Search

Morphologic interpretation is the standard in diagnosing myelodysplastic syndrome (MDS), but it has limitations, such as varying reliability in pathologic evaluation and lack of integration with genetic data. Somatic events shape morphologic features, but the complexity of morphologic and genetic changes makes clear associations challenging. This article interrogates novel clinical subtypes of MDS using a machine-learning technique devised to identify patterns of cooccurrence among morphologic features and genomic events. We sequenced 1079 MDS patients and analyzed bone marrow morphologic alterations and other clinical features. A total of 1929 somatic mutations were identified. Five distinct morphologic profiles with unique clinical characteristics were defined. Seventy-seven percent of higher-risk patients clustered in profile 1. All lower-risk (LR) patients clustered into the remaining 4 profiles: profile 2 was characterized by pancytopenia, profile 3 by monocytosis, profile 4 by elevated megakaryocytes, and profile 5 by erythroid dysplasia. These profiles could also separate patients with different prognoses. LR MDS patients were classified into 8 genetic signatures (eg, signature A had TET2 mutations, signature B had both TET2 and SRSF2 mutations, and signature G had SF3B1 mutations), demonstrating association with specific morphologic profiles. Six morphologic profiles/genetic signature associations were confirmed in a separate analysis of an independent cohort. Our study demonstrates that nonrandom or even pathognomonic relationships between morphology and genotype to define clinical features can be identified. This is the first comprehensive implementation of machine-learning algorithms to elucidate potential intrinsic interdependencies among genetic lesions, morphologies, and clinical prognostic in attributes of MDS.

Key Points 5-hydroxymethylcytosine and TET2 messenger RNA (mRNA) downregulation are common in myelodysplastic syndromes irrespective of TET2 mutations.TET3 mRNA expression levels are associated with distinct clinical outcomes in myelodysplastic syndromes with and without TET2 mutations., Decrease in DNA dioxygenase activity generated by TET2 gene family is crucial in myelodysplastic syndromes (MDS). The general downregulation of 5-hydroxymethylcytosine (5-hmC) argues for a role of DNA demethylation in MDS beyond TET2 mutations, which albeit frequent, do not convey any prognostic significance. We investigated TETs expression to identify factors which can modulate the impact of mutations and thus 5-hmC levels on clinical phenotypes and prognosis of MDS patients. DNA/RNA-sequencing and 5-hmC data were collected from 1665 patients with MDS and 91 controls. Irrespective of mutations, a significant fraction of MDS patients exhibited lower TET2 expression, whereas 5-hmC levels were not uniformly decreased. In searching for factors explaining compensatory mechanisms, we discovered that TET3 was upregulated in MDS and inversely correlated with TET2 expression in wild-type cases. Although TET2 was reduced across all age groups, TET3 levels were increased in a likely feedback mechanism induced by TET2 dysfunction. This inverse relationship of TET2 and TET3 expression also corresponded to the expression of L-2-hydroxyglutarate dehydrogenase, involved in agonist/antagonist substrate metabolism. Importantly, elevated TET3 levels influenced the clinical phenotype of TET2 deficiency whereby the lack of compensation by TET3 (low TET3 expression) was associated with poor outcomes of TET2 mutant carriers.

Idiopathic aplastic anemia (IAA) is a rare autoimmune bone marrow failure (BMF) disorder initiated by a human leukocyte antigen (HLA)-restricted T-cell response to unknown antigens. As in other autoimmune disorders, the predilection for certain HLA profiles seems to represent an etiologic factor; however, the structure-function patterns involved in the self-presentation in this disease remain unclear. Herein, we analyzed the molecular landscape of HLA complexes of a cohort of 300 IAA patients and almost 3000 healthy and disease controls by deeply dissecting their genotypic configurations, functional divergence, self-antigen binding capabilities, and T-cell receptor (TCR) repertoire specificities. Specifically, analysis of the evolutionary divergence of HLA genotypes (HED) showed that IAA patients carried class II HLA molecules whose antigen-binding sites were characterized by a high level of structural homology, only partially explained by specific risk allele profiles. This pattern implies reduced HLA binding capabilities, confirmed by binding analysis of hematopoietic stem cell (HSC)-derived self-peptides. IAA phenotype was associated with the enrichment in a few amino acids at specific positions within the peptide-binding groove of DRB1 molecules, affecting the interface HLA-antigen-TCR β and potentially constituting the basis of T-cell dysfunction and autoreactivity. When analyzing associations with clinical outcomes, low HED was associated with risk of malignant progression and worse survival, underlying reduced tumor surveillance in clearing potential neoantigens derived from mechanisms of clonal hematopoiesis. Our data shed light on the immunogenetic risk associated with IAA etiology and clonal evolution and on general pathophysiological mechanisms potentially involved in other autoimmune disorders.

TCR repertoire diversification constitutes a foundation for successful immune reconstitution after allogeneic hematopoietic cell transplantation (allo-HCT). Deep TCR Vβ sequencing of 135 serial specimens from a cohort of 35 allo-HCT recipients/donors was performed to dissect posttransplant TCR architecture and dynamics. Paired analysis of clonotypic repertoires showed a minimal overlap with donor expansions. Rarefied and hyperexpanded clonotypic patterns were hallmarks of T cell reconstitution and influenced clinical outcomes. Donor and pretransplant TCR diversity as well as divergence of class I human leukocyte antigen genotypes were major predictors of recipient TCR repertoire recovery. Complementary determining region 3-based specificity spectrum analysis indicated a predominant expansion of pathogen- and tumor-associated clonotypes in the late post-allo-HCT phase, while autoreactive clones were more expanded in the case of graft-versus-host disease occurrence. These findings shed light on post-allo-HCT adaptive immune reconstitution processes and possibly help in tracking alloreactive responses.

Idiopathic aplastic anemia (IAA) is a rare autoimmune bone marrow failure disorder initiated by a human leukocyte antigen (HLA)-restricted T-cell response to unknown antigens. Immunogenetic patterns associated with self-antigenic presentation remain unclear. Herein we analyzed the molecular landscape of HLA complexes and T-cell receptor (TCR) repertoires of a large cohort of IAA patients and controls. We show that antigen binding sites of class II HLA molecules in IAA are characterized by a high level of structural homology, only partially explained by specific risk allele profiles, implying reduced binding capabilities compared to controls. Few amino acids within the synapsis HLA-DRB1-antigen-TCR, are identified as strongly associated with IAA phenotype. Those structural patterns may affect TCR repertoires, promoting immunological cross-reactivity and autoimmunity. These findings inform on the immunogenetic risk associated with IAA and on general pathophysiological mechanisms potentially involved in autoimmunity.Key pointsClass II human leukocyte antigen (HLA) loci in idiopathic bone marrow failure (BMF) syndromes are characterized by low functional divergence and decreased peptide binding capabilities, only partially explained by enrichment in risk alleles.A superstructure at the interface with the peptide binding site of DRB1 locus, potentially involved in the presentation of self-antigenic specificities, can be identified in BMF patients.This immunogenetic pattern may contribute to decrease T-cell receptor repertoire diversity, expand autoreactive T-cell clones and increase autoimmune propensity in BMF.

Extranodal NK/T cell lymphomas (ENKTCLs) are aggressive Epstein-Barr virus-associated T/NK neoplasms that predominantly affect Asians. To explore the causative somatic events, we conducted a comprehensive genetic analysis of 19 ENKTCL patients by whole-genome (

Introduction/Methods: Enzymes that modify histone H3 at lysine 27 (H3K27) to thereby regulate gene transcription (epigenetic enzymes) are recurrently inactivated by deletion and/or mutation in myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN) and acute myeloid leukemias (AML). Frustrating understanding of mechanisms is that writers and erasers of the same modification, e.g., methyltransferases that create, and demethylases that remove, H3K27 trimethylation (H3K27me3), a repression ('off') mark, are recurrently inactivated. Moreover, acetyltransferases that write H3K27 acetylation (H3K27ac), an 'on' mark mutually exclusive with H3K27me3, are also recurrently inactivated. One clue to underlying mechanisms is that MDS/MPN/AML present with diverse lineage and maturation phenotypes - perhaps these emerge from, or select for, the diverse epigenetic enzyme mutations. We therefore identified genes upregulated with specific myeloid lineage-commitment, maturation and function fates (~500 genes each) and then examined distributions of H3K27me3 and H3K27ac at these gene-loci in: (i) embryonic stem cells (ESC); (ii) hematopoietic stem and progenitor cells (HSPC); (iii) mature myeloid cells (monocyte [mono], pro-erythroblast, megakaryocyte [MK]); and (iv) AML cells. Results: Terminal-myeloid programs underwent substantial remodeling to gain H3K27ac 'on' mark from ESC/HSPC to mature myeloid (Fig.1A). Providing a mechanism for this, the H3K27 acetyltransferases EP300 and CREBBP were recruited into the RUNX1/SPI1 myeloid-lineage master transcription factor (MTF) hub by cooperation between their transcription activating domains. Mutated/translocated RUNX1, or mutated-NPM1 that cytoplasmically dislocated SPI1, disrupted this cooperation, and reverted hub content to default recruitment of histone deacetylases (HDAC) instead. Demonstrating cause-effect, inhibiting these HDAC renewed AML cell maturation to terminal lineage-fates. Meanwhile, MYC-target (proliferation) genes have high baseline H3K27ac in ESC and HSPC and do not require major remodeling during ontogeny (Fig.1A). An H3K27ac remodeling requirement for lineage-maturation but not proliferation/housekeeping explains selection pressure for inactivating mutations in EP300 or CREBBP, that we found in ~1.2% of MDS/MPN/AML in our (n=690) and other series. Consistent with pan-lineage-maturation needs for H3K27ac, the mutations were found in all lineage sub-types. H3K27me3 'off' mark was mostly erased from myeloid programs in HSPC, but was greater at MK vs erythroid, and also at mono vs granulocyte genes (Fig.1B). This implied more need for H327me3 demethylase (KDM6A/UTX) for HSPC commitment into MK vs erythroid, or mono vs granulocyte, lineages. Accordingly, KDM6A was most upregulated in MK and mono-lineage cells (Fig.1C), and myeloid-conditional Kdm6a knockout decreased platelets and increased red cells in the spleen - reported by others: https://doi.org/10.1182/blood.V128.22.1467.1467. RUNX1-ETO has been shown to specifically impede granulocytic but not mono differentiation - https://www.nature.com/articles/2403396 and KDM6A inactivating mutations were significantly more likely to occur secondary to RUNX1-ETO (4-9% BEAT and AMLSG case series) vs other cytogenetics (0.005%). Selection for KDM6A secondary mutations could thus be to channel myeloid precursors toward lineages most efficiently impeded by primary mutations. Although H3K27me3 was substantially erased at all myeloid-commitment and terminal programs (except MK) in HSPC vs ESC, subsequent lineage-commitment and maturation entailed rewriting H3K27me3 at preceding HSPC and alternate lineage-fate programs, including at MTF genes for alternate fates (Fig.1B, D). Primary MDS/MPN/AML and AML cell lines with inactivating mutations/deletions in H3K27 methyltransferase EZH2 thus displayed aberrant co-expression of lineage MTFs and gene expression programs of normally mutually exclusive lineages (Fig.1E-G). Conclusion. Epigenetic remodeling requirements vary by myeloid lineage and maturation stage. Thus, epigenetic enzyme mutations are selected by, and cause, lineage-context of transformation. This knowledge can guide choice of specific epigenetic enzyme inhibitors to remedy the lineage-maturation defects that drive and define myeloid malignancies. Figure 1 Figure 1. Disclosures Haferlach: MLL Munich Leukemia Laboratory: Other: Part ownership. Haferlach: MLL Munich Leukemia Laboratory: Other: Part ownership. Maciejewski: Novartis: Consultancy; Bristol Myers Squibb/Celgene: Consultancy; Regeneron: Consultancy; Alexion: Consultancy. Saunthararajah: EpiDestiny: Consultancy, Current holder of individual stocks in a privately-held company, Membership on an entity's Board of Directors or advisory committees, Patents & Royalties.

In recent decades, there has been a substantial increase in recognition of germline (GL) predisposition to MN. Some congenital blood disorders predispose to subsequent myeloid neoplasia (MN). Severe congenital neutropenia (SCN) and Fanconi anemia serving as prominent examples. There is emerging evidence that heterozygous variants of disease recessive conditions may constitute risk alleles for late developing MN. However, their low penetrance may preclude identification of such alterations. Our previous analysisof allelic burden among selected combined potential GL predisposition genes to MN identified that the myeloperoxidase (MPO) gene (17q22-23) carried the highest risk of pathogenic alterations (Li, S.T, Leukemia, 2020). GL MPO mutations cause MPO deficiency syndrome, one of the most common inherited disorder associated with phagocyte defects and variable clinical penetrance. Despite the high prevalence, inherited MPO deficiency patients are asymptomatic and may remain unrecognized. We investigated 3280 MN and bone marrow failure syndromes (BMF) pts including 1138 MDS, 260 MDS/MPN, 1661 AML and 221 AA for presence of MPO mutations. In total, 38 different germline MPO mutations were identified in 143 cases. With a stringent bioanalytic pipeline, 28 pathogenic/likely pathogenic variants from 100 MN patients were included in the study. Germline MPO variants were significantly enriched in MN compared to control population (294 vs. 125 per 10 4 individuals; PC) followed by R569W (13/100), M519fs* (13/100) and Y173C (6/100) with none being biallelic. While no differences were found in distribution of -7/7q, del5q, or del20q, MPO variants carriers harbored less tri-8 compared to wild type counterparts (2% vs. 12%; P=.008). Only FLT3 and NRAS were significantly associated with MPO mutations. We then investigated the effects of MPO deficiency on hematopoietic function in murine model using competitive repopulation assays, whereby the difference in CD45.1/CD45.2 isotypes transplanted in to ROSA26 (tdTomato-EGFP) recipients assayed by flow cytometry allowed distinction of the 2 grafts. Mpo-/- cells gained over time proliferative advantage over normal murine bone marrow cells. Reverse combinations (graft mix vs recipient) also replicated this result. This effect was not due to increased HSC content in Mpo-/- marrow as there were no significant differences in LSK, CMP and MEPs cells. We then investigated the clonogenic consequences of Mpo-/- cells in the setting of H 2O 2 induced oxidative stress, after exposing to hydrogen peroxide. Mpo-/- cells increased clonogenic potential after second serially replating. This prompted us to investigate an MPO inhibitor, AZD5904 in human leukemia cells with different MPO expression (HL-60, high; K562, low). High MPO expressors retained higher cell viability following H 2O 2 and MPOi addition compared to those without MPOi (85±10 vs.59±8, P In conclusion, for the first time we demonstrated that germline MPO variants constitute risk alleles for MN evolution and replicated the potential replicative advantage of MPO deficient cells in murine model and potential mechanisms of the MPO deficiency in vitro including activation of non-homologous DNA repair response and error-prone DNA repair favoring replication. Repeated cycles of stress hematopoiesis e.g., due to increased infection rate may provide conditions in which MPO variants contribute to the risk of MN. Disclosures Haferlach: MLL Munich Leukemia Laboratory: Other: Part ownership. Maciejewski: Regeneron: Consultancy; Alexion: Consultancy; Bristol Myers Squibb/Celgene: Consultancy; Novartis: Consultancy.

Downregulation of class I human leukocyte antigen (HLA)-restricted antigen presentation has been identified as mechanism of immune-escape in many malignant and non-malignant disorders. In idiopathic aplastic anemia (AA), evolution of immune-privileged paroxysmal nocturnal hemoglobinuria (PNH) clones has been attributed to immune escape due to deficiency of GPI-anchored protein in the context of T-cell mediated autoimmunity. However, other mechanisms of clonal selection may also operate with or independently of PNH. Our group first described the presence of both somatic uniparental disomy (UPD) and microdeletions of the HLA region leading to loss of heterozygozity (LOH) and/or haploinsuffciency.1 Later the proof-of-concept of somatic mutations in HLA class I was provided.2 Mechanistically, HLA LOH leads to loss of an allele involved in the presentation of immune-dominant peptides, while haploinsufficiency may decrease the presentation threshold. Moreover, the general level of individual structural diversity of HLA molecules may determine the ability to present diverse targets, eventually derived from auto-antigens, and functionally would operate in the opposite direction to HLA LOH. In this scenario, we hypothesize that defects in both class I and II HLA loci may constitute different patterns of immune escape, reducing respectively CD8+ and CD4+ related activation and thus contributing to rescue hematopoietic stem cells from the immune attack. Furthermore, our idea is that the immune-escape environment may be related to the grade of HLA evolutionary divergence (HED), a metric that, accounting for the degree of structural diversity within a particular locus, represents an indirect measure of the antigenic landscape that the hematopoietic target cell is able to present (see abstract #:142693). Using a deep targeted HLA NGS panel and a newly developed in-house bioinformatic pipeline (characterized by stringent criteria for alignment, preprocessing and variant calling in the HLA region, based on the IPD IMGT/HLA database, Fig.A), we studied a large cohort of patients with idiopathic bone marrow failures (AA n=75, AA/MDS=10). In addition, we determined the impact of inter-loci HED on the probability to acquire somatic hits in HLA genes. Overall, 29 somatic HLA mutations were found in 16 patients (18%) at a median VAF of 11% (range: 2-93%):12 in class I (41%) and 17 in class II (59%), with 5 patients carrying mutations in both classes (Fig.B, C, D). The majority of those events (N=21, 72%) occurred in subjects also harbouring a PNH clone of small size (12 out 16 patients, median PNH clone size 1% [range:1-46%]). Most mutated loci were A and C for class I and DQB1 for class II (Fig. C, D); 9 mutations were identified as missense, with disruptive changes, 7 were intronic indels while 13 hits were localized in 5' or 3' untranslated regions (UTRs) (Fig.E, F). Through a computational prediction of the HLA regulatory domains involved in the UTR aberrations, we identified domains essential for the binding of GATA-1, RXRbeta, SP-1 and NFKB. The impairment of those regions may affect the transcription of HLA complexes. AA HLA mutant cases had more frequently a severe disease at diagnosis (severe AA: 81% vs. 60%, respectively in HLA mutated vs non mutated cases) and were in most part responders to immunosuppressive therapy (complete/partial responses: 75% vs 50% in HLA mutated vs non mutated patients). Within the AA/MDS group instead HLA mutations were found in 4 out of 10 patients (40%), including of note three -7/del7q cases. Using Pierini and Lenz algorithm3 to determine inter-class HED, we found that HLA mutations tended to occur more often in patients with a high inter-class mean HED (94% vs 72% in non mutated group, p=.001, Fig. G), consistent with the idea that higher structural diversity of HLA molecules may induce more pervasive auto-immune responses, stronger immune pressure and ultimately the establishment of immune-escape mechanisms. In summary, our results indicate the importance of class-I and -II HLA loci somatic hits as markers of autoimmunity and thereby the severity of the immune selection pressure, configuring possibly alternative mechanisms of immune-escape, in addition to immune privileged PNH clones. This environment may ultimately facilitate leukemic clonal expansion in AA-MDS setting. Disclosures Patel: Alexion: Other: educational speaker. Peffault De Latour:Apellis: Membership on an entity's Board of Directors or advisory committees; Alexion Pharmaceuticals Inc.: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Pfizer: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Novartis: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Amgen: Research Funding. Maciejewski:Novartis, Roche: Consultancy, Honoraria; Alexion, BMS: Speakers Bureau.

Severe aplastic anemia (sAA) is bone marrow (BM) failure syndrome characterized by immune mediated BM hypoplasia and pancytopenia usually treated with immunosuppressive therapy (IST) or allogeneic BM transplant. Despite the successes of IST in treating acquired, idiopathic sAA, most of the available treatment options only convert severe into non-severe forms of disease. Recently, eltrombopag (Epag), a small molecule thrombopoietin receptor (TPOR) agonist has been introduced into the routine treatment of de-novo and refractory AA with robust clinical response. Epag treatment showed remarkable tri-lineage responses in AA, suggesting an effect on hematopoietic stem and progenitor cells (HSPCs) amplification. The use of hematopoietic growth factors in AA that increases HSPCs has been associated with justified fears of facilitating AA evolution to MDS. Similarly, Epag has parallel effects on the amplification of HSPCs and there has been recent signals that long term treatment of Epag may increase the MDS risk. Using an in-house developed highly stable cell free high throughput screening of bioactive small molecule chemical library, we identified Epag as one of the most potent inhibitor of TET2. TET2 is the most abundant DNA dioxygenase in HSPCs. The loss-of-function mutations of TET2 frequently occurs in myeloid neoplasia. Here using in vitro and in vivo model system we report that effect of Epag, in part, is mediated by its ability to inhibit TET-dioxygenase activity. We demonstrate that Epag directly binds to TET2 in the presence of Fe2+ and inhibits its dioxygenase function in cell free system. In cell culture model we observed that nearly 40% of Epag partitioned into to nucleus suggesting its potential TET2 inhibitory role in cells. To test if TPOR signaling activation by Epag has any impact on TET activity we engineered murine cell lines BaF3 and 32D by introducing human TPOR. Irrespective of the status of TPOR-JAK-STAT pathway activation, as reflected in STAT5 phosphorylation, Epag demonstrated a robust TET-inhibition activity. However, in similar experiments peptide agonist of TPOR or another small molecule TPOR agonist Avotrombopag did not have any TET inhibitory effects. Consistent with these observations, Epag but not TPO or Avotrombopag inhibited TET activity in murine BM. Thus, Epag mediated TET2 inhibition in these cells are specific and independent of TPOR signaling. Consistent with previous reports, we observed that Epag treatment can significantly expand murine HSPCs even in the absence of TPOR signaling activation. The hematopoietic activity of Epag in murine system is dependent on its ability to inhibit TET2 as confirmed by the lack of Epag effects on Tet2-/- HSPCs. We further confirmed that TET-inhibition by Epag is central for its ability to expand HSPCs in murine model under in vivo settings using BM transplant experiments. We transplanted a mixture of cells derived from Tet2+/+ Pep Boy (CD45.1, 95%) and Tet2-/- C57BL/6 (CD45.2, 5%) into lethally irradiated recipient PepBoy mice. Epag treatment of mice reconstituted with the mixture of BM cells increases the fraction of Tet2+/+ BM cells only. Interestingly, Epag mimic loss of Tet2 in vivo as observed in the expansion of myeloid compartments of Tet2+/+ fractions reflected in CD11b-CD11c+, CD11b+CD11c-, CD11b+CD11c-Ly6C+Ly6G- (monocytes) and CD11b+CD11c-Ly6ClowLy6G+ (neutrophils). We did not observe any significant change in B220+, CD4+ or CD8+ lymphoid populations. On the contrary, we observed a growth restrictive effect of Epag treatment on Tet2-/- myeloid cells. Consistent with our hypothesis that Epag inhibits TET-dioxygenase activity, we observed hypermethylation in the mononuclear cells derived AA patients (n= 16) after Epag treatment. This was further confirmed in ex vivo Epag treatment of BM cells derived from healthy donor. Epag treatment significant increase clonogenic expansion of normal BM cells but not TET2 mutant myeloid cells. In summary, here we demonstrate that Epag is a potent inhibitor of TET-dioxygenase and its biological consequences are independent of TPOR-JAK-STAT activation. Epag treatment mimics TET2 loss of function in AA cells, however it restricts clonal expansion of TET2 mutant myeloid cells. Disclosures Patel: Alexion: Other: educational speaker. Nazha:Jazz: Research Funding; Incyte: Speakers Bureau; MEI: Other: Data monitoring Committee; Novartis: Speakers Bureau. Saunthararajah:EpiDestiny: Consultancy, Current equity holder in private company, Patents & Royalties: University of Illinois at Chicago. Sekeres:Pfizer: Consultancy; BMS: Consultancy; Takeda/Millenium: Consultancy. Maciejewski:Alexion, BMS: Speakers Bureau; Novartis, Roche: Consultancy, Honoraria.

Somatic mutations in components of the RNA-splicing machinery (LUC7L2, PRFP8, SF3B1, SRSF2, U2AF1, ZRSR2) are detected in several subtypes of myeloid malignancies with some of them delineating a district a clinico-morphologic phenotype. RNA-splicing defects exist also in cases without splicing factor mutations suggesting that myelodysplastic syndrome (MDS) is a "spliceopathy" in which the presence of mutations represent the most extreme pole in the mis-splicing spectrum. Although some studies have shown the effects of individual splicing factor mutations on a variety of downstream targets, no key genes have been identified. Literature reports have shown that cells from patients with myeloid malignancies are highly susceptible to alterations in splicing factors with studies presenting an epistasis among RNA-splicing factor mutations (Blood.2020006868). The dilemma is whether this epistasis has a null effect or suppresses/ activates splicing factors in the context of double hits. More recently large scale genomic studies have led to the discovery of cells carrying mutations in more than one splicing factor. We took advantage of a molecular characterized collection of an in-house MDS and AML patients (n=1,742) and publicly available data of AML (n=2,302; Beat AML Master Trial, The Cancer Genome Atlas, German-Austrian Study Group) to describe the co-existence of a chimeric splicing factor phenotype and its clonal architecture. We identified a total of 40 patients carrying at least two hits in the most common components of the RNA-splicing machinery (PRPF8, SF3B1, SRSF2, U2AF1, ZRSR2). No patient harbored two hits in the same gene. In total, we detected 149 somatic mutations. A higher frequency of mutations was found in SRSF2 (60%; 24/40) followed by SF3B1 (48%; 19/40), ZRSR2 (42%; 17/40), U2AF1 (30%; 12/40) and PRPF8 (12.5%; 5/40). Other splicing factors were found mutated more rarely as in the case of LUC7L2 which was present in only 2 patients. Mutations occurred at common hotspots in SF3B1 (K666; K700; R625), SRSF2 (P95), U2AF1 (S34; Q157) and were missense, frameshift, and splice mutations in PRPF8 and frameshift in ZRSR2. Other splicing factor mutations were less frequently observed to co-occur (LUC7L2 with SF3B1 or SRSF2). In total, double splicing factor mutations were mostly enriched in male. Seventy-six% of ZRSR2 mutant patients were male. Notable the frequency of double splicing factor mutant patients was higher in AML compared to MDS and MDS/MPN (53% vs. 25% vs. 22). Thirty-two% of double mutant patients had sAML. Most common co-occurrent genes in this cohort included TET2 (32%; 13/40), RUNX1 (22%; 9/40) and ASXL1/ STAG2 (18%; 7/40 for both). The two splicing factors that were more concomitantly mutated were SF3B1/SRSF2. In fact, concomitant SF3B1/SRSF2 were found in 22% (9/40) of the patients. As a whole, SF3B1/SRSF2 average variant allele frequency (VAF)was similar (43% vs. 47%). In this group of patients, molecular hits were not preferential with half of the patients carrying one or the other as first clone. The second most concurrent mutation with SRSF2 after SF3B1 was ZRSR2 (18%; 7/40). SRSF2 was less frequently co-mutated with U2AF1 (4 patients) or PRPF8 (3 patients). U2AF1 was found co-mutated with SRSF2 and ZRSR2 (4 patients, each) and SF3B1 (3 patients). One patient with sAML had a triple splicing factor configuration (PRPF8 VAF: 29%, SF3B1 VAF: 26%, U2AF1 VAF: 20%). In terms of conventional cytogenetics, half of the patients carried normal karyotype. Among the patients with abnormal karyotype, the most common abnormality was +8 (12%; 5/40) with 4 patients having isolated +8. One patient was hemizygous carrying -17 and PRPF8 mutation. Two patients had del(7q) and del(5q) abnormalities each. The most frequent mutation in the SF3B1/SRSF2 mutant patients was RUNX1 (44%; 4/9); in two cases, SRSF2 was the dominant/ founder clone. Three patients with SF3B1/SRSF2 also carried STAG2 while two patients had TET2 and FLT3. In summary, we describe a subset of myeloid malignancies with a complex mutational associations involving splicing factor genes. The presence of these mutations might uncover novel biological clues in the mechanisms by which splicing factors drive the phenotypes of MDS and/or AML which might not represent only a random association but define new ways by which splicing factors might silence the activity of one versus another. Disclosures No relevant conflicts of interest to declare.

Germline alterations causing inherited myeloid neoplasia (MN) have become prominent in the 2016 WHO classification. Increasing evidence reveals that many patients with MN (MDS and AML) may indeed harbor unsuspected predisposing or direct driver germline alterations. Penetrance of such alterations may be variable and family history unreliable or absent. Our previous screening of a selection of genes known to be involved in the pathogenesis of human cancers directly or indirectly according to the difference in the allelic burden in 690 MN patients (Li, S.T, Leukemia, 2020) we identified the myeloperoxidase (MPO) gene as having the highest burden of pathogenic alterations. MPO (17q22-23) encodes a heme-containing enzyme exclusively expressed in myeloid cells. MPO is known for its role in host defense against common microbial infections. In the presence of H2O2, MPO catalyzes the formation of hypohalous acids which are potent oxidants and strong bactericides. Many alterations in MPO have been found in hereditary MPO deficiency, an autosomal recessive syndrome with variable clinical penetrance. MPO deficiency syndrome is characterized by increased susceptibility to infections. To date, formal cancer association has not been established. In particular due to relatively mild phenotype, hereditary deficiency may remain unrecognized. Inspired by results of an initial screen we investigated MPO mutations in 3262 patients with various bone marrow failure syndromes (AA/PNH, 213) and MN (AML, 1655; MDS, 1133; MDS/MPN, 161; MPN, 100). In a subset of patients (20%), paired normal and tumor samples were sequenced in order to confirm germline origin. Germline MPO mutations were identified in 105 cases with 52% of them in AML. The most common recurrent mutation was c.2031-2A>C (42%; 44/105) (3' splice site of intron11) followed by I642L/F 15% (16/105), R569W 14 % (15/105), M519fs 13% (14/105), R460Q 7% (7/105), Y173C 5% (5/105), and R480H 2% (2/105). None of the variants were biallelic. The expected frequency of overall MPO variants in our cohort was higher than gnomAD (0.26% vs. 0.18%; P= .15×10-3) with odds ratio of 1.4. Notably, MPO mutations were significantly enriched in patients with age < 60 (2.2%; odds ratio =12.1, 95%CI=9.3-17.3; PC, R569W, M519fs, Y173C) contributed to this significant enrichment. Co-occurring lesions of germline MPO mutants (MPOMT) were then investigated: DNMT3A, SF3B1, TP53, NRAS were commonly observed in MPOMT compared to MPO WT-MN (20 vs. 14%, 17 vs. 13%, 16 vs. 8%, and 12 vs. 7%). Normal karyotype was found in 47% of carriers without enrichment for complex, -7/del(7q), del(5q), del(20q) or +8 karyotypes compared to WT. We compared RNA expression patterns between patients with pathogenic MPO mutations and those WT for MPO. Examination of genes significantly overexpressed in MPOMT showed strong enrichment in pseudoperoxidases hemoproteins involved in oxygen transport-globin (P=10-3) and heme biosynthesis (P=.012) (Fig.1C). The overexpression of hemoglobin/myoglobin genes in MPOMT suggests an excessive compensatory process due to the lack of true peroxidase activity. The fact that MPO has microbicidal activity led us to investigate the consequences of MPO mutations. Using shotgun whole-microbiome sequencing of circulating DNA, we examined the overall bacterial load in each patient. We found that MPO pathogenic mutation carriers had significantly higher bacterial burden than non-carriers (Fig.1D). The high bacterial burden of MPO mutants possibly promoted chronic inflammation, known to correlate with cancer pathogenesis. Indeed, we found that the presence of pathogenic MPO mutations was correlated with decreased survival in AML (P = 1.5 x 10-5). In sum, this is the first cohort showing germline MPO variants as predisposing alterations in MN. MPO variants might cause partial or total MPO deficiency resulting in chronic inflammatory responses by excessive compensation of immunity against invading microbes. This chronic inflammation stages can increase proliferation of HSC or myeloid progenitors and give rise to DNA replication associated mutations. Disclosures Maciejewski: Alexion, BMS: Speakers Bureau; Novartis, Roche: Consultancy, Honoraria.

Somatic mutations of X-chromosomal PIGA gene in hematopoietic stem cells are the key molecular events in the pathogenesis of PNH. PIGA mutations lead to impaired biosynthesis of glycosylphosphoinositol (GPI)-anchor. Resultant lack of GPI-anchored proteins (GPI-AP) leads to a clonal expansion of PNH clone and increased sensitivity of PNH red cells (RBCs) to complement-mediated hemolysis. Based on immunofluorescent staining, the severity of GPI-APs deficiency on the RBCs can be distinguished in: PNH type-1 (TI) with a normal expression; PNH type-2 (TII) with a partial reduction; and PNH type-3 (TIII) PNH with a total absence of CD55 and CD59.1 Fractions of these RBC types are believed to be a result of the molecular spectrum of PIGA mutations ranging from missense in various amino acids positions to truncations and locus deletions, which can be concurrent in some patients. Consequently, the presence of TII and TIII cells indicates the overall severity of PIGA defect and thus individual phenotypic subtypes can be defined as predominantly TII or TIII according to the dominant clonal defect. We hypothesize that PIGA genotype may shape PNH phenotype trajectory with regard to the severity of GPI-AP deficiency and clinical course. For instance, incomplete TII RBCs PNH phenotype should result in slower expansion due to incomplete selection advantage as compared to complete loss of PIGA function indicted by TIII RBCs, typical of hemolytic PNH. However, the process of "GPI-AP painting", a passive transfer from normal to TIII RBC could result in the presence of pseudo TII RBCs with intermittent phenotype. Here, using deep targeted sequencing, we correlated PNH flow cytometric phenotype with the type of PIGA mutations. For improved resolution, our study focused on patients with large clones and hemolytic PNH. Among 295 diagnosed patients who had a PNH clone (>0.5% of PNH granulocytes), we identified 60 (median age 37 years, range 5-84) who fulfilled the criteria for hemolytic PNH (PNH clone size >20% and LDH>x2.5 ULN), tested before complement blocker therapy. Whole blood DNA was subjected to multi-amplicon deep NGS using primers covering all exons of PIGA gene.2 Accordingly, 13/60 (21%) were classified as having TII and 47/60 (79%) TIII PNH. Overall, median granulocyte clone size was 71.5% (21.5-99) with no differences between TII/TIII subtypes (p=0.872) as well as it was for baseline characteristics such as sex ratio, blood counts and previous history of aplastic anemia. A total of 137 PIGA mutations were found, 27 in TII and 110 in TIII dominant patients. (Fig.1A-B) Overall, >1 and >2 mutations (clonal mosaicism) were found in 85% and 57% of patients, respectively. In cases negative by NGS of PIGA gene (9/60, 89% TIII PNH), no mutations were found even after sorting to purity of GPI(+) cells. Frameshift (39%, n=54) and missense (30%, n=41) mutations were most common, followed by splice site (16%, n=22), non-sense (11%, n=15) and non frameshift insertion/deletion (4%, n=5). Although PIGA mutations are spread throughout the entire coding region, TIII patients harbouring non-frameshift hits had more often truncating mutations mapping within the first 200bps across the PIGA domain. We demonstrated that missense PIGA hits were enriched in TII PNH, while frameshift deletions in TIII dominant patients (p=.0021 and .0201, respectively- Fig.1C). Moreover, when only dominant clone was analyzed, missense and frameshift mutations were strong predictors of PNH RBC phenotype, defining TII RBC (p=.0002) and TIII RBC patients (p=.02), respectively. (Fig.1D-E) When patients harbouring 1 unique somatic hit were studied (16/60, 27%), missense mutations accounted only for 19% of cases. Indeed, in an internal cohort of 89 AA patients with small PNH granulocytes clones ( PIGA missense and truncating mutations dictate the susceptibility of PNH RBCs to complement. We show that while missense mutations may still lead to a deficient but not totally absent GPI-AP synthesis, truncating mutations seem to be responsible for an entirely disruptive phenotype, causing the complete absence of GPI-APs on RBC membrane surface. Sequencing of other genes in the GPI synthetic pathway (AbstractID# 139335) may help to identify other somatic hits explaining PNH phenotype of cases lacking PIGA. Figure Disclosures Patel: Alexion: Other: educational speaker. Maciejewski:Novartis, Roche: Consultancy, Honoraria; Alexion, BMS: Speakers Bureau.

The loss of chromosome 7 (monosomy 7, -7) or its long arm [del(7q)] is a common karyotypic abnormality found in MDS and are associated with a poor prognosis. However recent studies may suggest a difference in survival outcomes. These lesions occur both independently and in a complex karyotype (CK). The leukemogenic features of -7/del7q can be explained through loss of heterozygosity or haploinsufficiency (HI) of key deleted genes. In addition, the phenotype of del(7q) and -7 might be further modified by the clonal architecture determining the mode of molecular evolution. Based on the hypothesis that a deletion will exert its biologic consequences in genes with HI expression and by co-associated mutations, a comprehensive analysis of genes on chr.7 was performed. Within a cohort of 8,161 patients with myeloid neoplasms (MN), we identified 511 with -7 (6%) and 143 with del(7q) (2%). In 29% of cases, del(7q) was found as an isolated alteration, while 90% of CK had del(7q); -7 showed similar percentages of isolated cases but was less common in CK (20%, 55%; respectively) cases. Del(7q) was significantly (P The most common co-associated molecular lesions with -7/del(7q) were: TP53 (31%), DNMT3A (13%), TET2 (13%), RUNX1 (12%), ASXL1 (11%), del(5q) (25%) and CK (54%). Del(7q) vs. -7 differed in the frequency of TP53 (39 vs. 29%; P=.03), DNMT3A (19 vs. 11%; P=.03), and IDH1/2 (13 vs. 52%; P Comparing the -7/del(7q) clone (calculated using allelic imbalance bio-analytic pipeline) to co-associated mutations allow us to identify the rank of -7/del(7q) in the clonal hierarchy: -7/del(7q) was an ancestral event in 58% and a secondary hit in 42% of the patients. When -7/del(7q) was the dominant hit, patients had fewer associated mutations, while in cases with secondary -7/del7q somatic mutations of TP53 (60%), IDH1/2 (30%), and DNMT3A (20%) were the ancestral hits suggesting a different clonal architecture according to -7/del(7q) clone size. Expression data was available for 49 of -7/del(7q) and 643 diploid cases for chr.7. Of the 694 informative genes on chr.7: 147 were deleted in all patients according to CNV analysis. We restricted the analysis to 15 uniformly HI genes selected because the expression inversely correlated with the level of 7q ploidy (ACTR3B, FASTK, GSTK1, IMPDH1, LRRC4, MEST, MGAM, NUP205, PRRT4, SLC37A3, SSBP1, TMEM209, TNPO3, ZC3HC1, ZNF277). Expression for these genes was then analyzed following normalization by -7/del7q clonal size. Unsupervised clustering grouped 72% of -7/del(7q) in one expression cluster. Classical -7/del(7q) signature genes fulfilled these criteria and with uniform HI in all cases. However, when we also included CUL1, CUX1, EZH2, KMT2C, and LUC7L2 to this analysis, the precision of -7/del(7q) clustering was 75% and defined a subcluster (#6). We were also able to identify cases previously misclassified (n=9) which were dispersed into 3 other clusters (14%, 8%, 3%) while 4% (n=23) of diploid cases were also misclassified and re-assigned into cluster 6. When we restricted the analysis to -7 cases only, a signature of 6 unique genes (C7orf43, CDK14, POLM, STAG3L5P, TRIM4) were found. Interestingly only POLM was on 7p, while the rest were on 7q22.1. In sum, we describe an integrated genomic and transcriptomic analysis of -7/del(7q) and we present data showing a complex heterogeneity of genes with HI expression which suggest that the clinical picture of patients carrying -7/del(7q) might be the result of a combination of clonality and HI. Disclosures Patel: Alexion: Other: educational speaker. Sekeres:Pfizer: Consultancy; Takeda/Millenium: Consultancy; BMS: Consultancy. Maciejewski:Alexion, BMS: Speakers Bureau; Novartis, Roche: Consultancy, Honoraria.

Somatic TP53 mutations are found in 10% of adult patients with MDS and de novo AML and in up to 20% of patients with therapy-related myeloid neoplasms. TP53 status is associated with complex karyotype (CK), aberrations of chromosome 5 and poor survival. Moreover, mutated TP53 (TP53MT) may be an indication for hematopoietic cell transplantation, but also predictive of relapse following the procedure, making this particular category of myeloid neoplasms (MN) a conundrum of clinical management. Unlike other tumor suppressor genes, missense (ms) mutations within the DNA-binding domain (DBD) are the most common genetic alterations in TP53 gene representing up to 80% of somatic hits, with involvement of canonical hotspots (R175H, Y220C, M237I, R248Q, R273H, R282W) in around 30% of cases. A loss-of-function (LOF) dominant-negative effect (DN) may explain the ability of TP53MT to interfere with wild type (WT) functions. Moreover, TP53 germ line (GL) mutations are responsible for Li-Fraumeni syndrome, and GL contamination may also exist in adult MN. Here we comprehensively characterize TP53MT MNs to better dissect the role of specific mutational configurations and identify the selective forces affecting outcomes in this poor prognostic MN category. A total of 764 TP53 mutations were found in 632 MN patients. Ms mutations were the most common (75%) followed by frameshift (11%), splice site (7%), nonsense (5%) and insertion/deletions (2%), with 20% of patients harboring more than 1 lesion. Topographical annotation revealed that ms mutations typically (98%) occurred within the DBD (residues 102-292) and only 2% occurred outside this region (vs. 28% in case of truncating mutations, p We then hypothesized that truncating (frameshift/nonsense/deletion) alterations require additional hits as the presence of one allele may be partially protective. Consequently, the VAF for these lesions may be a result of a UPD in a smaller fraction of cells; biallelic truncation hits thus may be truly biallelic rather than clonally mosaic, which can be demonstrated only by single cell DNA sequencing. In contrast, DN mutations in canonical hotspots decrease the function beyond 50% of the activity, with further inactivation would having less of an effect, thus exerting less selection pressure for acquisition of additional lesions. Indeed, second truncating hits (including UPD and del(17p)) were common (30%), while none of the dominant ms hits had a double-mutant hotspot configuration (vs. 14% of non-canonical ms double mutant), and these canonical dominant hits were less likely to be paired with del(17p) or truncating mutations (8%). Only 25% of CK had a WT configuration of TP53, consistent with our theory that dominant ms hits were more likely to be present without del(17p). It is possible that the inability to assert clear survival differences according to the number or types of TP53 lesions may be due to an inability to resolve the intraclonal configuration of mutations using VAF calculations. We also conclude that non-canonical ms mutations (many of them classified as VUS) may have a variable impact, with functional consequences ranging from those that are less severe than truncations to various degrees of negative dominance. Analyses of the impact of ms mutations on TP53 tetramers (which may contain various doses of mutant vs WT monomer), will shed further light on the biology of TP53MT MN. Disclosures Voso: Bristol Myers Squibb: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. Sekeres:Takeda/Millenium: Consultancy; BMS: Consultancy; Pfizer: Consultancy. Carraway:Abbvie: Other: Independent Advisory Committe (IRC); Stemline: Consultancy, Speakers Bureau; ASTEX: Other: Independent Advisory Committe (IRC); BMS: Consultancy, Other: Research support, Speakers Bureau; Novartis: Consultancy, Speakers Bureau; Jazz: Consultancy, Speakers Bureau; Takeda: Other: Independent Advisory Committe (IRC). Maciejewski:Alexion, BMS: Speakers Bureau; Novartis, Roche: Consultancy, Honoraria.

Subcutaneous panniculitis-like T-cell lymphoma (SPTCL) is a rare subtype of peripheral T-cell lymphoma affecting younger patients and associated with hemophagocytic lymphohistiocytosis. To clarify the molecular pathogenesis of SPTCL, we analyzed paired tumor and germline DNAs from 13 patients by whole-exome sequencing. All cases were Asians and were phenotypically sporadic with no family history of SPTCL. Consistent with a recent report, germline mutations in HAVCR2, encoding T-cell immunoglobulin mucin 3 (TIM3), were identified in 11 of 13 (85%) cases. All mutated cases were primary SPTCL, whereas the 2 cases without mutation were secondary SPTCL associated with underlying diseases, including viral infection and autoimmune disease. Ten patients harbored homozygous p.Y82C mutations, and 1 showed compound heterozygous mutations (p.Y82C and p.T101I). Both missense mutations altered highly conserved residues located in the extracellular immunoglobulin variable–like domain. According to the Genome Aggregation Database of >138 500 general individuals, both mutations were documented with minor allele frequencies < 0.007, indicating remarkable enrichment of these HAVCR2 alleles in SPTCL. SPTCL cells also harbored somatic mutations (6.2 per patient) that are frequently identified in genes associated with epigenetic regulation and signal transduction. In conclusion, individuals harboring biallelic HAVCR2 (TIM3) germline mutations were highly susceptible to sporadic SPTCL, which was also associated with clonal somatic mutations.

Genomic advances have initiated a new era of precision oncology. Targeted therapeutics can rationally select potentially responsive patient populations and exclude resistant patients based on molecular mechanisms of action. The application of immunomodulatory drugs (IMiDs) evolved from unselective usage to the identification of del(5q) as a target, which then led to mechanistic clues and clarification of its mode of action. Lenalidomide (LEN) has been used in non del(5q) patients with a variable response rate, but the molecular underpinnings for LEN efficacy in this setting are not known. Genomic analyses (Negoro et al., Leukemia 2016) have provided only a marginal improvement in response prediction, suggesting that other factors might be relevant in sensitizing cells to LEN. The effects of LEN are related to its binding affinity to cereblon (CRBN), the substrate adaptor of the CRL4CRBN E3 ubiquitin ligase, a cullin-ring ligase consisting of damaged DNA-binding protein1(DDB1), cullin4a (CUL4A), and regulator of cullins-1 (ROC1). LEN selectively activates CRL4CRBN E3 ubiquitin ligase to selectively degrade CSNK1A1 leading to synthetic lethality in patients with del(5q) that are haploinsufficient (HI) for CSNK1A1. However, LEN and the new generation compounds pomalidomide and iberomide (CC-220) also trigger recruitment and degradation of the zinc finger transcription factors IKAROS (IKZF1, 7p12.2) and AIOLOS (IKZF3, 17p12-q21.1) through CRBN, and therefore promote their proteosomal degradation. The synthetic lethality of these immunomodulator agents relies on expression levels of AIOLOS and IKAROS. Response to those agents depends strictly on low expressions of AIOLOS and IKAROS sensitizing certain cell types (e.g., myeloma). We hypothesized that levels of LEN sensitivity might correlate with expression levels of its targets in selected disease subtypes of myeloid neoplasia (MN). We studied a large cohort of patients with MN (n=3,328) and we identified 122 patients with -7 and 194 patients with del(7q). We noted that for those cases with informative RNA-seq expression, IKAROS exhibited haploinsufficient (HI) mRNA levels in 67% of patients with -7 and 40% in del(7q), as defined based on levels Our preliminary results suggest that selective application of new generation IMiDS targeting IKAROS have clinical potential in patients with -7. Supportive data from in vitro cellular models will be comprehensively presented. Disclosures Walter: MLL Munich Leukemia Laboratory: Employment. Hutter:MLL Munich Leukemia Laboratory: Employment. Meggendorfer:MLL Munich Leukemia Laboratory: Employment. Nazha:Tolero, Karyopharma: Honoraria; Incyte: Speakers Bureau; Abbvie: Consultancy; Jazz Pharmacutical: Research Funding; Daiichi Sankyo: Consultancy; Novartis: Speakers Bureau; MEI: Other: Data monitoring Committee. Sekeres:Celgene: Membership on an entity's Board of Directors or advisory committees; Millenium: Membership on an entity's Board of Directors or advisory committees; Syros: Membership on an entity's Board of Directors or advisory committees. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Maciejewski:Alexion: Consultancy; Novartis: Consultancy.

Background. As compared to somatic mutations, our understanding of germline (GL) variants' contribution to myeloid neoplasia (MN) is less advanced. To date, only a limited number of leukemogenic GL alterations have been identified in adult cases, but WHO Classification of Leukemias does distinguish a separate entity of MN with GL predisposition. Conceptually, pathogenetic variants can be sub-classified according to their penetrance ranging from "leukemia genes" (e.g., CEBPA, NPM1, RUNX1) to "leukemia risk genes" (e.g., DKC, FANCA, SBDS). Methods. Using publicly available data we compiled a cohort of 600 MN for which whole exome sequencing was available for both tumor and GL DNA. With the aid of an extensive calling pipeline we investigated the presence of pathogenic GL variants in 677 genes known to be involved in cancer. Application of a stringent pathogenicity score and somatic contamination filters allowed for identification of 5880 variants (with frequency Results. Overall 14 rare allele homozygous alterations and compound heterozygotes were found in 7/23 Fanconi anemia (FA) genes (e.g., BRCA2, SLX4, FANCI, FANCD2, FANCA, ERCC4, and BRIP1). Biparental inheritance in FA genes occurred more than would be expected given the total number of rare variants in the patient cohort suggesting a disease predisposition for individuals with biparental rare FA gene variants constituting a non-clonal first hit with a long latency period. This finding prompted us to prospectively analyze the frequency of FA variants in an independent MN cohort of 729 [MDS, MDS/MPN and secondary AML (sAML)] and 268 marrow failure (BMF) cases (AA, PNH). We have defined Tier-1 GL variants as those known variants (described in various FA data bases), pathogenic variants according to ASHG and new frame-shift or missense variants scored deleterious by using >4/6 applied scoring algorithms. All other GL variants including possibly pathogenic according to ASHG were assigned to Tier-2 group. The analysis revealed a total of 39 mutations (Tier 1); 80 Tier 2 variants were not included in this stringent analysis. There were 13 frame-shifts and 17 stop-gain, and there was 1 case in which 2 genes or 2 alleles were affected. The most commonly affected gene was BRCA2 with 6/39 alterations (6 variants) followed my BRCA1 in 5/39 (5 variants), FANCD2 with 5/39 alterations (4 variants), FANCA/FANCB and FANCG each with 3/39 (3 variants), FANCC in 2/39 (2 variants to a total frequency of 3.2% (23/729) and 6% (16/268) in MN and BMF, respectively. If combined, the frequency of these variants in control cohort would be expected to be 0.09%. The average age of presentation of FA carriers was 62 vs. 67 yrs. in all other MN (P=.01). The average age of presentation in BMF with FA variant was 30 vs. 45 yrs (P=0.02) in the rest of BMF group. Remarkably, we have identified 33 patients with AA who progressed to MDS, mostly del(7q), within this group: 9/33 FA Tier-1 carriers were found among these patients vs. 16/238 in non-progressors. Conclusions. The prevalence of FA gene variants are enriched in patients with MN (primarily MDS and sAML) and in BMF (primarily AA and PNH) suggesting that the carrier status of these genes may not follow an autosomal recessive trait and that the presence of a healthy allele is not entirely protective. The impact of carrier configuration may be missed due to long latency, competing morbidities and the resultant incomplete penetrance. Mono-allelic forms under genomic insult and replicative stress conditions, such as emergent increased hematopoiesis, endogenous or exogenous DNA damaging toxins and other environmental factors, might be responsible for instability in DNA repair responses. The clinical implications of it includes responsiveness to chemotherapy and in fact the choice of chemotherapy, transplant conditioning and for tumor surveillance e.g., in patients with AA at risk for MDS progression. Disclosures Nazha: Tolero, Karyopharma: Honoraria; Incyte: Speakers Bureau; Abbvie: Consultancy; Novartis: Speakers Bureau; Jazz Pharmacutical: Research Funding; Daiichi Sankyo: Consultancy; MEI: Other: Data monitoring Committee. Saunthararajah:Novo Nordisk: Consultancy; EpiDestiny: Consultancy, Equity Ownership, Patents & Royalties. Sekeres:Millenium: Membership on an entity's Board of Directors or advisory committees; Syros: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees. Maciejewski:Alexion: Consultancy; Novartis: Consultancy.

Chromosomal abnormalities can be founder lesions (e.g., t (8; 21), inv (16), inv (3)), initiate or advance disease progression (both founder and secondary hits e.g., ASXL1, TP53, RUNX1) or can be obligatory secondary hits (FLT3, NPM1). Hence, the rank of these mutations may determine the biological properties and clinical outcomes. However, while many mechanistic studies have been undertaken without identifying the key pathogenetic factors resulting from SF3B1 mutations, important biological clues can be derived from the consequences of SF3B1 alterations in the context of the clonal architecture of myeloid neoplasia (MN). SF3B1 mutant patients often have a homogeneous phenotype with isolated erythroid dysplasia, ring sideroblasts (RS) and favorable prognoses. Studies in primary MDS cells have suggested that SF3B1 mutations are initiating lesions and provide a marked clonal advantage to MDS-RS cells by propagating from rare lympho-myeloid hematopoietic stem cells. However, there is significant diversity of clinical phenotypes and outcomes including the observation that the disappearance of RS can be observed during the disease course of clonal MN and might suggest cellular shifts due to acquisition of additional hits. In such scenarios, the cell's fate in the context of SF3B1 mutations is pre-defined by the predominance of expanded hits. We took advantage of our detailed database of molecularly and clinical annotated cases with MN to study the SF3B1 mutatome and describe whether the clonal nature (ancestral vs. secondary) might change the clinical and phenotypic trajectories of MDS cells and whether the concatenation of mutations decreases the competitiveness of SF3B1 clones, leading to the dominance of other driver genes and subsequently to clonal evolution. The clonal hierarchy was resolved using our in-house designed VAF-based bioanalytic method and confirmed by the PyClone pipeline, which showed a high level of concordance. We first assigned clonal hierarchy to SF3B1 mutations by using VAFs (adjusted for copy number and zygosity) and classifying the mutations into dominant (if a cutoff of at least 5% difference between VAFs existed), secondary (any subsequent sub-clonal hit) and co-dominant hits (if the difference of VAFs between two mutations was In conclusion, our study of the clonal architecture of SF3B1 mutations highlights that clonal progression of cases with MN harboring SF3B1 mutations might be inferred by the rank of additional genetic lesions cooperating with SF3B1. Disclosures Meggendorfer: MLL Munich Leukemia Laboratory: Employment. Advani:Abbvie: Research Funding; Macrogenics: Research Funding; Pfizer: Honoraria, Research Funding; Amgen: Research Funding; Glycomimetics: Consultancy, Research Funding; Kite Pharmaceuticals: Consultancy. Nazha:Tolero, Karyopharma: Honoraria; Novartis: Speakers Bureau; MEI: Other: Data monitoring Committee; Daiichi Sankyo: Consultancy; Jazz Pharmacutical: Research Funding; Incyte: Speakers Bureau; Abbvie: Consultancy. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Sekeres:Syros: Membership on an entity's Board of Directors or advisory committees; Millenium: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees. Maciejewski:Alexion: Consultancy; Novartis: Consultancy.

Large granular lymphocytic leukemia (LGL) is a lymphoproliferative cytotoxic T cell (CTL) that typically presents in elderly population. The spectrum of the disease ranges from semi-reactive oligoclonal to clonal CTL expansion, and from silent to a chronic leukemia with para-neoplastic manifestation. However the description of the disease still preceded the new genetic landscape mapping techniques. In lieu of this, we analyzed the genetic landscape of LGL using a set of 33 genes. Interestingly, not only did the deeper analysis help with a better understanding of the complex CTL proliferative processes but indicated a suspected association with conditions such as clonal hematopoiesis of indeterminate potential (CHIP). Association of LGL with age related CH has been described in the past by us and others. Our results indicate that LGL is often associated with the presence of myeloid mutations, generally consistent with the most common typical CHIP mutations (DNMT3A, TET2, ASXL1). We then analyzed a set of 13 patients with coexistent MDS/LGL. This observation raises the question that LGL is, at least in some cases, a tumor surveillance response. All LGL patients were diagnosed using a stringent single center diagnostic criteria and at the time of diagnosis, were also checked for other conditions including MDS to prevent any risks of overlooked diagnoses. A previously well described diagnostic standard of >3/5 positive criteria was used; (i) presence of large granular lymphocytes (>500/μL, by differential counts from complete blood count) for >6 months; (ii) abnormal cytotoxic T lymphocytes expressing CD2, CD56 and CD57 and lacking CD28; (iii) preferential usage of a TCR Vβ family by flow cytometry; and (iv) TCR gene rearrangement by PCR; (v) bone marrow infiltration with LGL is another diagnostic criterion but a designated marrow biopsy is often not performed in clear LGL which are otherwise asymptomatic. Coexistent presence of MDS and LGL was confirmed in 13/240 (5%) of the patients. These patients displayed typical features of MDS including abnormal cytogenetics (8/13), dysmorphia of myeloid cells in the bone marrow (13/13), ringed sideroblasts (1/13) and the presence of highly clonal somatic myeloid mutations (8/13) with an average VAF of 28+3%. When compared to LGL only, the MDS/LGL series had 15% vs. 39% (P=0.014) STAT3/5b mutations. We detected the presence of mutant myeloid clones in 25% (41/161) of the patients with pure LGL with an average VAF of 35+ 2%. Suspecting that these mutations may be present early in MDS and CHIP, we compared the mutational spectrum of LGL against MDS (n=835) and CHIP. Indeed, an overlap between the typical CHIP mutations was noted with LGL; in addition the clonal burden in CHIP was observed to be lower, 13% vs. 33% in LGL indicating a more advanced myeloid disease in LGL. This could be due to overlapping demographics or through a common pathophysiologic link. Spliceosomal mutations, cohesion complex, PRC2 and RAS mutations were overrepresented in MDS and CHIP in comparison to LGL. Our results indicate that LGL is often associated with myeloid mutations that are also typical in CHIP. Though, age-related changes or other mutagenic stressors, including inflammatory changes could be contributing factors to both LGL and CHIP outgrowth, conversely LGL may evolve as a consequence of a tumor surveillance response to myeloid neoplasms or CHIP. While the coexistence of MDS and LGL has been described in the past, and our analysis corroborates this, we observed that LGL may also coexist with CHIP in some patients. The implication of this would be beneficial to our understanding of, not only, the individual diseases but could also impact care in CHIP survivorship. Disclosures Nazha: MEI: Other: Data monitoring Committee; Novartis: Speakers Bureau; Jazz Pharmacutical: Research Funding; Incyte: Speakers Bureau; Tolero, Karyopharma: Honoraria; Daiichi Sankyo: Consultancy; Abbvie: Consultancy. Saunthararajah:EpiDestiny: Consultancy, Equity Ownership, Patents & Royalties; Novo Nordisk: Consultancy. Sekeres:Celgene: Membership on an entity's Board of Directors or advisory committees; Millenium: Membership on an entity's Board of Directors or advisory committees; Syros: Membership on an entity's Board of Directors or advisory committees. Maciejewski:Novartis: Consultancy; Alexion: Consultancy.

Systematic application of NGS has led to the discovery of a myriad of somatic mutations. One important class of genes affected by mutations in myeloid neoplasia (MN) is involved in epigenetic regulation either directly by modifying DNA or by enzymes modifying histones such as histone methyltranferases and demethylases (e.g., EZH2, MLL, UTX). This work focuses on mutations of EP300, a histone acetyl transferase. EP300 (22q13.2) encodes the adenovirus E1A associated cellular p300 transcriptional co-activator protein with lysine acetyltransferase activity as does its paralog CBP (16p13.3). Ablation of EP300 in mice results in embryonic lethality. Heterozygous germline EP300 mutations were described in Rubinstein-Taybi syndrome (RBTS), a congenital neurodevelopmental disorder characterized by facial dysmorphology, distal limb abnormalities and mental retardation. In addition, RTBS patients appear to have a predisposition for childhood malignancies. Somatic EP300 mutations have been found in solid tumors and lymphoid neoplasia but have been less studied in MN. We analyzed DNA sequencing results of 2382 MN patients [AML (n=1646), MDS (n=344), MDS/MPN (n=79), and MPN (n=31)]. Germline status of lesions was confirmed using skin-biopsy or CD3+ derived DNA, when such specimens were available. New nonsense, deletion, and exceedingly rare, individual, missense mutations (4/6 applied scoring algorithms (CADD, PolyPhen2, SIFT, LRT, MutationAssesor and MutationTaster within Annovar) were included in downstream analyses. EP300 mutations were detected in 2% of patients (47/2382); 61% (28/46) were somatic and 39% (18/46) germline. Mutation types in somatic vs. germline were: missense (43 vs. 68%), nonsense (25 vs. 21%), splice site (14 vs. 0%), deletion (7 vs. 5%) and insertion (10 vs. 0%). No patients with hemizygous deletions were detected by SNP array. Median age of EP300 somatic and germline mutants were similar compared to EP300 wild type [63 (56-81) vs. 65 (29-79) vs. 65 (18/92) years]. For somatic EP300 mutants, 43% of cases (10/23) were characterized by normal and 22% complex cytogenetics without enrichment for -7/7q, del5q, del20q or +8. Of note, 67% of MDS patients (4/6) had a prior history of other malignancies treated with chemoimmunotherapy or radiation (lymphoma, prostate cancer). Using VAF-based bioanalytic methods, we were able to roughly recapitulate the clonal architecture and identify founder lesions. We compared the expected vs. observed VAFs according to the corresponding variability in read counts and a stringent binomial distribution algorithm. EP300 mutations were more likely to be dominant lesions (62.5%) while 37% were subclonal. Serial sample analysis was performed; EP300 mutant subclones expanded during progression of the diseases suggesting a tumor suppressor function of EP300. In normal hematopoiesis, EP300 expression increases proportionally with myeloid precursors' maturation. We found that EP300 mRNA levels were significantly lower in AML patients vs. healthy controls (log2CPM: 7.92 vs. 8.27, P=.03; Beat AML). One therapeutic approach would be to alleviate the effects of the mutations using HDACi. However, HDACi had equal sensitivity in EP300 mutant and WT cells in publicly available databases of drug sensitivity and also according to expression levels (low vs. high). Another alternative therapeutic approach is based on the potential synthetic lethality of a total loss of EP300 concluded from embryonic lethality in mice. Using in vitro proliferation assays we found that the HATi S7152 was more effective in growth inhibition (GI) in the EP300 mutant cells, Jurkat (M2118I) at 10 µM (GI: 84%) compared to EP300 wild type cells e.g., KG-1 (43%), K562 (54%), U937 (24%) and TF-1 (15%) at 10mM while NB-4 (L2393V) cells only reached a GI (27%) at 20µM. In sum, we characterized for the first time the spectrum of somatic and germline EP300 mutations in MN and the possible therapeutic implications of using HATi to modulate EP300 activity. Figure Disclosures Sekeres: Millenium: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Syros: Membership on an entity's Board of Directors or advisory committees. Maciejewski:Alexion: Consultancy; Novartis: Consultancy.