Improved prediction of chronological age from DNA methylation limits it as a biomarker of ageing

Abstract

DNA methylation is associated with age. The deviation of age predicted from DNA methylation from actual age has been proposed as a biomarker for ageing. However, a better prediction of chronological age implies less opportunity for biological age. Here we used 13,661 samples (from blood and saliva) in the age range of 2 to 104 years from 14 cohorts measured on Illumina HumanMethylation450/EPIC arrays to perform prediction analyses. We show that increasing the sample size achieves a smaller prediction error and higher correlations in test datasets. We demonstrate that smaller prediction errors provide a limit to how much variation in biological ageing can be captured by methylation and provide evidence that age predictors from small samples are prone to confounding by cell composition. Our predictor shows a similar or better performance in non-blood tissues including saliva, endometrium, breast, liver, adipose and muscle, compared with Horvath’s across-tissue age predictor.

OSCA: a tool for omic-data-based complex trait analysis

“…A by-product of the OREML application is to estimate the joint effects of all measures of one or multiple omic profiles (i.e., OBLUP analysis) to predict the phenotype in a new sample. This has been shown to be a powerful and robust approach in age prediction using gene expression or DNAm data [53,54]. We have also provided computationally efficient implementations in OSCA to manage large-scale omic data, and to perform omic-data-based quantitative trait locus (xQTL) analysis and meta-analysis of xQTL summary data.…”

Accurate prediction of cell composition, age, smoking consumption and infection serostatus based on blood DNA methylation profiles

“…The performance of models, together with the number of predictors that is optimal in terms of prediction accuracy, is shown in Table 1 for each cell proportion, as well as 23 other continuous traits, including age and morphometric and physiological measures. DNA methylation levels can accurately predict age and sex (18), intrinsic factors that are predictive of many traits, including immune cell counts in whole blood (14). It is therefore important to discern when predictors based on methylation probes give additional information to these two commonly available factors.…”

“…Age %B cells in CD45 cells %CD8 naive T cells (18,23). From Table 1 it appears that our elastic net models are also able to estimate red blood cell counts, height and weight with high accuracy.…”

“…Interestingly, out-of-sample prediction error decreases rapidly with the number of predictors, and plateaus at around 50 predictors ( Figure 1). This indicates that accurate predictions can be achieved with much fewer predictors than the hundreds of DNAm probes used by current prediction models of cell composition (9,12) and age (18,23). These results suggest that blood cell composition can be predicted well using only a few number of probes that are markers for differentiation events.…”

“…Therefore, to infer optimal models in terms of prediction accuracy, and to investigate how prediction accuracy depends on the number of predictors, we use the elastic net method (16,17). Similar models have previously been used for the prediction of age, smoking status, alcohol consumption and educational attainment based on DNAm (18,19). We believe that elastic net regression will be able to find both the predictors that mark differentiation events for the lineage of the cell, but also the numerous probes that are correlated with such predictors.…”