Search

Gastric marginal zone B-cell lymphoma of MALT type (MALT lymphoma) arises in the context of chronic inflammation induced by the bacterial pathogen Helicobacter pylori. Although generally considered an indolent disease, MALT lymphoma may transform to gastric diffuse large B-cell lymphoma (gDLBCL) through mechanisms that remain poorly understood. By comparing microRNA expression profiles of gastric MALT lymphoma and gDLBCL, we have identified a signature of 27 deregulated microRNAs(miRNAs) that share the characteristic of being transcriptionally repressed by Myc. Myc overexpression was consequently detected in 80% of gDLBCL but only 20% of MALT lymphomas spotted on a tissue microarray. A highly similar signature of Myc-repressed miRNAs was further detected in nodal DLBCL. Small interfering RNA–mediated knock-down of Myc blocked proliferation of DLBCL cell lines. Of the Myc-repressed miRNAs down-regulated in malignant lymphoma, miR-34a showed the strongest antiproliferative properties when overexpressed in DLBCL cells. We could further attribute miR-34a's tumor-suppressive effects to deregulation of its target FoxP1. FoxP1 overexpression was detected in gDLBCL but not in gastric MALT lymphoma; FoxP1 knock-down efficiently blocked DLBCL proliferation. In conclusion, our results elucidate a novel Myc- and FoxP1-dependent pathway of malignant transformation and suggest miR-34a replacement therapy as a promising strategy in lymphoma treatment.

Introduction. The majority of SMZLs display an indolent course, however the disease is still incurable and a significant proportion of patients (~25-30%) experience poor outcomes surviving 5 years, and for whom tumor material collected before initiation of medical therapy was available. Mutation analysis was performed by CAPP-seq targeted deep next generation sequencing of tumor genomic DNA. A stringent bioinformatic pipeline was applied to suppress the background noise allowing to call variants with a sensitivity of 5x10-2 in FFPE derived DNA. Copy number variations (CNVs) were identified by using the sequencing reads-based GATK4-CNV algorithm. IGHV rearrangements were obtained by using LymphoTrack® IGH FR1 Assay Panel kit. Molecular clusters were identified by an iterative algorithm that maximizes genetic distinctiveness of subgroups by reassigning patients between clusters that are created a priori based on the co-occurrence of genetic lesions. Relative survival, defined as the ratio between actuarial survival observed in the SMZL cohort and expected survival of the general population matched to patients by geographical origin, sex, age and calendar year of diagnosis, was calculated using the Ederer II method. Results. The analysis included 303 patients with a SMZL diagnosis confirmed on spleen histology. The sample size allowed to identify 30% differences in survival for molecular subgroups comprising at least 5% of cases with a statistical power between 80-100%. Median follow-up was 9.2 years. At 10 years, 85% of patients were alive, consistent with the known indolent behavior of this lymphoma. Genes recurrently affected by non-synonymous somatic mutations in >10% of SMZL included KLF2 (24%), NOTCH2 (19%), KMT2D (15%), TNFAIP3 (13%), EP300 and TP53 (10%). Deletion 7q was documented in 25% of cases and IGHV1-2*04 usage in 32%. By cluster analysis, three major molecular subgroups were identified, each of them characterized by a NOTCH pathway mutated gene (Fig. 1A). The first cluster was defined by NOTCH2 and/or KLF2 mutations and was enriched in TNFAIP3 mutations and IGHV1-2*04 gene usage (Fig. 1A). The second cluster was defined by SPEN mutations, and was enriched in KMT2D and other epigenetic gene mutations (Fig. 1A). The third cluster was enriched in NOTCH1 mutations (Fig. 1A). By relative survival analysis, the NOTCH2/KLF2 cluster showed a lower survival compared to the matched general population, indicating a significant impact of the disease on patients' expected survival (Fig. 1B). Conclusions. The large sample size and inclusion of SMZL confirmed by spleen histopathology review allowed for precise estimation of the prevalence of KLF2 and NOTCH2 mutations in this lymphoma. Three molecular clusters were identified in SMZL, each of them containing a NOTCH pathway gene, supporting the relevance of NOTCH signaling in the pathogenesis of SMZL. Patients belonging to the NOTCH2/KLF2 cluster had a lower relative survival compared to the matched general population. Disclosures Traverse-Glehen: Astra Zeneca: Other: Travel; Takeda: Research Funding. Gomes da Silva:Roche: Other: Institution's payment for consultancy, Travelling support; BMS: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: lecture fees; Celgene: Other: Travelling support; Janssen: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: lecture fees, Institution's payment for consultancy, Travelling support; Abbvie: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: lecture fees; Gilead Sciences: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: lecture fees, Research Funding. Ladetto:Celgene: Honoraria; Jannsen: Honoraria; Acerta: Honoraria; Abbvie: Honoraria; Sandoz: Honoraria; Roche: Honoraria. Rambaldi:Pfizer: Consultancy; Novartis: Consultancy; Omeros: Consultancy; Italfarmaco: Consultancy; Amgen Inc.: Consultancy; Roche: Consultancy; Celgene: Consultancy. Vitolo:Takeda: Speakers Bureau; Sandoz: Speakers Bureau; Gilead: Speakers Bureau; Janssen: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Roche: Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau; Celgene: Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau. Zinzani:MSD: Honoraria, Speakers Bureau; Gilead: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; SERVIER: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; TG Pharmaceuticals: Honoraria, Membership on an entity's Board of Directors or advisory committees; Bayer: Membership on an entity's Board of Directors or advisory committees; PFIZER: Honoraria, Membership on an entity's Board of Directors or advisory committees; Merck: Honoraria, Membership on an entity's Board of Directors or advisory committees; Janssen: Honoraria, Speakers Bureau; Bayer: Membership on an entity's Board of Directors or advisory committees; TG Pharmaceuticals: Honoraria, Membership on an entity's Board of Directors or advisory committees; BMS: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Celltrion: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Takeda: Membership on an entity's Board of Directors or advisory committees; PFIZER: Honoraria, Membership on an entity's Board of Directors or advisory committees; Verastem: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Celgene: Honoraria, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Merck: Honoraria, Membership on an entity's Board of Directors or advisory committees; Roche: Honoraria, Membership on an entity's Board of Directors or advisory committees; Astra Zeneca: Speakers Bureau. Gaidano:Roche: Consultancy, Honoraria; Gilead: Consultancy, Honoraria; AbbVie: Consultancy, Honoraria; Morphosys: Honoraria; Amgen: Consultancy, Honoraria; Janssen: Consultancy, Honoraria. Salles:Servier: Honoraria; Abbvie: Honoraria; F. Hoffmann-La Roche Ltd: Consultancy, Honoraria, Research Funding; Novartis: Consultancy, Honoraria; Amgen: Honoraria; BMS: Honoraria, Other: Advisory Board; Celgene: Honoraria, Other: Advisory Board, Research Funding; Acerta: Honoraria; Janssen: Honoraria, Other: Advisory Board; Merck: Honoraria; Pfizer: Honoraria; Morphosys: Honoraria; Gilead: Honoraria, Other: Advisory Board; Epizyme: Honoraria; Servier: Honoraria, Other: Advisory Board; Takeda: Honoraria. Zucca:Celltrion: Consultancy; AstraZeneca: Consultancy.

The survival of diffuse large B-cell lymphoma patients varies considerably, reflecting the molecular diversity of tumors. In view of the controversy whether cytologic features, immunohistochemical markers or gene expression signatures may capture this molecular diversity, we investigated which features provide prognostic information in a prospective trial in the R-CHOP treatment era. Within the cohort of DLBCLs patients treated in the RICOVER-60 trial of the German High-Grade Lymphoma Study Group (DSHNHL), we tested the prognostic impact of IB morphology in 949 patients. The expression of immunohistochemical markers CD5, CD10, BCL2, BCL6, human leukocyte antigen (HLA)–DR, interferon regulatory factor-4/multiple myeloma-1 (IRF4/MUM1), and Ki-67 was assessed in 506 patients. Expression of the immunohistochemical markers tested was of modest, if any, prognostic relevance. Moreover, the Hans algorithm using the expression patterns of CD10, BCL6, and interferon regulatory factor-4/multiple myeloma-1 failed to show prognostic significance in the entire cohort as well as in patient subgroups. IB morphology, however, emerged as a robust, significantly adverse prognostic factor in multivariate analysis, and its diagnosis showed a good reproducibility among expert hematopathologists. We conclude, therefore, that IB morphology in DLBCL is likely to capture some of the adverse molecular alterations that are currently not detectable in a routine diagnostic setting, and that its recognition has significant prognostic power.

Nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL) accounts for approximately 5% of all Hodgkin lymphoma cases. The aim of this study was to evaluate the prognostic implication of histopathologic NLPHL variants. Biopsies of 423 NLPHL patients treated within 9 prospective clinical trials performed by the German Hodgkin Study Group were classified as tumor cell-rich cases (n = 10), typical NLPHL (n = 308), or histopathologic variants (n = 105). Histopathologic variants were characterized by the presence of lymphoma cells outside the B-cell nodules or B-cell depletion of the microenvironment. Compared with typical NLPHL, histopathologic variants were associated with advanced disease (29.5% vs 14.6%, P = .0012) and a higher relapse rate (18.1% vs 6.5% at 5 years, P = .0009). Variant histology represented an independent prognostic factor (odds ratio = 2.955) in a multivariate model of progression/relapse. A prognostic score, including the risk factors variant histopathologic growth pattern, low serum albumin, and male gender, was derived from this model and allowed the definition of 3 distinct risk groups. NLPHL patients presenting with histopathologic variants have a poorer outcome compared with those showing typical histology. The newly developed prognostic score combining histologic and clinical features allows allocating NLPHL patients to defined risk groups.

MYC rearrangements occur in 5% to 10% of diffuse large B-cell lymphomas (DLBCL) and confer an increased risk to cyclophosphamide, hydroxydaunorubicin, oncovin, and prednisone (CHOP) and rituximab (R)-CHOP treated patients. We investigated the prognostic relevance of MYC-, BCL2- and BCL6-rearrangements and protein expression in a prospective randomized trial. Paraffin-embedded tumor samples from 442 de novo DLBCL treated within the RICOVER study of the German High-Grade Non-Hodgkin Lymphoma Study Group (DSHNHL) were investigated using immunohistochemistry and fluorescence in situ hybridization (FISH) to detect protein expression and breaks of MYC, BCL2, and BCL6. Rearrangements of MYC, BCL2, and BCL6 were detected in 8.8%, 13.5%, and 28.7%, respectively. Protein overexpression of MYC (>40%) was encountered in 31.8% of tumors; 79.6% and 82.8% of tumors expressed BCL2 and BCL6, respectively. MYC translocations, MYChigh, BCL2high, and BCL6low protein expressions were associated with inferior survival. In multivariate Cox regression modeling, protein expression patterns of MYC, BCL2 and BCL6, and MYC rearrangements were predictive of outcome and provided prognostic information independent of the International Prognostic Index (IPI) for overall survival and event-free survival. A combined immunohistochemical or FISH/immunohistochemical score predicts outcome in DLBCL patients independent of the IPI and identifies a subset of 15% of patients with dismal prognosis in the high-risk IPI group following treatment with R-CHOP. Registered at http://www.cancer.gov/clinicaltrials: RICOVER trial of the DSHNHL is NCT 00052936.

The Lymph2Cx Cell-of-origin (COO) assay enables the identification of molecular GCB and ABC subtypes of diffuse large B-cell lymphomas (DLBCL) using RNA from formalin-fixed paraffin-embedded (FFPE) samples. In order to explore the prognostic power of this COO assay within the frame of prospectively randomized clinical trials for the first time, we retrospectively performed COO analyses in patient samples from two clinical trials (RICOVER-60: n=360 and R-MegaCHOEP: n=92) of the German High Grade Lymphoma Study Group (DSHNHL) using the Nanostring nCounter®-platform. Immunohistochemistry was done for MYC, FOXP1, LMO2, GCET1, CD5, CD10, BCL2, BCL6, IRF4/MUM1 protein expression, and fluorescence in-situ hybridization was performed for MYC, BCL2 and BCL6 genes with break-apart probes. COO classification was successfully obtained from 414 samples (326/360 (91%) and 88/92 (96%) FFPE samples from the RICOVER60 and R-MegaCHOEP trials, respectively). In the RICOVER-60 study (patients aged > 60 and In RICOVER-60, the ABC-DLBCL group was associated with inferior clinical outcome compared with the GCB subtype (p=0.041, p=0.022 and p=0.034 for EFS, PFS and OS, respectively). A significant survival difference was also seen in CHOP-treated patients. This effect, however, was completely abrogated in R-CHOP treated patients with no significant difference observed in EFS, PFS and OS. This result was confirmed in multivariable analyses adjusted for IPI factors. We found the same results for all RICOVER-60 patients after adjustment for IPI factors (EFS: HR=1.1, PFS: HR=1.2, OS: HR=1.2 for ABC vs. GCB) and after adjustment for IPI factors, bulky disease, gender and age > 70 years. In keeping with this result, no survival differences were observed between ABC-DLBCL and GCB-DLBCL patients in the R-MegaCHOEP trial for EFS, PFS and OS, and this was also confirmed in multivariable analysis (data not shown). In the RICOVER-60 trial, patients with concomitant over-expression of both BCL2 (50-100%) and MYC (40-100%) (MYC+/BCL2+ DE DLBCL) showed significantly inferior survival in all R-CHOP treated patients and especially in the GCB subgroup compared with non-expressers [MYC-/BCL2- DLBCL; MYC-/BCL2+] (p In summary, COO profiling alone failed to identify prognostic subgroups, whereas MYC/BCL2 double expression was highly predictive of poor survival. These data from prospective randomized trials demonstrate that older patients and young high-risk patients show significant differences in molecular profiles and, furthermore, indicate that complete subtyping using different molecular methods will be necessary to predict survival under current standard therapy. Only comprehensive molecular analysis of individual patient samples will allow identification of optimal therapy including new molecules. Disclosures Scott: NanoString Technologies: Patents & Royalties: named inventor on a patent for molecular subtyping of DLBCL that has been licensed to NanoString Technologies. Klapper:Roche, Novartis, Amgen, Takeda: Research Funding.

Approximately 5-10% of diffuse large B-cell lymphomas (DLBCL) harbor a 8q24/MYC rearrangement (MYC+). We determined the prognostic significance of MYC rearrangement in patients with relapsed/refractory DLBCL prospectively treated by R-ICE or R-DHAP followed by high-dose therapy and autologous stem cell transplantation. Twenty-eight (17%) of the 161 patients analyzed presented a MYC+ rearrangement, targeted as either simple hit (25%) or complex hits (n=75%) including MYC/BCL2, MYC/BCL6, and MYC/BCL2/BCL6. Results were statistically highly concordant in matched primary and relapsed biopsies (n = 45). Compared to the MYC− DLBCL patients, the MYC+ DLBCL patients presented with a more elevated lactico-deshydrogenase level (P = .0006) and a more advanced age adjusted international prognostic index (P = .0039). The 4-year PFS and OS were significantly lower in the MYC+ DLBCL patients than those in the MYC− DLBCL patients, with rates of 18% vs 42% (P = .0322), and of 29% vs 62% (P = .0113), respectively. Type of treatment, R-DHAP or R-ICE, had no impact on survivals, with 4-year PFS rates of 17% vs 19% and 4-year OS rates of 26% vs 31%. In conclusion, MYC rearrangement is an early event in DLBCL. MYC+ DLBCL patients have a significant inferior prognosis than MYC− DLBCL patients. Their outcome was not influenced by the proposed salvage therapy.

The potential benefits of extended rituximab treatment have been investigated in a randomized trial comparing the standard schedule with prolonged treatment in 202 patients with newly diagnosed or refractory/relapsed follicular lymphoma (FL). All patients received standard treatment (rituximab 375 mg/m(2) weekly x 4). In 185 evaluable patients, the overall response rate was 67% in chemotherapy-naive patients and 46% in pretreated cases (P

Purpose A number of retrospective analyses have looked into the prognostic implication of MYC, BCL2 and BCL6 rearrangements and protein expression for MYC, BCL2 and BCL6 in patients with diffuse large B-cell lymphoma (DLBCL). Many of these studies suffer from small patient cohorts and differing or unknown treatment strategies (with or without Rituximab) administered to patients under study. Furthermore, the median age of these patients was relatively high. We for the first time report on the prognostic consequences of MYC, BCL2 and BCL6 alterations in younger (18-60 years) high-risk patients (aaIPI 2 or 3) with DLBCL. Patients and Methods The MegaCHOEP study (Schmitz et al. Lancet Oncology 2012: 13, 1250) randomized patients to 8xCHOEP-14 or sequential high-dose therapy supported by repeated infusions of autologous stem cells. Both treatment arms included 6 infusions of R (375 mg/ sqm). The median patient age was 48 years, and 27% of patients scored an aaIPI of 3. Paraffin-embedded tumor samples from 112 de novo DLBCL were investigated using immunohistochemistry for MYC, BCL2, and BCL6, and fluorescence in-situ hybridisation (FISH) to detect breaks of MYC, BCL2 and BCL6. Results Rearrangements of MYC, BCL2 and BCL6 were detected in 13.6%, 20.7% and 30.9% of DLBCL, respectively. A double or triple hit constellation occurred in 10.8%. Presence of BCL2 breaks (RR=4.7, 95% CI: 1.8-12.2) and MYC breaks (RR=2.4, 95% CI: 0.8-7.5), but not of BCL6 breaks were associated with inferior overall survival in univariate and multivariate analyses adjusted for aaIPI and treatment arm. Protein overexpression of MYC ≥30% (RR=2.4, 95% CI: 0.9-6.5), but not BCL2 (≥60%) or BCL6 (≥30%) indicated inferior overall survival. BCL2 overexpression was associated with inferior EFS (RR=2.2, 95% CI: 0.9-5.5) and PFS (RR=2.8, 95% CI: 1.0-8.2). If the same cutpoint for BCL2 used for a similar analysis in elderly patients treated on the RICOVER-60 trial (Horn et al. Blood 2013) was applied, no differences in EFS or PFS were observed. Conclusion Rearrangements of MYC and BCL2 seem to be relevant prognostic factors also in young high-risk patients. The frequencies of MYC and BCL2 rearrangements are not substantially higher than reported for other, mostly elderly patient groups carrying IPI scores from 0 - 5; it seems unlikely, therefore, that the rearrangements described here (completely) explain the poor prognosis of young, high-risk patients. The prognostic role of BCL2, BCL6, and MYC as well as other biological risk factors must be validated in independent data sets. Disclosures: No relevant conflicts of interest to declare.

Abstract 2747 Background: Single-agent immunotherapy with rituximab is a viable treatment option for low risk FL, with limited toxicity and a long duration of response in some patient subsets. We have previously shown that high expression of FcγRIIB promotes rituximab internalisation on various B cell targets, including FL (Blood 2011 118:2530–2540), something not seen with type II anti-CD20 antibodies. The SAKK 35/98 trial examined rituximab monotherapy in FL and now has long-term follow-up data of almost 10 years (JCO 2010 28:4480–4484). We analysed diagnostic tumour samples from this trial to determine the relationship of FcγRIIB expression to responses and clinical outcomes after rituximab treatment in FL. Methods: 202 patients (pts) with newly diagnosed or relapsed FL received induction treatment with rituximab 375 mg/m2 weekly for 4 weeks. Pts with stable or responding disease at week 12 were randomized into 2 groups: no further treatment or prolonged treatment with single infusions of rituximab 375 mg/m2 at weeks 12, 20, 28 and 36. Archived tissue samples from 135 evaluable pts were stained using an anti-human FcγRIIB antibody (clone EP888Y, Abcam) at a dilution of 1:3000 on a Dako autostainer. The samples were pretreated with the Dako EnVisionFLEX target retrieval solution high pH and detection using the Dako AS-Link 48 with Dako EnVision flex plus detection kit. Positive samples were graded into negative/low intensity staining (n=120) versus medium/high (n=13) by an expert lymphoma histopathologist blinded to the clinical outcomes. Data from 2 slides and response at week 12 data for 4 pts were unavailable (1 of whom also has missing slide data), resulting in 130 pts available for analysis. Failure-free survival (FFS) was defined as time from registration until failure to achieve complete/partial response at week 12, progression, relapse, a second cancer or death from any cause. Objective response rate (ORR) was associated with intensity staining levels using Fisher's exact test. All time-to-event endpoints were evaluated using the Kaplan-Meier method; groups were compared using the log-rank test. The hazard ratio (HR) was assessed using Cox proportional hazards models. Results: Registered and randomised pts had very similar baseline characteristics; previously untreated pts had slightly more favourable characteristics but were balanced between the 2 treatment arms. Pts expressing medium/high levels of FcγRIIB were less likely to respond to rituximab by week 12 (ORR 58.1% vs 23.1%, Fisher's exact test, p=0. 02), a finding independent of prior therapy. For FFS, there was a statistically significant difference (p=0.001; HR=0.42; 95% confidence interval (C.I.): 0.23–0.77) between the negative/low staining group (median: 21.4 months; 95% C.I.: 7.0–34.2) and the medium/high staining group (median: 7.0 months; 95% C.I.: Not calculable). The interaction between staining levels and randomised treatment groups for FFS was not statistically significant. There was a non-significant trend towards better overall survival in the low/negative group (median: 140.0 vs 50.0 months; p=0.15; HR=0.57; 95% C.I.: 0.27–1.23); however the event rate was lower (36.8% vs 61.5%). Conclusion: Elevated FcγRIIB expression level is associated with poor response to rituximab in pts with FL. This group may show better results with non-internalising type II antibodies, a hypothesis for validation in future prospective clinical trials. Disclosures: Ghielmini: Roche: Honoraria, Speakers Bureau. Johnson:Roche: Honoraria.

Abstract 594FN2 Background . Approximately 5% to 10% of DLBCLs harbor a MYC/8q24 oncogene rearrangement. We determined the prognostic significance of MYC+ DLBCL in a selected population of patients with relapsed/refractory DLBCL included in the CORAL and prospectively treated by RICE or RDHAP followed by BEAM and ASCT1. METHODS. Among the 396 patients included, 150 patients were analysed with break-apart FISH probes for MYC/8q24, BCL2/18q21, and BCL6/3q27. Correlations with clinical characteristics at diagnosis and at relapse, immunohistochemistry (IHC) for CD10, BCL6, MUM1, FOXP1, and BCL2 expression, cell of origin determined according to different IHC algorithms2-5 and by transcriptomic analysis, response to treatment and survivals were analysed considering presence or absence of MYC rearrangement. RESULTS. Twenty-four (16%) cases exhibited a MYC rearrangement, associated in 87.5% (n=21) of the cases with a BCL2/18q21 rearrangement (n=16), with a BCL6/3q27 rearrangement (n=4), and with both BCL2/18q21 and BCL6/3q27 rearrangements (n=4). Among the matched cases analysed (n=57), biological characteristics including tumor immunophenotype and breakpoints occurrence, were identical between primary and secondary biopsies. Presence of MYC+ rearrangement was significantly associated with CD10 expression (p= 0.0067) and a GC phenotype, based either on Hans algorithm (p Disclosures: No relevant conflicts of interest to declare.

Abstract 263 RS represents the development of a diffuse large B-cell lymphoma (DLBCL) in the context of chronic lymphocytic leukemia (CLL). The pathogenesis of RS is still largely unknown. Analysis of RS has been often focused on the study of lesions previously identified in de novo DLBCL. However, we have previously shown that RS lacks many of the typical genetic lesions shown in DLBCL (Rossi et al, Blood 2011). Here, we have applied an unbiased, genome-wide approach, searching for DNA copy number alterations in a large series of RS, comparing them with de novo DLBCL, CLL-phase of RS and untransformed CLL. Methods. The study included 50 RS, 30 CLL-phases preceding RS, 100 DLBCL, 318 CLL. All RS were classified as DLBCL. DNA was extracted from frozen biopsies and was processed using the Affymetrix GeneChip Human Mapping SNP6 arrays. Differences in frequencies between subgroups were evaluated using Fisher's exact test. Clinical data were available in half of the cases and genomic lesions were evaluated for their impact on clinical outcome with the log-rank test. Results. In RS, the most common gains were +12 (40%), +4q12 (14%), +2p25.3-25.2 (13%), +8q23.3-qter (12%), +1q32.1, +2p16.1-13.2 (BCL11A/REL), +15q22.31-26.3, +18q21.33 (BCL2) (10%), and +3q24.1-q26.3 (9%). Amplifications were observed at the BCL2 locus and at 1q32.1. The most common losses were at 17p13.1 (TP53; 29%), 8p21.3 (TNFRSF10A/B; 15%), 8p23.3 (16%), 11q22.3 (ATM) (15%), 14q23.3 (13%), 7q33–35 (11%), 14q32.11 (10%), 14q32.31-32.33 (TRAF3), and 9p21.3 (CDKN2A) (9%). The latter locus was also the target of recurrent homozygous deletions. RS appeared intermediate between CLL-phase and de novo DLBCL in terms of number of recurrent lesions. When compared with de novo DLBCL, RS presented less -6q23 (TNFAIP3; P=2.54E-05), -15q15.1-q21.1 (B2M) (P=5.64E-04), +7/7q (P=5.71E-04), -6q21 (PRDM1; P=6.53E-04), +2p16.1 (BCL11A; P=3.22E-03), -1p36 (TNFRSF14; P 9.57E-03), copy-neutral LOH at 6p25.3-21.32 (P =1.01E-02), +5p (P=1.39E-02), +11q14.1-q25 (P=1.63E-02), +1q22-q41 (P 2.13E-02), +6p21 (P 2.48E-02). Conversely, RS presented more often +4q12 (P=8.96E-03), -7q33-35 (P=2.04E-02), -11q22.3 (ATM; P=2.14E-02), -14q23.3 (P=2.69E-02). No statistical differences were observed for lesions such as +3/+3q, +8q, -8p, - 9p21.3 (CDKN2A), +12q, +15q, -17p (TP53). Compared to CLL-phase, RS had significantly more gains at 1q32.1 (P=2.28E-02), 11q23.3 (P=5.38E-03), 11q24 (P=1.10E-02) and 9p21.3 losses (CDKN2A; P=1.10E-02). Compared to the large series of CLL with no history of transformation, CLL-phase samples presented less frequently losses at 13q14.3 (DLEU2, MIR15A, MIR16-1a; 6.58E-03), more commonly gains at 12 (P=2.66E-02), 4q12 (P=1.103E-02), +2p (P=2.97E-02), and loss at TP53 locus (P=2.23E-03). The presence of losses of CDKN2A (P=4.6E-03), TNFRSF10A/B (P=3.2E-02) and BCL2 gains (P=4.0E-02) determined a poor overall survival in RS. Conclusions. DNA copy number changes in RS are intermediate between de novo DLBCL and CLL-phase. Recurrent lesions were identified that appeared more common in RS than in DLBCL, and that seem to be involved in the progression from CLL to an aggressive lymphoma. Also, the acquisition of mono- and bi-allelic losses of CDKN2A appears as one of the most important event in the transition from CLL-phase to RS. CDKN2A and TNFRSF10A/B losses and BCL2 gains were associated with a poor outcome. AR & EC equally contributed to the work. Disclosures: No relevant conflicts of interest to declare.

Abstract 993 Purpose: Cell of origin (COO) has a major prognostic impact in DLBCL. However this impact has never been analysed in the context of relapsed/refractory DLBCL. The purpose of this study was to evaluate the prognostic value of COO immunohistochemical markers and BCL2, BCL6 and c-MYC translocations in a well-defined cohort of patients included in the CORAL trial1. Methods: Enrolled patients had CD20+ DLBCL in first relapse or refractory after upfront therapy. Treatment was randomized to either R-ICE (rituximab, ifosfamide, etoposide, carboplatin) or R-DHAP (rituximab, dexamethasone, aracytine, cisplatin). Responding patients had high-dose chemotherapy and ASCT. Paraffin-embedded tumor samples of 249 patients were analysed by immunochemistry for CD10/MME, BCL6, MUM1/IRF4, FOXP1, BCL2 expression and for BCL2, BCL6 and c-MYC breakpoints by FISH, and were categorized according to different COO algorythms2-5. Primary biopsy at diagnosis and relapse biopsy were available for 189 and 147 patients respectively, including 87 matched pairs. Results: With a median follow-up at 27 months, there was no significant difference between R-ICE and R-DHAP for 2-year progression-free survival (PFS) (56.7%+/−5% vs 60%+/−5%, p=0.24) or overall survival (OS) (46%+/−5% vs 60%+/−5%, p=0.24). We confirmed that early relapse, prior rituximab exposure and secondary aaIPI had adverse prognostic value. In univariate analysis, absence of bcl6 expression, non-GC phenotype2, ABC phenotype4, at diagnosis were associated with an unfavourable PFS (p=0.038, p=0.0044, p=0.044, respectively), whereas only non-GC phenotype2 was associated to a worse OS (p=0.027). At relapse, expression of Foxp1 and bcl2, c-MYC breakpoint, ABC phenotype4, immunoFISH index5 was associated with an unfavourable PFS (p=0.02, p=0.027, p=0.03, p=0.032, p=0.02, respectively), and c-MYC breakpoint was associated with a worse OS (p=0.01). Tumor immunophenotype and FISH analysis at diagnosis and at relapse were statistically concordant in matched pairs (Wilcoxon's test between 0.42 and 1). In the whole group, only c-MYC breakpoint present in 13% of the patients was correlated with worse PFS (p= 0.02) and OS (p =0.04). Statistical interaction were found between GC/non-GC classification2 and the treatment R-ICE vs R-DHAP. Patients with GC B-cell like DLBCL subtype treated with R-DHAP have a better 2-year PFS (64% +/−7%) than those with GC B-cell like DLBCL treated with R-ICE (42% +/−7%), and the patients with ABC B-cell like DLBCL treated either with R-DHAP (35%+/−7%) or R-ICE (45%+/−7%) (p= 0.04). In multivariate analysis, only GC/non-GC phenotype2 interaction with the treatment, FOXP1 expression, prior rituximab exposure, secondary aaIIPI, time to relapse were statistically significant. Preliminary results of gene expression profiling seem to confirm these data and are currently under validation. Conclusion: COO remains a major factor in relapsed patients with a better response to R-DHAP salvage chemotherapy in GC-B DLBCL, underlying the importance of molecular DLBCL subtyping. Treatment of ABC subtype is still unsatisfactory. Disclosures: Schmitz: Roche: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. Gisselbrecht:Roche: Research Funding.

Mantle cell lymphoma (MCL) has a very poor prognosis. Remissions are often incomplete and short with a median survival of less than 3 years and a 5 year survival of < 30%. Young patients are usually treated with intensive chemotherapy regimens or a stem cell transplant. In elderly, unfit patients aggressive treatments are not a realistic approach. This trial set out to assess Gemcitabine in the treatment of patients with newly diagnosed, relapsed or chemotherapy resistant MCL not fit enough for intensive chemotherapy regimens. The primary endpoint was the objective response. Eligibility criteria were a WHO performance status < 2, age ≥ 18 years, a histologically confirmed diagnosis of MCL along with at least one measurable lesion of diameter ≥ 11 mm, ≤ 2 previous lines of chemotherapy. Gemcitabine was given in doses of 1000mg/m2 as a 30 minute infusion on days 1 and 8 of each 3 week cycle for a total of 9 cycles. To prevent flu-like symptoms, one dose of steroids per Gemcitabine infusion was administered. Response duration was read as the time from trial registration to a relapse for patients reaching a remission on the trial therapy. Progression-free survival was defined as the time from trial registration to death or disease progression, treatment failure as early termination of the trial treatment. Patient enrollment began in Aug 2004. Stage I analysis concluded the trial treatment was not promising for further investigation, due to only 1 (10%) patient reaching a remission, and patient accrual stopped after the inclusion of 18 patients in March 2006. Median age at enrollment was 70 years. 66% of the patients were male. MCL was newly diagnosed in 50% of the patients and relapsed in the remainder. Ann Arbor stage IV rating was present in 15 patients, stage II in 1 patient and stage I in 2 patients. Bone marrow involvement was reported in 10 patients. Gastrointestinal tract involvement was reported in 4 patients. Involvement of > 1 extranodal organ was seen in 3 patients. One CR, 4 PRs, 8 SDs and 4 PDs were recorded as a best response. The patient achieving a CR was one of the patients presenting with a stage IV disease. Most myelotoxicities occurred during the first chemotherapy cycle. Neutropenia CTC grade 1 in 24 (19.8%), grade 2 in 35 (28.9%) and grade 3/4 was seen in 24 (19.8%) of the cycles. Thrombocytopenia CTC grade 1 was recorded in 55 (45.5%) cycles and grade 2/3 in 9 (7.7%) cycles. Three patients developed non-hematological serious adverse events, defined as an inpatient or prolonged hospitalisation or a life threatening illness related to the trial medication; dyspnea, glomerular microangiopathy with hemolytic uremic syndrome, and hyperglycemia. The median time-to-progression and response duration was 8.0 (95% confidence interval: 5.5 – 9.3) and 10.6 (95% confidence interval: 5.5 – 10.9) months respectively. The median time-to-treatment failure was not reached. We conclude that Gemcitabine is well tolerated and that it can stabilise MCL in elderly patients. However, Gemcitabine as a monotherapy has only limited activity in MCL patients in terms of treatment response. Further trials should therefore assess gemcitabine in frail patients with MCL.

We analyzed the clinical and biological prognostic factors for efficacy and described the toxicity of treatment by study arm for 306 patients participating in two identical and parallel trials, one in follicular (FL) and one in mantle cell lymphoma (MCL). Chemotherapy naïve or pre-treated patients were treated with 4 weekly doses of rituximab (induction). Those responding or with stable disease were randomized to either no further treatment (arm A) or 4 more doses of rituximab given at 2-month intervals (arm B). Characteristics of the patients (4 variables), previous treatment (6 variables), disease (9 variables), histology (26 variables), blood cell counts (6 variables), lymphocytes subsets (5 variables), blood chemistry values (8 variables) and Fc receptor phenotype (3 variables) were collected at trial entry and during follow-up.We constructed multivariate models to identify prognostic factors for response and for event-free survival (EFS). In 273 eligible and evaluable cases the response rate was 44%. In these patients, the favorable prognostic factors for response were a disease diameter < 5 cm (odds ratio =2.5, p=0.0006), a follicular histology (OR=2.4, p=0.0029), a normal hemoglobin (OR= 2.4, p=0.0033) and a low blood lymphocyte count (OR=2.0, p=0.02), all determined at treatment start. At a median follow up of 54 months, the EFS for all 212 randomized cases was 9.1 months in arm A and 15.4 months in arm B (p=0.005). Factors prognostic for EFS in a multivariate model based on 162 patients were: having responded to rituximab induction (hazard ratio=0.44, p