A Reduced Daunorubicin Dose during Induction Is Equally Effective As Standard Dose in Children with Acute Lymphoblastic Leukemia (ALL) and Favorable Prognostic Features: Long-Term Results of the Phase 3 Trial AIEOP-BFM ALL 2009

Background: Anthracyclines have contributed to the development of effective antileukemic treatment of ALL. Modern ALL therapy, however, is shifting its attention towards reduction of acute and late toxicities while maintaining the same cure rate. Acute and late sequelae of anthracyclines are of major concern, thus, this trial aimed to clarify their role in low-risk patients (pts).

Patients and Methods: The randomization R1 was included in trial AIEOP-BFM ALL 2009 for pediatric pts (1 to < 18 years of age) with newly diagnosed ALL with the aim to safely reduce the daunorubicin (DNR) dose in induction with non-inferior event-free survival (EFS) and reduced toxicity. Pts were eligible for R1 if they had non-high-risk B-ALL (i.e. good response to the prednisone pre-phase, absence of hypodiploidy or KMT2A::AFF1 rearrangement) with either positivity for ETV6::RUNX1 or rapid response to the first two weeks of induction treatment (flow cytometry d15 with < 0.1% marrow blasts). After two weeks of induction therapy including 2 DNR doses (30 mg/m² on days 8 and 15), pts were randomly assigned to receive either a 3^rd and 4^th DNR dose (on days 22 and 29, control arm, CA) or no further DNR during induction (experimental arm, EA). Other treatment components have been previously described (Rizzari C et al, Hemasphere 2023, 7:e893). Depending on minimal residual disease (MRD, evaluated by real-time quantitative PCR) at the end of induction (EoI) and consolidation (EoC), pts were finally stratified into standard risk (SR), medium risk (MR), or high risk (HR) (Conter V. et al, J Clin Oncol 2023, 42:915-926). Pts who were treated as assigned by randomization R1 were included in the primary analyses on EFS, cumulative incidence of relapse (CIR), and overall survival (OS). Secondary outcome analyses were done by intention to treat (ITT). Incidences of protocol-defined adverse reactions (AR) of special interest, occurring during the randomized treatment phase (from day 22 of induction to start of consolidation) were analyzed in the as-treated population.

Results: Out of 6136 pts enrolled in AIEOP-BFM ALL 2009, 2514 pts (41.0 %) were eligible for R1. Of those, 2079 pts (82.7 %) were randomized to receive either the EA (n=1040) or the CA (n=1039). Forty randomized pts were not included in the treated-as-assigned population because they were either retrospectively not eligible (n=5), died before day 22 (n=1) or did not receive the assigned arm (switch from CA to EA (n=13) or vice versa (n=19) or 3^rd DNR dose given by mistake in EA (n=2)) resulting in 1016 pts in EA and 1023 pts in CA. Patient characteristics were equally distributed between the arms. With a median observation time of 7.0 years, the probability of 5-year EFS in the treated-as-assigned population was 92.7% (standard error, SE 0.8%) in CA and 92.2% (SE 0.9%) in EA. The lower limit of the one-sided 95% confidence interval for the difference was -2.3%, well above the pre-defined non-inferiority margin of -4%. CIR at 5 years was 5.6% (SE 0.7%) in CA compared to 5.9% (SE 0.8%) in EA. Five-year OS was 97.4% (SE 0.5%) in CA and 97.7% (SE 0.5%) in EA. Additional analyses in various subgroups (gender, age, ETV6::RUNX1 status, MRD status at EoI or at EoC, final risk groups SR, MR, or HR) did not reveal any differences in outcome between the randomization arms. Outcome results of ITT analyses were almost identical: 5y-EFS 92.7% (SE 0.8%) in CA, and 92.3% (SE 0.8%) in EA. AR of special interest were observed in 7.4% and 6.1% of pts in CA and EA, respectively (p=0.25). Life-threatening AR occurred in 1.4% in CA and 1.1% in EA (p=0.45), and were fatal in 0.7% and 0.5% of pts, respectively. Analyses of AR types revealed a trend towards a higher incidence of infection-related AR in CA than in EA (2.9% vs 1.7%, p=0.061) due to a four times higher incidence of invasive fungal infections (IFI) in CA than in EA (1.5% vs 0.4%, p=0.0076).

Conclusion: The results of this unique prospective trial in a large patient cohort clearly show that a 50% reduction of the cumulative DNR dose during induction did not compromise the outcome. Although the two different schedules showed almost no difference in the overall incidence of serious toxicity, the risk of IFI could be significantly mitigated by reducing the DNR dose. Our study demonstrates, that one can safely recommend reduced anthracyclines in ALL induction therapy if patients are selected by favorable prognostic factors, such as presence of ETV6::RUNX1, or very fast early treatment response.