Investigating causality in the association between DNA methylation and prevalent T2D using a bidirectional two-sample Mendelian Randomization

ABSTRACT Aim Several studies have identified associations between type 2 diabetes (T2D) and DNA methylation (DNAm). However, the causal role of these associations remains unclear. This study aims to provide evidence for a causal relationship between DNA methylation and T2D. Methods We implemented a bidirectional two-sample Mendelian randomization (2SMR) to evaluate causality at 58 CpG sites previously detected in a meta-analysis of epigenome-wide association studies (meta-EWAS) of prevalent T2D in Europeans. We retrieved genetic proxies for T2D and DNAm from the largest GWAS available. We also used data from the Avon Longitudinal Study of Parents and Children (ALSPAC, UK) when associations of interest were not available in the larger datasets. We identified 62 independent SNPs as proxies for T2D, and 39 methylation quantitative trait loci or mQTL as proxies for 30 of the 58 T2D-related CpGs. We applied correction for multiple testing using Bonferroni and inferred causality based on a P < 1.0×10 −3 or P < 2.0×10 −3 for the T2D⟶ DNAm direction, and the opposing DNAm ⟶ T2D direction of the 2SMR, respectively. Results We found strong evidence of causality of DNAm at cg25536676 ( DHCR24 ) on T2D, where an increase in transformed residuals of DNAm at this site were associated with 43% (OR=1.43, 95%CI=1.15-1.78, P =0.001) higher risk of T2D. We infer a likely causal direction for the remaining CpG sites assessed. In silico analyses showed that CpGs analyzed were enriched for eQTMs, and for specific traits dependent on the direction of causality predicted by 2SMR. Conclusions We identified one CpG mapping to a gene related with the metabolism of lipids ( DHCR24 ), as a novel causal biomarker for the risk of T2D. CpGs within the same gene-region have previously been associated with T2D-related traits in observational studies (BMI, waist circumference, HDL-cholesterol, insulin) and in MR analyses (LDL-cholesterol). Thus, we hypothesize that our candidate CpG in DHCR24 may be a causal mediator of the association between known modifiable risk factors and T2D. Formal causal mediation analysis should be implemented to further validate this assumption.