A State-of-the-Science Review on High-Resolution Metabolomics Application in Air Pollution Health Research: Current Progress, Analytical Challenges, and Recommendations for Future Direction

Abstract

Background: Understanding the mechanistic basis of air pollution toxicity is dependent on accurately characterizing both exposure and biological responses. Untargeted metabolomics, an analysis of small-molecule metabolic phenotypes, may offer improved estimation of exposures and corresponding health responses to complex environmental mixtures such as air pollution. The field remains nascent, however, with questions concerning the coherence and generalizability of findings across studies, study designs and analytical platforms. Objectives: We aimed to review the state of air pollution research from studies using untargeted high-resolution metabolomics (HRM), highlight the areas of concordance and dissimilarity in methodological approaches and reported findings, and discuss a path forward for future use of this analytical platform in air pollution research. Methods: We conducted a state-of-the-science review to a ) summarize recent research of air pollution studies using untargeted metabolomics and b ) identify gaps in the peer-reviewed literature and opportunities for addressing these gaps in future designs. We screened articles published within Pubmed and Web of Science between 1 January 2005 and 31 March 2022. Two reviewers independently screened 2,065 abstracts, with discrepancies resolved by a third reviewer. Results: We identified 47 articles that applied untargeted metabolomics on serum, plasma, whole blood, urine, saliva, or other biospecimens to investigate the impact of air pollution exposures on the human metabolome. Eight hundred sixteen unique features confirmed with level-1 or -2 evidence were reported to be associated with at least one or more air pollutants. Hypoxanthine, histidine, serine, aspartate, and glutamate were among the 35 metabolites consistently exhibiting associations with multiple air pollutants in at least 5 independent studies. Oxidative stress and inflammation-related pathways—including glycerophospholipid metabolism, pyrimidine metabolism, methionine and cysteine metabolism, tyrosine metabolism, and tryptophan metabolism—were the most commonly perturbed pathways reported in of studies. More than 80% of the reported features were not chemically annotated, limiting the 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. Our review of the 47 existing untargeted HRM–air pollution studies points to an underlying coherence and consistency across a range of sample analytical quantitation methods, extraction algorithms, and statistical modeling approaches. Future directions should focus on validation of these findings via hypothesis-driven protocols and technical advances in metabolic annotation and quantification...

Metabolic Perturbations Associated with an Exposure Mixture of Per- and Polyfluoroalkyl Substances in the Atlanta African American Maternal-Child Cohort

“…Metabolomics is the systemic study of all metabolites associated with exogenous exposure and endogenous processes, and has emerged as an innovative and powerful analytical platform in environmental epidemiology. , Recent investigations demonstrate the applicability of using metabolomics as a central platform to link human exposure with internal dose and biological response. − Specifically, an increase in circulating concentrations of PFAS has been consistently associated with several endocrine disruption- and oxidative stress-related pathways. ,,− In one of our recent metabolome-wide association studies (MWAS), we also identified and verified several biomechanisms and biomarkers mediating the association between serum PFAS concentrations and fetal growth restriction . Moreover, in two separate analyses within the same pregnant population, we found an inverse association between prenatal exposure to PFAS, modeled as single chemicals and a mixture, with fetal growth measures. , …”

“…Further, our results may be subject to recall bias for specific confounders (e.g., substance use), where participants may be more likely to under-report usage. Lastly, given the purpose of this proof-of-concept analysis is to demonstrate the feasibility and utility of applying mixture analysis in metabolomics application, we used unadjusted p -values for the pathway enrichment analysis due to insufficient power, which is common in environmental metabolomics studies . Despite the use of a less conservative significance cutoff, we identified pathways and metabolites similar to other PFAS MWAS that corrected for multiple comparisons. , …”

“…Liquid chromatography-high resolution mass spectrometry (LC-HRMS) was operated in full-scan mode at 120k resolution to cover a range of mass-to-charge ratio ( m / z ) from 85 to 1275, which includes features with a Level-1, -2, -3, or -4 confidence. Briefly, the evidence for chemical identifies confirmed with Level-1 confidence includes comparison to an authentic standard by mass spectrum and retention time; Level-2 confidence includes comparison of the mass spectrum to the library spectrum data where the spectrum structure match is unambiguous; Level-3 confidence includes tentative matches proposed with insufficient information for only one exact structure; and, Level-4 confidence includes only one reported unequivocal molecular formula. , Two internal standards, which include pooled serum and standard reference material for human metabolites in plasma (NIST SRM 1950), were added at the beginning and end of each batch of 20 samples for quality control and standardization. , The raw instrument files generated from the untargeted high-resolution metabolomics profiling were first converted to mzML format then metabolic features were extracted and aligned using apLCMS with modification of xMSanalyzer. , This improved data quality control and reduced intra- and interbatch effects. Before statistical analyses, additional quality control measures were performed to optimize the data quality.…”

Multiomic Signatures of Traffic-Related Air Pollution in London Reveal Potential Short-Term Perturbations in Gut Microbiome-Related Pathways

“…All of the annotated features were associated with NO 2 exposure. Of these 14 compounds, 4 compounds (caffeine, phenylalanine, 2 acyl-carnitines), were also found in our previous work . Our new analysis revealed higher levels of cortisol and cortisone, indoleactic acid (IAA), and indole-3-propionic acid (IPA) associated with TRAP exposure, as well as lower levels of isoleucine (SI Table 4).…”

“…Such studies have shown that a period of physical activity has beneficial health effects, − although these benefits may be offset by the impact of TRAP exposure on cardiopulmonary and lung function, with increases in inflammatory blood cells observed. − However, the underlying biological mechanisms have not been fully elucidated, and classical epidemiological methods cannot accurately ascertain the short-term impacts of exposures to TRAP, largely due to the absence of robust and specific biomarkers. The advent of omics-based high-throughput technologies allows for molecular changes and biological pathways associated with TRAP exposure to be identified at a cellular level, and underlying pathomolecular mechanisms to be mapped more precisely. − Metabolomics and transcriptomics provide powerful analytical methods to understand the molecular and biochemical pathways triggering a systemic response that can be observed in the peripheral blood. , Metabolomics systematically investigates metabolites, such as amino acids, fatty acids, and lipids, and their impact on oxidative stress and inflammation pathways, in turn affecting human health and disease risk. , Untargeted metabolomics offers a valuable approach to identifying and quantifying the effect of TRAP exposure on the blood metabolome. ,, Transcriptomics allows the identification of gene expressions that are differentially induced and can be used as exposure biomarkers . Combining metabolomics and transcriptomics obtained from the same samples offers a more comprehensive approach to better understand the biological response and the pathways affected by exposure to TRAP, as it will extend the coverage of the molecules assayed and include a large variety of both endogenous and exogenous molecules.…”