The epigenetic landscape of paediatric acute myeloid leukaemia
AuthorMeyer, Braydon Ashley
Document TypePhD thesis
Access StatusThis item is embargoed and will be available on 2022-05-08.
© 2019 Braydon Ashley Meyer
Paediatric acute myeloid leukaemia (AML) is a cancer of the blood and bone marrow. It is currently one of the leading causes of cancer-related mortality in children. While induction therapy is largely successful in achieving patient remission, the relatively high mortality rate is driven by the large genetic heterogeneity of AML and recurrence of disease. Disease relapse rate is higher than other childhood leukaemias, is fast acting and often chemotherapy resistant. While much of the genetic contribution to disease has been described, there is still a component of AML pathogenesis that has yet to be discovered. Many of the genetic lesions found in adult AML directly affect epigenetic modifying genes, however this is not the case in children. Despite this, previous research has shown vast epigenetic alteration in paediatric AML. As such it is possible that some of the unexplained pathogenesis in childhood AML can be elucidated by modulation of gene activity via aberrant changes in the most widely studied epigenetic process, DNA methylation (DNAm). Few studies have comprehensively interrogated the DNA methylome of paediatric AML, nor has the prognostic utility or biomarker potential of DNAm been explored. In this study, we explored the global methylation profile of paediatric AML in comparison to non-leukaemic controls and subtype-dependant and independent biomarkers of disease that may have functional relevance. Furthermore, we described DNAm signatures with potential prognostic utility, to accurately identify predisposition to relapse at diagnosis. Genome-wide DNAm was interrogated via the HumanMethylation450 BeadChip Array (HM450K) on a cohort comprising of 128 archival and fresh bone marrow tissue sourced from multiple hospitals around Australia. This data was then combined with the TARGET AML cohort comprising of a further 231 bone marrow samples. Targeted replication and validation of findings was undertaken on a reduced cohort using SEQUENOM MassArray EpiTYPER. Bioinformatic and machine learning analyses were undertaken in R. The findings revealed subtype-independent genome-wide average methylation (GWAM) to be increased in diagnostic samples compared to non-leukaemic controls. This was further verified by differences in the global methylation proxy genes known as LINE1 and Alu. Deeper interrogation of these differences demonstrated wide-spread differential methylation in previously implicated genes in AML pathogenesis including WT1 and DGKG, both of which were validated in an independent cohort. Other genes identified to be differentially methylated included ZSCAN1, REC8 and IRX1. Subtype analysis validated previous studies showing inv(16)-specific differential methylation in MN1 and MEIS1. Finally, DNAm was used as the primary feature for a machine learning model designed to predict patient relapse at diagnosis. The final model achieved an area under the curve (AUC) of 94% with correct identification of 91% of all cases involved (F-measure=0.914). To date, this study represents the largest and most comprehensive insight into aberrant DNAm in paediatric AML. Results have increased our understanding of genes that are differentially methylated and highlight the potential utility of DNAm as a future prognostic biomarker. It is anticipated that these findings will serve as a foundation for future functional studies aimed at delivering truly personalised treatment regimens for children with AML.
KeywordsCancer; Leukaemia; Paediatrics; Epigenetics; Methylation; Machine Learning
- Click on "Export Reference in RIS Format" and choose "open with... Endnote".
- Click on "Export Reference in RIS Format". Login to Refworks, go to References => Import References