Abstract Abstract 1585 Background: Diffuse Large B-cell Lymphoma (DLBCL) is a heterogeneous disease with variable outcome within international prognostic index (IPI) categories. Previous studies indicate expression of germinal centre B-cell (GCB) markers predict good outcome and that non-GCB sub-types may benefit from alternate therapies. A variety of immunohistochemical combinations have been proposed. However results are inconsistent and there is limited data comparing strategies. The need for a robust, practical method to distinguish GCB-DLBCL remains. We evaluated co-expression of the GCB proteins LMO2 and BCL6 with the Choi, Hans and Tally algorithms in patients treated with R-CHOP chemo-immunotherapy. Patients and Methods: Seventy-five DLBCL patients, chosen solely on availability of material for tissue microarray, were studied using immunohistochemistry for FOXP1, MUM1, CD10, BCL6, LMO2 and GCET1 on original biopsies. Clinical data was obtained from a prospectively maintained database. Results were correlated with event free (EFS) and overall survival (OS). Results: The cohort was typical for DLBCL described in the literature with 63% of patients aged over 60. The International Prognostic Index (IPI) predicted overall survival (p<0.001) with 36% of patients with a low IPI score (0,1), 41% with intermediate score (2,3) and 23% with a high IPI score (4,5). The three year event free and overall survival (EFS and OS) were 68% and 73% respectively with a median follow up of 37 months. None of the cell of origin algorithm predicted for EFS or OS. The Hans (49%) and Choi algorithms (43%) typed similar number of patients as GCB but the Tally method (26%) typed patients less frequently as GCB. This difference was statistically different (p=0.014). No single marker predicted for outcome, however there was a trend for patients who were BCL6 positive to have improved outcome (p=0.06). 53% of patients were positive for BCL6. BCL6 was a significant predictor of improved event free survival on univariate analysis (p=0.016). Interestingly, patients who were positive for both LMO2 and BCL6 (26% of total) had a significantly improved outcome on univariate analysis for both event free (p=0.01) and overall survival (p=0.0036). LM02/BCL6 dual positivity retained significance for event free survival independent of IPI on multivariate analysis (p=0.017). It was not possible to demonstrate independence from IPI for overall survival using Cox regression as none of the nineteen dual positive patients have died. There was no significant differences between patients who were LMO2/BCL6 positive and the rest of the cohort with respect to any of the IPI factors. Of note, LMO2 on its own was independent of IPI with regards to overall survival despite not reaching significance at the univariate level (p=0.049). In the subgroup of patients who were positive for BCL6 approximately half were also positive for LMO2. There was a significant difference in overall survival between these BCL6 single positive and BCL6/LMO2 dual positive tumours (p=0.003) with improved outcome seen in the dual positive patients. Similarly those LMO2 positive patients who were BCL6 negative had very poor outcome compared to LMO2/BCL6 dual positive tumours (p=0.0002). Dual positivity for BCL6/LMO2 was strongly associated with germinal centre phenotype with eighty per cent of these patients typing for this cell of origin (p=0.0004). Germinal centre DLBCL as typed by the Choi algorithm could be separated into two significantly different groups with different overall survival based on LMO2/BCL6 dual positivity (p=0.01) Conclusion: LMO2/BCL6 dual positivity is a prognostic marker in DLBCL treated with R-CHOP that is independent of IPI, and predicts outcome in GCB typed tumours. Should these findings be prospectively validated, the strategy would be easily transferable to the diagnostic laboratory. Disclosures: No relevant conflicts of interest to declare.