Abstract
Understanding the links between metabolism, ageing, and age-related phenotypes may clarify the role of ageing in disease onset and improve risk prediction. We conducted a cross-cohort assessment of biological age using broad-spectrum LC–MS metabolomics of 3,686 plasma samples in 2,295 participants, aged 20–89, from the UK Airwave study ( N = 960) and the Irish Longitudinal Study of Ageing ( N = 1,335). The nucleoside N 2 ,N 2 -dimethylguanosine, C-glycosyltryptophan, bile acid glucuronides, and the antioxidant zeta-carotene were associated with chronological age, frailty, and mortality. The noradrenergic metabolite 3-methoxy-4-hydroxyphenylglycol sulphate and the oligosaccharide sialyllactose were strongly associated with both age and mortality. We developed a metabolomic clock that was highly predictive of chronological age ( r = 0.92) in test samples. Metabolomic age acceleration was strongly correlated between study visits ( r > 0.6). Each standard deviation increase in metabolomic age acceleration (~ 5 years) was associated with 43% higher mortality risk, 27% higher risk of mild cognitive impairment, and 10% increased risk of a higher frailty score in fully adjusted models. The metabolites identified here may link ageing and age-related vulnerability and should be further investigated in mechanistic studies. The metabolomic clock has potential for translational applications, including as a prognostic and response marker of generalised age-related disease risk.