BACKGROUND: L-asparaginase is a crucial drug in childhood acute lymphoblastic leukemia (ALL) induction therapy, but much debate remains regarding the optimal formulation and dosage. As minimal residual disease (MRD) can accurately measure extremely low levels of lymphoblasts, it is a sensitive reflection of leukemia cell kill. We utilized MRD to compare the efficacy of Erwinia-asparaginase (Erwinia-asp) and E. coli-asparaginase (E. coli-asp) during induction therapy for childhood ALL. PROCEDURE: Of 116 precursor-B ALL patients, 22 were treated with Erwinia-asp, 90 with E. coli-asp, and 4 were switched from E. coli-asp to Erwinia-asp. MRD levels at the end of induction were analyzed for 90 patients (Erwinia-asp = 16; E. coli-asp = 74). Patients were stratified into MRD > or =10(-2), between 10(-2)-10(-4) and < or =10(-4). Toxicity information during induction was available for 110 patients. RESULTS: MRD was the only significant prognosticator compared to conventional criteria. Patients treated with Erwinia-asp were 6.7 times more likely to have MRD levels > or =10(-2) (P = 0.031), reflecting slower lymphoblast clearance. While non-asparaginase related toxicities were similar in both groups, more E. coli-asp patients experienced severe asparaginase-related toxicity. CONCLUSION: E. coli-asp is superior to Erwinia-asp in childhood ALL induction. Although E. coli-asp is more toxic, this is balanced by better response to therapy. Early response to treatment as measured by MRD is a direct reflection of leukemic cell kill and is a significant prognosticator of eventual outcome, making it a good surrogate marker to evaluate the efficacy of induction drugs in childhood ALL.