Event-free survival (EFS) and overall survival (OS) of paediatric patients with acute myeloid leukaemia (AML) has increased due to chemotherapy intensification, increased allocation to hematopoietic stem cell transplant (HSCT) in first complete remission (CR1) based upon high-risk genetic features or measurable residual disease (MRD) positivity and enhanced supportive care measures. Successful retrieval of the 30–35 % of children with subsequent relapsed AML largely depends upon the ability to proceed to consolidative HSCT in second complete remission (CR2), ideally with negative pre-transplant MRD [1,2]. While conventional chemotherapy salvage regimens can be effective, significant interest exists in the emerging success of precision medicine approaches with biologically relevant targeted inhibitors or immunotherapies (Figure 1), which are summarized herein [3,4].

At time of relapse, restaging of bone marrow, central nervous system (CNS), and other extramedullary sites (if clinically applicable) is necessary to guide treatment decision-making and subsequent response assessment. Complete cytomolecular genetic characterisation of relapsed AML is also essential to identify major oncogenic drivers (eg, translocations/genetic fusions) and potential mutational evolution that may confer prognostic significance and/or highlight potential opportunities for precision medicine therapies [1,2].

No current standard of care exists for paediatric patients with first relapse of AML, and approaches are even more heterogeneous at second or greater relapse. Prior clinical trials studied the efficacy of clofarabine-, bortezomib- [5], or liposomal daunomycin-based [6] salvage chemotherapy regimens in children with relapsed AML with varying success. More recently, CR2 rates of ≥ 70 % have been achieved in children with relapsed AML treated with fludarabine and cytarabine without or with granulocyte colony-stimulating factor support FLA(G). The recent Children's Oncology Group (COG) AAML1421 phase 2 study utilised CPX-351, a liposomal preparation of cytarabine and daunomycin, in cycle 1 and FLAG in cycle 2 to achieve an overall response rate of 81 % [7]. High rates of remission induction have also been achieved in adults with relapsed/refractory or newly-diagnosed AML with addition of idarubicin to fludarabine and cytarabine (FLAG-ida) [8] without or with the BCL-2 inhibitor venetoclax [9]. A recent paediatric institutional case series further reported safety and efficacy of FLAG-ida as induction therapy for children and adolescents with newly diagnosed AML [10].

The choice of salvage regimen for pediatric patients with first relapse of AML depends upon timing and site of relapse, cumulative anthracycline chemotherapy exposure, prior HSCT in CR1 (for high-risk leukemia genetics or end-induction 1 MRD positivity), history of significant end-organ toxicities, and goals of care. Anthracycline-containing reinduction regimens (Figure 2, pale green) should be prioritized for patients with prior cumulative anthracycline exposure <450 mg/m2 and may be considered with caution on a patient-specific basis for those with ≥ 450 mg/m2 exposure with appropriate cardiac function by echocardiogram assessment, although non-anthracycline-based regimens (Figure 2, pale blue) are preferentially initially recommended for the latter population. For patients with very early relapse after haematopoietic stem cell transplantation (HSCT) in first remission (CR1) for whom intensive myelosuppressive chemotherapy is clinically inadvisable, reinduction with hypomethylating agents (eg, azacytidine) in combination with BCL-2 inhibitors (eg, venetoclax; discussed in further detail below) may be considered (Figure 2, pale grey) if accessible [11].

Advanced molecular diagnostics and improved understanding of leukaemia biology potentially targetable by specific small molecule inhibitors have appreciably advanced precision medicine approaches for children with relapsed AML with advancement of some drugs to the frontline setting [12]. For instance, first-generation (eg midostaurin, sorafenib) and second-generation (eg, crenolanib, gilteritinib, quizartinib) FLT3 inhibitors have demonstrated efficacy and/or safety in combination with multi-agent chemotherapy in paediatric patients with relapsed or newly-diagnosed FLT3-mutant AML. Several FLT3 inhibitors are also now used or under study as post-HSCT maintenance therapy (Table 1) [13, 14, 15, 16].

Early-phase relapsed paediatric AML clinical trials have recently reported high rates of remission induction with venetoclax in combination with high-dose cytarabine ± idarubicin, and other studies of venetoclax in combination with multi-agent chemotherapy are ongoing [17]. The randomised APAL2020D phase 3 trial for paediatric patients with first or greater relapse of AML is currently assessing the therapeutic potential of venetoclax addition to a FLAG and CD33 antibody-drug conjugate gemtuzumab ozogamicin (GO)-based regimen (NCT05183035). Hypomethylating agents (eg, azacytidine, decitabine) in combination with histone deacetylase inhibitors (eg, panobinostat, vorinostat) have also been promising in the relapse setting [18,19]. Combining hypomethylating agents with checkpoint inhibitors (eg, nivolumab) has been less successful in the paediatric domain despite initially promising adult data [20].

Menin inhibitors (eg, bleximenib, enzomenib, revumenib, ziftomenib) have demonstrated 20–30 % CR rates in adults and children with relapsed KMT2A-rearranged or NPM1-mutant acute leukaemias and 80–100 % CR rates when used in combination with multi-agent chemotherapy in the frontline setting [21∗∗, 22, 23∗∗, 24, 25, 26, 27]. Revumenib was approved in 2024 by the United States Food and Drug Administration (FDA) for patients ≥ 1 year of age with relapsed/refractory KMT2A-rearranged acute leukemias. Additional paediatric-inclusive or paediatric-specific trials of menin inhibitors are underway, some with broader genetic eligibility relevant to children and adolescents with relapsed AML, including NUP98 rearrangements and other alterations known to activate HOX family genes/MEIS1 pathways (eg, UBTF tandem duplication) (Table 1) [28,29].

Additional small molecule inhibitors have been investigated in children with multiply-relapsed/refractory AML via recently-completed or current early phase clinical trials, including XPO1 (selinexor) [30], CREB (niclosamide), e-selectin (uproleselan), IDH2 (enasidenib), MDM2/MDMX (ALRN-6924), NEDD8 (pevonedistat), and telomerase (imetelstat) targets (Table 1) and represent additional therapeutic considerations if biologically appropriate and/or available.

Appreciable interest in immunotherapeutic approaches to overcome chemoresistance in relapsed/refractory AML exists. The heterogenous genetic and immunophenotypic landscape of childhood AML and the need to balance desired on-target/on-tumor efficacy with potential on-target/off-tumor toxicity with antigen targeting have been major challenges to date, although recent progress has been made in several domains [31∗, 32∗, 33∗].

GO is FDA- and European Medicines Agency-approved for children with newly-diagnosed and/or relapsed AML following favourable safety data and improved survival demonstrated in international paediatric clinical trials and prior reports of inferior clinical outcomes in children with AML and highest CD33 cell surface expression [34, 35, 36, 37]. GO addition to multi-agent chemotherapy appears particularly beneficial for children with FLT3-mutant or KMT2A-rearranged AML [38, 39∗, 40]. The active APAL2020D phase 3 clinical trial (NCT05183035) will further establish the baseline efficacy of the FLA + GO regimen in children with relapsed/refractory AML in addition to answering a randomized venetoclax question as aforementioned (Table 1). Various CD33-redirected chimeric antigen receptor (CAR) T cell immunotherapies are also under early-phase clinical investigation in children and AYAs with relapsed/refractory AML based upon promising preclinical data (Table 2) [41, 42, 43].

Similarly inferior clinical outcomes of children with AML with highest CD123 (IL-3 receptor alpha) surface expression have been reported [44]. Various CD123-targeting immunotherapies have been studied or are under active investigation in paediatric-specific or paediatric-inclusive phase 1 clinical trials, including the CD123xCD3 bispecific antibody flotetuzumab [45], CD123 ADC tagraxofusp [46], CD123-redirected CAR T cells [47,48], and CD123 NK cell engager SAR443579 [49] (Table 2).

Bench-to-bedside translation of additional promising antibody-based and cellular immunotherapeutic strategies targeting other cell surface antigens, including CD4, CD7, CD38, CD47, CD70, CLEC2A, CLEC12A (CLL-1, CD371), FLT3 [50], and FOLR1 [51], are also being pursued in relevant genetic and biologic subtypes of childhood AML. CAR T cell therapies directed against immunomodulatory receptors in the tumor microenvironment (eg, LILRB4) or natural killer (NK) cell ligands (eg, NKG2DL) have also demonstrated activity in preclinical pediatric acute leukemia models [52,53]. Allogeneic CD33-or NKG2DL-targeting CAR-NK cells have also been developed for adults, but they have not yet been studied in children [54, 55, 56]. AML antigen-agnostic cellular immunotherapy approaches, including HA-1 T cell receptor T cells [57] and cytokine-induced memory-like NK cells have also been studied in children with relapsed/refractory AML (Table 2) [58,59].

At the present time, evaluation of new immunotherapies via phase 1 clinical trials in children and adolescents with AML is restricted to those with second or greater relapse/refractory disease. Consideration of such trials is encouraged if biologically and clinically appropriate given the known chemoresistance and poor salvage of pediatric patients with multiply-relapsed/refractory AML with conventional therapies [60].