The primary objective of this study is to evaluate the efficacy and safety of induction therapy R-ICE in comparison to R-DHAP after 3 cycles adjusted to successful mobilization of stem cells in patients with previously treated diffuse large B-cell lymphoma CD20.

The goal is to detect a difference in mobilization adjusted response rate of 15% between R-ICE and R-DHAP.

The other objective is to evaluate the efficacy and safety of MabThera maintenance therapy after transplantation as measured by the event free survival.

The goal is to obtain a 15% increase of event free survival at 2 years.

In vitro, the addition of rituximab to standard anticancer drugs increases cell lyses even in chemoresistant cell lines. This chemosensitization effect was also demonstrated in vivo by the results of the GELA trial in elderly patients with DLCL. Reported phase II study results on the RICE regimen for treatment of patients with relapsed DLCL and comparison with historical controls being treated with ICE suggests that this effect (15% improvement in response rate) is likely in relapsing DLCL and had led the SWOG to stop a randomized trial comparing ICE vs RICE in patients with relapsed aggressive lymphoma.

In the setting of relapsed DLCL, high dose therapy (HDT) followed by autologous stem cell transplantation (ASCT) remains the standard to improve survival in highly selected chemosensitive patients. In the Parma study, only 58% of the patients with relapsed aggressive NHL were good responders after DHAP and 24% were in complete remission. Moreover, the quality of response depended on prognostic factors such as IPI and relapse > 12 months after treatment, and only patients responding to salvage therapy benefited from HDT + ASCT. As shown in the PARMA study. The goal in relapsed DLCL is to improve complete response rates before transplantation as it is the main parameter for eligibility for HDT + ASCT and the main prognostic factor. Unlike first line treatment with CHOP, no standard chemotherapy exists for relapsing patients. DHAP has been the most frequently used regimen for decades but incorporates only two drugs, and has dose-limiting renal toxicity. The ICE regimen was developed at several dosages and studies consistently produced CR rates that were 10-15% superior to DHAP. It is expected that this difference will remain the same with the addition of rituximab to both regimens. Recent phase II data in patients with relapsed DLCL not previously treated with rituximab showed that RICE produced a response rate of 78% with a complete remission rate of 58% and was active in primary refractory disease as well as in intermediate-high risk patients (IPI 2-3). Association of DHAP to Rituximab, R-DHAP has been done on small series of patients by investigators, including patients relapsing after autotransplant. Despite numerous phase II studies, no randomized study has been performed comparing the two regimens (DHAP/ICE) or others in relapsing DLCL. Treatment of first line DLCL has been changed in the past 10 years with more intensive regimens, often followed by ASCT, and very recently with the addition of rituximab to chemotherapy and therefore the population of relapsing patients might be different from the one in the initial PARMA study. A large lymphoma intergroup study working on a large prospective data base might help to find the best salvage regimen and to assess the role of retreatment with monoclonal antibodies in these patients. Finally, the role of rituximab maintenance therapy after HDT + ASCT in prolonging second complete response should be evaluated.

Inclusion Criteria:

• Patients with CD20-positive diffuse large B-cell lymphoma. Disease must be histologically proven in case of relapse or partial response.
• Aged 18 to 65 years
• First relapse after complete remission (CR), less than partial remission (PR) or partial response to first line treatment not achieving documented or confirmed complete remission.
• Eligible for transplant
• Previously treated with chemotherapy regimen containing anthracyclines with or without rituximab.
• ECOG performance status 0 to 2.
• Minimum life expectancy of 3 months.
• Signed written informed consent prior to randomization.

Exclusion Criteria:

• Burkitt, mantle-cell and T-cell lymphoma.
• CD20-negative diffuse large cell lymphoma
• Documented infection with HIV and hepatitis B virus [HBV] (in the absence of vaccination)
• Central nervous system or meningeal involvement by lymphoma.
• Not previously treated with anthracycline-containing regimens
• Prior transplantation
• Contra-indication to any drug contained in the chemotherapy regimens.
• Any serious active disease or co-morbid condition (according to the investigator's decision and information provided in the Investigational Drug Brochure [IDB]).
• Poor renal function (creatinine level > 150µmol/l or 1.5-2.0 x upper limit of normal [ULN]); poor hepatic function (total bilirubin level > 30mmol/l [> 1.5 x ULN], transaminases > 2.5 maximum normal level) unless these abnormalities are related to the lymphoma; poor bone marrow reserve as defined by neutrophils < 1.5G/l or platelets < 100G/l, unless related to bone marrow infiltration.
• Any history of cancer during the last 5 years with the exception of non-melanoma skin tumors or stage 0 (in situ) cervical carcinoma.
• Treatment with any investigational drug within 30 days before planned first cycle of chemotherapy and during the study.
• Pregnant women
• Adult patients unable to provide informed consent because of intellectual impairment.