Background: The chemical exposome includes exposure to numerous environmental and endogenous molecules, many of which have been linked to reproductive outcomes due to their endocrine-disrupting properties. As several breast cancer risk factors, including age and parity, are related to reproduction, it is imperative to investigate the interplay between such factors and the chemical exposome prior to conducting large scale exposome-based breast cancer studies.

Objective: This pilot study aimed to provide an overview of the chemical exposome in plasma samples from healthy women and identify associations between environmental exposures and three risk factors for breast cancer: age, parity, and age at menarche.

Material and methods: Plasma samples (n = 161), were selected based on reproductive history from 100 women participating in the Northern Sweden Health and Disease Study, between 1987 and 2006. Samples were analyzed by liquid chromatography high-resolution mass spectrometry (LC-HRMS) for 77 priority target analytes including contaminants and hormones, with simultaneous untargeted profiling of the chemical exposome and metabolome. Linear mixed effects models were applied to test associations between risk factors and chemical levels.

Results: Fifty-five target analytes were detected in at least one individual and over 94,000 untargeted features were detected across all samples. Among untargeted features, 430 could be annotated and were broadly classified as environmental (246), endogenous (167) or ambiguous (17). Applying mixed effect models to features detected in at least 70% of the samples (16,778), we found seven targeted analytes (including caffeine and various per- and poly-fluoroalkyl substances) and 38 untargeted features, positively associated with age. The directionality of these associations reversed for parity, decreasing with increasing births. Seven separate targeted analytes were associated with age at menarche.

Significance: This study demonstrates how a comprehensive chemical exposome approach can be used to inform future research prioritization regarding associations between known and unknown substances, reproduction, and breast cancer risk.

For comprehensive chemical exposomics in blood, analytical workflows are evolving through advances in sample preparation and instrumental methods. We hypothesized that gas chromatography-high-resolution mass spectrometry (GC-HRMS) workflows could be enhanced by minimizing lipid coextractives, thereby enabling larger injection volumes and lower matrix interference for improved target sensitivity and nontarget molecular discovery. A simple protocol was developed for small plasma volumes (100-200 μL) by using isohexane (H) to extract supernatants of acetonitrile-plasma (A-P). The HA-P method was quantitative for a wide range of hydrophobic multiclass target analytes (i.e., log Kow > 3.0), and the extracts were free of major lipids, thereby enabling robust large-volume injections (LVIs; 25 μL) in long sequences (60-70 h, 70-80 injections) to a GC-Orbitrap HRMS. Without lipid removal, LVI was counterproductive because method sensitivity suffered from the abundant matrix signal, resulting in low ion injection times to the Orbitrap. The median method quantification limit was 0.09 ng/mL (range 0.005-4.83 ng/mL), and good accuracy was shown for a certified reference serum. Applying the method to plasma from a Swedish cohort (n = 32; 100 μL), 51 of 103 target analytes were detected. Simultaneous nontarget analysis resulted in 112 structural annotations (12.8% annotation rate), and Level 1 identification was achieved for 7 of 8 substances in follow-up confirmations. The HA-P method is potentially scalable for application in cohort studies and is also compatible with many liquid-chromatography-based exposomics workflows.

Chemical exposomes can now be comprehensively measured in human blood, but knowledge of their variability and longitudinal stability is required for robust application in cohort studies. Here, we applied high-resolution chemical exposomics to plasma of 46 adults, each sampled 6 times over 2 years in a multiomic cohort, resulting in 276 individual exposomes. In addition to quantitative analysis of 83 priority target analytes, we discovered and semiquantified substances that have rarely or never been reported in humans, including personal care products, pesticide transformation products, and polymer additives. Hierarchical cluster analysis for 519 confidently annotated substances revealed unique and distinctive coexposures, including clustered pesticides, poly(ethylene glycols), chlorinated phenols, or natural substances from tea and coffee; interactive heatmaps were publicly deposited to support open exploration of the complex (meta)data. Intraclass correlation coefficients (ICC) for all annotated substances demonstrated the relatively low stability of the exposome compared to that of proteome, microbiome, and endogenous small molecules. Implications are that the chemical exposome must be measured more frequently than other omics in longitudinal studies and four longitudinal exposure types are defined that can be considered in study design. In this small cohort, mixed-effect models nevertheless revealed significant associations between testosterone and perfluoroalkyl substances, demonstrating great potential for longitudinal exposomics in precision health research.

The chemical exposome is the cumulative sum of environmental chemical exposures over an individual’s lifespan, including pollution, dietary substances, and the metabolic products of gut microbiota. Specific environmental chemicals are known to influence health and disease risk, but overall knowledge has advanced too slowly due to a previous focus on a limited number of targeted chemicals and the large volumes of blood required for sensitive analyses. Measurement of the chemical exposome in blood is strategic due to the simultaneous presence of dietary substances, drugs and environmental contaminants, as well as endogenous molecules whose profiles may be impacted by such exposures. To facilitate routine chemical exposomics in health studies, trace analytical methods for small volumes of blood are needed that can quantify a wide range of multiclass target analytes, while also discovering unexpected chemicals in a complex matrix dominated by endogenous molecules. Recognizing that our environment is dynamic, and that human susceptibility to disease changes over the life course, the exposome has always been envisaged as a parameter requiring repeated measures over time. However, fundamental questions remain on the longitudinal stability of the chemical exposome, including its relative stability compared to other omic profiles routinely measured in health studies today. 

The foundation of this doctoral thesis is a chemical exposomics analytical workflow, involving: a sample preparation method for ≤ 200 µL of human blood plasma that minimizes endogenous interferences, a combined targeted/untargeted liquid chromatography-high resolution mass spectrometry (LC-HRMS) acquisition, and a data processing workflow with open science tools to discover and annotate hundreds of small molecules in large datasets. Workflow applications in Swedish cohorts are also demonstrated, including the first cohort-scale application of longitudinal exposomics in blood. 

In Paper I, the selective removal of high abundance phospholipids from plasma enabled the sensitive and quantitative multiclass targeted analysis of 83 priority analytes. In untargeted acquisition, 109 and 28% more non-phospholipid molecular features in positive and negative mode, respectively, were detected with the new method compared to a control method without phospholipid removal. In Paper II, the same method was applied to a longitudinal multiomic wellness cohort, resulting in 519 confident molecular annotations, including novel exposures and correlated co-exposures (i.e. mixtures). A data resource containing the longitudinal stabilities for hundreds of environmental molecules in blood over 2 years revealed that the chemical exposome has low stability compared to other omic profiles in the same individuals, thereby urging repeated exposome measurement in future studies. In Paper III the workflow was applied to plasma from 100 healthy women in a pilot study for exposome and breast cancer, revealing associations between known and unknown chemicals and breast cancer risk factors. Overall, this thesis provides a powerful workflow for plasma chemical exposomics that can be applied at cohort-scale, and the combined products of this thesis will contribute to the design and execution of future exposome studies.

Background Exposure to perfluoroalkyl substances may affect offspring immune development and thereby increase risk of childhood asthma, but the underlying mechanisms and asthma phenotype affected by such exposure is unknown.

Methods In the Danish COPSAC2010 cohort of 738 unselected pregnant women and their children plasma PFOS and PFOA concentrations were semi-quantified by untargeted metabolomics analyses and calibrated using a targeted pipeline in mothers (gestation week 24 and 1 week postpartum) and children (age 1/2 , 11/2 and 6 years). We examined associations between pregnancy and childhood PFOS and PFOA exposure and childhood infections, asthma, allergic sensitization, atopic dermatitis, and lung function measures, and studied potential mechanisms by integrating data on systemic low-grade inflammation (hs-CRP), functional immune responses, and epigenetics.

Findings Higher maternal PFOS and PFOA exposure during pregnancy showed association with a non-atopic asthma phenotype by age 6, a protection against sensitization, and no association with atopic asthma or lung function, or atopic dermatitis. The effect was primarily driven by prenatal exposure. There was no association with infection proneness, low-grade inflammation, altered immune responses or epigenetic changes.

Interpretations Prenatal exposure to PFOS and PFOA, but not childhood exposure, specifically increased the risk of low prevalent non-atopic asthma, whereas there was no effect on atopic asthma, lung function, or atopic dermatitis. 

Chemical exposomics in human plasma wasenhanced by an optimizedphospholipid removal step that increased targeted method sensitivitywhile also revealing >13,000 new molecular features by LC-HRMSnon-targetedacquisition. The challenge of chemical exposomics in human plasmais the 1000-foldconcentration gap between endogenous substances and environmentalpollutants. Phospholipids are the major endogenous small moleculesin plasma, thus we validated a chemical exposomics protocol with anoptimized phospholipid-removal step prior to targeted and non-targetedliquid chromatography high-resolution mass spectrometry. Increasedinjection volume with negligible matrix effect permitted sensitivemulticlass targeted analysis of 77 priority analytes; median MLOQ= 0.05 ng/mL for 200 & mu;L plasma.In non-targeted acquisition, mean total signal intensities of non-phospholipidswere enhanced 6-fold in positive (max 28-fold) and 4-fold in negativemode (max 58-fold) compared to a control method without phospholipidremoval. Moreover, 109 and 28% more non-phospholipid molecular featureswere detected by exposomics in positive and negative mode, respectively,allowing new substances to be annotated that were non-detectable withoutphospholipid removal. In individual adult plasma (100 & mu;L, n = 34), 28 analytes were detected and quantified among10 chemical classes, and quantitation of per- and polyfluoroalkylsubstances (PFAS) was externally validated by independent targetedanalysis. Retrospective discovery and semi-quantification of PFAS-precursorswas demonstrated, and widespread fenuron exposure is reported in plasmafor the first time. The new exposomics method is complementary tometabolomics protocols, relies on open science resources, and canbe scaled to support large studies of the exposome.

Studies indicate that phthalates are endocrine disruptors affecting reproductive health. One of the most commonly used phthalates, di-n-butyl phthalate (DBP), has been linked with adverse reproductive health outcomes in men, but the mechanisms behind these effects are still poorly understood. Here, adult male mice were orally exposed to DBP (10 or 100 mg/kg/day) for five weeks, and the testis and adrenal glands were collected one week after the last dose, to examine more persistent effects. Quantification of testosterone, androstenedione, progesterone and corticosterone concentrations by liquid chromatography-mass spectrometry showed that testicular testosterone was significantly decreased in both DBP treatment groups, whereas the other steroids were not significantly altered. Western blot analysis of testis revealed that DBP exposure increased the levels of the steroidogenic enzymes CYP11A1, HSD3 beta 2, and CYP17A1, the oxidative stress marker nitrotyrosine, and the luteinizing hormone receptor (LHR). The analysis further demonstrated increased levels of the germ cell marker DAZL, the Sertoli cell markers vimentin and SOX9, and the Leydig cell marker SULT1E1. Overall, the present work provides more mechanistic understanding of how adult DBP exposure can induce effects on the male reproductive system by affecting several key cells and proteins important for testosterone biosynthesis and spermatogenesis, and for the first time shows that these effects persist at least one week after the last dose. It also demonstrates impairment of testosterone biosynthesis at a lower dose than previously reported.

Background Perfluoroalkyl substances PFOS and PFOA are persistent and bioaccumulative exogenous chemicals in the human body with a range of suspected negative health effects. It is hypothesised that exposure during prenatal and early postnatal life might have particularly detrimental effects on intrauterine and childhood growth. In a Dan-ish longitudinal mother-child cohort we investigate effect of PFOS and PFOA in pregnancy and infancy on intrauter-ine and childhood growth and anthropometry.

Methods COPSAC2010 is an ongoing population based mother-child cohort of 738 pregnant women and their children followed from 24 week gestation with longitudinal deep clinical phenotyping until age 10 years. In this observational cohort sub study plasma PFOS and PFOA concentrations were semi-quantified by untargeted metabolomics in the mothers at week 24 and 1 week postpartum and in the children at ages 6 and 18 months and calibrated using a targeted pipe-line. We examined associations to intrauterine and childhood growth and anthropometry, including interactions with child sex. Untargeted and targeted blood metabolomics profiles were integrated to investigate underlying mechanisms.

Findings Pregnancy plasma PFOA concentrations were associated with lower birth size -0.19 [-0.33; -0.05] BMI z-score per 1-ng/mL and increased childhood height (z-scored) at age 6: 0.18 [0.05; 0.31], but there was no association between childs' own infancy plasma PFOA concentration and height. Pregnancy plasma PFOS concentrations were also associated with lower birth BMI (-0.04 [-0.08; -0.01]), but in childhood pregnancy plasma PFOS con-centration interacted with child sex on BMI and fat percentage at 6 years with negative associations in girls and positive in boys. The effect of maternal plasma PFOS concentration on lower girl BMI was borderline mediated through increasing child plasma lactosyl-ceramide levels (p-mediation=0.08). Similarly the effect of maternal plasma PFOS concentration on higher boy fat percentage was borderline mediated through increasing child plasma lactosyl-ceramide levels (p-mediation=0.07). Infancy concentrations of plasma PFOS associated with lower height in childhood, -0.06 z-score at age 6 [-0.19; -0.03].

Interpretation Higher PFOS and PFOA plasma concentrations during pregnancy had detrimental effects on fetal growth. The effects on childhood growth were not similar as PFOA increased child height, opposite of PFOS in mul-tipollutant models suggesting a differing fetal programming effect. Sex specific growth effects were borderline medi-ated through an altered lactosyl-ceramide metabolism, proposing a possible mechanism of PFOS that has long-lasting health consequences in this observational study.

A combination of high-resolution mass spectrometry and computational molecular characterization techniques can structurally annotate up to 17% of organic compounds in fine particulate matter in highly polluted air sampled in the Maldives. Fine particulate-matter is an important component of air pollution that impacts health and climate, and which delivers anthropogenic contaminants to remote global regions. The complex composition of organic molecules in atmospheric particulates is poorly constrained, but has important implications for understanding pollutant sources, climate-aerosol interactions, and health risks of air pollution exposure. Here, comprehensive nontarget high-resolution mass spectrometry was combined with in silico structural prediction to achieve greater molecular-level insight for fine particulate samples (n = 40) collected at a remote receptor site in the Maldives during January to April 2018. Spectral database matching identified 0.5% of 60,030 molecular features observed, while a conservative computational workflow enabled structural annotation of 17% of organic structures among the remaining molecular dark matter. Compared to clean air from the southern Indian Ocean, molecular structures from highly-polluted regions were dominated by organic nitrogen compounds, many with computed physicochemical properties of high toxicological and climate relevance. We conclude that combining nontarget analysis with computational mass spectrometry can advance molecular-level understanding of the sources and impacts of polluted air.

The concept of the exposome was introduced over 15 years ago to reflect the important role that the environment exerts on health and disease. While originally viewed as a call-to-arms to develop more comprehensive exposure assessment methods applicable at the individual level and throughout the life course, the scope of the exposome has now expanded to include the associated biological response. In order to explore these concepts, a workshop was hosted by the Gunma University Initiative for Advanced Research (GIAR, Japan) to discuss the scope of exposomics from an international and multidisciplinary perspective. This Global Perspective is a summary of the discussions with emphasis on (1) top-down, bottom-up, and functional approaches to exposomics, (2) the need for integration and standardization of LC- and GC-based high-resolution mass spectrometry methods for untargeted exposome analyses, (3) the design of an exposomics study, (4) the requirement for open science workflows including mass spectral libraries and public databases, (5) the necessity for large investments in mass spectrometry infrastructure in order to sequence the exposome, and (6) the role of the exposome in precision medicine and nutrition to create personalized environmental exposure profiles. Recommendations are made on key issues to encourage continued advancement and cooperation in exposomics.