Exploring a potential mechanistic role of CpG-specific methylation in the relationship between environmental exposures and childhood acute lymphoblastic leukaemia

The aetiology of childhood acute lymphoblastic leukaemia (ALL) remains unclear. Whilst genetic aberrations have been suggested as initiating events, additional factors likely play a role in leukaemia development. Our previous gene-level analysis suggests DNA methylation may be a mediating mechanism through which some environmental factors may contribute to ALL manifestation. Here we use a meet-in-the-middle approach to perform CpG-level analysis to investigate DNA methylation as a mediating mechanism between risk exposures and ALL.

We used published data to identify differentially methylated CpGs (DMC’s) associated with risk factors related to ALL and for ALL itself, where DMCs conserved across all ALL subtypes were termed "constitutive". DMCs associated with constitutive ALL were integrated with DMCs associated with risk exposures to determine overlapping DMCs. Hypergeometric tests were used to assess the probability of relationships for overlapping methylation change. Where exposures are hypothesised to increase ALL risk (i.e. maternal smoking, radiation and alcohol), we assess the same directionality for exposure-related and ALL methylation changes. Where exposures are hypothesised to be protective (i.e. maternal folate, daycare, reported colds) we assess directionally opposing methylation changes.

CpG analysis reinforced earlier findings with significant overlap between methylation changes observed in maternal plasma folate, radiation and alcohol exposure and ALL itself, and lack of associations for maternal folate supplementation and coffee consumption. Significant overlapping methylation changes, previously seen at gene-level, were no longer observed for sugary caffeinated drinks or smoking related methylation. Previously, significant overlaps were identified for gene-level methylation associated with daycare and ALL, however no DMCs were found in common. Reported colds, previously found to be non-significant at gene-level, observed significant overlapping DMCs between exposure and ALL.

We suggest that CpG-level analysis may be more robust due to increased sensitivity to pinpoint the genomic position of altered methylation, in comparison with our previous gene-level approach.