1394

Background: Understanding the mechanistic basis of air pollution toxicity is dependent on accurately characterizing both exposure and health along a complete dose-to-response pathway. Untargeted metabolomics, an analysis of small-molecule metabolic phenotypes, may offer improved estimation to complex environmental mixtures. The field remains nascent, however, with questions concerning the coherence and generalizability of findings across study cohorts and analytical platforms. Methods: We conducted a systematic review to 1) summarize recent research of air pollution studies utilizing untargeted metabolomics and; 2) identify gaps in the peer-reviewed literature and opportunities for addressing these gaps in future designs. Using PRISMA guidelines, we screened articles published within MEDLINE and Web of Science between 1/1/05 and 12/21/19. Two reviewers independently screened a total of 1,074 abstracts, with discrepancies resolved by a third reviewer. Results: 22 studies fulfilled eligibility criteria and were included in this review. Over 300 metabolic features were associated with at least one or more air pollutant, with 52 features validated using authentic reference standards. Hypoxanthine, arginine, and histidine were among 12 metabolites consistently exhibiting associations with fine particulate matter and nitrogen dioxide exposure in at least three independent studies. Oxidative stress and inflammation related pathways, including glutathione, leukotriene and vitamin E metabolism, were the most commonly perturbed pathways reported in over 55% of studies. Challenges and gaps exist in metabolite annotation; over 80% of the reported features were not chemically annotated, limiting interpretability and generalizability of the findings. Conclusions: Numerous investigations have demonstrated the feasibility of using untargeted metabolomics as a platform linking exposure to internal dose and biological response. Metabolic perturbations in oxidative stress and acute inflammation pathway were associated with both short- and long- term air pollution exposure. Future directions should focus on validation of these findings via hypothesis-driven protocols and technical advances in metabolic annotation and quantification.