ALBERT

Description: Bortezomib (BTZ) was recently evaluated in a randomized Phase 3 clinical trial which compared standard chemotherapy (cytarabine, daunorubicin, etoposide; ADE) to standard therapy with BTZ (ADEB) for de novo pediatric acute myeloid leukemia. While the study concluded that BTZ did not improve outcome overall, we examined patient subgroups benefitting from BTZ-containing chemotherapy using proteomic analyses. The proteasome inhibitor BTZ disrupts protein homeostasis and activates cytoprotective heat shock responses. We measured total heat shock factor 1 (HSF1) and phosphorylated HSF1 (HSF1-pSer326) in leukemic cells from 483 pediatric patients using Reverse Phase Protein Arrays. HSF1-pSer326 phosphorylation was significantly lower in pediatric AML compared to CD34+ non-malignant cells. We identified a strong correlation between HSF1-pSer326 expression and BTZ sensitivity. BTZ significantly improved outcome of patients with low-HSF1-pSer326 with a 5-year event-free survival of 44% (ADE) vs. 67% for low-HSF1-pSer326 treated with ADEB (P=0.019). To determine the effect of HSF1 expression on BTZ potency in vitro, cell viability with HSF1 gene variants that mimicked phosphorylated (S326A) and non-phosphorylated (S326E) HSF1-pSer326 were examined. Those with increased HSF1 phosphorylation showed clear resistance to BTZ vs. those with wild type or reduced HSF1-phosphorylation. We hypothesize that HSF1-pSer326 expression could identify patients that benefit from BTZ-containing chemotherapy.

Description: CD123 is a cell surface protein expressed on hematopoietic progenitors and the surface of most AML blasts, making it a valuable therapeutic target for clinical intervention. As such, antibody-drug conjugates or CAR T cells against this antigen have been developed including tagraxofusp-erzs, recently approved for blastic plasmacytoid dendritic cell neoplasm (BPDCN). CD123 is the alpha subunit of the interleukin 3 receptor and is encoded by the pseudoautosomal IL3RA gene. Recent work demonstrated that different monoclonal antibodies directed against CD123 show sizable discrepancies when used to quantify this antigen on AML patient samples. (Cruz et al. 2018) Given these results and the variability in patient response to anti-CD123 therapeutics, we hypothesized that heterogeneity in IL3RA mRNA isoform expression may induce epitope variation on the cell surface, modulating antibody and therapeutic response. To better understand the heterogeneity, we analyzed long and short read transcriptomics data from normal bone marrow along with pediatric AML samples known to harbor translocations. The combination of these two types of RNA expression data afford both a look at full length isoforms produced in patients and the relative expression levels of each. To define the isoforms expressed in pediatric AML, we augmented short read RNAseq with long read transcriptomics on the PacBio platform. Following up on short RNAseq data generated from 4 clinical study cohorts of pediatric AML samples (N = 1,394) collected and normal bone marrow controls (NBM, N = 68), we chose diagnostic AML samples (N=10) and one NBM with high RNA integrity (RIN 〉9) for polyA transcript profiling using Pacific Biosciences (PacBio) long read RNA sequencing. This method gives full isoform sequences that can be reliably translated into open reading frames. It also adds new utility to our wealth of short read RNA-seq as the long read data can be used in a reference fashion to quantify and compare isoforms across cohorts. After profiling and classifying the novel isoforms, we honed in on transcripts from the IL3RA locus since these encode the CD123 antigen targeted by immunotherapy approaches. PacBio long read RNA sequencing detected 8 unique full-length transcript isoforms that mapped to the IL3RA gene: 4 known and 4 novel IL3RA transcripts. Three abundant known isoforms aligned to the canonical annotated IL3RA (Isoform 1, Figure 1A), an isoform missing exons 3 and 4 (Isoform 2) or a third isoform (Isoform 3, not shown) which does not encode a transmembrane domain. We focused on 3 novel isoforms (Figure 1, Isoforms A-C) encompassing a variety of splicing changes, but all of which are predicted to harbor a transmembrane domain and dramatically alter the extracellular peptide sequence in comparison to annotated isoforms. (Figure 1, domains predicted and colored in the legend) The novel isoforms were found independently in multiple patients, but as additional validation we PCR amplified cDNA from patient samples using an inclusive primer set directed to constitutive exons that flank the alternative splicing events and thus designed to capture multiple isoforms. (Figure 2A, arrows) Products were separated by gel electrophoresis with amplicons cloned, Sanger sequenced and analyzed through alignment with human reference sequences. The non-specific isoform amplification detects multiple isoforms indicating heterogeneity in splice site choice between patients. Fragment analysis from patient 2 (Figure 2B) confirms the presence of isoform variation with peaks corresponding with the expected products from isoforms 1, 2, A, B, and C. In an effort to further validate and quantify novel isoforms of IL3RA, we employed kallisto which utilizes short read RNAseq data from the entire cohort to get a count estimate for each isoform in pediatric AML patient samples and normal controls. These data (Figure 3) indicate that while the annotated isoform 1 is the most abundant, a wide range of novel isoform expression is detected in both normal and pAML samples. In conclusion, changes in protein length and peptide sequence may affect the efficacy of therapeutic anti-CD123 approaches since some patients express alternative isoforms with a wide range of abundance. We anticipate that the computational and experimental pipeline used to discover and characterize these isoforms will be of high value in the study of many cell surface antigens with therapeutic potential. Disclosures Underwood: Pacific Biosciences: Employment, Equity Ownership. Tseng:Pacific Biosciences: Employment, Equity Ownership. Farrar:Novartis: Research Funding.

Description: Infants with acute leukemia present special challenges, due in part to the unusual features of their disease. A clonal oncofusion protein is detectable in the vast majority of infants presenting with acute leukemia, whether myeloid (AML), lymphoblastic (ALL), or mixed-phenotype (MPAL). Many recurrent oncofusion proteins seen in this age group involve the KMT2A (formerly MLL, Mixed Lineage Leukemia) gene, located on chromosomal band 11q23, which is notorious for both lineage switching and poor clinical prognosis. Other fusions involve components of the nuclear pore complex (NUP98), chromatin modifiers (KAT6A), or lineage-associated pioneer factors (such as MYB and GLIS2). This diversity of oncofusion proteins has hampered efforts to develop targeted agents, and over half of infant cases eventually succumb to their disease. Previous work showed that infant ALL is distinct from childhood or adult ALL. Here we show that infant AML is similarly distinct from AML in other age groups, and this difference yields transformative opportunities to repurpose existing targeted and immune therapies. mRNA sequencing (RNAseq) analysis of over 1500 participants in Children's Oncology Group clinical trials revealed hundreds of infant-AML-specific transcripts, many associated with early B cell development. 18 of the affected genes form a canonical B-cell regulatory circuit, including the epigenetic regulators BRD4 and POU2AF1, and their onco-fetal targets LIN28B and IGF2BP3. These four genes also host hypo-methylated super-enhancers in most infant AML cases, but in few older patients. The MYC proto-oncogene is directly implicated in a regulatory loop involving microRNA let7a-2, which is expressed in a mutually exclusive manner with its target and regulator LIN28B (Fig 1A). By contrast, the WT1 gene, abundantly expressed in over 90% of AML cases from other age groups, is transcribed at low levels in most infant AML cases. Single-sample gene set enrichment analysis revealed that most cases of infant AML bear more similarity to infant B-cell ALL or MPAL than to AML in older children, adolescents, and adults (Fig 1B). Higher-order chromatin conformation further emphasizes these differences. The MYC proto-oncogene, a critical node in the gene regulatory network (Fig. 1A) observed across infant acute leukemia cases, is routinely amplified via trisomy 8 in this age group (and in fact all childhood leukemia). The IGF2BP3 gene, another critical node (Fig. 1A), is rarely deleted in infants (deletions of 7q tend to spare 7p15, and complex karyotype cases often retain two or more copies on derivative chromosomes). The immature humoral and adaptive immune system in infants complicates treatment in this age group, but also yields metabolic vulnerabilities which can be exploited to break the nearly ubiquitous positive feedback loop described herein. These age-specific differences suggest that alternative therapeutic targets and modalities are warranted not only in infant AML, but across infant acute leukemia more generally. The relatively small numbers of infant acute leukemia cases by lineage, along with common mechanisms that appear to sustain most cases of infant acute leukemia, and frequent clinical reports of lineage switching in infants regardless of morphology at presentation, all support this contention. Our results indicate that combination therapies coupling experimental ages (such as DFMO, which targets metabolic regulation of MYC, LIN28B, and let-7a via ornithine decarboxylase, and has shown substantial efficacy in MYCN-amplified neuroblastoma) with already approved agents such as BET inhibitors (which target BRD4) and blinatumomab (targeting CD19 as a canonical B-cell antigen) may be warranted. These immune-targeting combinations represent promising and potentially lineage-agnostic approach to improving outcomes for infants with acute leukemia, a group of patients for whom novel therapeutic strategies are desperately needed, and for whom less toxic, more effective, targeted combination therapies may spare a lifetime of late effects. Disclosures Pardo: Hematologics, Inc: Employment. Kaeding:Celgene: Employment. Loken:Hematologics, Inc: Employment, Equity Ownership. Farrar:Novartis: Research Funding.

Description: Introduction : Myeloproliferative neoplasms (MPNs) are rare clonal bone marrow disorders in children characterized by high blood counts, predisposition to clotting events, and the potential to transform to myelofibrosis or acute myeloid leukemia (AML). Children with MPNs have lower rates of the known driver mutations (in JAK2, MPL, and CALR) than adult patients, and the underlying pathways and molecular derangements in young patients remain unknown. Given the lack of knowledge about pediatric MPNs, it is critical that we gain a better understanding of the dysregulated pathways in these diseases, which is necessary for improving disease understanding and broadening treatment options in children. Therefore, the objective of this work was to identify differentially expressed genes and pathways between children with MPNs and healthy controls, as well as children with AML, to guide further study. Methods : Mononuclear cells were extracted from peripheral blood of pediatric MPN patients (n=20) and pediatric and young adult AML patients (n=1410), and bone marrow of normal controls (NC, n=68). AML patient samples were being evaluated as part of a Children's Oncology Group planned analysis. To identify an expression profile unique to MPNs, transcriptome data from MPN patients was contrasted against NC and AML patients. All samples were ribodepleted and underwent Illumina RNA-Seq to generate transcriptome expression data. All analyses were performed in R. Differentially expressed genes were identified using the voom function from the limma package (v. 3.38.3), and enriched pathways were identified using the pathfindR package (v. 1.3.1). Unsupervised hierarchical clustering and heatmap generation was performed using the ComplexHeatmap package (v. 1.20.0). Results : MPN patient samples showed a unique expression signature, distinct from both AML patients and normal controls. Unsupervised PCA plot (Figure 1A) and heatmaps (Figure 1B) show that MPN samples cluster together. There were 4,012 differentially expressed (DE) genes in MPNs compared to NC and 6,743 DE genes in MPNs compared to AML patients. There were 2,493 shared genes between the 2 groups (Figure 1C.) Significantly DE genes between MPNs and other groups included multiple platelet-relevant genes including PF4 (CXCL4), PF4V1, P2RY12, and PPBP (CXCL7). Interestingly, PF4V1 was the most DE gene in MPNs compared to AML, and third highest versus NC. Dysregulation of some of these genes has been seen in adult MPNs, as well as thrombosis. Further comparison of transcriptome profiles between children with (n=13) and without (n=7)JAK2 mutations showed upregulation of three genes, CFB, C2, and SERPING1, which are all known complement genes, implicating complement activation in JAK2-mutated MPN patients. Complement activation has previously been reported in adult MPNs. Pathway enrichment analysis shows a number of immune and inflammatory pathways as enriched in MPN patients compared to both AML and NC. There were 179 enriched pathways in MPNs compared to AML and 142 compared to NC, with 134 common pathways (Figure 1D.) The systemic lupus erythematosus pathway was the most heavily enriched pathway in MPNs compared to both AML and NC. Additional pathways with significant enrichment include hematopoietic cell lineage, cytokine-cytokine interactions, DNA replication, and various infection-relevant pathways. The JAK-STAT signaling pathway was also enriched in MPNs compared to both AML and NC, as was the platelet activation pathway. Conclusion: Transcriptome evaluation of childhood MPNs shows enrichment of numerous inflammatory and immune pathways, highlighting that, as in adult MPNs, inflammation is implicated in pediatric MPNs. Furthermore, specific complement genes were upregulated in JAK2-mutant MPN. Upregulation of platelet-specific genes implies potential insights into disease mechanisms and warrants more study. Variations in the cell populations may account for some of the differences seen, however all samples were largely mononuclear cells, making their comparisons reasonable. Further analysis of this early data is needed to better assess inflammatory changes and platelet activation in pediatric MPNs, as are larger sample sizes. Individual cells may have differential expression of various genes, and future experiments with single-cell RNA-seq would be helpful to further elucidate differences. Disclosures Levine: Novartis: Consultancy; Loxo: Membership on an entity's Board of Directors or advisory committees; Celgene: Consultancy, Research Funding; Gilead: Consultancy; Roche: Consultancy, Research Funding; Lilly: Honoraria; Amgen: Honoraria; Qiagen: Membership on an entity's Board of Directors or advisory committees; Imago Biosciences: Membership on an entity's Board of Directors or advisory committees; C4 Therapeutics: Membership on an entity's Board of Directors or advisory committees; Prelude Therapeutics: Research Funding; Isoplexis: Membership on an entity's Board of Directors or advisory committees.

Description: E-selectin (E-sel) is a cell adhesion glycoprotein that is expressed on endothelial cells and has been implicated in therapeutic resistance. In most myeloid leukemias, leukemic blasts express E-sel ligands (EsL), which contain the glycan epitope of the carbohydrate sialyl Lex (sLex). This expression increases the likelihood of adhesion to vascular endothelial cells and facilitates sequestration in the bone marrow vascular niche, leading to cell adhesion-mediated drug resistance and poor clinical outcome. E-sel antagonists like uproleselan, interrupts leukemic cell homing to the vascular niche, increases susceptibility to cytotoxic and targeted therapies and can be potent adjuncts to therapeutics. Recent data demonstrated a correlation between leukemic cell surface levels of EsL and response to uproleselan, linking EsL expression to response. We questioned whether transcriptome profiling of EsL-forming glycosylation genes can be used to identify elevated EsL expression in patients with acute myeloid leukemia (AML), and subsequently which patients might best respond to uproleselan. RNA-seq data from patients treated in COG AAML1031 (N = 1,074) was available for evaluation. We examined transcriptome expression of 24 genes that code for enzymes involved in glycosylation of EsL. All analyses were performed in R. Cox proportional hazards models were generated using the survival package. Multidimensional flow cytometry (MDF) was used to detect cell surface EsL expression by two techniques: direct binding of an E-sel/hIg, PE labeled chimera, and the anti-sLex antibody HECA-452. Seven of the 24 genes examined had minimal expression (mean

Description: The MLLT10 gene, a known fusion partner for KMT2A, encodes AF10 protein, a transcription factor that binds unmodified histone H3 and regulates DOT1L expression. KMT2A-MLLT10 fusion portends adverse outcome, but MLLT10 function and prognostic implications in partnership with other genes has not been defined. In comprehensive transcriptome and karyotype evaluation of 2226 children and young adults (0-30 years), we defined the full spectrum of MLLT10 fusions, identified new fusion partners, and correlated MLLT10 structural variants with clinical outcome. We also evaluated transcription and methylation profiles to identify genes dysregulated in MLLT10 fusions with and without KMT2A. 2226 patients treated on Children's Oncology Group (COG) trials AAML0531 and AAML1031 were evaluated by transcriptome profiling and/or karyotyping to identify leukemia associated fusions and copy number changes associated with prognosis. Collectively, 127 patients (5.7%) had primary fusions involving MLLT10: 104 (82%) involving KMT2A (KMT2A-MLLT10), and 23 patients (18%) revealed other fusion partners (MLLT10-X). Alternate, recurrent fusion partners included PICALM (n=13), DDX3X (n=2), and TEC (n=2), while fusions with 6 other partner genes (DDX3Y, CEP164, NAP1L1, SCN2B, TREH, and XPO1) were each identified in single patients. Given the known association of KMT2A-MLLT10 fusions with adverse outcome, we sought to determine whether MLLT10-X had distinct characteristics and comparable outcomes. Initial comparison of disease characteristics in patients with and without KMT2A as fusion partner showed significant differences in age at diagnosis. Those with KMT2A-MLLT10 had a median age of 1.7 years (range 0-21.3), compared to 12.7 years (range 1.4-18.9) in those with MLLT10-X (p ≤ 0.001). There was no significant difference in gender, race, mutational status, or white blood cell count between these two cohorts. MLLT10 rearranged patients (n=127) demonstrated adverse outcomes, with 5-year event-free survival (EFS) of 18.6% vs. 49% in non-MLLT10 rearranged patients (N=1953, p6 logFC, or over 400x higher on average in MLLT10 rearranged patients. To determine if patients with MLLT10 fusions had distinct epigenetic profiles, we performed differential methylation analyses on samples from normal bone marrow and patients with 4 high-risk molecular features: MLLT10 rearranged, KMT2A rearranged, NUP98-NSD1 fused, and FLT3-ITD, across nearly 1 million CpG sites on the Infinium EPIC array (Illumina, CA). After fitting a multivariate model with all of the interacting molecular features, the 250 most discriminative regions were extracted and plotted (ComplexHeatmap) (Fig 1D). Strikingly, patients with MLLT10-X fusions cluster discretely with ultra-high-risk NUP98-NSD1 fusion patients, showing a broadly hypermethylated profile, while KMT2A-MLLT10 patients cluster within the larger KMT2A category and show far fewer hypermethylated regions. We identified patients with MLLT10 fusion partners not previously described, and compared them to other AML patients, as well as patients with known MLLT10 partners KMT2A and PICALM. All MLLT10-aberrant cases had poor EFS and OS, high RR, overexpressed HOXA genes, and distinct DNA methylation profiles, while patients with MLLT10-X fusions tend to be older children. Regardless of fusion partner, patients with MLLT10 fusions exhibit very high risk, and should be prioritized for alternative therapeutic intervention. Disclosures Farrar: Novartis: Research Funding. Deshpande:A2A Pharmaceuticals: Consultancy; Salgomed Therapeutics: Consultancy.

Description: Childhood AML is a heterogeneous hematologic disease with a multitude of subtypes characterized by varying morphology, structural alterations, and recurrent mutations. Such heterogeneity and staggering number of genomic and transcriptional alterations has precluded appropriate risk classification. We investigated the expression of long non-coding RNAs (lncRNA) in childhood AML and explored its potential utility for more precise risk characterization at diagnosis. We inquired whether lncRNAs aberrantly expressed in AML compared to healthy normal bone marrows may be utilized for predicting outcome without knowledge of somatic variants, creating a variant-agnostic prognostic indicator. Ribodepleted RNA-sequencing data from normal bone marrows (N=68), as well as diagnostic primary samples (N=1300) from patients with detailed clinical annotations and outcome were included for study. To this end, the study population was divided into training (N=781), test (N=261), and validation (N=258) cohorts using blocked randomization. Upregulated lncRNAs compared to normal marrows in the training set (fold-change 〉 2, FDR 〈 0.05, Fig. 1A) were input for a LASSO cox proportional hazards regression which identified a set of 37 lncRNAs whose expression (log2 scale, TMM normalized) associated with patient outcome. The trained model coefficients were applied to the test and validation cohorts to produce a weighted sum of the 37 lncRNAs expression per patient (lncScores, range: -1.44 to +1.41). The distribution of lncScores revealed approximately equal numbers of patients with positive and negative scores in the training, test, and validation cohorts (Fig. 1B). In the training set, those with positive lncScores had an overall survival (OS) of 42.7% at 5-years from diagnosis compared to 76.3% for those with negative scores (hazard ratio (HR) = 3.15, p 〈 0.001). Positive lncScores were also associated with adverse event-free survival (EFS, HR = 2.47, p 〈 0.001). LncScore based outcome analysis in the independent test and validation cohorts showed nearly identical outcome results for OS (HR = 2.86 and 2.99 resp., p 〈 0.001) and EFS (HR = 2.37 and 2.35, p 〈 0.001, Fig. 1C) as was seen in the training set, demonstrating stability of the predictive power of the lncScore across independent cohorts. Next, the lncScore's performance was evaluated in relation to cyto/molecular risk groups (CMrisk: high, standard, and low) defined by karyotype, clinically relevant mutations, and cryptic fusions as presented previously (Cooper et al. ASH 2017). A multivariable cox regression model (N = 1300) that included lncScore, CMrisk, and white blood cell count revealed that both lncScore and CMrisk groups remained independent prognostic factors for OS (p 〈 0.001) and EFS (p ≤ 0.001), suggesting the lncRNA signature may provide additional prognostic information to that defined by CMrisk status. Application of the lncScore to the CMrisk groups demonstrated that 173 of the 541 of patients classified as CMrisk high (32%) would be reallocated to the standard risk arm based on negative lncScores (OS 57.3% vs 31.8%, p 〈 0.001). Similarly, 40% of patients with CMrisk standard (132/327) would be reclassified as low risk by the lncScore system (OS 73.8% vs 52.4%, p 〈 0.001). lncScores did not revise risk status in CMrisk low patients. The ribbon plot (Fig. 1E) demonstrates reallocation of patients using the integrated risk stratification system. Integration of the CMrisk and lncScore demonstrated that in combination the two risk classification systems can provide a more precise assessment of risk status, since both CMrisk high and CMrisk standard cohorts can be further stratified by the lncScore (Fig. 1D). The integrated approach provided a comprehensive risk classifier that incorporates both cytogenetic data, as well as transcriptional evidence, which benefits more than 1 out of 5 patients (23%, 305/1300) with a more accurate determination of risk at the time of diagnosis (Fig. 1F). This study demonstrates development and validation (in two independent cohorts) of a lncRNA based, variant-independent prognostic signature (p 〈 0.001) that can effectively partition patients into high and low risk groups at diagnosis. It can also significantly enhance the predictive power of conventional cyto/molecular markers for more precise prediction of outcome prior to the start of therapy. Disclosures Farrar: Novartis: Research Funding.

Description: Monosomy 7 (mono7) alterations in acute myeloid leukemia (AML) are associated with poor outcome and disease progression. Through advancements in multi-omic approaches, more specific treatment strategies may be available for high-risk cohorts. Here we describe pediatric cases of AML with mono7, co-occurring fusions, associated outcome, and potential therapeutic treatments. Of the 2200 patients treated in 3 consecutive Children's Oncology Group protocols (AAML03P1, AAML0531, and AAML1031), 45 patients (2%) had karyotypic evidence of mono7 with full complement of clinical data for analysis. RNA sequencing was available for 37 and whole genome sequencing (WGS) for 8 cases. Fusions were identified using TransAbyss, STARfusion, and Cicero algorithms, while structural variants were analyzed by CREST in the WGS samples. Differential expression comparing mono7 AML (N=28) vs. other AML (N=1064) was performed and epigenetic profiling was evaluated using Illumina's EPIC array (N=1025, N=79 normal bone marrow controls). Of the 45 patients with mono7, 22 had additional karyotypic alterations including copy number variants in 7 and translocations in 15 (6 had confirmatory evidence by RNA seq). RNA seq also identified 9 additional cryptic fusions. In total, the cohort contains 5 cases (11.1%) with KMT2A fusions, 3 (6.7%) with CBF fusions, 7 (15.7%) with 3q26 alterations, and 7 (15%) with copy number alterations (Fig. 1A). In 14 patients (31%), mono7 was the sole karyotypic alteration. In 28 patients with ribo-depleted RNA seq data, cryptic fusions involving the ALK gene were identified in 4 patients (14.3%) with ALK fused to either SPTBN1 (n=3) or RANBP2 (n=1) genes. Cryptic ALK fusions were not detected in the 1064 cases of other AML (p=1.5x10-37), suggesting a unique association between ALK fusions and mono7 and potential genomic cooperation. A differential expression analysis compared patients with mono7 (N=28) to all other AML (N=1064). This analysis identified 1547 dysregulated genes. Of these, MECOM (MDS1/EVI1 COMplex) was identified as the top upregulated gene (logFC 7.24; p=4.4x10-74) with a median expression 5.45 TPM (range 0-89.2 TPM) in patients with mono7 vs. 0.013 TPM in other AML patients. Evaluation of outcome based on MECOM expression demonstrated that patients with high MECOM expression (greater than median) had a 3-yearOS of 22%±20% compared to that of 68%±20% with low MECOM expression, (p=0.026, Fig. 1B) All patients with 3q26 variants had elevated MECOM expression (n=7). In addition, 11 patients without 3q26 alterations/MECOM fusions had MECOM overexpression. Given lack of underlying genomic etiology, we sought to determine whether epigenetic factors might mediate MECOM expression. A panel of 6 CpGs was sufficient to distinguish hematopoietic stem cells (HSCs) from granulocyte-monocyte progenitors (GMPs). HSC-like hypermethylation of these CpGs was strongly associated with high MECOM expression. Further, there was high correlation between total MECOM expression and the methylation status of a CpG island proximal to the short EVI1 transcript variant of MECOM (Spearman's rho = -0.51, p 〈 0.001, Fig. 1C), suggesting a regulatory underpinning for permissive expression of EVI1. The stem-like epigenetic signature and concordant high MECOM expression of poor-prognosis mono7 AML are consistent with a "stemness" signature and portend poor survival (Fig. 1D). Stemness mRNA signatures have been implicated in high-risk pediatric AML (Smith JL, ASH 2017) and the epigenetic signatures of these genes further corroborate cell of origin as an independently prognostic factor even within high-risk AML. Focused interrogation of the MECOM locus thru integration of the RNA Seq and WGS identified allele specific MECOM expression, adding additional potential mechanism of modulation to MECOM expression. Structural variant analysis confirmed no other CNVs on chromosome 3, indicating cis dysregulation of MECOM. Here, by interrogation of the genome, phenome, transcriptome, and epigenome of mono7 AML a substantial phenotypic and prognostic heterogeneity exists defining a cohort of patients, regulated by genomic and epigenomic alterations. Further, discovery of cryptic ALK fusions in mono7 present a target for ALK inhibitors (FDA approved for non-small cell lung cancer) in this high-risk population. Disclosures No relevant conflicts of interest to declare.

Description: Childhood acute myeloid leukemia (AML) is an aggressive myeloid malignancy characterized by mutational and cytogenetic abnormalities. Chromosomal rearrangements involving the NUP98 gene have come to light for its' significant impacts on outcome and response to treatment. NUP98-rearranged (NUP98-R) AML includes NUP98-NSD1, NUP98-KDM5A, and various less common NUP98 fusion partners - such as HOX, SET, and Bromodomain genes. NUP98-R account for 6.8% of patients across Children's Oncology Group studies CCG2961, AAML03P1, AAML0531, and AAML1031. The majority are NUP98-NSD1 (4.6%), then NUP98-KDM5A (1.4%), and the various partners, termed NUP98-X (0.83%). However, the biological implications of NUP98-X have yet to be investigated. We define transcriptional clusters and report transcriptional profiling results to reveal similarities and differences between the diverse NUP98-R fusions. RNA sequencing was completed for 1,492 pediatric AML patients and 68 healthy bone marrow controls (NBM). The algorithms STARfusion, TransAbyss, and CICERO detected 156 NUP98-R from RNA seq: NUP98-NSD1 (N=104), NUP98-KDM5A (N=32), and NUP98-X (N=20). NUP98-X encompassed 13 unique fusion partners and 45% (9/20) had a homeobox gene partner - HOXA9 (N=4), HOXD13 (N=3), HOXA13 (N=1), and PRRX1 (N=1). Unsupervised clustering via uniform manifold approximation and projection (UMAP) with input genes selected by jackstraw PCA revealed NUP98-X cluster more closely with NUP98-NSD1, but their trajectory is dispersed within NUP98-NSD1 and NUP98-KDM5A, indicating a potential hybrid transcriptional profile (Fig 1 A). NUP98-NSD1 clustered remotely from the majority of NUP98-KDM5A, highlighting their unique transcriptional profiles despite sharing NUP98-R. Five transcriptional clusters were identified by the Leiden algorithm, followed by selection of genes significantly associated with cluster assignment. Cluster marker genes were filtered to be unique for each cluster prior gene-set enrichment analysis (Fig 1 B). Cluster C3 encompassed the largest proportion of NUP98-X, with 7 homeobox and both PHF23 cases (9/20, 45%), but also comprised of 19 NUP98-NSD1 and 2 NUP98-KDM5A. Importantly, C3 was highly associated with the expression of long non-coding RNAs, Z69666.2, MT1XP1, and RP11-455O6.2 (≥ 37% specificity, FDR 〈 0.001), and pathways in enriched in oncogenic microRNAs (FDR 〈 0.001), suggesting a non-coding signature may define this group. NUP98-NSD1 divided into two primary clusters, C1 (52/104) and C2 (31/104), differentiated by the expression of mannose receptor genes MCR1/ MRC1L1, which detect products of the lysosome pathway, significantly activated in C1 (FDR 〈 0.001). NUP98-KDM5A also segregated into two clusters, C4 (N=22/32) and C5 (N=7/32) based on an FAB M7/AMKL signature, with 78.6% of M7 NUP98-R located in C4 (p=0.003). Whereas, C5 cases were non-M7 and uniquely enriched in the regulation of stem cells pathway (FDR = 0.005); activation was driven by the expression of POU5F1, BMP4, WNT7, and WNT10A/B. Transcriptional profiling of NUP98-R cohorts, independently compared to other AML (N=1,336), found 27 upregulated differentially expressed genes (DEGs) shared between NUP98-X and NUP98-NSD1, including MYCN oncogene, homeobox PBX3, and DNA methyltransferase DNMT3B; on the other hand, shared overexpression of MLLT3, homeobox IRX3, and CD79a characterized similarities between NUP98-X and NUP98-KDM5A (Fig 1C) and may contribute to the hybrid expression profile observed by UMAP clustering. There were 38 DEGs shared between all NUP98-R cohorts. Differential expression analysis comparing NUP98-R cohorts vs NBM (N=68) showed that 22/38 shared DEGs were dysregulated in normal hematopoiesis and 15/38 genes were concordantly overexpressed in NUP98-R. The minimal set of 15 genes strongly implicates dysregulation at the HOX locus; these targets include hsa-mir-10a, whose genomic locus is within the HOXB cluster, CACNG4 located on chr17q, the same chromosome arm as HOXB, and the remaining shared targets were HOXA/B genes (Fig 1D). In summary, we found that NUP98-X are dispersed between NUP98-NSD1 and NUP98-KDM5A by transcriptional clustering. Clustering revealed significant pathways and marker genes that may contribute to segregation of NUP98-R cohorts, and DEGs that contribute in part to their hybrid transcriptional profile, including HOX genes, PBX3, and IRX3. Disclosures No relevant conflicts of interest to declare.