Abstract
DNA methylation as a source for epigenetic variability has been implicated in a variety of different cancer types. Often these studies are confounded by inter-individual differences in the epigenetic profiles. The pattern of epigenetic marks can be altered by factors like age, nutrition, behavior or other environmental factors, which are difficult to control. We had the unique opportunity to study DNA methylation profiles in a pair of monozygotic twin boys who developed ETV6-RUNX1 B-progenitor acute lymphoblastic leukemia at 2 years of age within 3 weeks of each other. ETV6-RUNX1 ALL is characterized by a high frequency of recurring genetic alterations, but the full complement of genomic and epigenetic alterations contributing to leukemogenesis is unknown. For these twin cases, environmental influences upon epigenetic variation are largely eliminated. We used a mass spectrometry-based quantitative DNA methylation analysis technique (Sequenom’s® EpiTYPER™ application) to investigate 597 amplicons covering the promoter regions of 190 genes. The genomic target regions were selected to be enriched for genes involved in transcriptional regulation (n=130) and/or genes known to be targeted by recurring DNA copy number alterations in childhood leukemia (n= 60). Methylation analysis were performed on DNA extracted from cryopreserved, Ficoll enriched bone leukemic blasts obtained from diagnostic bone marrow aspirates, and non-leukemic peripheral blood leukocytes obtained at remission. We also examined DNA copy number alterations (CNAs) and loss-of- heterozygosity (LOH) using Affymetrix single nucleotide polymorphism (SNP) 6.0 arrays, which examine over 1.8 million loci, in both tumor and normal tissue for both twins. Analysis of SNP array data identified different somatic CNAs in the tumor samples of the two twins involving 9p21.3 (the CDKN2A/B tumor suppressor locus), 12p13.2 (ETV6) and trisomy 21, indicating that the shared ETV6-RUNX1 positive pre-leukemic clone acquired different secondary genetic alterations during leukemogenesis in each twin. Despite these genetic differences, the methylation profiles of the tumor samples were remarkably similar. Unsupervised two-dimensional clustering of quantitative methylation data revealed that the tumor samples clustered separately from the control samples. Based on these findings we calculated the methylation differences in each genomic target region. A total of 51 genomic regions were significantly differentially methylated between tumor and control samples (paired t-test P<0.001, and an average methylation difference > 10%). Within the differentially methylated genomic regions, a subset of approximately 20 exhibited strong regional differences, indicating that DNA methylation changes can be limited to certain areas of the promoter. In the group of genes known to be involved in transcriptional regulation, 32% were differentially methylated, including the HOXA, HOXB, HOXC and HOXD regions, while in the remaining genes only 15% were differentially methylated. This enrichment is significant on the level of 0.05 (Fisher’s exact test, odds ratio: 2.7). This represents the first study comparing genomic and epigenetic alterations in B-precursor ALL involving monozygotic twins. Notably, different DNA copy number alterations are acquired in each twin during leukemogeneis. In contrast, the tumor samples exhibit similar methylation patterns that are strikingly different to control samples obtained from the same individuals. These results indicate that combined genomic and epigenetic analyses will be important to characterize the full repertoire of genomic alterations in acute lymphoblastic leukemia.
Multivariate Boosting for Integrative Analysis of High-Dimensional Cancer Genomic Data
