Beydoun May A, Noren Hooten Nicole, Asefa Nigus G, Georgescu Michael F, Song Minkyo, Beydoun Hind A, Banerjee Sri, Khubchandani Jagdish, Meirelles Osorio, Launer Lenore J, Evans Michele K, Zonderman Alan B
Laboratory of Epidemiology and Population Sciences, National Institute on Aging, Baltimore, Maryland, USA.
VA National Center on Homelessness Among Veterans, U.S. Department of Veterans Affairs, Washington, District of Columbia, USA.
Aging Cell. 2025 Sep;24(9):e70159. doi: 10.1111/acel.70159. Epub 2025 Jul 6.
Telomere length and DNA methylation (DNAm) clocks serve as markers of biological aging and have been linked to mortality risk. This study applies additive Bayesian networks (ABNs) to examine associations between DNAm clocks, telomere length, and mortality, with a focus on racial and sex differences in aging. Data from three US cohorts-NHANES (n = 2522), HRS (n = 1029), and HANDLS (n = 92-470)-were analyzed using correlation matrices, Cox models, ABNs, and generalized structural equation models (GSEM) with mortality from the National Death Index. Epigenetic clocks, particularly GrimAgeEAA, HannumAgeEAA, and DunedinPoAM (or DunedinPACE), were stronger mortality predictors than telomere length. ABNs highlighted key relationships, consistently linking age and GrimAgeEAA to mortality in NHANES and HRS. GSEM models derived from ABNs indicated an inverse association between female sex and GrimAgeEAA in NHANES (β = -0.500) and HRS (β = -0.563), suggesting slower biological aging in women, although GrimAge clock incorporates sex in its definition. GrimAgeEAA strongly predicted mortality (LnHR, β ± SE of +0.476 ± 0.0393 in NHANES and +0.511 ± 0.0775 in HRS). Non-Hispanic Black adults exhibited accelerated aging via DunedinPoAM, partially mediating their higher mortality risk. Hispanic adults in NHANES had unique associations with PhenoAgeEAA (β = +0.197), a mortality predictor. DNAm clocks, particularly GrimAgeEAA, outperform telomere length in predicting mortality. Second-generation epigenetic aging markers offer insights into demographic disparities in aging and mortality, with ABNs revealing complex interrelations among aging biomarkers, sex, race, and mortality risk.
端粒长度和DNA甲基化（DNAm）时钟作为生物衰老的标志物，与死亡风险相关联。本研究应用加法贝叶斯网络（ABN）来检验DNAm时钟、端粒长度和死亡率之间的关联，重点关注衰老过程中的种族和性别差异。使用相关矩阵、Cox模型、ABN以及广义结构方程模型（GSEM），结合来自国家死亡指数的死亡率数据，对美国三个队列——美国国家健康与营养检查调查（NHANES，n = 2522）、健康与退休研究（HRS，n = 1029）和青少年及成人纵向健康与生活方式研究（HANDLS，n = 92 - 470）的数据进行了分析。表观遗传时钟，特别是GrimAgeEAA、HannumAgeEAA和达尼丁多学科衰老量表（DunedinPoAM，或达尼丁PACE），在预测死亡率方面比端粒长度更强。ABN突出了关键关系，在NHANES和HRS中始终将年龄和GrimAgeEAA与死亡率联系起来。从ABN推导出来的GSEM模型表明，在NHANES（β = -0.500）和HRS（β = -0.563）中，女性性别与GrimAgeEAA之间存在负相关，这表明女性的生物衰老速度较慢，尽管GrimAge时钟在其定义中纳入了性别因素。GrimAgeEAA强烈预测死亡率（NHANES中LnHR，β ± SE为+0.476 ± 0.0393，HRS中为+0.511 ± 0.0775）。非西班牙裔黑人成年人通过达尼丁多学科衰老量表表现出加速衰老，部分介导了他们较高的死亡风险。NHANES中的西班牙裔成年人与PhenoAgeEAA（β = +0.197）存在独特关联，PhenoAgeEAA是一种死亡率预测指标。DNAm时钟，特别是GrimAgeEAA，在预测死亡率方面优于端粒长度。第二代表观遗传衰老标志物为衰老和死亡率方面的人口统计学差异提供了见解，ABN揭示了衰老生物标志物、性别、种族和死亡风险之间的复杂相互关系。
