Adult patients with advanced non-Hodgkin's lymphoma in whom conventional chemotherapy has failed are seldom cured thereafter. We studied 100 such patients with intermediate-grade or high-grade non-Hodgkin's lymphoma who were subsequently treated with high-dose chemotherapy (61 patients) or high-dose chemotherapy plus total-body irradiation (39 patients), with bone marrow transplantation used for hematologic support. Thirty-four patients had disease that had been refractory to primary chemotherapy, and 66 patients had had a complete remission with primary chemotherapy but later relapsed. Before autologous bone marrow transplantation and high-dose chemotherapy, the 66 relapsed patients had also received conventional salvage chemotherapy; 22 had had no response or had had disease progression (a response termed "resistant relapse"), and 44 patients had responded partially or completely (a response termed "sensitive relapse"). After high-dose therapy and bone marrow transplantation, the actuarial three-year disease-free survival was zero in the refractory group, 14 percent in the resistant-relapse group, and 36 percent in the sensitive-relapse group. Patients who had had a complete remission in response to initial chemotherapy had a higher disease-free survival rate than those who had not (P less than 0.001), and patients with sensitive relapse had a higher disease-free survival rate than those with resistant relapse (P less than 0.003). These results should be considered in the planning or interpretation of trials of salvage chemotherapy in adults with non-Hodgkin's lymphoma.

Patients with PCL treated with regimens that included HDMTX followed by radiotherapy have an improved survival, but not a higher risk of late neurotoxicity as compared with other treatment modalities in this series.

Conformity with EBM was similar to previous reports. Elaboration of treatment strategy within a formal multidisciplinary staff and treatment within a cancer network are both important prognostic factors for optimal clinical care.

Subgroups with different survival expectancy can be identified among patients who are eligible for phase I clinical trials. If confirmed, the proposed prognostic model may be useful for therapeutic decision making in palliative oncology.