Background: Relapses continue to be problematic for adults with acute lymphoblastic leukemia (ALL). New therapies generally are first tested in the salvage setting prior to incorporation into frontline regimens. Defining the prognosis at relapse (or at failure of induction) and subsequently predicting outcome would be useful to select the population in whom to test new strategies, rather than attempting traditional reinduction therapy.

Methods: Between March 1980 and March 1997, 314 eligible adults with primary refractory (24%) or primary relapsed (76%) ALL were treated with various chemotherapy or stem cell transplantation (SCT) regimens. The Cox proportional hazards model was used to assess biologic factors and disease history in relation to survival.

Results: A complete remission (CR) was achieved in 97 patients (31%), 21% died prior to a response, and 49% were refractory to salvage therapy. Of the 76 patients refractory to induction therapy for their de novo ALL, 26 patients (34%) achieved a CR with salvage therapy. The median overall CR duration was 6 months. The median overall survival was 5 months; 24% of the patients were alive at 1 year, and the projected survival at 5 years was 3%. Nineteen patients were alive at the time of last follow-up, 10 with 6 weeks to 10 years of continuous CR from the time of their first salvage therapy. SCT consolidation in second CR was performed in 25% of patients; 28% of those who received allogeneic SCT remain in continuous CR at 4 months, 2(1/2) years, 3(1/2) years, and 10 years, whereas all 8 who received autologous SCT have relapsed. Favorable factors for longer survival by multivariate analysis were age <40 years, absence of circulating blasts, and first CR duration longer than 1 year. Patients were stratified into 4 risk groups: Group 1, with no unfavorable features or only short initial CR duration; Group 2, with only increased age or peripheral blasts; Group 3, with any 2 unfavorable features; and Group 4, with all 3 unfavorable features. The median survival times for each group were 11, 6, 4, and 2 months, respectively; 1-year survival rates were 44%, 25%, 12%, and 9%, respectively (P < 0.01). The resulting model was also predictive for CR rates; the corresponding CR rates were 47%, 35%, 14%, and 9%, respectively (P < 0.01).

Conclusions: Salvage therapy for adult ALL patients continues to yield poor results, but it is an area of research where it may be possible to discover new agents or strategies to be incorporated into frontline therapy. The prognostic model derived will be utilized prospectively to select patients for new therapeutic strategies involving such novel agents as liposomal compounds, purine nucleoside phosphorylase inhibitors, and monoclonal antibodies.