Abstract LB-073: Epigenome evolution in relapsed acute myeloid leukemia

Abstract

Abstract Acute myeloid leukemia (AML) is a predominantly fatal hematopoietic malignancy with high inter-patient and intra-patient genetic and epigenetic heterogeneity. The prognosis of relapsed AML remains dismal, yet the epigenetic basis of relapse is still unclear. Here we investigated whether and how the epigenome evolution impacts AML progression with biological and clinical relevance. Methods: We obtained clinical annotation and AML specimens from 138 patients with paired diagnosis and relapsed samples. We used normal bone marrow (NBM) as epigenetic/transcriptomic controls and patients’ matched germline DNA as genetic controls. We then performed DNA methylation sequencing (ERRBS), RNA-seq, and Exome-seq. For one patient with 5 serial time points, we performed whole genome sequencing (WGS), ERRBS, and single cell RNA-seq. We measured the epigenetic allele burden using a compositional entropy-based approach (Methclone) and methylation heterogeneity using epipolymorphism. Results: We found that diagnosis stage epigenetic allele burden (ΔS < -90) was linked to an inferior clinical outcome (p = 0.0064, log-rank test of relapse-free survival). The higher significance in promoter regions implies the functional impact of epigenetic dynamics. Promoter epiallele shift was associated with more differential expression events (p = 3.8 × 10−6, Wilcoxon signed-rank test) and promoter epiallele diversity is significantly associated with single cell resolution transcriptional heterogeneity (p < 2.2 × 10−16, ANOVA test). The global methylation heterogeneity is decreased from diagnosis to relapse, indicating a selective impact of chemotherapy on epigenetic variability (p = 0.0056, paired Wilcoxon test). We investigated epigenetic allele burden progression from diagnosis to relapse by classifying patients into three clusters using K-means clustering: those with 1) decreased, 2) stable, or 3) increased abundance of epiallele burden. No association was seen between epigenetic clusters and patterns of genetic evolution, and the genetic abundance is higher in Cluster 3 than Cluster 1 (p = 0.048, Wilcoxon test), indicating divergent paths of genetic and epigenetic evolution. We next examined differential expression in the epigenetic cluster samples at diagnosis compared to NBM. Cluster 1 specific genes were enriched for cell cycle processes, while Cluster 3 genes were enriched for immune responses (p < 0.001, gene ontology hypergeometric tests). Integrating WGS and ERRBS data showed that epiallele burden is more dynamic than somatic mutations; a significant increase in epiallele burden preceded a major increase of somatic mutational abundance. Summary: Our results indicate that epigenetic dynamics may provide leukemia cells greater evolutionary fitness via transcriptional adaptation and is associated with clinical outcome. This provides an alternative mechanism of AML resilience during progression and a potential predictor of relapse. Citation Format: Sheng Li, Francine E. Garrett-Bakelman, Stephen S. Chung, Todd Hricik, Franck Rapaport, Jay Patel, Richard Dillon, Priyanka Vijay, Anna L. Brown, Alexander E. Perl, Joy Connon, Mathijs A. Sanders, Peter J.M. Valk, Lars Bullinger, Selina Luger, Michael W. Becker, Ian D. Lewis, Luen Bik To, Richard J. D’Andrea, David Grimwade, Ruud Delwel, Bob Löwenberg, Hartmut Döhner, Konstanze Döhner, Monica L. Guzman, Duane C. Hassane, Gail J. Roboz, Martin Carroll, Christopher Y. Park, Donna S. Neuberg, Ross L. Levine, Ari M. Melnick, Christopher E. Mason. Epigenome evolution in relapsed acute myeloid leukemia. [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 LB-073.