Breath Analysis by Mass Spectrometry-Based Technologies for Biomonitoring Environmental Exposures

Abstract

Environmental exposures throughout the life of the subjects (exposome) could have a negative effect on their health outcomes. From this perspective, analysis of volatile organic compounds (VOCs) in human exhaled breath is emerging as a non-invasive tool to identify and check exposure to harmful agents. Breath analysis is also a helpful technique for human metabolism assessment, which allows for examining the impact of environmental exposures on organisms (biomonitoring). In this paper, a comprehensive review has been carried out to assess the use of breath analysis by mass spectrometry-based technologies for monitoring environmental exposures. Records of the last 20 years from three databases (PubMed/Medline, Scopus, and Web of Science) have been evaluated independently by two reviewers. A total of 38 studies fulfilled the criteria for eligibility. It has been compiled information about environmental exposures that have been monitored by breath analysis using mass spectrometry-based analytical platforms, as well as the most commonly used protocols of breath sampling, analytical techniques, and statistical methods. In addition, special emphasis has been placed on the huge range of VOCs selected as potential markers of environmental exposures. Despite the potential of breath analysis for monitoring human exposure, further research is needed to identify useful markers to establish it as a routine tool.

1. Introduction

Human exhaled breath is a complex mixture that provides information about the organism, since its composition can be influenced by the health status and the habits of each subject [1,2,3]. Moreover, a small percentage of exhaled breath (less than 1%) is made up of volatile organic compounds (VOCs), whose origin can be exogenous or endogenous [4]. Apart from exhaled breath, VOCs can also be detected in other human matrices such as urine, blood, or skin, and their chemical nature is very diverse (saturated and unsaturated hydrocarbons, nitrogenous compounds, organosulfur compounds, alcohols, aldehydes, or ketones, among others) [5,6].

Exhaled breath has been used as a source of disease biomarkers since ancient times. Greek physicians associated the sweet smell of breath with diabetes or a fishy smell with liver disorders [7,8]. Currently, Helicobacter pylori infection is commonly diagnosed through breath analysis, and breathalyzer tests are used by competent authorities to assess the blood alcohol levels of drivers [9]. Furthermore, in recent years, breath analysis has been implemented as a non-invasive tool for the search for biomarkers for diagnosis and monitoring of several respiratory diseases, gastrointestinal and metabolic disorders, as well as bacterial and viral infectious diseases, including tuberculosis, influenza virus, and COVID-19 [10,11,12,13,14,15].

Regarding the analytical platforms used for breath analysis, techniques based on mass spectrometry and techniques based on sensors are the most common [16,17]. In recent years, the clinical interest in sensor-based techniques for real-time breath analysis has increased, owing to their portability, simplicity, absence of need for qualified personnel, and short analysis times [16,18]. The most frequent sensor-based technique is the electronic nose (e-Nose). However, it has several drawbacks with respect to technologies based on mass spectrometry (e.g., gas chromatography coupled with mass spectrometry (GC/MS)), most important of which is the difficulty of individual identification of VOCs, and only allows detection of VOC patterns or “breath prints” in a complex mixture. Therefore, it is impossible to establish a relationship between specific VOCs in exhaled breath and different metabolic pathways or disease [17,19].

VOCs inhaled from ambient air have diverse sources such as fossil fuels, industrial processes, or disinfectants used in the water purification process [20]. Aromatic compounds such as benzene, toluene, ethylbenzene, xylene, or naphthalene can be derived from industrial activities or traffic, among others [21,22]. Hence, breath analysis can also be used to determine or examine occupational and environmental exposures. Monitoring by analysis of biological samples (biological monitoring or biomonitoring), such as exhaled breath, allows a more accurate assessment of the internal human dose of contaminants and their effect on individuals, rather than by direct environmental measurements [23,24]. In addition, analysis of endogenous VOCs in exhaled breath could enable the evaluation of the impact of environmental exposures at a physiological level on the organism [25].

Nowadays, it is well-known that environmental exposures can be associated with the development of diseases in the population, which may trigger major public health issues [23,26]. The exposome can be defined as the whole set of internal and external exposures suffered by a person, including, among others, radiation as well as chemical and biological agents. Research on the exposome contributes to knowledge of the environmental effects on the development of diseases [27]. Environmental pollution is linked to an increase in the development of respiratory and cardiovascular diseases, since exposure to particulate matter may entail oxidative stress and inflammatory response [28]. Not only is there an influence from outdoor air pollution, but pollution inside buildings (indoor air pollution) can also lead to health problems. Currently, people spend a lot of time indoors, in places such as schools, gyms, workspaces, or homes, so indoor air quality has an impact on health and the development of respiratory diseases (e.g., asthma) [29,30]. Indoor air is influenced by harmful agents derived from outdoor sources (e.g., traffic or industrial activities) and from indoor sources (e.g., environmental tobacco smoke (ETS), combustion of candles, household products, or air conditioning systems) [31].

This paper aims to summarize the possible applications of breath analysis for monitoring exposure to external agents. So, a comprehensive review of the studies carried out in the last 20 years focused on environmental pollution assessment by breath analysis using mass spectrometry has been performed. Special attention has been paid to the protocols followed for the collection and analysis of exhaled breath and statistical analysis, as well as to the VOCs selected as potential markers of environmental exposures.

2. Materials and Methods

2.1. Search Strategy and Study Selection

The search for papers was carried out in three databases (PubMed/Medline, Scopus, and Web of Science) and the following keywords with Boolean operators were used: (“environmental exposure” OR “exposome” OR “tobacco smoke” OR “exposure to tobacco” OR “pollutants” OR “environmental pollutants” OR “occupational exposure” OR “biomonitoring” and “human biomonitoring”) AND (“exhaled”) AND (“exhaled breath” OR “breath” OR “exhaled air” OR “volatilome” OR “volatile organic compounds” OR “VOC”) NOT (“condensate”). The literature search in WOS was performed in the “Web of Science Core Collection” by “topic”.

Studies were included if (i) they were original research articles published in the English language during the last 20 years (2004-March 2024); (ii) they were focused on VOC analysis in human exhaled breath in order to assess and monitor environmental exposures; and (iii) exhaled breath samples were analyzed by mass spectrometry-based technologies.

Studies were excluded if (i) they aimed only to analyze exhaled nitric oxide (FeNO), carbon monoxide (CO), and/or exhaled breath condensate (EBC).

Study selection was conducted independently by two reviewers (R.A.S-M. and A.P-G.) in two steps: (i) title and abstract screening, as well as (ii) full-text screening. Any disagreements between reviewers were resolved by consensus.

2.2. Data Collection

Next the datawere extracted from each study: (i) study design (target/untarget study); (ii) type of environmental exposure; (iii) study population; (iv) samples collected; (v) exhaled breath portion and breath container type; (vi) analytical platform type; (vii) statistical methods; (viii) main outcomes related to exhaled VOCs; and (ix) research location (country).

3. Results

3.1. Study Selection

The literature search identified a total of 1283 records (401 from PubMed/MEDLINE, 440 from Scopus, and 442 from WOS), of which 679 were duplicates and were discarded, leaving 604 entries. Of these, 531 did not fulfill the eligibility criteria, and 73 were selected for full-text screening. The main reasons for excluding studies include exclusive analysis of FeNO or CO, a topic other than environmental exposure monitoring, or the use of methodologies not based on mass spectrometry for breath analysis. Overall, 35 articles were excluded for several reasons, for example, use of other analytical methodologies for breath analysis instead of mass spectrometry-based techniques (n = 22), exclusive analysis of FeNO or EBC (n = 4), or lack of clear information on the methodology or results (n = 4). Finally, 38 studies were included in this review [32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69]. Figure 1 displays the flow diagram followed for study selection.

3.2. Data Collection

Data collected from the selected studies are detailed in

Table 1. Data extracted from the selected studies.

Ref.	Study Design	Environmental Exposure	Study Population	Samples	Exhaled Breath Portion/Breath Container	Analytical Platform	Statistical Methods	Summary of Findings Related to VOCs in Exhaled Breath
[32]	Target	Tobacco smoke exposure	268 subjects (48 smokers and 220 non-smokers)	-EB -AA	-Mixed expiratory breath -Teflon bag	PTR-MS	Mann–Whitney test and ROC curves	Acetonitrile (smokers vs. non-smokers)
[33]	Target	Tobacco smoke exposure	204 healthy subjects (100 smokers and 104 non-smokers)	-EB -AA	-Late expiratory breath -Tedlar bag	In-house capillary TD-GC/MS	Mann–Whitney test and Spearman correlation	2,5-dimethylfuran, benzene, toluene, o-xylene, and m/p-xylene (smokers vs. non-smokers) 2,5-dimethylfuran (significant differences after more than 24 h without smoking)
[44]	Untarget	Tobacco smoke exposure	137 patients with lung cancer and 143 healthy subjects (41 active smokers, 16 passive smokers, and 86 non-smokers)	-EB -AA	-End-tidal breath -Tedlar bag	SPME-GC/MS	Discriminant analysis and CHAID model	Higher concentration of acetonitrile, benzene, and furan derivatives in smokers without lung cancer Butyrolactone (passive smokers vs. non-smokers and smokers) 3-methylfuran and 2,5-dimethylfuran (smokers vs. non-smokers) Carbon disulfide (passive smokers and non-smokers)
[55]	Untarget	Tobacco smoke exposure	115 subjects (47 smokers and 68 non-smokers)	-EB -EA	-End-tidal breath -Tedlar bag and multibed sorption tube	TD-GC/MS	Kruskal–Wallis test and linear regression	Aromatic compounds, furan derivatives, alkenes, alkynes, dienes, ketones, VNCs, and VSCs (smokers vs. non-smokers and ex-smokers)
[64]	Target	Tobacco smoke exposure	89 healthy subjects (35 non-smokers, 24 ex-smokers, and 30 smokers)	-EB -AA	-Late expiratory breath -Bio-VOC and TD tubes (Tenax TA + graphitised carbon black + carbonized molecular sieve)	TD-GC/MS	Kruskal–Wallis test and Mann–Whitney test	Nonanal (smokers and ex-smokers vs. non-smokers)
[65]	Target	Tobacco smoke exposure	100 healthy subjects (non-smokers, ex-smokers, and smokers) and 57 subjects with COPD	-EB -AA	-Late expiratory breath -Bio-VOC and TD tubes (Tenax TA + graphitised carbon black + carbonized molecular sieve)	TD-GC/MS	Logistic regression and odds ratio calculation	Hexanal (COPD patients vs. healthy controls) Nonanal (smokers and ex-smokers vs. non-smokers)
[66]	Untarget	Tobacco smoke exposure	26 healthy subjects (16 smokers and 10 non-smokers)	-EB (in the morning before smoking—“blank smokers”) -EB (in the morning after 1 h abstinence after smoking) -EB (in the night after 1 h abstinence after smoking)	-Late expiratory breath -QUINTRON breath sampling system (discard bag and collecting bag)	SPME-GC/MS	Mann–Whitney test and predictive Probit model	Toluene, pyridine, and pyrrole (smokers vs. non-smokers) Nonane, 2,3-dimethyl (“blank smokers” vs. non-smokers) Toluene, pyridine, pyrrole, benzene, 2-butanone, 2-pentanone and 1-methyldecylamine (“blank smokers” vs. smokers after cigarette smoking)
[67]	Target	Tobacco smoke exposure in indoor environments	20 smokers	-EB -AA of room contaminated by cigarette smoke	-Mixed expiratory breath	CF-SPME-GC/MS	ANOVA test, linear regression and Pearson correlation	Acrolein (VOC levels in exhaled breath and indoor air were correlated)
[68]	Untarget	E-cigarette smoke exposure	5 non-vaping subjects	-EB from the day without vaping -EB from the vaping day -AA	-Late expiratory breath -Bio-VOC and TD tubes (Tenax TA)	TD-GC/MS		Higher concentrations of ethanol, ethyl acetate, and 1,4-Dichlorobenzene (vaping day vs. non-vaping day)
[69]	Target	Tobacco smoke exposure	Training group: 377 subjects (174 smokers and 203 non-smokers) Validation group: 64 subjects (self-reported smoking)	-EB	-Late expiratory breath -Tedlar bag	In-house capillary TD-GC/MS	Mann–Whitney test, ROC curves, and Multivariate logistic regression analysis	2,5-dimethylfuran, benzene, toluene, o-xylene, and m/p-xylene (smokers vs. non-smokers) 2,5-dimethylfuran (smoking status)
[34]	Untarget	Tobacco and e-cigarette smoke exposure	Training group: 48 healthy subjects (10 smokers, 18 vapers, and 20 non-smokers) Validation group: 4 smokers and 4 vapers	-EB after smoking or vaping -AA from breath sampling room	-Mixed expiratory breath -Tedlar bag	SPME-GC/MS	PCA with HCA and PLS-DA	Aromatic compounds, furan derivatives, VNCs, ketones, and alkenes are related to smokers Esters, terpenes, and oxygenated compounds (vapers vs. smokers)
[35]	Target	Occupational exposure: BTX exposure in gas station attendants	29 gas station attendants and 16 office workers	-EB (pre- and post-shift samples) -AA (personal air sampling)	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 1TD and Carbograph 2TD)	TD-GC/MS	Wilcoxon signed-rank test, Mann–Whitney test, Kruskal–Wallis test, and Spearman correlation	Benzene, toluene, m/p-xylene, and o-xylene (post-shift vs. pre-shift) Benzene, toluene, and m/p-xylene (gas station attendants vs. controls)
[36]	Target	Occupational exposure: benzene in gas stations and in gasoline quality control laboratories	25 subjects exposed to benzene in gasoline (workers in gas stations and in gasoline quality control laboratories) and 25 non-exposed subjects	-EB in the end of morning or in the middle of the work shift	-Late expiratory breath -SPME	SPME-GC/MS	ANOVA test and Brown–Forsythe’s Test	Benzene (exposed group vs. non-exposed group)
[37]	Target	Occupational exposure: benzene from gasoline in different workplaces	15 workers exposed to low levels of benzene (employees of restaurants, coffee shops, offices, park guards, and teachers) and 30 workers in retail gasoline stations—“exposed group”	-EB -AA (urban air samples and ambient air of the workplace)	-Late expiratory breath -SPME	SPME-GC/MS	Mann–Whitney test, correlation analysis, and linear models constructed by the least squares method, weighted by the experimental variance	Benzene (VOC levels in exhaled breath and in ambient air were correlated) Benzene (exposed group vs. non-exposed group)
[38]	Target	Exposure to disinfection by-products in chlorinated swimming pools	116 healthy and non-smoking subjects (non-professional swimmers)	-EB before and after swimming -AA from different locations of the swimming pool	-Late expiratory breath -Bio-VOC and TD tubes	TD-GC/MS	Kruskal–Wallis test, Spearman correlation, and linear regression models	Median level of exhaled total trihalomethanes (chloroform, dibromochloromethane, bromodichloromethane and bromoform) increased after swimming Dibromochloromethane, Bromodichloromethane, and bromoform (VOC levels in exhaled breath and in pool water were correlated) (VOC levels in exhaled breath and trichloramine in air were correlated)
[39]	Target	Exposure to disinfection by-products in chlorinated swimming pools	43 healthy and non-smoking subjects (non-professional swimmers)	-EB before and after swimming -AA (room and swimming pool)	-Late expiratory breath -Bio-VOC and TD tubes (Tenax TA 35/60 mesh)	TD-GC/MS	Paired t-test	Trihalomethanes (chloroform, dibromochloromethane, bromodichloromethane and bromoform) (after swimming vs. before swimming)
[40]	Target	Residential exposure: by-products of water chlorination	7 healthy subjects (4 in a residence with water with a high concentration of trihalomethanes and 3 in a residence with water with a low concentration of trihalomethanes)	-EB before all water use activities (baseline) and during or after water use activities -AA	-End-tidal breath -1 L Silcosteel stainless steel canisters	GC/MS		Higher concentration of chloroform in exhaled breath of participants from the site with water, with high concentration of trihalomethanes
[63]	Target	Residential exposure: by-products of water chlorination	7 healthy subjects that performed 12 common water-use activities in 2 residences	-EB before water use activities and 5 min after the end of the activity -EB during the shower event (n = 2) -AA	-End-tidal breath -1 L Silcosteel stainless steel canisters	GC/MS	Dixon’s outlier test, Mann–Whitney test, and Spearman correlation	Chloroform (after showering/bathing vs. before showering/bathing) Chloroform (VOC levels in exhaled breath and indoor air were correlated in showering) (VOC levels in exhaled breath and water were correlated in bathing) (VOC levels in exhaled breath and blood were correlated in showering and bathing)
[41]	Target	Residential exposure: indoor dampness	337 women (86 with dampness in home and 251 without dampness in home) and 337 children (87 with dampness in home and 250 without dampness in home)	-EB -AA from breath sampling room	-Mixed expiratory breath -Tedlar bag (women)/ Tedlar bag + Quintron bag (children) and TD tubes (Tenax TA/carbograph 5td)	TD-GC/MS	Mann–Whitney test and Student’s t-test	2-ethyl-hexanol (women with home dampness exposure vs. women without home dampness exposure)
[42]	Target	Residential exposure: chlorinated VOCs caused by groundwater contamination plumes	38 healthy non-smokers from 26 residences located in different areas (12 on the superfund site, 11 on other plumes, and 3 outside any plumes)	-EB -AA at homes located on the superfund site	-Late expiratory breath -Tedlar bag	TD-GC/MS	ANOVA test and mixed linear models	Tetrachloroethylene (VOC levels in exhaled breath and in indoor air were correlated)
[43]	Target	Soil contamination with chlorinated hydrocarbons in a bookshop that was a dry-cleaning shop	2 smoking workers in bookshops (shop owner who works 7 days per week and employee who works 3 days per week)	-EB (pre-shift and post-shift samples) -AA inside bookshop and outdoor sample close to ventilator air intake point (EB and AA in summer and in winter)	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 1 and Carbosieve SIII)	TD-GC/MS	Paired t-test	Tetrachloroethylene, benzene, and toluene (post-shift vs. pre-shift)
[45]	Target	Occupational exposure: tetrachloroethylene in different workplaces	24 workers of dry cleaners, 1 worker in an electroplating industry, 1 worker of in a research laboratory, and 1 worker in an automotive paint preparation shop	-EB at the end of the work shift -AA (samples of the different workplaces)	-Mixed expiratory breath	CF-SPME-GC/MS	Linear regression and correlation analysis	Tetrachloroethylene (VOC levels in exhaled breath and in ambient air were correlated)
[46]	Untarget	Occupational exposure: long-term professional exposure to asbestos	39 subjects (13 patients affected by malignant pleural mesothelioma (MPM), 13 subjects with long-term professional exposure to asbestos “exposed group”, and 13 healthy controls)	-EB	-Mixed expiratory breath -Tedlar bag and sorbent cartridges (Carboxen 1003, Carbopack B, and carbopackY)	TD-GC/MS	ANOVA test, PCA, DFA and CP-ANN	Cyclopentane, methyl-octane, and dimethyl-nonane (exposed group vs. MPM and healthy controls) Cyclopentane (long-term asbestos exposure)
[47]	Target	Occupational exposure: monitoring of sevoflurane exposure levels in hospital staff	5 anesthesiologists working in different operating rooms	-EB (at beginning of first day of working week, at end of same day, and at end of working week)	-Mixed expiratory breath -Nalophan bag and TD tubes (60/80 mesh Tenax GR phase (70% Tenax TA, 2,6-diphenyl-p-phenylene oxide and 30% graphite))	TD-GC/MS		Inconclusive results
		Occupational exposure: monitoring of isopropyl alcohol exposure levels in hospital staff	9 nurses	-EB (before beginning of work shift, and 90 and 180 min later) -AA			Linear regression and correlation analysis	Isopropyl alcohol (VOC levels in exhaled breath and in ambient air were correlated)
[48]	Untarget	Occupational exposure: VOCs in operating room personnel	12 workers of surgical operations (surgeons, surgical assistants, or nurses) and 1 administration nurse	-EB (before and after surgery) -AA (during the surgeries in process)	-Mixed expiratory breath -Bottle-Vac	GC/MS	Wilcoxon signed rank test	Sevoflurane (after surgical operations vs. before surgical operations)
[49]	Target	Exposure to anesthetic gases and disinfectants in hospital environments	100 subjects (24 hospital staff, 45 hospital visitors, and 31 external controls)	-EB	-Late expiratory breath -Tedlar bag	In-house microtrap-GC/MS	Mann–Whitney test	Isopropyl alcohol (hospital staff vs. external controls) and (hospital visitors vs. external controls) Sevoflurane (hospital staff vs. hospital visitors) 2,5-dimethylfuran (smoking status)
[50]	Target	Occupational exposure: acetonitrile in a university chemical laboratory	14 healthy non-smokers (6 workers at chemistry department laboratory—“exposed group”; 8 workers at geography department—“non exposed group”)	-EB (in the morning and early and late afternoon) -AA	-Mixed expiratory breath -Tedlar bag	SIFT-MS		Acetonitrile (exposed group vs. non-exposed group)
		Acetonitrile exposure testing	4 healthy non-smokers sat for 30 min in laboratory	-EB (before exposure, after exposure, and 30 min after exposure)				Higher concentration of acetonitrile after 30 min exposure
[51]	Target	Occupational exposure: solvent emissions in university chemical laboratories	76 subjects (55 researchers from 4 laboratories of chemistry department—“exposed group”; 21 non-exposed subjects)	-EB -AA (samples for each laboratory and for each sampling day)	-Late expiratory breath -Tedlar bag	In-house capillary TD-GC/MS	Kruskal–Wallis test and Mann–Whitney test	Diethyl ether, acetone, n-hexane, 2-methylpentane, methylene chloride, 3-methylpentane, methylcyclopentane, ethyl acetate, chloroform, n-pentane, n-heptane, benzene, and toluene (exposed group vs. non-exposed group)
[52]	Target	Occupational exposure: toluene used as solvent in a chemical factory	36 workers of a chemical factory exposed to toluene	-EB (16 h after shift) -AA in breathing zone during work-shift with personal passive dosimeters	-Late expiratory breath -Glass vial (Tenax TA) and TD tube (Carbotrap 201)	TD-GC/MSD	Linear regression and correlation analysis	Toluene (VOC levels in exhaled breath and in environmental air were correlated)
[53]	Untarget	Occupational exposure: crystalline silica dust	69 subjects (20 workers exposed to silica, 4 silicosis patient—“positive”, 20 healthy non-smokers, and 25 healthy smokers)	-EA -AA	-Late expiratory breath -Tedlar bag	SPME-GC/MS	Kruskal–Wallis test, ANOVA test, and Student’s t-test	Acetaldehyde, 2-propanol, decane, 1,3 butadiene, propanthiol, 3-hydroxy-2-butanone, hexanal, pentadecane, butanoic acid, and nonanal (exposed subjects vs. negative control groups)
[54]	Target	Occupational exposure: firefighters in two rounds of controlled structure burns	18 firefighters	-EB before, shortly after, and 6 h after specific firefighting tasks (planned exposure)	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Heatmap, within-subject and between-subject variance components, and ICC	Benzene, toluene, ethylbenzene, styrene, 1,3,5-trimethylbenzene, and naphthalene (post firefighting tasks vs. pre firefighting tasks) Benzene, ethylbenzene, m,p-xylene, styrene, and naphthalene (post firefighting tasks vs. 6 h after firefighting tasks)
[56]	Target	Occupational exposure: firefighters in two rounds of controlled structure burns	18 non-smoking firefighters	-EB before, shortly after, and 6 h after the controlled burn -AA (personal air sampling)	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Non-parametric sign tests and Spearman correlation	Benzene (post firefighting tasks vs. pre firefighting tasks) ΔBenzene (change in pre- to post-firefighting tasks) was correlated with personal air concentrations of polycyclic aromatic hydrocarbons in firefighters from round 2
[57]	Target	Occupational exposure: VOCs off-gassing from personal protective equipment (PPE) of firefighters in controlled structure burns	6 non-smoking firefighters	-EB before and shortly after controlled burn -AA inside structure (before each burn, during each burn, and during last burn) -Off-gas sampling from PPE Ensembles (before each burn and after each burn)	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Paired t-tests, linear regression and Pearson correlation	Benzene, toluene, ethylbenzene, xylenes, and styrene (post-burn exhaled breath concentrations and off-gassing air concentrations from used PPE ensembles were correlated)
[58]	Target	Occupational exposure: firefighters with different tasks in controlled structure burns	40 non-smoking firefighters	-EB before, immediately after, and 1 h after participation in controlled structure burns -AA	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Student’s t-test	All firefighter samples: benzene and ethylbenzene (post-exposure vs. pre-exposure) Firefighters participating in attack and search: benzene, ethylbenzene, m/p-xylene, and o-xylene (post-exposure vs. pre-exposure) Firefighters participating in outside ventilation: benzene (post-exposure vs. pre-exposure)
[59]	Untarget	Occupational exposure: firefighters in controlled structure burns	40 non-smoking firefighters	-EB before, immediately after, and 1 h after participation in controlled structure burns -AA	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Student’s t-test and heatmap	Decane (post-exposure vs. pre-exposure) Trifluorobenzene and 1,2,4-trimethylbenzene (post-exposure vs. pre-exposure)
[60]	Target	Occupational exposure: firefighters with different tasks and fire attack tactics in controlled residential fires	36 non-smoking firefighters	-EB before, immediately after, and 1 h after each fire	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Mixed linear models	Firefighters assigned to attack and search: benzene, Ethylbenzene, and xylenes (post-exposure vs. pre-exposure) Firefighters assigned to outside vent: benzene (post-exposure vs. pre-exposure) Firefighters assigned to overhaul: benzene (post-exposure vs. pre-exposure) No difference in concentrations of exhaled benzene associated with tactic (interior attack vs. transitional attack)
[61]	Target	Occupational exposure: firefighters with three different PPE ensembles and two treatments (new/laudered) during fire exposure	24 firefighters	-EB before and immediately after each fire -AA (personal air sampling of the outside and inside of the turnout jacket) -AA	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Paired t-test, mixed models and Pearson correlation	Benzene (post-fire vs. pre-fire) There were no significant differences among 3 conditions Toluene (post-fire vs. pre-fire) in firefighters with new Nomex® knit hood, new turnout jacket, and new turnout pants Firefighters who did not wear new knit hood: ΔBenzene (change in pre- to post-fire) was correlated with outside and inside jacket personal air concentrations ΔBenzene (change in pre- to post-fire) was correlated with outside and inside jacket personal air concentrations in firefighters with laundered turnout jacket, pants, and particulate-blocking hoods
[62]	Target	Occupational exposure: firefighters with three PPE and base-layer configurations during fire exposure	23 non-smoking firefighters	-EB before and immediately after removing the equipment after each burn scenario -AA (personal air sampling) -AA	-Late expiratory breath -Bio-VOC and TD tubes (Carbograph 2TD and Carbograph 1TD)	TD-GC/MS	Mixed models	Benzene and toluene (post-fire vs. pre-fire) with all PPE configurations and zip statuses.

Ref.: references, VOCs: volatile organic compounds, EB: exhaled breath, AA: ambiental air, MS: mass spectrometry, PTR: proton transfer reaction, GC: gas chromatography, TD: thermal desorption, CF: cold fiber, SPME: solid-phase microextraction, ROC: receiver operator characteristic, CHAID: chi-squared automatic interaction detector, VNCs: volatile nitrogen-containing compounds, VSCs: volatile sulfur-containing compounds, COPD: chronic obstructive pulmonary disease, e-cigarette: electronic cigarette, BTX: benzene, toluene, and xylenes (p-/m-/o-xylene), PCA: principal component analysis, HCA: hierarchical cluster analysis, PLS-DA: partial least squares discriminant analysis, DFA: discriminant function analysis, CP-ANNs: counterpropagation artificial neural networks, SIFT: selective ion flow tube, MSD: mass selective detector, ICC: intra-class correlation co-efficient. The red words represent significantly high levels in the exposed group, the purple words represent discrimination between groups, the blue words represent VOC levels positively correlated, and the green words represent significantly high levels in the non-exposed group.

. The geographical location of the selected studies is indicated in Figure 2.

3.3. Environmental Exposures Assessed by Breath Analysis

Most of the studies focused on the assessment of exposure to harmful agents in workplaces (21 studies) [35,36,37,43,45,46,47,48,49,50,51,52,53,54,56,57,58,59,60,61,62]. Indeed, current society is very concerned about indoor exposures, including occupational exposures, since people spend most of their time indoors, making indoor air quality a significant factor in public health [29,31]. Of these 21 studies, 8 evaluated the fire exposure of firefighters [54,56,57,58,59,60,61,62]. This high percentage of studies stems from the large number of potentially carcinogenic toxic compounds that are generated during burning. Exposure to these compounds can lead to the development of health problems, hence the interest in monitoring them [70]. In addition, breath analysis could also be used to assess the effectiveness of the protective equipment that firefighters wear.

Other studies have addressed household exposures (4 studies) [40,41,42,63]. This is another indoor exposure that is relevant to monitor, as it is experienced daily by the whole population. It highlights the assessment of exposure to water disinfection by-products, which could be potentially hazardous [71]. Exposure to these by-products has also been studied in indoor swimming pools (two studies) [38,39]. Breath analysis could be useful for the identification of disinfectants that generate a lower concentration of harmful compounds, and for the determination of the optimum doses of these disinfectants.

In addition, several studies assessed the effect of smoking status or ETS exposure on human exhaled breath (10 studies) [32,33,34,44,55,64,65,66,67,69]. A couple of studies examined exhaled breath from people who use electronic cigarettes (vapers) [34,68], whose popularity has grown in the last decade [72].

3.4. Collection and Analysis of Exhaled Breath

None of the selected studies conducted real-time breath analysis [73], in other words, direct analysis of exhaled breath. This may be a consequence of the fact that subjects must be transported to a laboratory with an analytical platform for real-time breath analysis by mass spectrometry [73,74]. The laboratory can often be far from the exposure scenario, which may decrease the applicability of the real-time breath analysis approach by mass spectrometry for biomonitoring environmental exposures. Thus, all studies performed offline breath analysis [17], using a breath container for transport and storage. Table 1 shows that the most popular container for exhaled air collection was the Bio-VOC (15 studies) [35,38,39,43,54,56,57,58,59,60,61,62,64,65,68]. There were also many studies using gas sampling bags, such as Tedlar bags (12 articles) [33,34,41,42,44,46,49,50,51,53,55,69], perhaps due to their low cost and reusability.

In terms of the exhaled breath portion (Figure 3), a few of the selected studies collected end-tidal breath. To collect end-tidal breath, a capnograph is used to determine CO₂ levels and to identify the alveolar air of exhalation. Alveolar air is enriched in endogenous VOCs [7,17,75]. Therefore, the collection of end-tidal breath is very interesting to find biomarkers of internal damage by environmental exposures. In addition, alveolar air analysis is useful to determine which exogenous VOCs may have passed from the lungs into the bloodstream [76]. Nine studies [32,34,41,45,46,47,48,50] collected the mixed expiratory breath, which includes all air emitted on exhalation (dead space air, airway air, and alveolar air). Apart from alveolar air analysis, evaluation of upper airway air can provide valuable information about exposure to environmental pollutants, such as exposure to tobacco smoke [17,76]. The procedure of the mixed expiratory breath collection is very simple; however, there is a greater risk of interference from VOCs of the sampling room [17,75]. Therefore, measuring VOCs in the ambient air of the sampling room is an essential strategy for reducing their potential influence and ensuring the reliability of the findings [34,41]. On the other hand, most studies collected the late expiratory breath (25 studies) [33,35,36,37,38,39,43,49,51,52,53,54,56,57,58,59,60,61,62,64,65,66,68,69,77]. This is in line with the widespread use of Bio-VOC. This device allows for easy collection of late expiratory breath. Late exhaled breath collection aims at the exclusion of dead space air through the elimination of the first part of exhaled breath [75]. Despite the fact that the collection of late expiratory breath reduces the influence of exogenous VOCs from the sampling room, the main issue is the reproducibility. The reason is that there is no precise reference point, like CO₂ in end-tidal breath collection, to indicate when the exhaled breath sample should be collected. So, the collection of this portion does not ensure the exclusive presence of alveolar air. In this sense, the use of Bio-VOC is beneficial in improving accuracy [75,78].

In addition, 94.7% of the studies [33,34,35,36,37,38,39,40,41,42,43,45,46,47,48,49,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69] used GC/MS as the analytical technique, which is the gold standard in breath analysis due to its high resolution and sensitivity. GC/MS requires a preconcentration step prior to sample analysis [10]. In the selected studies, thermal desorption (TD) was the most frequently used preconcentration method (Table 1). In this regard, in many of the studies, exhaled air was transferred from breath containers to thermal desorption tubes (TD tubes) [35,38,39,41,43,46,47,52,54,55,56,57,58,59,60,61,62,65,68]. Most of the tubes had multi-bed sorbents (e.g., Tenax TA, carbograph 1TD, 2TD or 5TD, carboxen, etc.), which increases the range of VOCs that can be retained. The use of TD tubes is attractive because sample storage is much more stable than in other breath containers (e.g., gas sampling bags). Thus, storage of exhaled breath in TD tubes is beneficial for transport from the collection site to the analysis platform [75,79]. In addition, several studies used solid-phase microextraction (SPME) as a preconcentration method [34,36,44,45,53,66,67].

Only a couple of studies used analytical techniques other than GC/MS. One study used PTR-MS (proton transfer reaction—mass spectrometry) [32] and another used SIFT-MS (selective ion flow tube—mass spectrometry) [50]. However, although these techniques can be implemented for real-time breath analysis, they were used for offline breath analysis in the selected studies. This option has the advantage of not requiring the subject to move to the analytical platform [17,80].

3.5. Statistical Analysis in Biomonitoring by Breath Analysis

As can be seen in Table 1, most of the studies conducted a statistical analysis with a traditional approach focusing on univariate analyses comparing exhaled levels of a compound between different study groups. For instance, a control group that was not exposed, and another group that was exposed, or groups with different degrees of exposure. Thus, both parametric (Student’s t-test or ANOVA test) and non-parametric (Mann–Whitney test or Kruskal–Wallis test) tests were commonly used. None of the studies reported the use of corrections for multiple comparisons (e.g., Bonferroni correction) in order to avoid false positives [7]. Differences in VOC levels in the same group before and after exposure were also frequently assessed using tests such as a paired t-test. In studies that collected an ambient air sample, correlation analysis (Pearson correlation or Spearman correlation) was usually performed to determine the relationship with exhaled breath samples. Nevertheless, the implementation of multivariate analyses was not widespread. Indeed, only a few studies assessed the influence of environmental exposures on the whole pool of VOCs or volatilome in exhaled breath. Multivariate statistical techniques provide a more global approach. For this reason, a successful strategy for monitoring environmental exposures could be to perform multivariate analysis to have a more comprehensive view and to generate models that can be validated and thus extrapolated to another population [12,17,81].

3.6. Selected VOCs for Environmental Exposure Monitoring by Breath Analysis

Several VOCs were selected as possible markers for monitoring different environmental exposures (Table 1). Figure 4 summarizes some of the associations observed between exhaled VOCs and environmental exposures in the selected studies. Many studies assessed exhaled levels of aromatic compounds such as benzene, toluene, xylenes (m-xylene, p-xylene, and o-xylene), and ethylbenzene. Specifically, the term “BTEX” is often used to refer to these four compounds together. They are a group of solvents widely used in industry and have been associated with air pollution [21,24,82]. “BTEXS” is often used as a term that also includes styrene, another aromatic compound [83]. Exhaled levels of benzene and their relationship to diverse environmental exposures were examined in many studies [33,35,36,37,43,54,56,57,58,60,61,62,66,69]. Benzene is considered a carcinogen by the International Agency for Research on Cancer (IARC), so it is essential that exposure to this compound is monitored [24]. Three studies analyzed in this review found an association between levels of benzene in exhaled breath and occupational exposure at petrol stations [35,36,37], as it is a component of gasoline and is emitted by automobiles [84]. Likewise, a large number of papers observed a relationship between exhaled benzene levels and occupational exposure to fires in firefighters [54,56,57,58,59,60,61,62]. Benzene can be a by-product of combustion, which could explain its association with fire exposure [85,86]. Other studies have found significantly elevated levels of benzene in the exhaled breath of smokers [33,66,69]. Like benzene, higher levels of toluene and xylenes were also observed in smokers [33,66,69] and gas station attendants [35]. In addition, there was a relationship between occupational exposure of firefighters and exhaled levels of toluene, xylenes, ethylbenzene, and styrene [54,57,58,59,60,61,62]. Another study monitored toluene levels in exhaled breath in workers of a chemical factory exposed to this solvent [52]. Therefore, levels of BTEX or BTEXS in human exhaled breath could be influenced by different external agents.

Naphthalene in exhaled breath was also determined to evaluate the influence of firefighter tasks [54]. Naphthalene is a polycyclic aromatic hydrocarbon (PAH) that can be produced by burning processes and is found ubiquitously in the environment [22].

Acetonitrile [32], 2,5-dimethylfuran [33,44,69], or 3-methylfuran [44] were associated with active smoking in the selected studies. Indeed, 2,5-dimethylfuran was highlighted as a possible biomarker of smoking status after being tested in a validation group by Castellanos et al. [69]. Carbon disulfide and butyrolactone were identified as possible discriminants between passive smokers and non-smokers [44].

In a couple of studies [38,39], trihalomethanes (THMs), such as chloroform, dibromochloromethane, bromodichloromethane, or bromoform, were detected in exhaled breath to assess exposure to disinfection by-products in chlorinated swimming pools. In addition, chloroform has been measured in exhaled breath to check residential exposure to water chlorination by-products [40,63]. THMs are water disinfection by-products (DBPs), and prolonged exposure to them has been linked to bladder cancer [24,71].

Furthermore, tetrachloroethylene [42,43,45] and sevoflurane [48,49] were measured in exhaled breath samples in several studies. Tetrachloroethylene is commonly used as a solvent in dry-cleaning shops and can be found in contaminated groundwater and soil [42,87]. It is a potent neurotoxicant and carcinogen and has been associated with infertility. Two studies confirmed the correlation between ambient air and exhaled breath levels of tetrachloroethylene [42,45]. Sevoflurane is one of the most widely used inhalation anesthetics in clinical practice and a common hospital environmental contaminant [88].

On the other hand, several studies examined VOCs that could have an endogenous origin, such as alkanes, aldehydes, ketones, or alcohols. Many of these compounds could be by-products of different human metabolic pathways and be related to increased oxidative stress, lipid peroxidation, or inflammation [89,90]. Levels of endogenous VOCs could be modified due to internal damage caused by exposure to pollutants. Therefore, the detection of these VOCs in exhaled breath in the monitoring of environmental exposure is of great utility, since it allows for determining the effects that these environmental exposures cause in the organism.

4. Discussion

In this paper, an exhaustive review of the studies published in the last 20 years focused on breath analysis using mass spectrometry-based techniques as a tool for biomonitoring environmental exposures has been carried out. Specifically, 38 studies have been selected (Figure 1), and information on sampling protocols, analytical techniques, statistical techniques, and VOCs selected as possible biomarkers of exposure was extracted (Table 1).

In the selected studies, breath analysis was successfully used for monitoring different environmental exposures, including occupational exposure in diverse workplaces or indoor contaminant exposure at home. However, most studies focused on specific exposures, such as fire exposure in firefighters or tobacco smoke. Therefore, further studies are needed to examine the role of breath analysis in monitoring other environmental exposures.

Moreover, some of the selected studies tested the correlation between VOC levels in ambient air samples from hazardous areas and in exhaled breath samples [37,38,42,45,47,52,67]. More accurate exposure assessment is the main advantage of biomonitoring techniques over other common approaches, such as analysis of ambient air samples. Although ambient air sampling involves simple protocols, it does not take into account some key factors that affect the internal body dose of the contaminant, such as physiological factors or the mobility of the subjects [24,82]. Collection of ambient air by a device close to the breathing zone (personal air sampling) is a strategy that allows carrying out ambient air measures in the periods and places where subjects suffer from exposure [91]. However, despite overcoming some of the limitations of conventional ambient air analysis, this approach does not consider what happens when the contaminant is introduced through the airways. The analysis of biological specimens, such as exhaled breath, provides insight into the real impact of contaminant exposure on the metabolism and health status of the organism [23,24,82]. Despite the fact that it is a less established strategy than the analysis of other biological matrices (e.g., blood and urine), breath analysis has some attractive advantages that could be helpful for biomonitoring environmental exposures. The strengths of breath analysis over blood analysis are the non-invasiveness and the lack of need for professional staff, such as nurses, to collect the samples [82,92]. Analysis of exhaled breath, especially alveolar air analysis, could provide a non-invasive method for detecting the presence of exogenous and/or endogenous VOCs in the blood after exposure to an environmental pollutant [76,92]. For example, Gordon et al. [63] observed a correlation in chloroform levels between exhaled breath and blood samples after showering or bathing in a study on residential exposure to by-products of water chlorination. Concerning the analysis of urine samples, breath analysis can provide a reflection of gas exchange during or immediately after exposure to a hazardous agent. Urine samples usually contain intermediate or final products of contaminant metabolism. So, there is a time-lag from environmental exposure to urine detection due to the need to wait for the contaminant metabolism [82,92].

Furthermore, the 38 selected studies carried out offline breath analysis, which means that the sample was collected in a breath container prior to analysis. Analytical platforms based on mass spectrometry are expensive, large, and difficult to transport [17]. The use of portable breath containers is very attractive for environmental exposure surveillance because it allows the sample to be collected in situ, that is, at the same place and time of exposure. However, it could be interesting to implement real-time breath analysis using mass spectrometry-based technologies such as SIFT-MS, PTR-MS, or SESI (secondary electrospray ionization)-MS in future studies on biomonitoring of environmental exposures, especially those where exposed subjects can easily go to the laboratory (e.g., tobacco smoke exposure).

Most of the studies conducted a targeted analysis and therefore focused exclusively on some VOCs in exhaled breath. However, analysis of the exhaled volatilome is more complex, and an untargeted analysis could extract relevant findings and provide more information. Not only can the pollutant compounds be detected in exhaled breath, but their derivatives produced upon metabolism and endogenous VOCs that provide evidence of damage caused by exposure can also be determined [25,93,94,95]. In volatilome analysis, a common phenomenon is multicollinearity, or the existence of many VOCs that are related to each other (e.g., compounds derived from the same metabolic pathway). High dimensionality is also typical in volatilome data, i.e., “fat matrices” with numerous variables, such as the hundreds of VOCs present in exhaled breath [81,96,97]. For all these reasons, the use of multivariate techniques is mandatory for volatilome analysis. Currently, there are many advanced multivariate statistical methods, including both unsupervised and supervised learning methods, that can be used for volatilome data analysis [17,81]. Nevertheless, only a few of the selected studies implemented multivariate statistical techniques, which is in line with the small number of studies that performed an untargeted analysis. In this regard, it is strongly recommended that more multivariate techniques be utilized in future studies for environmental exposure monitoring by breath analysis.

Regarding the reported VOCs, at least 50 VOCs were highlighted for their potential for environmental exposure biomonitoring using breath analysis. However, common limitations of existing studies, for instance, lack of validation of findings in independent groups of subjects, small study populations, or lack of longitudinal studies, remain a challenge in the identification of true biomonitoring markers. Several papers involved case–control studies (exposed group versus unexposed group) with a small number of participants. Most studies drew conclusions based exclusively on comparisons between these two groups, without validating their results in an independent set of participants. In this regard, it is worth highlighting the study by Castellanos et al. [69], which validated the usefulness of 2,5-dimethylfuran in a new group of subjects as a marker of smoking status. Notwithstanding this, the same associations between certain VOCs and some environmental exposures have been observed in different studies (e.g., benzene and occupational exposure in firefighters, nonanal and tobacco smoke exposure, among others) (Figure 4). As a result, several findings were consistent in different studies. Another controversial point is the low specificity of exhaled VOCs, in other words, the association of the same VOC with different environmental exposures. While some VOCs were associated with a specific source (e.g., 2,5-dimethylfuran and smoking status), many VOCs, such as BTEX, were linked to multiple environmental exposures. Although this may seem like a limitation a priori [24], the measurement of ubiquitous pollutant VOCs can provide a holistic view of the different environmental exposures that affect an individual (exposome). Besides being able to examine the actual impact of various environmental exposures on the organism as a whole, breath analysis could also indicate their influence on metabolism through the measurement of endogenous VOCs. Thus, breath analysis provides an attractive approach for biomonitoring the human exposome [24,25,82,93]. Nevertheless, more studies focusing on the use of exhaled volatilome analysis to assess the human exposome are needed. It is essential to determine the influence of a wide range of exposure sources on exhaled VOCs and to evaluate the possible interaction between different factors.

5. Limitations

The aim of this review was to compile and examine studies conducted over the last 20 years on biomonitoring of environmental exposure using breath analysis by means of mass spectrometry analytical platforms. Nevertheless, this review may have some limitations due to its search strategy and inclusion and exclusion criteria. Thus, the study search was only carried out in three databases (PubMed/Medline, Scopus, and Web of Science) and was limited to studies written in English, which means that some studies may have been omitted. Similarly, studies using methodologies other than mass spectrometry, such as sensor-based technologies, were excluded. Thus, a more in-depth analysis of the advantages and disadvantages of other technologies compared to mass spectrometry could not be performed.

6. Conclusions

A comprehensive review of the use of breath analysis using mass spectrometry-based techniques for environmental pollution exposure assessment was carried out in this paper. Research studies conducted over the last 20 years have shown a prevalence of offline exhaled air analysis using breath containers, the implementation of GC/MS, and a targeted approach focusing on a few VOCs. Several exhaled VOCs have been linked to different environmental exposures, such as exposure to tobacco smoke, exposure to fires in firefighters, exposure to water chlorination by-products, among others. Thus, breath analysis, as a non-invasive strategy for biological specimen analysis, could play a key role in offering a more accurate estimation of the organism level of environmental exposure and its influence on human health. However, multiple challenges must be addressed, such as the recruitment of large cohorts for longitudinal studies, the use of multivariate statistical techniques, further studies with an untargeted approach, the implementation of analysis of the volatilome and not of specific VOCs, and a rigorous assessment of the different factors that can influence the levels of exogenous VOCs and their interaction. All this could contribute to the establishment of breath analysis as a technique for biomonitoring the exposome.