Chapter 1: Origin and Emission of Volatile Biomarkers in Breath: End-tidal Perspective Free

The volatile organic compound (VOC) composition of exhaled breath is complex and determined by many internal and external factors, such as metabolic, inflammatory and oxidative stress processes and environmental conditions, respectively.¹  Nearly 1500 trace VOCs have been found in exhaled breath,²  with detected concentrations ranging from subparts per trillion by volume (pptv) up to hundreds of parts per million by volume (ppmv). These VOCs originate from several sources, including the respiratory tract, organs, tissues and microbiomes, as schematically represented in Figure 1.1. They are the subject of intense international research mainly focused on biomarker discovery.

The main source of volatiles in breath is from systemic circulation, with VOCs passing through the barrier between the blood capillaries and alveoli into the lungs. From there, they are carried out in the exhaled breath. VOCs also originate from the lungs themselves, the upper airways, the gut and the oral cavity. VOCs coming from the mouth and gut are generally a result of symbiotic and pathogenic bacterial presence and from the intake of foods, drinks, medications and oral products (e.g., toothpaste and mouthwashes) and metabolites of them. Hence, a VOC in breath can have endogenous (systemic) origins, which result from normal and abnormal metabolic activities or pathological disorders, or is exogenous in nature, resulting from the ingestion, inhalation or dermal absorption of chemicals and substances, or has contributions from both endogenous and exogenous sources. For breath test applications, the key point is that the VOC profile created from all possible sources can change because of a disease or an infection, reflecting the physiological state of an individual at the time of sampling.

Gas chromatography equipped with an ionization detector was used as early as 1960 to investigate VOCs in expired air.³  In this pioneering study, 11 compounds were identified but not quantified, including acetone, isoprene and ethanol, with the paper identifying potential uses for these VOCs in medical diagnosis, nutritional studies and physiological research. However, the advent of modern breath trace VOC analysis is generally attributed to Linus Pauling and his coworkers, who used gas chromatography–mass spectrometry (GC–MS) in the early 1970s.⁴  Since then, breath research has predominantly focused on the discovery of endogenous VOCs as potential biomarkers of disease. However, in recent years there has been a growing realization that exogenous VOCs provide useful probes to metabolic processes, particularly for use in personalized medicine. This approach is sometimes referred to as “induced volatilomics”.

Volatiles present in the breath, including hydrogen (H₂), carbon dioxide (CO₂) and trace VOCs, can result from the ingestion, adsorption and/or inhalation of a volatile that is not metabolized or from the metabolism of an ingested compound (e.g., drugs or a chemical substrate). The major advantage is that the intake of an exogenous VOC or an ingested substrate can be selected so that the associated VOCs in breath are not present beforehand in the body, which greatly simplifies breath analysis. For example, exogenous VOCs in the breath can be used for in vivo pharmacokinetics studies to determine washout characteristics resulting from a VOC's excretion, skin emission, exhalation, distribution, absorption and metabolism.^5–8  In this context, the application of isotopically labeled metabolites as biochemical probes is a very promising approach but has limitations that are discussed in Section 1.1.3.

For any given VOC, large variations in its concentration can occur between breath samples taken from an individual. This is due to not only changes in metabolic and physiological processes but also the type of sampling in terms of whether mixed-expired or end-tidal breath is taken for analysis. Preservation of the breath integrity during sampling is probably one of the most demanding challenges in breath analysis.

Three key phases are associated with an exhalation of a VOC. As identified in Figure 1.2, Phase I (A–B) is the anatomical dead-space starting from the end of inhalation (the intensity of which can be above zero if there is, for example, room contamination or if VOCs are emitted from the oral cavity and upper airways). Phase II (B–C) is where the VOC is coming from the upper airways and lungs, mixing with the dead-space gas. Phase III (C–D) contains the VOC coming from the alveoli, with the end-tidal VOC concentration being measured at the end of phase III (D). Phase III is followed by the inspiratory phase (D–E). End-tidal concentrations of a VOC are close to the alveolar concentrations, but how close depends on its interaction with the upper airways, which in turn predominantly depends on a VOC's solubility in mucus (see Chapter 12).

In breath analysis, the end-tidal concentration of a VOC (C_end-tidal) can be defined as the average concentration of a VOC in phase III, namely:

whereas the mixed expired concentration of a VOC (C_{mixed-expired}) can be defined as the average concentration of a VOC in phases I–III, i.e.:

How one can estimate the end-tidal concentration from the mixed expired concentration has been demonstrated for methane by Szabó et al.⁹  using the expression:

where C_A is the methane concentration in alveolar air considering the dead-space (V̇_dead-space) and alveolar (V̇_A) ventilations, and C_I is the methane concentration of inhaled air.⁹  The dead-space volume was estimated at ∼1 ml lb⁻¹ of body weight for adults.

To measure VOCs that originate in the mouth or upper airways, measurements need to be taken through sampling in phase 1 or phase 2, respectively. For blood-borne VOCs with low blood:air partition coefficients (λ_b:air < 10, e.g., isoprene), the end-tidal concentration equals the average alveolar concentration, since there is little interaction with the upper airways (unless production occurs here and/or in the mouth). For blood-borne VOCs with higher values of λ_b:air, it is still advisable to measure the “end-tidal” concentrations because of the improved quality of the data and because it best represents the alveolar concentration. However, this is not always practicable for offline sampling owing to the need to monitor when the start of phase III is reached. This difficulty is removed when taking real-time breath-by-breath measurements, which are also possible with soft chemical ionisation mass spectrometric analytical techniques, such as proton transfer reaction–mass spectrometry (PTR-MS) and selected ion flow tube–mass spectrometry (SIFT-MS), both techniques being useful for targeted VOC analysis and continuous monitoring.^10–12 

Whereas endogenous VOCs are generally products of metabolic activity, exogenous VOCs result from external exposure, including the intake of drugs, a substrate, food or drink, and from the environment through inhalation or skin absorption of chemical compounds that can be metabolized, excreted, exhaled or stored in the body. The combination of exogenous and endogenous VOCs makes up a person's breath volatilome, which is a complex and dynamic system, constantly in flux owing to changes in many internal and external factors, including the person's health status, diet, environmental exposure, medication and physical activity. Examples are breath concentrations of acetone and isoprene increasing as a result of fasting and sudden physical activity, respectively.^13,14 

Continually changing breath profiles can cause problems in unambiguously determining VOC biomarkers for specific medical applications, especially when relying on VOCs that are endogenous in origin. Nevertheless, changes in the concentrations and/or types of exhaled VOCs that make up the volatilome have the powerful potential of acting as noninvasive biomarkers for disease, infections (viral, bacterial and fungal), physical activity, the gut microbiome, the presence of drugs, the effects and efficacy of medication, as well as for determining environmental exposure to chemicals. An example is the exploitation of subtle changes in the volatilome for diagnostic purposes that has been adopted in a breath test (Heartsbreath) used in screening for heart transplant rejection, approved by the Food and Drug Administration in 2004.¹⁵ 

Identifying, detecting and monitoring endogenous VOCs in exhaled breath is very attractive in developing noninvasive screening techniques for the diagnosis and monitoring of diseases or infections. To determine the correct endogenous VOCs and their concentration levels, the discovery phase of any research program invariably involves the comparison of exhaled breath between healthy controls and people with well diagnosed diseases or infections. Even then, to tease out the differences in the VOC patterns, multivariate statistics often have to be employed. Associated with this approach are the problems of exogenous VOCs as confounders of genuine endogenous VOCs, either from the exogenous compound itself or from its metabolic products, and the lack of knowledge on the biochemical pathways leading to the production of endogenous VOCs.¹⁶ 

Despite decades of intense research, breath tests involving endogenous VOCs have found limited use in clinical settings. This highlights the challenges in the use of such VOCs as biomarkers, which include a lack of any unique VOC(s) associated with a given disease and with issues of reproducibility and biological variations (inter- and intra-individual) in their concentration levels. These make the use of changes in concentration levels of breath VOCs as indicators of disease or infection problematic. An example is illustrated with lung cancer; many endogenous VOCs in exhaled breath have been associated with this disease, yet of the numerous reports there has been little overlap in the VOCs identified.^17,18  This deficiency of consistency among individual studies has contributed to the lack of any clinical breath VOC test being developed to specifically detect lung cancer.

Another problem with endogenous VOCs is that the biochemical pathways leading to their production and distribution, as well as their relationships to disease, are often either not known or at best only poorly understood. Without such knowledge, it is difficult to interpret the message the VOCs might bring and which VOCs are genuinely associated with a given disease or infection, limiting their validity and credibility. Investigations of VOCs that are emitted from well characterized single-cell lines can help in developing the underpinning knowledge of the biological processes involved (cf. Chapters 9 and 10), but such studies are limited. Therefore, the link of the origins of exhaled VOCs to specific cellular metabolic processes is often missing. Other challenges are the large normal ranges and variations observed, with factors – age, gender, ethnicity, body mass index and medication – complicating the breath analysis of endogenous VOCs.

Owing to the many difficulties associated with the use of endogenous VOCs in exhaled breath as biomarkers, there is a growing interest in the use of substrates or exogenous VOCs for breath tests. Once taken into the body, they can act as simple probes to the metabolic and physiological processes through the generation or lack of production of unique VOCs. Arguably, this provides the best chance for a much needed bridge to translate breath research to clinical applications and hence the promise of the development of useful tests for use in the health sector.¹⁹ 

Nearly all currently adopted clinical breath tests involve those that generate volatiles following the ingestion of a substrate. In the gut or bloodstream, these substrates are metabolized to produce a known gas, such as hydrogen (H₂), methane (CH₄) or, with the use of ¹³C-labeled substrates, ¹³CO₂.^20–22  Hydrogen and methane breath tests are used with considerable success to assess carbohydrate malabsorption, and, on their own, hydrogen breath tests are used to determine small intestinal bacterial overgrowth. The origin of these gases is from bacterial metabolism of ingested fructose for use in diagnosing malabsorption, and glucose or lactulose to assess small intestinal bacterial overgrowth.

For ¹³C-substrate breath tests, the labeled substrate is selected so that, once metabolized, ¹³CO₂ levels are substantially increased above normal levels in the bloodstream, resulting, in turn, in increased ¹³CO₂ concentration in exhaled breath, which can then be easily measured in real time by infrared spectroscopy. The substrate-derived increase in levels of ¹³CO₂ provide an indirect measure of enzyme activity and hence the health status of the person being tested. A well-known example is the quantitative detection of Helicobacter pylori infection through the ¹³C-urea breath test.²³  Increased ¹³CO₂ production is due to the degradation of ¹³C-urea through the bacterial action of urease.

Two other ¹³CO₂ clinical breath tests are available, one involving ¹³C-labeled Spirulina platensis as a substrate for investigating gastric emptying (gastroparesis).²⁴  It relies on changes in the kinetics of metabolism for diagnostic purposes. The other test involves the use of ¹³C-methacetin for determining liver function by being metabolized to acetaminophen and ¹³CO₂ by the cytochrome P450 1A2 (CYP1A2).²⁵ 

Directly related to the ¹³C-methacetin test are the ¹³CO₂ breath tests that have been evaluated using labeled substrates, ¹³C-dextromethorphan and ¹³C-pantoprazol, for the determination of the efficacy of CYP2D6 and CYP2C19 phenotypes, respectively.^26,27  Although these and other ¹³C-substrate breath tests have been neither clinically developed nor approved, such tests hold much promise for developing personalized medicine by providing a method that readily determines responders and nonresponders for drug metabolism.

Instead of using the biotransformation of a ¹³C-labeled substrate to produce ¹³CO₂ for use in breath tests, an alternative solution is the use of unlabeled exogenous VOCs for assessing in vivo metabolic processes. These exogenous VOCs either are regularly being introduced into the body, for example through food and drink intake or medication, or can be deliberately ingested, absorbed or given intravenously. The concentrations of these compounds can increase above healthy expected ranges because of ineffective metabolism. This takes the direction and focus of breath research away from endogenous VOC discovery for disease diagnosis and instead places breath tests on a firmer scientific and clinical footing through the administration or ingestion of exogenous compounds (intentionally or unintentionally) that specifically target certain enzymes of disease-specific processes. These compounds can directly enter the bloodstream through inhalation/dermal absorption, injection or via the gut, with the compound (if volatile) and/or its metabolic VOC products being directly measured in breath. This method of producing VOC biomarkers in the breath that are not normally present in exhaled breath (or perhaps present only in low pptv levels) provides a pathway for the development of simple and viable specific breath tests. A promising example of this is the presence of limonene in breath for assessing liver metabolism.

Limonene (C₁₀H₁₆) is a monoterpene that enters the body predominantly through food and drink, e.g., from the consumption of citrus fruits and drinks. Once in the bloodstream, limonene should be rapidly metabolized by the P450 enzymes, CYP2C9 and CYP2C19, to perillyl alcohol, trans-carveol and trans-isopiperiten. For healthy people, efficient metabolism results in limonene concentrations in the breath generally found in the low parts per billion by volume (ppbv) concentration levels.

The rapid conversion of limonene in the body needed to maintain low breath concentrations changes dramatically in sufferers from liver cirrhosis. This results in elevated levels of limonene in their breath compared to healthy controls,^28–34  with concentrations that can reach hundreds of ppbv.³²  This observation is attributed to the reduction in limonene metabolism that results from levels of CYP enzymes being produced in a cirrhotic liver that are lower than in a healthy liver, leading in turn to reduced clearance of limonene in the blood. Although some of the nonmetabolized limonene is absorbed into body fat, increased levels in blood occur, which explain the higher limonene concentrations in the exhaled breath of people suffering from a dysfunctional liver. Although higher levels of limonene (and other VOCs) in breath have been associated with cirrhotic patients, their origins are not necessarily a direct result of the disease itself. They could, for example, be biomarkers simply associated with a person being seriously ill and not specific to or resulting from the disease. However, clear proof that limonene and a small number of other key VOCs are directly associated with a cirrhotic liver comes from the study of Fernández del Río et al.³² 

Rather than just relying on a comparison of the end-tidal breath of patients and healthy controls, Fernández del Río et al. used a 3-phase biomarker discovery procedure. Phase 1 consisted of the standard discovery method involving the comparison of breath samples taken from 31 patients with liver cirrhosis and 30 healthy controls to establish a set of putative VOC biomarkers. Nearly all controls were the patients’ own partners, which minimized the influence of any exogenous VOCs present in the home and hospital as possible confounding factors. From the discovery phase, 7 VOCs were elevated in the breath of patients versus the controls. Phase 2 of the study took the discovery to a new level, which consisted of having 12 patients’ breath profiles analyzed pre– and post–liver transplant to investigate intra-individual differences in these VOCs. This led to a hypothesis-led study that used patients as their own controls, thereby reducing the risk of false discovery. Of the 7 potential VOC biomarkers discovered in phase 1, 5 VOCs showed a statistically significant decrease post-transplant: limonene, methanol (CH₃OH), 2-pentanone (C₅H₁₀O), 2-butanone (C₄H₈O) and carbon disulfide (CS₂). Phase 3 of the study involved following the 5 patients longitudinally as inpatients during the post-transplant period. Within days of transplant, the concentrations of 4 of the identified VOCs had reached normal levels in the exhaled breath. Only limonene concentration levels dropped gradually following the liver transplant, reaching normal levels many days later. This is illustrated in Figure 1.3 for the 5 patients. The data presented in this figure provide evidence that limonene had been stored in the body, most probably in fat tissue owing to its lipophilic nature, and that it was gradually released into the bloodstream for metabolic clearance. Thus, following transplant, limonene washout can be used noninvasively to assess the functioning of a liver graft. Overall, Fernández del Río et al.'s study raised the potential of using limonene as a biomarker for early-stage liver disease and pharmacokinetic-based breath tests to assess liver function. The study also highlighted a major advantage of a test involving breath VOCs, namely that they can be used to assess the overall functioning of the liver rather than conducting a localized test to provide a liver biopsy. Importantly, Fernández del Río et al. established a new direction for breath research and potentially new types of tests, namely using unlabeled exogenous VOCs (limonene in the case of liver disease) rather than isotopically labeled VOCs or substrates as probes to explore metabolism and a disease-related imbalance.

As previously mentioned, endogenous VOCs in exhaled breath come from biological processes in the body, with their characteristics in terms of concentrations and types, changing because of chronic disease, infection, metabolic disorder, diet and physical activity (or simply movement, passive or otherwise). Despite the difficulties of identifying these endogenous VOCs and using their footprints as biomarkers possibly correlated to a specific disease or infection, a significant number of studies have reported differences in VOC profiles between sick people and healthy controls.

VOC fingerprinting, i.e., where the VOCs are not necessarily identified, has had some notable success. A fine example is the study of Nakhleh et al.³⁵  Using an artificially intelligent sensor array based on molecularly modified gold nanoparticles and a random network of single-walled carbon nanotubes, they reported the noninvasive diagnosis and classification of 17 diseases from exhaled breath using pattern analysis. More recently, pattern recognition of endogenous VOCs has been effectively used to detect viral infections, e.g., influenza-A and SARS-CoV-2, and in the differentiation of bacteria.^36–39  Other endogenous VOCs have been associated with a number of diseases, which include cancer, asthma, chronic obstructive pulmonary disease, cystic fibrosis and chronic inflammatory diseases.^17,40–42 

The types of breath VOCs that have been identified and associated with some diseases are summarized and discussed by Drabinska et al.²  However, it is worth emphasizing that the use of endogenous VOCs for clinical breath tests is currently limited, owing to issues of the diagnostic specificity of VOCs, their unknown biological origins, uncertain tentative identification, limited or no validation; as previously mentioned, there is lack of reproducibility among the studies. These curb the credibility and usefulness of volatilomics in general and hence significantly hamper the clinical development and adoption of reliable breath tests involving endogenous VOCs. It is therefore not surprising that to date the detection of endogenous VOCs as diagnostic disease biomarkers has achieved only limited success.

Of all the many endogenous VOCs present in exhaled breath, isoprene and acetone need special mention because they are generally at much higher concentrations than found for any other endogenous breath VOC and because their dynamics have been extensively and accurately measured, modeled and simulated. As with all VOCs, the breath dynamics of isoprene and acetone are very much dependent on their solubilities (see Chapter 12). Whereas acetone is highly soluble in blood (λ_b:air ∼ 340),^43,44  isoprene (λ_b:air ∼ 0.95) is not.⁴⁵  This means that, whereas end-tidal breath measurements provide an excellent measure of the alveolar concentration for isoprene, they do not do so for acetone. Isoprene and acetone can therefore be viewed as prototypical examples of 2 major classes of VOCs that represent extremes of the physicochemical properties of breath VOCs.⁴³ 

Isoprene (2-methyl-1,3-butadiene, C₅H₈) is the main hydrocarbon in breath that is endogenously produced in the human body. It is lipophilic and presumed to be correlated with cholesterol biosynthesis.^46–49  However, Sukul et al.⁵⁰  found, by quantitative gene expression analysis of the mevalonate pathway enzymes, that endogenous isoprene does not come from cholesterol synthesis.

Owing to its volatility and low affinity for water and blood, the exchange of endogenous isoprene occurs mainly in the alveolar region, and hence the end-tidal concentration closely represents the alveolar concentration. Exhaled breath has large differences in concentration, with ranges in adults from 70 parts per billion (ppb) up to 133 ppb (25–75% quantile) at rest.⁵⁰  Children have lower levels,⁵¹  and only rarely are isoprene levels deficient or absent in exhaled breath.^47,50–53 

Isoprene has received widespread attention in breath research because of its potential as a sensitive and noninvasive biomarker in detecting and monitoring several metabolic effects. In response to physical exercise or sudden exertion, isoprene (thought to be produced in muscle tissue⁵⁴ ) is suddenly released into the blood circulation, resulting in higher concentrations due to increased blood flow through the muscle. This leads to an almost instantaneous increase of isoprene breath levels, as illustrated in Figure 1.4 with exhaled acetone and CO₂ levels shown for comparison.⁵⁵ 

Isoprene levels in breath are very sensitive to changes in pulmonary ventilation and perfusion due to isoprene's lipophilic behavior and its low blood:air partition coefficient (dimensionless Henry constant), whereas hydrophilic acetone has a more stable behavior. At the onset of any exercise period, isoprene levels in breath increase by three- to fivefold within ∼1 min. Following the first exercise period, the isoprene levels showing lower sharp rises at the start of subsequent exercise periods mean that the loss of isoprene from the muscles tissues during exercise is not rapidly replaced.

King et al.⁵⁴  developed a 3-compartment model that dynamically describes the transition from a state of rest (constant exhaled isoprene concentration, constant cardiac output and constant breath flow) to a new equilibrium state at a moderate constant exercise workload (e.g., at ∼75 W, blood flow approximately doubles and breath flow increases ∼fourfold). For the isoprene breath concentration to reach a new equilibrium takes ∼15 min, whereas blood and breath flow reaches a new equilibrium within ∼1 min. The essence of the King et al. model is that it predicts the main production of isoprene as originating in the muscle tissue. Hence, blood passing through a muscle has a much higher isoprene concentration than blood flowing through other parts of the body. At rest, the contribution of the blood that has passed through the muscle to the mixed venous blood is small (<20%), but when exercise starts, blood flow through the working muscle immediately increases, resulting in a larger contribution (up to ∼80%)of blood containing higher concentrations of isoprene in the mixed venous blood. This results in an overall higher concentration of isoprene in the mixed venous blood. As the production of isoprene in the muscle tissue is constant, the level of isoprene in the blood coming from the muscle decreases to a new steady-state during exercise. Overall, this leads to an isoprene peak in exhaled breath at the onset of exercise (compare this with Figure 4 of King et al.,⁵⁴  with the peripheral compartment containing working muscles).

During sleep, exhaled isoprene concentrations tend to increase by a factor of 2 overnight.^56,57  Recent and real-time unpublished measurements we have made demonstrate that volunteers who had rested and avoided any movements show that this increase can also occur without sleeping.

Isoprene's value as a potential disease marker is very limited. Lower levels of isoprene occur in the exhaled breath of patients suffering from lung cancer compared to those of healthy controls; hence, isoprene could be used as a marker for lung cancer, but there are two problems. First, isoprene is insufficiently specific to lung cancer; e.g., lower levels of isoprene have also been associated with chronic heart failure and muscle dystrophy.^58,59  Second, the concentrations of breath VOCs are altered with changes in breathing patterns, and people suffering from lung disease often have impaired breathing compared to healthy individuals. Since in this case blood flow is unchanged, any drop in the breath isoprene concentration can be explained by the simple Farhi equation (see Chapter 12 for more details). To illustrate this second point, Figure 1.5 shows two data sets for the concentration of end-tidal isoprene in human breath.⁶⁰  The box plots provide a summary of the population data collected from a large clinical study investigating lung cancer.

The results in the box plots in Figure 1.5 seem to imply that lung cancer can be detected based on decreased isoprene levels, with a tentative threshold isoprene concentration of ∼70 ppbv differentiating between people with and without lung cancer. However, the continuous data set shows that the end-tidal concentrations of one individual (in red) can be changed by altering the breathing pattern from normal to a more rapid breathing, which leads to an instantaneous drop in isoprene concentration. This does not mean that isoprene is not a biomarker for lung cancer but that it is not suited to discriminating lung cancer patients from people with healthy lungs. This serves as a warning regarding the difficulties associated with interpreting breath VOC data, which is often aggravated by volunteers tending to hyperventilate when providing a breath sample.⁶¹  As a consequence of these observations, Koc et al.⁶⁰  suggested that a sensible interpretation of any breath VOC emission can only be obtained if information on its variability with changes of ventilation and blood flow is provided. Certainly, assessment of the underlying exhalation kinetics are required for the successful adoption of endogenous VOCs in clinically acceptable breath tests.

Acetone (C₃H₆O) is the most dominant ketone produced by the human body. Therefore, it is always a major constituent of exhaled breath. In comparison to isoprene, acetone is highly water soluble, and hence it will interact with the water-like mucus membranes lining the conductive airways. This means that the end-tidal exhaled breath measurements never provide the true alveolar acetone concentrations (except when using isothermal rebreathing).^62–64  Instead, the end-tidal measurements provide a lower value, by about a factor of 3.⁶⁵  The amount of reduction depends on a number of factors, such as airway temperature profiles, airway perfusion, mucus thickness and volumetric flow.

Acetone is generally present in exhaled breath in greater concentrations than isoprene, with end-tidal levels having been observed in the range of approximately a few hundreds of ppbv up to several tens of parts per million by volume (ppmv). Its production is mainly linked to fatty acid metabolism, although it also comes from cholesterol, urea and glycogen metabolism.^66,67  High levels of exhaled acetone can result from uncontrolled diabetes, and therefore it is tempting to consider developing a breath test to monitor acetone as a biomarker for diabetes. However, given that exhaled acetone is related to fat burning, dieting or fasting will also result in high acetone concentrations. High concentrations are found in the breath of people who are eating high-fat diets compared to those on low-fat diets. This simply reflects increased fat metabolism, resulting from a ketogenic diet.^68,69  In support, patients on a ketogenic diet have elevated levels of acetone in their plasma and brains.⁷⁰  Higher concentrations of plasma acetone results in higher rates of elimination through exhalation, both in starvation (e.g., during sleep) and in diabetes, up to 4 mmol L⁻¹ and 10 mmol L⁻¹, respectively.^71,72  That large intra-individual variation in acetone breath concentration occurring throughout the course of a day means that an acetone breath test on its own would be of little use for monitoring diabetes. However, it could contribute to help in managing diabetes more effectively via better discrimination between hyperglycemia (high glucose levels in the blood) caused by overeating and insufficient insulin, if combined with measurements of blood glucose levels, as currently used to monitor diabetes.^10,73 

Using H₃O⁺ as the reagent ion, soft chemical ionization techniques can monitor protonated breath acetone in real time in order to determine the period over which phase III of the breathing cycle occurs, which can then in turn be used to determine the time intervals for the analysis of other breath VOCs for integration purposes. To illustrate this, Figure 1.6 provides a real-time single-breath profile for an individual recorded using a PTR-ToF-MS with buffered end-tidal (BET) sampling.⁷⁴  The signal recorded at m/z 59.050 is protonated acetone (C₃H₇O⁺).

With regard to the application of breath VOCs, studies by O'Hara et al.^75,76  show that the concentrations of acetone (and isoprene) give a more consistent measure than the analysis of a single aliquot of blood over a short period. Far less intersubject variability in breath levels than in the concentration determined from the blood measurements is seen. Figure 1.7 illustrates this by providing the measured concentrations of acetone in the exhaled breath and blood (radial arterial and peripheral venous) of an individual volunteer over ∼1 h.⁷⁵  The lower variability in breath measurements may have resulted from the large volume of blood that flowed through the lungs during the sampling period.

VOC biomarkers of disease, infection and treatment present in the bloodstream can pass through the barrier between the blood capillaries and alveoli of the lungs and then be released in the exhaled breath. Breath VOCs can also originate from symbiotic and pathogenic bacteria in the body, fungal infections, and the ingestion, inhalation or absorption of compounds into the body. Together, VOCs form specific signatures containing information on biochemical processes occurring in the body. The chemical analysis of these signatures provides a noninvasive window into the human body capable of revealing pathological processes due to diseased or infected states.

Although the physical origins of many VOCs in exhaled breath are well-defined, their biological origins are often less clear for a number of reasons. The breath VOC composition is clearly very complex and affected by numerous confounding factors, such as diet, environmental VOCs, highly contaminated clinical environment, microbiota activity or interindividual variability, making the data noisy and difficult to interpret. Furthermore, breath VOCs have large concentration ranges, and relevant information is frequently provided by less abundant components with low signal-to-noise ratios. These can be affected by noninformative species of high abundance, the extraction of relevant information requiring advanced multivariate statistical methods.

Advancement in our understanding of the biological origins of exhaled VOCs and the many factors that influence end-tidal concentrations should lead to major progress in the field of breath research and lay a much needed foundation to help realize the full potential that VOC breath analysis can bring to the fields of medicine and personalized healthcare.