An Investigation into the Temporal Reproducibility of Tryptophan Metabolite Networks Among Healthy Adolescents

Abstract

Tryptophan and its bioactive metabolites are associated with health conditions such as systemic inflammation, cardiometabolic diseases, and neurodegenerative disorders. There are dynamic interactions among metabolites of tryptophan. The interactions between metabolites, particularly those that are strong and temporally reproducible could be of pathophysiological relevance. Using a targeted metabolomics approach, the concentration levels of tryptophan and 18 of its metabolites across multiple pathways was quantified in 24-hours urine samples at 2 time-points, age 17\u2009years (baseline) and 18\u2009years (follow-up) from 132 (52% female) apparently healthy adolescent participants of the DOrtmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study. In sex-specific analyses, we applied 2 network approaches, the Gaussian graphical model and Bayesian network to (1) explore the network structure for both time-points, (2) retrieve strongly related metabolites, and (3) determine whether the strongly related metabolites were temporally reproducible. Independent of selected covariates, the 2 network approaches revealed 5 associations that were strong and temporally reproducible. These were novel relationships, between kynurenic acid and indole-3-acetic acid in females and between kynurenic acid and xanthurenic acid in males, as well as known relationships between kynurenine and 3-hydroxykynurenine, and between 3-hydroxykynurenine and 3-hydroxyanthranilic acid in females and between tryptophan and kynurenine in males. Overall, this epidemiological study using network-based approaches shed new light into tryptophan metabolism, particularly the interaction of host and microbial metabolites. The 5 observed relationships suggested the existence of a temporally stable pattern of tryptophan and 6 metabolites in healthy adolescent, which could be further investigated in search of fingerprints of specific physiological states. The metabolites in these relationships may represent a multi-biomarker panel that could be informative for health outcomes.