Epigenomic Analysis of Acute Myeloid Leukemia Identifies Specific Patterns and Markes with Clinical and Biological Relevance.

Genomic aberrations resulting in activation of oncogenes, inactivation of tumor suppressor genes or in the formation of novel chimeric genes are currently considered the main cause of the malignant phenotype of the AML cells. There is now increasing evidence that in addition to genetic aberrations, therapeutically reversible epigenetic events also play a critical role in the pathogenesis of human leukemias.

We used a high-throughout methylation profiling to explore systematically the epigenomic variation underlying the biologic and clinical heterogeneity in AML.

Using the Illumina GoldenGate Methylation Cancer Panel that spans 1,505 CpG loci, a detailed methylation profile of 116 AML patients distributed along all the cytogenetic prognostic subgroups was established. In addition, controls (BM and CB) and human progenitor cells expressing AML1/ETO, CBFβ/MYH11 or MLL/AF9 fusion proteins were analysed. Unsupervised and supervised hierarchical cluster were performed, and a selection of the most significantly differentially methylated loci (Δβ of at least 0.34 and FDR <0.05) calculated as ψβ=(sample mean bvalue)–(control mean bvalue) was done. Candidate genes were validated by MSP in an independent cohort of 244 AML cases. Bisulphite sequencing and quantitative RT-PCR were carried in selected cases.

AML samples were correctly separated from BM controls and segregated in two main categories. While one of them showed a rather plane profile (Group I) similar to the one observed in the control bone marrow samples (only 7 probes showed a mean Δβ>0.34), the other (Group II) presented dramatic changes with an aberrant methylation signature (24 probes showed a mean Δβ>0.34). These two methylation categories showed significant differences in their distribution accross the prognosis cytogenetic groups. Eighty percent of the cases included on the adverse cytogenetic prognostic group clustered in one arm (Group I) and 80% of the cases included on the good prognosis cytogenetic group clustered in the other one (Group II). The normal karyotype (NK) cases were evenly distributed among the Groups I and II. No significant differences were observed for other variables as FLT3 mutational status. Kaplan-Meier analysis did not identified significant differences on the overall survival between the AML Group I and II.

Focussing on the NK cases, two CpGs (from DBC1 and CDKN2B genes) were identified as statistically significant predictors of 5-year overall survival (OS). On the independent AML series, only the promoter methylation status of DBC1 retained statistical significance as predictor of the EFS and OS on NK cases. Expression studies showed a significant silencing of DBC1 on the aberrantly methylated samples.

Taking advantage of our model of hematopoietic stem cells, stably transfected with the fusion proteins^1-3 we observed that the epigenetic signature of the MLL leukemias is also observed on human progenitor cells fully transformed “in vitro” by this single oncogenic event. However, HSC expressing the AML1/ETO and CBFβ/MYH11 fusion proteins, which did not showed a full transformed phenotype, did not recapitulate the methylation signature observed on the AML primary cases.

These results allowed us to conclude that: (1) The aberrant methylation of DBC1 gene among the NK AML cases is associated with a poor outcome. Therefore, the identification of patients with DNA epigenetic aberrations that have predictive value in determining survival should be essential in the forthcoming designs of clinical trials with demethylating agents. (2) A larger number of epigenetically modified genes is observed in the presence of single genetic abnormalities as t(8;21), t(15;17) or MLL rearrangements. However, a full leukemic transformation is require for the acquisition of a specific aberrant methylation profile, suggesting than the presence of fusion proteins as AML1/ETO or CBF/MYH11 is not sufficient to acquire a full aberrant methylation signature.

1. Wunderlich M, et al .et al Blood 2006;108:1690-7.

2. Wei J, et al . Cancer Cell

3. Mulloy JC et al Blood 2002;99:15-23.

No relevant conflicts of interest to declare.