Refining epigenetic prediction of chronological and biological age

Abstract

Epigenetic clocks can track both chronological age (cAge) and biological age (bAge). The latter is typically defined by physiological biomarkers and risk of adverse health outcomes, including all-cause mortality. As cohort sample sizes increase, estimates of cAge and bAge become more precise. Here, we aim to refine predictors and improve understanding of the epigenomic architecture of cAge and bAge. First, we perform large-scale (N = 18,413) epigenome-wide association studies (EWAS) of chronological age and all-cause mortality. Next, to improve cAge prediction, we use methylation data from 24,673 participants from the Generation Scotland (GS) study, the Lothian Birth Cohorts (LBC) of 1921 and 1936 and 8 publicly available datasets. Through the inclusion of linear and non-linear age-CpG associations from the EWAS, feature pre-selection/dimensionality reduction in advance of elastic net regression, and a leave-one-cohort-out (LOCO) cross validation framework, we arrive at an improved cAge predictor (median absolute error = 2.3 years across 10 cohorts). In addition, we train a predictor of bAge on 1,214 all-cause mortality events in GS, based on epigenetic surrogates for 109 plasma proteins and the 8 component parts of GrimAge, the current best epigenetic predictor of all-cause mortality. We test this predictor in four external cohorts (LBC1921, LBC1936, the Framingham Heart Study and the Women's Health Initiative study) where it outperforms GrimAge in its association to survival (HRGrimAge = 1.47 [1.40, 1.54] with p = 1.08 x 10-52, and HRbAge = 1.52 [1.44, 1.59] with p = 2.20 x 10-60). Finally, we introduce MethylBrowsR, an online tool to visualize epigenome-wide CpG-age associations.

Blood-based DNA methylation study of alcohol consumption

“…Previous studies have found that pre-filtering ahead of elastic net greatly improves predictor performance when using this training method 36 . Our current results echo this, with a significant increase in prediction accuracy found when training on CpGs with a previously established association to alcohol consumption (Table 1).…”

An overview of DNA methylation-derived trait score methods and applications

“…Such prospective DNAm data sets are not yet common. One of the largest cohorts to date is Generation Scotland (an adult community cohort) which has 18,413 individuals and electronic health linkage spanning 15 years after blood sampling and has been used to investigate MPS associations with the incidence of 11 major morbidities, including type 2 diabetes [3,20]. The CHARGE consortium has reported MPS in relation to incident coronary heart disease in a follow-up of 11.2 years, on average, following sample collection using data from 9 cohorts comprising 11,461 individuals [26].…”

“…However, in current commercial arrays, these probes have been selected to be informative, being annotated to 96% of coding genes, as well as targeting known enhancer and promoter elements. The measurement of DNAm by array technology is cheaper and more high-throughput than by sequencing [2] and so relatively large human data sets have been generated from array technology [3] to identify probes which show differential DNAm at CpG sites associated with traits. These methylation-wide association study (MWAS) data sets are many-fold smaller than genome-wide association study (GWAS) data sets that rely predominantly on SNP array data.…”

Probabilistic inference of epigenetic age acceleration from cellular dynamics

“…In recent years, advances in the field have led to numerous improvements. First, population-based cohorts used for training have increased from modest sizes to include and combine large cross-sectional studies with thousands of participants 3,4,7,8 . Second, the complexity of models has advanced in parallel on two fronts.…”