Distinct Genomic Landscape of Chinese Pediatric Acute Myeloid Leukemia

Although comprehensive genomic studies have revealed key genomic aberrations in pediatric acute myeloid leukemia (AML), knowledge about Chinese patients remains lacking. Here we report the genomic landscape of Chinese pediatric AML by analyzing the sequence mutations and fusions from transcriptome sequencing (RNA-seq) of 292 cases diagnosed through 2009 to 2018 in Shanghai Children's Medical Center. Informed consents were obtained from parents for all patients.

A total of 1831 non-synonymous mutations that were predicted somatic and/or associated to pediatric cancer were identified in 972 genes, including 1597 single nucleotide variants (SNV), 210 insertion/deletion (indels) and 24 internal tandem duplications (ITD), with a median of 6 mutations per case (ranging 0 to 15). Among these abnormalities, 7 aberrations occurred in more than 5% of cases in current cohort, including mutations in KIT (n=54, 18.5%), FLT3 (n=46, 15.8%), NRAS (n=28, 9.6%), CEBPA (n=23, 7.9%), ASXL2 (n=20, 6.8%), KRAS (n=16, 5.5%) and CSF3R (n=15, 5.1%). 444 potential driver variations were identified affecting 66 genes by a combined strategy of mutation pathogenicity and hotspot analysis. Each patient carried a median of one driver mutations per case (ranging 0 to 7). In addition, RNA-seq identified 227 fusions involving 99 genes in 203 out of 292 patients (69.5%), and CBL exon8/9 deletion in 12 patients (4.1%). The most prevalent fusions detected in current cohort included RUNX1-RUNX1T1 (n=82, 28.1%), KMT2A rearrangements (n=45, 15.4%) and NUP98 rearrangements (n=17, 5.8%). Furthermore, novel gene rearrangements were identified in current study, including PTPRA-FUS, ZEB2-ATIC, MSI2-UBE3C (n=1 each).

Distinct genomic aberration profile was revealed while comparing our results to the mutation profile characterized in Children's Oncology Group (COG)-National Cancer Institute (NCI) TARGET AML initiative representing the Western pediatric AML cohort. A total of 16 recurrently mutated genes were identified with significantly (two-sided fisher exact test) different mutation frequency. Among these, 7 genes mutated more frequently in Chinese patients, including KIT (18.5% vs 12.8% in Chinese and Western cohort, respectively. p=0.027), ASXL2 (6.8% vs 3.6%, p=0.043), CSF3R (5.1% vs 2.4%, p=0.044), JAK2 (3.4% vs 0.0%, p<0.001), DNM2 (2.7% vs 0.0%, p<0.001), KDM6A (2.1% vs 0.0%, p<0.001) and KMT2C (1.7% vs 0.0%, p=0.003). On the other hand, mutations in FLT3 (15.8% vs 33.0%, p<0.001), NRAS (9.6 vs 30.9%, p<0.001), KRAS (5.5% vs 12.8%, p<0.001), WT1 (2.4% vs 13.6%, p<0.001), NPM1 (2.4% vs 10.3%, p<0.001), PTPN11 (3.8% vs 8.1%, p=0.016), TET2 (1.0% vs 5.2%, p=0.001), CBL sequence mutation (0.0% vs 3.0%, p<0.001) and IKZF1 (0.3% vs 2.7%, p=0.018) were occurred less frequently in Chinese patients. Notably, the RAS signaling pathway as a whole was significantly less frequently mutated in Chinese patients (35.6% vs 71.0%, p<0.001). Furthermore, distinct associations between mutations and FAB subtypes were also observed. For example, NF1 mutations were significantly enriched with subtype M5 in Chinese patients (p=0.003), which was previously reported as co-mutated with CBFB-MYH11 fusion with associated with subtype M4.

Survival analysis revealed key genomic aberrations associated with patient prognosis. Variants significantly (log-rank test) associated with better event free survival rate included mutations in CEBPA (p=0.023), NPM1 (p=0.026) and GATA2 (p=0.016). On the other hand, CBFA2T3-GLIS2 (p=0.028), nucleoporin gene family related fusions (including NUP98, NUP214 and NUP153, p<0.001), FUS related fusions (p=0.030), mutations in RUNX1 (p<0.001) and FLT3 (p=0.003) were associated with worse prognosis. A revised risk stratification model was proposed based on these associations observed.

Characterized a first comprehensive genomic landscape of Chinese pediatric AML, our results reveal a distinct mutation profile as compared to the Western cohort, in terms of both mutation frequency and patterns of mutation co-occurrence. These findings further reveal the complexity of pediatric AML and highlight the importance of tailored risk stratification for Chinese patients in clinical management.