Genome-Wide Profiling of DNA Copy Number Abnormalities and Loss-Of-Heterozygosity in Pediatric Acute Myeloid Leukemia Using High-Resolution Single Nucleotide Polymorphism Microarrays.

To identify a comprehensive registry of oncogenic lesions in pediatric acute myeloblastic leukemia (AML), we used Affymetrix single nucleotide polymorphism (SNP) arrays to examine changes in DNA copy number and loss-of-heterozygosity (LOH) in leukemic blasts from 112 cases of pediatric AML and corresponding remission samples from 63 of these cases. The analyzed cases included t(8;21)[AML1-ETO] (n=20), inv(16)[CBFβ-MYH11](n=16), t(15;17)[PML-RARα] (n=7), MLL rearranged (n=17), FAB M7 (n=9) and normal cytogenetics or miscellaneous cytogenetic abnormalities (n=43). Four SNP arrays (50K Hind and Xba, 250K Sty and Nsp) were used to interrogate over 615,000 markers at a mean inter-marker distance of 4.8 kb. Combined data were analyzed using dChipSNP and a modified array normalization algorithm using only those SNPs from regions known to be diploid by routine karyotyping. These analyses not only detected known whole or partial chromosomal losses or gains, but also detected numerous copy number abnormalities that were not evident by conventional cytogenetics. Somatic DNA copy number abnormalities were identified in 102 (91.1%) cases. The mean number of lesions per patient was 3.2 (range 1–12), with a mean of 2.13 deletions/whole chromosome losses and 1.02 amplifications/whole chromosome gains per patient. Deletions were detected in the leukemic blasts from over 90% of patients, whereas amplifications were only seen in the leukemic blasts from 54% of patients. The vast majority of deletions were focal (<20 Mb) with less than 20% of cases containing larger deletions or losses of whole chromosomes or chromosomal arms. No differences in the frequency of deletions were observed among the different genetic sybtypes of AML. Lesions identified in 2 or more cases included deletions of CDKN2A/B (9p21.3, n=4) and FOXE1 (9p22, n=3), and amplifications of ETS1 (11q24.3, n=3) and MYST4 (10q22.2, n=2). For each of the listed genes, at least one case was identified harboring a focal deletion or amplification confined to the specific gene, thus definitively identifying the gene as the target of the alteration. In addition to these recurrent lesions, copy number changes were identified in regions containing 5 or fewer genes in single cases, including deletions involving the tumor suppressor candidate TUSC3 (8p22), alpha 3 catenin (10q21.3), and amplifications involving FGFR activating protein 1 and RAS homolog G (11p15.4). Importantly, within our cohort of de novo AMLs no focal sub-microscopic lesions involving 5q or 7q were identified. Similarly, copy-neutral LOH (uniparental disomy) that was not adjacent to an identified region of deletion or amplification was uncommon. Taken together, these data demonstrate a surprisingly low frequency of copy number changes in pediatric AML with focal deletions measured at an average resolution of 5-10 KB across the genome predominating over focal amplifications. Moreover, the identified lesions appear to target a rather large number of different genes. Correlating the identified copy number changes with mutations in other genes known to be involved in leukemogenesis including, NRAS, KRAS, FLT3, NPM1, CEBPA, BRAF, PTPN11, AML1 and KIT, should provide valuable insights in the molecular pathology of AML.