The impact of preventive central nervous system irradiation (CNS-RT) in childhood acute myeloid leukemia (AML) is still discussed. As results of study AML-BFM87 revealed an increased risk for relapse when CNS-RT was not performed, studies AML-BFM98 and -2004 randomized CNS-RT of 18 or 12 Gy in order to evaluate the efficacy of the lower dose and to reduce late effects.

To achieve a power of 80% for non-inferiority (range 11%) 240 patients per group were required. Out of 722 eligible patients, 486 patients <18 years were randomized to receive 12 Gy (n = 249) or 18 Gy (n = 237). Since this was a non-inferiority study, the analysis was performed for patients as treated (12 Gy: n = 252 and 18 Gy: n = 219).

Five-year survival, event-free survival and cumulative incidence of relapse were similar in patients who received 12 or 18 Gy, respectively (82 ± 3% vs. 79 ± 3%, 68 ± 3% vs. 63 ± 3%, and 30 ± 3% vs. 34 ± 3%). The lower limit of the one-sided confidence interval for the -5% difference in 5-years pEFS was 2%. There were six relapses with CNS involvement (one in the 12 Gy, and five in the 18 Gy group).

Results demonstrate no disadvantage for patients irradiated with a reduced CNS dose of 12 Gy.