TranslAGE: Translating Geroscience to Humans by Validating and Refining Aging Biomarkers in Longitudinal Studies

The TranslAGE research project aims to evaluate and optimize DNA methylation (DNAm)-based aging biomarkers for geroscience clinical trials. We will benchmark existing biomarkers to identify those that are: (1) responsive to aging interventions, (2) prognostic for age-related diseases and outcomes, and (3) stable over time without intervention. This will guide the development of next-generation biomarkers.

Study Design: We will analyze DNAm and phenotypic data from the Health and Retirement Study (HRS), evaluating how various DNAm biomarkers relate to mortality, disease, and physical and cognitive function. All pipelines and variable names will be harmonized with other major studies, including the Baltimore Longitudinal Study of Aging (BLSA) and Framingham Heart Study, enabling cross-study comparison.

Analysis Plan:
1. Data Harmonization: Core variables (age, sex, death, follow-up time) will be standardized using a common data dictionary.
2. Biomarker Computation: Multiple DNAm biomarkers will be calculated using https://github.com/HigginsChenLab/methylCIPHER and Z-scored for comparability.
3. Model Fitting: Cox models for mortality and disease onset, Logistic regression for incident disease, Linear models for continuous traits like cognition, BMI, gait speed.
4. Performance Evaluation: Predictive performance will be assessed using hazard ratios, concordance index (C-index), and AUCs. Biomarkers will be ranked across domains, and visualizations will summarize associations.
5. Cross-Study Integration: Summary statistics will be compared with those from external cohorts, intervention studies (e.g., effect sizes), and test-retest reliability datasets (e.g., intraclass correlation)
6. Composite Development: Top biomarkers with strong prognostic power, responsiveness, and stability will be combined into a composite score. Novel mortality, morbidity, and frailty predictors will be trained in external datasets and validated in HRS.

Planned collaborations: Sofiya Milman at Albert Einstein College of Medicine (relating DNAm biomarkers to frailty), Jessica Lasky-Su at Brigham and Women’s Hospital (relating DNAm proxies of serum proteins and metabolites to aging outcomes).

The TranslAGE knowledgebase aims to improve how we measure biological aging, helping scientists evaluate new treatments that target the aging process. We will use DNA methylation from these and other studies to identify biomarkers that predict future health, respond to anti-aging therapies, and remain stable over time. We will calculate many different biological age scores from blood samples and test how well each one predicts outcomes like memory, physical ability, disease, and mortality. We will combine this information with results from other studies to identify the best scores that not only predict future age-related disease, but are also responsive to treatments intended to prevent future disease. We will also use insights from this study to develop better biological age scores that may eventually be used to assess whether interventions are successfully modifying the aging process to promote healthier, longer lives.