gms | German Medical Science

Introduction: DNA methylation (DNAm) is one of several epigenetic mechanisms to control gene expression. It is well known that environmental factors can have an impact on DNA methylation. In the past years, several so-called “epigenetic clocks” have been developed to predict biological age of humans. Identifying “age accelerators” (i.e. individuals with a higher predicted biological age compared to chronological age) may help develop prevention strategies for persons at risk for disease or premature death. The aim of this study was to analyze the impact of smoking on DNAm and biological age estimates.

Methods: We generated genome-wide DNAm data via the Human MethylationEPIC array (Illumina, Inc) from whole blood samples of ~1,100 healthy probands (age at baseline >60 years) of the Berlin Aging Study II (BASE-II) [1] and of ~600 individuals (>50 years, ~300 healthy, ~300 with depression at baseline) of the BiDirect Study [2]. DNAm data were available at two and three time points, respectively. Using these data, we performed epigenome-wide association analyses (EWAS) on smoking status, pack years of smoking and years of smoking using linear models. In this context, DNAm-based “polyepigenetic scores” (PES) were calculated to assess the variance explained by earlier EWAS results [3]. Biological (epigenetic) age was calculated utilizing the online DNA Methylation Age Calculator [4] and according to other published methods (e.g. [5]). Epigenetic age acceleration (EAA) was estimated based on residuals of the regression analysis of DNAm age on chronological age.

Results: Our EWAS on current vs. never smokers in the BASE-II baseline data yielded 35 epigenome-wide significant (α=9E-08) DNAm signals, which encompassed many previously described smoking-related loci (e.g., AHRR, RARA, and F2RL3). PES analyses suggest that CpGs from the Joehanes et al. EWAS on current vs. never smokers [3] explain 21.5 to 28.5% of variance (p<8.0E-18) in the equivalent BASE-II smoking phenotype. Furthermore, several - but not all - of the calculated EAA estimates showed evidence for association (p-values ranging from 1.0E-11 to 0.73) supporting prior evidence of the (detrimental) impact of smoking on biological age.

Discussion: We validated several previously described DNAm association signals using EWAS by analyzing smoking data in the BASE-II dataset and quantified the association of biological clocks with smoking behavior. Results are currently being assessed in the independent BiDirect dataset.

Conclusion: At the conference, we will provide a detailed summary of the EWAS results as well as poly-epigenetic score and epigenetic clock analyses related to smoking.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.