Abstract

ABSTRACT © F e r r a t a S t o r t i F o u n d a t i o n BCL2 translocations are more frequently found in the GCB subtype, whereas 18q21 locus amplification is more common in the ABC subtype of DLBCL. 3, Design and Methods Patients We studied 327 cases of previously untreated de novo DLBCL, diagnosed between January 2002 and October 2009, and collected as part of the International DLBCL Rituxan-CHOP Consortium Program Study. These cases were analyzed for Bcl-2 protein expression, and BCL2 and MYC gene abnormalities, and gene expression profiling (GEP) was performed. All cases were reviewed by a group of hematopathologists (SMM, MAP, MBM, AT, and KHY), and the diagnoses were confirmed based on World Health Organization classification criteria. Patients with transformation from low grade lymphoma, those with composite follicular lymphoma, primary mediastinal large B-cell lymphoma, primary cutaneous and primary central nervous system DLBCL were excluded from the analysis due to the unique biological features of these types of lymphoma. All patients were adults who were negative for human immunodeficiency virus and had sufficient clinical data and clinical follow-up. Patients in this study were treated with R-CHOP (n=291, 89%) or R-CHOP-like regimens (n=36, 11%; CHOP scheme adopting different anthracyclines i.e. novantrone or epirubicin). All patients with advanced stage disease received six (92%) or eight (8%) cycles, every 21 days, with or without radiotherapy for residual disease or initial bulky disease; localized cases received at least three cycles followed by radiotherapy or six cycles without radiotherapy. The current study was approved by each of the participating centers' Institutional Review Boards, and the overall collaborative study was approved by the Institutional Review Board at The University of Texas MD Anderson Cancer Center in Houston, Texas, USA. Immunohistochemistry for Bcl-2 and cut-off determination Bcl-2 protein expression was evaluated in all patients using a monoclonal anti-Bcl-2 antibody (Clone-124, Dako, Carpinteria, CA, USA) and standard immunohistochemical methods. The formalin-fixed, paraffin-embedded tissue slides underwent deparaffinization and heat-induced antigen retrieval techniques. An endogenous biotin-blocking kit (Ventana) was used to decrease background staining. Following antigen retrieval and primary antibody incubation, the reaction was completed in a Ventana ES instrument using a diaminobenzidine immunoperoxidase detection kit (Ventana). Immunoreactivity was determined without knowledge of the patients' survival, clinical data, or GEP data. The samples were analyzed independently by a group of four hematopathologists/pathologists in addition to the hematopathologist of each of the contributing centers, and disagreements were resolved by joint review at a multi-headed microscope. An average of 300-400 cells in four to five fields were counted in the tissue microarray cores. A percentage of tumor cell staining ≥50% was considered positive after receiver operating characteristic (ROC) curve analysis was implemented to assess the discriminatory accuracy of Bcl-2 protein in recognizing patients with different overall survival (OS) and progression-free survival (PFS). The 50% value was established from the analysis of the area under the ROC curve (AUROC) and had the maximum specificity and sensibility for OS and PFS discrimination in our patients (AUROC=0.564, P=0.017 for OS and AUROC=0.564, P=0.015 for PFS). 31 Gene expression profiling analysis RNA was extracted from 327 formalin-fixed, paraffin-embedded tissue samples using a HighPure Paraffin RNA Extraction Kit (Roche Applied Science). Fifty nanograms of RNA were transcribed into cDNA, linearly amplified using the WT-Ovation™ FFPE System (Nugen), and biotin-labeled using FL-Ovation™ cDNA Biotin Module V2 (Nugen) in all cases. For GeneChip hybridization, 5 μg of WT-Ovation amplified cDNA were applied to HG-U133 Plus 2.0 GeneChips (Affymetrix) and hybridized overnight. GeneChips were washed, stained, and scanned using the Fluidic Station 450 and GeneChip Scanner 3000 (Affymetrix) according to the manufacturer's recommendations. For data analysis and classification, the microarray DQN (trimmed mean of differences of perfect match and mismatch intensities with quantile normalization) signals were generated and normalized to the quantiles of beta distribution with parameters p=1.2 and q=3 as previously described. 32 A Bayesian model was also utilized to determine the class probability. The classification model was built on the 47 paired formalin-fixed, paraffin-embedded tissue sample dataset previously generated with a confidence rate of 90-100% in fresh frozen tissue and 92-100% in formalin-fixed, paraffinembedded tissue. The same methodology developed during this study has been validated and demonstrated to be applicable by using the LLMPP dataset in the Gene Expression Omnibus (GEO) database GSE#10846 that has 181 CHOP-treated and 233 R-CHOP-treated DLBCL patients with fresh-frozen samples. 3, Validation set To validate our observations in predicting survival in an independent series of cases, we analyzed a second group of 120 archival DLBCL cases studied similarly to the first cohort except for MYC analysis that was not available (GCB 49%, ABC 40%, unclassified 11%; BCL2 translocations in 18%; Bcl-2 overexpression in 54%). All these patients had been treated with R-CHOP and the same selection criteria as those for the first cohort were applied. The clinical characteristics at presentation of the patients in the validation set were not significantly different from those of the patients in the test set. Statistical analysis Following pre-defined criteria, 33 PFS was measured from the time of diagnosis to the time of progression or death from any cause. OS was measured from the time of diagnosis to last followup or death from any cause. Only patients with a follow-up of longer than 12 months were included in the survival analysis. The actuarial probabilities of PFS and OS were determined using the Kaplan-Meier method, and differences were compared using the log-rank test. A Cox proportional-hazards model was used for multivariate analysis. The χ 2 test or Mann-Whitney test was applied to assess differences between variables. The interobserver agreement for FISH was assessed using the κ statistic; a κ value of >0.75 implied excellent agreement. All statistical calculations, except for ROC and the κ statistic which were performed with SPSS 18.0 (SPSS Inc., Chicago, IL, USA), were conducted using StatView (Abacus Concepts, Berkeley, CA, USA). Results Patients' characteristics and outcome The median age of the patients at diagnosis was 62 years (range, 18-86). Their clinical characteristics are reported in BCL2 and MYC genes in the subgroups defined by gene expression profiling Sixty patients (18.3%) had DLBCL with BCL2 gene translocations, and 50 (15.3%) had BCL2 gene amplifications. The presence of BCL2 translocations was not associated with any clinical prognostic variable at diagnosis, except for Ann Arbor Stage (70% versus 49% with stage III-IV, P=0.004), as shown in The OS and PFS rates of patients with BCL2 translocations were similar to those of patients without BCL2 translocations, irrespectively of MYC status. When we restricted the analysis to the GCB subtype, patients with BCL2 translocations alone, in the absence of MYC breaks, had a significantly worse outcome than GCB patients without BCL2 translocations (3-year PFS of 53% versus 76%, respectively; P=0.0002). The outcome of patients with BCL2 rearranged GCB subtype was similar to that of the patients with the ABC subtype of DLBCL (52%, P=0.30), but still better than that of the patients with double hit lymphomas (P<0.0001, The presence of MYC breaks alone in the 19 patients without concomitant BCL2 translocations was not associated with impaired PFS (P=0.70) or OS (P=0.66) in the whole cohort, but was associated with inferior OS (P=0.03), but not PFS (P=0.22), in patients with GCB-DLBCL (only 9 with isolated MYC breaks). As shown in BCL2 gains were not prognostic in any of the subgroups of patients. Particular consideration of high-level amplifications was of no additional prognostic value. Bcl-2 protein expression, clinical characteristics, fluorescence in situ hybridization and gene expression profiling None of the common clinical characteristics of our patients at the time of presentation was significantly associated with Bcl-2 protein expression except age, with older patients more often being Bcl-2 positive (≥60 years old, P=0.02). Bcl-2 protein expression in GEP-and FISHdefined subgroups is shown in © F e r r a t a S t o r t i F o u n d a t i o n patients without the BCL2 translocation (range, 0-100%; median 60%). Bcl-2 protein expression was significantly associated with worse PFS (P=0.01) and OS (P=0.02) in the whole cohort, but when patients were divided according to GEPdefined subtypes, we observed that higher Bcl-2 expression was associated with significantly inferior PFS in the GCB subgroup (P=0.04), but not in the ABC subgroup (P=0.57), as shown in Multivariate analysis Multivariate analysis of all 137 patients with the GCB subtype of DLBCL showed that BCL2 translocations (HR 0.40, 95% CI: 0.18-0.89; P=0.02), but not Bcl-2 expression (HR 1.01, 95% CI: 0.45-2.21; P=0.98), MYC breaks (HR 0.25, 95% CI: 0.10-0.59; P=0.001), and IPI score (HR 0.41, 95% CI: 0.20-0.84; P=0.01), were independently associated with patients' outcome. Results were not modified after each molecular feature was computed with age as a continuous parameter. C. Visco et al. 258 haematologica | 2013; 98(2) Overall survival Progression-free survival Four-hundred and forty-four genes were found to be differentially expressed (>1.5 fold and P<0.005) in DLBCL patients with or without BCL2 translocations including both GCB and ABC subtypes. In the GCB group, however, only 43 genes were differentially expressed among patients with and without BCL2 translocations ( Interestingly, a number of genes overexpressed in the BCL2 translocated group are involved in the control of angiogenesis and the inflammatory response (AIMP1, PPIA, and ALOX), while others are involved in promoting apoptosis or regulating B-cell signaling (STK17A, RAL-GPS2, NCOA3, STRBP, and ZNF117). 35-37 C. Visco et al. 260 haematologica | 2013; 98(2) © F e r r a t a S t o r t i F o u n d a t i o n Discussion We addressed the clinical impact of BCL2 aberrations and their relationship to Bcl-2 protein expression in a large series of patients with DLBCL homogeneously treated with R-CHOP, with known MYC gene status and molecularly characterized according to GEP analysis. We were able to establish the role of the BCL2 gene in different subtypes of DLBCL, irrespectively of concomitant MYC aberrations. We found that isolated BCL2 translocations, in the absence of MYC breaks, were associated with a poor outcome in the subset of patients with GCB-DLBCL, and that the prognosis of these patients was similar to that of patients with ABC-DLBCL. The concomitant presence of MYC breaks (double hit lymphoma) further worsened the outcome of these patients. The role of Bcl-2 protein expression appeared dependent on its association with BCL2 translocations, as outlined by multivariate analysis and survival curves As determined by FISH break apart probe analysis, the overall frequency of BCL2 translocations in de novo DLBCL was 18.3%. The BCL2 translocations were almost exclusively associated with GCB-DLBCL, found in 34.5% of cases The impact of BCL2 translocations on survival in our series could not be explained by differences in the clinical features of the patients because there was no association between the presence of BCL2 translocations and IPI risk groups (P=0.90, A C B FIGURE A COLORI SOLO ONLINE © F e r r a t a S t o r t i F o u n d a t i o n with isolated BCL2 or MYC lesions. Confirming previous findings, In this series, Bcl-2 protein was overexpressed in half of the patients with GCB-DLBCL and in 72% of patients with ABC-DLBCL 18,39 However, Bcl-2 overexpression had prognostic value only in the GCB subtype, as already observed by others in the era of R-CHOP therapy. Iqbal et al. 22 Secondly, Bcl-2 protein expression in Iqbal's study was significantly associated with adverse clinical prognostic factors (stage III-IV, elevated lactate dehydrogenase, high IPI risk group) in GCB-DLBCL, which was not the case in our study. Finally, no mention was made about exclusion of possibly confounding DLBCL subtypes such as double hit lymphoma, primary cutaneous or primary central nervous system DLBCL. We also acknowledge that different findings in the literature regarding BCL2 rearrangements or protein expression could very well be related to lack of uniformity between different studies in terms of Bcl-2 staining and scoring. Moreover, patients' characteristics in the different series, differences in the management of the cases as they were not in clinical trials, data collection regarding outcome, and sometimes short follow-up times may also have contributed to different results. Our GEP analysis revealed that patients with BCL2 translocations substantially differed with respect to important recurrent oncogenic events, which may contribute to the adverse outcome of the subgroup of GCB-DLBCL patients with BCL2 translocations. Up-regulation of the BCL11A gene occurred exclusively in the group of patients with BCL2 translocations We confirm that the outcome of GCB-DLBCL patients should be interpreted in the context of abnormalities of the MYC and BCL2 genes. While the MYC rearrangement is quite rare, it is rarely found as the sole genetic abnormality, and its clinical relevance is mainly related to a double hit mechanism, BCL2 rearrangements are present in a considerable fraction of patients with the GCB subtype who have similar outcomes to those of patients with the ABC subtype. Our results confirm that the GCB and ABC subtypes of DLBCL have distinct pathogeneses, and support the rationale for further classification of different subgroups. © F e r r a t a S t o r t i F o u n d a t i o n Funding CV is a hematologist supported by San