Primary mediastinal large B-cell lymphoma (PMLBCL): long-term results from a retrospective multicentre Italian experience in 138 patients treated with CHOP or MACOP-B/VACOP-B

The optimal treatment of primary mediastinal large B-cell lymphoma (PMLBCL) is still undefined. In the absence of randomised studies, we retrospectively analysed: (a) the effectiveness of two chemotherapy regimens (CHOP vs MACOP-B/VACOP-B) in complete remission (CR) achievement and event-free survival (EFS) and (b) the role of mediastinal involved-field radiotherapy (IF-RT) as consolidation. From 1982 to 1999, 138 consecutive patients affected by PMLBCL were treated in 13 Italian institutions with CHOP (43) or MACOP-B/VACOP-B (95). The two groups of patients were similar as regard to age, gender, presence of bulky mediastinal mass, pleural effusion, stage and international prognostic indexes category of risk. Overall, 75.5% of patients in CR received IF-RT as consolidation. Complete remission was 51.1% in the CHOP group and 80% in MACOP-B/VACOP-B (P<0.001). Relapse occurred in 22.7% of CHOP- and in 9.2% of MACOP-B/VACOP-B-treated patients (n.s.). Event-free patients were 39.5% in CHOP and 75.7% in the MACOP-B/VACOP-B group (P<0.001). The addition of IF-RT as consolidation improved the outcome, irrespectively of the type of chemotherapy (P=0.04). At a multivariate analysis, achievement of CR (P<0.0001) and type of CT (MACOP-B/VACOP-B) retained the significance for OS (P=0.008) and EFS (P=0.03). In our experience, MACOP-B/VACOP-B appears to positively influence OS and EFS in patients affected by PMLBCL, as compared to CHOP. Consolidation IF-RT on mediastinum further improves the outcome of CR patients.

The high proliferation rate of neoplastic cells (Todeschini et al, 1990;Falini et al, 1995) correlates with the rapid growth and with the frequent presence of necrotic areas in the lymphoma mass (Aisenberg, 1999). Sclerosis can explain the frequent persistence of variable degrees of residual mass even after successful treatment (Rohatiner et al, 1994).
Most published reports on PMLBCL mainly focus on the clinical -pathological features, and few papers only analyse the role of different chemotherapy regimens on the clinical outcome. As a result, the optimal therapy has not been defined yet, as pointed out by a recent review (van Besien et al, 2001).
A possible superiority of third-generation regimens in DLCLs has not been validated by a large prospective randomised study (Fisher et al, 1993). However, this study was not specifically focused on PMLBCLs. By contrast, data from other series suggest that, in this setting, third-generation therapies (usually MACOP-B and VACOP-B) may be more effective than CHOP (Levitt et al, 1982;Todeschini et al, 1990;Bertini et al, 1991;Falini et al, 1995;Cazals-Hatem et al, 1996;Bieri et al, 1999;van Besien et al, 2001;Zinzani et al, 2001Zinzani et al, , 2002.
In absence of randomised trials, we retrospectively analysed a large series of PMLBCL consecutive patients treated with CHOP or MACOP-B/VACOP-B observed in 13 different Italian institutions.
The first aim of this analysis was to compare the long-term results of CHOP vs MACOP-B/VACOP-B, the two most commonly employed regimens in PMLBCL patients in USA and Europe, respectively (van Besien et al, 2001). Patients treated with other regimens and those receiving stem cell transplant (SCT) as consolidation were not analysed. The two therapeutic approaches have been compared, taking into account the recognised international prognostic indexes (IPI) (The International Non-Hodgkin's Lymphoma Prognostic Factors Project, 1993).
The second objective of our analysis was to evaluate the role of consolidation IF-RT on the long-term outcome after CR achievement.
This paper reports our multicentre retrospective experience.

Patients
Our review includes 138 consecutive patients observed in 13 Italian institutions between 1982 and 1999. Patients received routine staging procedures (complete physical examination, laboratory tests including serum LDH and serum beta-2 microglobulin, chest X-ray, CT scan of the chest, abdomen and pelvis, bone marrow aspirate, and biopsy). Stage was defined according to the Ann Arbor staging system. Bulky disease was defined as a mediastinal mass larger than 1/3 of thoracic diameter or a mediastinal mass larger than 10 cm (major diameter).

Diagnostic criteria
All biopsy specimens were reviewed and confirmed according to the WHO criteria for PMLBCL diagnosis. Immunophenotypical analysis was always performed. Other NHLs involving the mediastinum and not fulfilling these criteria were not considered.

Selection criteria
Only patients treated with CHOP or MACOP-B/VACOP-B were included in this report. Patients who underwent other regimens or patients who received SCT after completing CHOP or MACOP-B/ VACOP-B were excluded. Each institution chose treatment (CHOP or MACOP-B/VACOP-B) according to the local policy at the time.
Chemotherapy was based on six cycles of CHOP or on a full (12 weeks) cycle of MACOP-B/VACOP-B. In each group of therapy, patients were divided into low/low-intermediate-and highintermediate/high-risk groups according to IPI. Involved-field radiotherapy was administered in patients in complete remission (CR)/near-complete remission (NCR). The decision to employ IF-RT or not was made according to the local policy of each centre. The modality of IF-RT administration was the following: (a) when the mass involved only the mediastinum, the whole original disease was irradiated; (b) when the mass spread beyond the mediastinum displacing the lung, IF-RT was administered to involve the mediastinum volume plus 1.5 cm outside its borders. The median dosage of IF-RT was 34 Gy (range 30 -45). Only patients who underwent IF-RT after achieving CR or NCR were considered for IF-RT significance on disease outcome.

Response criteria
CR was defined as the disappearance of lymphoma lesions and resolution of symptoms for at least 3 months after the end of CT. Near-complete remission (i.e. unconfirmed CR) was defined as the reduction of mass (lymphoma lesions) 490%, without new growth at physical and imaging examination, without signs and symptoms of disease for at least 3 months. Partial remission (PR) was defined as reduction of at least 50% of the original measurable mass. No remission (NR) was defined as (a) minimal modification (a response less than 50%) or (b) no modification or (c) progression of the mass under CT or (d) new lymphoma growth during CT or within 3 months after the end of CT.
In our review, gallium scan could not be considered in the assessment of CR, since only a minority of patients underwent this procedure. The possible persistence of some degree of fibrosis after CT even in responders makes it difficult to define the real incidence of CR. However, the long-term observation of our study allowed us to clarify this point, since NR or low responsive cases had invariably a rapid fatal outcome. Therefore, the long-term EFS had been judged a posteriori as an indicator of CR achievement. OS was calculated from the beginning of chemotherapy to the death or the last follow-up. EFS was defined as survival from the beginning of chemotherapy, in the absence of unfavourable events such as (1) toxic death, (2) NR, PR or progression, (3) relapse and (4) death due to any cause.

Statistical analysis
Statistical analysis was performed using the w 2 test with Yates correction and with Fisher's exact test for categorical variables. Survival curves were calculated from the beginning of the treatment according to the actuarial method proposed by Kaplan and Meier. Differences between curves were evaluated by log-rank test. Cox regression analysis was performed to determine the independent contribution of the following variables: achievement of CR, type of treatment (CHOP vs MACOP-B/VACOP-B) and IPI score ( 6%); high, eight out of 138 (5.7%). The two risk categories were balanced in CHOP-and MACOP-B/VACOP-B-treated groups ( Table 1).
Of the 37 patients in stage IV, only 17 (45.9%) became event-free survivors. Among the 15 patients with organ involvement beyond the diaphragm (four treated with CHOP, 11 with MACOP-B), only five (33.3%) were event-free (all treated with MACOP-B/ VACOP-B).

Involved-field radiotherapy was considered for patients in CR/NCR
The majority of patients who achieved CR with CT received IF-RT on the mediastinum as consolidation (74 out of 98: 75.5%).
Kaplan -Meier survival curve performed on the population who achieved CR showed a statistically significant difference in eventfree survival time in patients receiving IF-RT (P ¼ 0.04) (Figure 2).
Considering the achievement of CR, the type of treatment and the IPI levels as covariates, a multivariate Cox regression analysis for survival indicated that the achievement of CR and the type of treatment (MACOP-B/VACOP-B) were statistically significant. Differences between IPI levels did not achieve this condition (Table 4).

DISCUSSION
Our analysis was focused on CHOP vs MACOP/VACOP-B, the most commonly regimens used in PMLBCL. The long-term results provided by this retrospective experience indicate that, in this setting, MACOP-B/VACOP-B achieve better results than CHOP, and confirm our previous experience (Todeschini et al, 1990). Additional retrospective data are in agreement with our observation (Bertini et al, 1991;Lazzarino et al, 1993;Falini et al, 1995;Cazals-Hatem et al, 1996;Lazzarino et al, 1997;Martelli et al, 1998;Bieri et al, 1999;Zinzani et al, 2001). Similar results were recently observed in a large retrospective study (Zinzani et al, 2002), which included patients treated with chemotherapy other than CHOP or MACOP/VACOP-B. Limiting the observations to CHOP and third-generation therapies, the authors found a projected 10-years progression-free survival (PFS) and OS of 35 and 44% with CHOP, and 67 and 71% with third-generation therapies, respectively. Although a large randomised study (Fisher et al, 1993) showed no difference in outcome for intermediate-high-grade NHLs, in that study, the subgroup of PMLBCL was not separately considered.
In our series, advantages were statistically significant in the MACOP-B/VACOP-B-treated group at low/low-intermediate risk (P ¼ 0.001). In the small group of high-intermediate/high-risk patients, an advantage from MACOP-B/VACOP-B as compared to CHOP was observed in terms of CR rate and EFS, but it was not statistically significant (P ¼ 0.068).
Based on the results of our experience, we suggest that MACOP-B/VACOP-B can be more effective than CHOP in the treatment of PMLBCL. This assumption is supported by three observations. First, although this is a retrospective review, the clinical characteristics of the two groups of patients were balanced (age, bulky mediastinum, pleural and pericardial effusion, presence of SVCS, stage I-II vs III-IV, type of chemotherapy during the study period) (Table 1).
Second, the IPI risk categories were equally distributed in the two therapeutic groups (Table 1).
Third, the two CT regimens were administered almost uniformly in the study time, thus avoiding the occurrence of possible biases due to different supportive measures.
The achievement of CR was the most significant prognostic factor both for OS and EFS (Po0.0001) ( Table 4). A high percentage of patients who reached CR remained event-free, thus suggesting that a therapeutic approach able to increase the CR rate could influence the long-term OS and EFS. In fact, in both chemotherapy groups of our series, patients unresponsive or progressing under CT had invariably a fatal outcome (26 out of 26 died of lymphoma).
Relapse rarely affects PMLBCL patients, although it is the main cause of failure in other large-cell lymphomas. In our series, a very poor outcome was observed in the few relapsed patients (12.2%), since all of them died of lymphoma, independently of first-line therapy and subsequent salvage approaches (Table 2). Involvedfield radiotherapy was ineffective in controlling the disease progression too. These findings are in agreement with the majority of reports (Moller et al, 1987;Jacobson et al, 1988;Cazals-Hatem et al, 1996;Zinzani et al, 2001), but contrast with the relatively good results described by other authors (Popat et al, 1998;Sehn et al, 1998) in small groups of refractory/relapsed patients treated with stem cell rescue.
In PMLBCL, the dismal outcome of relapsed/refractory patients seems even worser than in other DLCLs, and it appears as a further characteristic of this lymphoma entity. In fact, in other aggressive NHLs, a variable percentage of patients achieve a second CR and a small but not negligible proportion of them obtain a prolonged survival.
In our experience, patients with advanced stage of disease (i.e. subdiaphragmatic organ involvement) had a lower achievement of CR and a very poor outcome, suggesting that in these patients more intensive approaches, including stem cell transplantation, should be early considered.
Consolidation IF-RT is largely employed in PMLBCL patients, but its role is unclear. In our experience, the administration of IF-RT after CR achievement showed a positive impact on EFS, irrespective of the type of chemotherapy administered (P ¼ 0.04) ( Figure 2). The favourable role of IF-RT is suggested by a recent large retrospective analysis (Zinzani et al, 2002).
The relative high percentage of NR in CHOP-treated patients and the extremely poor outcome of NR/relapsed patients suggest that the maximum effort should be done in the early phase of the disease, given the critical importance of the first-line therapy.
In fact, at the multivariate analysis, CR achievement and type of chemotherapy (MACOP-B, VACOP-B) were independent prognostic factors for EFS, while IPI did not result to be significant (Table 4).
According to these results and in the absence of randomised studies, MACOP-B/VACOP-B followed by IF-RT should be provisionally recommended for PMLBCL patients.
Randomised prospective studies are needed to define the optimal first-line chemotherapy and to assess the role of mediastinal IF-RT in PMLBCL treatment.