High Frequency of Copy Number Variations in Myeloid Leukemia of Down Syndrome.

Abstract 3242

Poster Board III-179

Pediatric acute myeloid leukemia (AML) is a heterogenous disease, with current therapy resulting in cure rates of approximately 60%. The prognosis depends on cytogenetics and early response to therapy. Recently, Radtke et al. (PNAS 2009) reported that (by using single-nucleotide-polymorphism microarrays and candidate resequencing) de novo AML is characterized by a low frequency of copy number variations (CNVs), with a mean of 2.38 somatic CNVs per case. This is in contrast to pediatric acute lymphoblastic leukemia (ALL), with an average of 6 per case (Mullighan et al, Leukemia 2009). The only exception in AML was acute megakaryocytic leukaemia (AMKL, n=9 cases), which is characterized by a high number of CNVs (average of 9.33). Of these 9 AMKL cases, only one patient had a GATA1-mutation, typical of Myeloid Leukemia in Down syndrome (ML-DS). Most ML-DS cases are classified as AMKL, and can be preceded by a pre-leukemic clone in newborns (transient leukaemia- TL), which in approximately 80% resolves spontaneously. Treatment outcome for non Down syndrome AMKL is poor, whereas Down AMKL has a good prognosis.

We performed array-based comparative genomic hybridization (Array-CGH) to detect copy number variations (CNV) in Down syndrome related leukemias, with the aim to study whether progression from TL to ML-DS is associated with clonal evolution. We also compared the data obtained with array-CGH from DS patients with two other subgroups of non-DS pediatric AML, i.e. MLL-rearranged AML, and cytogenetically normal AML (CN-AML). This may provide insight in differences in genomic stability between these subgroups. For this analysis, 66 MLL-rearranged AML, 41 CN-AML, 7 TL and 14 ML-DS samples taken at initial diagnosis were available. TL and ML-DS samples were unpaired. Genomic aberrations were scored as a minimum of three (44K arrays, Agilent) and as a minimum of five (105K arrays, Agilent) adjacent probes deviating beyond the threshold of ∼0.5. Results are given in table 1 and 2 below.

ML-DS patients had a significantly higher frequency of samples with CNVs compared to TL-patients (100% vs. 57%; p=0.026). Most amplifications were (sub)chromosomal (e.g. extra copies of chromosome 8, 11 and 21), whereas the deletions were mostly focal. Except for the (sub)chromosomal amplifications, the other aberrations were mostly found in single cases. Overall, deletions were more common than amplifications. When we analyzed the number of CNVs per case, ML-DS patients had a trend to have a higher number of CNVs per patient (p=0.06) than TL-patients. When the TL and ML-DS patients were taken together and compared to the other non-DS AML-subgroups taken together, there was a significantly higher number of patients with copy variations (86% vs. 58%; p=0.013) in the Down leukemia group. Also, the number of CNV per case was significantly higher in the Down syndrome patients compared to the non-DS AML subgroups (p=< 0.001).

In conclusion, we observed significantly less patients with CNVs and less CNVs per patient in MLL- rearranged and CN-AML compared to Down syndrome related leukemia. This suggests that the Down syndrome related leukemias are more genetically unstable when compared with their non-DS controls. Our data are consistent with the findings in AMKL in non-Down syndrome patients as reported by Radtke et al, and may suggest that genomic instability is a general phenomenon of AMKL. This is noteworthy because ML-DS and non-DS AMKL differ in various other aspects. The mechanisms behind the genetically more instable character of the AMKL subgroup is still unclear, and requires further study. Within the Down syndrome patients, there was a statistically significant increase in patients with CNVs in ML-DS compared to TL, and a trend for more CNVs per patient, suggestive of clonal evolution.