Genomic Landscape of Pediatric Non-Down's Syndrome Acute Megakaryoblastic Leukemia in China

Acute megakaryoblastic leukemia (AMKL) accounts for 4% to 15% of newly diagnosed pediatric acute myeloid leukemia (AML), and different from Down's syndrome AMKL (DS-AMKL) and non-AMKL, clinical outcomes of non-DS-AMKL are characterized by an inferior prognosis. Although comprehensive genomic studies have revealed key genomic aberrations of non-DS-AMKL in Western cohort, there is still no reports about Chinese patients. Here we report a comprehensive transcriptomic study on 100 Chinese pediatric non-DS-AMKL cases through 2009 to 2022 in Shanghai Children's Medical Center.

A total of 133 rearrangements were detected involving 78 patients, including previously reported recurrent fusion events which defining subgroups of AMKL, namely CBFA2T3-GLIS2 (n=20, 20%), NUP98-KDM5A (n=16,16%), rearrangements of HOX cluster genes (HOXr, n=11, 11%), RBM15-MKL1 (n=5, 5%) and MLLr (n=3, 3%). Two new recurrent gene translocations, XPO1-TRNC18 and STAG2 rearrangements (STAG2r) were detected in 5 (5%) and 11 (11%) cases, respectively. Further unsupervised clustering analysis was done with the most variant coding genes defined by standard deviation, the results showed that cases carrying STAG2r and XPO1-TNRC18 fusion exhibited distinguishable gene expression pattern from other subgroups and were clustered together respectively. It suggested these two rearrangements as potential driver genomic alterations in AMKL. By introducing XPO1-TNRC18 fusion and three STAG2 aberrations (STAG2 knock-out, STAG2-GNA12 and STAG2-OCRL) into 32D cell line, we proved these gene alterations promoted cell proliferation and colonies formation, suggesting potential function in tumor transformation. Based on these evidences, we defined XPO1-TNRC18 and STAG2r as two new subgroups of AMKL.

We identified 222 nonsynonymous sequency mutations affecting 137 genes totally, including 202 single-nucleotide variants and 20 insertion/deletion, with a median of two mutations for per case (range 0-6). Among driver abnormalities, 7 aberrations occurred in more than 3% of cases in current cohort, including JAK2 (n=15, 15%), MPL (n=13, 13%), PTPN11 (n=7, 7%), NRAS (n=5, 5%), GATA1 (n=4, 4%), CTCF (n=4, 4%) and JAK3 (n=3, 3%). All the mutations could be clustered into 8 pathways, and the most frequently mutated pathways were JAK signaling (30%), transcription regulation (29%) and RAS signaling (14%). Mutation frequency varied across different subgroups. For example, JAK/STAT lesions were enriched in STAG2r subgroup (p<0.05).

Distinct genomic aberration profile was revealed while comparing results of this non-DS-AMKL cohort to the mutation profile of non-AMKL cohort which we have published before. CBFA2T3-GLIS2, NUP98-KDM5A, STAG2r, XPO1-TNRC18 and RBM15-MKL1 subgroups were merely detected in AMKL, while driver aberrations like FLT3-ITD, CBFB-MYH1 and so on were not detected in AMKL. For driver mutations, JAK2 (15% vs 3.7%, p<0.0001), MPL (13% vs 0%, p<0.0001 ) and CTCF (4% vs 0%, p=0.007) were detected more frequently in non-DS-AML, while NRAS (5% vs 18.8%, p<0.001), KRAS (2% vs 10.6%, p=0.008), KIT (1% vs 21.6%, p<0.001), CSF3R (1% vs 6.1%, p=0.047), FLT3 (0% vs 22.5%, p<0.001), CEBPA (0% vs 9.8%, p<0.001), AXSL2 (0% vs 8.6%, p<0.001), GATA2 (0% vs 6.1%, p=0.007) and DHX15 (0% vs 4.5%, p=0.038) occurred less frequently in non-DS-AMKL cohort.

Survival analysis revealed association between key aberrations and clinical outcomes. Carrying CBFA2T3-GLIS2 (p=0.018), NUP98-KDM5A (p<0.001) and FUS-ERG (p=0.018) was associated with unfavorable prognosis. Driver mutations in JAK2 (p=0.03) and RAS pathway (p=0.01) were associated with shorter overall survival. Among the two new subgroups, the new XPO1-TNRC18 subgroup demonstrated similar clinical outcomes to MLLr cases, with 5-year EFS and OS of 50%, and STAG2r was characterized by a favorable clinical outcome, with 5-year EFS and OS over 75%.

As the first comprehensive genomic landscape of Chinese pediatric non-DS-AMKL, our results defined two new subgroups with distinct expression patterns and prognosis characteristics. We further revealed a distinct mutation profile between AMKL and non-AMKL and shed light on association between driver aberrations and clinical outcomes. These findings revealed non-DS-AMKL as a heterogenous disease, and suggested the importance of exploration of risk-stratification and targeted therapy in different mutations driven cases.