Abstract 623Gene expression profiling has successfully distinguished three subtypes of DLBCL with different biology and response to treatment: 1) germinal center B-cell (GCB); and 2) non-germinal center lymphomas, that include: activated B-cell-like (ABC) and Type 3 subtypes. Currently, immunohistochemical (IHC) analysis of lymphoma biopsy specimens appear to be a more widely applicable methodology (i.e. compared to gene microarray analysis) to use in order to differentiate between subtypes of DLBCL. While the clinical benefit of adding rituximab to CHOP or CHOP-like chemotherapy as a front line treatment of DLBCL is beyond dispute, it also requires a re-evaluation of previously accepted biomarkers of response to CHOP or CHOP-like chemotherapy alone. The predictive value of “IHC–defined” GCB phenotype in rituximab-chemotherapy-treated patients continues to be controversial, as retrospective studies have reported conflicting results. In an attempt to define the predictive value of using the Han's algorithm in newly diagnosed DLBCL patients undergoing frontline immunochemotherapy, we retrospectively analyzed differences in progression free survival (PFS) and overall survival (OS) between patients with GCB and non-GCB DLBCL treated with equivalent doses of rituximab and anthracycline-based therapy at our institution. Using the tumor registry and the pharmacy database, we identified patients with DLBCL treated at our Institution between 2000 and 2008. Demographic, clinical, pharmacologic and pathological characteristics were obtained for each patient. Patients were classified into GCB or non-GCB DLBCL according to the Han's algorithm based on the expression of CD10, Bcl-6 and MUM-1 in the large cell component of the tumor specimen. Cumulative doses of rituximab (R), cyclophosphamide (C), doxorubicin (H), vincristine (O), etoposide (E,; when used) and prednisone (P) were calculated for each patient, as well as the number of cycles, dose delays, and growth factor use. A total of 192 patients were included in the study. The average age was 58.65 years (F/M:73/119). Using the Hans algorithm, n=55 (28.6%) and n=57 (29.7%) were classified as non-GCB or GCB, respectively. Inadequate information was availabel to classify 80 patients (undetermined group). Clinical indictors such as clinical demographics, international prognostic index (IPI) score, extra-nodal disease, performance status, Ann Arbor stage, therapy delays, cumulative rituximab and chemotherapy doses were not significantly different between groups (non-GCB, GCB, and undetermined DLBCL). The majority of patients received R+CHOP (90%) or R+ dose adjusted–EPOCH. On follow-up, a total of 42 (21.8%) patients relapsed or were found to have primary-refractory disease. The complete remission rate to front-line therapy was 81% for the entire cohort of patients and was not different between patients with GCB or non-GCB DLBCL. On the other hand, significant differences in PFS and OS were observed between patients with non-GCB versus GCB DLBCL. The 5-year progression free survival (PFS) and overall survival (OS) were significantly better in the GCB DLBCL subtype (75.4% vs. 56.4%, p=0.017 and 84.2% vs. 70.9 %, p=0.037; respectively). As no differences in clinical parameters, CR rate, or rituximab-chemotherapy dose/schedule were observed between non-GCB and GCB DLBCL patients, it is possible that intrinsic biological pathways involved in lymphomagenesis and/or “resistance” of these subtypes of DLBCL may play a role in their responsiveness to rituximab-based immunochemoimmunotherapy. In summary, our data suggest that the Hans algorithm can predict the clinical outcome of patients with DLBCL undergoing front-line therapy with R-chemotherapy. Patients with non-GCB DLBCL while having a comparable initial complete response rate to R+CHOP had a shorter PFS and OS than GCB DLBCL patients. Non-GCB DLBCL represents a subgroup of DLBCL for which innovative therapeutic strategies targeting key regulatory pathways in the induction and/or maintenance setting are needed in an attempt to prolong PFS and improve OS. Disclosures:No relevant conflicts of interest to declare.