There has been a steady increase in the number of studies aiming to identify DNA methylation differences associated with complex phenotypes. Many of the challenges of epigenetic epidemiology regarding study design and interpretation have been discussed in detail, however there are analytical concerns that are outstanding and require further exploration. In this study we seek to address three analytical issues. First, we quantify the multiple testing burden and propose a standard statistical significance threshold for identifying DNA methylation sites that are associated with an outcome. Second, we establish whether linear regression, the chosen statistical tool for the majority of studies, is appropriate and whether it is biased by the underlying distribution of DNA methylation data. Finally, we assess the sample size required for adequately powered DNA methylation association studies. We quantified DNA methylation in the Understanding Society cohort (n = 1175), a large population based study, using the Illumina EPIC array to assess the statistical properties of DNA methylation association analyses. By simulating null DNA methylation studies, we generated the distribution of p-values expected by chance and calculated the 5% family-wise error for EPIC array studies to be 9 × 10− 8. Next, we tested whether the assumptions of linear regression are violated by DNA methylation data and found that the majority of sites do not satisfy the assumption of normal residuals. Nevertheless, we found no evidence that this bias influences analyses by increasing the likelihood of affected sites to be false positives. Finally, we performed power calculations for EPIC based DNA methylation studies, demonstrating that existing studies with data on ~ 1000 samples are adequately powered to detect small differences at the majority of sites. We propose that a significance threshold of P < 9 × 10− 8 adequately controls the false positive rate for EPIC array DNA methylation studies. Moreover, our results indicate that linear regression is a valid statistical methodology for DNA methylation studies, despite the fact that the data do not always satisfy the assumptions of this test. These findings have implications for epidemiological-based studies of DNA methylation and provide a framework for the interpretation of findings from current and future studies.
Characterizing the complex relationship between genetic, epigenetic, and transcriptomic variation has the potential to increase understanding about the mechanisms underpinning health and disease phenotypes. We undertook a comprehensive analysis of common genetic variation on DNA methylation (DNAm) by using the Illumina EPIC array to profile samples from the UK Household Longitudinal study. We identified 12,689,548 significant DNA methylation quantitative trait loci (mQTL) associations (p < 6.52 × 10-14) occurring between 2,907,234 genetic variants and 93,268 DNAm sites, including a large number not identified by previous DNAm-profiling methods. We demonstrate the utility of these data for interpreting the functional consequences of common genetic variation associated with > 60 human traits by using summary-data-based Mendelian randomization (SMR) to identify 1,662 pleiotropic associations between 36 complex traits and 1,246 DNAm sites. We also use SMR to characterize the relationship between DNAm and gene expression and thereby identify 6,798 pleiotropic associations between 5,420 DNAm sites and the transcription of 1,702 genes. Our mQTL database and SMR results are available via a searchable online database as a resource to the research community.
Background: The Horvath epigenetic clock is widely used. It predicts age quite well from 353 CpG sites in the DNA methylation profile in unknown samples and has been used to calculate “age acceleration” in various tissues and environments. Results: The model systematically underestimates age in tissues from older people. This is seen in all examined tissues but most strongly in the cerebellum and is consistently observed in multiple datasets. Age acceleration is thus age-dependent, and this can lead to spurious associations. The current literature includes examples of association tests with age acceleration calculated in a wide variety of ways. Conclusions: The concept of an epigenetic clock is compelling, but caution should be taken in interpreting associations with age acceleration. Association tests of age acceleration should include age as a covariate.
Accelerated DNA methylation age is linked to all-cause mortality and environmental factors, but studies of associations with socioeconomic position are limited. Researchers generally use small selected samples, and it is unclear how findings obtained with 2 commonly used methods for calculating methylation age (the Horvath method and the Hannum method) translate to general population samples including younger and older adults. Among 1,099 United Kingdom adults aged 28–98 years in 2011–2012, we assessed the relationship of Horvath and Hannum DNA methylation age acceleration with a range of social position measures: current income and employment, education, income and unemployment across a 12-year period, and childhood social class. Accounting for confounders, participants who had been less advantaged in childhood were epigenetically “older” as adults: In comparison with participants who had professional/managerial parents, Hannum age was 1.07 years higher (95% confidence interval: 0.20, 1.94) for participants with parents in semiskilled/unskilled occupations and 1.85 years higher (95% confidence interval: 0.67, 3.02) for those without a working parent at age 14 years. No other robust associations were seen. Results accord with research implicating early life circumstances as critical for DNA methylation age in adulthood. Since methylation age acceleration as measured by the Horvath and Hannum estimators appears strongly linked to chronological age, researchers examining associations with the social environment must take steps to avoid age-related confounding.