Primary mediastinal B-cell lymphoma (PMBCL) is a relatively rare clinico-pathologic subtype of diffuse large B-cell lymohoma. The optimal management, the prognostic factors and the role of PET/CT scan in this entity remain a matter of debate. While several retrospective studies suggested that dose-dense regimens are more effective than standard CHOP, the impact of adding rituximab (R) on the outcome of patients (pts) with PMBCL has not been fully evaluated. In this retrospective analysis we reviewed the clinical and radiological records of 81 consecutive pts with PMBCL treated in Sheba Medical Center between August 1985 and October 2006. Chemotherapy in the pre-rituximab era (−R cohort) included VACOPB (n=47) or 6 courses of standard CHOP (n=5). Since October 2002, 6 cycles of R were added concurrent with the treatment in another 29 pts (+R cohort): R-VACOPB (n=21) and R-CHOP (n=8). Radiotherapy was not administered following initial chemotherapy to any of the pts. Median age at diagnosis was 31 years (yrs) (range 17–61). Stage I/II and bulky mediastinum (≥ 10 cm in diameter) were present in 88% and 33%, respectively, and extranodal involvement was evident in 44% of all pts. After a median follow-up of 85 months (range 9–240), the overall (OS) and progression-free (PFS) survival at 5 yrs for the entire cohort were 89% and 66%, respectively, with a plateau 1–2 yrs following treatment. PFS at 5 years was significantly better with +R (81%) than with -R cohort (58%, P=0.03). Five-year PFS in pts treated with R-VACOPB, R-CHOP, VACOP-B and CHOP were 84%, 74%, 62%, 20%, respectively (P=0.025). Yet, there was no significant difference in OS between +R and -R cohorts (96% vs. 88% at 5 yrs, p=0.29). Direct survival comparisons demonstrated that 5-yr OS and PFS were significantly better in VACOPB than in CHOP (P=0.04 and 0.05, respectively) and that R-VACOPB was significantly superior to VACOPB in terms of PFS (p=0.05). In contrast, there was no difference in 5-yr PFS between R-VACOPB and R-CHOP (p=0.44). Univariate analysis revealed that aaIPI was not predictive of OS (p=0.51). Age above 31 yrs (p=0.02) and pericardial effusion (p=0.04) were the only predictors of reduced OS. Furthermore, beginning in 2003, 16 consecutive pts in the +R cohort, who were scanned by PET/CT-FDG before starting and after completion of therapy, were also evaluated in the middle (mid-PET) of treatment. The estimated 3-year PFS rate for mid-PET negative pts (n=8) and for mid-PET positive pts (n=8) was 86% and 75%, respectively (P=0.48). In terms of treatment failure, the negative predictive value of mid-PET was 100%, while the positive predictive value was only 25%. In conclusion, our population-based historical comparison demonstrates that the addition of R to anthracycline-based therapy significantly improved PFS in pts with PMBCL. We observed superior PFS with VACOPB compared with CHOP, but this superiority was abrogated by the introduction of R as part of initial therapy. These findings merit further study in randomized prospective studies.

Author notes

Disclosure: No relevant conflicts of interest to declare