Next generation sequencing profiling of pediatric Acute Myeloid Leukemia reveals distinct genetic landscapes across cytogenetic subgroups

Abstract Next-generation sequencing (NGS) transformed the molecular diagnosis of pediatric acute myeloid leukemia (AML), enabling simultaneous detection of multiple clinically relevant genetic alterations. We performed targeted NGS using the Archer VariantPlex Myeloid panel (75 genes) in 175 pediatric AML samples collected at diagnosis in AIEOP-AML2013 protocol. Pathogenic, likely pathogenic and variants of uncertain significance (Tier I, II or III as per ACMG guidelines) were analyzed. Gene mutations were detected in 159/175 patients (91%), involving 38 different genes. The average number of variants per AML was 2 (range 0–5); the variant number did not correlate with prognosis. The mean number of mutations was significantly lower in patients 10 years (mean 1.75) compared to those ≥10 years (mean 2.12, p=0.03). The most frequently mutated genes were NRAS (50/175, 29%), FLT3 (40/175, 23%, including ITD, TKD, and other), WT1 (31/175, 18%), KRAS (25/175, 14%), KIT (23/175), PTPN11 (17/175, 10%), NF1 (16/175, 9%) and CEBPA (13/175, 7%). By categorizing mutations into functional classes, we observed 122 AML (70%) with at least one mutation in signaling pathway genes, 40 (23%) in epigenetic regulators and 30 (17%) in transcription factor genes, while only 3 cases had mutations in cohesin-complex genes. Mutations in epigenetic regulators were never observed in AML cases with FLT3-ITD, KMT2A-rearrangements or inv(16)CBFB::MYH11. Analysis of mutation distribution across genetic subgroups revealed specific associations. In t(8;21)RUNX1::RUNX1T1 AML (n=39), mutations were detected in KIT (17 cases), CCND2 (7), NRAS (6), KRAS (6), ASXL1 (6), FLT3 (3), WT1 (3), and KDM6A (3). Interestingly ASXL1 and CCND2 mutations were strongly enriched in this subgroup (ASXL1 6/8 and CCND2 7/8 mutated cases in the whole cohort), although they had no impact on outcome. In inv(16)CBFB::MYH11 cases (n=14), we found KIT mutations in 4 cases, NRAS in 10, KRAS in 3 and WT1 in 2. Of note, among KIT-mutated AML (n=23), most cases (21) were associated with core-binding factor (CBF) alterations, underscoring a strong correlation between KIT mutations and CBF AML. Additionally, KIT-mutated AML within the CBF subgroup had worse event-free survival compared to non-mutated cases (40% versus 96%, p=0.0001). Among KMT2A-rearranged AML (n=36), NRAS was mutated in 10 cases, KRAS in 8, NF1 in 8, FLT3 in 6, WT1 in 3, PTPN11 in 3, and RUNX1 in 3. Interestingly, 50% of NF1 mutations (8/16) were found in KMT2A-rearranged AML, with no impact on outcome. In NPM1-mutated cases (n=29), 15 had FLT3 aberrations, including 11 with FLT3-ITD. Other mutations were found in NRAS (9), WT1 (9), PTPN11 (8), IDH1 (5), IDH2 (3), and DNMT3A (3) and KRAS (2). Notably, DNMT3A mutations were exclusively observed in NPM1-mutated AML. Moreover, among IDH1/2-mutated AML cases (n=16), 8 cases (50%) were associated with NPM1 mutations. Similarly, 47% of the PTPN11-mutated cases (8/17) were also associated with NPM1 mutations. None of these variants had a prognostic impact in this cohort. Among FLT3-ITD AML (n=24), co-occurring mutations were observed in NPM1 (11), WT1 (11), RUNX1 (4), NRAS (3) andPTPN11 (3). Importantly, among the 31 patients with WT1 mutations, 11 also harbored FLT3-ITD (35%). Interestingly, among the 4 TP53 mutated AML identified in this cohort, 3 experienced an adverse event. Moreover, 45/175 (26%) AML harbored at least one mutation (range 1-3) in a predisposing gene (CEBPA, RUNX1, GATA2, DDX41, ETV6, IKZF1, TP53, NF1, JAK2, SH2B3, CDKN2A), but these patients did not have a different outcome. Moreover, by analyzing n=35 diagnosis-relapse paired samples, we observed differential clonal dynamics. Some variants were often lost at relapse (NRAS, present at diagnosis in 9 cases but retained in only 1 at relapse), suggesting reduced relevance for disease progression. Conversely, mutations in genes such as PTPN11, IDH1 and TP53 were maintained at relapse (4/4 cases for each gene). Other genes showed a mixed pattern, with comparable probability of being maintained or lost (e.g. FLT3 mutations were retained in 4/6 cases, KIT in 3/8). Notably, the emergence of de novo mutations at relapse (e.g. WT1) highlights possible clone acquisition. In conclusion, NGS profiling in pediatric AML reveals a high prevalence of somatic mutations with subtype-specific patterns. Future collaborative studies will help clarify mutation frequencies, associations, clonal evolution, and prognostic impact.