Henderson, Kimberly J.

Abstract Activation of NFkB has been noted in late B-cell malignancies, but no genetic basis for this activation has been shown except for rare mutations of the NFkB pathway in lymphoma and none have been described in MM. Bortezomib, highly tumoricidal in a subset of MM patients, inhibits NFkB suggesting it’s role as a therapeutic target. Here we show for the first time the critical importance of mutations that dysregulate this potential survival pathway in human myeloma cell lines (HMCL) and new diagnosis MM patients. We performed high-density oligonucleotide array CGH (Agilent 44k) on CD138+ selected MM cells from 68 patients and 42 unique HMCL. Initially we focused on small loci, bi-allelically deleted, as markers for inactivation of potential tumor suppressor genes. We next developed FISH probes for the identified loci to confirm the bi-allelic deletions seen by aCGH. We next sequenced all samples with monoallelic deletions, and additional unselected patients, and identified additional inactivating mutations. Multiple novel tumor suppressor genes (TSG) were identified and validated, whose loss of function (both in patients and HMCL) leads to activation of the non-canonical NFkB pathway including: TRAF2 (9q34), TRAF3 (14q32), BIRC2/BIRC3 (11q22) and CYLD (16q12). Functional studies showed that HMCL with deletion or mutations in these genes have markedly increased NFkB2 p52/p100 ratios relative to other HMCL (and for BIRC2/BIRC3 relative to an isogenic line that retains one copy of BIRC2/BIRC3), indicating genetic activation of the non-canonical NFkB pathway. Reintroduction of deleted TSG into HMCL inhibited cell growth and resulted in cell death (see data on abstract by J. Keats et al). We next focused our analysis (of combined aCGH and GEP data) on positive regulators of this pathway. Rare IgH translocations of NIK have recently been identified in MM. By aCGH we identified amplification and rearrangements of NIK. We identified dysregulated expression of TNF family receptors that activate the non-canonical pathway: CD40 (translocation), TACI (amplification) and LTBR. All of the HMCL with these mutations also showed increased NFkB2 p52/p100 ratios, consistent with activation of the non-canonical NFkB. These mutations correlate with a gene expression profile of NFkB activation that is detected in one half of patients. In summary, we have determined for the first time the genetic basis for constitutive NFkB activation - present in 19/42 HMCL, and we estimate in at least one quarter of newly diagnosed patients (approximately one half of those with evidence of NFkB activation by GEP). We propose a mechanism in which positive (TACI, CD40, LTBR) and negative regulators (TRAF2, TRAF3, BIRC2/BIRC3, CYLD) of the pathway converge on NIK to mediate the processing of NFKB2 p100 to p52. The preponderance of genetic evidence suggests that drugs that target the non-canonical NFkB pathway, will be most effective in treating the 50% of MM patients with NFkB activation.

Abstract H-MM affects about 40–50% of patients, and a similar proportion of MGUS/SMM. The genetic evolution of the disease and the clinical implications of the co-existence of 13q- or t(14q32) are unknown. We conducted a global survey of genetic abnormalities in H-MM in order to describe genetic evolution pathways. H-MM cases from the Mitelman cancer genetic database were combined with cases from the Mayo Clinic MM genetic database. All abnormalities (regional/whole chromosome gains/losses, breakpoints and translocations) were plotted against a 344-chromosome band profile created in a matrix format, and the number of genetic abnormalities per tumor (NAPT) logged. The abnormalities present in more than 5% of tumors were considered recurrent and each karyotype was scored for the presence or absence of each of the recurrent abnormalities. To investigate the temporal pattern of acquisition of genetic abnormalities, all tumors with a given abnormality were selected and the distributions of NAPT plotted. The modes of these distributions were used as a value of ‘lateness’, referred to as time of occurrence (TO). The impact of co-existence of 13q- and t(14q32) on progression free survival (PFS) and overall survival (OS) in H-MM was determined for MM patients entered into an Eastern Cooperative Oncology Group chemotherapy trial. Four hundred and sixty-nine H-MM patients were included in the karyotypic analysis. The median chromosome number was 53 (range 48–74) and the median NAPT was 10 (range 1–25). The distribution of NAPT revealed a normal distribution, suggesting an obligatory acquisition of a given number of genetic abnormalities. The most common abnormalities were trisomies of chromosomes 3,5,7,9,11,15,19 and 21 affecting a third or more of the patients. Deletions and abnormalities of even-numbered chromosomes occur in 20% or less of patients. None of the more than 300 specific translocations detected were recurrent. However, 5 translocation breakpoints were recurrent including 14q32, 8q24 and 11q13. TO analysis suggest that for H-MM, the trisomies are early genetic events, translocations occur midway and deletions tend to be later events. Amongst the trisomies, correlation analysis does not suggest a particular order of acquisition. Clustering analysis using principal component analysis suggests that in H-MM, IgH translocations are secondary events associated with disease evolution rather than defining unique biological entities. Consistent with this, survival of H-MM patients is not affected by 13q- status but is negatively impacted by the presence of t(14q32) especially those with unknown partners (p<0.05). The data presented suggest that H-MM fit a proposed aneuploid model of cancer due to auto-catalytic chromosome reassortment (Duesberg et al Cell Cycle2003:2:202–210). The presence of tumor specific trisomies and the apparent similar phenotype induced by various combinations of trisomies are consistent with this model.

Abstract Background: The free light chain (FLC) assay allows a nephelometric measurement of this free kappa and lambda light chains that circulate as monomers or dimers, not bound to the immunoglobulin heavy chain. Estimation of the free light chain levels and their ratio has utility in screening, prognostication and treatment response assessment in multiple myeloma, and is particularly useful in light chain myeloma. Previous studies have suggested that absence of a functional rearrangement on at least one of the IgH allele is the basis of light chain myeloma. Translocations involving the immunoglobulin heavy chain (IgH) locus on chromosome 14 can be detected using FISH in over half of the patients with myeloma. We hypothesized that presence of an IgH translocation may be associated with compromised production of heavy chains and leads to unbalanced, excess production of clonal light chain. Methods: From February 1988 to May 1992, 653 patients with previously untreated multiple myeloma were enrolled on the phase III clinical trial E9486. Serum free light chains were run on stored sera obtained at baseline in 495 patients as part of a previous study. Among these patients, 314 patients were previously studied using FISH for presence or absence of IgH translocations, p53 abnormalities and deletion 13 and were included in the current analysis. Results: The median difference between the involved and uninvolved FLC (FLCdiff) was higher among the group of patients with any IgH translocation abnormality compared to those with none (Figure 1). This was particularly evident among the patients with a t(14;16) who had the highest FLCdiff. In a multivariate analysis, we examined if the prognostic value seen with FLCdiff was independent of the presence of IgH translocations t(4:14) and t(14;16), which are associated with a worse outcome. The presence of t(4;14) or t(14;16) was associated with FLCdiff groups using a cutoff point of 185, with 13% of those with FLCd ≤185 and 27% of those with FLCd > 185 having the translocations (P= 0.008). At the cutoff value of 185, the prognostic value of the FLC-diff on progression free and overall survival appears to be independent of these high risk IgH translocations (P-value=0.03 and 0.003, respectively for PFS and OS) We then examined the prognostic value of the FLC-diff in the group of patients without these two high risk abnormalities and were able to confirm that they remained prognostic for PFS (P=0.009) and OS (p=0.005). However, among the group of patients with either of the two high risk IgH abnormalities, FLCdiff estimate added little to assessment of their outcome (P=0.97 and 0.20, respectively for PFS and OS). Conclusion: Elevated free light chain levels are associated with presence of IgH translocations and these patients are more likely to harbor the high risk genetic abnormalities. Routine FISH testing for patients with newly diagnosed myeloma will allow us to confirm these and allow better risk stratification of the patients. Additional work needs to be done to help us better understand the molecular basis for unbalanced light chain excretion by myeloma cells and the current finding of the association with IgH translocation abnormalities should form the basis for this work. Figure 1. Distribution of FLC ratio (inv vs. uninv) and FLC difference (inv − uninv) among patients. Solid squares represent patients with t(4;14) or t(14;16) or t(11;14) or other 14q32 translocation; open circles represent patients without t(4;14), t(14;16), t(11;14) and any other 14q32 translocations. The horizontal dashed lines refer to FLC ratio original values of 100 and 1000, and vertical dashed lines refer to FLC difference original values of 10 and 100. Solid line indicates significant association (P<=0.05) between the dichotomized FLC groups and translocation status. Figure 1. Distribution of FLC ratio (inv vs. uninv) and FLC difference (inv − uninv) among patients. Solid squares represent patients with t(4;14) or t(14;16) or t(11;14) or other 14q32 translocation; open circles represent patients without t(4;14), t(14;16), t(11;14) and any other 14q32 translocations. The horizontal dashed lines refer to FLC ratio original values of 100 and 1000, and vertical dashed lines refer to FLC difference original values of 10 and 100. Solid line indicates significant association (P<=0.05) between the dichotomized FLC groups and translocation status.

Abstract Table 1. Samples to p53 deletions p 53 deletion FISH aCGH FISH and aCGH MGUS 142 - - SMM 108 - - Newly diagnosed MM - 182 - Relapsed/refractory MM 62 156 19 HMCLs - 48 - Total 312 386 19 Background: Inactivation of p53 by mutation or allelic loss is a rare event in multiple myeloma (MM) at the time of disease diagnosis and believed to be more common in the late stages of the disease. Here we defined the prevalence of p53 deletions in monoclonal gammopathy of undetermined significance (MGUS), smoldering multiple myeloma (SMM), newly diagnosed MM, relapsed MM and human myeloma cell lines (HMCLs). Indirect mechanisms of p53 inactivation such as amplification of MDM2, or deletion of CDKN2A (p14ARF) were also investigated. Patients and Methods: A total of 631 MM samples and 48 HMCLs were analyzed for p53 abnormalities (Table 1) and compared with a cohort of plasma cell leukemia (PCL) previously published by us. Interphase fluorescence in situ hybridization (FISH) and highresolution array-based comparative genomic hybridization (aCGH) were used to detect p53 deletions. MDM2 amplification and p14ARF loss were evaluated by aCGH. Overall survival (OS) was estimated through Kaplan Meier method. Results: The prevalence of p53 deletions was 1.4%, 3.7%, 8.9%, 20% and 87.5% in MGUS, SMM, newly diagnosed MM, relapsed MM and HMCLs respectively (Figure 1). Of interest patients in first relapse showed a p53 deletion prevalence of 18% in comparison with 33% for patients in second relapse or more. We obtained p53 deletion status in 8 patients noted to have deletion at the time of relapse and in whom we had previous samples stored. In 7 of 8 cases deletions had been acquired (absent in the previous samples). The median OS for relapsed MM patients with or without p53 deletion were 4.2 months and 37.8 months respectively (p <0.001). MDM 2 amplification and p14ARF loss was detected in 0.4% and 2% of new diagnosed MM, 0.6% and 5.1% of relapsed MM and 2% and 29% of HMCLs respectively. Discussion: Emergence of p53 deletion/mutations denotes progression genetic events in MM. Importantly we show for the first time the overriding importance of p53 deletions as prognostic markers in relapsed MM. A molecular staging system that is based on the presence of p53 inactivation may be more powerful than classifying patients based on the loss of p53. Other abnormalities involving p53 pathway regulators like MDM2 gain and p14ARF loss are infrequent events in MM, even in advanced disease. Figure 1. Prevalence of p53 deletions Figure 1. Prevalence of p53 deletions

Abstract Constitutive activation of NF-κB signaling pathways has been identified in a variety of malignancies, however the molecular basis has remained largely unknown. Recently, we have shown mutations in 10 genes encoding positive and negative regulators of NF-κB signaling pathways in 20% of multiple myeloma (MM) patients, with TRAF3 being the most commonly affected gene. TRAF3 is a negative regulator of NF-κB signal and its inactivation results in the constitutive activation of the alternative or noncanonical NF-κB signaling pathway. With the exception of two recent studies in MM, there is no available data regarding the mechanisms that dysregulate NF-κB in other B-cell malignancies. The aim of this study was to analyze the TRAF3 status in a cohort of Waldenström’s macroglobulinemia (WM) patients. Forty-nine WM were analyzed by using a combination of array-based comparative genomic hybridization (aCGH) and cIgM-FISH. Genome sequencing from 28 of these patients was performed on all coding exons of TRAF3 and the adjacent intron-exon junctions. A NF-κB index, as a measure of NF-κB transcription activity, was calculated from 21 patients with gene expression data. The index is the mean expression level of four probe sets differentially expressed between human myeloma cell lines with identified mutations compared to those without identified NF-κB pathway mutations (CD74, IL2RG and 2x TNFAIP3). By performing aCGH we identified a small monoallelic deletion (∼460 Kb), including the TRAF3 locus in one of 22 patients. Genome sequencing identified a missense mutation (D483N) in the remaining allele. This substitution affects a highly conserved amino acid throughout vertebrate evolution, suggesting its functional relevance. We used cIgM-FISH to screen 27 patients with no available aCGH data and we found one biallelic TRAF3 deletion and one chromosome 14 monosomy. In the later case, no DNA was available to sequence the remaining allele. Both samples with biallelic inactivation of TRAF3 possess the two highest NFKB index values of our WM cohort, suggesting NF-κB transcription activity. Moreover, the case with biallelic deletion showed the lowest level of TRAF3 gene expression. Additionally, the entire TRAF3 coding region was sequenced in 28 of these 49 patients and one homozygous nonsense mutation (K286X) was identified. No GEP was available in this sample to calculate the NF-κB index. TRAF3 was only recently implicated in a malignant process, when inactivating mutations and biallelic deletions were reported in approximately 13% of MM patients, making it the most frequently inactivated tumor suppressor in MM. Our study has identified biallelic TRAF3 inactivation in at least 3 of 49 WM patients, suggesting that the constitutive activation of NF-κB signaling pathways may be involved in the pathogenesis of WM. Recent findings showing an association between TRAF3 loss of function and good response to bortezomib in MM patients suggest that proteasome inhibitor-based therapy may be an attractive option in the treatment of the WM patients with constitutive activation of NF-κB signaling pathways. Additional mutational screening of TRAF3 and other regulators of these pathways are ongoing to determine the extent of constitutive activation present in WM and other B-cell malignancies.

Abstract INTRODUCTION The bone marrow plasma cell labeling index (PCLI) and the beta 2 microglobulin (β2M) when combined provide a powerful staging system highly predictive of outcomes in myeloma patients (Greipp et. al., Blood, 81:3382–87, 1993). This staging system lacks information on cytogenetics and it has been demonstrated that deletion 13 (Δ13) by conventional cytogenetics is associated with poor outcomes. In conventionally treated patients Δ13 by FISH is an independent prognostic variable. We analyzed the importance of Δ13 by FISH in a cohort of patients receiving high dose therapy and assessed its value relative to known prognostic parameters, PCLI and β2M. PATIENTS AND METHODS 238 patients undergoing high dose therapy between January 1990 and September 2001 were studied. The minimum follow-up of patients was 34 months. The median age of the cohort 56 years. Interphase fluorescence in situ hybridization with Probes LSI13/D13S319 was performed with simultaneous immunofluorescent detection of bone marrow plasma cells by the presence of cytoplasmic immunoglobulin. PCLI was done by a slide-based immunofluorescence method with antibodies to bromodeoxyuridine. RESULTS Of the 238 patients studied 215 had a successful FISH analysis and 212 of these had a bone marrow labeling index. Deletion 13 was absent in 104 patients (48%) and present in 111 patients. The overall survival and time to progression for patients lacking and having deletion 13 was 47.8 vs 26.4 (OS) and 19.0 vs 11.7 (TTP)months respectively. A Cox model was constructed with the PCLI, the presence or absence of deletion 13 and serum β2M. Although all three variables were independently predictive of survival and time to progression, the β2M was the weakest of these three variables. The hazard function for survival for PCLI, deletion 13 and β2M was 1.4, 1.6 and 1.1 respectively. A staging system was constructed assigning 1 point for the presence of deletion 13 or a PCLI ≥ 1%. Patients fell into 3 categories of 0 points (N=69), 1 point (N=104) and 2 points (N=39). The overall survival for the three groups respectively were 56.5, 36.6 and 13.7 months (figure). The time to progression for the three groups were 22.3, 15.4 and 8 months (P<0.001). When analysis was limited to those patients transplanted within 12 months of diagnosis the results remained significant (P<0.003) with median survivals of stage 0, 1 and 2 of 72.5, 59.1 and 15.4 months respectively. TTP for stage 0, 1 & 2 in this same cohort of patients transplanted < 12 mos after diagnosis is 30.6, 21.5 & 8.2 months respectively. CONCLUSION Deletion 13 by FISH is an independent predictor of survival and time to progression providing information independent of the β2M and the PCLI. Combining the PCLI with FISH del 13 identifies patients who following transplant have a median time to progression of 8 months and a median OS of 13.7 months. This group may not benefit from autologous transplant and are candidates for novel therapeutic approaches. Figure Figure

Abstract Events mediating transformation from the pre-malignant MGUS to MM is largely unknown. Previous studies have highlighted RAS mutation as a progression event. However, this abnormality is present in only a small subset of patients. Global gene expression differences between MGUS and MM have been reported but provided little coherent biological insights. To define progression events, we examined the differences in gene expression profiles (GEP, Affymetrix U133A genechip) between 101 MM and 22 MGUS using a powerful functional genomic tool, Gene Set Enrichment Analysis (Subramanium et al PNAS 2005), that allow extraction of biological information based on enrichment of genes associated with a phenotype within gene-sets derived experimentally or extracted from database. We found that a predominant theme relating to MM progression is the over-expression of proliferation related genes in MM highlighted by the enrichment of cell cycle related gene-sets. Another common theme is the enrichment of gene-sets relating to MYC activation. Seven of eight gene-sets related to MYC activation in the database are significantly enriched (p-value <0.05 and false detection rate <0.05). Leading edge (enriched genes) analysis of enriched MYC-related and proliferation-related gene-sets revealed a core, overlapping signature of 8 genes specific to MYC and not proliferation. These genes are further verified as MYC transcriptional targets using Metacore, a network analysis tool. In addition, 7 of the 8 genes had been validated as MYC target genes by chromatin immunoprecipitation previously (www.myccancergene.org). We validated this signature in independent publicly available cell lines and mouse gene expression experiments where this signature is present in mammary epithelial cell lines engineered to over-express MYC but not E2F3, SRC and BCAT, and mouse embryonic fibroblast over-expressing MYC but not HER2 or with RB knocked-out. In addition, analysis of proliferation index in these datasets showed that there is dissociation of proliferation from this signature. This showed conclusively that our signature is MYC specific and not related to proliferation. Of interest, the signature is also present in cell lines with activating RAS mutation, which also have high expression of MYC, suggesting that MYC activation is a downstream effect of activated RAS signaling. Indeed, in our patient cohort, most patients with RAS mutation express this signature. When applied to 2 large GEP MM patient cohorts, this signature was not present in normal plasma cells, rarely present in MGUS, and present in a significant proportion of MM and almost all the human myeloma cell lines. When analyzed in the context of other B-cell malignancies, the signature is very strong in Burkitt’s lymphoma, present in most myeloma but absent in chronic lymphocytic leukemia, Waldenstrom Macroglobulinemia, and normal B-cells. Our analysis therefore suggests that MYC activation may be a unifying pathological event in MGUS to MM progression. As MYC translocations occur in only about 13% of newly diagnosed MM patients, activating RAS mutations and other trans-mechanisms may mediate MYC activation. Our recent demonstration that sporadic MYC activation in a mouse strain prone to monoclonal gammopathy led to MM development provides important functional proof.

Introduction: While the molecular target of immunomodulators such as pomalidomide (POM) and lenalidomide (LEN) has been identified, the mechanisms underlying therapeutic resistance remain incompletely understood. The uniformly emerging resistance to therapy over time in the absence of identifiable cereblon pathway mutations in the majority of patients raises questions about alternative mechanisms including aberrant gene expression. Methods: We performed gene expression profiling using an Affymetrix GeneChip Human Genome U133 Plus 2.0 microarray on CD138+ bone marrow cells from patients with relapsed / refractory (RRMM) and newly diagnosed (NDMM) multiple myeloma prior to initiating treatment. Patients were treated on two phase II clinical trial protocols (MC0789: POM ± dexamethasone in RRMM; MC0884: LEN ± dexamethasone in NDMM) between 2007 and 2012. We categorized patients based on their IMWG response as non-responders (SD) and responders (VGPR+). We selected 15 responders and 15 non-responders from MC0789 (n = 30) and compared overall survival, gene expression patterns, and involved cellular pathways between the two groups. We selected 5 responders and 5 non-responders from MC0884 (n = 10) for targeted validation of differentially expressed candidate genes. After data quality control and normalization of gene expression values, differential gene expression was estimated using limma. Statistical significance was adjusted for multiple testing in the discovery set using a false discovery rate-based approach for genome-wide experiments (q-value). We used Gene Ontology and PANTHER pathway analysis for functional annotation of differentially expressed genes. Overall survival estimates were calculated using the Kaplan-Meier method. Computation and visualization were performed in R. Results: Median age at treatment initiation on MC0789 was 65 years (40 - 82), 65% of the patients were male. Pomalidomide resistance was associated with an increase in mortality (median overall survival 1.6 versus 6.4 years, p = 0.009, Kaplan-Meier plot). There were 1076 differentially regulated genes between responders and non-responders (521 up- and 555 down-regulated, q < 0.050 for all genes, volcano plot). Expression of CRBN was 1.5-fold down-regulated in non-responders (q = 0.005). Supervised hierarchical clustering of the top 500 differentially expressed genes demonstrated distinct patterns in pomalidomide-resistant disease (heatmap). Gene ontology analysis revealed protein synthesis as one of the most enriched biological processes (bar graph). Pathway analysis showed a 6-fold enrichment (FDR = 0.007) of the ubiquitin proteasome pathway in pomalidomide-resistant disease. Differentially expressed genes involved key protein degradation pathways, epigenetic modifiers, and transcription factors. Targeted validation in MC0884 revealed 13 common genes with at least 1.5-fold differential expression (5 up- and 8 down-regulated), 12 of which have previously been implicated in the regulation of apoptosis, tumor glucose metabolism, Rho and Wnt signaling, miRNA-driven resistance to chemotherapy, and ubiquitin-dependent protein degradation (Table and Sankey diagram). The most up-regulated gene in non-responders was MYRIP, a gene coding for a vesicle trafficking protein associated with platinum resistance and suppression of pro-apoptotic BCL-2 family members in solid malignancies. The most down-regulated gene was FRZB, a gene coding for a negative regulator of Wnt signaling, previously implicated in the progression of monoclonal gammopathy of undetermined significance to multiple myeloma. Conclusions: Overall survival of patients with pomalidomide-resistant RRMM remains poor. Pomalidomide resistance was associated with differential gene expression in several potentially targetable cellular pathways beyond the known drug target cereblon. Targeted validation of candidate genes revealed common cellular pathways in immunomodulator-resistant disease. Elucidating the exact molecular mechanisms underlying immunomodulator resistance is of considerable interest for biomarker development and the identification of novel therapeutic targets and warrants further exploration. Figure Disclosures Lacy: Celgene: Research Funding. Dispenzieri:Celgene: Research Funding; Alnylam: Research Funding; Intellia: Consultancy; Janssen: Consultancy; Pfizer: Research Funding; Akcea: Consultancy; Takeda: Research Funding. Kumar:Takeda: Research Funding; Celgene: Consultancy, Research Funding; Janssen: Consultancy, Research Funding.

Introduction: Recent advancements in the treatment of multiple myeloma (MM) have centered on engaging the immune system to target multiple myeloma cells. Although these therapies are being combined with immunomodulatory imide drugs (IMiDs) and corticosteroids, the individual contributions of these drugs on the immune system of MM patients has not been examined in the upfront setting. In this study, we examined the peripheral blood immunophenotypes of newly diagnosed multiple myeloma (NDMM) patients receiving the IMiD lenalidomide with or without the corticosteroid dexamethasone. Methods: To characterize immunophenotypes, we utilized flow cytometry to profile white blood cell populations from 35 patients enrolled in a clinical trial testing the efficacy of lenalidomide with and without dexamethasone in NDMM (NCT00772915). In this trial, all patients were initiated on single-agent lenalidomide. Dexamethasone was initiated in patients that did not meet desirable responses or for disease progression. At each cycle, peripheral blood was stained with a 17-marker antibody panel against several immune lineages and functional surface markers. We grouped patients into two groups: 1) lenalidomide alone or 2) lenalidomide with dexamethasone according to their treatment regimen at each cycle timepoint. Results: First we confirmed anti-myeloma cell activity for both groups by measuring a steady decline in circulating plasma cells in both groups. Examining peripheral blood immunophenotypes showed an expected decrease in T cells and a smaller decrease in B cells in both groups of patients. Closer inspection of B cell populations revealed a switch towards a more immature B cell phenotype in both treatment groups. This was measured as a switch from CD19-CD20+ cells to CD19+CD20- B cells. Inspection of T cell subsets revealed that patients receiving single-agent lenalidomide had a sustained decrease in the levels of CD4+ T cells and increase in the levels of CD8+ T cells. This was seen in both naïve and regulatory T cells evidenced by a decrease in the CD4/CD8 ratio among CD28+ T cells as well as CD25+ T cells. Importantly, this alteration did not lead to sustained alterations in the overall level of CD25+ or CD28+ T cells, and the addition of dexamethasone reverses this trend. In addition to the effects seen on T and B cell numbers, we detected expansions of NK cell populations in patients receiving lenalidomide alone. This expansion is detected as an overall increase in CD56+ mononuclear cells with the majority of cells being CD56+CD3- cells. Conclusions: Our data show that lenalidomide and dexamethasone therapy have shared but distinct effects on peripheral blood immunophenotypes in NDMM. Both drugs alter B cells numbers and populations leading to an expansion of CD19+CD20- B cells. However, lenalidomide alone decreases the CD4/CD8 T cell ratio; and, lenalidomide more strongly expands NK cell populations. The addition of dexamethasone reverses this trend and leads to a restoration of the CD4/CD8 ratio. This suggests that lenalidomide without dexamethasone might be counterproductive in immunotherapies intended to recruit CD4+ T cells. Conversely, lenalidomide alone could increase the efficacy of immunotherapies dependent on NK cell recruitment such as antibody-dependent cellular cytotoxicity (ADCC). This information may benefit future investigations of immune responses in MM patients and improve the adoption of immunotherapies to MM patients. Figure Disclosures Kumar: Celgene: Consultancy, Research Funding; Janssen: Consultancy, Research Funding; Takeda: Research Funding.

Introduction: High-risk multiple myeloma can be defined by the presence of specific cytogenetic abnormalities (structural) or by characteristic changes in bone marrow and peripheral blood biomarkers (functional). While both entities are characterized by therapeutic resistance, frequent disease relapses, and adverse survival outcomes, the underlying molecular mechanisms remain incompletely understood. Methods: We performed gene expression profiling (GEP) using an Affymetrix GeneChip Human Genome U133 Plus 2.0 microarray on CD138+ bone marrow cells from 137 patients diagnosed with multiple myeloma between 2004 and 2012. All patients underwent Fluorescence In-situ Hybridization (FISH) evaluation, plasma cell labeling, International Staging System (ISS) risk stratification, and GEP prior to initiating treatment with novel agents. The presence of del(17p), t(4;14), t(14;16), and t(14;20) on FISH, a plasma cell labeling index (PCLI) > 2%, and ISS stage III were considered high-risk abnormalities: FISH-HR (n = 15, structural high-risk, at least one high-risk FISH lesion), PCLI-HR (n = 20; functional high-risk, PCLI > 2%), and ISS-HR (n = 12; functional high-risk, ISS stage III). For each HR group we sampled standard risk (SR) controls in a 4:1 ratio. After data quality control and normalization, differential gene expression was estimated using limma. Statistical significance was adjusted for multiple comparisons using a false discovery rate-based approach for genome-wide experiments (q-value). We employed PANTHER pathway analysis for the differentially expressed genes in each HR group. We implemented a simple gene expression score (GES) by calculating the sum of quartiles of the normalized gene expression values for genes differentially expressed in more than one HR group (GES = ΣUP(quartile - 1) + ΣDN(4 - quartile)) and externally validated its prognostic significance (UAMS TT2 / TT3, GSE24080). Survival outcomes were analyzed using the methods described by Kaplan, Meier, and Cox. Computation and visualization were performed in R. Results: Median age at diagnosis was 63 years (32 - 87), 53% of the patients were male. High-risk disease was associated with inferior overall survival, regardless of the used definition (left Kaplan-Meier plots): FISH-HR (HR 4.3, 95% CI 1.9 - 9.8, p < 0.001), PCLI-HR (HR 2.7, 95% CI 1.4 - 5.3, p = 0.004), and ISS-HR (HR 2.8, 95% CI 1.2 - 6.5, p = 0.015). There were 59 (FISH-HR), 424 (ISS-HR), and 507 (PCLI-HR) differentially expressed genes (q < 0.050 for all genes, volcano plots). PCLI-HR and FISH-HR demonstrated a predominance of transcriptional up-regulation while ISS-HR had a balanced gene expression profile with a similar number of genes being up- and down-regulated. The involved cellular pathways were different across the HR groups except for anti-apoptotic signaling (bar graphs). All HR groups had distinct gene expression profiles with no complete overlap between all HR groups. There were 71 genes with overlap between two HR groups (69 up-regulated, 2 down-regulated, Venn diagrams). The median GES was 97 (18 - 206, higher numbers indicating higher expression of up-regulated and lower numbers of down-regulated high-risk genes) in 559 patients treated on UAMS TT2 / TT3 (GSE24080). Tertiles of the GES were associated with event-free survival (HR 1.4, 95% CI 1.2 - 1.6, p < 0.001) and remained independently prognostic after adjusting for age, sex, and ISS stage (HR 1.3, 95% CI 1.1 - 1.5, p < 0.001). Conclusions: High-risk multiple myeloma remains associated with inferior overall survival, regardless of the used definition (structural or functional). The subtypes of high-risk disease have distinct gene expression profiles and involve different cellular pathways, providing important clues to the underlying biology. A 71 gene signature derived from the different high-risk subtypes was of prognostic significance in a clinical trial population after adjusting for known prognostic factors. Figure Disclosures Lacy: Celgene: Research Funding. Dispenzieri:Akcea: Consultancy; Intellia: Consultancy; Janssen: Consultancy; Pfizer: Research Funding; Takeda: Research Funding; Celgene: Research Funding; Alnylam: Research Funding. Stewart:Takeda: Consultancy; Seattle Genetics: Consultancy; Roche: Consultancy; Ono: Consultancy; Celgene: Consultancy, Research Funding; Ionis: Consultancy; Janssen: Consultancy, Research Funding; Oncopeptides: Consultancy; Amgen: Consultancy, Research Funding; Bristol Myers-Squibb: Consultancy. Bergsagel:Celgene: Consultancy; Ionis Pharmaceuticals: Consultancy; Janssen Pharmaceuticals: Consultancy. Kumar:Celgene: Consultancy, Research Funding; Janssen: Consultancy, Research Funding; Takeda: Research Funding.