Maximizing insights from longitudinal epigenetic age data: simulations, applications, and practical guidance

Area of Science:

• Genomics
• Epigenetics
• Aging Research

Background:

• Epigenetic age (EA) estimates biological age using DNA methylation (DNAm).
• Epigenetic age acceleration (EAA) signifies deviations from chronological aging and is linked to health outcomes.
• Longitudinal studies are increasingly used to analyze EA dynamics, but methodological variations exist.

Purpose of the Study:

• To evaluate the robustness of different analytical approaches for longitudinal epigenetic age studies.
• To identify optimal statistical models for analyzing changes in epigenetic age over time.
• To guide future research in epigenetic aging methodology.

Main Methods:

• Simulations were conducted to compare various statistical models, outcome variables (EA vs. EAA), and time variables (chronological age vs. age-independent).
• Linear mixed models and generalized estimating equations were assessed.
• Two real-world datasets were analyzed using biological sex and birthweight as predictors of longitudinal EA.

Main Results:

• Linear mixed models or generalized estimating equations using chronological age as the time variable provided the most accurate effect size estimates.
• The choice between EA and EAA as an outcome had minimal impact on the estimates.
• Analysis of real-world data revealed that GrimAge accelerates in males at birth but decelerates over time.

Conclusions:

• The findings offer guidance for selecting appropriate methodologies in longitudinal epigenetic age research.
• Accurate estimation of epigenetic age dynamics is crucial for reliable association studies.
• Methodological choices can significantly influence the interpretation of epigenetic aging patterns.