Primary Mediastinal Large B Cell Lymphoma: Elucidating Optimal Therapy and Prognostic Factors; an Analysis in 141 Consecutive Patients Treated at Memorial Sloan Kettering from 1980–1999.

Introduction: Primary Mediastinal Large B cell Lymphoma (PMLBL) is a distinct clinicopathologic entity. The optimal treatment and prognostic factors remain to be determined. We report long-term follow-up of 141 consecutive patients (pts) treated at MSKCC from 1980–1999 with a goal of elucidating the optimal treatment strategy and identifying predictive factors.

Methods: Pts were defined as having PMLBL if they had B-cell diffuse mixed, diffuse large, or immunoblastic histology; a mediastinal mass greater than 5 cm ± contiguous extranodal or supraclavicular involvement with no concomitant extra-mediastinal mass greater in size than the primary mediastinal lesion; and no prior therapy for lymphoma. From 1980–1995 pts were treated on IRB approved clinical trials that were either CHOP-like, NHL-15 or an upfront ASCT; after 1995 pts received either CHOP-like chemotherapy or the NHL-15 program, as per MSKCC treatment guidelines. The NHL-15 program (Ann Oncol., in press, 2004) comprises dose-dense sequential induction with doxorubicin 60mg/m² i.v. weeks (wks) 1,3,5 & 7 and vincristine 1.4 mg/m² i.v. wks1,2,3,5 & 7, followed by consolidation with cyclophosphamide 3000 mg/m² i.v. on wks 9, 11 & 13 with G-CSF support. Clinical evaluation included LDH, age-adjusted IPI (aaIPI) and IPI assessment, CT scan and gallium imaging. A CR required complete normalization of radiographic and laboratory values, including gallium scan, or a biopsy proven remission if <75% reduction in mediastinal mass or residual gallium avidity.

Results: Median age 32 (16–73); M:F ratio 1.14:1. Stage I/II/III/IV (n=141): 1/83/5/52; 75% bulky, 77% elevated LDH, and 40% with B symptoms. Extranodal disease (ENS) was present in 69% of pts (multiple sites in 28%). Fifty-four percent of pts had high-intermediate/high risk by aaIPI, with no significant differences between treatment groups. With a median follow-up of 11 years, event-free survival (EFS) and overall survival (OS) as analyzed by intent-to-treat for all pts was 50% and 66%, respectively. EFS/OS for CHOP/CHOPlike (56 pts), NHL-15 (68 pts), and upfront ASCT (17 pts) was 34/51%, 60/84%, and 60/78%, respectively. In pairwise comparison, CHOP/CHOPlike regimens had an inferior EFS and OS compared with NHL-15 or upfront ASCT (p<0.001 all comparisons). Radiotherapy (RT), utilized in 35% of patients, was not associated with survival differences. Multivariate analysis for prognostic factors revealed the following predictors: for EFS, >1 ENS (hazard ratio, HR=1.86) and initial therapy received (NHL-15 HR=0.45, upfront ASCT HR=0.40); for OS only initial therapy with NHL-15 was predictive (HR=0.28). Neither the aaIPI nor the IPI were prognostic in this analysis.

Conclusions: In PMLBL:

1. dose-dense chemotherapy with NHL-15 may be superior to CHOP (with or without radiation);

2. patients treated with chemotherapy only can have excellent long term outcomes; the impact of RT will require a phase III trial comparing dose-dense chemotherapy +/− RT;

3. in this analysis, the aaIPI did not predict survival; it may not be and ideal model for risk stratification

4. the use of high dose therapy and ASCT should be reserved for those patients who fail upfront anthracycline-based therapy or should be tested in the very poor risk patients with multiple extranodal sites.