Pediatric acute myeloid leukemia (AML) is a heterogeneous clonal disorder characterized by malignant transformation of the hematopoietic stem cell (HSC). Pediatric AML is less common than acute lymphoblastic leukemia (ALL); while ALL accounts for 80% of all pediatric acute leukemias, AML barely reaches 15–20%. Pediatric AML has a bimodal age distribution, with higher incidence in patients younger than 2 years and a second peak in adolescents up to 10 to 20 years old [1].

Pediatric AML is generally characterized by poorer prognosis compared to ALL although outcomes have improved in the recent years. The 3-year overall survival (OS) is 70% or even higher ^[2][3][4][2,3,4]. Notably, 30% of children with AML relapse after treatment with very poor OS. The AML BFM study group reported a 5-year OS of 45%± 4%, with fewer early deaths (8.1%vs.2.2%) ^[5][6][5,6].

While the majority of pediatric AML occurs de novo, there is an increased incidence of AML reported in certain hereditary disorders such as Fanconi anemia, Kostmann syndrome, Shwachman–Diamond syndrome, and Diamond–Blackfan anemia, which may be related to alterations of DNA repair or detoxification of reactive oxygen species genes to pathogenetic variants in telomere biology genes or ribosome function ^[7][8].Children with Down syndrome (DS) have an increased risk of developing AML, particularly the megakaryoblast (MK) subtype. Indeed, the most common genetic factor associated with the development of AML is trisomy 21 ^[8][9].Germ-line mutations were found in families with a high risk of AML, suggesting a familial predisposition ^[9][10].DS patients also harbor a GATA1 mutation that, as broadly established, can lead to the development of a transient myeloproliferative disease (TDM), which commonly resolves without any treatment. Residual cells may undergo apoptosis or acquire additional mutations leading to acute megakaryocytic leukemia (AMKL) with an average latency of 3 years. Nevertheless, these patients typically have less-aggressive disease, with a significantly longer disease-free survival (DFS) compared to other types of pediatric AML. This is possibly due the fact that DS-AMKL requires less intense therapy to achieve a cure compared to non-DS AMKL, showing that children with AML-DS are more responsive to chemotherapy ^[10][11][11,12].

AML is derived from a clonal proliferation and the abnormal differentiation of myeloid stem cells as a consequence of two sequential genetic events. This process is known as the two-hit model of leukemogenesis ^[12][13]. In AML, there are two different types of relevant mutations: class I mutations frequently result in uncontrolled proliferation and include receptor or cytoplasmatic–nuclear tyrosine kinase mutations such as FLT3, K/NRAS, TP53, and c-KIT, in 20–25%, 40%, 2%, and 12% cases, respectively, and do not affect differentiation. In contrast, class II mutations involve cellular differentiation arrest and/or self-renewal ^[13][14][7,14]. The most common type II cytogenetic abnormalities in children are t(8;21)(q22;q22), involving core-binding factor (CBF) β AML, as well as inv 16 or t(16;16), and t(15;17)(q22;q21) in acute promyelocytic leukemia (APL). Specific pediatric translocations rarely detected in adults include t(1;22)(p13;q13) and t(11;12)(p15;p13). The NPM1 and CEBPA, class II mutations, are observed in 5–10% and 4% of cases, respectively, and they are associated with a better prognosis ^[15][16][15,16]. Although attractive, this model represents an oversimplification of AML pathogenesis; many approaches revealed the presence of novel mutations and absence of tyrosine kinase lesions, particularly in normal-karyotype AML, reducing the relevance of the two-hit model [17].

The WHO classification for hematological malignancies, revised in 2016, integrated genetic characteristics, such as karyotypes and molecular aberrations, with morphology, immunophenotype, and clinical presentation, but has limited application in children, since cytogenetic and genetic abnormalities are uncommon as compared to adult AML [18]. As such, pediatric AML is classified as “not-otherwise-specified” [19].

New discoveries in AML genetic alterations were applied to pediatric patients, improving risk stratification and therapeutic approaches. The TARGET project, an analysis of the molecular aberrations in pediatric AML, showed that the mutation rate is lower in pediatric than in adult AML, with different somatic aberrations including structural changes, aberrant DNA methylation, and specific pediatric mutations [20]. In particular, RAS, KIT, and FLT3 class I mutation types are the most common mutated genes while DNMT3, IDH1, and IDH2 gene mutations are rare [21].

Fusion genes specifically identified in pediatric AML and associated with a grim prognosis include CBFA2T3-GLS2 and NUP98-NSD1.CBFA2T3-GLS2 is a chimeric transcript derived from a cryptic inversion of the telomeric region of chromosome 16 and expressed especially in non-DS FAB M7 AML. This gene mutation is present in patients younger than 5 years old and is characterized by high bone marrow blast count and extra-medullary involvement [20]. In contrast, NUP98-NSD1 can be found in 3.8% cases of pediatric AML and is the most frequent NUP98 rearrangement. It is usually associated with other chromosomal abnormalities, particularly trisomy 8, and other genetic mutations, such as FLT3-ITD, WT1, and CEBPA, and appears to play a role in histone methylation and acetylation [22].

Diagnostic approaches in pediatric AML may be influenced by experience in adults ^[13][7]. Treatment decisions are based on risk stratification and further guided by the initial response to treatment ^[5][6][23][5,6,23]. Most international pediatric study groups (AIEOP/BFM/FRANCE/UK/COG/Japan) define risk classification according to genetic/molecular abnormalities and response to treatment by the measurement of residual disease by flow and morphology [24]. In particular, favorable prognostic factors include t(8;21)(q22;q22)/RUNX1-RUNX1T1, t(15;17)(q22;q21)/PML-RARA, NPM1-mutated AML, and CEBPA double mutation [6].AML with MLL translocations has variable outcomes, depending on the associated translocation and occurs more frequently in children compared to in adults [25]. Chromosome 3q and 5q abnormalities and monosomy karyotype and high blast count at diagnosis are predictors of poor outcome ^[6][26][6,26].

FLT3 mutation is also associated with dismal prognosis in adult AML [27] and remains controversial in pediatrics ^[28][29][30][28,29,30]. However, a large meta-analysis of 10 trials including 1661 patients with pediatric AML showed a shorter OS for FLT3-ITD mutated AML [31]. AIEOP-BFM AML 2020 proposed risk stratification in three groups (standard, intermediate, and high-risk AML); AML is high risk when minimal residual disease (MRD) is ≥1% after induction course 1 or ≥0.1% at induction 2 or blast count is ≥5% at induction 1 (only if flow results are not informative) with one of genetic/molecular aberration at diagnosis, as described in Table 1 [24].

Few changes were observed over the last forty years in the treatment algorithm of both pediatric and adult AML. The standard therapeutic approach is a combined cytarabine and anthracycline-based regimen followed by consolidation with allogeneic stem cell transplantation (allo-SCT) for high-risk AML and allo-SCT for non-high-risk patients only in second CR after relapse ^[32][33][32,33].In the last decade, several drugs such as epigenetic treatments, i.e., hypomethylating agents or histone-deacetylaseinhibitors, anti-CD33–ozogamicin conjugated monoclonal antibodies, and FLT3 and isocitrate dehydrogenase (IDH) inhibitors were used to treat pediatric and adult AML (Table 2).