Supplementary Material for: Plasma metabolomic signature predicting deviation from healthy vascular aging in two European cohorts

Introduction: Biological rather than chronological age (C-age) drives deviation from healthy aging. This study aimed to identify a plasma metabolomic signature indicative of age-related cardiovascular risk (pMTB-age) predicting vascular morbidity and mortality. Methods: Nuclear magnetic resonance identified 38 plasma metabolites in two population cohorts: the Flemish cohort examined as discovery (N = 719 [2005–2010]) and internal replication cohort (N = 580/719 [2009–2013]) and the external replication Spanish HORTEGA cohort (N = 811 [2001]). Results: The trained model (pMTB-age), relating C-age to the plasma metabolome derived by elastic net regression, included 18 metabolites (six amino-acids) and explained from 28.6% to 22.9% of C-age in the discovery and replication data. Feature importance of the retained metabolites derived by SHapley Additive exPlanation showed large interindividual variability in the relation between C-age and pMTB-age. In Flemish and Spanish, cardiovascular risk factors were significantly associated with C-age, pMTB-age and pMTB-age uncorrelated from C age (pMTB-age-R). In Flemish (median follow-up 12.3 years) and Spanish (18.8 years) mortality and cardiovascular complications correlated with C-age and pMTB-age. In Flemish, cardiac endpoints (hazard ratio [95%CI] 1.28 [1.00-1.63]) and its components kept significance in relation to pMTB-age-R. The pathway analysis revealed overrepresentation of glycine, serine and threonine. Conclusions: pMTB-age is a multidimensional biomarker, which identifies individuals with accelerated vascular aging with high precision and combined with the pathway analysis highlights the role of amino-acids in vascular disease. Therefore, pMTB-age can guide risk stratification and the personalized and timely prevention and treatment tailored to an individual's unique pMTB-age profile.