Link, Brian K.

Abstract Abstract 259 Despite recent progress, our understanding of the genomic basis of diffuse large B cell lymphoma (DLBCL) is incomplete. We have performed massively parallel whole exome sequencing with deep coverage of 57 primary tumor samples from patients with DLBCL and compared them to matched normal tissue to gain insight into the genetic defects leading to DLBCL. We define genes that are significantly enriched for somatic mutations, and thus, may be drivers of the disease. In addition to identifying genes that are well established to be functionally relevant in DLBCL, such as MyD88, CARD11, EZH2, and BCL2, we also discovered new genes that are significantly mutated that have not been described to play a role in DLBCL previously, such as MLL2, 3 and4 genes and PCLO. One of the challenges for the understanding of DLBCL biology is to determine the relative contribution of different genes with somatic mutations to the pathogenesis of the disease in an individual patient. Identification of functionally relevant mutated genes that only occur in a small number of patients based on prevalence alone may require thousands of patient samples. We therefore devise several strategies to discover genes with a high likelihood of functionally relevant mutations independent of their prevalence. Genes that do not co-occur with other mutated genes in the same patients, i.e. are mutually exclusive, are more likely to represent strong drivers, and to occur early in the development of DLBCL. We determined that TP53, BCL2, PIM1, and CD79B fall into this group. We apply pathway-focused analysis to identify mutated genes with low prevalence that are highly likely to have a functional impact. For instance, we identified G13D mutations in KRAS in two patients, and mutations in BRAF, NOTCH1, SOCS1, and PTEN in other patients that were previously described in other malignancies. These mutated genes are identified as part of significantly enriched gene sets, including the B cell receptor signaling/NFkB pathway, JAK/STAT, and antigen processing gene sets. In addition to these pathways that are important for physiological B cell activation, we find that many histone methyltransferases are enriched as a geneset in addition to MLL2. We identified a number of hotspots that are subject to somatic mutation. Our data imply that many of these hotspots may be targets of activation-induced cytidine deaminase (AID), which is usually involved in somatic hypermutation, suggesting a possible mechanism of how somatic mutation in DLBCL is mediated. In summary, our data define novel genetic defects in DLBCL by whole exome sequencing of 57 patients. Our data suggest that driver mutations may not be randomly distributed across the genome, but may occur in a directed fashion. We identified new genes not previously implicated in DLBCL, and have developed strategies to prioritize mutations that are most likely to be drivers of the disease. Disclosures: No relevant conflicts of interest to declare.

Abstract Background Relatively few studies have evaluated the role of host (germline) genetics in predicting outcome in diffuse large B-cell lymphoma (DLBCL), and of these, most were small candidate gene studies without replication and/or from the pre-R-CHOP era. We report the first genome-wide association study (GWAS) of outcome in DLBCL patients treated with rituximab and anthracycline-based chemotherapy. We used two independent studies for discovery, and validated the results in two additional independent studies. Methods In the discovery phase, we conducted a meta-analysis of two GWAS studies of DLBCL patients treated with rituximab and anthracycline-based chemotherapy, the first study as part of the prospective clinical trials of the LNH2003B program (N=540) of the Lymphoma Study Association (LYSA) and the second as part of the Molecular Epidemiology Resource (N=316), a prospective observational study from the University of Iowa/Mayo Clinic Lymphoma Specialized Program of Research Excellence (SPORE). Genotyping for the GWAS was performed using the Illumina 610 (LYSA) and 660 Quad (SPORE) BeadArrays (Illumina, San Diego, USA). HapMap2 was used to impute additional SNPs. Association testing was conducted assuming a log-additive genetic model and using Cox regression analyses. We calculated the hazard ratios (HRs) and 95% CIs for event-free survival (EFS), with adjustment for age, sex and age-adjusted IPI. The two studies were then combined using the fixed effects inverse variance method based on the beta estimates and standard errors from each study. From the meta-analysis, 89 SNPs were selected for validation, based on a minor allele frequency (MAF) >0.05 and LD pattern for multiple SNPs from the same region. Validation cohorts were an independent set of DLBCL patients from the SPORE (SPORE-2, N=366) and DLBCL patients from the prospective GOELAMS 075 phase III study (N=294), and validation genotyping was conducted using the Illumina VeraCode platform. We then conducted a meta-analysis of the 4 studies. We report SNPs with both a statistical significance of p<5x10-6 and effect sizes in the same direction across the 4 studies in the meta-analysis. For the latter SNPs, we also report the results for overall survival (OS), adjusted for age, sex and age-adjusted IPI. Results In the final meta-analysis of all four studies for the prediction of EFS, the top SNP was rs7765004 (MAF=0.30; HR=1.45; p=9.7 x 10-8) in the 6q21 region, which is near the MARCKS (myristoylated alanine-rich protein kinase C substrate) and HDAC2 genes. MARCKS encodes for a protein involved in cell mobility, while HDAC2 belongs to the histone deacetylase family. Additional SNPs from this region included rs6918103 (MAF=0.25; HR=1.38; p=7.6 x 10-6), which marks an EZH2 histone binding site, and rs11969684 (MAF=0.12; HR=1.43; p=3.3 x 10-5). The next hit was rs7712513 (MAF=0.31; HR=1.40; p=3.2 x 10-7), which is near arginine-fifty homeobox pseudogene 1 (ARGFXP1) on chromosome 5q23.2. This was followed by a region on chromosome 8q13.3, with the SNP rs9298183 (MAF=0.29; HR=1.41; p=1.4 x 10-6) from LOC100132891 showing the strongest association, along with 3 other SNPs from this region (rs2035376, rs6472637, and rs1600797; all p<5 x 10-6). These SNPs are near the musculin gene (MSC) a transcriptional repressor of E2A frequently silenced by promoter methylation in DLBCL. Finally, rs2253986 (MAF=0.27; HR=1.39; p=3.4 x 10-6) at chromosome 16q12.2, along with rs2253971 and rs6499810, were associated with EFS. These SNPs are located in CES5A (carboxylesterase 5A), which encodes a member of the carboxylesterase family. The family members are responsible for the hydrolysis or transesterification of various xenobiotics. The top SNPs associated with EFS also predicted OS, including rs7765004 (HR=1.41; p=1.2 x 10-5), rs7712513 (HR=1.39; p=8.4 x 10-6), rs9298183 (HR=1.36; p=3.5 x 10-5), and rs2253986 (HR=1.35; p=5.2 x 10-5). Conclusions Although our top SNPs on did not reach genome-wide significance (p<5x10-8), they showed clear consistency across the four large prospective studies and predicted both EFS and OS independently of IPI. The strongest findings were from loci on 6q21, 8q13.3, 5q23.2, and 16q12.2, and several genes from these regions are involved in methylation and xenobiotic metabolism, supporting a role for host genetic variation in these pathways in DLBCL prognosis. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 996 Background: Treatment with R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) has improved survival in diffuse large B-cell lymphoma (DLBCL) lymphoma, although a significant percentage of patients do not achieve a remission or they relapse. One potential explanation for this observation is that host pharmacogenetic background may impact metabolism, detoxification and transport of R-CHOP. However, there are few prospectively collected data that address pharmacogenetics in DLBCL, particularly for event-free and overall survival in the rituximab era. We therefore tested the hypothesis that functional candidate single nucleotide polymorphisms (SNPs) from genes involved in the metabolism of CHOP (CYP2C19, CYP3A5, ABCB1, ABCC2, ABCG2, RAC2, CYBA, NR3C1, GSTA1, GSTP1, TP53) and rituximab (FCGR2A, FCGR3A) influence prognosis in DLBCL. Methods: We genotyped 19 SNPs from 13 genes in a prospective cohort of newly diagnosed DLBCL patients with germline DNA enrolled at the Mayo Clinic and University of Iowa from 2002–2008 as part of the Molecular Epidemiology Resource of the Iowa/Mayo Lymphoma SPORE. All patients were systematically followed through 2009 for overall survival (OS) and event-free survival (EFS), defined as disease progression, retreatment or death due to any cause. All SNPs were genotyped on the Illumina Golden-Gate platform. Cox regression was used to estimate Hazard Ratios (HRs) and 95% confidence intervals (CI) for individual SNPs with OS and EFS. Each SNP was modeled with the most prevalent homozygous genotype used as the reference group. An ordinal test was used to assess the trend across genotypes, and a p<0.05 was considered statistically significant. All Cox models adjusted for International Prognostic Index (IPI) and treatment. Results: The median age at diagnosis of the 439 DLBCL patients was 62 years (range 18–92). The IPI was distributed as follows: 0–1 (34%), 2 (26%), 3 (24%) and 4–5 (16%). All patients were treated with chemotherapy, and 91% received R-CHOP. Through 2009, there were 171 events (39%) and 122 (28%) deaths, with a median follow-up for living patients of 49 months (range 1–98). The SNP rs1045642 in the ATP-binding cassette (ABC) protein gene ABCB1 was associated with OS (p=0.012). The variant T allele for ABCB1 rs1045642 (minor allele frequency (MAF)=0.47) is associated with a silent I→I substitution, and the T allele predicts low enzyme expression. Compared to CC homozygotes, patients with CT (HR=0.66, 95% CI 0.44–0.99) or TT (HR=0.53, 95% CI 0.31–0.90) genotypes had superior OS. We also observed an association of the GSTP1 SNP rs1695 with OS (p-trend=0.018). The variant G allele (MAF of 0.31) is associated with an A→V substitution, and this change predicts low enzyme expression. Compared to AA homozygotes, patients with the AG (HR=0.78, 95% CI 0.54–1.14) or GG (HR=0.39, 95% CI 0.17–0.90) genotypes had superior OS. Nearly identical associations for these two SNPS were observed with EFS. There were no significant associations with any of the other SNPs that we evaluated. Conclusions: In a large series of newly diagnosed DLBCL patients treated in the rituximab era, we found evidence that genetic variation in ABCB1 and GSTP1 is associated with improved OS and EFS after adjusting for IPI and treatment. The variant allele for ABCB1 is associated with slower enzyme activity, favorably impacting doxorubicin and vincristine pharmacodynamics, and the variant allele for GSTP1 is associated with slower detoxification of doxorubicin and vincristine, increasing bioavailability of these drugs. A comprehensive evaluation of the role of pharmacogenetics in DLBCL is warranted. Support: P50 CA97274, R01 CA129539. Disclosures: No relevant conflicts of interest to declare.

Abstract We have recently reported that event-free survival at 24 months from diagnosis (EFS24) is a robust endpoint for assessment of disease related outcome for newly diagnosed patients with diffuse large B-cell lymphoma (DLBCL) treated with anthracycline-based immunochemotherapy (IC). Patients who achieve EFS24 following IC have a subsequent overall survival that is equivalent to the age and sex matched general population. The IPI was developed in the pre-rituximab era using OS/EFS endpoints. While the IPI is associated with outcome in the IC era, the original survival estimates for the IPI risk groups are outdated. The IPI is also a discrete risk classifier, with only 6 possible groups. Here we remodel the IPI using EFS24 as the endpoint in a large international cohort of patients treated with IC. This model provides an individual level 0-100% risk of failing to achieve EFS24 for each patient and can be used for patient prognosis, treatment stratification, and risk assessment. Methods Newly diagnosed DLBCL patients treated with immunochemotherapy from the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource, NCCTG N0489 clinical trial, GELA LNH2003B program, and a Lyon, France based hospital registry were included in the original EFS24 study. Patients with a relapse, progression, unplanned re-treatment, or death due to any cause within 730 days of diagnosis were considered a failure for the EFS24 endpoint. Patients alive and progression/relapse free at 730 days from diagnosis with no subsequent treatment following initial IC were considered a success for the EFS24 endpoint. Variables selected for potential model inclusion were the IPI components (age, stage, ECOG PS, number of extranodal sites, and LDH), as well as standard clinical variables typically reported in studies of DLBCL: presence of B-symptoms, bone marrow involvement, bulky disease (>10 cm), and sex. The IPI24, a prognostic model for EFS24, was developed using logistic regression models. Model building began by fitting the classic IPI score (0-5), then allowing individual IPI variables to carry different model weights and number of categories, followed by dropping insignificant variables, finally adding any additional significant variables. Model performance was assessed using bias corrected c-statistics, AIC, and R2 values. Model selection was done using the complete case set with multiple imputation on the full dataset used to refine parameter estimates for the final model. Results 1596 total patients were available for modeling. The median age was 62 (range 18-93) and 53% were male; 33% were IPI 0-1, 25% were IPI 2, 25% were IPI 3, and 17% were IPI of 4-5. The failure rate for achieving EFS24 was 29%. All variables but sex (p=0.33) were univariately associated with EFS24 (p<6.4x10-5). The original IPI had good prognostic ability for EFS24 (c=0.696, p=2.2x10-32). Allowing multiple categories for ECOG PS, stage, and age (continuous) improved model fit and prognostic ability (c=0.727). The number of extranodal sites was not significant when IPI variables were modeled separately and was removed from the model. Sex and bulky disease were additional significant variables when added to the model. Thus, the final IPI24 model contained the following risk predictors: male sex, ECOG PS (0 vs. 1 vs 2 vs 3-4), stage (I vs. II, vs. III vs. IV), LDH (elevated vs. normal range), bulky disease > 10cm, and age (continuous). This model resulted in improved fit and prognostic ability (c=0.749) over the standard IPI (c=0.696, bootstrap p<0.001). Model validation was performed on an additional 158 patients from the SPORE MER and showed good concordance (c=0.697). Additional validation is planned in a larger set of independent patients. Conclusion The IPI24, generated using standard clinical variables, yields an individual risk prediction for failing to achieve EFS24 following IC for treatment of DLBCL. The IPI24 has superior prognostic ability to the standard IPI and can be implemented in the clinic using a nomogram or as a simple electronic application. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 1591 Background: The serum free light chain (FLC) assay quantitates free kappa and free lambda immunoglobulin light chains. We previously showed elevated FLC are present in approximately 1 in 3 diffuse large B cell lymphoma (DLBCL) patients from both a phase II cooperative group clinical trial and a large clinic based epidemiology cohort. Approximately 10% of DLBCL patients had elevated FLC that was monoclonal in these series. Patients with monoclonal elevated FLC have very poor outcome with almost 50% of patients progressing in the first 15 months following initial treatment with immunochemotherapy. Unlike patients with polyclonal FLC elevation, where a variety of host effects may cause a rise in both kappa and lambda, it is likely the excess FLC in monoclonal patients is directly related to the tumor. DLBCL are a heterogenous group of tumors that can be classified into germinal center (GCB) or activated B cell (ABC) groupings using immunohistochemistry (IHC) staining algorithms. Here we examine the association of monoclonal free light chains with ABC vs. GCB DLBCL tumors. Methods: Newly diagnosed DLBCL patients were prospectively enrolled in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER) or NCCTG clinical trial N0489. Serum FLC was quantitated from enrollment research serum draws using the Freelite FLC assay (The Binding Site, Ltd., Birmingham, UK). Monoclonal elevated FLC was defined as an elevated kappa or lambda with a corresponding abnormal FLC ratio. Immunohistochemistry (IHC) staining was performed on paraffin tissue from research tissue microarrays (TMAs). Results: IHC tissue staining was available from 12 patients with monoclonal elevated FLC unrelated to renal failure. Germinal center markers were predominantly negative: 2 of 11 were LMO2 positive (18%), 1 of 12 was CD10 positive (8%), and zero cases out of 11 were GCET1 positive. FoxP1 was positive in 5 of 9 cases (56%) and MUM1 was positive in 4 of 12 cases (33%). By the Tally algorithm, 10 of 11 cases (91%) were ABC type DLBCL The percentage of ABC patients by the Tally algorithm in the corresponding DLBCL patients without monoclonal FLC was 45% (N=108, p=0.004). Conclusions: Approximately 10% of DLBCL patients have elevated monoclonal FLC and these patients have poor prognosis. Elevated monoclonal FLC may be a serum marker for ABC-type lymphoma, as the tumors from patients are predominantly negative for germinal center stains and ABC by the Tally algorithm. The secretion of light chains by the tumor suggests an activated B cell clone and the poor prognosis is characteristic of ABC type DLBCL and. Given the likelihood of the excess monoclonal FLC being tumor related, the potential exists for tracking of tumor response and progression via serum FLC in these patients. Replication of these findings in an additional cohort of DLBCL patients is planned. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 991 Background: The outcome of patients with DLBCL has significantly improved with the addition of rituximab to CHOP chemotherapy. While standard clinical prognostic factors such as those included in the International Prognostic Index (IPI) still predict outcome in DLBCL in R-CHOP treated patients, predicting the outcome of patients might be further refined using biological factors. We therefore tested whether pretreatment serum cytokines, that measure the biological activity of the tumor as well as the host response, could provide additional prognostic information in DLBCL patients. Methods: Newly diagnosed DLBCL patients were prospectively enrolled in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER). Serum cytokines were measured from pre-treatment blood draws using a multiplex ELISA (Upstate Cell Signaling Solutions, Lake Placid, NY). Thirty cytokines, including pro-inflammatory, Th1 and Th2 associated cytokines, were analyzed using the Luminex-100 system Version 1.7 (Luminex, Austin, TX). Data analysis was performed using the MasterPlex QT 1.0 system (MiraiBio, Alameda, CA). The assay was also performed on research serum draws from 400 non-lymphoma epidemiology controls enrolled on a corresponding lymphoma etiology case-control study. The upper limit of normal for each cytokine was defined as the 95th percentile from the control distribution. Clinical data was abstracted using a standard protocol for the MER and patients were followed systematically for event-free (progression, retreatment, or death due to any cause) and overall survival (EFS and OS, respectively). Cox proportional hazards models were used to evaluate the association between cytokines and outcome. Principal components analysis (PCA) was performed on the combined DLBCL and control cytokine dataset, agnostic to clinical characteristics and outcome, to identify cytokine profiles. Results: 206 DLBCL patients diagnosed between September 2002 and February 2008 with available serum were included. The median age was 64 years (range 20–92) and 52% were male. At a medium follow-up of 47 months (range 5–94), 75 patients (36%) had relapsed and 63 patients had died (31%). A cytokine profile was determined based on the first two principal components (PC) of the PCA analyses with PC1 corresponding to a general (host) cytokine elevation (also seen in normal controls) and PC2 corresponding to an increase predominantly in cytokines that were commonly elevated in DLBCL patients. 75 DLBCL patients (36%) had a normal cytokine profile similar to the general population of controls; 114 DLBCL patients (55%) had elevation in tumor specific cytokines (CXCL10, CXCL9, IL-2R, IL-12, IL-8, and IL-10) and 17 (8%) had a general cytokine elevation. Patients with any elevation of cytokines (tumor specific or general elevation) had a significantly worse event-free (HR=3.96, 95% CI: 2.13–7.37, p<0.00001) and overall survival (HR=3.75, 95% CI: 1.99–7.09, p<0.0001) than those with a normal cytokine profile (see figure). While the cytokine elevations were associated with advanced stage, poor performance score and elevated LDH, they were not associated with advanced age or extranodal disease and the results remained significant after adjustment for the IPI and treatment in the Cox model (EFS HR=3.11, 95% CI: 1.66–5.85, p<0.0001; OS HR=2.93, 95% CI: 1.53–5.63, p=0.001). Elevated serum cytokines also identified patients who were unlikely to complete initial therapy or who relapsed early. In low-risk patients (IPI 0–2), only 3% of patients with normal serum cytokines relapsed in the first 12 months compared to 23% of patients with elevated cytokines. Similarly in the high-risk group (IPI 3–5), none of the patients with normal cytokines relapsed in the first year compared to 43% with elevated cytokines. Conclusion: Elevated pretreatment serum cytokines, predominantly related to the malignancy, are associated with an increased likelihood of disease relapse and an inferior survival in patients with DLBCL. Furthermore, the pretreatment cytokine profile may also identify DLBCL patients at high risk for early disease relapse. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 3708 Background: Rituximab maintenance (RM) following induction therapy has been demonstrated to increase PFS in randomized trials in both first-line and relapsed settings (ECOG 1496, EORTC 20981, and most recently, PRIMA), and RM has been frequently used in clinical practice in the U.S. and other countries. We examined the various RM approaches used by practicing physicians in the U.S., factors that predicted whether RM was used, and compared the “real-world” effectiveness of RM to the outcomes of observation (Obs) following front-line R-based induction in FL patients (pts). Methods: The National LymphoCare Study is a prospective, multi-center, longitudinal, observational study that collects data on treatment and outcomes for pts diagnosed with FL in the U.S. from 2004–2007. Pts who achieved complete response (CR), partial response (PR) or stable disease (SD) following initial treatment with R-based therapy and who did not initiate second line therapy during the 215-day period following the date of last dose of initial therapy were categorized as Obs. Pts meeting the same criteria who started maintenance R treatment in the 215-day period comprise the RM group. Multivariable logistic regression analyses were performed to identify characteristics related to receiving RM. We compared RM and Obs groups in progression-free survival (PFS), time to next treatment (TTNT), and overall survival (OS) using Cox proportional hazards models controlling for confounding factors of clinical interest (sex, FLIPI risk group, treatment setting [academic vs. community], and induction therapy). PFS was defined as the number of days from the 215 day period following induction that defined the Obs and RM groups up to and including the date of disease progression as assessed by the treating physician or death from any cause. Pts who had not yet experienced a PFS event at the time of analysis were censored at the date of the most recent response assessment. Results: Of the 2,732 pts enrolled in NLCS, 1,536 pts completed R-based induction therapy. Among these,1,236 pts achieved CR/PR/SD and met other criteria described above to be included in analysis. 557 pts started RM treatment in the 215-day post-induction work up period (128 following R and 429 following R-chemo), and 679 pts were observed (125 following R and 554 following R-chemo). The mean duration of RM was 528 days, and the mean number of RM doses received was 13. 23% received Rx4 every 6 months, 20% Rx1 every 2 or 3 months, and 57% received another schedule. 14% of pts discontinued RM early (n=79). The most common reasons for early discontinuations were disease progression (29%), toxicity (11%), or death (7%). Pts who received R induction were more likely to receive RM than those who received R-chemo as induction (51% vs. 44%). Pts treated in community practices were slightly more likely to receive RM compared with those who were treated in academic centers (46% vs. 40%). In multivariable models, the most significant predictors of receiving RM were grade (I/II), stage (III/IV), geographic region (other than West), community practice setting, and R induction treatment (all p<0.05). After adjusting for FLIPI risk group, induction treatment, and clinical factors, RM significantly improved PFS (HR= 0.72; p= 0.010) and TTNT (HR= 0.74; p= 0.03). At 4 years of follow-up, 11% of pts in the combined population have died and there was no significant difference in OS between RM and Obs groups. 22% (n=124) of RM pts had received 2nd line treatment at the time of analysis, of whom 36% received R alone and 31% received R-chemo. The response rate for 2nd treatment was similar between RM and Obs groups (60% CR/PR for both groups). Conclusions: RM following front-line R-based induction was common in this period. However, a large percentage of pts received a RM schedule that was not consistent with standard published schedules reflecting variation among community practices. Nevertheless, RM use in clinical practice produced significant differences in PFS and TTNT compared to Obs. Importantly, response to second-line treatment was similar between RM and Obs groups. Additional followup is needed to determine whether RM produces OS benefits in this context. Disclosures: Sinha: Celgene: Research Funding. Zhou:RTI-HS: Employment; Novartis: Research Funding. Friedberg:Genentech: Consultancy; astellas: ; Lilly: ; Abbott/Trubion: ; Seattle Genetics: Honoraria; Cephalon: Consultancy. Hirata:Genentech Inc/Roche: Employment; Roche: Equity Ownership. Flowers:Spectrum: Consultancy, Research Funding; Millennium/Takeda: Consultancy, Research Funding; Celgene: Consultancy, Research Funding; Genentech/Roche (unpaid): Consultancy; Novartis: Research Funding; Seattle Genetics: Consultancy.

Abstract Gene expression profiling has shown that diffuse large B-cell lymphoma (DLBCL) clusters into three major subtypes based on similarity in expression patterns to their cell of origin (COO): germinal center B-cell-like (GCB-DLBCL), activated B-cell-like DLBCL (ABC-DBLCL) and primary mediastinal B-cell lymphoma (PMBCL). These subtypes of DLBCLs are associated with distinctly different overall survival rates after standard immunochemotherapy. However, clinical and prognostic heterogeneity remains within COO subsets and strategies are needed to further stratify patients to identify and target high-risk subsets. A comprehensive genomic analysis of COO on a clinically defined set of DLBCL cases has not been performed and the aim of this study was to use whole-exome sequencing (WES) data from 58 paired tumor-normal DLBCL samples to assess association of known DLBCL genomic alterations with cell COO as well as for identification of novel and relevant genetic biomarkers. To investigate genomic alterations associated with DLBCL subsets, we analyzed WES and genome wide copy number data from 58 paired tumor-normal DLBCL tumors. Gene expression profiling or Hans classification was performed to determine DLBCL COO subtype; 31 patients were classified as GCB and 27 as non-GCB. The WES data were used to 1) assess the association of known DLBCL genomic alterations with COO, and 2) identify novel alternations associated with COO. Statistical analysis was performed and the data were ranked by significance (p≤ 0.05) within each DLBCL subtype. In total, 45 genomic abnormalities were analyzed for their association with either GCB, non-GCB or both. Mutations in CREBBP, EZH2, MEF2B, FOXO1 and REL have also been reported as GCB driver mutations and we observed GCB patients with these mutations, but the mutation clustering was not wholly associated with GCB. MYC double-hits were exclusively found in the GCB-subtype group. For the non-GCB cases we found that mutations in MAP2K3 and MYD88 were significantly associated with this subtype(p< 0.05). In addition to mutational patterns, we identified several copy number alterations (CNA) across both groups. Chromosomal losses in GCB patients were found at chromosomes 10q11.21-10q24.23, 4q12-4q35.2, 3q12.1-3q29, 4p12-4p16.3, 10p11.21-10p15.3, and 14q11.2-14q24.3 whereas gains were localized to 7q11.1-7q36.3, 7p11.2-7p22.3, and 1q21-1q32.1 (p < 0.05). No CNA was observed to directly associate with non-GCB patients, however, a loss at 9p21 and gains at 9q24.1 and 18q21.33 trended with the non-GCB subtype, supporting previous reports. Loss at 10q23.31 or a gain in 2p13-2p12 have been reported as being specific for GCB and our data confirmed the association of 10q23.31 with GCB while a gain at 2p13-2p12 (REL)was found in both subtypes. To further understand genomic differences between DLBCL subtypes, we evaluated the relative percentage of each genomic feature. 18/45 (40%) were only observed in GCB patients whereas 2/45 (5%) were specific to the non-GCB subtype. The majority (25/45, 55%) overlapped between the two subtypes. Throughout our analysis we noted that 7 non-GCB cases lacked any of the driver mutations analyzed in the study. While all cases carried mutations, they consisted of low frequency mutations that were not specifically associated with COO. 2/7 cases had a gain at 9p24.1 that included CD274 and JAK2. Because 9p24.1 gains have not been fully defined in DLBCL, we reviewed all cases and identified 4 (7%) with a 9p24.1 gain, 3 of which were non-GCB and 1 GCB. One non-GCB case was EBV+ and none of the cases showed evidence of PMBCL. Of the 9p24.1 cases, three had RNAseq data available and we found that PDL1 and JAK2 expression was elevated (12 fold p< 0.01 and 7 fold p< 0.01, respectively) when compared to the 9p24.1 normal cases (n=32). While outcome was not the focus of the study, we did note that 6/7 cases that lacked driver mutations achieved event free survival at 24 months (EFS24). Taken together, this analysis has further characterized the genetic profile of each COO subtype and has identified novel GCB CNAs which require independent replication. Additionally, we identify a subgroup of non-GCB DLBCL patients that do not harbor known driver mutations and require further genomic study to better resolve the biology of these tumors. Together, these data provide insight on the genetic heterogeneity of DLBCL and identify genetic variants that may inform subtype specific therapy. Disclosures Nowakowski: Morphosys: Research Funding; Celgene: Research Funding; Bayer: Consultancy, Research Funding. Rimsza:NCI/NIH: Patents & Royalties: L.M. Rimsza is a co-inventor on a provisional patent, owned by the NCI of the NIH, using Nanostring technology for determining cell of origin in DLBCL..

Abstract Background: DLBCL without an event (disease progression, retreatment or death) within 24 months post diagnosis have a subsequent overall survival equivalent to that of the age- and sex-matched general population. In contrast, patients with an event with 24 months post diagnosis have poor survival. Patients experiencing early events are a heterogeneous group, consisting of patients with primary refectory disease and responding patients who later developed relapse. While the outcome of patients with primary refractory disease have been recently studied, little is known about or patients who have a response following immunochemotherapy (IC) and proceed to surveillance. These patients are the subject of recently completed and ongoing post-IC maintenance trials. To provide perspective, we analyzed outcomes of patients from a prospective observational study entering post IC. Methods: Newly diagnosed patients with DLBCL and treated with anthracycline-based immunochemotherapy were enrolled in the University of Iowa/Mayo Clinic SPORE Molecular Epidemiology Resource (MER) from 2002-2009. End of immunochemotherapy (EOIC) was defined as the date the patient entered surveillance after completion of anthracycline-based IC as per our previous study of surveillance imaging. Response status at EOIC was abstracted as reported from the clinical and radiographic (CT and/or PET) reports. EFS24 was defined being event-free at 24 months from diagnosis, where events consisted of progression, re-treatment, or death (any cause) after initial immunochemotherapy. Survival from EOIC was compared to the general United States population using standardized mortality ratios (SMR) and expected survival. Logistic regression with EFS24 as the endpoint was used to identify pretreatment variables that inform prognosis at the time of EOIC, using Odds ratios (ORs) and 95% confidence intervals (CI) as the measure of association. A subset analysis was performed in the patients aged 60-80 with aaIPI 1-3. Results: 680 patients with DLBCL were enrolled in the MER and initiated R-CHOP or comparable IC from 2002-2009. 552 pts (81%) completed therapy and proceeded to surveillance. Median age of the 552 patients was 61 years (range 18-92) and 51% were male. At a median FU of 89 months from EOIC (range 5-156), 146 pts (26%) had died, and 467 (86%) achieved EFS24. From EOIC, the 5 year overall survival was 82% compared to 91% expected based on age and sex matched general population; SMR was 1.88 (95% CI: 1.60-2.22). In the patients age 60-80 with aaIP1 1-3, (N=203), 79% achieved EFS24 and the 5-year OS from EOIC was 75% compared to 89% expected; SMR was 2.08 (95% 1.65-2.61). When examining all 552 patients proceeding to surveillance at end of EOIC, several high-risk disease characteristics at diagnosis were associated with higher risk of failing EFS24 after completion of therapy: ECOG PS 2-4 (OR = 2.72, 95% CI: 1.55-4.78) , LDH> ULN (OR = 3.54, 95% CI: 1.99-6.31), Stage III/IV (OR=2.43, 95% CI: 1.41-4.16), and aaIPI (per-point OR = 2.02, 95% CI: 1.54-2.64). Patients reported to be in CR at EOIC were more likely to achieve EFS24 (90%) compared to patients not in CR (72%, p<0.001). Conclusion: Patients entering surveillance after IC have shortened overall survival when compared with the general US population, with significant proportion of patients having an event within 24 months of diagnosis. Increased aaIPI and response less than CR are risk factors for inferior EFS24 in this group. Assessment of the prognostic value of EOIC response (PR vs. CR based on central review) is ongoing. Disclosures Nowakowski: Celgene: Research Funding; Morphosys: Research Funding; Bayer: Consultancy, Research Funding. Maurer:Celgene: Research Funding; Kite Pharma: Research Funding. Howlett:Celgene: Employment, Equity Ownership. Ansell:BMS, Seattle Genetics, Merck, Celldex and Affimed: Research Funding.

Abstract Background: There are a number of ongoing clinical trials assessing next-generation immunochemotherapy (IC) regimens (i.e. RX-CHOP) in DLBCL. In addition to standard clinical trial workup, some trials require central pathology review and/or molecular phenotyping before treatment assignment and/or initiation of therapy. There is concern that these studies may be biasing patient selection due to missing patients with aggressive disease who require immediate therapy and cannot delay their treatment to enroll on a study. Here we examine clinical characteristics and outcome stratified by the time from diagnosis to initiation of therapy from a large observational cohort of patients with DLBCL from the R-CHOP era. Methods: Patients were prospectively enrolled in the University of Iowa / Mayo Clinic SPORE Molecular Epidemiology Resource (MER) within 9 months of diagnosis and followed for relapse, retreatment, and death. Clinical management at diagnosis and subsequent therapies were per treating physician. This analysis includes patients with stage II-IV DLBCL or primary mediastinal B-cell lymphoma (PMBCL) who underwent front-line anthracycline based IC; patients with primary CNS lymphoma, PTLD, or a component of low-grade lymphoma were excluded. Time from diagnosis to treatment was defined as the time from date of first lymphoma-containing biopsy to the initiation of IC therapy; delayed therapy was defined as initiating therapy more than 14 days after diagnosis. Event-free survival was defined as time from diagnosis until progression, retreatment, or death due to any cause. EFS24 was defined as progression free status 24 months from diagnosis. Results: 720 patients with stage II-IV newly diagnosed DLBCL or PMBCL and treated with IC were enrolled in the MER from 2002-2012. Median age at diagnosis was 62 years (range 18-92) and 399 patients (55%) were male. 541 patients (75%) had stage III/IV disease and IPI at diagnosis was 0-1 in 166 patients (23%), 2 in 221 patients (31%), 3 in 222 patients (31%) and 4-5 in 111 patients (15%). 233 of 395 patients (59%) were GCB per Hans. At a median follow-up of 73 months, (range 0-163), 349 (49%) patients had an event and 267 patients died (37%); 37% of patients failed to achieve EFS24. Median time from initial lymphoma diagnosis to initiation of IC was 14 days (range 0-79, IQR=8-23). Patients with delay in therapy (>14 days from diagnosis) were more frequently female (50%, p=0.0051) and older than patients who initiated therapy within 14 days from diagnosis (median age at diagnosis of 63 years vs. 60 years, p=0.0008). Patients with delay in treatment initiation had universally less aggressive disease characteristics, including earlier stage, non-elevated LDH, 0-1 extranodal sites, absence of B-symptoms, lower ECOG PS, and lower IPI (see table). In addition, patients with delayed therapy were enriched for GCB subtype per Hans algorithm (p=0.056). Patients with initiation of therapy within 14 day of diagnosis had significantly worse outcome (EFS24 failure=44%) compared to patients with delayed time to initiation of therapy (EFS24 failure=28%, p<0.0001, figure), which remained significant after adjusting for either IPI or aaIPI (both p<0.005). The association between delayed time to initiation of therapy and EFS24 was observed in both GCB (EFS24 failure = 42% vs. 27%, p=0.019) and non-GCB (EFS24 failure = 46% vs. 27%, p=0.014) subsets by Hans. The lower event rate in delayed therapy patients results in an approximately 10% loss of power if the study was powered based on all patients regardless of timing from diagnosis to therapy. Conclusions: Patients with delayed therapy from diagnosis have less aggressive clinical characteristics compared to patients who initiate treatment within 14 days from diagnosis. Studies with a lengthy trial work-up including central pathology review period may be selecting patients with less aggressive disease based on the patient's ability to delay treatment to complete study enrollment requirements. Furthermore, selection of patients with less aggressive disease may result in underpowered studies due to a lower event rate (fewer events) than expected. This retrospective analysis would suggest that trials in DLBCL should consider streamlined enrollment and therapy initiation to avoid potential selection bias and loss of power. Further assessment of implications on trial design and outcomes by cell of origin in this setting is ongoing. Table Table. Figure Figure. Disclosures Maurer: Kite Pharma: Research Funding; Celgene: Research Funding. Ansell:BMS, Seattle Genetics, Merck, Celldex and Affimed: Research Funding. Nowakowski:Morphosys: Research Funding; Bayer: Consultancy, Research Funding; Celgene: Research Funding.