Primary mediastinal large B-cell lymphoma: optimal therapy and prognostic factor analysis in 141 consecutive patients treated at Memorial Sloan Kettering from 1980 to 1999

Abstract

Primary mediastinal large B-cell lymphoma (PMLBL) is a distinct clinicopathological entity with unclear prognostic factors and optimal treatment approach. To elucidate an optimal treatment and identify predictive factors, a retrospective analysis of 141 consecutive patients was undertaken. Patients received cyclophosphamide, hydroxydaunomycin, Oncovin, prednisone (CHOP)-like therapy, the non-Hodgkin lymphoma (NHL)-15 regimen or upfront autologous stem cell transplantation (ASCT) on Institutional Review Board approved trials or according to the institutional guidelines. Evaluation included lactate dehydrogenase, International Prognostic Index (IPI) assessment, computed tomography scan and gallium imaging. With a median follow-up of 10.9 years, event-free survival (EFS) and overall survival (OS) was 50% and 66% respectively. EFS/OS for CHOP/CHOP-like, NHL-15 and upfront ASCT was 34/51%, 60/84% and 60/78% respectively. CHOP/CHOP-like regimens had inferior EFS and OS versus NHL-15 or upfront ASCT (P < 0.001). A total of 23% of patients received radiotherapy. Multivariate analysis revealed the following outcome predictors: for EFS, greater than or equal to two extranodal sites and initial therapy received (NHL-15 or upfront ASCT); for OS, only initial therapy with NHL-15. We conclude: (i) dose-dense chemotherapy with NHL-15 may be superior to CHOP for PMLBL; (ii) The impact of consolidative radiotherapy requires randomised controlled trials; (iii) The age-adjusted IPI did not predict survival in this analysis; (iv) high-dose chemotherapy/ASCT should be reserved for upfront anthracycline-based therapy failure or in clinical trials for high-risk patients.