Keats, Jonathan J.

Abstract Abstract 2881 Multiple myeloma (MM) is characterized by a remarkable heterogeneity in outcome following standard and high-dose therapies. Significant efforts have been made to identify genetic changes and signatures that can predict clinical outcome and include them in the routine clinical care. Gene expression profiling (GEP) studies have achieved a central role in the study of multiple myeloma (MM), as they become a critical component in the risk-based stratification of the disease. To molecularly stratify disease-risk groups, we performed GEP on purified plasma cells (obtained from the immunobead selection of CD138+ cells) from 489 MM samples in different stages of the disease using the Affymetrix U133Plus2.0 array. A total of 162 probes were analyzed using an in house automated script to generate a GEP report with the most used risk stratification indices and signatures, including the UAMS 70-gene, UAMS class, TC classification, proliferation and centrosome signature, and NFKB activation indices. In a subset of 57 samples, IgH translocations were analyzed using FISH and results were correlated with GEP data. A macrophage index was calculated and used as a surrogate measurement of non-plasma cell contamination. A total of 49 samples (10%) were excluded from subsequent analysis as the macrophage index indicated a significant contamination with no plasma cells, hence potentially compromising the results. The percent of high-risk disease patients identified from different signatures ranged from 26.4% by using high proliferation index to 28.8% with high centrosome signature and 31.3% with high 70-gene index. This percent of high-risk cases based on the 70-gene index is similar to what was found in Total therapy 2 (TT2) and TT3 cohorts. A third of patients (33.2%) were classified as D1 in the TC class, followed by 11q13 (19.3%), D2 (16.4%), 4p16 (13.8%), MAF (6.1%), None (4.7%), D1+D2 (4.5%) and 6p21 (1.8%). The NF-kB pathway was likely activated in 45.5% to 59.5% of cases, depending on the index used for its calculation. High proliferation index and high centrosome signature significantly correlates with 70-gene high-risk group (p<0.0001). Conversely, the activation of NF-kB pathway was not significantly different between high- and low- risk subgroups. TC subgroups D1 (p<0.0001) and 11q13 (p=0.01) were significantly more common in the 70-gene low-risk group. Similarly, TC subgroups 4p16 (p=0.0004), Maf (p=0.02) and D2 (p=0.05) were enriched in the high-risk group. Translocations t(4;14)(p16;q32), t(11;14)(q13;q32) and t(14;16)(q32;q23) were precisely predicted by the TC classification (100% correspondence). Cases with IgH translocations with unknown partner were classified in subgroups D1 (33%), D2 (25%), 6p21 (25%) and Maf (16%). Here we summarized the associations between the most significant gene expression indices and signatures relevant to MM risk-stratification. The multiple variables simultaneously analyzed in an automated way, provide a powerful and fast tool for risk-stratification, helping in the therapeutic decision-making. Disclosures: Stewart: Celgene: Consultancy, Research Funding; Millennium: Consultancy, Research Funding; Novartis: Consultancy, Research Funding; BMS: Consultancy, Research Funding; Onyx Pharmaceuticals: Consultancy, Research Funding. Fonseca:Consulting :Genzyme, Medtronic, BMS, Amgen, Otsuka, Celgene, Intellikine, Lilly Research Support: Cylene, Onyz, Celgene: Consultancy, Research Funding.

Abstract Human myeloma cell lines (HMCL) provide both a discovery and validation platform to improve our understanding of the molecular pathogenesis of multiple myeloma. We have completed a project to characterize the underlying genetics of all commonly used HMCL with a primary goal of identifying appropriate model systems for findings from large scale patient studies like the multiple myeloma genomics initiative (MMGI). We first purchased all 37 commercially available HMCL from DSMZ, JCRB, ECACC, and ATCC. An additional, 32 non-commercial cell lines were obtained from Diane Jelinek and Leif Bergsagel. Subsequently each HMCL was thawed and cultured under strict parameters, which yielded cells for analysis, by comparative genomic hybridization, whole exome sequencing (Agilent SureSelect V4+UTR) and mRNA sequencing. The combination of these two assays provides a detailed map of the genetic complexity underlying this deadly disease. For variant discovery, alignment was done using BWA followed by realignment, quality recalibration and duplicate removal. High quality calls were identified using Mutect by requiring a minor allele frequency of 0.25 and a minimum depth of coverage of 10x. This identified a median of 81999 mutations per sample with 25th and 75thpercentile values of 76621 and 87440 respectively. To identify likely somatic mutations, we filtered out variants found in dbSNP, 1000 genomes project and the NHLBI Exome Sequencing Project when any of the SNPs within each database had a global minor allele frequency of greater than 0.1%. In addition, we included only those variants in gene space defined by RefSeq or GENCODE and targeted by Agilent exome capture V4+UTR. After these filtering steps a median of 1389 putative somatic mutations remained. To identify key events responsible for mutagenesis, we extracted non-synonymous variants in genes, which show substantially expression in RNASeq(FPKM>5). This identified a median of 371 and 135 variants per sample respectively. Overall, across all the 66 samples, there were at total of 9074 non-synonymous mutations in 4453 genes, which are significantly expressed. To identify potential oncogenes we focused on mutations that occurred at the same position in the genome or altered the same amino acid in expressed genes. This identified 63 genes including KRAS(n=16) and NRAS(n=11) as expected but also BRD9, CCNE1 and MAP2K3. To identify putative tumor suppressors, we looked for genes with homozygous mutations or genes with more than one mutation in the same sample. This identified 112 genes including TP53, TRAF3, FAM46C and DIS3 as expected along with LTB and KDM6B. Independently, to attempt to identify genes involved in MM pathogenesis, genes with mutations having a SIFT score of less than 0.05 which occurred in at least 5%(n>3) of the cell-lines were identified. This identified 185 genes of which 25 are in our putative oncogene list and 37 are in the tumor suppressor list. These study results provide the myeloma community with the unique resource of a fully characterized series of models that can be used for laboratory-based tests. To increase the utility of the dataset to the community all these data are freely available at www.keatslab.org or by request. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 602 The molecular target(s) which co-operate with proteasome inhibition in inducing drug sensitivity or resistance in Multiple Myeloma (MM) remain unknown. We therefore conducted a genome scale small interfering RNA (siRNA) lethality study in KMS11 MM cells in the presence or absence of bortezomib without regard to pre-conceived mechanistic notions. Primary screening was performed in a single-siRNA-per-well format with the human druggable genome siRNA set V4 comprising 13,984 siRNA targeting 6,992 genes and comprising two RNAi per gene. siRNA were transfected at low concentration (13nM) to minimize off-target effects using conditions that resulted in transfection of >95% cells and <5% background cytotoxicity. Bortezomib was added (at the IC10, 25, 70 and 90) 24 hours post transfection. After 96 hours (72 hours after bortezomib), viability was measured by ATP-dependent luminescence. In primary screening 320 candidate bortezomib sensitizing genes were identified and rescreened with four siRNA oligos each gene. 57 of the top sensitizer hits were selected for which at least two distinct siRNA decreased the EC50 by 2 standard deviations from cells treated with control siRNAs. By cross referencing gene expression profile (GEP) data of KMS11, we furthered pared the list to 37 plausibly expressed targets (0.5% of those genes originally screened) as bortezomib sensitizers. After silencing, 50% of these 37 genes also sensitized the lung cancer cell line A549 to bortezomib. After further parsing of genes which also modulated the sensitivity of MM to Melphalan (non specific chemosensitizers), 34 genes remained: The strongest sensitizers to bortezomib were the proteasome subunits PSMA5, PSMB2, PSMB3, PSMB7 but included less obvious targets such as BAZ1B, CDK5, CDC42SE2, MDM4, NME7, TFE3, TNFAIP3, TNK1, TOP1, VAMP2 and YY1 were also identified. Of these, the most potent synergetic effects were observed with siRNAs against the proteasome subunits and against cyclin dependent kinase 5 (CDK5), which caused the greatest shift in EC50. CDK5 is of particular interest as a therapeutic target as it is expressed at high levels in MM and neural tissues but has low expression in other organs. Using viral shRNA expression, silencing CDK5 consistently increased the sensitivity of genetically variable MM cell lines (n=5) to all of the proteasome inhibitors tested: bortezomib, carfilzomib and PR047, and the effect could be at least partially rescued by overexpression of an RNAi resistant CDK5. To explore therapeutic relevance the small molecule CDK5 inhibitor, Roscovitin, was shown to be synergistic or additive with bortezomib in both MM cell lines and primary patient samples. Gene expression profiling was then performed to seek an explanation for the CDK5 sensitization effect and regulation of a proteasome subunit PSMB5 by CDK5 was identified as a probable route to sensitization. In summary inhibition of the existing proteasome either directly by suppression of proteasome subunits, or indirectly by suppression of modulators such as CDK5 appears to confer the greatest sensitization effect suggesting that combinations of bortezomib with other unique proteasome inhibitor drugs or combinations with inhibitors of CDK5 is a logical avenue for clinical exploration. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 606 Multiple myeloma is a complex malignancy with multiple underlying genetic events. Our group has spent considerable effort over the last 15 years to elucidate the genetic underpinnings of myeloma. Most recently, we used array-based comparative genomic hybridization (aCGH) as a discovery tool in 62 myeloma patients and 46 myeloma cell lines. In that preliminary screen using the Agilent 44B aCGH platform (∼70kb resolution) we identified a diverse array of abnormalities, which resulted in constitutive activation of the NF-kB pathways. That initial analysis concentrated on the 43 genes we identified as potential targets of the 13 homozygous deletion events detected in the patient samples. A pathway analysis of these genes revealed a single pathway involving TRAF3, TRAF2, BIRC2, BIRC3, and CYLD. This first analysis focused exclusively on abnormalities present in the patient samples as we worried some abnormalities identified exclusively in the cell lines might not be relevant to the pathogenesis of myeloma in patients. However, several abnormalities were equally or more frequent overall but occurred exclusively in cell lines including CDKN2C (14 samples), CDKN1B (4 samples), KDM6A/UTX (4 samples), RB1 (3 samples), TP53 (3 samples). Given the fact that KDM6A/UTX deletions were as frequent as many of the best-described tumor suppressors it seemed like a good candidate but in the absence of patient events or a known function at the time it was not prioritized for further study. Recently, as part of the Multiple Myeloma Research Consortium (MMRC) Genomics Initiative, we have completed the analysis of a cohort of 250 myeloma patient samples by aCGH using the Agilent 244A aCGH platform (∼15kb resolution) and gene expression profiling using the Affymetrix U133Plus2.0 genechip. In this cohort with a significantly improved aCGH platform we identified 17 genes that are recurrently inactivated by homozygous deletions including DIAPH2 (15 samples), CDKN2C (14 samples), TRAF3 (11 samples), CYLD (8 samples), BIRC2/3 (7 samples), KDM6A/UTX (6 samples), and RB1 (5 samples). Based on the significant improvement in resolution and data quality achieved with the Agilent 244A aCGH platform we rescreen all of the cell lines on this improved platform. This significantly changed the frequency of several homozygous deletions in this population with the most frequently targeted genes now being CDKN2C (20 samples), KDM6A/UTX (13 samples), DIAPH2 (7 samples), RB1 (4 samples), TP53 (4 samples), CDKN1B (4 samples), and TRAF3 (4 samples). Moreover, as part of the genomic characterization of a spontaneous myeloma mouse model that we have developed, Vk*-Myc, we have identified recurrent (∼50%) homozygous deletions of Kdm6a/Utx. Therefore, one of the genes most commonly targeted by a homozygous deletion in human and mouse myeloma is KDM6A/UTX. In late 2007 after we had identified the first patients with KDM6A/UTX deletions it was shown that UTX is a functional histone demethylase that removes methyl groups from histone H3 lysine 27 (H3K27me). Given the high incidence of deletions and the fact that MMSET, the overexpressed target gene of t(4;14) in myeloma, is predicted to methylate H3K27, H3K36, and/or H4K20 by evolutionary conservation we developed the hypothesis that myeloma is characterized by abnormalities that result in excessive H3K27me (typically a repressive chromatin mark). Given the extensive whole genome sequencing occurring in the MMRC genomics initiative we elected to focus our resequencing efforts on the cell lines exclusively. These studies identified an additional 4 samples with LOH and an inactivating mutation bringing the total percentage of inactivated cell lines to 33%. Clearly, in the expanded patient and cell line cohorts the inactivation of KDM6A/UTX is not independent of MMSET overexpression suggesting they may act independently or synergistically. We are currently attempting to identify the genes controlled by KDM6A/UTX inactivation to better understand the functional consequences of this highly recurrent event. However, in the mouse model unlike the patient or cell lines, the gene expression profiling has identified a gene expression signature that differentiates UTX inactivated and functional samples suggesting an oncogenic function of inactivation. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 604 We have conducted a pilot study to interrogate both CpG hyper- and hypomethylation in Multiple Myeloma (MM) using the Illumina GoldenGate (GG) BeadArrayTM high-throughput assay. Specifically, we utilized the Illumina Methylation Cancer Panel I, which spans 1,505 CpG loci selected from 807 genes including known tumor suppressor genes and oncogenes. To date we have profiled 50 CD138+ MM specimens, 4 MM cell lines and 6 normal CD138+ plasma cell samples (each representing a pool of 3-5 individuals) obtained from the bone marrow of patients who have undergone hip or knee replacement surgeries. We have obtained highly reproducible and reliable DNA methylation profiles with an average r2 of greater than 0.98 when the β values were compared between technical replicates. Further, we generated an internal control panel for the array by performing methylation specific PCR (MSP) to determine the methylation status of several genes found on the array (CDKN2A, E-Cadherin, MGMT, hMLH1) in 4 MM cell lines. A differential methylation analysis was conducted between normal plasma cells and MM samples using the BeadStudio Methylation Module v3.2. Differential scores for each CpG locus on the array were generated that correspond to significant differences between groups. Scores greater than (hypermethylation) or less than (hypomethylation) 13 represent ‘Diff scores' that are statistically significant (p values < 0.05) and this cut-off threshold is used to determine differentially methylated loci. Globally, when comparing MM to normal plasma cells, hypomethylation occurred at 42% of probes on the array, and was more common than hypermethylation, which occurred at only 21% of probes. Interestingly, 77% of probes situated in non-CpG islands were hypomethylated compared to 26% or probes in CpG islands. On the contrary hypermethylation occurred mainly in CpG islands (26% of probes) whereas only 7.4% of probes in non-CpG islands were considered hypermethylated. A total of 946 CpG loci were differentially methylated between normal plasma cells and MM. Among the most significantly differentially methylated CpG loci between normal plasma cells and MM were FZD9, ALOX12, MMP-3, MMP-8, and p16. Array comparative genomic hybridization (aCGH) data availabel for this same sample set was used to generate a hyperdiploid index for each sample by identifying those samples with whole chromosome gains at minimally three odd numbered chromosomes. There were 751 CpG loci that were differentially methylated between hypderdiploid and non-hyperdiploid tumors with WEE1, FRZB, ELK1, CLDN4, and SHH being among the most significant. We also applied the pathway enrichment analysis tool GeneGo v5 (Agilent) to these data, which identified cell adhesion/ECM remodeling, developmental pathways and several immune response pathways as the most significantly enriched pathways between normal and plasma cells with ephrin signaling pathways in the context of cell adhesion as the most significantly enriched pathway between hyperdiploid and non-hyperdiploid tumors. We hope to integrate methylation data with orthogonal datasets including gene expression, aCGH, and DNA sequence data being developed as part of the MMRF Genomics Initiative. Further validation of hits will also be followed up by functional studies to assess the true biological relevance of differentially methylated genes. This study represents one of the first methylome interrogation studies in MM and points towards global hypomethylation as possibly an important mechanism driving myelomagenesis. Our data also suggest that differential methylation of genes may be a significant determinant of hyperdiploid biology. Determining the set of critical genes and pathways based on the myeloma methylome is likely to lead to an improved understanding of biological mechanisms involved in myelomagenesis, ultimately leading to new and improved methods of diagnosing and treating this disease. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 4160 We propose that the pathophysiology of marginal zone B-cell lymphomas (MZL) could allow for their classification into lymphoid organs located in MZL and extra-nodal MZL of MALT type; the former category encompassing splenic MZL (SMZL) and nodal MZL (NMZL). The rarity of these lymphomas and difficulty in distinguishing them from other low-grade B-cell lymphoma such as lymphoplasmacytic lymphomas hampers further understanding of disease biology. In an attempt to better characterize the genetic basis of these entities and compare acquired genetic changes based on this categorization we performed array-based comparative genomic hybridization (using the 244A array, Agilent) in a cohort of MZL cases. In total we analyzed 101 MZL, including 46 MALT (including 8 cases with t(11;18)(q21;q21)), 35 SMZL and 20 NMZL. Overall, 90.1% cases have copy-number abnormalities (CNA). Genomic complexity was greatest in MALT without t(11;18)(q21;q21)(4 CNA), followed NMZL (3.5 CNA), SMZL (3 CNA) and MALT with t(11;18)(q21;21)(1 CNA). Seventeen minimal deleted regions (MDRs) and 9 minimal amplified regions (MARs) were defined based on the presence in at least one entity with a frequency of 10% or higher. Gains of chromosomes 3, 12 and 18 were found in all entities. Deletion of 6q was the most common abnormality in MALT (21.1%) and, to less extent, SMZL and NMZL. Biallelic and focal monoallelic deletions on 6q23.3-q24.1 identified TNFAIP3 as the sole gene inside the common MDR between entities. Deletion of chromosome 13 was observed in all entities, but NMZL and MALT were characterized by whole chromosome deletions, whereas focal deletions on 13q14.3 (MIR16-1/MIR15A) were identified in SMZL. A subset of abnormalities was shared for more than one entity: deletions 9p21.3 (CDKN2A) and 11q21-q22 (ATM) were found in NMZL and SMZL but not in MALT; deletion 7q32.1-q33 was the most common CNA in of SMZL but was also found in MALT; gain of 6p was found in MALT and NMZL but not in SMZL. Other CNA were only found in one entity such as deletion of 17p13.3-p12 (TP53) in SMZL, deletion of 15q25.3-q26.2 in NMZL. In addition to 17% of cases with the t(11;18)(q21;q21), extra copies of MALT1, focal amplifications of REL and/or monoallelic/biallelic deletions of TNFAIP3 were identified in an additional 46% of MALT lymphomas. These CNA were identified alone or in combination, but all of them were mutually exclusive with the t(11;18)(q21;21), thus totalizing 63% of MALT lymphomas with abnormalities affecting regulator genes from the NF-kB signaling pathway. Moreover, depending on the anatomic site of the MALT lymphoma, the prevalence of CNA affecting the NF-kB pathways varies from 50% in lacrimal glands to 75% in salivary glands. Abnormalities affecting regulator genes of the NF-kB signaling pathways were identified in a significantly lower frequency in NMZL (30%) and SMZL (31%) entities, suggesting its differential role in the pathogenesis of different entities. Using this high-resolution approach we show that while many abnormalities are shared unique patterns of genomic gains and losses characterize subgroups of MZL. Extranodal MZL seems to have a higher propensity of constitutive genetic upregulation of NF-kB pathways. Furthermore, this study remarks the importance of this pathway also in the subgroup of MALT lymphomas without translocations affecting the pathway. Disclosures: Fonseca: Genzyme: Consultancy; Medtronic: Consultancy; BMS: Consultancy; AMGEN: Consultancy; Otsuka: Consultancy; Celgene: Consultancy, Research Funding; Intellikine: Consultancy; Cylene: Research Funding; Onyx: Research Funding; FISH probes prognostication in myeloma: Patents & Royalties.

Abstract Background: IMiDs are now recognized to promote the proteasomal degradation of IKZF1/3 and the transcriptional repression of MYC and IRF4. MYC locus rearrangement is a recurrent somatic event in MM and results in MYC repositioning near superenhancers (such as IGH@, IGL@ and others). The mechanisms of IMiDs-mediated downregulation of MYC in MM cells is not well understood, in particular it is unclear how IMiDs alter the activity of the super-enhancers driving MYC transcription. Methods and results: Transcriptome analysis (RNAseq) of MM cells (OPM2 and MM1S) treated with lenalidomide or upon the silencing of IKZF3 (lentiviral delivery of dox-inducible IKZF3 shRNA) confirmed the downregulation of MYC, IRF4 and among others the upregulation of other genes of interest such as CD38, DKK1, PDL1, FOXO3 and a host of genes involved in interferon response. Gene set enrichment analysis (GSEA) confirmed the enrichment of genes signatures associated with MYC as well as the interferon response. Notably, in t(4;14) MM cell lines we observed a significant downregulation of FGFR3 while the expression of WHSC1 was not affected. This finding was confirmed in a library of t(4;14) MM cell lines by qRT-PCR analysis. As FGFR3 is driven by the 3' IGH enhancer while WHSC1 is under the control of the intronic Em enhancer in t(4;14) MM cells, and in view of the known role of IKZF1/3 in class switch recombination, we postulated that IKZF1/3 regulate the 3' IGH enhancer activity and MYC expression in cells harbouring a MYC-IGH rearrangement. To validate this hypothesis we established the genome-wide distribution of IKZF1 in OPM2 and MM1S cells (both cell lines harbour an IGH-MYC rearrangement) by ChIP-Seq. Ikaros bound to 17660 loci of which 43% were associated with gene targets. Ikaros-binding motifs and other motifs such as STAT1, E2A, RUNX1 were also identified (MEME Tomtom and Jasper motif analysis) in the vicinity of Ikaros enrichment peaks. Importantly IKZF1 peaks also mapped to the IGH 3' enhancer loci and these results were confirmed by ChIP-PCR analysis. Of interest while no Ikaros peaks mapped to MYC (within 5 kb of TSS) by ChIP-seq, modest enrichment of Ikaros at MYC promoter was identified by ChIP-PCR in MM1S and KMS11 cells, and this enrichment was significantly attenuated by lenalidomide treatment. Since inhibition of MYC occurs as a consequence of depletion of the acetylated chromatin reader BRD4 at enhancers that drive MYC expression (Delmore et al., 2011), we examined the interaction between BRD4, IKZF1 and IMiDs treatment. Chip-Seq analysis revealed that BRD4 and IKZF1 are associated with most active enhancers and promoters in MM1S tumor cells. In particular, BRD4 and IKZF1 were enriched at the IGH 3'α2 enhancer locus. Importantly, treatment with lenalidomide (10μM, 4 hours), dramatically depleted IKZF1 and BRD4 enrichment at the 3' IGH enhancers (ChIP-PCR analysis). In contrast, treatment with the bromodomain inhibitor JQ1 depleted BRD4, but not IKZF1 from the IGH 3' enhancer locus. These findings suggest that IKZF1 regulates BRD4 binding to the 3' IGH enhancers and explain the selective effect of IMiDs on MYC transcription in transformed cells harboring a MYC rearrangement. In addition, these results suggest that IMiDs will affect MYC expression only in cells were MYC is driven by an IKZF1/IKZF3 responsive enhancer. Lastly we sought to explain how IKZF1 may regulate BRD4 enrichment at H3K27Ac marks in super-enhancer loci. Since Ikaros is a known integral component of the NuRD (nucleosome remodelling and histone deacetylase) complex, we examined whether lenalidomide treatment and the ensuing Ikaros depletion lead to local gain of NuRD at MYC TSS and depletion of H3K27 acetylation. CHD4 (helicase subunit of the NURD complex) and H3K27Ac ChIP-PCR analysis revealed significant accumulation of the NuRD complex at MYC TSS as well as reduced H3K27Ac at 3'IGH enhancers after IMiDs exposure in MM1S and OPM2 cells. Importantly, CHD4 silencing abrogated lenalidomide induced MYC transcriptional repression and cell death. Conclusion: In summary, our studies suggest that Ikaros drives MYC expression in MM cells harbouring an IGH-MYC rearrangement by regulating the accumulation of the acetylated chromatin reader BRD4 at the IGH 3' enhancer and preventing the NURD complex mediated depletion of H3K27Ac marks at these enhancers. Our findings also explain the IKZF1/3 dependent mechanism of IMiDs-mediated MYC transcriptional repression. Disclosures Neri: Celgene: Research Funding. Duggan:Jansen: Honoraria; Celgene: Honoraria. Keats:Translational Genomic Research Institute: Employment. Bahlis:Celgene: Consultancy, Honoraria, Research Funding, Speakers Bureau; Johnson & Johnson: Consultancy; Johnson & Johnson: Speakers Bureau; Johnson & Johnson: Research Funding; Amgen: Consultancy.

Abstract Abstract 1923 Our understanding of the genetic abnormalities associated with the development of multiple myeloma has increased significantly in the last decade. However, very little is known about how, or if, myeloma tumor genomes change with time and if therapeutic interventions influence these events. To address these issues we studied a cohort of 29 patients for whom at least two serial samples (1-65 months, median 19 months) were available for analysis. Each serial pair was analyzed by both array-based comparative genomic hybridization (aCGH) and microarray gene expression profiling (GEP) to identify DNA copy number abnormalities (CNA) at a 25kb resolution and gene expression differences present in the bulk of the tumor mass. Though this does not address the intra-clonal heterogeneity that may exist at a given time point, it does answer if the bulk of the tumor mass is changing with time. This study has unearthed several surprising and clinically relevant findings. First, myeloma tumor genomes are not as unstable as previous cytogenetic analyses suggest. In 40% of patients we observed no detectable CNA changes (1-37 months, median 12 months). In 24% of patients we observed the exclusive acquisition of new CNA (1-12, median 3.5) (3-22 months, median 18 months). In 36% of patients we observed both the loss (1-20, median 3) and gain (1-33, median 21) of CNA (5-43 months, median 20 months). Because time was not a significant influence on the detection of stable or unstable genomes we compared CNA changes with TC class and found patients with the high-risk 4p16 and maf IgH translocations were over-represented in the latter subset of patients. These observations raise the question of what happens between multiple rounds of therapy and if different regimens influence these phenotypes differently. For two patients with no CNA changes between the first two time points there was an additional sample that extended the follow-up by 52 and 12 months. Again no CNA changes were seen between diagnosis and these final samples taken 63 and 50 months later. For one patient with CNA changes (5 shared, 29 lost, and 32 gained) we have a detailed time course of 5 samples from diagnosis through to end-stage plasma cell leukemia. This patient received continuous lenalidomide-dexamethasone (Rd) for 20 months and progessed with a clone containing a BIRC2/3 deletion, which activates the NFKB pathway. The patient received single agent PR-171 and a bortezomib containing regimen and unexpectedly, the tumor genome observed in the third sample was almost identical (32 shared, 2 lost, and 4 gained CNA) to the first time point, including two copies of BIRC2/3. Subsequently, the patient received melphalan-prednisone-bortezomib (MPV) and the tumor genome observed in the fourth and fifth samples, which were identical, were similar to that seen in the second sample (24 shared, 13 lost, and 39 gained CNA). To understand these observations better we performed FISH to ascertain if the observed clones were detectable earlier, albeit at a low frequency. These experiments proved that the two dominant subclones observed at time points 1 and 3 versus 2, 4, 5 were mutually exclusive at the single cell level. Moreover, both of these clones were detectable at diagnosis with 12% of the tumor mass being the second subclone that eventually evolved into plasma cell leukemia. Interestingly, we assayed 5 of the 39 unique CNA observed in the final two samples and only one, the 17p13 deletion, was detectable earlier. This suggests the MPV regimen effectively eliminated a clone that was previously sensitive to Rd and selected for a dramatically evolved subclone that was previously sensitive to two different proteasome inhibitors. Although it is clear that the high-risk patients are enriched in the subset with the most changes, it is not clear if the specific drugs used (Melphalan vs IMID vs proteasome inhibitor) or intervention strategy (Cycled vs continuous/maintenance) and perhaps the response achieved (PR vs CR) influences these events. These observations do highlight two important clinical concepts that need to be considered in the future. First, the meaning of a partial response needs further investigation as this may reflect effective elimination of one subclone but not another. Second, because some patients are not changing or can revert back to a previous subclone we need to consider re-chanllenging patients with previously effective regimens when patients progress. Disclosures: Fonseca: Genzyme: Consultancy; Medtronic: Consultancy; BMS: Consultancy; AMGEN: Consultancy; Otsuka: Consultancy; Celgene: Consultancy, Research Funding; Intellikine: Consultancy; Cylene: Research Funding; Onyx: Research Funding; FISH probes prognostication in myeloma: Patents & Royalties. Stewart:Millennium: Consultancy; Celgene: Honoraria. Bergsagel:Amgen: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Genentech: Membership on an entity's Board of Directors or advisory committees; Millennium: Speakers Bureau; Novartis: Speakers Bureau.

Abstract Abstract 1246 Poster Board I-268 B-cell chronic lymphocytic leukemia (B-CLL) is characterized by a variable clinical course. Genomic aberrations as detected by FISH analysis and the mutational status of the IgVH genes have been found to have prognostic clinical significance with unmutated IgVH genes associated with clinically more aggressive disease. Previous studies have suggested that unmutated IgVH B-CLL is associated with increased chromosomal instability and increased chromosomal aberrations leading to accelerated disease progression. We recently performed a high-throughput, high-resolution array-based comparative genomic hybridization (aCGH) in a cohort of B-CLL patients. Fifty-eight patients have been analyzed by using the Sureprint G3 microarray (one million probes, Agilent). Our working hypothesis was that we would see increased genomic aberrations and instability in the unmutated IgVH group. The total number of copy number aberrations and total megabases (Mb) deleted or gained per genome were used as a surrogate marker for genomic instability. Variations in chromosomal aberrations in various VH families were also evaluated. B-CLL with unmutated IgVH genes utilized members of the VH1, VH3, VH4 and VH5 families with VH1-69 the most prevalent. B-CLL with mutated IgVH genes utilized VH families VH1, VH2, VH3, VH4, and VH6. Considerable variation was seen in the number of genomic abnormalities between cases. Overall, 315 abnormalities have been identified (236 deletions and 79 gains), with a median of 4 copy-number abnormalities per patient (range 0-32). In patients with IgVH mutational status available the median number of chromosomal aberrations in the mutated group was 5 (range 1-17) and unmutated group was 4 (range 0-20). Copy number deletions were significantly more common than chromosomal gains for the total group and in both the unmutated and mutated IgVH gene groups. Mean number of megabases (Mb) of DNA deleted in the unmutated IgVH group was 17.80Mb (standard deviation 24.17 Mb, median 6.63 Mb, range 0 - 90.55Mb) and 40.65 Mb in the mutated IgVH group (standard deviation 86.04 Mb, median 15.35Mb, range 0 – 307.57Mb). Mean number of Mb gained in the unmutated group was 37.74Mb (standard deviation 57.58Mb, median 0.44Mb, range 0 – 132.47Mb) and 32.99Mb (standard deviation 59.76Mb, median 0.14Mb, range 0 – 172.50Mb) in the mutated group. Nine patients had B-CLL with VH1-69 genes and 8 of them were unmutated. Median number of chromosomal deletions was 2 and gains also 2 (ranges 0 – 8 deletions and 0 – 4 gains) for this group. Similar findings were seen for the other VH families. These results initially suggest that total copy number variations do not account for the observed clinical differences between mutated and unmutated IgVH B-CLL. The catalogue of potentially clinically relevant chromosomal abnormalities found in this aCGH study is being currently evaluated. Disclosures Zent: Genentech, Bayer, Genzyme, Novartis: Research Funding. Shanafelt:Genentech: Research Funding; Hospira: Membership on an entity's Board of Directors or advisory committees, Research Funding; Polyphenon E International: Research Funding; Celgene: Research Funding; Cephalon: Research Funding; Bayer Health Care Pharmaceuticals: Research Funding. Kay:Genentech, Celgene, Hospira, Polyphenon Pharma, Sanofi-Aventis: Research Funding; Biogenc-Idec, Celgene, Genentech, genmab: Membership on an entity's Board of Directors or advisory committees. Fonseca:BMS: Consultancy; Amgen: Consultancy; Otsuka: Consultancy; Celgene: Consultancy; Medtronic: Consultancy; Genzyme: Consultancy.

Abstract Abstract 2810 Poster Board II-786 We wished to explore the genetic events associated with disease progression and development of drug resistance in multiple myeloma (MM). To do so 11 patients were studied in whom at least two (range 2-3) temporally distinct samples of tumor DNA and RNA were available. The baseline genetic initiating event was defined for all patients (3 were genetic high risk; one with t(14;16) two with t(4;14)) as well as the gene expression profile (GEP) defined risk score using the Little Rock 70 or 17 gene panel (only one, the t(14;16) was GEP defined high risk). High resolution array CGH and gene expression were then performed on each sample. Of the 8 patients with a “low risk” tumor initiating event and low risk GEP score, 6 patients had no, or only one, copy number abnormality (CNA) change between the two temporally distinct MM samples. In stark contrast the 3 genetic high risk at baseline had between 17 and 40 distinct CNA changes at the time of progression. For all 11 patients 89 CNA were acquired with progression whereas 19 previously abnormal regions disappeared suggesting clones with these abnormalities were extinguished by the therapy received. In total we detected 0-40 CNA changes between the various timepoints, median 1, mean 10.7. The acquisition of new CNA was much more common than the loss of CNA. We then focused more specifically on the t(4;14) patient with the highest number of CNA changes. This patient has a well documented clinical course of having a sustained two year VGPR to Len/dex and then progressing while still taking Len/dex. Comparison of the pre and post-Len/dex samples identified 40 CNA changes(the most of any pair studied to date). Only six CNA were shared between the two samples, which included deletions of chr4, 9, 12, 13, and X plus a t(4;14) translocation. These likely represent the initiating “driver” tumor events. The new CNA we identified originated from both remodeled genomic changes and the emergence of unique changes, indicating a new tumor clone had emerged while the previously dominant clone had regressed (e.g. a deletion of a large segment of chromosome 8 at diagnosis was no longer observed in the relapse sample). The newly acquired CNA encompassed 3968 genes (13.7% of the genes in the genome), however, only 1235 of these genes (4%) were expressed in this patient at diagnosis (1188 in the typical myeloma patient). Since 1235 genes is still a large number we hypothesized that whole genome sequencing (WGS) would help elucidate the mechanism of lenalidomide resistance. We isolated DNA from germline tissue and CD138 purified tumor cells including: diagnostic, first relapse and second relapse samples. Utilizing SOLiD (Applied Biosystems, Foster City, CA) sequencing technology, we have completed fragment library WGS on both the germline and the final tumor samples. Quality control measures report the average number of sequence reads per start point to be less than 1.2, indicating the library is primarily composed of unique molecules. In addition, approximately 40% of the sequence reads map uniquely to the genome. Together these quality measures indicate our sample libraries are complex and provide good representation of the genome. Data on the whole genome sequence of myeloma at diagnosis and at the time of progression will be presented. Disclosures: Bergsagel: Celgene: Consultancy.