ALBERT

Description: NGS has led to the discovery of somatic mutations in splicing factors (SF), a group of functionally related genes previously not implicated in leukemogenesis. At least one genetic aberration in the most commonly affected 7 splicing factor genes is present in the majority of patients with MDS and related diseases (MDS/MPN and AML). The most popular and plausible hypothesis is that individual splicing mutations are associated with mis-splicing of key pathogenic genes in leukemia. However, searches for the essential mis-spliced gene or pathway in engineered cell lines and murine models have not been successful despite identification of many downstream gene targets. We have designed a strategy that overcomes pitfalls and advances results of previous attempts to identify the most essential targets. First, we have collected an expansive dataset (RNA-Seq and WGS of 1258 patient samples and 63 healthy controls) which allowed us to overcome sample size limitations and exclude cases with low tumor burdens, decreasing the analytic noise. In addition to studying the common mutant SRSF2 (n=208), SF3B1 (n=282), and U2AF1 (n=69) cases, we have also studied LOH lesions (fs, ns, deletions) in the less frequently affected splicing factors LUC7L2, DDX41, PRPF8, and ZRSR2 (n=211) (Fig.1A). Unsupervised hierarchical clustering segregated patient splicing signatures by disease type, SF mutation, and SF expression. To detect significantly dysregulated alternative splicing (AS) events, samples from each disease subtype, with mutations in SF3B1 (various), SRSF2P95, U2AF1S34, or U2AF1Q157, were compared to patients without SF mutations and also healthy controls. The disease cohorts were also stratified by LUC7L2, DDX41, PRPF8, and ZRSR2 expression levels, and the lower expression groups were compared to both the higher expression groups and healthy controls. Meta-analysis revealed over 17,000 splicing variations that were significantly dysregulated in at least one of 64 comparisons (PSI≥5%, q≤.05). Statistically significant AS events in each group were overlaid to identify commonly dysregulated AS events when compared to both the disease control and the healthy bone marrow controls (Fig.1B). We characterized AS events that were unique to the myeloid neoplasm subtypes as well as specific to genetic aberrations in SFs. We also identified genes and transcripts mis-spliced in multiple groups, suggesting a convergence of splicing factor mutations on a common target gene. The vast majority of our analysis identified alterations in isoform balance, however some splice sites that were activated only in the MDS and AML cohorts but never utilized in healthy controls. Examples of these tumor-specific splicing events are found in CERS2, which was found in a majority of patient samples, and in FMNL1, which was overwhelmingly mis-spliced in SF3B1 mutant patient samples (data not shown). We have highlighted the 52 AS events that were changed most often in comparisons against disease controls and/or healthy controls. Examples of targeted exons and introns included those in ubiquitination factors, transcription factors, DNA repair factors, and oncogenes. We classified significantly changed exons by the functional domains of the translated protein. The cohorts were then stratified according to the inclusion level of the exon or intron. The inclusion groups were compared to distinguish differences both in gene expression and in dysregulation of downstream pathways. Furthermore, the exons and introns were examined for any correlation with survival in the myeloid neoplasm subtypes. Integration of these datasets provided insights into the functional impact of AS in myeloid neoplasms, e.g., TDP1 exon 12, or exon 10b of NCOR1 inclusion both is positively correlated with expression of MYC targets and negatively associated with survival in AML patients (Fig.1C-D) In sum, we have identified strong isoform candidates for the practical study of AS driven pathogenesis, utilizing both RNA-seq and the integration of publicly available exon annotation and pathway databases. Notably, our analyses have unveiled hundreds of splicing changes dysregulated at a statistically significant level, thus warranting further assessments. This assemblage of splicing patterns found in myeloid neoplasms patients' samples is the largest in existence and should greatly advance the study of pathogenic AS. Disclosures Walter: MLL Munich Leukemia Laboratory: Employment. Hutter:MLL Munich Leukemia Laboratory: Employment. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Maciejewski:Alexion: Consultancy; Novartis: Consultancy.

Description: Background: Interpreting the pathogenic potential of an amino-acid changing single nucleotide variant (SNV) in a disease related gene can be challenging, especially for rare variants for which little or no information is available in clinical databases. In silico predictors, tools that predict the functional impact of an SNV algorithmically, can be useful in this scenario, and guidelines for variant interpretation recommend their inclusion in the interpretation process. Resources such as the dbNSFP database, which contains pre-calculated prediction scores for dozens of different algorithms, are readily available today. However, individual predictors rarely come to the same conclusion, and even for well-known disease causing SNVs results can be heterogeneous or even contradictory, which complicates their interpretation. Ensemble predictors such as REVEL, MetaLR/SVM or CADD combine the knowledge/information from multiple individual sources. These predictors use machine learning methods and training sets of pre-defined pathogenic and benign SNVs to integrate individual algorithms into a single, easy to interpret score. However, current training sets are based on pathogenic germline variants, which might cause these predictors to underperform when testing somatic variants. Aim: Development of HePPy (Hematological Predictor of Pathogenicity), an ensemble in silico predictor trained on somatic disease causing variants for use in a hematological setting. Methods: We followed the approach laid out by REVEL and used 10 in silico predictor scores and 4 phylogenetic conservation scores from the dbNSFP data base to train a random forest model. Our training set consisted of 371 unique missense SNVs from 61 hematologically relevant genes that were recurrently identified (in at least 10 patients) during routine diagnostics. All were consistently and unambiguously characterized by hematological experts as either a pathogenic somatic variant (n = 268) or a benign germline variant (n = 103) using a rigorous manual classification process within a data set of 69,879 cases studied between 2005 and 2018. Model accuracy was assessed by 10-fold cross-validation and further evaluated using a test data set consisting of 335 rare missense SNVs from routine diagnostics for which control germline material (buccal swabs, finger nail clippings) from the respective patients was available. Variants originating in the germline were expected to be mainly benign (n = 123), while somatic variants were considered pathogenic (n = 212). We compared the performance of this new tool to REVEL, MetaLR/SVM, CADD and the popular individual predictors SIFT and Polyphen2 by generating receiver operating characteristic (ROC) curves and calculating the area under the curve (AUC). Model implementation and analysis was performed using the R libraries "randomForest", "caret" and "pROC". Results: HePPy scores range from 0 (benign) to 1 (pathogenic) and cross-validation on the training set indicates a high accuracy of 0.968, which is also reflected by the clear separation in the distribution of obtained scores for benign and pathogenic training SNVs (see figure B). Application of the model to the test data set of rare SNVs shows that HePPy (AUC = 0.873) outperforms all other prediction tools in separating germline from somatic variants (see figure A). Surprisingly, both MetaLR (AUC = 0.717) and MetaSVM (AUC = 0.703) performed worse than the individual predictors SIFT (AUC = 0.794) and Polyphen2 (AUC = 0.821), while CADD (AUC = 0.831) and REVEL (AUC = 0.850) showed better performance. HePPy scores for somatic test variants were heavily skewed towards very high values (mean = 0.917). Germline variants had significantly lower scores (mean = 0.466), but their distribution was much more uniform than for somatic variants (see figure C). This suggests, to consider a significant proportion of the rare germline variants to have pathogenic potential. This is in line with the growing awareness of pathogenic germline variants and familial predisposition and emphasizes the importance of in silico predictions and other tools to replace the simple "tumor vs. normal" comparison. Summary: We developed HePPy, a new in silico ensemble predictor that is trained on 371 well-defined hematopathological somatic missense variants, which outperforms other currently available methods for in silico prediction in a hematological setting. Figure Disclosures Hutter: MLL Munich Leukemia Laboratory: Employment. Baer:MLL Munich Leukemia Laboratory: Employment. Walter:MLL Munich Leukemia Laboratory: Employment. Kern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.

Description: Lenalidomide (LEN) has established a new paradigm of targeted therapy in MDS. The mechanistic underpinnings of LEN efficacy are related to the synthetic lethality of this agent through its ability to bind cereblon (CRBN). LEN induces degradation of CK1α, which is encoded by the CSNK1A1 gene located on the del(5q) CDR, whereby haploinsufficient levels of this gene allow for selective toxicity to deletion mutants. Another common cytogenetic abnormality present in patients with myeloid neoplasia (MN) is -7/del(7q). To date no selective therapies exist for -7/del(7q), an urgent unfulfilled need, given the poor prognosis associated with this cytogenetic abnormality. We were interested to explore if this same notion of selective toxicity may be possible in del(7q) myeloid patients and sought to screen drugs for this focused population. From a large cohort of patients with MN (n=3,328), we found -7/del(7q) in 10% (n=316) of patients. We first identified a signature pattern of haploinsufficient genes on -7/del(7q) based on NGS. We then searched for haploinsufficient genes which, if targeted by investigational drugs, could provide a therapeutic window for selected MN subtypes in analogy to LEN in del(5q). For the purpose of our analysis, haploinsufficient expression was defined as

Description: Monosomy (-7) and deletions of the long arm of chromosome 7 (del7q) are frequently found in patients with myeloid neoplasms, suggesting a crucial role of this region in disease pathogenicity. -7/del7q conveys a poor prognosis and no targeted therapies exist for patients harboring this defect. We previously characterized the most common deleted regions (CDR) of del7q, including 7q22, 7q34, and 7q35-36, as well as micro deletions on del7q indicative of pathogenic genes. Unlike del5q, -7/del7q is affected both by deletions and somatic UPD, suggesting that loss of heterozygosity (LOH), rather than or in addition to haploinsufficient (HI) gene expression, may play a pathogenic role. Previous studies identified possible driver genes contributing to the pathogenesis of -7/del7q including CUX1, EZH2, LUC7L2, MLL3, and SAMD9/9L. Some of these genes may be affected by somatic mutations in hemizygous (LUC7L2), or homozygous (EZH2) configurations while others are affected by germ line (GL) mutations wherein a disease-prone allele is eliminated upon somatic acquisition of -7/del7q. The pathogenic mechanisms driving evolution of -7/del7q neoplasms, though, have not been clarified and specific therapies similar to lenalidomide in del5q have not been developed yet. The outstanding open questions include: i) the rank of -7/del7q in the clonal hierarchy; ii) genetic predispositions to -7/del7q; iii) new important genes affected by LOH or HI; iv) genetic differences between -7 and del7q; and v) co-associated somatic hits. We performed a complex molecular analysis of -7/del7q patients (N=316) using NGS: 67% with -7 (211/316) and 33% with del7q (105/316). First we performed analyses of clonal architecture using an allelic imbalance pipeline which facilitated reconstruction of clonal hierarchy. In 58% of patients -7/del7q was an ancestral lesion, while in 42% it was preceded by somatic hits such as TP53 (60%), IDH1/2 (30%), and DNMT3A (20%). The frequency of a chr7 abnormality as ancestral vs. secondary was no different in -7 vs. del7q. We then studied somatic hits associated with -7/del7q. The most commonly mutated genes were TP53 (43%), TET2 (17%), DNMT3A (15%), and ASXL1 (11%) and the most frequent additional lesion was del5q (30%). About 40% of -7 and 58% of del7q occurred in the context of complex karyotype (CK). We then compared various clinical and mutational features of patients with -7 vs. del7q: -7 was significantly associated with +8 (p=.002), while del7q was associated with CK (p=.004). Isolated -7/del7q was associated with mutations in TET2 (35% vs. 10%, p

Description: Background Patients (pts) with myelodysplastic syndromes (MDS) have heterogeneous outcomes that can range from months for some pts to decades for others. Although several prognostic scoring systems have been developed to risk stratify MDS pts, survival varies even within discrete categories, which may lead to over- or under-treatment. Deficits in discriminatory power likely derive from analytic approaches or lack of incorporation of molecular data. Here, we developed a model that uses a machine learning approach to analyze genomic and clinical data to provide a personalized overall outcome that is patient-specific. Method Clinical and mutational data from MDS pts diagnosed according to 2008 WHO criteria were analyzed. The model was developed in a combined cohort from the Cleveland Clinic and Munich Leukemia Laboratory and validated in a separate cohort from the Moffitt Cancer Center. Next generation targeted deep sequencing of 40 gene mutations commonly found in myeloid malignancies was performed. Pts who underwent hematopoietic cell transplant (HCT) were censored at the time of transplant. A random survival forest (RSF) algorithm was used to build the model, in which clinical and molecular variables are randomly selected for inclusion in determining survival, thereby avoiding the shortcomings of traditional Cox step-wise regression in accounting for variable interactions. Survival prediction is thus specific to each pt's particular clinical and molecular characteristics. The accuracy of the proposed model, compared to other models, was assessed by concordance (c-) index. Results Of 2302 pts, 1471 were included in the training cohort and 831 in the validation cohort. In the training cohort, the median age was 71 years (range, 19-99), 230 pts (16%) progressed to AML, 156 (11%) had secondary/therapy-related MDS, and 130(9%) underwent HCT. Risk stratification by IPSS: 391 (27%) low, 626 (43%) intermediate-1, 280 (19%) intermediate-2, 104 (7%) high, 104 (7%) missing, and by IPSS-R: 749 (51%) very low/ low, 336 (23%) intermediate, 190 (13%) high, 92 (6%) very high, and 104 (7%) missing. Cytogenetic analysis by IPSS-R criteria: 65 (4%) very good, 1060 (72%) good, 193 (13%) intermediate, 60 (4%) poor, and 93 (6%) very poor. The most commonly mutated genes were: SF3B1 (26%), TET2 (25%), ASXL1 (20%), SRSF2 (15%), DNMT3A (12%), STAG2 (8%), RUNX1 (8%), and TP53 (8%). All clinical variables and mutations were included in the RSF algorithm. To identify the most important variables that impacted the outcome and the least number of variables that produced the best prediction, we conducted several feature extraction analyses which identified the following variables that impacted OS (ranked from the most important to the least): cytogenetic risk categories by IPSS-R, platelets, mutation number, hemoglobin, bone marrow blasts %, 2008 WHO diagnosis, WBC, age, ANC, absolute lymphocyte count (ALC), TP53, RUNX1, STAG2, ASXL1, absolute monocyte counts (AMC), SF3B1, SRSF2, RAD21, secondary vs. de novo MDS, NRAS, NPM1, TET2, and EZH2. The clinical and mutational variables can be entered into a web application that can run the trained model and provide OS and AML transformation probabilities at different time points that are specific for a pt, Figure 1. The C-index for the new model was .74 for OS and .81 for AML transformation. The new model outperformed IPSS (c-index .66, .73) and IPSS-R (.67, .73) for OS and AML transformation, respectively. The geno-clinical model outperformed mutations only (c-index .64, .72), mutations + cytogenetics (c-index .68, .74), and mutations + cytogenetics +age (c-index .69, .75) for OS and AML transformation, respectively. Addition of mutational variant allelic frequency did not significantly improve prediction accuracy. When applying the new model to the validation cohort, the c-index for OS and AML transformation were .80, and .78, respectively. Conclusion We built a personalized prediction model based on clinical and genomic data that outperformed IPSS and IPSS-R in predicting OS and AML transformation. The new model gives survival probabilities at different time points that are unique for a given pt. Incorporating clinical and mutational data outperformed a mutations only model even when cytogenetics and age were added. Disclosures Nazha: MEI: Consultancy. Komrokji:Celgene: Honoraria, Research Funding; Novartis: Honoraria, Speakers Bureau; Novartis: Honoraria, Speakers Bureau; Novartis: Honoraria, Speakers Bureau; Novartis: Honoraria, Speakers Bureau; Celgene: Honoraria, Research Funding. Meggendorfer:MLL Munich Leukemia Laboratory: Employment. Walter:MLL Munich Leukemia Laboratory: Employment. Hutter:MLL Munich Leukemia Laboratory: Employment. Sallman:Celgene: Research Funding, Speakers Bureau. Roboz:Otsuka: Consultancy; Orsenix: Consultancy; Celgene Corporation: Consultancy; Daiichi Sankyo: Consultancy; Pfizer: Consultancy; Cellectis: Research Funding; Argenx: Consultancy; Roche/Genentech: Consultancy; Celltrion: Consultancy; Sandoz: Consultancy; Aphivena Therapeutics: Consultancy; Bayer: Consultancy; Pfizer: Consultancy; Aphivena Therapeutics: Consultancy; Eisai: Consultancy; Sandoz: Consultancy; Eisai: Consultancy; Roche/Genentech: Consultancy; AbbVie: Consultancy; Novartis: Consultancy; Janssen Pharmaceuticals: Consultancy; Bayer: Consultancy; Celltrion: Consultancy; Novartis: Consultancy; Janssen Pharmaceuticals: Consultancy; Astex Pharmaceuticals: Consultancy; Daiichi Sankyo: Consultancy; Celgene Corporation: Consultancy; Jazz Pharmaceuticals: Consultancy; Jazz Pharmaceuticals: Consultancy; Cellectis: Research Funding; Otsuka: Consultancy; Orsenix: Consultancy; Argenx: Consultancy; Astex Pharmaceuticals: Consultancy; AbbVie: Consultancy. List:Celgene: Research Funding. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Maciejewski:Apellis Pharmaceuticals: Consultancy; Ra Pharmaceuticals, Inc: Consultancy; Alexion Pharmaceuticals, Inc.: Consultancy, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Ra Pharmaceuticals, Inc: Consultancy; Alexion Pharmaceuticals, Inc.: Consultancy, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Apellis Pharmaceuticals: Consultancy. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Sekeres:Opsona: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Opsona: Membership on an entity's Board of Directors or advisory committees.

Description: Background: Genome instability is a hallmark of cancer. Mutations in DNA repair pathway genes are frequent in a number of solid tumors. Defects in DNA repair or damage response can weaken response to conventional chemotherapy and are frequently regarded as poor prognostic markers. However, a high tumor mutation burden (TMB, number of somatic mutations per mega base) was recently found to correlate with better response to immune checkpoint inhibitors e.g. in colon cancer. Patients with defects in the DNA mismatch repair (MMR) pathway in solid tumors are among the cases with the highest TMB. Hematological malignancies are generally expected on the lower end of the TMB spectrum. We used whole genome sequencing (WGS) for 3256 patients with hematological malignancies (lymphatic and myeloid) to determine factors of genetic instability across all entities. Aim: To determine the number of known mutations in genes from the DNA repair pathway To estimate TMB using WGS and identify cases with high TMB in hematologic malignancies Methods: We investigated a cohort of 3256 patients with hematological malignancies, who were analyzed according to WHO diagnostic gold standards for routine purposes (incl. 584 acute myeloid leukemia [AML] and 635 myelodysplastic syndromes [MDS] samples). We performed amplification-free library preparation and sequencing on HiseqX and NovaSeq 6000 with a median coverage of 106x. Mapping and variant calling was performed with standard pipelines via BaseSpace (all Illumina, San Diego, CA). A pool of gender-matched genomic DNA (Promega, Madison, WI) was used for a tumor-unmatched normal variant calling. (a) In detail we evaluated 180 genes involved in DNA repair. We filtered on (likely) pathogenic variants from ClinVar and for TP53 on protein-truncating variants and (likely, possibly) pathogenic variants from the IARC database. (b) For TBM calculation we determined protein-altering changes and then subtracted all gnomAD listed variants in order to eliminate most germline variants. Results: We found 479 of 3256 (15%) patients with at least one pathogenic variant according to current database annotations in DNA repair or damage response genes. Most pathogenic variants were found in TP53 (330/3256; 10%) and ATM (25/3256, 1%), however, this is probably the effect of the already available systematic database annotation for both genes. For routine diagnostic purposes TP53 mutation status had been analyzed for 1184 patients with Sanger sequencing (7%) or amplicon next-generation sequencing (93%). A 98% and 99% concordance of the pathogenic and non-pathogenic TP53 status was found in comparison to WGS. Mutations in genes from the DNA double-strand break repair (and/or homologous DNA pairing and strand exchange) pathway were found in 93 patients (3%). Pathogenic and potentially germline MMR gene mutations were found in only 3 patients (0.1%, 2 MLH1, 1 MSH6), which equals the expected frequency in the Western population (0.05-0.3%). Next, we calculated TMB. The average was 2.4 [range: 0.4-39.2]. Only samples above the 95th percentile were defined as "TMBhi" (TMB ≥5). TMB was lowest in chronic myeloid leukemia (CML) and essential thrombocythemia (ET) (〈 2) and no ET or CML patient was found among the TMBhi. We then focused on MDS, which is our largest subcohort: 56 of 635 (9%) patients were in TMBhi. Furthermore, among MDS patients, a significantly higher TMB was found in MDS-EB-2 (average 3.3 vs. 2.3 for non EB-2, p