RESUMO
Potential predictors of event-free survival (EFS) were assessed in 251 consecutively diagnosed children treated for acute myeloid leukemia (AML) on three successive clinical trials. The lack of significant differences in 4-year EFS for these studies (20% +/- 4%, 29% +/- 4%, and 20% +/- 7%) permitted combined analysis of presenting features. Splenomegaly (P = .002), coagulation abnormalities (P = .001), leukocyte count > or = 10 x 10(9)/L (P = .002), and age > 14 years (P = .01) were statistically significant predictors of a poorer EFS by univariate analysis and retained significance in multivariate analysis. Age < 2 years and monocytic leukemias (often cited as adverse factors in AML) showed no prognostic influence in this study. The estimated relative risk of failure for a child with a single adverse feature at diagnosis was at least 1.4 times greater than that for a patient with no adverse features. For children with two or more adverse features, the relative risk increased by more than threefold. These clinical variables, alone or in combination, may identify important subgroups of patients with AML at high risk for failure and for whom improved or alternative therapies are especially important.