DNA methylation trajectories and accelerated epigenetic aging in incident type 2 diabetes

Abstract

DNA methylation (DNAm) patterns across the genome changes during aging and development of complex diseases including type 2 diabetes (T2D). Our study aimed to estimate DNAm trajectories of CpG sites associated with T2D, epigenetic age (DNAmAge), and age acceleration based on four epigenetic clocks (GrimAge, Hannum, Horvath, phenoAge) in the period 10 years prior to and up to T2D onset. In this nested case-control study within Doetinchem Cohort Study, we included 132 incident T2D cases and 132 age- and sex-matched controls. DNAm was measured in blood using the Illumina Infinium Methylation EPIC array. From 107 CpG sites associated with T2D, 10 CpG sites (9%) showed different slopes of DNAm trajectories over time (p < 0.05) and an additional 8 CpG sites (8%) showed significant differences in DNAm levels (at least 1%, p-value per time point < 0.05) at all three time points with nearly parallel trajectories between incident T2D cases and controls. In controls, age acceleration levels were negative (slower epigenetic aging), while in incident T2D cases, levels were positive, suggesting accelerated aging in the case group. We showed that DNAm levels at specific CpG sites, up to 10 years before T2D onset, are different between incident T2D cases and healthy controls and distinct patterns of clinical traits over time may have an impact on those DNAm profiles. Up to 10 years before T2D diagnosis, cases manifested accelerated epigenetic aging. Markers of biological aging including age acceleration estimates based on Horvath need further investigation to assess their utility for predicting age-related diseases including T2D.

Keywords: Aging markers; Biological age; DNA methylation; Epigenetic clock; Type 2 diabetes.

Fig. 1

Timeline diagram showing sampling of 132 T2D case–control pairs from the Doetinchem Cohort Study for each timepoint. T-10 and T-5 represent approximately 10 and 5 years, respectively, before T2D diagnosis; T0 = T2D diagnosis

Fig. 2

DNAm trajectories of four CpG sites showing significantly different slopes and significant differences in methylation levels for at least two timepoints between incident T2D cases and controls. Trajectories were extracted from linear mixed effects models adjusted for age at time of diagnosis (T0), sex, cell type proportion, and batch effects. Ribbons represent 95% confidence intervals

Fig. 3

DNAm trajectories for eight CpG sites showing significantly (p < 0.05) different DNAm levels at each time point (T-10, T-5, and T0) between incident T2D cases and controls. Trajectories were extracted from linear mixed effects models adjusted for age at time of diagnosis (T0), sex, cell type proportion, and batch effects. Ribbons represent 95% confidence intervals

Fig. 4

Trajectories and levels of DNAmAges and age acceleration based on four epigenetic clocks in incident T2D cases and controls. Trajectories were extracted from linear mixed effects models. DNAmAge models were adjusted for sex, estimated white cell types and batch effects and age acceleration models for sex. Ribbons represent 95% confidence intervals. The dotted line in the upper plots represents mean chronological age (which was the same for cases and controls due to age-matching); in the lower plots, the dotted line is y = 0, indicating no age acceleration