S4.4 DNA methylation as a biomarker of aging in longitudinal, population based cohorts
Chair: Sara Hägg
Epigenetic mechanisms are characterized by chemical modifications to DNA altering the expression of genes, and have been studied using genome-wide DNA methylation arrays in population based cohorts. Several thousands of methylated sites across the genome have been found to associate with the aging process. Changes in methylation have also been reported for age-related traits and in response to environmental factors such as smoking. Moreover, methylation at specific sites have been used for accurate predictions of an individual’s biological age “the epigenetic clock”. The objective of our symposium is to present an overview of the link between physical and biological aging using data from longitudinal, population based cohort studies.
S4.4.1 Investigations of the epigenetic drift, clock and mutations in a longitudinal study
Karolinska Institutet, Stockholm, Sweden
Background: Epigenetic changes associated with the aging process can be described by the epigenetic drift, the epigenetic clock and epigenetic mutations, but it is unclear how these processes interact. Methods: DNA methylation associations with age were conducted in 375 individuals (994 samples) from the Swedish Adoption/Twin Study of Aging collected up to five times (1992-2012) using leukocyte DNA methylation data (Illumina 450k). The epigenetic clock and burden of epigenetic mutations – abnormal methylation levels defined as outliers exceeding three times the interquartile range from the 25th/75th percentiles – were defined in each sample. Results: Mean age of participants was 69 years at baseline. 1316 age-associated methylation sites (p<1.3e-7) were identified.The epigenetic clock and epimutation burden correlated with age (r=0.67 and r=0.76 respectively, p<1e-4 for both). Only a few methylation sites overlapped between the clock CpGs and the age-associated sites. 1185 frequently appearing (> 50 samples) epimutation sites were identified and the epimutation burden increased exponentially with age (p=1.7e-13). Of the frequently occurring epimutation sites, two were also age-associated. Conclusions: Age-related epigenetic processes can be described in many ways and they represent different functions to the most parts.
S4.4.2 Do different biological clocks tell the same time? Comparing telomere length,
brain-age, and epigenetic age in the Lothian Birth Cohort 1936
University of Edinburgh, UK
Background: Different individuals age biologically at different rates. Ageing clocks have been derived using multiple biological measures, including: telomere length, DNA methylation, and sturctural brain imaging. Deeply phenotyped longitudinal cohorts of ageing allow us to track changes in these clocks over time, determine the correlations between them, and investigate how well they predict age-related outcomes, such as longevity. Methods: Telomere length, (MRI-based) brain age, and the epigenetic clock were assessed longitudinally at up to 4 time points (ages 70, 73, 76, and 79 years) in the Lothian Birth Cohort 1936 (n=1,091). Linear regression and linear mixed effects models were used to determine the strength of association between the measures and the rate of change over time. Cox regression models were used to model the associations of the biological clocks with longevity. Results: Small correlations were found between the three biological clocks – maximum Pearson r=0.05, all P>0.05. Brain age and the epigenetic clock were accuarate predictors of chronological age, and showed similar sized correlations with fitness measures such as walking speed, lung function, and cognitive ability (greater biological age was associated with lower fitness). All three clocks predicted mortality independently of each other. Conclusions: Different biological clocks are not correlated with each other in a large cohort of ageing yet all independently predict lifespan.
S4.4.3 Leisure-time physical activity shows no effect on dna methylation age in
discordant twin study
University of Jyväskylä, Jyväskylä, Finland
Background: DNA methylation (DNAm) age as described by the epigenetic clock is a novel marker of biological age. Few studies have investigated physical activity and DNAm age. Methods: We performed quantitative genetic modeling in young and older monozygotic (MZ) and dizygotic (DZ) twin pairs to investigate the contributions of genetic and shared/non-shared environmental variation on age acceleration (residuals from the regression of DNAm Age on chronological age). We tested the hypothesis that leisure-time physical activity is one of the non-shared environmental factors that affect epigenetic aging. We performed a co-twin control analysis with 16 same-sex twin pairs who had persistent discordance in physical activity for 32 years. Results: More of the variation in age acceleration was explained by non-shared environmental factors among older twin pairs 47 (95%CI 35, 63)% compared with younger pairs 26 (19, 35)%). DNAm age did not differ among active and inactive co-twins (60.7 vs. 61.8 yrs). Conclusions: Leisure-time physical activity during adult years has at most minor effects on epigenetic aging. This supports recent findings that long term leisure-time physical activity in adulthood has little or no effect on mortality after controlling for genetic factors.
S4.4.4 Epigenetic signature of frailty – discovering the longitudinal changes in DNA
methylation that underlie the development of frailty
Karolinska Institutet, Sweden
Background: Frailty is an adverse aging outcome that is associated with an increased risk of disability, hospitalizations and death. The level of frailty is also considered as a marker of biological age. However, the genomic determinants for the development of frailty are largely unknown. Methods: In the Swedish Adoption/Twin Study of Aging (n=365; age 48-94 years at baseline) we sought to identify the genome-wide DNA methylation changes that associate with the development of frailty over the follow-up period of up to 20 years. Linear mixed modeling was used in the analysis and the Rockwood frailty index (FI) was used to assess the level of frailty. Results: We found evidence for cross-sectional and longitudinal associations between the FI and DNA methylation sites only in men. The top five methylation sites (in men) that passed the multiple testing correction (FDR<0.05) and adjustment for confounders will be further tested in a growth curve model for their association with the developmental trajectories of frailty. Furthermore, we will examine to what extent the methylation levels in the frailty-associated sites are genetically determined. Conclusions: The development of frailty seems to be accompanied with DNA methylation changes only in men. Our further analyses will shed more light into the sex-specific aspects and the causality of this relationship.
S4.4.5 Allostatic load and aging
University of Edinburgh, UK
Background: Allostatic load is defined as the physiological wear and tear on the body caused by repeated exposure to stressors. The nature of load as a longitudinal process across the life-courses positions it as especially interesting with respect to later life effects. In addition, no studies have addressed the association with the epigenetic clock. Methods: Allostatic load (AL) will be defined in the Lothian Birth Cohort 1936 and discussed in the nature of AL as a construct. Epigenetic age will be calculated in the same samples across four waves covering 9 years using the online Horvath calculator. Results: Results will be presented as novel dual trajectory models of the relations between AL and epigenetic age across four waves (9 years) of later life in the Lothian Birth Cohort 1936. Conclusions: The overall association of AL and the epigenetic clock will be summarized.