Prognostic Significance of Time to Relapse in Pediatric AML: Results from the International Randomised Phase III Study Relapsed AML 2001/01

The duration of the first complete remission is an accepted prognostic factor in pediatric relapsed AML, but a detailed analysis of the time to relapse: very early (<6 months from diagnosis), early (6–12 mo), late (12–24 mo) and very late (>24 mo from diagnosis) is lacking. This was therefore studied in patients who were enrolled in study Relapsed AML 2001/01. This prospective, randomised study for pediatric relapsed AML, excluding AML M3 and those >18 years of age at initial diagnosis, opened in most countries in 2002/2003. FLAG is given for 2 consecutive courses: fludarabine 30 mg/m²/day × 5 days, cytarabine 2 g/m²/day × 5 days, G-CSF 200 μg/m²/dose × 6 days, starting day −1. Liposomal daunorubicin (DaunoXome^®, DNX; a new anthracycline with potentially less cardiotoxicity), at 60 mg/m2/day on days 1, 3 and 5 is added or not in a 1:1 randomised fashion to the first course of FLAG. After two courses, patients are eligible for allogeneic stem cell transplantation (allo-SCT), sometimes bridged by high-or low- intensity consolidation chemotherapy. Efficacy data for both arms are still blinded as the study remains open until late 2008. Early treatment response is defined as good in patients with ≤20% blasts in bone marrow (BM) at “day 28” (taken shortly before the start of the 2^nd reinduction course), and as poor in patients with >20% BM blasts at day 28. Complete remission is determined after 2 reinduction courses using conventional criteria. Thirteen groups worldwide are enrolling patients. More than 500 patients were registered by April 2008, but this analysis relates to 427 eligible and fully evaluable patients diagnosed before 1 July 2007, also excluding patients who had not achieved documented first CR with initial firstline treatment. The majority of patients in study Relapsed AML 2001/01 progressed to an allo-SCT, with the exception of the subgroup of very early relapsed patients, many of whom had an early death and progressive disease which precluded allo-SCT. A total of 34 patients relapsed very early, 173 early, 162 late and 58 very late. So-called favourable cytogenetics, i.e. t(8;21) or inv(16), are found in all subgroups, but more frequently in the late/very late relapses. The table shows the outcome by time to relapse and treatment response. There is a significant trend (p<0.0001) for a higher rate of good early treatment response, rate of CR2 achieved and for an improved survival for patients who had a longer CR1 and relapsed late.

Outcome in the very early relapsed AML subgroup is particularly poor, even in the patients with a good early response (BM blasts day 28 ≤20%), in contrast to the relatively favourable outcome for good early responders in all other subgroups (see table). Overall probability of survival for the whole cohort at 5-years is 32% (S.E. 3%). In conclusion, the rate of good early treatment response at relapse, achievement of CR2 and probability of survival from relapsed AML all significantly increase with longer duration of CR1. Hypothetically, this may be explained by differences in drug resistance and relapse potential of residual leukemia cells. Patient-selection may have resulted in fewer very high-risk patients not being enrolled into this study, however the overall results are encouraging. High-dose cytarabine as part of the FLAG regimen, potentially DNX, and the use of allo-SCT may have contributed to the overall reasonably favourable outcome in this study on pediatric relapsed AML, with a probability of long-term survival of more than 30%. This justifies initiation of intensive reinduction chemotherapy in all pediatric patients with relapsed AML, perhaps with the exception of patients relapsing within 6 months of initial diagnosis (very early relapse), in whom palliative therapy or more experimental studies should be considered.