) and relapse risk (RR). Of the 1,412 pediatric patients evaluated, CD276 mRNA expression was positive (defined as mRNA transcripts per million (TPM) &amp;gt;2) in 289 patients (20%). Higher prevalence (58%) was seen in adult AML patients (BEAT AML N=209/451, SWOG N=191/199, TCGA-LAML N=85/179), Fig 1A. Flow cytometry validated cell surface expression in pediatric AML, with 22% of patients expressing CD276. Cell surface expression was also observed in AML cell lines. CD276 mRNA expression is absent in normal hematopoietic cells (TPM &amp;lt;1) with absent cell surface expression in normal bone marrow as assessed by flow cytometry, allowing for targeted CD276 therapies to be utilized with no/minimal on-target off-tumor hematopoietic toxicity. Evaluation of co-occurring cytomolecular aberrations demonstrated a significant enrichment of CD276+ pediatric patients with high-risk genetic alterations, most notably KMT2A fusions (N=150/343, 43.7%), KAT6A-CREBBP (N=11/12, 91.6%) and CBFA2T3-GLIS2 (N=22/39, 56.4%). We evaluated the significance of CD276 mRNA expression on clinical outcomes in childhood AML. 5 year event free survival (EFS) for patients with CD276 expression was 35% compared to 44% for patients without CD276 expression (p=0.0022). Overall survival (OS) was 52% for patients with CD276 expression compared to 61% for patients without CD276 expression (p=0.0015). Relapse was seen in 40.5% of patients with CD276 expression compared to 35.8% of patients without CD276 expression and relapse occurred on average earlier in CD276 positive patients (0.89 years vs. 1.06 years, p=0.02). Given the poor clinical outcomes for patients with high CD276 expression, novel targeted therapies are needed for this high risk cohort. With multiple CD276 immunotherapies under clinical evaluation in solid tumors, we tested vobra duo in OCI-AML2, a CD276 positive KMT2A-rearranged AML cell line, which demonstrated robust cytolytic activity with an IC50 less than 1nM (IC50=0.9306 nM) supporting the potential efficacy of CD276 targeted therapies in AML (Fig 1B). In a large cohort of AML patients, we identified CD276 to be overexpressed in 20% of pediatric AML patients and 58% of adult AML patients with a significant enrichment in high risk fusion subtypes including KMT2A-rearranged AML, KAT6A-CREBBP and CBFA2T3-GLIS2 fusions. Patients with CD276 expression have poor outcomes with significantly worse EFS and OS compared to patients who do not have CD276 expression. With multiple CD276 immunotherapies under clinical evaluation for solid tumors, we hypothesized that targeting CD276 in AML would result in specific cytolytic activity. Vobra duo showed robust in vitro cytolytic activity against CD276 positive AML cells highlighting the need for ongoing preclinical evaluations of CD276 targeted therapies in AML. Given the established safety profile for vobra duo this provides a clear path for rapid translation to clinical use for high risk AML patients.

CLEC2A Is a Novel AML-Restricted Immunotherapeutic Target Enriched in KMT2A-Rearranged Acute Myeloid Leukemia

KMT2A rearrangements (KMT2A-r) are a frequent event in childhood acute myeloid leukemia (AML) and in treatment related AML in adults leading to a highly refractory and deadly subtype of AML regardless of age. Targeted therapies for KMT2A-r AML have had limited success with novel therapies urgently needed. Despite successes in other leukemias, immunotherapy in AML is in its infancy in part due to significant overlap of antigens expressed in AML and normal hematopoietic cells, which would limit efficacy and increase hematopoietic toxicity. We interrogated the AML transcriptome compared to normal hematopoiesis to identify optimal targets that are silent in normal hematopoieisis and are highly expressed in AML. Here we describe a novel immunotherapeutic target, CLEC2A, a member of the C-type lectin family, that is highly enriched in high-risk AML, particularly KMT2A-r AML and absent in normal hematopoeisis. We evaluated the association between KMT2A fusions and CLEC2A expression identifying CLEC2A as a putative oncoprotein and the potential for CLEC2A to serve as an immunotherapeutic target in high-risk AML via antibody-mediated cytotoxicity. Diagnostic specimens from 1,864 patients with AML (Pediatric 0-29 years from COG Phase III trials N=1486; Adult 18-88 years SWOG N=199 and TCGA-LAML N=179) were interrogated using contemporary NGS to identify genetic aberrations. Conventional karyotyping and mutational analysis were used to determine additional cytogenetic and molecular abnormalities. As CLEC2A was entirely absent in normal bone marrow (TPM 0-0.15), positive CLEC2A expression was defined as mRNA transcripts per million (TPM) &amp;gt;1. Although, CLEC2A was detected in only 18% of AML patients (252/1486; 17% pediatric and 90/378; 24% adult patients), it was highly enriched in those with KMT2A-r AML with 77% (N=193) of CLEC2A positive patients having KMT2A-r, accounting for 48% of the total KMT2A-r cohort, Fig 1A. CLEC2A expression was enriched in specific fusion subgroups: MLLT10 (80%), MLLT4 (71%), and MLLT3 (48%), Fig 1B. CLEC2A expression was low to absent in a majority of other patients, though was seen in other high-risk groups including non-KMT2A MLLT10 fusions, NUP98 fusions, CBFA2T3-GLIS2 fusions, and monosomy 7. CLEC2A is also expressed in a subtype of T-ALL (HOXA activated), which is a fusion driven T-ALL. We further evaluated a potential causal link between KMT2A fusions and CLEC2A expression and inquired whether KMT2A fusion proteins might directly bind to the CLEC2A promoter and induce CLEC2A expression. We used AutoCUT&amp;RUN profiling of genome-wide occupancy of KMT2A oncoproteins to evaluate direct binding of the KMT2A fusion protein to the CLEC2A promoter in primary AML patient samples with three different KMT2A fusions (MLLT10, MLLT4 and MLLT3). We found direct binding of KMT2A fusion proteins to the CLEC2A promoter region, indicating a causal relationship between KMT2A fusions and CLEC2A expression. We then evaluated the potential of therapeutic targeting of CLEC2A by an antibody-drug conjugate using a HumZap cytotoxicy assay. CLEC2A antibody conjugated to saporin (a toxin with cytolytic activity when internalized) was incubated with CLEC2A+ and CLEC2A- cells with target-specific cytotoxicity evaluated by colorimetric assay. CLEC2A+ OCI-AML2 cells had significantly higher rate of cytotoxicity compared to CLEC2A- control cells at 10nM Ab-toxin conjugate. Clinical outcomes for patients with and without CLEC2A expression was evaluated. Patients with CLEC2A had significantly worse outcomes (EFS 23% vs. 45.5%, p&amp;lt;0.0001 and OS 38% vs. 63%, p&amp;lt;0.0001 for patients with and without CLEC2A expression, respectively). Given high rates of co-occurring KMT2A-r AML, particularly high-risk fusions (MLLT10 and MLLT4), we evaluated clinical outcomes within the KMT2A-r cohort as these patients have poor EFS and OS. KMT2A-r patients with CLEC2A had significantly worse EFS (24% vs. 39%, p=0.0063) and OS (38% vs. 60%, p=0.0002). Within the KMT2A-r cohort, high rates of relapse were seen for patients with CLEC2A expression compared to those without CLEC2A (72% vs. 55%, p=0.0039). Here we describe CLEC2A, a novel AML-restricted cell surface target that is an ideal immunotherapeutic target. CLEC2A is highly expressed in KMT2A-r AML, entirely absent in normal hematopoietic cells, directly and causally linked to the KMT2A fusion, and can be used for target-directed cytotoxicity.

Acute myeloid leukemia with RAM immunophenotype: A new underdiagnosed entity.

Acute myeloid leukemia (AML) with RAM immunophenotype is a distinct subtype of AML, as described by the Children's Oncology Group (COG), with characteristic morphological and immunophenotypic properties. It is characterized by strong CD56 expression with dim to negative CD45, HLA-DR, and CD38 expression. It is an aggressive leukemia with a poor response to induction chemotherapy and/or frequent relapses. Seven cases with the characteristic RAM immunophenotype were identified in this retrospective analysis of newly diagnosed pediatric AML cases from January 2019 to December 2021. Herein, we have critically analyzed their clinical, morphological, cytochemical, immunophenotyping, cytogenetic, and molecular profiles. The patients were traced and followed for their current disease and treatment status. Of 302 cases of pediatric AML (age <18 years), seven cases (2.3%) with the distinct RAM phenotype were observed, with age ranging from 9 months to 5 years. Two patients were misdiagnosed earlier as small round cell tumor because of the strong CD56 positivity and the absence of leukocyte common antigen (LCA), but they were later correctly identified as granulocytic sarcoma. The bone marrow aspirate showed blasts with unusual cohesiveness and clumping with nuclear moulding, mimicking non-hematologic malignancies. Flow cytometry revealed blasts with low side scatter, dim to negative CD45 and CD38, negative cMPO, CD36, and CD11b; moderate to bright CD33, CD117, and bright CD56. The Mean fluorescence intensity (MFI) of CD13 expression was significantly lower as compared to the internal controls. Cytogenetic and molecular studies did not show any recurrent abnormalities. Reverse transcription polymerase chain reaction for CBFA2T3-GLIS2 fusion was performed in 5/7 cases, with one positive result. On clinical follow-up, two patients were refractory to chemotherapy. Six of the seven cases had succumbed to death (duration of survival: 3-343 days after initial diagnosis). AML with RAM immunophenotype, a distinct form of pediatric AML with a poor prognosis, may pose a diagnostic challenge if presented as a soft tissue mass. A comprehensive immunophenotypic evaluation, including stem cell and myeloid markers, is critical for an accurate diagnosis of myeloid sarcoma with the RAM-immunophenotype. Our data demonstrated weak CD13 expression as an additional immunophenotypic finding.

Abstract LB122: Arima-HiC sequencing accurately detects clinically-relevant structural variants in pediatric leukemia samples

Abstract Genetic structural variants (SVs), especially those leading to gene fusions, are well-known oncogenic drivers. These SVs can produce overexpression or loss-of-function of certain genes, or generate chimeric fusion proteins. Thus, they serve as important disease biomarkers across several cancers and can also represent therapeutically targetable alterations. Detecting these SVs at a gene-level resolution can be challenging with lower resolution karyotyping approaches, or even RNA sequencing approaches, due to difficulties with culture, sample stability, low transcript abundance, or low-quality RNA. In addition, the ability of fluorescence in situ hybridization (FISH) techniques to detect gene fusions is limited by the targeted nature of these assays as they depend heavily on the design and selection of each probe. Here we use a novel method, Arima-HiC sequencing, that utilizes DNA to evaluate 12 pediatric leukemia samples and determine this assay’s effectiveness in detecting clinically-relevant SVs. We first selected 5 archived (cryopreserved) pediatric acute myeloid leukemia (AML) samples (archival period range: 1-4 years) known to be either fusion-positive (n=3) or fusion-negative (n=2) via prior clinical genetic testing (i.e., chromosomes, FISH, and/or microarray). All samples underwent Arima-HiC sequencing. Briefly, chromatin digestion, end-labeling, and proximity ligation were performed prior to DNA purification per the Arima-HiC protocol. Purified DNA was next prepared as a short-read sequencing library and sequenced on a HiSeq X. The raw reads were aligned and deduplicated, and SVs were called using HiC-Breakfinder software. For a discovery set, we then additionally selected 7 pediatric leukemia samples—6 precursor B-cell acute lymphoblastic leukemias (ALL) and 1 AML—for Arima-HiC sequencing (as above). These cases had undergone standard-of-care cytogenetic (karyotyping, FISH, microarray) and/or molecular (targeted cancer NGS sequencing panel) testing clinically, and a genetic driver/known gene fusion had not been identified. Using Arima-HiC sequencing, we identified the clinically-relevant SV in each of our 3 fusion-positive AML cases, consistent with the original diagnostic cytogenetic finding (n= 1 RUNX1-RUNX1T1 fusion, 1 CBFB-MYH11 fusion, and 1 CBFA2T3-GLIS2 fusion). The 2 fusion-negative AML cases were also negative for structural gene fusions by HiC data. In our discovery sample set, Arima-HiC sequencing was able to find clinically-relevant SVs that were not previously detected in 3 of 6 samples: a KMT2A-MLLT10 fusion was found in an AML case, and a ZNF384-EP300 fusion was found in each of 2 ALL cases. A ABHD17B-PTK2B fusion was found in another ALL case, although ABHD17B may be a novel partner PTK2B and is undergoing validation. Rearrangements involving KRAS and EGFR were detected in the final two ALL cases, and are also undergoing validation. Overall, this study demonstrates how Arima-HiC sequencing can provide diagnostic value in pediatric leukemia specimens via the identification of clinically relevant SVs. Citation Format: Anthony Schmitt, Shadi Melnyk, Kristin Sikkink, Lisa Lansdon, Tomi Pastinen, Erin Guest, Midhat Farooqi. Arima-HiC sequencing accurately detects clinically-relevant structural variants in pediatric leukemia samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB122.

Clinical Experience with Azacitidine and Venetoclax +/- Gemtuzumab in Pediatric Patients with Relapsed or Refractory Acute Leukemias: A Report from the Children's Hospital of Philadelphia

Background: Children with refractory or relapsed (R/R) leukemia, particularly acute myeloid leukemia (AML), have limited curative options. Allogeneic hematopoietic stem cell transplantation (HSCT) for patients who achieve subsequent complete remission (CR) after relapse is generally required for long-term cure, but many children with multiply-R/R AML are unable to achieve CR with conventional salvage regimens. Dual therapy with the hypomethylating agent azacitidine and BCL-2 inhibitor venetoclax (aza/ven) is well-tolerated in adults with AML and has become the standard of care for elderly patients or those with co-morbid conditions precluding intensive chemotherapy. The safety of venetoclax in combination with cytarabine +/- idarubicin was established in children with R/R AML (Karol Lancet Oncology 2020), but pediatric experience with the aza/ven combination remains limited. We sought to describe our institutional experience with aza/ven-based therapy in children and adolescents/young adults with R/R acute leukemias. Methods: We conducted a retrospective chart review of patients aged 0-21 years at initial leukemia diagnosis who were treated with commercially-available aza/ven at the Children's Hospital of Philadelphia (CHOP) from January 1, 2019 to June 30, 2022. Data were collected for baseline patient demographics, leukemia immunophenoptyping and genetics, therapy administration, clinical response, and post-aza/ven outcomes. Azacitidine 100 mg/m2 daily intravenously (IV) was given on days 1-5 of each 28-day cycle. Venetoclax was given daily as oral tablets with 3-day ramp-up dosing to minimize tumor lysis syndrome (TLS) risk as 200 mg (day 1), 400 mg (day 2), and 800 mg (days 3-28) adult equivalent dosing (Place Future Oncology 2018) with 50% dose reduction for patients receiving concomitant azoles. Results: Thirty-seven pediatric patients with R/R acute leukemia received aza/ven-based therapy at CHOP during the study period; 27 (72.9%) had AML, 7 (18.9%) had B-cell acute lymphoblastic leukemia (B-ALL), 1 had T-cell ALL (2.7%), and 2 (5.4%) had mixed phenotype acute leukemia (MPAL). Six children did not complete a full 28-day cycle of aza/ven due to progressive disease (n=4) or severe TLS (n=2). Within the cohort of patients who completed 1 cycle, 22/31 (71%) had multiply-relapsed leukemia, and 9/31 (29%) had primary chemorefractory disease. The median age at initial leukemia diagnosis was 8 years (range 0.5-21). Patients in this cohort received a median of 2 aza/ven cycles (range 1-6). Eight patients with AML also received one dose of gemtuzumab ozogamicin (GO) 3 mg/m2 IV (given on day 4, 5, or 8) with cycle 1 aza/ven; 3 patients received additional GO in subsequent cycles. Twelve of 31 patients who completed 1 cycle aza/ven had cycle 2 or later chemotherapy held or delayed, most commonly to facilitate resolution of drug-, infection- or leukemia-induced cytopenias. Amongst the 29 patients who underwent bone marrow response assessment after aza/ven cycle 1 (n=25) or cycle 2 (n=4), 14 were in CR with negative measurable residual disease (MRD) by flow cytometry (<0.1% for AML, <0.01% for ALL/MPAL), including young children with high-risk AML genetics such as CBFA2T3-GLIS2 and ETV6-EP300 fusions. One additional patient with residual leukemia after cycle 1 aza/ven achieved MRD-negative CR after cycle 2. The remaining patients had either persistent leukemia detectable in peripheral blood or were MRD+ in bone marrow at first response assessment.After receipt of aza/ven, 11 patients with AML and 1 with MPAL proceeded to HSCT, 2 patients with B-ALL were treated with autologous CD19 chimeric antigen receptor T cells (CD19CART), and 1 patient with AML received allogeneic CD123CART. The study cohort was followed for a median of 7.8 months (range 1-32 months). At last follow up, 39% (12/31) of patients were alive (Figure 1). Conclusions: We report the experience of a single institution with 37 pediatric patients with multiply-R/R AML, ALL, and MPAL treated with aza/ven-based therapy, often in the outpatient clinic with good tolerability. Importantly, MRD-negative CR was achieved in 14/37 children (37.8%), nearly always after one cycle of aza/ven, which enabled consolidative HSCT in most responders and long-term survival of some patients. Formal clinical trial evaluation of aza/ven in children with R/R acute leukemias, ideally at an earlier stage of relapse, is warranted. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal

CBFA2T3-GLIS2 model of pediatric acute megakaryoblastic leukemia identifies FOLR1 as a CAR T cell target.

The CBFA2T3-GLIS2 (C/G) fusion is a product of a cryptic translocation primarily seen in infants and early childhood and is associated with dismal outcome. Here, we demonstrate that the expression of the C/G oncogenic fusion protein promotes the transformation of human cord blood hematopoietic stem and progenitor cells (CB HSPCs) in an endothelial cell coculture system that recapitulates the transcriptome, morphology, and immunophenotype of C/G acute myeloid leukemia (AML) and induces highly aggressive leukemia in xenograft models. Interrogating the transcriptome of C/G-CB cells and primary C/G AML identified a library of C/G-fusion-specific genes that are potential targets for therapy. We developed chimeric antigen receptor (CAR) T cells directed against one of the targets, folate receptor α (FOLR1), and demonstrated their preclinical efficacy against C/G AML using in vitro and xenograft models. FOLR1 is also expressed in renal and pulmonary epithelium, raising concerns for toxicity that must be addressed for the clinical application of this therapy. Our findings underscore the role of the endothelial niche in promoting leukemic transformation of C/G-transduced CB HSPCs. Furthermore, this work has broad implications for studies of leukemogenesis applicable to a variety of oncogenic fusion-driven pediatric leukemias, providing a robust and tractable model system to characterize the molecular mechanisms of leukemogenesis and identify biomarkers for disease diagnosis and targets for therapy.

Mechanisms of the CBFA2T3-GLIS2 Fusion in AML Maintenance

Mechanisms of the CBFA2T3-GLIS2 Fusion in AML Maintenance

CLINICAL AND MOLECULAR REPORT OF A RARE RAM POSITIVE NON-DOWN SYNDROME INFANT CBFA2T3-GLIS2 NEGATIVE, PRESENTING THE DRIVER P210-BCR-ABL1

Pediatric Acute Megakaryoblastic Leukemia (AMKL) comprises 4%‒15% of Acute Myeloid Leukemia (AML) cases. Down Syndrome (DS)-AMKL usually displays a favorable outcome, whereas the prognosis of non-DS-AMKL is generally poor. Here we present the clinical and genomic data of a rare non-DS-AMKL infant with p210-BCR-ABL1 fusion, co-occurring with a high-risk WT1 mutation. The baby was RAM positive and CBFA2T3-GLIS2 negative, showing refractoriness to treatment. A 1-year-old girl was referred to the Hospital with a history of fever and motor developmental delay for four months. On admission, she presented ecchymosis, swelling, and organomegaly (spleen and liver). The WBC count was 26.5 × 10 3 /μL, hemoglobin level was 7.6 g/dL, platelet count was 25 × 10 3 /μL, and LDH was 2.887. At diagnosis, she showed bilateral myeloid sarcoma. Morphological analysis of the Bone Marrow (BM) showed clustered cytoplasmic blebs. Flow cytometry analysis revealed 83% of blast cells and a profile compatible with AMKL and RAM phenotype. The patient was treated under the high-risk AML-BFM-2004. Imatinib was incorporated into the treatment after BCR-ABL1 detection. However, minimal residual disease remained positive throughout the treatment, making BM transplantation unfeasible. Disease refractoriness was persistent even with the administration of Dasatinib, FLAG-IDA and Venetoclax. The patient died one year after diagnosis due to sepsis during BM aplasia. We performed G-banding, FISH, and Multiplex RT-PCR experiments. The molecular response was assessed by absolute quantification via RT-qPCR assays. Sanger sequencing and NGS were performed on a targeted approach. The karyotype was 46,XX,t(9;22)(q21;q34). We observed polyploidy (3‒20N). FISH in polyploid Nuclei (8N) showed four BCR-ABL1 fusion signals. In diploid metaphases, we observed two BCR-ABL1 signals. Multiplex RT-PCR detected the p210- BCR - ABL1 . RT-qPCR monitoring detected persistent high levels of BCR-ABL1 transcripts associated with disease refractoriness. Sanger sequencing showed no GATA1 mutation. NGS revealed a missense WT1 pathogenic mutation, and CBFA2T3-GLIS2 was not present. RAM phenotype is characterized by overexpression of CD56 and under-expression of CD38, CD45, and HLA-DR. This phenotype is characteristic of children, usually associated with AML-M7 (38%) and intermediate-risk cytogenetics not associated with prognostic subgroups. RAM is related to a poor prognosis with high MRD and induction failure rates compared to non-RAM patients. This phenotype is associated with the CBFA2T3-GLIS2 fusion, which is related to extramedullary involvement, a particular blast morphology, and a worse outcome as compared to fusion-negative AML pediatric patients, including a high frequency of disease refractoriness and relapse. NGS studies did not show CBFA2T3-GLIS2 in our patient. Thus, our results indicate that RAM+ associated with the driver p210-BCR-ABL1, even with specific targeted therapy, may lead to treatment resistance regardless of CBFA2T3-GLIS2+. Here we highlight for the first time the presence of p210-BCR-ABL1 associated with RAM phenotype in the absence of CBFA2T3-GLIS2 and disease refractoriness in a non-DS-AMKL infant.

Abstract 1075: Therapeutic targeting of CBFA2T3-GLIS2 infant AML with ELU001 - folate receptor alpha-directed C’Dot-drug-conjugate

Abstract Acute myeloid leukemia (AML) in infants is a unique subtype of AML that is driven by oncogenic fusions, resulting in a highly proliferative disease. The CBFA2T3-GLIS2 oncogenic fusion that is seen exclusively in infants and young children &amp;lt;3 years is the most refractory AML with near uniform mortality with conventional therapies. As part of our Target Pediatric AML (TpAML) discovery effort to define novel therapeutic targets, we discovered folate receptor alpha (FRα, FOLR1) transcript to be expressed in CBF-GLIS positive AML and verified cell surface expression of FOLR1 on leukemic cells. We present pre-clinical efficacy of a novel FOLR1 directed therapy (ELU001) in CBF-GLIS AML.ELU001 is a novel ~6 nm-diameter FOLR1-targeted C’Dot-Drug-Conjugate (CDC) incorporating ~21 exatecan molecules as a payload that is designed to treat cancers with improved safety and efficacy as compared to antibody-drug conjugates. A Phase 1 clinical trial of ELU001 in adult patients with tumors that overexpress FOLR1 is currently underway (NCT05001282). In this study, we determined the potential utility of ELU001 for CBFA2T3-GLIS2 AML by evaluating its preclinical efficacy in AML models. We confirmed the binding of ELU001 to AML cells expressing FOLR1 by flow cytometry using MV4;11 AML cell line transduced with an FOLR1 expression construct (MV4;11 FOLR1+). We show that MV4;11 FOLR1+ cells readily bind to ELU001 but not MV4;11 parental cells. Consistent with target-dependent binding, in vitro cytotoxicity studies demonstrated that MV4;11 FOLR1+ cells were highly susceptible to ELU001 (IC50=30 pM) while MV4;11 parental cells were only minimally affected by treatment. To evaluate in vivo efficacy, we transplanted MV4;11 FOLR1+ cells transduced with Luciferase-expression construct into immune deficient NSG mice at 1M cells per mouse and treated the mice the following week with different doses of ELU001 (vehicle, 0.3mg/kg Q3Dx6, 0.5 mg/kg Q3Dx3, and 0.65mg/kg Q3Dx3). Leukemia burden was assessed by whole-body bioluminescence (IVIS) imaging, performed weekly. IVIS imaging showed that ELU001 effectively eradicated the leukemia in vivoat all three doses tested resulting in enhanced survival while mice that received the vehicle treatment exhibited disease progression and eventually succumb to the disease. We confirmed that the leukemia was absent in ELU001-treated mice at the study endpoint of 120 days post treatment by flow cytometric analysis of the bone marrow, spleen and liver tissues harvested at necropsy. ELU001 was well tolerated with an apparent maximum tolerance dose of 1.50 mg/kg/mouse (0.5 mg/kg exatecan, Q3DX3). In summary, we demonstrated that ELU001 is highly effective at eliminating FOLR1 positive AML cells in vitro and in vivo, providing the preclinical data to support further assessment of ELU001 in clinical trials for pediatric patients with CBFA2T3-GLIS2 fusions overexpressing FOLR1. Citation Format: Quy Le, Feng Chen, Thao Tang, Cynthia Nourigat McKay, Fei Wu, Melik Turker, Thomas Gardinier, Vaibhav Patel, Aranapakam Venkatesan, Tin Khor, Kai Ma, Gregory P. Adams, Soheil Meshinchi. Therapeutic targeting of CBFA2T3-GLIS2 infant AML with ELU001 - folate receptor alpha-directed C’Dot-drug-conjugate [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1075.

3173 – HUMAN MODELS OF PEDIATRIC CBFA2T3-GLIS2 ACUTE MEGAKARYOBLASTIC LEUKEMIA CONTRIBUTE TO UNCOVER THERAPEUTIC VULNERABILITIES

Acute megakaryoblastic leukemia (AMKL) is a poor prognostic subtype of AML afflicting children mostly under 3 years of age. Chromosomal translocations generating chimeric fusion genes are considered critical transforming events and are present in a high proportion of cases. Among them, the CBFA2T3-GLIS2 (CG2) fusion is associated with a high risk of relapse and accounts for ∼15% of pediatric AMKL cases. Using overexpression of CG2 in human cord blood CD34+ stem /progenitor cells (hHSPC) followed by xenotransplantation in immunodeficient mice, we are reporting the generation of multiple synthetic AMKL models that phenocopy the human disease in term of immunophenotype, morphology and gene expression. CG2 models revealed a stable genomic landscape with few copy number anomalies and lack of recurrent cooperating mutations. Leveraging from local and published transcriptomic datasets of CG2 AMKL samples (14 patients, 10 models), non-CG2 pediatric AMKL (n=66), NUP98-KDM5A synthetic models (n=5) and normal hHSPC (n=4), we have determined a robust gene expression signature defining CG2 AMKL. Gene set enrichment analysis of CG2 AMKL expression profile identified the JAK2-STAT5 axis as a potential pathway that is upregulated in CG2 leukemia, confirmed by downstream STAT5 phosphorylation in CG2 cells. Cell surface proteomic analysis by LC-MS identified multiple membrane proteins upstream of JAK-STAT signaling in CG2 model leukemia compared to normal hHSPC. Single-agent dose-response curves conducted with multiple FDA-approved drugs targeting JAK2 showed specific grow inhibition of CG2 AMKL models compared to normal hHSPC and non-AMKL cell line. Progress in leukemia research is hampered by the paucity of patient samples and the generated AMKL models pave the way toward a better understanding of leukemogenesis and development of targeted therapy. Acute megakaryoblastic leukemia (AMKL) is a poor prognostic subtype of AML afflicting children mostly under 3 years of age. Chromosomal translocations generating chimeric fusion genes are considered critical transforming events and are present in a high proportion of cases. Among them, the CBFA2T3-GLIS2 (CG2) fusion is associated with a high risk of relapse and accounts for ∼15% of pediatric AMKL cases. Using overexpression of CG2 in human cord blood CD34+ stem /progenitor cells (hHSPC) followed by xenotransplantation in immunodeficient mice, we are reporting the generation of multiple synthetic AMKL models that phenocopy the human disease in term of immunophenotype, morphology and gene expression. CG2 models revealed a stable genomic landscape with few copy number anomalies and lack of recurrent cooperating mutations. Leveraging from local and published transcriptomic datasets of CG2 AMKL samples (14 patients, 10 models), non-CG2 pediatric AMKL (n=66), NUP98-KDM5A synthetic models (n=5) and normal hHSPC (n=4), we have determined a robust gene expression signature defining CG2 AMKL. Gene set enrichment analysis of CG2 AMKL expression profile identified the JAK2-STAT5 axis as a potential pathway that is upregulated in CG2 leukemia, confirmed by downstream STAT5 phosphorylation in CG2 cells. Cell surface proteomic analysis by LC-MS identified multiple membrane proteins upstream of JAK-STAT signaling in CG2 model leukemia compared to normal hHSPC. Single-agent dose-response curves conducted with multiple FDA-approved drugs targeting JAK2 showed specific grow inhibition of CG2 AMKL models compared to normal hHSPC and non-AMKL cell line. Progress in leukemia research is hampered by the paucity of patient samples and the generated AMKL models pave the way toward a better understanding of leukemogenesis and development of targeted therapy.

The Role of ETO and CtBP1 in CBFA2T3-GLIS2 Mediated Transcriptional Regulation and Leukemogenesis

CBFA2T3-GLIS2 is the most prevalent fusion oncogene in pediatric acute megakaryoblastic leukemia patients without Down syndrome and is associated with an event free survival of only 8%. A cryptic inversion event on chromosome 16 joins the three nervy homology regions (NHR) of CBFA2T3 to the five zinc fingers of GLIS2. This configuration enables the encoded chimeric transcription factor to bind GLIS2 consensus sequences throughout the genome and recruit transcriptional activators and repressors to alter gene expression and enhance self-renewal capability. Few cooperating mutations have been identified in patients harboring this fusion which suggests it is the sole oncogenic driver. The molecular mechanism by which CBFA2T3-GLIS2 drives leukemogenesis is not fully understood. Identification of components critical to the transcriptional complex and their role in gene regulation may reveal novel therapeutic targets to improve patient outcomes. Studies on the wild type CBFA2T3 and GLIS2 proteins have demonstrated interactions with the transcriptional regulators ETO and CtBP1 respectively. Further p300 has been shown to play a role in transcriptional regulation imparted by both transcription factors. We therefore hypothesize the fusion promotes transcriptional activation when the histone acetyltransferase p300 and the transcription factor ETO are recruited through NHR1 and NHR2, respectively. When the co-repressor CtBP1 is recruited through the PXDLS motif, located in the GLIS2 portion of the fusion, transcriptional repression predominates. Association of these co-factors with the fusion was confirmed through co-immunoprecipitations. Site-directed mutagenesis was then used to systematically delete NHR1 and NHR2 and mutate the PXDLS motif to evaluate the resultant effects on transcriptional regulation, self-renewal, and leukemogenesis imparted by the fusion. A luciferase reporter assay was used to assess transcriptional activation of the BMP2 promoter, a gene which is known to be upregulated by the CBFA2T3-GLIS2 fusion. Loss of NHR1, NHR2, or NHR3 did not alter the ability of the fusion to activate transcription. In contrast, loss of NHR1 and NHR2 in combination (NHR1-2Δ) and mutation of the PXDLS domain decreased transcriptional activation compared to the wild type fusion. The effect of the mutations on self-renewal capability was then evaluated through colony formation assays. Consistent with the luciferase reporter assay, NHR1-2Δ and PXDLS mutants decreased the number of colonies at week six compared to the unmanipulated fusion. Next, we investigated the effect of these mutations on leukemogenesis. Murine models harboring the CBFA2T3-GLIS2 fusion without cooperating mutations have been unsuccessful and patient-derived xenograft models are limited and difficult to manipulate. Therefore, we developed a novel in vivo model of CBFA2T3-GLIS2 driven leukemia. CD34+ stem cells were isolated from human cord blood and transduced with a lentivirus construct encoding the fusion and a GFP reporter. The cells were then differentiated to megakaryoblasts using human TPO and IL1-beta and sorted for purity prior to injection into immunodeficient NSG-SGM3 recipient mice. The fusion positive human primary megakaryoblasts induced a serially transplantable leukemia within 180 days that recapitulates CBFA2T3-GLIS2 positive patient specimens on a transcriptional and protein level. In contrast to our in vitro studies where NHR2 deletion alone did not alter transcriptional activation and self-renewal, the loss of this domain abrogated leukemogenesis in vivo, suggesting a dependency on the association of ETO with the fusion. Mice that received PXDLS mutant cells, however, developed leukemia at a normal latency suggesting that CtBP1 is not required. This study confirms the CBFA2T3-GLIS2 fusion is sufficient for oncogenic transformation of human CD34+ stem cells. We demonstrate that disruption of ETO, p300, and CtBP1 recruitment to the transcriptional complex decreases transcriptional regulation and self-renewal imparted by the fusion. The loss of ETO was the most detrimental to leukemogenesis in our murine model, uncovering a potential new pathway for the development of targeted therapies. Ongoing studies include CUT&RUN sequencing for the fusion, ETO, and CtBP1 to determine co-occupancy of target genes to further understand those that are critical in transformation. DisclosuresGruber: Kura Oncology: Consultancy.

Project Stella: Development and Preclinical Assessment of FOLR1-Directed Chimeric Antigen Receptor T Cells in CBF2AT3-GLIS2/RAM AML

Background: A rare but highly aggressive type of AML that is only seen in infants with a unique immunophenotype (RAM phenotype) is caused by cryptic CBFA2T3-GLIS2 (CBF/GLIS) fusion. This infant AML is highly refractory to conventional chemotherapy with near uniform fatality despite highly intensive and myeloablative therapy (PMID 23153540). Transcriptome profiling of CBF/GLIS AML has revealed new insights into the pathogenesis of the fusion and uncovered fusion-specific molecular biomarkers that could be used for risk stratification and to inform treatment (PMID 30592296). Studying the largest cohort of these high-risk infants, we demonstrated several alterations in gene expression and transcriptional networks in these CBF/GLIS-positive patient samples that have potential for therapeutic targeting (PMID 31719049). FOLR1, which encodes for folate receptor alpha, was highly and uniquely expressed in CBF/GLIS AML but was entirely absent in AML with other cytogenetics abnormalities and in normal hematopoietic cells. Furthermore, we recently demonstrated that forced expression of CBF/GLIS enhances the proliferation and alters differentiation in cord blood (CB) CD34+ early precursors towards megakaryocytic lineage that recapitulates acute megakaryocytic leukemia seen in infants (PMID 31719049). Of significance, we showed that FOLR1 surface expression is causally linked to CBF/GLIS-induced malignant transformation, thus making it an attractive antigen for targeted therapies against CBF/GLIS AML cells. Given that chimeric antigen receptor (CAR) T cells are extremely effective at eradicating relapsed/refractory B-ALL malignancies, we developed FOLR1-directed CAR T cells for pre-clinical evaluation in CBF/GLIS AML.Methods: We generated a FOLR1-directed CAR using anti-FOLR1 binder (Farletuzumab), IgG4 intermediate spacer and 41-BB/CD3zeta signaling domains. The pre-clinical efficacy of FOLR1 CAR T cells was evaluated against CBF/GLIS AML cell lines in vitro and in vivo. CBF/GLIS AML models include CB CD34+ cells transduced with CBF/GLIS expression construct (CBF/GLIS-CB) and WSU-AML cell line. We also engineered Kasumi-1 cell line to express FOLR1 (Kasumi-1 FOLR1+) to evaluate target specificity (Figure 1A).Results: We tested the target specificity of FOLR1-directed CAR T cells against FOLR1-positive (CBF/GLIS-CB, WSU-AML, Kasumi-1 FOLR1+) and FOLR1-negative (Kasumi-1) cells. CD8 FOLR1 CAR T cells demonstrated cytolytic activity against FOLR1 positive but not FOLR1 negative cells (Figure 1B). Furthermore, both CD8 and CD4 FOLR1 CAR T cells produced higher levels of IL-2, IFN-γ, and TNF-α and proliferated more robustly than did unmodified T cells when co-incubated with FOLR1 positive but not FOLR1 negative cells (Figure 1C). These results indicate highly specific reactivity of FOLR1 CAR T cells against AML cells expressing FOLR1. We next investigated the in vivo efficacy of FOLR1-directed CAR T cells. In CBF/GLIS-CB, WSU-AML, and Kasumi-1 FOLR1+ xenograft models, treatment with FOLR1 CAR T cells induced leukemia clearance, while disease progression occurred in all mice that received unmodified T cells (Figure 1D). Activity of FOLR1 CAR T cells in vivo was target specific, as they did not limit the leukemia progression nor extend the survival of Kasumi-1 xenografts (Figure 1D).To determine whether FOLR1 is expressed on normal HSPCs, we characterized FOLR1 expression in normal CB CD34+ samples. FOLR1 expression was entirely silent in HSPC subsets (Figure 1E). Consistent with lack of expression, no cytolytic activity was detected against HPSCs Moreover, FOLR1 CAR T cells did not affect the self-renewal and multilineage differentiation capacity of normal HSPCs as compared to unmodified control T cells (Figure 1F), whereas significant eradication of colonies were detected in the CBF/GLIS-CB cells (Figure 1G).Conclusion: In this study, we demonstrate that FOLR1 CAR T effectively eradicates CBF/GLIS AML cells without compromising normal HSPCs, providing a promising approach for the treatment of high-risk CBF/GLIS AML. Transition of this CAR T to clinical development for infant AML is underway. [Display omitted] DisclosuresHylkema: Moderna: Current equity holder in publicly-traded company; Quest Diagnostics Inc: Current equity holder in publicly-traded company. Pardo: Hematologics, Inc.: Current Employment. Eidenschink Brodersen: Hematologics, Inc.: Current Employment, Other: Equity Ownership. Loken: Hematologics, Inc.: Current Employment, Other: current equity holder in a privately owned company.

Endothelial Cell Niche Promotes Leukemic Transformation of Human Cord Blood Stem/Progenitor Cells Expressing CBFA2T3-GLIS2 Oncogenic Fusion

The CBFA2T3-GLIS2 (CBF/GLIS) fusion is a product of a cryptic translocation exclusively seen in refractory infant AML. Lack of relevant model systems that accurately recapitulate this infant AML has limited progress. To overcome this barrier, we developed an endothelial cell (EC) co-culture system to support malignant transformation, self-renewal, and propagation of leukemia-initiating cells (LIC) in CBF/GLIS-transduced human cord blood hematopoietic stem/progenitor cells (CB HSPCs) ex vivo.Lack of recurrent cooperating mutations suggests that CBF/GLIS fusion might be sufficient for malignant transformation. To test this, we expressed the CBF/GLIS fusion or GFP control in CB HSPCs (CBF/GLIS-CB or GFP-CB) by lentiviral transduction and placed transduced cells in either EC co-culture or myeloid-promoting culture (MC). CBF/GLIS-CB cells expanded faster with prolonged lifespan in EC co-culture compared to MC (Figure 1A). Proliferation of CBF/GLIS-CB cells declined after transfer to either an EC trans-well culture or in suspension culture (Figure 1B), suggesting that direct contact as well as secreted factors are required for optimal growth of transduced cells.The CBF/GLIS fusion has been shown to confer enhanced megakaryocytic differentiation. At 6 weeks, CBF/GLIS-CB cells in EC co-culture formed significantly more megakaryocytic colonies than CBF/GLIS-CB cells grown in MC or CBF/GLIS-GFP cells grown in either condition (Figure 1C). At 12 weeks, CBF/GLIS-CB cells cultured in EC co-culture continued to produce numerous megakaryocytic colonies, demonstrating long lived self-renewal and enhance megakaryocytic differentiation of CBF/GLIS-CB cells co-cultured with ECs.To determine whether the EC niche promotes generation and propagation of LICs, we evaluated the murine engraftment of CBF/GLIS-CB cells expanded on ECs or in MC following 3, 6, 9 and 12 weeks of culture. CBF/GLIS-CB cells cultured in EC co-culture at each time point exhibited robust engraftment that progressed to frank leukemia in vivo (Figure 1D), demonstrating that EC co-culture promotes long-term maintenance of functional LICs. CBF/GLIS-CB cells grown in MC also induced leukemia from 3- and 6-week cultures but then became senescent at 9 and 12 weeks, suggesting limited preservation of the LICs.Flow cytometric analysis of CBF/GLIS-CB cells identified a malignant population that is of the RAM immunophenotype (CD56 hi, CD45 dim, and CD38 dim/-) previously reported in infants with CBF/GLIS AML in both culturing conditions. However, CBF/GLIS-CB cells in EC co-culture constituted an almost homogeneous population that expressed the RAM immunophenotype, whereas only a subset was detected in MC at week 6 (Figure 1E).To determine the fidelity of transformation to primary leukemia, we performed RNA-sequencing of CBF/GLIS-CB cells cultured with ECs or in MC. Unsurpervised clustering analysis demonstrated that the CBF/GLIS-CB cells from weeks 6 and 12 in EC co-culture clustered with primary CBF/GLIS-positive patient samples, but not CBF/GLIS-CB cells cultured in MC nor GFP controls (Figure 1F). Further transcriptome analysis revealed CBF/GLIS and HSC signature genes, previously identified to be associated with CBF/GLIS AML, were both significantly enriched in CBF/GLIS-CB cells grown in EC culture relative to MC (Figure 1G). These results suggested that the signaling pathways that are aberrantly dysregulated in primary CBF/GLIS leukemia are faithfully recapitulated in CBF/GLIS-CB cells co-cultured with ECs.Despite concerted efforts, previous attempts to model CBF/GLIS AML in murine hematopoietic cells have failed to generate overt leukemia. In this study, we demonstrate that in an EC co-culture system, the CBF/GLIS oncogenic fusion is sufficient to transform human CB HSPCs that faithfully recapitulates the morphology, transcriptome and immunophenotype of CBF/GLIS AML as well as highly aggressive leukemia in xenograft models. Furthermore, the EC co-culture system provides a tractable model system to further interrogate the mechanisms of leukemogenesis and identify biomarkers for disease diagnosis and targets for therapy in CBF/GLIS AML. [Display omitted] DisclosuresHylkema: Quest Diagnostics Inc: Current equity holder in publicly-traded company; Moderna: Current equity holder in publicly-traded company. Pardo: Hematologics, Inc.: Current Employment. Eidenschink Brodersen: Hematologics, Inc.: Current Employment, Other: equity ownership. Loken: Hematologics, Inc.: Current Employment, Other: current equity holder in a privately owned company.

Loss of PTEN in Pediatric AML Confers Sensitivity to PARP Inhibition

Pediatric Acute Myeloid Leukemia (AML) is a rare, but deadly cancer. Outcomes over the last 20 years have remained stagnant with an overall 5-year survival rate < 70% and relapse rates around 50%. Further, few new therapies have been successfully introduced to improve these outcomes. Here we report that exploiting deficiencies in DNA damage repair (DDR) is a potential therapeutic strategy for AML. Poly-ADP Ribose Polymerase (PARP) inhibitors were initially developed to target deficient homologous recombination (HR) in BRCA1/2 mutated cancers by blocking single stranded base repair following DNA damage, leading to an accumulation of double stranded DNA breaks, thereby inducing apoptosis.To evaluate the activity of PARP inhibition in pediatric AML, talazoparib was tested as a single agent and in combination with standard chemotherapeutic agents in human AML cell lines representing low (Kasumi-1 and ME-1), intermediate (AML193), and high-risk (CTS, CMS, MOLM-13, and CHRF288-11) disease based on their genomic mutations. Talazoparib showed the highest efficacy as a single agent in all four cell lines with genomic lesions found in high-risk AML subtypes.After combination drug screens, topotecan (synergistic) and gemcitabine (additive) were chosen to move forward to in vivo testing. Our investigational combination was tested in vivo in four murine models representing pediatric AML subtypes harboring AML1-ETO9a (low risk), MLL-AF6 (high risk), CBAF2T3-GLIS2/JAK2 V617F (high risk) and NUP98-KDM5A (high risk) oncogenes. Mice received a backbone of either current standard of care chemotherapy (SOC; anthracycline plus cytarabine) or topotecan plus gemcitabine. NUP98-KDM5A and MLL-AF6 positive mice receiving single agent talazoparib were found to have prolonged survival compared to vehicle alone (p=0.019 and p<0.0001, respectively) which was further enhanced by the addition of chemotherapy irrespective of backbone (p <0.0001). Conversely, mice with AML1-ETOa positive leukemia had no response to single agent PARP inhibitor. While a few mice benefitted from the addition of talazoparib to SOC, this result was not statistically significant (p= 0.42). Early response by bioluminescent imaging confirmed that mice with MLL-AF6 and NUP98-KDM5A driven leukemias who received talazoparib in combination with chemotherapy had the lowest leukemia burdens while the AML1-ETOa cohort did not benefit from the addition of this targeted agent. Interestingly, mice harboring CBAF2T3-GLIS2/JAK2 V617F were not responsive to PARP inhibitors, which was inconsistent with the CMS cell line that has same oncogenic fusion gene but lacks the JAK2 V617F mutation.Synergy experiments with ATM inhibitor AZD0156 demonstrated tremendous synergy with talazoparib in sensitive cell lines with almost no synergy in those that were resistant, suggesting that sensitive cell lines are unable to efficiently activate the HR pathway to repair double stranded breaks induced by PARP inhibition whereas resistant cells can overcome inhibition. To determine the HR response to DNA damage in our cell lines, we exposed them to 1uM topotecan for 2 hours and then measured γH2AX response at 0, 4 and 24 hours. γH2AX is a sensor of DNA damage and therefore increases with DNA damage and decreases with repair. PARP inhibitor sensitive cell lines had persistence of gamma H2AX at 24hrs while resistant cell lines had at least partial resolution of damage, confirming that PARP inhibitor sensitive cell lines have aberrant DNA damage response through HR.RNA sequencing of our cell lines revealed a correlation between Phosphatase and tensin homolog (PTEN) transcript levels and PARP sensitivity. Western blotting confirmed that PTEN was downregulated or absent in both cell lines and murine leukemias that were sensitive to PARP inhibitors. In contrast to the CMS cell line that carries the CBFA2T3-GLIS2 fusion, murine leukemias with CBAF2T3-GLIS2/JAK2 V617F had high levels of PTEN, supporting the hypothesis that sensitivity to PARP inhibitors is due to loss of PTEN.In conclusion, we report that a subset of pediatric AML with high- risk features are sensitive to PARP inhibition due to deficient DDR through HR. Downregulation of PTEN is a candidate biomarker of response to PARP inhibitors in these patients. This data illuminates a promising therapeutic vulnerability in a patient population where new targeted treatments are vital to improve outcomes. DisclosuresNo relevant conflicts of interest to declare.

Transcriptomic and Network Signatures of Cryptic Fusion CBFA2T3-GLIS2 in Pediatric AML Reveals Clinical Considerations and Implications

IntroductionA cryptic inv(16)(p13.3q24.3) encoding the CBFA2T3-GLIS2 fusion was initially reported in childhood acute megakaryoblastic leukemia (AMKL) and was shown to be associated with unfavorable outcomes. Although this fusion onco-protein is more prevalent in FAB M7 AML, it is reported in younger patients with all subtypes of AML. We present a comprehensive transcriptomic signature of an unselected cohort of children with CBFA2T3-GLIS2 fusion treated on COG AML trials.MethodsCBFA2T3-GLIS2 fusions were identified in 26 of 1042 (2.5%) patients enrolled in Children's Oncology Group (COG) studies AAML03P1 and AAML0531, with a prevalence of 8% in patients < 3 years of age. Whole transcriptome sequence data was available from 466 patients and 20 normal bone marrows (NBM) for expression profiling. Differentially expressed genes were identified using the Limma v.3.30.12 R package. Unsupervised hierarchal clustering was performed using Euclidean distance with the WPGMA linkage algorithm. Gene-set enrichment analysis was completed using the Broad Institute software v.2-2.2.4 with MSigDB C2 curated gene sets.ResultsOf 26 patients with CBFA2T3-GLIS2 fusions, 21 (81%) were < 5 years of age (Fig. 1A). KMT2A fusions co-occurred in only two patients, and none of the positive tumors carried core binding factor fusions. Few other variants accompanied CBFA2T3-GLIS2 fusions and overall, this subclass of AML had been categorized as cytogenetically normal (42% CN-AML, Fig. 1B). CBFA2T3-GLIS2 AML fared worse than the AML population as a whole; more than 75% of fusion-positive cases relapsed within two years (Fig. 1C, p = 0.001).Gene-set enrichment analysis between CBFA2T3-GLIS2 fusion and fusion-negative tumors revealed the top three positively enriched gene sets as 1) basal cell carcinoma (FDR = 0.003), 2) ECM receptor interactions (FDR = 0.006), and 3) hedgehog signaling pathways (FDR = 0.029). Consistent with previous reports, genes associated with WNT, hedgehog (HH), and BMP signaling were significantly up-regulated in CBFA2T3-GLIS2 tumors (Fig. 1D). CBFA2T3-GLIS2 fusions have been associated with induction of BMP signaling through downstream HH targets, BMP2 and BMP4 in pediatric AMKL. Moreover, concurrent dysregulation of WNT, BMP and HH signaling is a feature of CBFA2T3-GLIS2 AMKL (Gruber et al., 2015). Unsupervised hierarchical clustering of differentially expressed genes effectively clustered the fusion-positive samples, suggesting CBFA2T3-GLIS2 fusions drive a distinct transcriptional profile from other AML subtypes (Fig. 1E). CBFA2T3-GLIS2 tumors compared to NBM had upregulated expression of 4 cancer testes antigens (CTAs): CCNA1, PRAME, IGF2BP3, and KDM5B (fold-change > 2, adj. p < 0.001). CTAs are promising candidates for targeted T-cell therapies due their low expression in NBM and surrounding tissues, which could limit treatment related toxicity ( Ochsenreither et al., 2012) .We further performed unsupervised hierarchical clustering in 437 patients for whom flow cytometry data were available (Fig. 1F). Most fusion-positive tumors (7/10) clustered together by mean fluorescence intensity (MFI), indicating that CBFA2T3-GLIS2 tumors have highly distinctive immune profiles. Of particular interest, fusion-positive tumors expressed CD56 at high levels (median CD56 in CBFA2T3-GLIS2 : 1,505.0 MFI [range: 6.4 - 4,290.0] versus median CD56 in fusion-negative AML: 9.0 MFI [range: 2.4 - 3,360.0], p = 0.001). A novel immunophenotype in pediatric AML, termed RAM, has been associated with high CD56 expression and much lower event-free survival. A preponderance of CBFA2T3-GLIS2 AML displayed the RAM phenotype, with 70% of fusion-positive tumors identified as RAM compared to 0.23% (1/427) in the fusion-negative population (p < 0.001).ConclusionCBFA2T3-GLIS2 AML tends to have an aggressive disease pathology in much younger patients, who had been primarily classified as CN-AML. This suggests screening for CBFA2T3-GLIS2 in younger patients could be valuable as standard practice, though not yet recommended by the World Health Organization. Our associations with CBFA2T3-GLIS2 fusions provides a potential link between genomic and immunologic characteristics of a high-risk group of pediatric AMLs, which could be employed in the future for clinical screening through molecular or flow-cytometric methods. [Display omitted] DisclosuresBrodersen:Hematologics Inc: Employment. Loken:Hematologics Inc: Employment, Equity Ownership.

EP066 - Trisomy 3 and CBFA2T3-GLIS2 fusion in a child with acute megakaryoblastic leukemia without Down syndrome

eP066 - Trisomy 3 and CBFA2T3-GLIS2 fusion in a child with acute megakaryoblastic leukemia without Down syndrome

Clinical impact of genomic characterization of 15 patients with acute megakaryoblastic leukemia-related malignancies.

Acute megakaryoblastic leukemia (AMKL) is a rare subtype of acute myeloid leukemia but is approximately 500 times more likely to develop in children with Down syndrome (DS) through transformation of transient abnormal myelopoiesis (TAM). This study investigates the clinical significance of genomic heterogeneity of AMKL in children with and without DS and in children with TAM. Genomic evaluation of nine patients with DS-related TAM or AMKL, and six patients with non-DS AMKL, included conventional cytogenetics and a comprehensive next-generation sequencing panel for single-nucleotide variants/indels and copy-number variants in 118 genes and fusions involving 110 genes. Recurrent gene fusions were found in all patients with non-DS, including two individuals with complex genomes and either a NUP98–KDM5A or a KMT2A–MLLT6 fusion, and the remaining harbored a CBFA2T3–GLIS2 fusion, which arose from both typical and atypical cytogenetic mechanisms. These fusions guided treatment protocols and resulted in a change in diagnosis in two patients. The nine patients with DS had constitutional trisomy 21 and somatic GATA1 mutations, and those with DS-AMKL had two to four additional clinically significant somatic mutations. Comprehensive genomic characterization provides critical information for diagnosis, risk stratification, and treatment decisions for patients with AMKL. Continued genetic and clinical characterization of these rare cancers will aid in improving patient management.