Introduction

Exhaled breath testing is a broad topic that of conjures up images of police officers and forensic alcohol tests, hospitals and anesthesiology, pulmonary function examinations (spirometry) for asthma and emphysema, and oxygen uptake training in sports medicine as illustrated in Fig. 1. The trace-level analytical techniques used for toxicology and the high-tech medical, clinical, and environmental applications are rarely what come to mind first. However, breath has now joined blood and urine as an important diagnostic biofluid for all kinds of analyses for chemical exposures, biological response, metabolism kinetics, health state assessment, multiple time point assessments, and toxicological effects monitoring (Perbellini et al. 2003; Wallace et al. 1987a, b, c). In fact, breath biomarkers have demonstrated particular advantages over blood and urine in that breath sampling provides an essentially inexhaustible supply, does not require medical personnel or privacy, is noninvasive, and does not produce potentially infectious waste such as needles, bandages, and glassware (Wallace et al. 1986; Amann and Smith 2005; Pleil 2008).

Fig. 1
figure 1

Commonly held conceptions of applications for breath monitoring: law enforcement, anesthesia, spirometry, and exercise physiology

Full size image

The human exposome

The interaction of exogenous chemicals with human life processes or “systems biology” has taken on a new and overarching meaning for long-term latency or chronic diseases like cancer and cardiopulmonary disease, autoimmune and neurodegenerative diseases including Parkinson’s, Alzheimer’s, Lupus, and scleroderma, and a variety of developmental disorders such as autism, spina bifida and cerebral palsy (Pleil 2008, and references therein). The conventional wisdom that disease incidence is “bad luck” is losing traction; it is now recognized that the “gene × environment” interaction is likely the cause and that chronic exposures to toxic environmental compounds, whether individually or in concert, instigate long-term processes that can ultimately overwhelm repair and defense mechanisms (Smith and Rappaport 2009; Rappaport and Smith 2010). As such, the human exposome has been proposed as the repository of the chemicals and interaction products that need to be cataloged and interpreted to assess disease processes, and in the long run direct intervention measures (Lioy and Rappaport 2011; Wild 2012; Miller and Jones 2014; Sexton et al. 2004). Originally, human exposome research focused on blood-borne chemicals as the blood contacts all living cells and is considered the “central compartment” (Rappaport et al. 2014). Certain constituents of the exposome are now monitored in other biological fluids for easier access, higher concentrations, and closer proximity to specific organs (e.g., urine for kidney and bladder, lung lavage fluid and breath for lung, cell lines for microbiome and tissues) (Vrijheid et al. 2014; Pleil et al. 2014a, b; Pleil and Stiegel 2013; Amann et al. 2014a, b). The use of breath patterns has even been proposed as a form of ultimate fingerprint like the genome by using portable devices (Risby and Pleil 2014; Stix 2003; Fridman et al. 2015). In the following discussions, the focus is on the breath exposome as a tool for toxicity testing and evaluation.

Gas-phase breath sampling and analysis

Breath biomarkers, pulmonary physiology, and various gas-phase analytics are now emerging as important toxicological research applications (Manolis 1983; Droz and Guillemin 1986; Lippmann and Schlesinger 2000; Miekisch et al. 2004; Cao and Duan 2007). Breath biomarkers have been implemented in a variety of occupational applications as well, for example firefighters’ turnout gear has been tested for safety, fuel systems personnel in the military have been monitored for jet fuel exposures, and industrial workers have been tested for nanoparticle exposures (Pleil et al. 2014a, b; Tu et al. 2004; Pelclova et al. 2015). Although not discussed in detail here, the dental research community has also embraced gas-phase analytical technology to assess and mitigate causes of halitosis both as an adverse condition and as a symptom of underlying periodontal disease (Seemann et al. 2014; Laleman et al. 2014). The medical research community is focusing on exhaled breath patterns to indicate preclinical disease state for lung cancer, COPD, and other diseases (Filipiak et al. 2014; Lázár et al. 2014; Arasaradnam et al. 2016) and has even embraced canine olfaction as an alternative analytical tool (Jezierski et al. 2015).

The aforementioned techniques are all based on the exhaled breath, which can be measured using spirometry, analyzed with various real-time instruments, collected in canisters (illustrated in Fig. 2) or transferred onto adsorbent tubes for subsequent analysis by gas chromatography–mass spectrometry (GC–MS). Typically, these methods focus on volatile organic compounds (VOCs) related to exposure. Specifically, the gas-phase analyses are useful for:

Fig. 2
figure 2

Canister-based breath sampling: non-rebreathing device for multiple-breath collections (upper left); different examples of single-breath collection device (1-l evacuated canister), and simultaneous blood and breath sampling in an exposure study (lower right)

Full size image

  • Pulmonary function/inflammatory response testing.

  • Estimating exposure and dose of VOCs from fuels, consumer products, solvents, disinfection by-products, etc.

  • Developing toxicokinetic models for absorption, distribution, metabolism, and excretion (ADME) of VOCs.

  • Linking toxicity from in vitro test to in vivo effects.

  • Assessing metabolism with isotopically labeled compounds.

Exhaled breath is the best-suited biofluid for direct access to physiological changes in lung function and the “exposure to effect” pathway of the toxicity of exogenous gas-phase chemicals. This is particularly true for any analytical schemes that require rapid response; unlike blood and urine sampling, breath can be sampled essentially in real time (adjacent breaths) and is inexhaustible (Amann et al. 2014a, b; Beauchamp 2011; Pleil et al. 2011; Vaittinen et al. 2013); in fact, these features of “quick and noninvasive” were the original reasons for converting forensic alcohol analysis (for drunk driving) from blood to breath testing (Dubowski 1974).

Vapor, aerosol, and particulate phase breath sampling and analysis

There is an additional arena of analytical procedures that use exhaled breath condensate (EBC) or the subset of exhaled breath aerosol (EBA) fraction of whole breath (Grob et al. 2008; Mutlu et al. 2001; Johnson and Morawska 2009; Almstrand et al. 2008). EBC is collected by passing breath through a chilled tube and freezing out the water vapor, dissolved (polar) organics, and any associated particulates, bacteria, proteins, etc., in the exhaled aerosols and vapors; although researchers have developed various “home-built” devices, self-administered kits are available for routine collection of EBC. In our laboratory, for example, we use hardware from Respiratory Research, Austin TX, and instruments and methods modified to exploit chilled tube collection as from Loccioni Gruppa, Santa Rosario, Italy (Liu et al. 2007; Pleil et al. 2008; Soyer et al. 2006) as illustrated in Fig. 3. EBA is a subset of EBC and is extracted from total EBC, or collected directly using filter materials or particle impingers to capture bacteria, protein fragments, cell fragments, and other particulates directly and exclude gases and vapors (Holmgren et al. 2010; Horvath et al. 2005). EBC is a versatile medium; depending upon the goal of the study, polar volatile compounds can be analyzed by GC–MS, by liquid chromatography–mass spectrometry (LC–MS), or by various large molecules methods such as immunochemistry. EBC and EBA provide additional value in that they give access to the semi- and non-volatile constituents in breath for applications such as:

Fig. 3
figure 3

Collection of EBC and EBA samples: standard “R-tube” kit for EBC (upper left); manual EBC collection with dry ice coolant (top center); automated feedback-controlled EBC and EBA collection using Loccioni device (top right and lower left); Tenax tube collection of polar VOCs from 1-ml EBC samples

Full size image

  • Testing airway pH in liquid EBC for airway viscosity, asthma status, and other preclinical disease indicators.

  • Assaying exhaled biomarkers of pulmonary inflammation (e.g., cytokines) in EBC or EBA with immunochemistry assays.

  • Collecting EBC for noninvasive genotyping of the pulmonary microbiome.

  • Collecting EBC for developing ex vivo cultures of individual strains of pulmonary bacteria for VOCs fingerprinting via GC–MS.

  • Assaying polar organic materials dissolved in water aerosol that are otherwise lost in gas-phase analyses via GC–MS or LC–MS.

  • Assaying partition of non-volatile constituents (e.g., PM 2.5) that indicate previously inhaled particulate matter via LC–MS or immunochemistry.

Investigations of EBC and EBA are relatively new, especially for toxicological purposes. However, they do provide a noninvasive window into the larger non-volatile biomarkers and bioindicators previously thought to be accessible only through blood or urine, and also to polar organic and inorganic compounds difficult to assay using only the gas phase of breath (Hunt 2002).

Toxicological applications for breath analysis

Essentially, all blood, breath, and urine biomonitoring applications link to toxicology in some way. More specifically, the following breath-based topics have emerged as important in the past years and will be the focus of this review:

  • Evaluation of smoking activity and toxic compounds.

  • Interpreting the human exposome using case–control and community toxicity discovery approaches.

  • Development of toxicokinetic models (ADME).

  • Development of adverse outcome pathways (AOPs) and linking in vitro test to in vivo effects.

  • Pulmonary function and inflammatory response.

  • In vivo testing with isotopically labeled tracer compounds.

  • Cell line testing for chemical toxicity prioritization.

As mentioned above, breath biomonitoring has advantages from multiple logistical perspectives. This is especially valuable for the regime of population-based monitoring, wherein one needs series of temporal or spot samples from a large number of people (or cell lines in petri dishes) (Cao and Duan 2006; Lin et al. 2005; Filipiak et al. 2010). Secondly, the use of gas-phase versus liquid-phase analysis provides advantage for high-throughput efficiency, especially with modern real-time instrumentation, wherein analyses can be multiplexed both longitudinally and in parallel with a single analytical platform (Smith and Španel 2005).

With few exceptions, the newer applications using EBC and EBA require some form of liquid-based analysis, which is more time- and labor-intensive; however, the logistical sampling advantages remain as opposed to bronchoalveolar lavage (Jackson et al. 2007). In contrast to blood and urine collection, one can still collect breath EBC and EBA samples without extensively trained personnel, without special infrastructure (clinic, bathroom, coolers) and without creating biohazardous wastes.

The topics presented in the ensuing text are certainly not all-inclusive because the breath research field and the study of toxicology are each enormously complex and expanding rapidly. However, this review provides a relatively broad overview of breath-based applications for toxicology within the context of contemporary use. Due to the breadth of this topic, a best effort is made to draw example references from classical articles and contemporary research; these should not be considered an exhaustive list.

Current applications

Toxicity of smoking

Historically, the earliest breath-based assessments of exogenous compounds toxicity revolved around tobacco smoking and specific compounds including carbon monoxide (CO), nitric oxide (NO), benzene, as well as a variety of other chemicals known to be associated with adverse health outcomes. (Jarvis et al. 1980; Kharitonov et al. 1995; Wester et al. 1986a, b, 1987a, b, c; Gordon et al. 1995; Gordon 1990). To date, smoking status is often used as a major factor in the validation of new technologies for breath analyses as it is well known that toxicity markers of all sorts should be differentiated in smokers versus non-smokers (Santini et al. 2016; Kushch et al. 2008; Basanta et al. 2010; Španel and Smith 1996).

More recently, smoking status has been considered to be more of a confounder for diagnostic breath analysis, rather than the object of toxicity testing. In fact, many studies now deliberately exclude smokers or at least categorize them as a separate cohort to avoid erroneous correlations with disease diagnosis. This factor and other confounders have posed challenges for developing preclinical lung cancer diagnostics as it is difficult to find a large enough “never-smoker” control group of lung cancer patients, and it has been suggested that such patients actually suffer from a different type of cancer than do smokers (Kischkel et al. 2010; Sun et al. 2007).

Exposome: case–control (non-targeted/discovery analysis)

Traditional toxicological and pharmacological practice has generally implemented the targeted case–control study, wherein the initiating event is controlled, and the outcome is observed. Researchers split the subject pool into two groups; one group gets the chemical or drug, the other a placebo, and they wait to see the outcome. For toxicology, the groups are often rats or mice, and in pharmacology, they could be humans as well, once the drug reaches clinical trial status. Similarly, in classical environmental and epidemiological studies, subjects may be sorted by exposure category, say urban versus rural, farmer versus office worker, male versus female, etc., but again, the exposure is targeted a priori for observing the groups, for example automobile exhaust, pesticides, or consumer products, respectively (Rappaport 2012).

The new emerging technology for identifying toxicological effects of chemicals, especially those present in the environment, comprises the concept of “non-targeted” or “discovery” analysis (Pleil and Stiegel 2013; Rager et al. 2016). Here, there is no preconception as to which chemicals are associated with health outcome. The underlying mechanism is still some form of case–control comparison, but now the analytical strategy is to measure all possible chemicals, and try to discover empirically which differentiate the groups. This research revolves around the concept of the human exposome proposed in 2005 by Christopher Wild, Director of the International Agency of Research on Cancer (IARC, Lyon, France). The exposome represents the totality of chemical space in human media, both from environmental and internal metabolism sources and, as such, is thought to embody the chemical record of human environment interaction (Wild 2005; Rappaport 2011; Pleil 2012). An outcome of this model is the newest concept, the environmental-wide association study (EWAS), wherein the standard case–control framework is opened up to mining of bigger databases (Patel et al. 2010; Rappaport et al. 2014).

Exhaled breath, and especially the EBC fraction, has joined blood and urine as a primary biological fluid expressing the human exposome. There are three classes of toxicological stressors that are preferentially accessible in breath:

  • volatile organics (both exogenous and endogenous),

  • inflammatory markers, proteins, and inorganic particulates,

  • pulmonary bacteria, viruses, and their associated toxins.

All of these classes represent some portion of the human exposome and can be used for disease and toxicity diagnoses. For example, patterns discovered in cytokines, polar volatile organics, fine particles, aerosols, and bacterial off-gas products have been used to establish “what is unremarkable” so that outlier effects could also be detected (Stiegel et al. 2015; Hubbard et al. 2009; Benor et al. 2015; Moritz et al. 2015; Greguš 2015).

Toxicokinetics

The first kinetics applications for breath biomarkers in toxicology actually evolved from medical anesthesiology. These approaches implemented absorption, distribution, metabolism, and excretion (ADME) of volatile anesthetics such as isoflurane, enflurane, and halothane (Fiserova-Bergerova and Holaday 1979). Breath analysis tools are still in use for assessing the progression of anesthesia and evaluation of organ function. (Miekisch et al. 2008; Ghimenti et al. 2013; Lirk et al. 2004). Breath ADME approaches have been extended to exploring toxicity of exogenous VOCs by determining the rate constants of uptake and elimination for suites of compounds implicated in human carcinogenicity including chloroform, benzene, methyl tertiary butyl ether, toluene, naphthalene, vinyl chloride, carbon tetrachloride, trichloroethylene, dichlorobenzene (Pellizzari et al. 1992; Pleil and Lindstrom 1997; Pleil et al. 2007). As the rate constants in the first compartment (blood) reflect half-times of 0.5–2 min, only high-frequency collection of breath samples could keep up (Gordon et al. 1988; Chinery and Gleason 1993; Pleil and Lindstrom 1998; Ashley et al. 1996).

The ability to take fast measurements to assess temporal uptake and elimination profiles from controlled breath studies is crucial for interpreting steady-state blood/breath partitions, minimum dose, and multiple-compartment capacities (Buckley et al. 1997; Dobrev et al. 2001; Wallace et al. 1993). All of these parameters are crucial for assessing toxicological properties of important environmental and occupational organics. Furthermore, it was the toxicological applications that fostered the new technologies of real-time mass-based instruments including proton transfer reaction mass spectrometry (PTR-MS), ion mobility spectrometry (IMS), and selected ion flow tube mass spectrometry (SIFT-MS) (Lindinger et al. 1998; Baumbach and Eiceman 1999; Smith and Španel 2005).

In addition to developing kinetic parameters for use in toxicokinetic models, there are other toxicological applications that rely on rapid breath assessments based on a variety of chemical, optical, and mass-based sensors. Some examples are ingestion of volatile foods or substances (Beauchamp et al. 2010; Traucher et al. 1996), breath diagnosis of life-threatening hydrogen cyanide (HCN) toxicity in victims of fire (Stamyr et al. 2015), alcohol and ethylene glycol poisoning (Johanson 2000; Walsh and Macleod 1983; Eder et al. 1998), carbon monoxide exposures (Cunnington and Hormbrey 2002), and toluene abuse (Garriott et al. 1981).

Adverse outcome pathways (AOPs)

The concept of the AOP as a mechanism for high-throughput toxicity testing is relatively new. The general idea is that the toxicological impact of an exogenous chemical exposure is initiated in the human systems biology and then creates a cascade of subsequent events (typically at the molecular or cellular level) that ultimately lead to an adverse preclinical or clinical outcome (Adeleye et al. 2015; Pleil et al. 2015; Tollefsen et al. 2014). Once an AOP is established from in vivo laboratory study, one small section of the chemical reaction chain is reproduced in vitro to represent the full pathway. To date, AOP tests are mostly implemented in multi-well plates and the analyses are based on liquid-phase immunochemistry, although some liquid chromatography work may also be performed (Kavlock et al. 2012; Parker et al. 2000; Inglese et al. 2007). A complementary approach has been proposed termed the aggregate exposure pathway (AEP), which is meant to serve as the precursor analog to the AOP (Teeguarden et al. 2016).

Neither AOP nor AEP approaches have as yet been implemented using breath-based toxicity testing. However, recently, the concept of “cell breath” has been proposed that would complement standard liquid-phase high-throughput toxicity testing with a headspace gas component (Pleil 2016). The use of gas-phase probes has a number of advantages. Analyses could be multiplexed longitudinally as the sample is not consumed as in liquid testing. Gas-phase analyses using real-time instrumentation are extremely fast, typically less than a second or so per data point, and can be highly specific and diagnostic if coupled with high-resolution mass spectrometry (Smith and Španel 2011; Pleil and Isaacs 2016). Certain metabolic pathways that have been characterized in vivo could be probed directly in vitro or ex vivo in a gas-phase analog of liquid-phase AOP testing (Angrish et al. 2015). Finally, the use of cellular respiration analysis provides a platform for direct toxicity testing of gas-phase toxicants, which is currently lacking for high-throughput testing.

Pulmonary function

Breath-based toxicity research does not necessarily require chemical biomarker analysis. One of the simplest methods for assessing the immediate toxic effects of exogenous (airborne) exposures or onset of pulmonary disease flare-up is a change in lung function (Penttinen et al. 2001; Madden et al. 2014). Although not a traditional chemical biomarker, a physiological marker is just as useful in assessing a perturbation (Pleil 2012). Here, the analytical finish is a straightforward measurement of exhaled volume, speed, and in some cases temperature (Lázár et al. 2014). The primary parameters are FEV1, the total volume of breath forcefully expelled in the first second, and FVC, the forced vital capacity or the maximum expiratory volume of a single breath, although a variety of other parameters have been used as well (Morris 1976). Both are very sensitive to pulmonary inflammation from asthma condition, chronic obstructive pulmonary disease (COPD), and air pollution, as well as interactions among them (Harving et al. 1991; Mannino et al. 2003; Scannell et al. 1996; Liu et al. 2009). This can be a large and noticeable effect; a number of studies have shown that chronic urban exposures to traffic and ozone can reduce pulmonary function by approximately 5–25 % (Götschi et al. 2008) and that acute co-exposures to urban levels of ozone and diesel exhaust can synergistically induce reductions in pulmonary function of about 17 % on average (Madden et al. 2014). Certainly, pulmonary function tests alone are not definitive for specific identifying causes; complementary measures of breath-borne inflammatory markers, exogenous compounds, and endogenous volatiles are required to establish toxic effects.

Isotopically labeled tracer compounds

The toxicity of exogenous compounds often depends on the particular human metabolic pathway or microbiome interaction that is affected in different ways based on an individual person’s genotype or phenotype; compounds can experience variability in detoxification for excretion, or their metabolites could become more toxic than the original compounds are alone (Bugrim et al. 2004; Thomas et al. 2001; Savides and Oehme 1983; Clayton et al. 2009). Standard procedures in pharmacology and toxicology generally rely on identifying metabolites in blood or excreted in urine to establish the metabolic paths, and then retrospective analysis is used to interpret genetic polymorphisms or phenotype that influence these outcomes. The ultimate goal is to interpret genomic patterns and epigenetic modifications to identify individuals at risk and provide advice on drug prescription or intervention/avoidance of certain environmental or commercial chemicals (Shimada et al. 1994; Easton et al. 2007; Baccarelli and Bollati 2009).

Exhaled breath and EBC tests have become an important feature in the realm of individualized or personalized medicine and toxicity testing (Modak 2010, 2013; Baraldi et al. 2009; Fens et al. 2009). However, rather than taking complex genetic and metabolite measurements, new breath tests now rely on specific probe molecules and the subsequent appearance of exhaled compounds like carbon dioxide (CO2), oxygen (O2), or hydrogen (H2). Most of the novel tests employ organic compounds with isotopically labeled carbon-13 (13C), oxygen-18 (18O), or deuterium (2H), wherein a more complex organic probe molecule with these substitutions is injected into or ingested by the patient (Schellekens et al. 2011). Exhaled breath is then monitored using mass spectrometry or optical instrumentation for perturbations in the normal isotopic abundances of water or CO2 that would indicate metabolism of the probe (Smith et al. 2002; Crosson et al. 2002; Logan et al. 1991). Of particular note in this category is the helicobacter pylori CO2 breath test based on ingestion of 13C-labeled urea, which has become a standard noninvasive alternative method for identifying an underlying cause of gastric and duodenal ulcers (Cutler et al. 1995).

From a chemical toxicological perspective, an important development for isotopic breath tests is personalized testing of chemotherapeutic agents (Polk and Peek 2010). Chemotherapy is designed to be more cytotoxic for cancer than human cells; however, there are individual genotypes for which this distinction is not sufficient. In fact, in some individuals, the treatment could prove fatal. Rather than inferring metabolism paths from detailed genetic testing, the patient is given a benign surrogate that is isotopically labeled, and then, the metabolism pathway is directly monitored via an exhaled gas, typically 13C-labeled CO2 (Modak 2007). For example, cancer patients are treated with 13C-labeled methionine to assess mitochondrial toxicity of antiretroviral drugs (Milazzo et al. 2005), or with [13C]-dextromethorphan to assess breast cancer therapy with tamoxifen (Opdam et al. 2015). A variety of chemical probes including [13C]-dextromethorphan, [13C]-pantoprazole, and [13C]-uracil have been used to evaluate activity of CYP2D6, CYP2C19, and DPD enzymes, which are important in about 35 % of FDA-approved medications (Modak 2009). As such, a simple 1-hour spot test of exhaled 13C-labeled carbon dioxide can establish specific enzymatic pathways without the effort and cost of genotyping or measuring a suite of metabolites.

In vitro to in vivo linkage; cell line toxicity prioritization tests

As mentioned above in the AOPs discussion, the implementation of cellular level respiration analysis has offered a new opportunity for linking chemical toxicity at the systemic (in vivo) level to a more pragmatic set of test performed in cell line bioreactors (Astashkina et al. 2012; Bhogal et al. 2005). This approach is currently relegated to liquid analysis (Dix et al. 2007; Soldatow et al. 2013); the ultimate breath (or cell breath) applications are still in their infancy, but are currently under laboratory development as a complementary technique.

The current focus for “cell breath” research is to measure volatile emissions from bacterial mono-cultures to identify patterns of VOCs in vitro and develop a “fingerprint” library (Shestivska et al. 2015; Zhu et al. 2013; Sethi et al. 2013). Subsequently, measurements of human exhaled breath could be explored for such patterns to diagnose infectious state and monitor treatment (Risby and Solga 2006; Barker et al. 2006; Christ-Crain and Müller 2007). These investigations do not have direct toxicological implications; however, they have provided the underlying technology for maintaining viable cell cultures and measuring gas-phase chemicals. The ultimate goal for toxicity testing is to use this technology to probe for changes in the baseline respiration patterns of cell cultures (bacterial, animal, human) upon administration of potentially toxic substances. The magnitude of the observed perturbations in vitro could be compared to the applied dose to prioritize chemicals for more detailed in vivo toxicity testing to complement current high-throughput tests at the molecular level (Judson et al. 2010; Tice et al. 2013).

Future applications

A number of future applications for breath biomarkers in toxicology have been alluded to above. These mostly involve novel collection methods, implementation of aerosol diagnostics, and new technologies for rapid measurement of cellular-based media. Briefly, they are encapsulated by these topics:

  • Breath spot analysis

  • Miniaturized bioreactors

  • Microbiome interaction

  • Crowd breath

Breath spot

The clinical/medical communities have long embraced the concept of the dried blood spot (DBS), or “Guthrie spot,” as a convenient medium for specific screening tests for childhood metabolic diseases (Guthrie and Susi 1963; Garrick et al. 1973; Crossle et al. 1979). More recently, archived and new DBS media have been tested for metals and chemicals in newborns that could be indicative of prenatal exposures or used to monitor drug dose (Spliethoff et al. 2008; Funk et al. 2015; Spooner et al. 2009). Neonatal DBS have also been used to assess inflammatory markers and growth factors and are considered new tools for case–control disease studies (Skogstrand et al. 2005). Based on these concepts, the next logical step is to implement the analog “dried breath spot” or DBrS (to differentiate it from DBS). The concept is to have subjects exhale through a standard disposable respiratory filter as used in spirometry, which would capture aerosols, bacteria, cell fragments, and large molecules carried by vapor. Upon air-drying, these could easily be shipped to a laboratory for extraction and analysis of captured analytes. For now, this is speculative, but could well become a very simple toxicological assessment tool especially for population-based studies where simplicity in the field is paramount.

Miniaturized bioreactor

Growing and maintaining viable cell lines ex vivo is a difficult endeavor. Commercial bioreactors used for cell and tissue cultures are fairly large (liter to m3 volumes), and even the term “mini-bioreactor” refers to volumes of 100 ml or so (Marks 2003; Kumar et al. 2004). As cells and tissues require constant gas and nutrient flows, it is possible to monitor processes via online analysis (Matz and Lennemann 1996; Luchner et al. 2012). Such large-scale processes, however, are not amenable to toxicity testing, and so standard AOP-based methods are implemented in microwell arrays (typically 96-well plates) and condensed-phase (liquid) analysis. The challenges for implementing microscale gas analysis are sensitivity and speed as volumes encountered therein are of the order of microliters (μl) and current systems for gas analyses need milliliter (ml) volumes. However, the advances in chemical array sensors, mid-infrared laser spectrometers, and bench-top time of flight mass spectrometers (ToF–MS) hold promise in approaching the needs of real-time gas-phase analysis at the microscale, and some progress is being made for these applications (Baranska et al. 2015; Filipiak et al. 2015).

Microbiome interaction

Human cells are not unique in experiencing toxic effects from exogenous chemicals. The human microbiome represents about 10 times as many bacterial cells as those comprising an adult person (Sleator 2010). As such, the human exposome can be affected by direct bacterial metabolism of toxic chemicals, especially in the gut and lung, where exogenous exposures are most likely (Possemiers et al. 2009). Recent research has shown that certain toxicity pathways are activated by the microbiome, whereas human cells would otherwise be unaffected. A well-known example is the gut bacterial metabolic product l-carnitine, linked to cardiovascular disease, which occurs only in subjects who are regular meat eaters, but not vegetarians (Ussher et al. 2013). Diagnostic pattern analysis of cell breath, both in vivo and ex vivo, of the human microbiome as a source of toxic metabolism has the potential for developing antibiotic intervention treatments.

Crowd breath

At the opposite end of the philosophical spectrum from in vitro cellular respiration of monocultures is a new concept for using breath biomarkers in toxicology. Termed “crowd breath,” this is a non-targeted (discovery) approach, wherein the air surrounding crowds of people is repeatedly analyzed to develop a time course of as many organic molecules as possible (Williams and Pleil 2016). These samples represent a mixture of sources from the surrounding environment, commercial and personal products, local industrial emissions, as well as from human breath. To separate out the aggregate breath source, compounds are sorted by co-variance with known breath emissions of acetone, isoprene, and CO2. This technique was applied to monitoring people inside a movie theater, and the results were annotated with specific scenes that could arouse emotion, e.g., fear, humor, suspense, stress (Williams et al. 2016). The ultimate value for toxicological research would be linking microenvironmental contaminants back to exhaled breath to assess group exposures, and possibly health risk levels. Another potential use would be to evaluate the toxicological potential of non-chemical stressors (emotion) via endogenous breath markers emitted by groups.

Summary

Environmental health and clinical practice have embraced many of the concepts of breath biomarker testing for practical reasons: Breath testing is noninvasive and strategically straightforward, and samples are essentially limitless. These qualities have made breath a valuable complement to standard toxicological investigations using blood and urine samples. Breath has also carved a niche as a simple “stand-alone” sampling medium, especially when rapid sampling frequency is needed in the case of toxicokinetic studies. Furthermore, human breath investigations are no longer relegated to just the gas phase; the condensate and aerosol fractions of exhalations are now routinely used to gain access to protein fragments, microbiological media, dissolved inorganics, and highly polar compounds. Breath samples give immediate access to monitoring specific metabolism pathways indicative of toxicity by the use of pattern “fingerprinting” or isotopically labels probes. And finally, the broad field of breath analysis is now expanding to the arena of “cell breath” where individual strains of cells, bacteria, or tissue samples are explored, all the way to “crowd breath” where groups of individuals are tested for more generic toxicological responses.