Abstract Background: Acute myeloid leukemia (AML) causes the most leukemia-related deaths in the United States, and has frequent epigenetic aberrations, including a CpG island methylator phenotype (CIMP). CIMP defines unique molecular subtypes of other cancers and has been linked to mutations in IDH1/2, however the clinical consequences of CIMP and the role of IDH1/2 mutations in AML remain unclear. Methods: To measure genome-wide CpG methylation we used Digital Restriction Enzyme Analysis of Methylation (DREAM) on AML bone marrow samples and normal peripheral blood controls. For validation we used methylation data from patient samples from The Cancer Genome Atlas (TCGA) on the Illumina Infinium HumanMethylation450 platform. We also used RNA-seq data from TCGA, and microarray data from GEO (GSE6891). Statistical analysis was done using R. Results: Genome-wide analysis of variably methylated CpG sites in 96 AML bone marrow samples using DREAM revealed two distinct CpG island methylator phenotypes by hierarchical clustering: IDH-CIMP (I-CIMP) in which 7/10 cases had oncogenic IDH1/2 mutations, and AML-CIMP (A-CIMP), which lacked any mutations in IDH1/2. At median follow-up of 6.16 years, A-CIMP cases, but not I-CIMP cases were associated with longer overall survival (median OS, years: A-CIMP = Not reached, P = 0.08; I-CIMP = 3.35, P = 0.50; CIMP-negative = 1.17). We validated and extended these findings using TCGA data. In this cohort A-CIMP cases also had significantly longer OS compared to CIMP-negative (median OS, years: A-CIMP = 2.34, P = 0.01; I-CIMP = 1.25, P = 0.89; CIMP-negative = 1.00). Aberrant hypermethylation in A-CIMP occurred preferentially at CpG islands by a factor greater than 3, while I-CIMP cases demonstrated a slight preference for hypermethylation at sites outside CpG islands. Interestingly, A-CIMP was enriched in CEBPA (19%) and WT1 mutations (14%), but inversely correlated with IDH, TET2, and NPM1 mutations. Functional pathway analysis revealed that genes hypermethylated in A-CIMP are associated with pluripotency maintenance - including PAX6, GBX2, and HOXA9 - and RNA-seq data largely, but not entirely, recapitulated methylation-based patterns. There was a strong correlation between promoter CpG island methylation and gene expression for many A-CIMP genes. Finally, the transcriptional program associated with A-CIMP was found to correlate with outcome and genetic backgrounds in both TCGA and additional independent datasets. Conclusions: Taken together, our data suggest that CIMP in AML is complex, multifactorial and cannot be explained solely by coding gene mutations (e.g. IDH1/2, TET2). There is an association between A-CIMP and curability in multiple AML datasets that cannot be recapitulated by mutational data alone and that may be worth validating in prospective studies. Citation Format: Andrew D. Kelly, Heike Kroeger, Jumpei Yamazaki, Rodolphe Taby, Frank Neumann, Sijia Yu, Justin T. Lee, Rong He, Shoudan Liang, Yue Lu, Matteo Cesaroni, Sherry A. Pierce, Steven M. Kornblau, Carlos E. Bueso-Ramos, Farhad Ravandi, Hagop M. Kantarjain, Jaroslav Jelinek, Jean-Pierre J. Issa. A CpG island methylator phenotype in acute myeloid leukemia independent of IDH mutations and associated with a favorable outcome. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2779.
Read full abstract