AbstractAbstract 3107 Background:Survival of young high risk diffuse large B-cell lymphoma (DLBCL) is now approaching 80% due to implementation of rituximab and dose-dense chemotherapy protocols but the patients with relapsed or refractory disease continue to have a poor prognosis. Such patients could benefit from additional therapies if their clinical outcome could be more accurately predicted at the time of diagnosis. Recently, several gene-expression signatures with prognostic significance in DLBCL have been identified. To date, the accessibility of exon arrays that interrogate exon-level expression has enabled a new, more sensitive method of analysing gene-expression than the traditional 3′ arrays. In the present study, we have tested the utility of exon profiling to define novel prognostic markers for young high risk DLBCL patients. Patients:Study population consisted of 41 patients (36 DLBCL and 5 follicular lymphoma (FL) grade 3) less than 65 years old with high risk (age adjusted International Prognostic Index (aaIPI) Score 2–3) disease. The selection of the patients was based on the availability of freshly frozen lymphoma tissue containing adequate material for mRNA analyses. All tissue samples were taken before treatments. All patients were treated in the Nordic phase II protocol with six courses of R-CHOEP14 followed by systemic CNS prophylaxis with one course of high-dose methotrexate and one course of high-dose cytarabine. In the present report with a median follow-up of 29 months, (range 15–63 months), ten patients had relapsed and nine died. Relapse free survival (RFS) was 74% and overall survival (OS) 77%. Results:We identified differentially expressed exons between the relapsed patients and the patients in remission using criteria of p ≤ 0.05 and fold change ≥ 1.6. In order to estimate the gene-level expression, and to exclude false positives, the genes were considered as differentially expressed only if at least 20% of all exons were differentially expressed. Accordingly, 646 genes were identified, from which 119 encoded proteins. In a pathway network analysis (Laakso and Hautaniemi, 2010), 23 genes were found likely to be involved in conventional signalling pathways. The identified pathways included important events of lymphoma biology such as antigen processing and presentation, cell adhesion, chemokine signalling as well as TGF-beta, Toll-like receptor, Wnt and MAPK signalling. We also performed a gene level survival analysis with data combined of differentially expressed exons and follow-up information. 12 of the 23 genes were found to associate with RFS (p<0.05). Among these, high expression of HLA-DOA, HLA-DQB1 (both members of MHC class II family), RFXAP (MCH class II transcription regulator), SMAD7 (mediator of TGF-beta signalling), IRF5 (interferon regulatory factor) and CR1 (complement component) had a favourable impact on RFS. In contrast, high IL22 (interleukin 22) and DLG2 (Discs 2) levels were associated with adverse outcome. Similarly, 7 of the 23 genes were predictive for OS (p<0.05). Prognostic impact of one third of the transcripts could be confirmed in an independent gene array based data set of 233 DLBCL patients treated with immunochemotherapy (Lenz et al., 2008). Conclusions:The data suggest that exon-based transcriptome profiling of diagnostic tumor tissue can identify biologically relevant genes that discriminate the outcome of homogenously treated young high risk lymphoma patients. Such genes and involved pathways represent markers for improved patient risk stratification and potential targets for novel DLBCL therapies. Disclosures:Holte:Roche: Honoraria, Research Funding; Amgen: Honoraria, Research Funding. Leppa:Roche: Honoraria, Research Funding.