ALBERT

Description: Background: Diffuse Large B Cell Lymphoma (DLBCL) is the most common form of lymphoma in adults. Gene expression profiling has demonstrated that DLBCL can be classified into two distinct subgroups – activated B-cell-like (ABC) and germinal center B-cell-like (GCB) DLBCL. These subgroups arise through distinct normal cells of origin, activate different oncogenic pathways and display markedly different clinical outcomes. Deregulation of the transcriptome is believed to play a key role in the malignant transformation of B cells that culminates in the development of either ABC or GCB DLBCL. Here we describe global differences in RNA expression, mutation and splicing in relation to the pathogenesis of these subgroups of DLBCL. Methods: RNA sequencing (RNAseq) has emerged as a powerful tool for defining the cancer transcriptome. While mRNA sequencing is the most widely applied method for RNAseq, it overlooks non-coding RNAs, requires high-quality RNA and lacks strand-specificity. To overcome these limitations, we developed a method for strand-specific total RNA sequencing (ssRNAseq) to characterize the transcriptomes of 112 DLBCL tumors. Results: Through this work, we defined the entire spectrum of coding and non-coding RNAs expressed in DLBCLs including hundreds of lincRNAs, snoRNAs and microRNAs in addition to mRNAs. We found that the strand-specificity of our method was greater than 95% in all cases. This strand-specific sequencing strategy allowed us to maintain the orientation of the transcript to enable more accurate transcript annotation and better prediction of novel transcripts. Furthermore, we showed that our method had equal efficacy on frozen and FFPE tumor specimens from the sample patient in 24 cases. In addition, through simultaneous measurement of expression of diverse RNA types combined with mutations in MYD88, GNA13, EZH2, and BCL2, we demonstrated that we could distinguish the clinically important subgroups of DLBCL. Finally, we applied ssRNAseq to distinct training and validation sets of DLBCL cases (N=86 and N=112) to define alternative splicing events in DLBCL and found 1,021 genes that were preferentially spliced in a subgroup-specific manner. These alternatively spliced genes were selectively enriched in a number of different pathways important in lymphomas including those related to immune function, cell cycle progression and focal adhesion pathways, suggesting that alternative splicing regulates a number of important oncogenic processes in DLCBL. Conclusions: Strand-specific total RNA sequencing is a powerful method for defining the transcriptome and alternative splicing events in DLBCL. Here we define a complete coding and non-coding transcriptome of DLBCL and report the first characterization of subgroup-specific alternative splicing in DLBCL using high throughput sequencing. Our data demonstrate the power of our ssRNAseq method in defining the molecular patterns underlying DLBCLs and provide a starting point for defining the role of alternative splicing in this complex and heterogeneous disease. Disclosures Mann: Quiagen: Research Funding.

Description: Background: HDAC inhibitors (HDACi) are being investigated as treatment for relapsed/refractory non Hodgkin lymphoma (NHL) and other cancers. However, the mechanisms underlying sensitivity and resistance to HDAC inhibition in lymphomas have not been fully characterized. We probed the cellular and molecular response to HDACi in vitro and in vivo in order to determine factors that dictate the response to HDACi and to enable design of approaches to incorporate HDACi into novel combination therapeutics. Methods: High-throughput cytotoxicity screening was performed using two different HDAC inhibitors, LBH589 (panobinostat) and SAHA (vorinostat) in 52 lymphoid cell lines characterized through RNA-seq and microarray gene expression profiling. This screen revealed a greater than 50-fold range in concentration needed to induce cytotoxicity for the 2 different HDAC inhibitors and there was moderate correlation between the 2 compounds in this panel (Pearson correlation r = 0.76, p 〈 0.01). By pairing this chemosensitivity data with gene expression profiles of the screened cell lines, we developed a gene expression classifier for LBH589 that identified resistant and sensitive cell line groups. This predictor was applied to B-cell NHL cell lines tested with LBH589 in the Cancer Cell Line Encyclopedia (CCLE) and we found that the sensitive and resistant cell line groups distinguished by this method differed more than 5-fold in IC50 (0.021 vs. 1.24 nM, P 〈 0.01 by Wilcoxon rank sum), thus validating the ability of this approach to distinguish HDACi resistant cell lines. We further initiated a clinical trial of LBH589 in relapsed/refractory diffuse large B cell lymphoma patients combined with RNAseq profiling of their tumors prior to embarking on treatment. We treated nine patients with LBH589, and application of our response predictor to scaled RNAseq gene expression data revealed 4 predicted responders and 5 predicted non-responders. Two of the predicted responders had a clinical response to LBH589, whereas none of the predicted non-responders had a clinical response, thus our classifier was able to identify all of the LBH589-responsive patients from this cohort (P = 0.08 by Fisher's exact test). Analysis of differentially expressed molecular pathways in HDACi sensitive and resistant samples by gene set enrichment revealed the JAK-STAT pathway as the most differentially expressed pathway associated with HDACi resistance (at P 〈 0.001 and FDR 〈 0.20). We further identified a number of distinct mutations including STAT3, SOCS1 and JAK1 that were associated with activation of the JAK-STAT pathway by gene expression signatures and the LBH589 response signature in DLBCL cell lines and patient samples by analysis of RNA-seq data. Phosphoprotein analysis by Western blot and Sis-inducible-element (SIE) luciferase reporter assays were used to confirm JAK-STAT activation in these samples and we found that overexpression of STAT3 Src-homology domain mutations activated JAK-STAT3 signaling in isogenic cell lines and fostered resistance to LBH589 in vitro. Conversely, using in vivo DLBCL xenograft models, we found that combining JAK-STAT and HDAC inhibition by treatment with LBH589 and ruxolitinib resulted in synergistic reduction of tumor cell viability and tumor growth with tolerable toxicity in mice. Conclusions: Sustained JAK-STAT activation appears to mediate resistance to HDAC inhibition in DLBCL and other NHLs and several recurrent genetic lesions drive JAK-STAT activation in these diseases. This process can be overcome by JAK 1/2 inhibition with ruxolitinib and these findings demonstrate a role for combination therapy with HDAC inhibitors and small molecules targeting the JAK-STAT pathway in lymphoid malignancies. Disclosures No relevant conflicts of interest to declare.

Description: Introduction Enteropathy-associated T cell lymphoma (EATL) is an intestinal tumor of the intraepithelial T lymphocytes, with a median survival time of less than 1 year. It is a rare disease in general and has two main subtypes described. Type 1 EATL is a complication in patients with celiac disease, a chronic gluten-sensitive enteropathy. Type 2 EATL, characterized by smaller monomorphic lymphocytes, typically occurs sporadically in patients without celiac disease. Very little is known about the genetic mutations and gene expression signatures that define this disease, or the extent to which the two types of EATL are genetically distinct. It has been suggested that the two types of EATLs should be reclassified as separate diseases in future WHO categories. Methods In this study, we performed whole exome sequencing to 100-fold depth of 41 EATL tumors including 23 type 1 cases and 18 type 2 cases. Both alpha-beta (65%) and gamma-delta (35%) T cell receptor rearrangements were seen among these cases. Paired normal DNA was sequenced in most (N=30) cases. We defined somatic mutations, copy number alterations, and HLA genotypes in these cases from sequencing data. Additionally, we generated RNA sequencing data on the same EATL tumors. Corresponding clinical and outcome data was collected on the same cohort. Results We found that both type 1 and type 2 EATLs had overlapping patterns of mutations and similar overall survival. The most commonly mutated genes were chromatin modifier genes (34%) including ATRX and ARID1B. We also identified recurrent somatic mutations in signal transduction genes, including JAK1 and BCL9L. TP53 mutations were also recurrent (12%). Copy number amplifications in 9q, 1q, and 8q occurred most frequently and were present in both subtypes. We further compared the mutational profiles to peripheral T cell lymphoma, angioimmunoblastic T cell lymphoma, cutaneous T cell lymphoma, natural killer/T cell lymphoma, diffuse large B cell lymphoma, and Burkitt lymphoma. These comparisons identify EATL as a genetically distinct disease with a very different pattern of mutations. RNAseq identified the gene expression patterns that are unique to EATL and also identified gene expression signatures that distinguish the two types of EATL. The DQ2 or DQ8 HLA genotype is present in the majority of type 1 cases (73%) while occurring infrequently in type 2 cases (27%). Conclusions Our study defines the genetic landscape of enteropathy associated T cell lymphoma and highlights the genetic and clinical overlap between the two types. While the two types have differences in mutations and gene expression patterns, they have more in common with each other compared to other lymphoma types. Our data may inform future decisions regarding the potential separation of the two EATL types as distinct entities. Disclosures No relevant conflicts of interest to declare.

Description: Nonhodgkin lymphomas (NHLs) are among the most common cancer subtypes, with approximately 〉350,000 new cases diagnosed annually worldwide. The majority of NHLs arise from germinal center (GC)-derived B cells. Through network analysis we demonstrated that mutations in focal adhesion pathway genes, including RHOA and GNA13, are a defining characteristic of GC-derived B cell lymphoma, occurring in ~75% of the samples analyzed, and less than 10% of ABC DLBCLs and mantle cell lymphomas. We and others have previously described RHOA mutations in GC-derived B cell lymphomas including DLBCL and Burkitt lymphoma, but the functional consequences of these mutations remain unknown. We examined the patterns of RHOA mutation in DLBCL through targeted gene sequencing in 335 tumor-normal pairs. Cell of origin analysis revealed that there were 128 GCB DLBCLs in this group, with RHOA mutations occurring in 15 (11.7%) of GCB cases and none of the 207 ABC cases. RHOA mutations also occurred in 10% of Burkitt lymphoma cases (total N=66) and zero mantle cell lymphoma cases (total N=54). All the RHOA mutations were missense, and were predicted computationally to be of high impact. A significant proportion of the identified RHOA mutations affect evolutionarily conserved residues within the protein's GTP binding domain, further suggesting a functional importance. Interestingly, inactivating G17V mutations, found in 15% of peripheral T cell and 35-70% of angioimmunoblastic T cell lymphomas, never occurred in DLBCLs. RHOA is thought to signal downstream of GNA13 in a variety of cell types to regulate cellular adhesion and migration. In our study, human mutations in GNA13 and RHOA were mutually exclusive, and GC B cell specific GNA13 deletion in the AID-Cre Gna13 mouse model resulted in reduced levels of active RHOA, suggesting a pathway effect. To determine if the altered migration patterns noted in GNA13 deficient mice were due to loss of RHOA signaling, we crossed RhoA conditional knockout mice with mice expressing Cre under the AID promoter to generate deletion of RhoA exclusively in GC B cells. We then compared these mice with AID-Cre Gna13 knockout mice in parallel experiments. We found that GNA13 deficient mice demonstrate altered GC B migration dynamics, evidence of GC zonal disorganization, and reduced levels of filamentous actin. We also found that GNA13 is required for focal adhesion formation in the human Burkitt lymphoma derived cell line Raji. Similar to the GNA13 deficient state, we found that RHOA deficient GC B cells have altered dark zone and light zone dynamics, with an increased proportion of GC B cells expressing GC light zone markers, and reduced levels of filamentous actin. These data suggest that focal adhesion genes, including RHOA and GNA13, may promote lymphoma in GC B cells via aberrant migration and cellular adhesion, processes normally essential to intrazonal cycling, affinity selection, and GC B cell maturation. To our knowledge, this represents the in vivo characterization of RHOA in GC B cells and RHO mutations in B cell lymphomas and points to an important oncogenic role for RHOA in the germinal center niche. Disclosures No relevant conflicts of interest to declare.

Description: Background Over 90% of Ph-positive chronic myelogenous leukemia (“typical CML”) patients have breakpoints in the M-bcr, which typically result in b2a2 (e13a2) and/or b3a2 (e14a2) fusion mRNAs, both of which are translated into the p210 BCR-ABL protein. CML patients with the p190 BCR-ABL (m-bcr) or p230 BCR-ABL (μ-bcr) fusion genes have been reported. Atypical BCR breakpoints outside these cluster regions are extremely rare. For instance, only 8 cases have been described of e6a2 fusion CML. Very little is known about the clinical or biological characteristics of this subtype of CML, including the role of collaborating gene mutations in the development of disease. In this study, we defined the gene mutations that occurred in a rare e6a2 CML case and compared the observed gene mutations to those in “typical” chronic phase (CP)-CML cases. To our knowledge, this is the first comparison of the genetic mutations occurring in typical CML and in this rare atypical form of CML. Methodology We identified the index e6a2 CML patient, and eight additional typical CML patients for whom we had bone marrow aspirate, peripheral blood and paired normal tissue. We performed whole-exome sequencing for all of these samples using the Agilent solution-based system of exon capture, which uses RNA baits to target all protein coding genes (CCDS database), as well as ∼700 human miRNAs from miRBase (v13). In all, we generated over 3 GB of sequencing data using high throughput sequencing on the Illumina platform. Results We identified 15 candidate cancer genes that were somatically mutated in our e6a2 CML patient. Commonly implicated biological processes comprising these genes included transcription (STAT5A, TET2, GTF2F1), cellular differentiation (TP73), and signal transduction (GPR116). Interestingly, the majority of these mutations also occurred in typical CML, albeit at lower frequency. Thus, genes mutated common to our atypical case and typical CMLs included STAT5A, TET2, GTF2F1, ABL1 and CYP2A6. Thus, while atypical e6a2 BCR-ABL fusion CML cases are extremely rare, they appear to share many aspects of the biology with typical CMLs. Conclusion This study represents an in-depth analysis of a rare e6a2 CML in combination with one of the first analyses of gene mutations that occur in typical CML. Our data provide a significant first step to identifying genes that play a role in the pathogenesis along with BCR-ABL that perhaps contribute to drug resistance, and ultimately impact overall survival. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 898 Burkitt Lymphoma (BL) is a highly proliferative form of non-Hodgkin lymphoma and is characterized by translocation of the C-MYC gene to the immunoglobulin gene loci resulting in deregulation. The role of collaborating gene mutations in BL is largely unknown. We performed whole exome sequencing and gene expression profiling of 57 Burkitt lymphoma and 94 DLBCL exomes. Mutational analysis revealed that ID3 is recurrently mutated in 38% of Burkitt lymphoma samples. ID3 mutations did not occur in any of the 94 DLBCL cases. ID3 gene expression was also found to be a distinguishing feature of Burkitt lymphomas (P

Description: Introduction: Plasmablastic lymphoma (PL) is an aggressive variant of lymphoma, with strong association with HIV. Despite significant improvements in the survival of other lymphomas, PL has a short overall survival (14 months). The association with Epstein-Barr virus (EBV) infection and MYC chromosomal translocations are defining features of PL. However, the genetic causes and the role of specific mutation in PL are largely unknown. This limitation hinders the design of therapeutic approaches aimed to improve PL survival. Therefore, we performed a comprehensive analysis of the genetic landscape in PL. Methodology: Whole exon sequencing of 52 de novo PL tumors from HIV+ patients and matched normal tissues (15%) using high throughput sequencing on the Illumina platform. Of these 52 patient samples, 10 tumor and 4 normal control samples were removed due to poor sequencing quality. Mapping and variant calls were performed using BWA and Varscan softwares. Variants call filtering criteria include: exon location, minimum coverage of 5%, and onside T-test P value ≤ of 0.01 or 0.02). For immunohistochemistry, we use p-TAK1 Thr184-187 (Cell signaling) and p-BTK Tyr551(Termofisher) antibodies. Results: Analysis of exome sequencing data identified 1562 recurrent somatic mutations (p-value ≤0.01) in 711 genes, including 304 mutations previously identified in cancer driving genes. The most common recurrent pathways affected within the top 2000 gene mutations with p≤0.02 comprised mainly of NF-κB signaling followed by immune response (antigen presentation by MHC class II and the alternative and lectin induced complement pathways), reverse signaling by Ephrin B, mTOR/PTEN and EGFR/RAS pathways. Based on the lack of canonical NF-κB activation previously reported (Chapman J et al. Leukemia 2015; 29: 2270-2273) and the important role of MYD88-p100 signaling pathway in B cell differentiation into plasmablast (Guo et al. Oncogene 2016. 36(29):4224-4232), we investigated the status of p100 signaling in PL using a published gene expression dataset (Chapman J. et al Leukemia 2015). Our analysis demonstrated that most PL (70%) manifest constitutive p100 signaling. Therefore, we focused on mutations in genes involved in the NF-κB activation. Mutations in the NF-κB pathway were identified in all the analyzed cases with an average of 4 mutated genes in each tumor. Consistent with the previously reported downregulation in RNA expression of genes implicated in the BCR and canonical NF-κB signaling, we found deleterious mutations in genes in the BCR pathway that have been previously reported in lymphomas, including MATL1, FYN and SYK (24%, 16% and 15%, respectively). In contrast, we found frequent gene mutations in the MYD88-PI3P pathway that never have been reported in lymphomas, including SHIP2, DOCK8, PLCG2 in 50%, 39% and 37% of the cases, respectively. These findings are of relevance, as this mutations are expected to results in increased MYD88/TAK1 and BTK signaling (Shinners NP et al J. Imm. 2007, 179 (6) 3872-3880) and can be targeted by specific inhibitors. To evaluate the status of phosphorylation of TAK1 and BTK in these tumors, we performed immunohistochemistry analysis in 15 PL, demonstrating high levels of phosphorylation of these proteins in all tumors analyzed. Conclusion: To our knowledge, this is the first in-depth analysis of PL genome. Our data provide the most comprehensive genetic portrait of PL, provides potential genetic causes of this disease and identify potential druggable targets that deserve further clinical exploration. Disclosures Flowers: National Cancer Institute: Research Funding; Millennium/Takeda: Research Funding; Eastern Cooperative Oncology Group: Research Funding; OptumRx: Consultancy; Denovo Biopharma: Consultancy; Genentech/Roche: Research Funding; V Foundation: Research Funding; Abbvie: Consultancy, Research Funding; Acerta: Research Funding; Celgene: Research Funding; Pharmacyclics/ Janssen: Consultancy; Spectrum: Consultancy; Burroughs Wellcome Fund: Research Funding; Bayer: Consultancy; Karyopharm: Consultancy; Gilead: Research Funding; Genentech/Roche: Consultancy; Pharmacyclics: Research Funding; Janssen Pharmaceutical: Research Funding; Abbvie: Research Funding; BeiGene: Research Funding; TG Therapeutics: Research Funding; Gilead: Consultancy. Lossos:Affimed: Research Funding. Bernal-Mizrachi:Takeda Pharmaceutical Company: Research Funding; Kodikaz Therapeutic Solutions: Consultancy, Equity Ownership.

Description: Introduction: Diffuse large B cell lymphoma (DLBCL) is a clinically heterogeneous disease. While roughly half of the patients respond well to standard R-CHOP therapy, the majority of the remainder succumb to their disease. While many targeted therapies have been developed in DLBCL, resistance to single agents develops almost invariably. While drug combinations have proved to be effective approaches to overcoming resistance in infectious diseases, developing such combinations has proved to be difficult in diffuse large B cell lymphomas and other cancers owing to not only the heterogeneity of these diseases and overlapping toxicity profiles. We hypothesized that with advent of powerful new machine learning approaches combined with genomics, that we would be able to identify novel mechanisms of resistance and develop effective combination therapies to overcome resistance to single agents. Results We tested in vitro responses to all FDA-approved and Phase III cancer drugs (N=150) in six DLBCL cell lines carefully chosen to represent the heterogeneity with regard to cell of origin and common genetic alterations. We observed that roughly half of our drugs were active in at least 50% of the DLBCL cell lines. We then performed RNA sequencing on these cell lines before and after exposure to each of these drugs at their specific IC50 (concentration of drug required to kill 50% of the cells). In addition, we tested the effects of 38 cytokines and antibodies to assess their downstream biological effects. In all, we generated 1167 RNAseq profiles post-exposure to drug (N=900) or cytokines. Hierarchical clustering of our RNAseq data demonstrated clusters of drugs with shared mechanisms and targets (e.g. HDAC inhibitors, PI3K and mTOR inhibitors). We developed a machine learning approach using a combination of neural networks and Bayesian network propagation analysis to identify pathway activation and mechanisms of resistance associated with each of the drugs. Our approach identified 16 combinations of drugs that had different mechanisms and downstream targets. Surprisingly, we found that histone deacetylase inhibitors (HDACi, e.g. panobinostat) were predicted to be strongly synergistic in combination with JAK inhibitors (e.g. ruxolitinib). These findings were unexpected as ruxolitinib had very weak single agent effects and the JAK-STAT pathway is not thought to be specifically associated with response to HDACi. We verified the predictions of the machine learning algorithm by performing in vitro combination assays in six different cell lines. In each case, we found that the combination was highly synergistic using the Chou-Talalay method. We further verified the feasibility and efficacy of combining panobinostat (HDACi) and the JAK inhibitor ruxolitinib in vivo using xenograft models. Both single agents had relatively modest effects on tumor burden, but we found significant synergy with the combination (p