Basic Science and Pathogenesis

Advancing chronological age is a strong contributor to Alzheimer's disease (AD) risk, yet individual variability exists in aging trajectories. Novel AD biomarkers are needed to better stratify AD risk, especially in preclinical stages. To this end, we computed epigenetic "clocks," a measure of biological age based on DNA methylation markers, and tested associations with dementia and AD biomarkers.

We leveraged whole blood and brain DNA methylation data collected from the EFIGA aging and AD cohort along with the Methylclock R package to compute Horvath and Hannum aging clocks (N = 731). Next, we calculated age acceleration, by residualizing the regression of biological age on chronological age. We ran linear regression models testing associations between biological age and age acceleration with a panel of plasma dementia and AD biomarkers. All regressions were adjusted for multiple comparisons (false-discovery rate [FDR]).

Biological age was significantly correlated with chronological age (R² _Horvath=0.18, p_Horvath =4.74x10^-33; R² _Hannum=0.17, p_Hannum=3.48x10^-30) and with sex (β_Horvath=0.21, p_Horvath=8.44x10^-3; β_Hannum=0.27, p_Hannum=7.32x10^-4). We replicated correlations in brain (R² _Horvath=0.52, p_Horvath=6.27x10^-89; R² _Hannum=0.21, p_Hannum=6.46x10^-30), and in an independent aging and AD cohort, ROS/MAP (in brain - R² _Horvath=0.49, p_Horvath=6.51x10^-78; R² _Hannum=0.34, p_Hannum=1.03x10^-47). Seven plasma biomarkers were significantly associated with biological age calculated with the Horvath clock, the Hannum clock, or both: p-tau217 (β_Horvath=0.13, p.FDR_Horvath=2.69x10^-3; β_Hannum=0.19 p.FDR_Hannum=1.35x10^-5), p-tau181 (β_Horvath=0.13, p.FDR_Horvath=1.75x10^-3; β_Hannum=0.14 p.FDR_Hannum=1.75x10^-3), GFAP (β_Horvath=0.09, p.FDR_Horvath=2.07x10^-2; β_Hannum=0.15 p.FDR_Hannum=5.77x10^-4), p-tau231 (β_Horvath=0.11, p.FDR_Horvath=9.87x10^-3), NfL (β_Hannum=0.11, p.FDR_Hannum=9.87x10^-3), NfL (β_Hannum=0.11, p.FDR_Hannum=9.87x10^-3), Aβ40 (β_Horvath=0.09, p.FDR_Horvath=2.07x10^-2; β_Hannum=0.11, p.FDR_Hannum=1.66x10^-2), and total-tau (β_Horvath=0.09, p.FDR_Horvath=2.07x10^-2).

We demonstrated that biological age is a predictor of AD pathology in preclinical and clinical stages, and we will next evaluate biological age acceleration as a predictor of pathology. In addition, we will be replicating all biomarker associations in brain by testing associations with autopsy measures of neuropathology. Future directions will include conducting models stratified by sex and by APOE-ε4 carrier status to identify if the relationship between biological aging and AD biomarkers is modulated by biological sex or APOE-ε4 genotype.