Favorable outcome of primary mediastinal large B-cell lymphoma in a single institution: the British Columbia experience.

BACKGROUND: Primary mediastinal large B-cell lymphoma (PMBCL) is a distinct clinico-pathological subtype of diffuse large B-cell lymphoma (DLBCL). The optimal treatment is unknown, with some studies suggesting a superior outcome with dose-intensive chemotherapy regimens, and the role of radiotherapy remains ill-defined. PATIENTS AND METHODS: The British Columbia Cancer Agency lymphoma database was searched and records reviewed to identify those patients presenting with a prominent mediastinal mass and considered to be PMBCL based on the current REAL/WHO classifications. Patients were treated based on era-specific BCCA guidelines (1980-1992 MACOPB/VACOPB; 1992-2001 CHOP-type; 2001-present CHOP-R). Beginning in January 1998 involved-field radiotherapy was recommended to be routinely administered following chemotherapy. Prior to this, use of radiotherapy was individualized in advanced disease.
RESULTS: In total, 153 patients with newly diagnosed PMBCL were identified between 28 July 1980 and 30 June 2003. The median age was 37 years (range 13-82) and the majority had stage I/II (74%), bulky mediastinal disease (75%). Overall (OS) and progression-free (PFS) survival at 5 years for the entire cohort were 75% and 69%, respectively. In direct comparison with a cohort of patients with DLBCL (n = 1273), OS (P = 10(-4)) and PFS (P = 0.0001) favored PMBCL. The age-adjusted International Prognostic Index (aaIPI) was not predictive of survival (P = 0.18). Five-year OS in patients < 65 years old treated with MACOPB/VACOPB, CHOP-R and CHOP-type was 87%, 81% and 71% respectively (P = 0.048). In pair-wise survival comparisons, only MACOPB/VACOPB and CHOP-type treated patients were significantly different (P = 0.016). In Cox multiple regression analysis, poor performance status remained the only predictor of survival, with treatment received demonstrating a trend to worse outcome for patients treated with CHOP-type regimens (P = 0.09). In an intention-to-treat analysis comparing the era before radiotherapy was routinely administered with after, there was no significant difference in 5-year PFS (74% versus 62%; P = 0.09) or OS (78% versus 69%; P = 0.14).
CONCLUSIONS: In this single institution, population-based retrospective study, we found that PMBCL patients have excellent survival rates and a distinct plateau is observed in PFS, in striking comparison to DLBCL. The aaIPI was not predictive of survival in this population, suggesting that other prognostic models may be better suited for risk stratification. Dose-intensified chemotherapy with MACOPB or VACOPB demonstrated a trend to superior outcome over CHOP-type chemotherapy. However, further randomized studies are needed and the impact of rituximab on these comparisons must be considered. Finally, the routine addition of radiotherapy does not improve survival.