Acute Myeloid Leukemia is a highly diverse disease containing many cytogenetic and molecular abnormalities. We analyzed the DNA methylation
(DMP) and gene expression profiles (GEP) of 344 AML patients using an unsupervised and supervised approach. We hypothesized to better
characterize the disease phenotype by combing these features as these may result in specific patterns in cancer cells which reflect biological
differences. The unsupervised approach segregates patients into 18 clusters, among them six clusters that are defined by the World Health Organization, such as inv16, t(15;17), t(8;21) and CEBPA double mutants. In addition we identified four novel AML subtypes that could not be explained by the enrichment of any currently known recurrent cytogenetic, molecular, morphological or clinical feature. Two of these clusters are categorized with good stability. One of these cluster could be characterized with pathways that are involved in the accumulation of red blood cells and highly predictable using 21 GEP and 3 DMP features, whereas the other cluster is characterized with T-cell related pathways and highly predictable with 9 GEP and 4 DMP features.