Exploring the potential role of environmentally-associated DNA methylation to contribute to risk of different subtypes of childhood acute lymphoblastic leukaemia

Background: Acute lymphoblastic leukaemia (ALL) is a heterogenous disease with unclear aetiology. Various genetic aberrations have been retrospectively detected at birth and are suggested to be initiating events in disease development. ALL can be categorised into cytogenetic subtypes based on the alterations they possess. However, further factors, such as epigenetic modifications, are required for disease progression. Previously we provided evidence to suggest that DNA methylation may be a mediating mechanism through which some environmental factors may contribute to ALL manifestation. Due to the differing genetic profiles of the subtypes, it is plausible that different exposures may pose differing levels of risk for each subtype. Here we have employed our previously established, meet-in-the-middle approach, to perform CpG-based analysis to investigate DNA methylation as a mediating mechanism between potential risk exposures and specific ALL-subtypes.

Materials and methods: Differentially methylated CpGs (DMC’s) associated with ALL environmental risk factors were identified using previously published data. We then selected data measuring DNA methylation in ALL, where analysis of 10 cytogenetic subtypes were considered. DMCs associated with cytogenetic subtypes were integrated with DMCs associated with risk exposures. Hypergeometric tests were used to assess the probability of relationships between exposure-associated and ALL-associated methylation for any overlapping methylation change, and when considering the directionality of methylation. Where exposures are hypothesised to increase risk (i.e. maternal smoking, radiation and alcohol), we hypothesise observing the same directionality for exposure related methylation and methylation in ALL. However, where exposures are hypothesised to be protective (i.e. maternal folate status, daycare attendance, reported cold) we assess opposing methylation changes.

Results: Significant overlapping methylation differed between cytogenetic subtypes for different exposures. When DMCs for exposures associated with increased risk were analysed by subtype significant CpG overlaps in the same direction are observed for radiation and MLL, for alcohol intake with T-ALL and ETV6-RUNX1 and for smoking throughout pregnancy with MLL and undefined subtypes. For protective exposures analysed by subtype, significant overlaps in opposing directions are observed for methylation associated with reported colds and the MLL subtype. For maternal plasma folate, overlapping methylation changes in opposing directions are observed for 8/10 subtypes.

Conclusion: Environmental exposures may influence risk of ALL in a subtype-specific manner via exposure-associate methylation. Whilst the potential influence of maternal folate exposure during pregnancy on methylation patterns that may contribute to ALL appears to be fairly consistent across most subtypes, the epigenetic effect of other exposures may be more likely to contribute to the development of specific subtypes. This analysis is therefore useful in understanding which risk factors may contribute to specific subtypes of leukaemia and those which more generally influence ALL risk. Such knowledge may be useful to influence public health policy to aid and tailor prevention strategies.