Since long, it has been known that the human breath contains clues to many diseases. For example, the odor of acetone is a well known sign of increasing ketoacidosis in diabetes. However, breath from healthy humans also contains a number of short chain hydrocarbons, some of which simply represent exhalation of previously inhaled air pollutants, whereas others appear to derive from metabolic or other endogenous processes in the tissues. Among the latter type are pentane and ethane, stemming from peroxidation of fatty acids and thus representing oxidative tissue damage(13). Accordingly, several studies have reported elevated breath pentane in disease states where there is inflammation or tissue degradation(48). However, the most prominent volatile hydrocarbon in human expired air has been identified as isoprene (2-methyl-1,3-butadiene)(911). Although the physiologic role of isoprene is unknown, studies in animals, plants, and humans suggest that this unsaturated hydrocarbon is produced through the mevalonic acid pathway of cholesterol biosynthesis(12,13). On the other hand, attempts to correlate breath isoprene and cholesterol levels in humans were recently reported unsuccessful(14), and patients with acute myocardial infarction were found to have elevated breath levels not of pentane, but of isoprene(15). Recently, breath isoprene was proposed as a noninvasive marker of free radical-induced injury to epithelial membranes of the respiratory tract(16). This raises the question whether breath isoprene might be increased also in diabetes, as diabetes has been shown to be accompanied by free radical-induced injury to tissues(17).

One problem with several reports dealing with the analysis of exhaled volatile hydrocarbons is the failure to distinguish between isoprene and pentane. Because isoprene and pentane coelute on most chromatography columns(18,19), many previous reports may have unknowingly reported isoprene as the combination of isoprene and pentane. We recently described a highly sensitive and specific method for the analysis of isoprene in human breath(20), based on gas chromatography with diode array UV detection(21). With this method, unequivocal identification of isoprene is possible by matching the UV absorption spectrum of the biologic sample with an authentic isoprene standard. Moreover, with this method of analysis, breath isoprene cannot be confused with breath pentane, because saturated hydrocarbons such as pentane and ethane do not absorb UV radiation in the same range of wavelengths as does isoprene.

To test the hypothesis that breath isoprene might be increased in diabetes, we have now used this method to determine isoprene in breath of healthy children of different ages and diabetic children of different metabolic states. Moreover, to examine whether breath isoprene is influenced by metabolic state in the neonatal period, breath isoprene was determined in newborns of different metabolic states (catabolic and anabolic). Acetone was determined in the same breath samples, thereby making it possible to compare the exhalation of isoprene with that of a well established metabolic marker. The study was approved by the Ethics Committee of the Faculty of Health Sciences.


Subjects. The children participating in the investigation were grouped into four categories, referred to as newborn infants, preschool children, school children, and children with diabetes. As to newborn infants, expired air from eight full-term babies was analyzed. All newborns were healthy, without perinatal complications, and with birth weights appropriate for gestational age. Spontaneously expired air was collected after 1-3 d(when postnatal weight loss was between 3.0 and 10.8%) and on d 4-7 (as weight had increased by 1.1-6.4%). In 23 healthy pre-school children, aged 1.5 mo to 6 y, samples of expired air were taken at an ordinary check-up at the well baby clinic.

Expired air was also analyzed from 18 healthy school children, aged 8-16 y, and 13 diabetic children and adolescents, aged 9-20 y. The diabetic children had had insulin-dependent diabetes mellitus for 2-16 y and were 3-15 y of age at the time of diagnosis. In seven cases the exhaled air was sampled at an ordinary visit to the office, with the patients in their habitual states, having eaten and taken their ordinary insulin doses. Their glycosylated hemoglobin at those visits varied between 6.3 and 12.3% (normal range <5.4%). In eight cases (including one of the girls and one of the boys sampled at the office), exhaled air was collected when the patients were fasting, and then again 60 and 120 min after they had eaten a breakfast but not taken any insulin that morning. This subgroup of patients had a diabetes duration of 2-5 y. In the fasting state without any insulin they would be expected to be in a catabolic state with increased acetone production, at least those five patients with abnormal long-term metabolic control reflected by glycosylated Hb values between 6.4 and 12.3%.

Breath air sampling. A pilot study showed that holding the breath for about 1 s caused steady state levels of both isoprene and acetone that were the same as after holding the breath for up to 10 s. In school age and diabetic children the breath collection procedure therefore was as follows. After holding the breath for 2 s, each subject exhaled into a 5-L Teflon (PTFE) bag. Each child exhaled only once into the PTFE bag, producing a gas volume between 250 and 1000 mL, depending on the size of the child. In children less than 1 y old, a sampling method involving a breath ventilation equipment was used. In these cases, the subject was breathing normally in a face mask connected to a ventilation nonrebreathing valve system so that it was possible to collect the exhaled air into a PTFE bag. The dead space volume for this system was less than 15 mL. Sampling procedure for these children meant collecting several tidal breaths during quiet breathing.

In a second stage, the compounds to be analyzed (isoprene and acetone) were collected from the exhaled breath air sample onto a solid sorbent tube. Aliquots of breath air (250 mL) were drawn from the collecting bag, and compounds in the breath were trapped on an adsorbent tube containing two different adsorbent materials, Tenax TA and Anasorb CMS (Alltech, Deerfield, IL, and SKC, Inc., Eightyfour, PA). To prevent acetone loss due to dissolution in condensed water, the bag was kept at 37°C during this procedure. The adsorbent tubes were then transferred to the laboratory for immediate analysis.

Analysis. The chemical determination of isoprene and acetone was made with a recently developed technique that has been described in detail elsewhere(21). In brief, this technique involves gas chromatographic separation and UV detection and identification by way of a diode array spectrophotometer. The equipment used was an INSCAN 175 gas chromatograph/UV spectrophotometer (INSCAN AB, Järfälla, Sweden). The adsorbent tubes were connected to the carrier gas line and placed in the thermal desorption oven of the instrument. The temperature of the oven was kept at 190°C. After a time delay of 1 min, the desorbed compounds were flushed onto the separation column of the instrument by means of opening the valve for the carrier gas flow. The separation column was an INpolA, which is a porous polymer type of column suitable for low boiling organic compounds and inorganic gases. The carrier gas flow rate was kept at 25 mL/min, whereas the temperature program was a linear ramp starting at 50°C with an increment of 10°C/min. The overall separation time was 6 min; during that time UV spectra were recorded every 4th s. A typical three-dimensional chromatogram of breath isoprene and acetone is shown in Figure 1.

Figure 1
figure 1

Gas chromatography-UV analysis of isoprene and acetone in exhaled breath. The three-dimensional plot shows 1) the UV-spectral wavelength (in nanometers) along the x axis, 2) the absorbance of the two compounds(proportional to concentration) along the y axis, and 3) the retention time (in seconds) of the separated compounds along the z axis.

Quantifications were made from the chromatograms formed at 214.5 nm(isoprene) and 193.9 nm (acetone). The integrated values from the sample analysis were compared with those of standards injected directly into the instrument. Acetone and isoprene air vapor standards were prepared from stock solutions containing 0.2% (vol/vol) acetone and 0.02% (vol/vol) isoprene in hexane. Standard curves were obtained by injecting exactly measured volumes(1, 2, 3, and 4 µL) of this solution into the gas chromatograph for analysis by diode array UV detection. The detection response was linear over the concentration range encountered in breath samples. Moreover, when ambient room air (250 mL) was analyzed in control experiments, acetone and isoprene could not be detected.

Repeated analysis of the same breath sample showed a variation coefficient of 9% both for isoprene and acetone. Concentrations of isoprene and acetone in the Teflon bag were stable for 1 d but decreased thereafter; adsorption of the compounds onto the adsorbent tubes was therefore performed within 1 d after the breath sampling.

All chemicals and solvents used throughout the investigation were of analytical grade and were supplied by Merck (Darmstadt, Germany). Isoprene(99% purity) was purchased from Sigma Chemical Co. (St. Louis, MO).

Statistical determinations. Significance of differences, given as p values, were calculated according to Wilcoxon signed rank test (for related samples) and Wilcoxon test (for independent samples) using the SPSS computer package (SPSS Inc., Chicago, IL). Relations between age and breath concentrations were analyzed by linear regression, age being the independent variable.


Isoprene. The concentrations of isoprene and acetone in expired air from healthy and diabetic children are illustrated in Table 1 and Figures 2 and 3. Newborn babies had undetectable or very low levels of isoprene in exhaled breath, but after only a few weeks there was in general a somewhat larger excretion. At a few months of age, increasing concentrations of isoprene were found in the breath, and there was a tendency toward higher concentrations with increasing age (r = 0.55; p< 0.05) (Fig. 2). Accordingly, healthy school children showed significantly higher values than healthy preschool children(p = 0.038).

Table 1 Acetone and isoprene concentration in exhaled breath from healthy and diabetic children
Figure 2
figure 2

Isoprene levels in exhaled breath of 41 different-aged healthy children (23 preschool and 18 school children). Shown are the results of linear regression analysis of the relation between age and breath isoprene, age being the independent variable.

Figure 3
figure 3

Acetone levels in exhaled breath of 41 different-aged healthy children (23 preschool and 18 school children). Shown are the results of linear regression analysis of the relation between age and breath acetone, age being the independent variable.

No differences in breath isoprene were found neither between newborn babies in anabolic and catabolic states (p = 0.463) nor between diabetic and nondiabetic children (p = 0.229). Moreover, breath isoprene concentrations were about the same in diabetic patients regardless of whether they were fasting or had received insulin and food (p = 0.674).

Acetone. Figure 3 shows breath acetone in healthy children of different ages. There is a tendency toward lower concentrations with increasing age. Healthy school children had lower values than healthy preschool children (p = 0.010).

Newborn babies in catabolic states exhaled higher concentrations of acetone than did newborns in anabolic states (p = 0.012). Furthermore, diabetic children had much higher concentrations than did healthy children (p = 0.001). A further increase in breath acetone was found in some of the diabetic children after the breakfast load.


In this report we describe a new technique by which isoprene and acetone can be determined in exhaled breath from children. Whereas breath acetone is known to correlate to metabolic state in, e.g. diabetes, breath isoprene has not yet been clearly associated with the pathogenesis or pathology of a disease. As shown in Table 1, breath acetone apparently correlates to metabolic state in newborn babies. The most dynamic physiologic change in body weight and composition over a short period of time probably occurs in the immediate newborn period, when weight losses up to 10% of body weight within a few days can be normal. As illustrated in the present investigation, this metabolic change during the postnatal adaptive phase is reflected by a parallel change of acetone in exhaled breath. In addition, even reasonably well controlled, nonfasting, insulin-treated diabetic patients at their regular visits to the office showed increased concentrations of acetone in their breath compared with healthy control children (Table 1). Diabetic children who had fasted overnight and not taken any prebreakfast insulin showed a further increase in breath acetone, still without having ketonuria. After having had a standardized breakfast meal, but still no insulin injection, some of them had even higher breath acetone concentrations. Thus, as expected, breath acetone correlated to the metabolic state in diabetic children. These findings show that our technique for measuring breath acetone is not only noninvasive and extremely simple for the patient, but also sensitive.

As to the levels of isoprene, we found undetectable or very low levels in exhaled breath from newborn babies. These results confirm the early findings of Jansson and Larsson(9). Interestingly, these pioneer investigators reported that no isoprene at all were found in samples of expired air from newborn babies (up to 1 wk old). After only a few weeks of age we found increasing excretion of breath isoprene, and our data suggest a relation between breath isoprene concentration and age (Fig. 2). Diabetic children had a rather constant breath isoprene concentration irrespective of metabolic state, and of similar magnitude as healthy children(Table 1).

The biologic implications of breath isoprene remain to be elucidated. Earlier reports have suggested that the amount of isoprene eliminated via the breath is not influenced by diet, age, sex, or fasting(11), but is influenced by sleep patterns(22,23). Isoprene is considered to be an endogenous compound(24,25), but it is also affected by exposure to exogenous agents, such as tobacco smoke(26,27) or ozone(16), and was recently demonstrated to induce multiple organ neoplasia in rodents after inhalation exposure(28). Thus, the biosynthesis, turnover, and excretion pattern of isoprene might reflect some fundamental biochemical process that still remains to be explored. Moreover, even if isoprene has been determined in human blood(23,25) it is not yet clear whether breath isoprene reflects the concentration in blood circulating through the lungs or if it is generated locally in the lungs and/or upper airways.

In summary, we have thus demonstrated a technique by which components in exhaled breath can be determined in a simple way. The gas chromatography-UV absorption method is a new approach for biochemical analysis of volatile organic compounds which combines good sensitivity with high specificity. The gas phase UV spectrum represents a fingerprint of the molecule and unequivocal identification of a molecule is possible by matching the UV absorption spectrum of the breath sample with an authentic standard. This technique may have clinical applications for studying disorders associated with the exhalation of volatile compounds. For example, a practical and noninvasive method revealing slight ketosis in diabetes would be helpful. Breath acetone levels may also be a useful indicator in other catabolic situations such as during intensive care, after surgery and after cancer treatment. Intensive care patients on ventilator represent a major challenge, not only because of the severe illness per se but also because the oxygen treatment and the parenteral nutrition may influence the metabolic state. Hence, our technique for measuring isoprene and acetone in exhaled breath may open up new possibilities in dealing with these clinically important problems.