Introduction

As a potential non-invasive diagnostic method, breath analysis has attracted a lot of scientific clinical interest during the last decade. Breath testing dates back to the early history of medicine, as ancient physicians knew that the odour of a patient’s breath is altered with some diseases [1]. Modern breath testing began with the work of Linus Pauling in 1971. Pauling used a cold trap to freeze out the breath volatile organic compounds (VOCs), followed by heating the sample and injecting it into a gas chromatograph; he concluded that normal human breath contains many different VOCs in very low concentrations [2]. Pauling’s historical achievement was to provide the first evidence that human breath is an extremely complex gas, and he believed that the analysis of breath VOCs could open a valuable new window into human metabolism and illuminate its functions in health and disease processes [2].

We now know that there are approximately 3,000 VOCs that have been detected at least once in human breath, and most breath samples contain around 200 VOCs, and most of them are present in picomolar concentrations [2]. Breath testing challenges scientists in two ways. First, it is technically very difficult to analyze breath VOCs that are present in picomolar concentrations. Second, the source and biochemical significance of most breath VOCs are still unknown [3, 4].

Breath VOCs are either generated in the body or may be absorbed as contaminants from the environment. VOCs in the body are mainly blood borne, which allows for the monitoring of different processes within the body. Endogenous biomarkers are commonly used for diagnostic purposes [5].

Acetone is one of the most abundant VOCs in human breath. It is produced by hepatocytes via the decarboxylation of excess acetyl-CoA. Acetone is ultimately formed by the decarboxylation of acetoacetate, which is derived from lipolysis or lipid peroxidation. Isoprene is always present in human breath and is thought to be formed along the mevalonic pathway of cholesterol synthesis. Methane and ethane are produced by lipid peroxidation, which is initiated by the removal of an allylic hydrogen atom through reactive oxygen species [5].

Monitoring breath CO2 is very important during the breath analysis process. The respiratory process usually consists of three main events. The first is cellular metabolism; CO2 is generated by metabolic process during which the body utilizes O2. The second is transportation of O2 and CO2 between cells and pulmonary capillaries, and diffusion from or into alveoli. The third is ventilation of lungs between alveoli and atmosphere. All three components of respiration are involved in the appearance of CO2 [6].

Although there has been a lot of progress in furthering the understanding of the origin of breath VOCs, breath testing has not yet been introduced into the clinical practice. The main reason is the difficulty associated with the methodology, including sampling and analysis [5]. Membrane extraction with a sorbent interface (MESI) is a promising new technique that might be one solution to this problem. The MESI system integrates sampling and pre-concentration into one step and was developed to allow for the rapid routine analysis and long-term continuous monitoring of VOCs in various environmental matrices [7, 8]. This paper illustrates the advantages of applying MESI coupled with CO2 monitoring for breath analysis, and introduces a novel calibration method for breath analysis by MESI system.

For membrane extraction with good flow (agitation) conditions in both the sample matrix and the extraction phase (stripping gas) and at constant temperature, the rate-determining step for the mass transfer is the diffusion of the analytes through the membrane material. The concentration of the unknown (C s) at steady-state can be calculated by Eq. (1) as follows [8]:

$$C_{{\text{s}}} = \frac{{bn}}{{B_{{\text{2}}} AD_{{\text{e}}} K_{{{\text{es}}}} t}}$$
(1)

where

b :

is

the thickness of the membrane

n :

the extraction amount

B 2 :

a geometric factor defined by the shape of the membrane

A :

the surface area of membrane

D e :

the diffusion coefficient in the membrane material

K es :

the membrane material/sample matrix distribution constant

t :

the extraction time

In such cases, external calibration or a simpler on-line calibration method [9] exhibits good precision and a wide linear range. Equation (1) can thus be rewritten as:

$$C_{{\text{s}}} = Zn$$
(2)

where \(Z = \frac{b}{{B_{{\text{2}}} AD_{{{\text{es}}}} K_{{{\text{es}}}} t}}\)

Z is a calibration constant determined on the basis of calibration using known concentrations of standard gases when the rate-determining step of mass transfer, as mentioned, is only controlled by the membrane. However, analyte extraction rates into the membrane in real breath sampling may be affected by several variable factors including the velocity of the breath sample, different breath patterns and the sampling temperature [9], thus requiring formal testing of many different sampling conditions for a number of compounds. Otherwise, the calibration data can only be used under specific experimental conditions. However, it is impractical to conduct external calibration studies for all exposure scenarios.

Internal standardization is an important calibration method and can effectively compensate for the difference in extraction rates due to the aforementioned variables. But the standards need to be delivered into the sample matrix, which adds complexity to the sampling device. Meanwhile, it is important to minimize the amount of foreign substances that are added to the investigated system. A conceptually innovative internal calibration method was thus developed to meet the demands of field analyses, in which an analytically non-interfering compound was added to the carrier (stripping) gas to provide an overall correction factor for variations in the MESI extraction rates of the analytes [10].

The theoretical considerations related to the addition of the internal standard to the carrier (stripping) gas to calibrate the analyte extraction process are summarized as follows. The need for correcting the extraction rate is illustrated by studying the effects of sample velocity on the total mass transfer coefficient. The possibility of using an internal standard to correct the extraction rate of the target analyte is demonstrated by the isotropic mass transfer rate between two similar compounds, which were used as an internal standard and a target analyte, respectively. The ratio of the total mass transfer coefficients of the internal standard to the target analyte was calibrated for different experimental scenarios. Finally, acetone was fed with different velocities into the system.

The permeation of analytes through a nonporous polymer membrane is generally described in terms of a “solution-diffusion” mechanism, which consists of seven consecutive steps [8]. Permeation rates are independent of whether the receiving chamber contains a vacuum or a gas other than the penetrant [11]. Therefore, the internal standard, from the carrier (stripping) gas, and the target analyte, from the sample matrix, will permeate simultaneously through the membrane in opposite directions. When the diffusion reaches a steady state, these processes follow Fick’s first law of diffusion, as shown in Fig. 1.

Fig. 1
figure 1

Processes and concentration profiles during target analyte and internal standard permeation through a nonporous membrane

Full size image

For the analyte in the sample matrix [10]:

$$J_{{\text{s}}} = \frac{1}{{{\left( {\frac{1}{{h_{{{\text{c,S}}}} }} + \frac{1}{{h_{{{\text{s,S}}}} }} + \frac{1}{{K_{{\text{S}}} h_{{{\text{m,S}}}} }}} \right)}}}{\left( {C_{{\text{a}}} - C_{{\text{f}}} } \right)} = h_{{{\text{total,S}}}} {\left( {C_{{\text{a}}} - C_{{\text{f}}} } \right)}$$
(3)

where J s is the mass flux of the analyte permeated from the sample matrix to the stripping phase, C a is the concentration of the analyte in the bulk of the sample matrix, and C f is the concentration of the analyte in the bulk of the stripping phase.

\(h_{{{\text{c,S}}}} = \frac{{D_{{{\text{c,S}}}} }}{{\delta _{{{\text{1,S}}}} }}\) is the defined as mass transfer coefficient of the analyte in the inside boundary layer. Mass transfer coefficients of the membrane hm,S , and outside boundary layer hs,S are defined as, \(h_{{{\text{m,S}}}} = \frac{{D_{{{\text{m,S}}}} }}{{\delta _{{{\text{m,S}}}} }},\) and \(h_{{{\text{s,S}}}} = \frac{{D_{{{\text{s,S}}}} }}{{\delta _{{{\text{2,S}}}} }}.\)Ds,S, Dm,S, and Dc,S are diffusion coefficients of the analyte in the sample matrix, membrane, and carrier (stripping) gas, respectively. δ1,S, δm,S, and δ2,S are the thicknesses of the inside boundary layer, the membrane, and the outside boundary layer, respectively. KS is the distribution coefficient of the analyte between the membrane and the gas phase.

htotal,S is the total mass transfer coefficient for the analyte, and it is defined as

$$\frac{1}{{h_{{{\text{total,S}}}} }} = \frac{1}{{h_{{{\text{c,S}}}} }} + \frac{1}{{h_{{{\text{s,S}}}} }} + \frac{1}{{K_{{\text{S}}} h_{{{\text{m,S}}}} }}$$
(4)

Using the same procedure, the flux equation for the internal standard in the carrier (stripping) gas into the sample matrix can be deduced as:

$$J_{{\text{I}}} = \frac{1}{{\frac{1}{{h_{{{\text{c,I}}}} }} + \frac{1}{{h_{{{\text{s,I}}}} }} + \frac{1}{{K_{{\text{I}}} h_{{{\text{m,I}}}} }}}}{\left( {C_{0} - C_{5} } \right)} = h_{{{\text{total,I}}}} {\left( {C_{0} - C_{5} } \right)}$$
(5)

where h c,I, h m,I and h s,I are the mass transfer coefficients of the target internal standard in the inside boundary layer, membrane, and outside boundary layer, respectively.

KI is the distribution constant of the internal standard between the membrane and the gas phase. C0 is the concentration of the internal standard in the bulk of the stripping (carrier gas) phase. C5 is the concentration of the internal standard in the bulk of the sample matrix;

htotal,I is the total mass transfer coefficient for the internal standard and it is defined as:

$$\frac{1}{{h_{{{\text{total,I}}}} }} = \frac{1}{{h_{{{\text{c,I}}}} }} + \frac{1}{{h_{{{\text{s,I}}}} }} + \frac{1}{{K_{{\text{s}}} h_{{{\text{m,I}}}} }}$$
(6)

hc,s, hs,s, hc,I and hs,I are the mass transfer parameters that reflect the effects of the boundary layer between the two sides of the membranes.

In most cases, C 5 and C f remained negligibly small during the experiment. This is because (1)the amounts of penetrant (internal standard or target analyte) were very small, and (2) the carrier (stripping) gas continuously purged the inner surface of the membrane to prevent an increase in the penetrant concentration near the inner surface of the membrane, and the penetrant of the standard was immediately released to the surrounding sample matrix, minimizing the possibility of accumulation on the outer surface of the membrane.

Thus, Eqs. (3) and (5) can be simplified as:

$$J_{{\text{I}}} = \frac{1}{A}\frac{{n_{{\text{I}}} }}{t} = h_{{{\text{total,I}}}} C_{0} $$
(7)
$$J_{{\text{s}}} = \frac{1}{A}\frac{{n_{{\text{s}}} }}{t} = h_{{{\text{total,s}}}} C_{{\text{a}}} $$
(8)

If r is defined as:

$$r = \frac{{h_{{{\text{total,s}}}} }}{{h_{{{\text{total,I}}}} }}$$
(9)

and Eq. (7) is divided by Eq. (8), the target analyte concentration can be expressed as:

$$C_{{\text{a}}} = \frac{{n_{{\text{s}}} }}{{n_{{\text{I}}} r}}C_{0} = \frac{{f_{{\text{s}}} H_{{\text{s}}} }}{{f_{{\text{I}}} H_{{\text{I}}} r}}C_{0} $$
(10)

In Eq. (10), C 0 is the known concentration of the internal standard in the carrier (stripping) gas, r can be calibrated based on different scenarios, which are addressed below, and n s is the extracted amount of the target analyte in the sorbent trap. n I is the permeated amount of internal standard in the matrix, which is the difference between the accumulation of the internal standard with and without the membrane module.

$$r = \frac{{C_{0} }}{{C_{{\text{a}}} }}\frac{{n_{{\text{s}}} }}{{n_{{\text{I}}} }} = \frac{{C_{0} }}{{C_{{\text{a}}} }}\frac{{f_{{\text{s}}} H_{{\text{s}}} }}{{f_{{\text{I}}} H_{{\text{I}}} }}$$
(11)

During calibration in the lab, the r value can be determined by Eq. (11). Both f s and f I are GC calibration factors for the target analyte and the internal standard, respectively, and H s and H I are the peak areas of the target analyte and the internal standard, obtained individually by GC analysis.

Experimental

Materials and instruments

Helium (5.0 ultra high purity, 99.999%), nitrogen (5.0 ultra high purity, 99.999%) and hydrogen (5.0 Polymer, 99.999%) were purchased from Praxair (Kitchener, ON, Canada). Acetone (HPLC grade, ≥99.5%) was purchased from Merck KgaA (Germany). Isoprene (99%) was purchased from Sigma-Aldrich (Oakville, ON, Canada). The plexiglas tubular sampler (124 mL) and the glass vial sampler (250 mL) were custom-made at the University of Waterloo (Waterloo, ON, Canada). Flat sheet silicone polycarbonate membranes SSP-M213 (0.0005”) were purchased from Special Silicone Products (Ballston Spa, NY, USA). Tenax TA 80/100 mesh, Carboxen 569 20/45 mesh, Carboxen 1000 60/80 mesh were purchased from Supelco (Bellefonte, PA, USA). Rtx-U Plot column (30 m×0.25 mm i.d.×1.40 μm df), Hydroguard MXT guard columns & transfer lines (0.28 mm i.d. and 0.53 mm i.d.), coiled Silcosteel tubing (0.53 mm i.d.) and gastight syringes (Hamilton, 1.0 mL) were obtained from Restek (Bellefonte, PA, USA). A two-stage Peltier cooler was purchased from Melcor (Trenton, NJ, USA). The gas chromatograph coupled with FID (Chrompack CP9002) was supplied by Varian (Walnut Creek, CA, USA). A DC power supply (hp Harrison 6427B) from Hewlett Packard (Palo Alto, CA, USA), an electronic thermometer (Fluke 53II) from Fluke Corporation (Everett, WA, USA) and an electronic flow meter (ADM 2000 Intelligent flow meter) from J & W Scientific (Folsom, CA, USA) were also used. The membrane module, the power supply for the cooler (S & D - 066), the temperature controller for the DC power supply (S & D - 070) and the heating timer (S & D -073) were custom-made by the Science Shop of the University of Waterloo (Waterloo, ON). The NICO CO2 monitor, based on a CAPNOSTAT CO2 sensor was generously provided by Respironics Novametrix.

MESI coupled with CO2 monitoring system

The MESI system consists of five major sections: the membrane extraction module, the CO2 monitoring system, the trap and thermal desorption sorbent interface, the separation and detection system (GC/FID), and the data acquisition system (Fig. 2). The detailed structure of the MESI system has been described previously [12, 13].

Fig. 2
figure 2

Schematic of MESI-GC coupled with CO2 monitoring system

Full size image

Standard gas generating system

Since most breath VOCs are present in the picomolar concentration range, low concentration gas standards are required. One of the best methods for producing low concentration gas standards is using a permeation tube method. The analyte is held within a container, which usually is a polytetrafluoroethylene (PTFE) tube. The Teflon tube works as a permeable membrane. As the constant dilution flow passes through the permeation tube, a stable dynamic trace standard gas is generated.

In order to calibrate the dynamic standard gas generating system, the outlet of the permeation tube was directly connected with the sorbent trap, and the membrane module was bypassed. The flow rate of the carrier gas was 3.86 ml/min and was controlled by an electronic pressure control. The amount of the standard gas can be calibrated by the relationship of the peak area and the accumulation time of the standard gas on the sorbent.

At a flow rate of 3.86 ml/min, after 17,160 min the mass loss of the permeation tube was 4.0 mg, therefore the concentration of the internal standard gas pentane was 58.8 ng/ml. Consequently, the amount of pentane accumulated on the sorbent trap can be calculated by the accumulation time of pentane on the sorbent trap at a flow rate of 3.86 ml/min. For example, when the accumulation time of pentane on the sorbent trap is 2 min, the amount of pentane trapped on the sorbent tube is 58.9 ng/ml×3.86 ml/min×2 min=0.454 μg. By the same method, after 43,200 min, the mass loss was 2.0 mg while the flow rate was 4.0 ml/min, and the concentration of acetone was 11.57 ng/ml.

Experimental conditions

Helium (as both carrier gas and stripping gas) flow rate was 5.0 mL/min. The GC/FID oven temperature was isothermal at 100 °C. The trap cooling temperature ranged from −15 to +20.5 °C. Desorption heating parameters were 150–200 °C for 60 s for acetone and isoprene. The Carboxen sorbent trap was conditioned on-line at up to 180 °C by several 60-s desorptions in pure nitrogen gas until a sufficiently low contaminant background was achieved. When trace quantities are to be determined, the background pattern becomes important. The schematic of MESI-GC with an internal standard system is illustrated in Fig. 3.

Fig. 3
figure 3

Schematic of MESI-GC internal standard calibration system

Full size image

An electronic pressure control was set in front of the pentane permeation tube, so the constant flow passed through the permeation tube, and a constant concentration of the internal standard gas pentane was obtained. A similar design was used for the acetone permeation tube, and a constant concentration of analyte acetone was obtained as well. Different flow rates of acetone were acquired by introducing a split valve in front of the sampling chamber.

After setting up the experimental system, a stable base line was obtained. The membrane module was immersed in the N2 environment and the background of the system was very clean. There was no background noise when a heating pulse was applied to the trap module. These results illustrate that the membrane and the trap were clean and the instrument worked properly.

Results and discussion

The optimization of the flow rate of the feeding side to the MESI system

The acetone flow rate was changed with the split valve located on the inlet of the sampling chamber. The concentration of acetone was 0.64 ng/min. The acetone flow rate that passed through the membrane was controlled from 0.25 to 73 ml/min, and the linear velocity was 0.0014 and 0.40 cm/s, respectively. Figure 4 provides a plot of the amount of internal standard pentane vs. the flow rate of acetone.

Fig. 4
figure 4

Effect of the flow rate of analyte to extraction amount of internal standard pentane and analyte sample acetone

Full size image

From Fig. 4 we can see that the amount of internal standard pentane decreased dramatically with an increase in the acetone flow rate through the membrane. This is especially true when the flow rate was less than 5 ml/min, which corresponded to a linear velocity of 0.028 cm/s. However, when the flow rate was greater than 25 ml/min, the corresponding linear velocity was 0.14 cm/s, and the change in the internal standard was not obvious. This was due to the effect of the flow rate on the sampling side. The resistance of the analyte that permeated through the membrane consisted of three parts: the boundary layer of the outside of the membrane, the membrane, and the boundary layer of the inside of the membrane. When the membrane and temperature were chosen, the velocity of the carrier gas was fixed, and the mass transfer of the internal standard in the membrane and the boundary layer inside of the membrane remained the same. The mass transfer of the internal standard changed when the velocity of the sampling side was altered, especially when the flow rate was less than 5 ml/min, while the corresponding linear velocity was under 0.028 cm/s, and therefore additional internal standard pentane permeated through the membrane while the velocity of the sampling side increased. As a result, the internal standard that remained in the carrier gas decreased, and therefore the amount of internal standard decreased when the velocity of the sampling side increased. However, when the flow rate was greater than 25 ml/min, the corresponding linear velocity was greater than 0.14 cm/s, the boundary layer effect on the outside of the membrane was relatively small, the main resistance of the internal standard pentane permeating through the membrane was determined by the membrane and the boundary layer inside the membrane; therefore, the difference in the signal change of the internal standard was not obvious. The same results were observed with the analyte acetone profile.

The optimization of the flow rate of the stripping side to the MESI system

The effect of the different stripping gas velocities on the MESI system was also tested. The different velocities of the stripping gas can be obtained by increasing the pressure of the electronic pressure control system. Since the stripping gas also worked as the carrier gas, the maximum flow rate that could be achieved was 5.68 ml/min, while the velocity passing through the membrane at the stripping side was 23.67 cm/s.

Figure 5 illustrates the plot of the change in the internal standard pentane at different stripping-side gas velocities and different feeding-side gas velocities. It illustrates that increasing the stripping-side velocity did not change the amount of pentane that permeated through the membrane. This is mainly because the gas velocity of the stripping side is over 100 times higher than the gas velocity of the feeding side. The boundary layer effect is not obvious. The mass transfer is not determined by the resistance of the inside membrane boundary layer. The main resistance is the membrane itself and the outside membrane boundary layer. Therefore, by increasing the gas velocity of the feeding side, additional pentane permeated through the membrane. However, when the flow rate of the feeding side was greater than 25 ml/min, the corresponding linear velocity was greater than 0.14 cm/s, and the boundary layer effect outside the membrane was relatively small. The main resistance of the internal standard pentane permeating through the membrane was determined by the membrane itself, therefore, the difference in the signal change of the internal standard is not obvious.

Fig. 5
figure 5

Effect of stripping-side gas velocity on the extraction amount of internal standard pentane

Full size image

The effect of different stripping-side gas velocities on the analyte acetone was also evaluated. Figure 6 is the graph of the change in the mass of acetone at different stripping-side gas velocities and different feeding-side gas velocities. From this figure, we observe that by increasing the gas velocity of the stripping side, the amount of acetone that permeated through the membrane also increased. This is because the inside membrane boundary layer will decrease while the velocity of the stripping gas increases, therefore, the resistance of the inside membrane decreases, so there is more analyte acetone permeating through the membrane. However, the stripping gas also worked as the carrier gas, and when the gas velocity was 23.67 cm/s, the flow rate was as high as 5.68 ml/min, so the optimized stripping-side velocity was 16.17 cm/s, and the corresponding carrier gas flow rate was 3.88 ml/min. If the flow rate is less than 3.88 ml/min, the resistance of the inside membrane boundary layer will increase, and the retention time of the analyte will also increase.

Fig. 6
figure 6

Effect of stripping-side gas velocity on the extraction amount of acetone (analyte)

Full size image

Effect of agitation on the internal calibration system

The effect of agitating the MESI system was also evaluated, by placing an electric fan into the sampling chamber. The stripping-side gas velocity was set at 16.17 cm/s, the corresponding carrier gas flow rate was 3.88 ml/min. The effect of the agitation on the internal standard pentane and analyte acetone was tested at different feeding-side gas velocities.

Figure 7 provides a comparison of the effect of agitation and the absence of agitation on the MESI internal standard calibration system. The amount of internal standard pentane decreased by 53% after the fan was applied to the membrane module, and it is shown that more internal standard permeated through the membrane after the agitation. It is also quite obvious that when there was no fan, at a flow rate of about 10 ml/min for acetone and a corresponding feeding-side gas velocity of 0.056 cm/s, the amounts of pentane and acetone were equal. After this point, the change in the amount of pentane and acetone is not obvious when the flow rate increases; however, this point was changed to 6 ml/min, corresponding to a feeding-side gas velocity of 0.033 cm/s when the fan was applied to the membrane module. This is due to the effect of agitation on the mass transfer; agitation will decrease the boundary layer. Thus, additional analyte will permeate through the membrane, and additional internal standard will permeate through the membrane when there is agitation.

Fig. 7
figure 7

Comparison of the effect of agitation on the extraction amount of internal standard pentane and analyte acetone

Full size image

R values at different feeding-side gas velocities and different stripping-side gas velocities

The R value is the ratio of the amount of internal standard pentane that permeated through the membrane and the amount of the analyte acetone that permeated through the membrane.

Figure 8 is the relationship of the R value and the feeding-side gas velocity. It was observed that the R value was constant when the feeding-side gas velocity was changed, but slightly decreased when the stripping-side gas velocity was increased. The internal standard calibration method is an effective way to calibrate the MESI system. As the results show, the ratio of the amount of internal standard that permeated through the membrane to the amount of the analyte that permeated through the membrane was constant. The minimum feeding-side gas velocity was 0.14 cm/s, and the corresponding flow rate was 25 ml/min. The optimized stripping-side gas velocity was 16.17 cm/s, and the corresponding carrier gas flow rate was 3.88 ml/min.

Fig. 8
figure 8

R values at different stripping-side gas velocities

Full size image

Breath acetone profile

Breath acetone was quantitatively analyzed by the MESI-GC coupled with a CO2 monitoring system. A subject was asked to inhale and exhale into the sampling chamber. At the same time, the end tidal pressure of CO2 was monitored. The breath sample was collected in the sampling chamber. The extraction time was 20 min, followed by a heating pulse that was applied to the sorbent trap. The acetone chromatogram was obtained using an FID. The levels of breath acetone were compared at different end tidal pressures of CO2. Figure 9 illustrates the breath acetone profile of a healthy male at different ETCO2 pressures (end tidal CO2 pressure). The linear correlation between acetone and ETCO2 pressure was demonstrated. This linear relationship illustrates that CO2 can be used as an internal standard to monitor breath VOCs at different breathing patterns. The ratio of the amount of breath acetone (ng) and the concentration of CO2 is shown in Fig. 10.

$$\frac{{{\text{Amount}}\,{\text{of}}\,{\text{breath}}\,{\text{acetone}}}}{{{\text{ET}}\,{\text{CO}}_{{\text{2}}} \,{\text{pressure}}}} = C$$
Fig. 9
figure 9

Breath acetone profile of a healthy male at different ETCO2 pressures

Full size image
Fig. 10
figure 10

Ratio of amount of breath acetone to the concentration of CO2

Full size image

The ratio of the amount of breath acetone (ng) and the concentration of CO2 is a constant value. After the use of CO2 as the internal standard normalization has been validated with the higher concentration analytes, such as acetone or isoprene, the technique can be applied in the analysis of other breath VOCs.

Breath acetone profiles of a male and female at different times during the day

Acetone is formed by the decarboxylation of acetoacetate, which is derived from lipolysis or lipid peroxidation. Ketone bodies such as acetone are oxidized via the Krebs cycle in peripheral tissue [3, 7]. Ketone bodies in the blood (including acetoacetate and β-hydroxybutyrate) are elevated in ketonemic subjects during times of fasting or starving or during diet. Breath acetone concentrations are increased in patients with uncontrolled diabetes. As acetone is produced by the spontaneous decarboxylation of acetoacetate, it is impossible to quantify the fraction that arises from lipid peroxidation.

In this study, a healthy male subject was asked to inhale and then exhale air into the sample chamber through the inlet nozzle. Each time the subject was asked to exhale as much as possible until the maximum ETCO2 pressure was reached (43 mmHg). The breath samples were collected at 8 am (after breakfast), 1 pm (before lunch), and 8 pm (after dinner). A similar experiment was also conducted with a female subject. A healthy female subject was asked to inhale and then exhale air into the sample chamber through the inlet nozzle. Each time the subject was asked to exhale as much as possible until the maximum ETCO2 pressure was reached (44 mmHg). The breath samples were collected at 10 am (after breakfast), 1 pm (before lunch), and 5 pm (before dinner). Figure 11 illustrates the breath acetone profiles of a healthy male and a healthy female at different times of the day.

Fig. 11
figure 11

Comparison of the breath acetone of a male and female at different times of a day

Full size image

From Fig. 11, we observe that the concentration of breath acetone increased when the subject had fasted, and decreased after dinner, which is consistent with the literature evidence that supports the use of acetone as a diet biomarker.

Different subject breaths were also tested for different diet conditions, and the subjects were asked to breath as much as possible for these experiments (the ETCO2 was reached at about 45 mmHg). The breath samples were collected 3 h after breakfast and 1 h after lunch. From Table 1 and Fig. 12, we observe that these subjects had different acetones level in their breath. It is quite obvious that the acetone level decreased after the consumption of food, and the results show that membrane extraction with a sorbent interface (MESI) coupled with a CO2 monitoring system is a simple, rapid, sensitive and solvent-free method for the determination of low concentrations of acetone in human breath samples.

Table 1 Breath acetone monitoring
Full size table
Fig. 12
figure 12

Breath acetone monitoring for five volunteers

Full size image

Conclusion

Breath analysis is an attractive, non-invasive procedure for medical diagnosis, drug monitoring and employee screening. It has been used in numerous laboratory-based studies and for field research. Despite the obvious advantages it presents for routine biological monitoring, it has not been widely accepted as a tool in occupational hygiene. Suitable sampling methods and measurement techniques are the bottleneck. Overall, experts feel breath tests hold promise for the future. In 2001, Schubert noted [14], “We have a problem of standardization, defining markers, and sampling methods. But the prospects are so exciting, I think it’s worthwhile to continue and do the basic research”.

MESI has several unique advantages, which can overcome the problems encountered in breath analysis, such as low concentrations of components and high moisture content of the human breath. Therefore, MESI is a promising technique for breath analysis.

The internal standard calibration method is an effective way to calibrate the MESI system. As the results indicate, the ratio of the amount of internal standard that permeated through the membrane and the amount of the analyte that permeated through the membrane is constant. The minimum feeding-side gas velocity was 0.14 cm/s, and the corresponding flow rate was 25 ml/min. The optimized stripping-side gas velocity was 16.17 cm/s, and the corresponding carrier gas flow rate was 3.88 ml/min.

The use of breath CO2 as an internal standard for the analysis of breath VOCs is an effective method to solve the difficulties associated with variations in the analyte concentration in a breath sample that are due to mass losses and different breathing patterns of different subjects. The concentration of breath acetone was successfully expressed normalized to CO2 as in the alveolar air. The breath acetone profiles of a healthy male and a healthy female at different times of the day were plotted by the MESI system, and results are consistent with the literature. In summary, this technique can be used for the monitoring of the breath acetone concentration of diabetic patients and for other biomarker monitoring.