Essentially, aging is characterized by progressive degeneration and loss of function across multiple physiological systems. While chronological age is the most straightforward indicator of aging, the variability in aging across different organ systems results in a wide variation in aging characteristics among individuals of the same chronological age. In recent years, various algorithms based on DNA methylation (DNAm), such as HorvathAge and GrimAge, have provided new ways to estimate biological age and have shown potential in predicting mortality and age-related diseases. However, due to differences in objectives, methodologies, and tissue types used among these algorithms, it remains uncertain which tool best captures the true state of biological aging. This study focuses on twelve different DNAm signatures of aging to explore their associations with mortality and compare their predictive abilities.
The study included participants aged 50 or older from the U.S. National Health and Nutrition Examination Survey 1999-2000, with mortality data linked to the National Death Index through December 31, 2019. A total of 2,532 participants were included, with an average chronological age of 66.13 years. During a median follow-up period of 17.17 years, 1,361 deaths were observed. The study incorporated twelve DNAm estimators, including HorvathAge, HannumAge, SkinBloodAge, LevinePhenoAge, ZhangAge, LinAge, WeidnerAge, VidalBraloAge, GrimAge, GrimAge2, HorvathTelo, and DunedinPoAm. It was found that almost all DNAm estimators and chronological age were significantly correlated, with correlation coefficients ranging from 0.70 to 0.99. After adjusting for multiple confounding factors, we revealed that several algorithms, including HorvathAge, HannumAge, LevinPhenoAge, GrimAge, GrimAge2, HorvathTelo, and DunedinPoAm, were significantly associated with all-cause mortality. Notably, when all twelve algorithms were included in the same model, only GrimAge2 remained significantly associated with all-cause mortality, with a hazard ratio (HR) of 2.69 per standard deviation (SD) increase. Compared to chronological age, GrimAge2 and GrimAge demonstrated superior predictive abilities, with Harrell's C-statistics of 0.760 and 0.759, respectively.
Strengths of this study include the use of a large, nationally representative cohort with a long follow-up, providing insights into their utility for public health and aging research, and enhancing the accuracy and generalizability of the findings. However, limitations in this study should be noted. First, as an observational study, the relationships between DNAm signatures of aging and mortality might be biased by reverse causality and unknown confounding factors. Second, the DNAm signatures of aging were assessed once, so we cannot track their stability or changes over time. Last, our study focused on U.S. participants, with non-Hispanic white individuals accounting for 40.64%. Further validation in diverse populations (such as Asian population) with varying age ranges and risk profiles is needed to confirm the generalizability of our results.
In conclusion, our study revealed that DNAm signatures of aging are robustly and independently associated with all-cause mortality. Of these, GrimAge2 stands out as the strongest mortality predictor, surpassing other DNAm signatures and chronological age. Our findings underscore that GrimAge2 is a promising tool for assessing mortality risk and evaluating healthy aging interventions. Future studies are needed to validate these findings and explore the underlying mechanisms by which GrimAge2 affects mortality.
About Author: (Briefly outline the corresponding author's academic achievements and research contributions) (One author only)
Professor Xiangwei Li holds a PhD in epidemiology from the Heidelberg University School of Medicine/German Cancer Research Center. He was previously employed at the Chinese Academy of Medical Sciences/Peking Union Medical College; Heidelberg University Medical School/German Cancer Research Center. He is now a professor at the Shanghai Jiao Tong University School of Medicine, China. He has published more than 20 papers in peer-reviewed international journals, such as Lancet Infectious Diseases, JAMA Oncology, Nature Communications, eClinicalMedicine, and eBioMedicine. He also serves as Associate Editor of JMIR Public Health and Surveillance, and Youth Editor of the journal hLife.
