Background and Objective

Primary mediastinal large B-cell lymphoma (PMBL) accounts for 2 to 4% of non-Hodgkin lymphomas and is characterized by distinct clinical, pathological and genetic features. Although the utility of DA-EPOCH-R without radiotherapy (RT) and a PET-guided RT approach were recently reported, a standard therapy has not yet been established, mostly due to the lack of data from prospective randomized studies. In addition, the prognosis for patients (pts) with relapsed PMBL is not well understood. Therefore, we conducted a multicenter, cooperative retrospective study to evaluate the clinical outcome of pts with PMBL.

Patients and Methods

We analyzed a total of 345 pts with newly diagnosed PMBL from 65 institutes between May 1986 and September 2012 in Japan. Pts were treated according to each institutional protocol or physicians' decisions. In pts treated with R-CHOP, the role of PET before RT was analyzed. In addition, we analyzed prognostic factors for PMBL pts and constructed a novel prognostic model using data from patients treated with R-CHOP.


The median age was 32 (range, 17-83) years, and female pts were predominant (58%) among the patient population. Median tumor diameter was 10 cm. Stage I/II, low-risk by IPI, and PS 0/1 were also predominant (68%, 52% and 75%, respectively). The presence of pleural or pericardial effusion, elevated lactate dehydrogenase level and extra-nodal lesions were observed in 46%, 80% and 44% of pts, respectively.

With a median follow-up of 48 months in surviving pts, overall survival (OS) and progression-free survival (PFS) at 4 years were 87% and 70%, respectively. The OS and PFS were improved in pts treated with rituximab(R)-containing chemotherapy (n = 267) (4-year OS: 91% vs. 77%, P < 0.001; 4-year PFS: 75% vs. 54%, P < 0.001, respectively). The OS at 4 years for patients treated with CHOP (n = 44), R-CHOP (n = 187), DA-EPOCH-R (n = 9), second- or third-generation regimens (n = 45; 28 with R and 17 without R), and chemotherapy followed by ASCT (n = 57; 43 with R and 14 without R) were 67%, 90%, 100%, 91% and 92%, respectively (P < 0.001). The PFS at 4 years were 40%, 71%, 100%, 83% and 76%, respectively (P < 0.001) (Figure 1). Consolidative RT was given to 42% of the patient population.

A total of 119 of 187 pts treated with R-CHOP were assessed by PET/CT after the completion of R-CHOP, and 64 pts received consolidative RT after R-CHOP. In pts with negative PET after R-CHOP (n = 84), the OS (100% vs. 100%, P > 0.99) and PFS (92% vs. 72%, P = 0.36) at 4 years were similar when comparing pts treated with (n =25) or without RT (n =59). However, in pts with positive PET after R-CHOP (n = 28), both OS (100% vs. 60%, P = 0.010) and PFS (80% vs. 17%, P < 0.001) at 4 years were superior in pts who received RT (n = 21) than in pts who did not receive RT (n = 7).

A total of 97 pts (28%) relapsed or progressed after first-line therapy. Of these, 67 (19%) and 11 pts (3%) relapsed in the mediastinum and CNS, respectively. Median time from initial diagnosis to relapse or progression was 9 months. Median OS after relapse or progression was 16 months and was higher in patients treated with stem-cell transplantation (SCT) (n = 58; 44 ASCT, 14 allogeneic SCT) than in patients who did not undergo SCT (4-year OS: 67% vs. 31%, P < 0.001).

The IPI was predictive for OS (P = 0.002) and PFS (P < 0.001) in pts with PMBL treated with R-CHOP. Moreover, multivariate analysis showed that the presence of pleural or pericardial effusion was a significant and independent prognostic factor for PFS. We constructed a novel prognostic model (PMBL prognostic index; PMBIPI) and classified pts treated with R-CHOP into three different risk groups using these two factors (the presence of pleural or pericardial effusion, and IPI high/intermediate-risk or high-risk). For 93 pts (51%) classified as the low-risk group (0 factor), OS and PFS at 4 years were 97% and 89%, respectively. For 61 pts (34%) classified as the intermediate-risk group (1 factor), OS and PFS at 4 years were 85% and 59%, respectively. For 27 pts (15%) classified as the high-risk group (2 factors), OS and PFS at 4 years were 72% (P = 0.001) and 44% (P < 0.001), respectively (Figure 2).


The combination of R and chemotherapy improved outcomes for patients with PMBL. In addition, PET could predict the necessity for RT in pts with PMBL treated with R-CHOP. PMBIPI is a promising tool for risk-stratification of pts with PMBL. These findings require further validation in prospective studies.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract