Genomic Profiling Identifies Novel Mutations and Fusion Genes in Newly Diagnosed and Relapsed Pediatric FLT3-ITD-Positive AML

Pediatric cancers are distinct from adult cancers in both their genomic alterations and therapeutic responses. Fms-like tyrosine kinase 3 (FLT3) mutations, especially internal tandem duplications (ITD), are among the most common mutations in acute myeloid leukemia (AML). FLT3-ITD mutations occur in approximately 15% of pediatric and 25-30% of adult AML, and are generally associated with poor prognosis. However, a number of studies have suggested that FLT3-ITD-positive(+) AML requires additional cooperative mutations. The objective of this study was to characterize the mutational landscape in a cohort of FLT3-ITD+ pediatric AML patients (median age,12.6 years; range, 2.8-19.2 years) enrolled to the AML02 and AML08 trials using samples obtained at diagnosis (n=34) and paired diagnosis/relapse samples (n=5). Children with promyelocytic leukemia were excluded. Samples were analyzed by RNASeq, a targeted 95 gene next generation sequencing (NGS) panel, and whole exome sequencing (WES).

At diagnosis, 58.8% of the samples contained fusion genes; 41.2% were NUP98-NSD1, 11.8% were novel fusions (NSD1-CAPRIN1, NSD1-RALBP1, RUNX1-BCL11B, ZEB2-BCL11B), and 5.9% were previously reported fusions (CBFB-MYH11, DEK-NUP214). The NGS panel identified that WT1 and NPM1 were routinely mutated at a frequency of 32.4% and 20.6% respectively. While the NPM1 mutation was either a 4bp insertion at amino acid (a.a.) 287 or 288, WT1 mutations were heterogenenous with missense mutations, insertions and deletions all being reported. WT1 mutations and NUP98-NSD1 co-associated in 7 patients, 1 patient also harbored a TYK2 mutation; in the remaining 7 patients with NUP98-NSD1 fusions, a mutation in RAD21 or NRAS was observed in 2 patients. For samples with other fusions (n=6), we detected an average of 1 additional mutation per sample, which included mutations (variant allele frequency; VAF) in DNMT3A (0.44), IDH2 (0.49), KIT (0.37), NPM1 (0.51), PLCG2 (0.44), RAD21 (0.55), and SMC1A (0.47). No fusion genes were observed in 13 patients. In this latter subset, mutations in NPM1 (n=6) and WT1 (n=3) were observed. Other alterations that were identified in these samples included mutations in DNMT3A, IDH2, PLCG2, and PRKCB, which co-occurred with NPM1 mutations. Three patients did not harbor a fusion gene or a gene mutation by our analysis. When looking at cumulative incidence of relapse or resistant disease, our study results are concordant with previous reports where a NUP98-NSD1 fusion associated with worse prognosis (hazard ratio [HR] = 3.2, p = 0.02), but FLT3-ITD allelic ratio >0.4 was not prognostic (HR = 1.1, p=0.87). NPM1 mutations were not significantly associated with better prognosis (HR = 0.2, p = 0.11).

We next sought to identify relapse specific alterations by analysis of paired diagnosis/relapse samples by RNASeq, NGS panel, and WES. Notably, the FLT3-ITD mutation was maintained at relapse in all samples. From the NGS panel, we observed the emergence of a MED12 mutation (P1751Q, VAF 0.37) and WT1 mutation (p.S152*, VAF 0.19) at relapse; a mutational switch in WT1 from diagnosis to relapse was also observed (5bp insertion at a.a. 157 to 2bp insertion at a.a. 158). By RNASeq analysis, we found a novel relapse specific fusion gene, LUZP6-OSBL1A. From exome sequencing, mutations in transcription factors were observed at relapse such as CREBBP, GLI3, and TBX20. Our analysis of relapse specific genes showed recurrent mutations in HUWE1, OGT, NACAD, and UNC13A. Intriguingly, both OGT and HUWE1 have been implicated in cancer metabolic reprogramming, and regulate MYC transcriptional programs. OGT is an O-Linked β-N-acetylglucosamine (O-GlcNAc) transferase involved post-translational modification of serine and threonine residues. HUWE1 is an E3 ubiquitin ligase that has been established as a tumor suppressor and previously reported to be mutated in AML.

In conclusion, we demonstrate that additional genomic alterations are observed in the majority of pediatric FLT3-ITD+ AML samples evaluated, with a high proportion of samples containing fusion genes, WT1 and NPM1 mutations. We also identified novel fusion genes and mutations that have not been previously reported in pediatric FLT3-ITD+ AML, including relapse specific mutations. These results provide further biological insight into the genomic heterogeneity of pediatric FLT3-ITD+ AML, warranting further investigations in larger patient cohorts.