Acute myeloid leukemia have mainly been characterized by the presence or absence of recurrent cytogenetic aberrations. In order to improve diagnosis and therapy new studies have been performed more recently to supplement this molecular classification with gene expression profiling. It has also been established that expression levels of genes are often largely controlled by the state of cytosine methylation in the adjacent promoter region. A novel technology using a unique combination of base specific cleavage and MALDI TOF mass spectrometry enables quantitative high throughput DNA methylation analysis. We have employed this new DNA methylation analysis technology to analyze a set of over 100 genes recently introduced as a gene expression panel for the prediction of survival in AML patients (Bullinger et al. NEJM 2004). We determined the degree of methylation for selected CpG islands within in 120 promoter regions of these genes in a total of 96 individuals. The resulting data on the degree of methylation at each CpG islands/ sample was analyzed using a supervised and an unsupervised approach. A hierarchical cluster analysis was performed to identify a subset of genes that can be used for further molecular classification of AML. We then also evaluated the methylation data for its ability to predict survival. We first used a set of 48 individuals to train a statistical learning algorithm and a second set of 48 samples to validate the trained algorithm. Furthermore we wanted to evaluate the magnitude of gene silencing caused by promoter methylation. We therefore correlated the expression levels of all examined genes to the degree of methylation in their promoter regions. The results of this work demonstrate the importance of DNA methylation as biomarker in molecular classification of AML. We also show that large-scale DNA methylation studies can now be performed with reasonable efforts in a very short time period. These results provide groundwork for future research that will ultimately enable individualized therapy based on molecular characterization.

[Proc Amer Assoc Cancer Res, Volume 46, 2005]