Does the epigenetic clock GrimAge predict mortality independent of genetic influences: an 18 year follow-up study in older female twin pairs.

Background: Epigenetic clocks are based on DNA methylation (DNAm). It has been suggested that these clocks are useable markers of biological aging and premature mortality. Because genetic factors explain variations in both epigenetic aging and mortality, this association could also be explained by shared genetic factors. We investigated the influence of genetic and lifestyle factors (smoking, alcohol consumption, physical activity, chronic diseases, body mass index) and education on the association of accelerated epigenetic aging with mortality using a longitudinal twin design. Utilizing a publicly available online tool, we calculated the epigenetic age using two epigenetic clocks, Horvath DNAmAge and DNAm GrimAge, in 413 Finnish twin sisters, aged 63–76 years, at the beginning of the 18-year mortality follow-up. Epigenetic age acceleration was calculated as the residuals from a linear regression model of epigenetic age estimated on chronological age (AA_Horvath, AA_GrimAge, respectively). Cox proportional hazard models were conducted for individuals and twin pairs. Results: The results of the individual-based analyses showed an increased mortality hazard ratio (HR) of 1.31 (CI₉₅: 1.13–1.53) per one standard deviation (SD) increase in AA_GrimAge. The results indicated no significant associations of AA_Horvath with mortality. Pairwise mortality analyses showed an HR of 1.50 (CI₉₅: 1.02–2.20) per 1 SD increase in AA_GrimAge. However, after adjusting for smoking, the HR attenuated substantially and was statistically non-significant (1.29; CI₉₅: 0.84–1.99). Similarly, in multivariable adjusted models the HR (1.42–1.49) was non-significant. In AA_Horvath, the non-significant HRs were lower among monozygotic pairs in comparison to dizygotic pairs, while in AA_GrimAge there were no systematic differences by zygosity. Further, the pairwise analysis in quartiles showed that the increased within pair difference in AA_GrimAge was associated with a higher all-cause mortality risk. Conclusions: In conclusion, the findings suggest that DNAm GrimAge is a strong predictor of mortality independent of genetic influences. Smoking, which is known to alter DNAm levels and is built into the DNAm GrimAge algorithm, attenuated the association between epigenetic aging and mortality risk. [ABSTRACT FROM AUTHOR]