Epigenetic markers of middle-age: non-linear DNA methylation changes with aging in humans

Background Human DNA is known to exhibit an overall tendency toward demethylation with aging. However, assuming a simple linear relationship between DNA methylation and age does not align with the phenotype of human development and the aging process. This study aimed to investigate the existence of DNA methylation patterns with peaks or troughs at specific ages in addition to simple linear changes. Methods A large-scale dataset of genome-wide DNA methylation data from 10,420 individuals was analyzed. Hierarchical multiple regression models were applied to detect patterns of the association between age and DNA methylation: linear increase, linear decrease, U-shaped curve, and inverse U-shaped curve. Results Among the 864,627 CpG sites analyzed, 8.4% exhibited an increase in DNA methylation with age, 23.9% showed a decrease, and 5.5% were better explained by a quadratic model (P < 5.7815×10 ). Within the non-linear subset, inverse U-shaped CpG sites peaking in methylation during middle age were predominant. Genes exhibiting quadratic association patterns between DNA methylation and age, and those linked to diseases with common onset during middle age, were also detected. Conclusions Non-linear age-related DNA methylation patterns, with peaks or troughs occurring at specific ages, were detected. This suggests that humans do not simply age linearly, but that programmed mechanisms or cascade-like processes may exist to promote or suppress the expression of specific genes at certain ages, contributing onset of certain diseases at specific timings.