Abstract
Biological age (BA) and its residual relative to chronological age are popularly used to quantify individual aging. Although these residuals independently predict age-related health outcomes, conventional BA measures often lack robustness in heterogeneous populations, and their residuals are not directly derivable in clinical practice. To address these limitations, we introduce the Gompertz law-based residual (GOLD-R) framework, a method designed to directly estimate BA residuals and optimized for cross-sectional data. We demonstrated the applicability and robustness of GOLD-R across multiple data types and populations. First, training on DNA methylation data from the EWAS Data Hub, the framework outperformed established epigenetic clocks in predicting mortality in a pan-cancer dataset. Then, applied to UK Biobank proteomics data, GOLD-R generated organismal and organ-specific aging measures that proved more robust than conventional age-prediction approaches in forecasting incident diseases and mortality. Finally, extending the analysis to clinical biomarkers using the NHANES and HRS, we found that GOLD-R residuals, derived from clinical biomarkers, surpassed those from both epigenetic and phenotypic clocks in performance. In summary, our findings establish GOLD-R as a robust algorithm for biological age estimation, providing a practical tool for both research and clinical applications.
Introduction
While chronological age (CA) is a simple count of years lived, biological age (BA) reflects the actual physiological state and decline of an individual. The discrepancy between these two—often called the "age gap" or residual—is a critical indicator of whether an person is aging faster or slower than average. Identifying this gap is the primary goal of "aging clocks," which use various biomarkers to predict health outcomes.
However, existing aging clocks face significant hurdles. Most calculate the age residual as an afterthought using linear regression, which can introduce mathematical biases. Furthermore, many high-performing models require longitudinal data—tracking the same individuals over decades—which is prohibitively expensive and rare. Finally, many clocks are not designed for clinical utility, requiring complex sequencing that isn't available in a standard doctor’s office. To solve these issues, researchers developed the GOLD-R (Gompertz law-based residual) framework, a robust method optimized for cross-sectional data that directly predicts an individual's aging deviation.
The GOLD-R Framework
The foundation of GOLD-R is the Gompertz law of mortality, which states that the risk of death increases exponentially as we age. Unlike standard clocks that attempt to guess a person's age directly, GOLD-R focuses on the "residual"—the gap between an individual's actual health risk and the risk expected for their chronological age.
The framework employs a two-stage LASSO regression process:
- Stage 1: Biomarkers are used to predict an individual's mortality risk.
- Stage 2: The model calculates the "residual risk" (the difference between biomarker-predicted risk and age-expected risk) and trains a second model to predict this residual directly.
By using chronological age as a strong baseline and biomarkers to capture the "biological noise" that age alone cannot explain, GOLD-R provides a much cleaner signal of accelerated or decelerated aging.
Multi-Omic Validation
The study validated GOLD-R across three distinct biological layers to ensure its versatility and robustness:
1. Epigenetics (DNA Methylation)
Training on DNA methylation data from the EWAS Data Hub (7,313 samples) and validating on pan-cancer data from The Cancer Genome Atlas (TCGA), GOLD-R demonstrated remarkable parsimony. While the well-known "GrimAge" clock utilizes 1,030 CpG sites, the GOLD-R DNAmAge clock achieved superior results using only 34 CpG sites. Despite its smaller footprint, GOLD-R was more accurate at predicting mortality in cancer patients.
2. Proteomics
Using the UK Biobank dataset (53,014 participants), the researchers analyzed nearly 3,000 plasma proteins. GOLD-R allowed for the creation of organ-specific clocks (e.g., kidney, brain, heart). The study found that kidney and heart aging were particularly strong predictors of overall healthspan, and the GOLD-R approach proved more robust than conventional age-prediction methods in forecasting incident diseases.
3. Clinical Biomarkers
To ensure clinical relevance, the model was applied to standard lab results from the NHANES and the Health and Retirement Study (HRS). A version of the model using standard clinical biomarkers (GOLD-R DNAmCliAge) outperformed traditional phenotypic clocks in predicting 10-year survival probabilities, suggesting that expensive sequencing is not always necessary for accurate biological age estimation.
Key Results and Insights
- Superior Predictive Power: In pan-cancer datasets, GOLD-R achieved a higher C-index (0.623) for mortality prediction than established models like GrimAge (0.619) and PhenoAge (0.589).
- Robustness to Population Noise: In simulations, GOLD-R maintained its predictive power even when population-specific noise was introduced, whereas conventional models saw significant degradation in performance.
- Organ-Specific Insights: The flexibility of the framework allowed researchers to isolate which organs were aging most rapidly, providing a more granular view of an individual's health than a single aggregate score.
Conclusion and Future Directions
GOLD-R represents a significant step forward in the practical application of aging clocks. By focusing on the residual risk and leveraging the Gompertz law, the framework provides a more accurate and robust measure of biological aging that works across different data types and populations.
While the results are promising, future research is needed to determine if GOLD-R can predict non-lethal "geriatric syndromes" such as frailty or cognitive decline before they manifest as disease. Additionally, the integration of DNA, proteins, and clinical markers into a single "pan-biological" clock remains a high-priority goal for the next generation of aging research.
Summary updated