Prognostic Impact of Age in Children and Adolescents with Acute Lymphoblastic Leukemia: Data from the Trials ALL-BFM 86, 90, and 95

Abstract

Large progress has been made in the treatment of acute lymphoblastic leukemia (ALL) of childhood and adolescence over the past 30 years. Eighty percent of the patients can be cured, but clinical subgroups with a dismal outcome can still be identified. In this study, we investigated the association of age with prognosis in 5 181 patients with ALL under 18 years (y) of age enrolled in the three consecutive treatment trials ALL-BFM 86, 90 and 95 in more than 80 centers. Event-free survival (pEFS) of the total group was significantly associated with age. The most unfavorable outcome was found in infancy and the best results were achieved at toddler and pre-school age. Beyond 5 y of age, survival probability decreased (pEFS at 8 y: < 1 y = 0.45; 1-5 y = 0.82; 6-9 y = 0.75; 10-14 y = 0.63; > or = 15 y = 0.59). The proportion of T-ALL as compared to precursor B-cell ALL (pB-ALL) was lower in younger children, due to an incidence peak of pB-ALL in toddlers and at pre-school age compared to a constant incidence of T-ALL. Within the T-ALL group, no correlation of age with sex, initial white blood cell count, CNS disease, or early treatment response was found. Children under 10 y of age had a slightly lower relapse rate compared to older patients. Within pB-ALL patients, the proportion as well as the absolute incidence of TEL/AML1 rearrangement and DNA index of > or = 1.16 was higher in the younger children. A lower proportion of BCR/ABL-positive ALL was observed in the age group of < 6 y when compared to patients aged > or = 6 y, but the absolute incidence was constant across the age groups after the first year of life. More than half of the infants had a CD10-negative pB-ALL. The incidence was constant after a peak in the first year of life, yet the percentage of CD10 negativity increased with rising age in this subgroup. Adolescents with pB-ALL had a significantly higher proportion of prednisone poor-responders. Accordingly, outcome was worse in older patients. This pattern was also evident in the biologically heterogeneous group of patients with a DNA index of > or = 1.16. In contrast, no significant age-related outcome differences could be shown within TEL/AML1- or BCR/ABL-positive patients, as well as within CD10-negative pB-ALL beyond infant age. Analysis of the pB-ALL group in a Cox's regression model including age and the above-listed biological factors revealed age < 1 year and > or = 10 years as independent risk factors. This is in line with the poorer prognosis of these age groups in the pB-ALL subgroup without specific biological characteristics. This subgroup also had an incidence peak at toddler age, presumably containing other favorable biological subsets. An independent prognostic impact of age in pediatric ALL cannot be excluded by this study. However, our analyses show that the age-associated different prognosis in childhood ALL is at least partly related to the different distribution of relevant prognostic subgroups between the age groups.