Purpose: Using the technique of recursive partitioning and amalgamation analysis with verification, the Pediatric Oncology Group (POG) investigated the independent prognostic significance of previously published prognostic factors significantly associated with event-free survival (EFS) in B-progenitor cell acute lymphoblastic leukemia (ALL).
Patients and methods: Age, leukocyte count, sex, immunophenotype (expression of cytoplasmic immunoglobulin [Ig] and of surface antigens CD10 and CD34), and DNA index (ratio of the flow cytometry-determined DNA content of leukemia cells to that of normal diploid cells) were the variables used in the evaluation of four antimetabolite-based chemotherapy regimens in 1,535 children with the newly diagnosed B-progenitor cell ALL between February 1986 and May 1990.
Results: There were three subgroups at widely different risks of treatment failure. A DNA index greater than 1.16 was the most prognostic feature. The final prognostic subgrouping was as follows: (1) DNA index greater than 1.16; (2) DNA index less than or equal to 1.16, age less than 11.0 years, and leukocyte count less than 50 x 10(9)/L; and (3) DNA index less than or equal to 1.16, (age greater than 11.0 years, and/or leukocyte count greater than 50 x 10(9)/L). These groups made up 20%, 53%, and 27% of the patients and had 4-year EFS rates (SE) of 90.1% (6.3%), 80.5% (5.1%), and 50.4% (7.6%), respectively.
Conclusions: Use of the DNA index, leukocyte count, and age--data that are relatively inexpensive and simple to obtain--may be sufficient to stratify patients with B-progenitor cell ALL for risk-directed therapy. Patients at an extremely low risk of failing therapy (approximately 20% of cases in this study) can thus be identified and spared the toxic short-term and late effects of more intensive therapies that may be needed for children with less favorable clinical and biologic features.