Development of Acute Megakaryoblastic Leukemia in Down Syndrome Is Associated with Sequential Epigenetic Changes That Target the Down Syndrome Critical Region on Chromosome 21,

Abstract 3451 Down syndrome (DS) is associated with an increased frequency of Acute Megakaryoblastic Leukemias (AMKL). In the context of DS, AMKL is usually preceded by the development of a transient myeloproliferative disorder (TMD). Up to 10% of DS children may develop this disorder, and approximately 20–30% of those with TMD go on to develop AMKL, which may or may not be preceded by a return to normal hematopoiesis. While trisomy 21 (tri21) alone is not sufficient to develop AMKL, it is clear that the acquisition of somatic mutations in GATA1 is a requirement for the development of DS-AMKL. Much has been studied in relation with this genetic mutation, however not much is known about the epigenetic changes accompanying the development of AMKL in DS patients. Given the step-wise nature of the development of DS-AMKL, we hypothesized that distinct and cumulative epigenetic changes would be associated with each of the phases of the disease. In order to address this hypothesis we performed DNA methylation profiling of a cohort of 7 primary DS-AMKL samples using the latest design of the HELP microarray assay, which covers ∼230,000 CpG sites at ∼22,000 RefSeq promoter regions. In order to better understand how aberrant epigenetic patterns become established during the development of this disease we compared these DNA methylation profiles to those of mononuclear cells from normal fetal liver (FL-MNC) (n=4), tri21 FL-MNC (n=4) and TMD blasts (n=6). We identified a set of 711 unique genes that displayed significant loss of methylation in Tri21 FL-MNC with respect to control FL-MNC (FDR < 5% and methylation delta ≥ 25%). However, when tri21 FL-MNC were compared to the GATA1s positive TMD blasts, we observed that TMD samples acquired significant hypermethylation of 346 unique genes at the same significance cutoff. A direct comparison of the genes affected by these two opposing waves of methylation changes demonstrated that different sets of genes were being targeted at each stage. Next we compared the DNA methylation profiles of TMD and DS-AMKL blasts, and found them to be virtually identical, indicating that an epigenetic imprint becomes established at the TMD phase, which is still present at the leukemic phase. Finally we compared DS-AMKL to a cohort of 8 non-DS AMKL, and observed that DS-AMKL samples display significant hypomethylation of 267 unique genes, indicating that these two diseases are not only genetically but also epigenetically distinct. In determining the location of the epigenetic changes acquired at the different stages of the disease we found that, similar to what had been previously reported for gene expression, DNA methylation changes did not localize solely to chr 21, but that these changes were distributed along all chromosomes. However, detailed analysis of chr 21 revealed that changes on this chromosome preferentially targeted a specific region on the long arm, the Down Syndrome Critical Region (DSCR). This region becomes significantly hypomethylated in Tri21 FL-MNC and remains hypomethylated through all the stages of the disease. Several groups have sought to identify candidate genes on the DSCR that may contribute to the pathogenesis of the disease. Two genes, BACH1 and CXADR, have been identified in common by two different studies as being overexpressed in DS-AMKL vs. Non-DS AMKL. Both of these genes displayed aberrant hypomethylation at their promoter regions in Tri21 FL MNC, TMD and DS-AMKL samples: such hypomethylation was absent from all the control samples. Finally, given that GATA1 mutations are a requirement for the development of DS-AMKL, we compared the different aberrant DNA methylation signatures to the normal GATA1 targets identified by ChIP sequencing experiments, and found 20–25% overlap. In summary, we have completed the first comprehensive DNA methylation study of the developmental phases of DS-AMKL and have demonstrated that a) specific epigenetic changes are associated with the different stages of the disease, b) these changes occur in ‘waves' of sequential hypo and hypermethylation, c) aberrant DNA methylation on chr 21 preferentially targets the DSCR and results in aberrant overexpression of the associated genes, and d) epigenetic profiles of TMD and DS-AMKL are virtually identical. These observations lead us to believe that an epigenetic memory established at the TMD stage may be at least in part contributing to the development of full-blown AMKL at a later time point. Disclosures: No relevant conflicts of interest to declare.