Acute Myeloid Leukemia (AML) is a heterogeneous disease from the molecular and biological standpoints. In order to resolve some of this complexity, a recent microarray-based expression profiling study segregated cohorts of patients with common gene signatures. One of these signatures was associated with alterations of the CCAAT/enhancer-binding protein alpha (CEBPA) gene. Among these patients, a subset harbored CEBPA mutations, while the remainder failed to express CEBPA, which in a number of cases correlated with hypermethylation of its promoter. This latter subgroup of leukemias with silenced CEBPA presented with significant biological differences compared to CEBPA mutant patients, including expression of T-cell markers and activating mutations of NOTCH1 (Wouters BJ et. al., PMID:17671232). Since our preliminary data show that DNA methylation profiling is extremely accurate in identifying distinct biological phenotypes in AML and other tumors, we wondered whether genome-wide epigenetic analysis would identify the biological difference between these patients. In order to determine the DNA methylation profiles of these patients we performed HELP (HpaII tiny fragment Enrichment by Ligation-mediated PCR), a quantitative genome-wide DNA methylation method, using a 400,000 feature cutom-desinged microarray, representing 24,000 gene promoters with 50-mer oligonucleotides. We studied the complete previously identified cluster of AML cases presenting with a CEBPA expression signature, and compared and contrasted the DNA methylation profiles of cases carrying the CEBPA mutation and those presenting with CEBPA silencing. Remarkably, unsupervised (unbiased) clustering of DNA methylation profiles revealed that samples were readily segregated into two groups that overlapped perfectly with the presence or absence of the CEBPA mutation, indicating the presence of underlying genome-wide DNA methylation differences between these two groups. We next performed supervised analysis of the samples to compare the DNA methylation profiles of CEBPA mutated vs. CEBPA non-mutated samples. The analysis was performed using a moderated T test, and 291 genes promoters were identified as differentially methylated between the two groups at a significance level of p <0.001. Within this differentially methylated signature, we detected a clear predominance (90%) of hypermethylated genes in the CEBPA silenced group. The critical importance of CEBPA loss of function in mediating the phenotype of these tumors was underlined by the fact that multiple other members of the CEBPA network were likewise hypermethylated. Other than the CEBPA network, the two other most hypermethylated biological pathways involved p38MAPK signaling and PDGF/LCK signaling. Thus, we conclude i) that hypermethylation of the CEBPA promoter is not an isolated event, but rather part of a more widespread epigenetic regulation, and ii) that the original CEBPA signature group is composed of two distinct subgroups originating through two distinct mechanisms, a genetic one and an epigenetic one. The significance of these findings may be supported by the fact that the hypermethylated cohort of patients tended to show a worse prognosis than CEBPA mutant patients. These patients might be good candidates for DNA methyltransferase inhibitors in prospective clinical trials.