Mid- and long-term correlations of plasma metabolite concentrations measured by a targeted metabolomics approach

Introduction Plasma metabolites measured by metabolomics techniques have been shown to be associated with chronic disease risk in epidemiological studies. However, in most prospective studies metabolomic profiles can only be obtained at a single time point and data on intra-individual variation in metabolite levels over time are sparse. Objectives Here, we evaluated the intra-individual variation in the concentrations of 177 metabolites over time using repeat blood samples of 104 adults (50% female). Methods Blood samples were obtained at a baseline visit between 1994 and 1998 and during two further examinations 14 and 15 years later. Plasma metabolite levels were quantified by tandem mass spectrometry with the Meta-DisIDQ (TM) p180 Kit. Intra-individual variation was assessed by Spearman's correlation coefficients (rho). Results Mid-term correlations over 1 year were good (rho >= 0.7) for 5.1% and reasonable (rho >= 0.4 < 0.7) for 61.0% of the metabolites, while long-term correlations over 15 years were good for 2.8% and reasonable for 27.1%. The strongest mid-term correlations between metabolite concentrations were observed for the acylcarnitine C-3-OH (0.72) and metabolites from the amino acid/biogenic amine groups, i.e. creatinine (0.83), proline (0.79), lysine (0.77), isoleucine (0.76), and ornithine (0.74). C-3-OH (0.78) as well as several amino acids/biogenic amines, i.e. ornithine (0.76), sarcosine (0.76), lysine (0.75), spermine (0.73), and glutamine (0.69) showed the strongest long-term correlations. Conclusions The biological reproducibility of plasma concentrations of a majority of metabolites occurs reasonable over 1 year. In contrast, concentrations of many metabolites seem to be affected by substantial intra-individual variation over 15 years.