Abstract 303 Background: Primary mediastinal large B-cell lymphoma (PMBCL) was distinguished from DLBCL in the REAL/WHO classification by virtue of distinct clinical and pathologic features. In DLBCL, R-CHOP has demonstrated a clear benefit over CHOP chemotherapy in randomized trials, some of which have included a small subset of PMBCL pts (pts). However, large studies evaluating R-CHOP in PMBCL are lacking. Further, the role of consolidative radiotherapy (RT) to the mediastinum remains unclear particularly with the more frequent use of FDG-PET. At the BCCA, from 2001–2005, R-CHOP with consolidative RT was recommended in all pts with PMBCL. After 2005, FDG-PET scan was used to guide RT usage following 6 cycles of R-CHOP. We evaluated: 1) the outcome of pts with PMBCL in R-CHOP vsCHOP treated pts; 2) prognostic factors in R-CHOP treated pts; 3) the impact of the introduction of PET based approach to guide RT compared to the use routine RT. Methods: Using the BC Cancer Agency Centre for Lymphoid Cancer database, we identified all pts with PMBCL by the REAL/WHO classification treated with R-CHOP in addition to an earlier group treated with CHOP chemotherapy. For R-CHOP treated pts, from 2001–2005, consolidative RT to the mediastinumwas routinely administered following R-CHOP (‘RT’ era). Since 2005, a PET was planned at the end of chemotherapy to guide RT (‘PET’ era). If the PET was negative, pts were observed and if the PET was positive, consolidative RT was given, if feasible. Results: In total, 176 pts were identified: 96 received R-CHOP; 80 received CHOP. Baseline clinical features were similar between the treatment groups. Compared to those treated with CHOP, R-CHOP pts had an improved time to progression (TTP) (5 y 78% vs 65%, p=0.018) and overall survival (OS) (5 y 88% vs 70%, p=0.015). Further, refractory disease (progressive disease (PD) on chemotherapy or relapse For the R-CHOP treated pts, 46 were treated in the ‘RT era’; 80% received consolidative RT; 50 were treated in the ‘PET era’; 38% received RT. In the RT era, there was a greater proportion of pts with extranodal sites > 1 (p=0.007) and the presence of a pleural/pericardial effusion (p=0.009). In an era to era outcome comparison, there was no difference in the 5 y TTP (RT era 76% vs PET era 83%, p=0.626) or 5 y OS (82% vs 89%, p=0.773). In univariate analysis, the presence of B symptoms was associated with an inferior TTP (5 y 67% vs 90%, p=0.014) and OS (5 y 83% vs 94%, p=0.047). An effusion at diagnosis was associated with an inferior TTP (p=0.017) but not OS (p=0.277). Using recursive partitioning, age > 38 y was identified as the optimal cut-off and was associated with an inferior OS (p=0.003) and there was a trend to a worse TTP (p=0.096). The aaIPI was not predictive of TTP (p=0.357) or OS (p=0.386). In multivariate analysis (including variables with p value 38 y (HR 2.98, p=0.025) were associated with an inferior TTP. For OS, age > 38 y (HR 6.7, p=0.003) and B symptoms (HR 4.5, p=0.023) were associated with an inferior outcome. Treatment era (RT vsPET) did not affect TTP or OS in MVA. In total, 59 pts treated with R-CHOP had a PET scan at the end of treatment, 50 in the PET era and 9 pts in the ‘RT era’ by self-pay: 35 (59%) were PET-negative (neg) (2 received RT) and 24(41%) were PET-positive (pos) (23 received RT). With a median follow-up of 5.4 y, there was no difference in the TTP (5 y 78% vs 83%, p=0.735) or OS (5 y 88.5% vs 95%, 0.271) for PET-neg and PET-pos cases, respectively. In total, there were 6 relapses in PET-neg cases: 2 CNS and 4 mediastinal. There were 4 relapses in the PET pos cases (PD in the mediastinum at PET scan n=1; within in RT field n=1; within and outside RT field n=1; outside RT field n=1). Conclusion: We confirm the superior outcome using R-CHOP compared to CHOP chemotherapy in PMBCL, however, rituximab does not appear to impact the rare risk of CNS relapse. The introduction of a PET-guided RT approach in R-CHOP treated PMBCL pts has maintained excellent outcomes in the majority of pts while reducing the use of RT. The presence of B symptoms at diagnosis was associated with an increased risk of relapse and may reflect more aggressive underlying disease biology. Disclosures: Savage:Roche: Research Funding. Klasa:F Hoffmann-La Roche (Roche Canada): Consultancy, Research Funding. Villa:F Hoffmann-La Roche (Roche Canada): Consultancy, Research Funding. Slack:F Hoffmann-La Roche (Roche Canada): Consultancy, Research Funding. Gascoyne:F Hoffmann-La Roche (Roche Canada): Consultancy, Research Funding. Connors:Roche: Research Funding. Sehn:F. Hoffmann-La Roche (Roche Canada): Research Funding.