CardioMetAge, an aging clock constructed from chronological age and 12 common clinical biomarkers, outperformed existing aging clocks in predicting cardiometabolic disease mortality, incidence, and progression, and was slowed by two-year caloric restriction by 1.23 years relative to ad libitum control.
Key Findings
Methods
CardioMetAge was constructed as a linear combination of chronological age and 12 common clinical biomarkers, trained in NHANES-III.
The model was trained in the NHANES-III dataset and applied to continuous NHANES and UK Biobank cohorts.
The final model uses 12 common clinical biomarkers combined with chronological age.
The age deviation metric derived from this model is termed CardioMetAgeDev.
Results
CardioMetAgeDev showed stronger associations with cardiometabolic disease (CMD) mortality compared to deviations of PhenoAge and other traditional biological age models.
HR per SD for CMD mortality: 1.87 (95% CI: 1.83, 1.91).
CardioMetAgeDev outperformed deviations from PhenoAge and other traditional biological age models.
Associations were examined across the continuous NHANES and UK Biobank cohorts.
Results
CardioMetAgeDev was associated with CMD incidence.
HR per SD for CMD incidence: 1.35 (95% CI: 1.33, 1.37).
This association was stronger than those observed for deviations of PhenoAge and other traditional biological age models.
Results
CardioMetAgeDev was associated with disease progression transitions between cardiometabolic disease states.
HR per SD for transition from no CMD to first CMD: 1.34 (95% CI: 1.32, 1.35).
HR per SD for transition from first CMD to cardiometabolic multimorbidity: 1.25 (95% CI: 1.21, 1.30).
CardioMetAgeDev outperformed PhenoAge deviation and other traditional biological age model deviations for these transitions.
Results
CardioMetAge consistently outperformed PhenoAge and other traditional biological age models in predicting 10-year CMD incidence.
Performance was evaluated for 10-year CMD incidence prediction.
Comparisons were made against deviations of PhemaAge and other traditional biological age models.
The model was validated in the continuous NHANES and UK Biobank cohorts.
Results
Proteomic analyses linked CardioMetAgeDev to inflammatory activation and metabolic disorders.
Proteomic pathway analyses were conducted to identify biological determinants of cardiometabolic aging.
CardioMetAgeDev was associated with proteomic pathways related to inflammatory activation and metabolic disorders.
These analyses highlighted the biological underpinnings of cardiometabolic aging captured by the model.
Results
Lifestyle factors and socioeconomic status were associated with CMD risks partly via CardioMetAgeDev, with measurable mediation proportions.
Mediation proportion for lifestyle factors via CardioMetAgeDev: 34.5%.
Mediation proportion for socioeconomic status via CardioMetAgeDev: 10.7%.
Both modifiable factors were analyzed for their associations with CMD risk through the CardioMetAgeDev pathway.
Results
Two-year caloric restriction slowed the progression of CardioMetAge compared to ad libitum control.
Two-year caloric restriction slowed CardioMetAge progression by 1.23 years (95% CI: 0.61, 1.84) relative to the ad libitum control.
This finding demonstrates the responsiveness of CardioMetAge to dietary intervention.
The caloric restriction intervention analysis evaluated the impact of the intervention on CardioMetAge change over two years.
Li Y, Xu X, Zheng Y, He X, Wang J, Liu Z, et al.. (2026). CardioMetAge estimates cardiometabolic aging and predicts disease outcomes.. BMC medicine. https://doi.org/10.1186/s12916-026-04621-5