ALBERT

Description: Prevention is an essential component of cancer eradication. Next-generation sequencing of cancer genomes and epigenomes has defined large numbers of driver mutations and molecular subgroups, leading to therapeutic advances. By comparison, there is a relative paucity of such knowledge in premalignant neoplasia, which inherently limits the potential to develop precision prevention strategies. Studies on the interplay between germ-line and somatic events have elucidated genetic processes underlying premalignant progression and preventive targets. Emerging data hint at the immune system’s ability to intercept premalignancy and prevent cancer. Genetically engineered mouse models have identified mechanisms by which genetic drivers and other somatic alterations recruit inflammatory cells and induce changes in normal cells to create and interact with the premalignant tumor microenvironment to promote oncogenesis and immune evasion. These studies are currently limited to only a few lesion types and patients. In this Perspective, we advocate a large-scale collaborative effort to systematically map the biology of premalignancy and the surrounding cellular response. By bringing together scientists from diverse disciplines (e.g., biochemistry, omics, and computational biology; microbiology, immunology, and medical genetics; engineering, imaging, and synthetic chemistry; and implementation science), we can drive a concerted effort focused on cancer vaccines to reprogram the immune response to prevent, detect, and reject premalignancy. Lynch syndrome, clonal hematopoiesis, and cervical intraepithelial neoplasia which also serve as models for inherited syndromes, blood, and viral premalignancies, are ideal scenarios in which to launch this initiative.

Description: Introduction We investigate the role of Ezh2 in neutrophil function using murine progenitor cells differentiated into neutrophils lacking the Ezh2 gene. Ezh2 is the catalytic component of the polycomb repressive complex 2, which methylates lysine 27 of histone H3. It is frequently disrupted in myelodysplastic syndromes (MDS) leading to loss of function (Ernst et al., 2010). Mutations in EZH2 are found in 6% of MDS patients and while not strongly linked to cytopenias or blast proportion, they are independently associated with worse overall survival compared to patients with wildtype EZH2 (Bejar R. et al., 2011 and 2012). We hypothesize that Ezh2 mutations may cause qualitative defects in myeloid cells that impact their function and could contribute to the adverse prognosis observed in EZH2 mutant MDS. Methods Bone marrow from Ezh2 null (Ezh2-/-) and littermate control mice (WT) were transduced with HOXB8 fused to the estrogen receptor ligand-binding domain to produce immortalized myeloid progenitor cells. Removal of estrogen from the media allows these cells differentiate into mature neutrophils (Wang G.G., 2006). Differentiated cells were characterized for surface markers by flow cytometry and for gene expression by PCR of mRNA. Spontaneous cell death was measured by annexin/PI staining. Cell cycle patterns were determined by measuring the red emission of PI. Chemotactic function was assessed by counting cells that migrated across a transwell in presence/absence of the attractant zymosan. For phagocytosis experiments, cells were incubated with Fluoresbrite YG carboxylate beads at 37°C or 4°C. Reactive oxygen species (ROS) generation was measured by the oxidation of dihydrorhodamine 123 into fluorescent rhodamine 123. Results Estrogen withdrawal caused differentiation of both WT and Ezh2-/- lines into cells with mature neutrophil morphology after six days (Figure 1a). Both differentiated lines expressed the neutrophil surface markers CD11b and CD62L and the neutrophil-specific genes lactoferrin and Itgb2l. Ezh2 -/- cells had an increased rate of spontaneous cell death compared to WT in undifferentiated (32.81% vs. 20.33%) and mature cells (32.82% vs. 14.23%). Nevertheless, both progenitor cell lines showed similar cell cycle patterns, demonstrating that Ezh2 absence had no other effect on cell cycle progression. Ezh2 -/- neutrophils failed to migrate towards zymosan (Figure 1b). Expression of Tlr2, which binds zymosan, and other Toll-like receptors (Tlr4/5/9) were similar between the differentiated cell lines. Cells incubated with FITC-zymosan at 37°C showed no fluorescence differences between cell lines, indicating similar adherence. Experiments with neutrophils from an MDS patient with homozygous EZH2 mutations demonstrated a similar migration defect. Additional studies in MDS patient samples are ongoing and will be presented. Phagocytosis was reduced in Ezh2-/-cells. Unstimulated, the number of cells ingesting and adhering YG-beads was significantly greater with WT cells than with Ezh2-/-cells. When activated with fMLP, both lines showed increased adherence of YG-beads but the number of phagocytosing Ezh2-/- cells was reduced. The average number of beads ingested by each cell was lower for Ezh2-/- cells compared to WT (5.95 vs 2.94, p 〈 0.001) in resting cells, and 9.47 vs. 3.73 in fMLP-activated cells, p 〈 0.01. The fraction of Ezh2-/- neutrophils generating ROS when stimulated with PMA is 2.4-fold higher than for WT cells. ROS production was greatly reduced in the presence of diphenyleneiodonium (DPI), confirming the role of NADPH oxidase in the generation of ROS. Conclusion Our results indicate impaired function of neutrophils derived from Ezh2-/- mice, demonstrating increased spontaneous cell death, impaired migration, decreased phagocytosis, and overproduction of ROS. Qualitative defects observed in neutrophils deficient for EZH2 may help explain the adverse prognosis associated with these mutations in MDS patients. Disclosures: Bejar: Genoptix: Consultancy, Honoraria, Membership on an entity’s Board of Directors or advisory committees; Celgene: Consultancy, Membership on an entity’s Board of Directors or advisory committees.

Description: Abstract 1865 Cytogenetic changes, mainly deletions, can be found in about 30–50% of patients with Myelodysplastic Syndromes (MDS). To identify a tumor suppressor candidate within a commonly deleted region on chromosome 20q, we performed gene expression analysis on CD34+ bone marrow cells obtained from 8 patients with a 20q aberration and 18 with a normal karyotype. However, we were unable to identify genes that were significantly differentially expressed in aberrant 20q karyotype as compared to normal karyotype MDS patients. In contrast, a comparison of CD34+ cells from all MDS cases analyzed (n=26) with CD34+ cells obtained from normal bone marrow (n=4) revealed 108 genes that were differentially expressed. Interestingly, one of the top-scoring genes was MYBL2, which is located on chromosome 20q. MYBL2 levels were downregulated more than two-fold in 18 out of 26 cases. RNAi-mediated knockdown of MYBL2 in CD34+ normal bone marrow cells revealed a signature of genes functionally associated with the G2/M cell cycle phase confirming the well-documented role of MYBL2 as key transcription factor governing the onset of cell division. We hypothesize that in a subset of MDS cases the control of cell division may be impaired by low levels of MYBL2 such that altered cell fates established during cell division in early hematopoietic stem and progenitor cells will lead to clonal expansion with imbalanced or impaired differentiation. Indeed, gene set enrichment analysis revealed a strong enrichment of MYBL2 signature genes in MDS CD34+ cells. In support of a potential role as tumor suppressor, resequencing of MYBL2 (144 patients) identified 2 somatic mutations, pinpointing an additional mechanism to reduce expression of normal levels of wild-type MYBL2. Disclosures: No relevant conflicts of interest to declare.

Description: Introduction: Ezh2 is the catalytic component of the polycomb repressive complex 2, which methylates lysine 27 of histone H3 (H3K27). Loss of function mutations in EZH2 are found in 6% of MDS patients and are independently associated with worse overall survival compared to patients with wildtype EZH2 (Bejar R. et al., 2011 and 2012). Our group has described that neutrophils derived from Ezh2-/- mice have functional defects (Perez-Ladaga et al., 2013), including decreased phagocytosis, aberrant migration and overproduction of reactive oxygen species (ROS). To determine how loss of Ezh2 might contribute to these functional deficits, we performed gene expression profiling on immortalized myeloid cell lines capable of neutrophilic differentiation. Methods: Bone marrow from Ezh2 null (Ezh2-/-) and littermate control mice (WT) were transduced with HOXB8 fused to the estrogen receptor ligand-binding domain to produce immortalized myeloid progenitor cells. Removal of estrogen from the media allows these cells differentiate into mature neutrophils (Wang G.G., 2006). RNA from progenitor and mature neutrophils (WT and Ezh2-/-) was extracted each condition in duplicate and subjected to gene expression profile (Affymetrix). Transcriptome analysis was conducted with TAC software from Affymetrix and gene set comparisons between the different phenotypes were analyzed with Gene Set Enrichment Analysis (GSEA). Rescue by lentiviral re-introduction of Ezh2 into Ezh2-/- cells is currently ongoing. Results: Estrogen withdrawal causes differentiation of WT and Ezh2-/- lines into mature neutrophils after six days. Interestingly, WT neutrophils lose Ezh2 mRNA and protein expression as soon as three days after estrogen withdrawal. WT mature neutrophils lack Ezh2 and trimethyl-H3K27 (me3H3K27), showing similar amounts as Ezh2-/- derived neutrophils. Gene expression profiling of 65956 transcripts demonstrated that 1953 of them were differentially expressed between WT and Ezh2-/- mature neutrophils. Nearly 65% of these genes were upregulated in Ezh2-/- derived neutrophils when compared to WT. As Ezh2 levels in mature neutrophils are similar in both conditions, gene expression differences are likely due to EZH2 and me3H3K27 differences in the progenitor state. Among the differentially expressed genes, the transcription factor GATA1 was found upregulated in Ezh2-/- derived neutrophils, a result confirmed by qPCR. GATA1 regulates the expression of hundreds of genes and is essential for erythropoiesis. GATA1 target erythroid genes were also found upregulated in Ezh2-/- derived neutrophils when compared to WT, while no significant differences in neutrophil gene expression were detected. Similarly, GSEA analysis of Ezh2-/- vs. WT confirmed strong enrichment for erythroid associated expression programs. A Heme Metabolism Signature based on a panel of 182 genes showed a strong correlation with Ezh2-/- derived neutrophils (Figure 1A). GSEA was used to examine possible mechanisms behind the functional defects previously reported in Ezh2-/- derived neutrophils such as overproduction of ROS and impaired migration. A gene set based on 192 genes encoding proteins involved in oxidative phosphorylation demonstrated a significant correlation between this pathway signature and Ezh2-/- derived neutrophils (Figure 1B).On the other hand, GSEA showed a positive correlation between WT differentiated neutrophils and a panel of 115 genes involved in leukocyte transendothelial migration (Figure 1C). Conclusion: Our results show that HOXB8-ER immortalized myeloid progenitor cells are able to produce mature neutrophils even in absence of Ezh2. The loss of Ezh2 in myeloid progenitor cells is associated with the differential expression of 1953 genes in mature neutrophils, including the upregulation of genes involved in erythroid differentiation programs and oxidative phosphorylation, and the downregulation of genes involved in leukocyte migration. Ongoing rescue experiments re-introducing Ezh2 into Ezh2-/- progenitor cells are being performed to determine if this restores normal neutrophil functions and silences the aberrant erythroid gene expression in Ezh2-/- derived neutrophils. Our findings may help explain how Ezh2 loss causes neutrophil dysfunction and contributes to the adverse prognosis associated with EZH2 mutations in MDS patients. Disclosures Orkin: Editas Inc.: Consultancy. Ebert:genoptix: Consultancy, Patents & Royalties; Celgene: Consultancy; H3 Biomedicine: Consultancy. Bejar:Alexion: Other: ad hoc advisory board; Celgene: Consultancy, Honoraria; Genoptix Medical Laboratory: Consultancy, Honoraria, Patents & Royalties: MDS prognostic gene signature.

Description: One of the most commonly mutated genes in myelodysplastic syndrome (MDS) is TET2, which actively demethylates DNA through a 5-hydroxymethylcytosine (hmC) intermediate. MDS is a disease of impaired hematopoietic differentiation, and murine models of TET2 loss display an expansion of the hematopoietic stem and myeloid progenitor pool. Previous studies examining the effect of TET2 loss on DNA methylation in myeloid malignancies have reached conflicting conclusions. We therefore aimed to better characterize the effects of TET2 loss on methylation within a well-defined set of MDS patients. We compared the DNA methylation status of bone marrow aspirates from TET2 mutant or WT cases (n=74) from a large, well-characterized cohort of MDS patients by reduced-representation bisulfite sequencing (RRBS). In order to focus specifically on the effects of TET2, each TET2 mutant patient sample was matched with a TET2 WT sample for all other factors known to affect DNA methylation, including age, sex, and disease subtype. Importantly, samples were also matched for presence of other somatic mutations, including known oncogenes and tumor suppressors. We found that global methylation was significantly increased in the TET2 mutant group. In patients with the highest TET2 mutant allele burden, the differential methylation was even greater. We next examined specific sites of differential methylation. Interestingly, though promoter regions are enriched for hmC and Tet family binding in murine ES cells, we found no difference in methylation between groups at promoters (TSS +/- 1kb). We found no difference in methylation between TET2 mutant and WT patient samples within enhancers. When we examined intragenic regions, however, we found a significant increase in methylation at intron-exon boundaries (+/- 500bp, excluding promoters) in the TET2 mutant group, which accounted for the majority of global differences in methylation. We further examined our findings in a conditional murine model of Tet2 loss. RRBS of Tet2+/+, Tet2+/- and Tet2-/- Mx1Cre+ HSPC (lin- c-Kit+ Sca-1+; HSPC) showed analagous patterns of methylation to TET2 mutant MDS samples: hypermethylation was seen globally and locally at intron-exon boundaries, but not at promoters or enhancers, in Tet2+/- and Tet2-/- vs Tet2+/+ cells. We performed hmC-DNA immunoprecipitation followed by semi-quantitative RT-PCR for a subset of the regions hypermethylated in Tet2-/- HSPCs, and found enrichment of hmC at these sites in Tet2+/+ versus Tet2-/- DNA, demonstrating that the observed hypermethylation is due to loss of hmC conversion following loss of TET2. DNA methylation at intron-exon boundaries affects mRNA splicing through interactions with RNA polymerase II binding partners. RNA sequencing of bone marrow aspirates from 2 TET2 mutant and matched WT patient pairs revealed few changes in absolute transcript abundance. We defined a gene set comprised of intragenic regions hypermethylated in TET2 mutant patients that also involved known splice variants. Using this approach, we observed shifts in splice variant transcript abundance within individual genes, specifically involving regions of aberrant hypermethylation in TET2 mutant patients. We expanded on these results with RNA sequencing of Tet2+/+ and Tet2-/- HSPCs and observed similar shifts in splice variant abundance among genes with differential intragenic methylation. In aggregate, our data show that loss of TET2 in MDS patients and in a Tet2 knockout murine model results in a global increase in DNA methylation and that hypermethylation due specifically to loss of TET2 is localized to intron-exon boundaries. Furthermore, we find that increased methylation at sites of alternate splicing correlates with shifts in splice variant ratios across a broad subset of genes. We hypothesize that this alteration in splice variant abundance may affect hematopoietic differentiation and promote the development of an MDS phenotype. Disclosures Bejar: Genoptix: Consultancy, Honoraria, Licensed IP Other; Celgene: Consultancy, Honoraria.

Description: Introduction Myelodysplastic Syndromes (MDS) are a heterogeneous group of clonal myeloid stem cells disorders with high prevalence in the elderly characterized by inefficient hematopoiesis, peripheral blood (PB) cytopenias, and an increased risk of transformation to acute myeloid leukemia (AML). The karyotype is the clinical parameter with the strongest prognostic impact according the IPSS-R (Greenberg et al., 2012). The most frequent cytogenetic alteration is the chromosome 5q deletion (del[5q]) which as a single anomaly, confers a good prognosis and predicts an excellent response to lenalidomide. Whether other genetic abnormalities routinely cooperate with del(5q) is not known. Whole-exome sequencing (WES) is a powerful tool to identify somatic mutations in protein coding genes that might cooperate with del(5q). In order to better understand the genetic basis of MDS with del(5q), we performed whole-exome sequencing (assessing 334,378 exons) of tumor-normal paired samples from 21 MDS patients. Herein we describe the preliminary findings. The analysis is ongoing and the complete results will be presented in the meeting. Methods A total of 21 patients with MDS (16 with del(5q) as a sole abnormality, 3 with del(5q) and additional alterations and 2 with normal karyotype) were included in our study. We examined a total of 25 tumor samples (21 diagnostic bone marrow (BM) samples with matched CD3+ cells as a controls, additional BM samples from 3 patients during lenalidomide treatment and 1 bone marrow sample from a del(5q) patient after AML progression). DNA was extracted from BM samples and from isolated peripheral blood CD3+ cells (magnetic-activated cell sorting (MACS), MiltenyiBiotec GmbH, Germany). The purity of CD3+ cells was assessed by FC 500 flow cytometer (Beckman Coulter, Hialeah, Fl, USA). Only DNA that fulfilled quality controls required by WES were submitted. For each diagnostic sample, we performed Conventional G-banding cytogenetics and fluorescence in situ hybridization (FISH, to confirm or dismiss 5q deletions). Whole-exome targeted capture was carried out on 3 μg of genomic DNA, using the SureSelect Human Exome Kit 51Mb version 4 (Agilent Technologies, Inc., Santa Clara, CA, USA). The captured and amplified exome library was sequenced with 100 bp paired-end reads on an Illumina HiSeq2000. Whole-exome sequencing data were analyzed using an in-house bioinformatics pipeline as previously reported. Somatic mutations identified as alterations present in tumor but not in the matched CD3+ sample were validated by Sanger sequencing. Results In our preliminary analysis of WES from 12 patients (10 patients with 5q- and 2 patients with normal karyotype), a total of 249 non-silent somatic variant candidates were identified, of which 146 were confirmed as somatic mutations. Recurrent mutations were observed in three genes (ASXL1, NBPF10 and SF3B1) in 3 different patients. Seven genes (HRNR, JAK2, POTEG, MUC5B, PHLDA, TTN, ZNF717) were mutated in two patients. Mutations in several genes known to be mutated in MDS (ASXL1, JAK2, RUNX1, SF3B1, SRSF2 and TET2) were also identified. Patients with the 5q deletion had an average of 11 mutations whereas patients with normal karyotype had a higher mean (14.5). Mutated genes identified in both groups were HRNR, JAK2, MUC5B, NBPF10 and SF3B1. No mutations in TP53 were detected in this subset. Pathway analysis of the complete list of somatically mutated genes will be carried out once all 21 patients are analyzed. The four in-treatment samples will be examined from their matched diagnostic samples. Conclusions Whole-exome sequencing of largely del(5q) MDS patient samples identified both known and previously unreported somatic mutations. Analysis of additional samples will allow a more complete description of the genes and pathways that may cooperate with del(5q) in the development and progression of MDS. Acknowledgments Financial support: This work has been supported (in part) by a grant from Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain (PI 11/02010); by Red Temática de Investigación Cooperativa en Cáncer (RTICC, FEDER) (RD07/0020/2004; RD12/0036/0044); Acción COST BM0801: European Genomics and Epigenomics Study on MDS and AML; Sociedad Española de Hematología y Hemoterapia (SEHH) and MDS Celgene. Footnotes Rafael Bejar and Francesc Sole contributed equally. Disclosures: Díez-Campelo: Novartis and Celgene: Honoraria, Research Funding. Cañizo:Celgene Jansen-Cilag Arry Novartis: Membership on an entity’s Board of Directors or advisory committees, Research Funding. Sanchez:Celgene: Honoraria, Research Funding. Bejar:Genoptix: Consultancy, Honoraria, Membership on an entity’s Board of Directors or advisory committees; Celgene: Consultancy, Membership on an entity’s Board of Directors or advisory committees. Solé:Celgene: Consultancy, Membership on an entity’s Board of Directors or advisory committees, Research Funding.

Description: INTRODUCTION Myelodysplastic syndromes (MDS) are a heterogeneous group of clonal myeloid stem cell disorders that are highly prevalent in elderly populations. MDS are characterized by inefficient hematopoiesis, peripheral blood (PB) cytopenias, and increased risk of transformation to acute myeloid leukemia (AML; 20–30% of patients with MDS). Around 50% of MDS patients carry at least one karyotypic aberration. The interstitial deletion of the long arm of chromosome 5 ([del(5q)] is the most common aberration, accounting for almost 30% of abnormal MDS karyotype. Various studies supports a favorable prognosis of MDS with isolated del(5q) with an excellent response to lenalidomide treatment. In order to describe the molecular events associated with MDS and del(5q) we performed whole-exome sequencing (WES)(assessing 334,378 exons) of tumor-normal paired samples from 20 MDS patients to unravel the genetic basis of MDS with del(5q). The analysis is ongoing and the complete results will be presented in the meeting. METHODS A total of 50 samples from 20 patients with MDS, with del(5q) were collected. For each diagnostic sample, we performed Conventional G-banding cytogenetics and fluorescence in situ hybridization (FISH, to confirm or dismiss del(5q)) and SNP arrays with Cytoscan HD (Affymetrix). These samples included: 20 tumor samples at diagnosis, 20 control samples and 10 samples after diagnosis, during lenalidomide treatment (5) or at the moment of relapse (5) in order to compare the genetic status before and during the treatment. Genomic DNA from tumor cells was obtained from bone marrow (BM) samples or from PB granulocytes. As a source of constitutional DNA we used CD3+T cells from each patient by isolating by magnetic-activated cell sorting. WES targeted capture was carried out on 7μg of genomic DNA, using the SureSelect Human Exome Kit 51Mb version 4.Libraries were sequenced on an Illumina HiSeq2000. Sequencing data will be analyzed using an in-house bioinformatics pipeline as previously reported. RESULTS Our preliminary analysis of these 20 new patients by WES confirmed our previous analyses with mutations in well described genes as ASXL1, JAK2 and TET2, but not in genes RUNX1, SF3B1 and SRSF2. In those patients we found two patients with missense mutation in TP53, one of the patients had an isolated del(5q) and is receiving lenalidomide treatment, and the other one had a complex karyotype. According to our prior analyses, in which 249 non-silent somatic variants were detected, we look forward to validate these mutations in this new series of patients. CONCLUSIONS We envision to validate these previous results with the new sequencing data of more patients with MDS and del(5q). We expect to measure somatic mutations that vary in abundance after lenalidomide treatment, potentially identifying mutations associated with resistance or relapse. ACKNOWLEDGEMENTS: This work has been supported (in part) by a grants from Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain (PI 11/02010); by Red Temática de Investigación Cooperativa en Cáncer (RTICC, FEDER) (RD07/0020/2004; RD12/0036/0044); 2014 SGR225 (GRE) Generalitat de Catalunya; Fundació Internacional Josep Carreras; Obra Social “la Caixa”; Sociedad Española de Hematología y Hemoterapia (SEHH)and Celgene Spain. FOOTNOTES Rafael Bejar and Francesc Sole contributed equally. Disclosures Díez-Campelo: Novartis, Celgene: Honoraria, Research Funding. Xicoy:Celgene: Honoraria. Cañizo:Celgene, Jansen-Cilag, Arry, Novartis: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding. sanchez-Garcia:Celgene: Honoraria, Research Funding. Bejar:Celgene: Membership on an entity's Board of Directors or advisory committees; Genoptix Medical Laboratory: Consultancy, Honoraria, Licensed IP, no royalties Patents & Royalties, Membership on an entity's Board of Directors or advisory committees. Sole:Celgene: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding.

Description: Background: While treatment with the hypomethylating agents (HMAs) azacitidine (AZA) and decitabine (DAC) improves cytopenias and prolongs survival in MDS patients (pts), only 30-40% of pts respond. Genomic and/or clinical models that can predict which pts will respond could prevent prolonged exposure to ineffective therapy, avoid toxicities and decrease unnecessary treatment costs. Machine learning (ML), a field of artificial intelligence, is an advanced computational analysis of complex data sets that can overcome some of the limitations of standard statistical methods. ML uses computational algorithms to automatically extract hidden information from a dataset by learning from relationships, patterns, and trends in the data. Thus, ML can produce powerful, reliable and reproducible predictive models based on large and complex datasets. The aim of this project is to build a geno-clinical model that uses ML algorthims to predict responses to HMAs. Methods: We screened a cohort of 433 pts with MDS who received HMAs at multiple academic institutions for the presence of common myeloid somatic mutations in 29 genes. Responses were assessed per International Working Group 2006 criteria. Five popular supervised classification ML algorithms including: random forest (RF), tree ensemble (TE), naive bayes (NB), decision tree (DT), and support vector machine (SVM) algorithms were used individually and in combination to enhance the accuracy of the proposed model (bag of model approach). For each iteration, the dataset was divided randomly into training and validation cohorts. The partition of the dataset was repeated multiple times randomly to minimize biases in pt selection. A 10-fold cross validation was also used on the entire dataset to assure data reproducibility. Important variables were selected using backward feature elimination and tree depth scores. Performance was evaluated according to the area under curve (AUC) and accuracy matrix. All analyses were done using KNIME (an open analytic platform for ML). Results: Among 433 pts, 193 (45%) received AZA, 176 (40%) DAC, and 64 (14%) received HMA +/- combination. The median age was 70 years (range, 31-100) and 28% were females. Responses included: 95 (58%) complete remission (CR), 14 (3%) marrow CR, 16 (4%) partial remission (PR), and 59 (14%) hematologic improvement (HI). For the purpose of this analysis, pts with CR/PR/HI were considered as responders. The most commonly mutated genes were: ASXL1 (31%), TET2 (22%), SRSF2 (17%), RUNX1 (15%), and DNMT3A (14%). In univariate analyses, no single mutation was more prevalent in responders compared to non-responders except NF1 (more common in non-responders, p = .04). A logistic regression multivariate analysis did not produce a reliable and reproducible model. When applying ML algorithms on learner (80% randomly selected pts) and predictor cohorts, the accuracy rate in predicting responses for RF was 64%, for TE 60%, for NB 60%, for DT 66%, and for SVM 51%. When results from each model were combined (a bag of models approach), the accuracy increased to 69%. Backward feature elimination and tree depth scores identified the following factors as predictors of response: hemoglobin 69 years, TP53 with variant allelic frequency (VAF) 〉15%, CBL VAF 〉30%, and RUNX1 VAF 〉 25%. Only ASXL1mutations at any VAF were predictive of HMA resistance. Interestingly, none of the mutations were selected for response or resistance when the models did not include VAF. Neither treatment modality with azacitidine vs. decitabine vs. combination nor treatment center impacted response. When the analysis was restricted to pts with higher-risk disease by IPSS, the accuracy rate in predicting responses improved: for RF it became 71%, for TE 65%, for NB 60%, for DT 64%, and for SVM 76%. When the analysis was focused on pts who achieved CR vs. No CR, the models predicted the response differently. The RF and TE models were able to predict No CR with an accuracy rate of 75% and 76% respectively. Other models were able to predict CR and No CR with lower accuracy. Conclusion: We propose a novel geno-clinical model that uses machine intelligence to predict HMA response/resistance in pts with MDS. The model has a higher accuracy rate in higher-risk MDS pts. ML can open opportunities in translating genomic data into reliable predictive models that can aid physicians in clinical decision making. Disclosures Bejar: Celgene: Consultancy, Honoraria; Foundation Medicine: Consultancy; Genoptix: Consultancy, Honoraria, Patents & Royalties: No royalties.

Description: Recurring chromosome abnormalities are frequent events in cancer and are especially prevalent in hematologic neoplasms. Somatic heterozygous deletions on chromosome 20q are detected in a variety of hematopoietic malignancies including myelodysplastic syndrome (MDS), classical myeloproliferative neoplasm (MPN), MDS/MPN overlap disorders such as chronic myelomonocytic leukemia (CMML), and acute leukemias. Del(20q) is especially prevalent in MPN patients (~10-15%), where it is the most commonly detected cytogenetic abnormality associated with primary myelofibrosis (PMF) and post-polycythemia vera myelofibrosis (MF). This suggests that heterozygous loss of genes in the del(20q) common deleted region (CDR) may contribute to adverse MPN progression. Despite these observations, relatively few genes located within the CDR have been unambiguously implicated, highlighting a significant need for further investigation. To identify genes that may play an important role in the biology of del(20q)-associated malignancies we utilized a published gene expression dataset of bone-marrow derived CD34+ cells from MDS patients and healthy controls (Gerstung et al, 2015). Comparison of the patients harboring del(20q) to healthy controls revealed STK4 (encoding Hippo kinase MST1) to be the most significantly downregulated gene (mean: 3.5-fold) among those located within the chromosome 20q CDR. We therefore set out to assess the role of Hippo kinase inactivation in hematologic malignancy using conditional gene inactivation in mice. We found that complete inactivation of both Hippo kinases (Stk4 and Stk3) within the hematopoietic system using Vav1-Cre (Stk4-/-Stk3-/-) resulted in a lethal bone marrow failure (median survival: 7 weeks) associated with myelodysplastic features and frequent extramedullary hematopoiesis in the spleen. A single copy of Stk4 rescued the lethality due to bone marrow failure, however sub-haploinsufficient mice displayed thrombocytopenia with a trend towards mild anemia; phenotypes that closely resemble those observed in MDS patients with isolated del(20q). Both a reduced number of mature megakaryocytes and the presence of dysplastic megakaryocytes were apparent in bone marrow sections. Inducible Hippo kinase inactivation in adult mice using the Mx1-Cre system similarly recapitulated several phenotypic features of both MDS and MPN. In competitive bone marrow transplant assays we found that Stk4-/-Stk3-/- hematopoietic stem cells (HSC) completely lacked engraftment potential and failed to reconstitute normal hematopoiesis, revealing a potential role for Hippo kinase function in HSC homing and retention in the bone marrow. Heterozygous HSCs maintained relatively normal steady-state hematopoiesis in peripheral blood and bone marrow for up to 48 weeks in primary and secondary transplantations, although upon aging these mice were prone to development of thrombocytopenia with increased mean platelet volume. Given the high frequency of del(20q) in MPN, especially PMF, we asked whether heterozygous Hippo kinase inactivation may cooperate with the common driver mutation JAK2-V617F to accelerate disease progression. Using an HSC-enriched retroviral transduction/transplantation model in C57BL/6 recipient mice, we monitored MPN progression for 36 weeks in heterozygous Stk4+/-Stk3+/-, or control (Vav1-Cre-), cells with or without expression of JAK2-V617F. While both JAK2-V617F groups initially displayed a similar degree of polycythemia relative to controls, we found heterozygous Hippo kinase inactivation to promote accelerated disease progression towards lethal bone marrow fibrosis during the course of observation. Recipients in this group showed significantly reduced overall survival, which was associated with higher grade fibrosis in bone marrow, elevated peripheral granulocyte counts, enhanced splenomegaly, and increased frequencies of hematopoietic stem and progenitor populations in the spleen. Together, these findings implicate aberrant Hippo kinase loss-of-function in the pathogenesis of del(20q)-associated hematologic malignancies, and shed new light on the molecular events that contribute to adverse MPN progression. Disclosures Bejar: Genoptix: Consultancy; Modus Outcomes: Consultancy; Celgene: Consultancy, Honoraria; Takeda: Research Funding; Astex/Otsuka: Consultancy, Honoraria; AbbVie/Genentech: Consultancy, Honoraria; Foundation Medicine: Consultancy. Guan:Vivace: Equity Ownership.