Duration and intensity of maintenance chemotherapy in acute lymphoblastic leukaemia: overview of 42 trials involving 12 000 randomised children.

BACKGROUND: The effects on long-term outcome in childhood acute lymphoblastic leukaemia (ALL) of the duration and the intensity of maintenance chemotherapy need to be assessed reliably. With this objective the Childhood ALL Collaborative Group coordinated a worldwide overview of all randomised trials that began before 1987. METHODS: Individual patient data were sought for about 3900 children in trials of longer vs shorter maintenance (eg, 3 vs 2 years), 3700 in trials of intensive "reinduction" chemotherapy during maintenance, and 4400 in trials of various other questions, including 1300 in trials of pulses of vincristine and prednisone (VP) during maintenance. Analyses were of survival in first remission, overall survival, and cause-specific mortality. FINDINGS: Deaths during remission were increased by longer maintenance (2.7 percent vs 1.2 percent), VP pulses (4.0 vs 3.2 percent), and intensive reinduction (4.8 percent vs 3.3 percent), but these increases were counterbalanced by reductions in relapses. Total events (relapse or death) were significantly reduced by longer maintenance (23.3 percent vs 27.6 percent), VP pulses (31.2 percent vs 40.4 percent) and intensive reinduction (27.8 percent vs 35.8 percent) (each 2p<0.001). Many of those who relapsed were successfully re-treated, however, and only for intensive reinduction was overall survival significantly improved (18.5 percent vs 22.3 percent; 2p=0.01). INTERPRETATION: Intensive reinduction chemotherapy in these trials produced an absolute improvement of about 4 percent in long-term survival; if the extra deaths in remission had been avoided, this would have been a 5 percent benefit. Further improvements in survival seem more likely to be obtained with intensive treatment than with longer low-level maintenance.