To investigate the pathogenesis of acute megakaryoblastic leukemia, AML M7 (M7), we analyzed the gene expression profiles of 113 patient samples on Affymetrix U133A GeneChips. Classification by unsupervised clustering discriminates 70 M7 samples from 12 with other AML FAB subtypes, 3 normal controls, 10 normal and remission samples of patients with Down Syndrome (DS) and 18 ALL samples from DS patients.
Further, M7 subclasses can be identified. DS samples (21 DS-M7 and 9 transient myeloproliferative disease (TMD) samples) cluster apart from non-DS M7 samples, with a few exceptions, most notably the 4 M7 samples with the translocation t(1;22). Smaller subgroups can be detected by consensus clustering in DS and non-DS M7.
The DS-non-DS distinction is not driven by differential expression of chromosome 21 genes, in particular RUNX1 expression levels are not increased in DS. Consistent with differences observed by flow cytometry, DS samples show a higher expression of ANK1, GYPA, GYPB and CD36, while non-DS M7 samples have higher VWF and CD34 expression, reflecting known morphologic differences. The hematopoietic transcription factor GATA1 is overexpressed in DS M7/TMD when compared to non-DS M7s or controls, and is among the most significant markers of the DS class. The mutation of GATA1 that is characteristic for DS is likely to affect its transcriptome specifically. By nearest neighbour analysis, about 300 genes follow the pattern of the GATA1 gene expression within a significant range by permutation testing. This GATA-1 signature is highly enriched in the DS samples and includes several known target genes of GATA1, such as GATA2, GYPA, ANK1 and ALAD. This approach should lead to the identification of genes contributing to cellular proliferation and differentiation in the context of the GATA1 mutation of DS M7.