Allostatic Load of Glucose Homeostasis, Epigenetic Biomarkers, and All‐Cause and Cardiovascular Death: A Prospective Cohort Study

BACKGROUND: To evaluate the association between allostatic load of glucose homeostasis system (AL-GHS), a cumulative index of metabolic stress on glucose homeostasis, with all-cause and cardiovascular disease death, and to explore mediating roles of multiple epigenetic biomarkers in such association. METHODS: This study included 1021 US adults aged ≥50 years from the 1999 to 2002 National Health and Nutrition Examination Survey, with mortality follow-up through December 2019. AL-GHS (range, 0-6) was calculated on the basis of 6 glucose-related biomarkers: fasting glucose, insulin, homeostasis model assessment of insulin resistance and β-cell function, triglyceride-glucose index, and glycemic index. Association of AL-GHS with all-cause and cardiovascular death was assessed using weighted Cox proportional hazards models, supplemented by Kaplan-Meier curve and restricted cubic spline regression. Mediation analyses were performed to evaluate mediating effects through epigenetic biomarkers. RESULTS: Per 1-score increase in AL-GHS was linearly associated with a 9.5% and a 15.8% increment in the risk of all-cause (hazard ratio [HR], 1.095 [95% CI, 1.016-1.181]) and cardiovascular disease (HR, 1.158 [95% CI, 1.010-1.327]) death, respectively. Epigenetic biomarkers HorvathAge (HR, 1.292 [95% CI, 1.105-1.510]), HannumAge (HR, 1.307 [95% CI, 1.054-1.621]), PhenoAge (HR, 1.428 [95% CI, 1.119-1.822]), GrimAgeMort (HR, 1.753 [95% CI, 1.265-2.429]), GrimAge2Mort (HR, 1.725 [95% CI, 1.315-2.263]), and HorvathTelo (HR, 0.807 [95% CI, 0.688-0.947]) were linearly related to all-cause death while GrimAgeMort (HR, 1.753 [95% CI, 1.265-2.429]) and GrimAge2Mort (HR, 1.725 [95% CI, 1.315-2.263]) were linearly related to cardiovascular disease death. HorvathAge and PhenoAge mediated 7.53% and 18.00%, respectively, of the relationship between AL-GHS and all-cause death. CONCLUSIONS: AL-GHS was positively associated with both all-cause and cardiovascular death, and the association with all-cause death was partially mediated by HorvathAge and PhenoAge.