INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

CONCENTRATIONS OF VOLATILE substances in blood are reflected in their concentrations in the exhaled air because blood concentrations relate to alveolar concentrations by their blood-gas partition coefficient or solubility. Constituents of patients' exhaled gas include volatile anesthetics, acetone, n-pentane, and isoprene. Volatile anesthetics such as isoflurane are exhaled for a long period after exposure during anesthesia (2, 3, 8). Acetone is formed by decarboxylation of acetoacetate, pentane arises from peroxidation of n-6 polyunsaturated fatty acids (6), and isoprene (2-methylbutadiene-1,3) is thought to be formed along the mevalonic pathway of cholesterol synthesis (26). Exhalation of acetone, pentane, and isoprene has been related to metabolic disorders, inflammatory reactions (7, 9, 10, 12-14, 16, 17, 25, 28) and certain diseases such as acute respiratory distress syndrome and pneumonia (24). Results, however, are conflicting, and a systematic and reliable correlation between the composition of patients' exhaled air and their clinical status has not yet been established. Substance concentrations of interest fall in the range of 10¹² to 10⁹ mol/l. Because of these extremely low concentrations, quantitative chemical analysis of mixed exhaled gas may be affected considerably by errors due to dilution and contamination with dead space gas. In mechanically ventilated patients, dead space even comprises parts of the respiratory tubing that may contain additional contaminants. To analyze the relationship between chemical composition of exhaled gas and patients' clinical conditions, substance concentrations in alveolar gas must be determined because only alveolar concentrations reflect concentrations in blood. We therefore developed a sampling method using the expired CO₂ to separate dead space gas from alveolar gas.

The purpose of this study was 1) to evaluate the CO₂-controlled sampling method in mechanically ventilated patients and to test it with respect to efficacy of alveolar gas sampling and 2) to assess the effect of CO₂-controlled sampling on the measurement of exhaled substance concentrations. For this purpose, four volatile substances of different origins were analyzed in the exhaled gas. Two of the volatile substances came predominantly from the patients' alveoli; the other two had measurable concentrations in the inspiratory gas. Results of CO₂-controlled sampling were compared with those obtained by means of a mixed sampling method.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Patients. After approval by the local ethics committee and after informed consent by the patient or next of kin had been obtained, 12 patients (7 men, 5 women) were chosen randomly among the patients of our surgical intensive care unit. Ten patients had received isoflurane as an anesthetic. All patients were under controlled or assisted mechanical ventilation.

Gas sampling. A schematic drawing of the CO₂-controlled sampling device is shown in Fig. 1. The sensor (cuvette) of a fast-responding infrared absorption mainstream CO₂ analyzer (930, Siemens-Elema, Solna, Sweden; Fig. 1, no. 6) was inserted between the Y piece of the respiratory circuit and the patient. A stopcock (Fig. 1, no. 7) was mounted through a bore into the cuvette to prevent leakage of respiratory gas when the sampling system was not connected. The stopcock joined the cuvette with the electrically operated two-way valve (LFYA, Lee) and had an internal dead space of 88 µl (Fig. 1, no. 8). Adsorption traps (27) containing 80 mg of activated charcoal (Analyt, Müllheim, Germany) were mounted onto the two outlets of the valve. By means of a Y piece, the traps were connected to a roller pump (Model 7553-75, Cole-Parmer Instruments, Niles, IL) working at a constant flow of 200 ml/min (Fig. 1, no. 14). Volatile substances were collected and concentrated by adsorption onto the activated charcoal. Approximately 1 liter of exhaled gas was drawn through the traps by means of the roller pump. Sample volume and flow were measured by collecting the gas under water and recording the time of sampling. The preconcentrating effect of the charcoal was ~1,500-fold. One thousand milliliters of exhaled gas passed through the trap. Desorption took place in 2 s, and the gas flow through the trap during this period was 20 ml/min, i.e., the substances adsorbed on the charcoal were liberated into a volume of 2 s × 20 ml/ 60 s = 0.67 ml. Thus the cumulative sample volume-to-measured sample volume ratio was ~1,500/l. Precision and reproducibility of the preconcentration process have been evaluated in several studies (15, 22, 27). Relative standard deviations were 10%, and correlation coefficients for calibration curves were always found to be >0.95.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Schematic drawing of the CO2-controlled sampling device. 1, patient; 2, endotracheal tube; 3 and 4, respiratory tubing; 5, ventilator; 6, CO2 mainstream sensor; 7, stopcock; 8, electrical 2-way valve; 9, CO2 analyzer; 10, electronic processing unit; 11, trap; 12, alveolar sampling (expiratory) position; 13, mixed inspiratory and dead space sampling (inspiratory) position; 14, roller pump.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Tracing of exhaled CO2 concentrations (vol% CO2) and switching points of the valve. A: start of alveolar sampling at a static threshold of 3.5%. B: end of alveolar sampling at 90% of the preceding maximum CO2 concentration. C: maximum CO2 concentration during alveolar phase. I, CO2 free inspiratory phase; II, mixing phase; III, alveolar phase.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Schematic drawing of the sampling devices mounted in the respiratory circuit showing the CO2-triggered sampling device and the T pieces for inspiratory (site 1) and mixed exhaled (site 2) sampling. CO2 conc, CO2 concentration.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Simultaneous tracing of electrical impulses generated by the processing unit for switching the 2-way valve in mV (upper curve) and of analog CO2 analyzer output in mV and vol% CO2, respectively (lower curve) in a patient (8 in Table 1) under assisted mechanical ventilation. a, Beginning and b, end of alveolar sampling. I, CO2 free inspiratory phase; II, mixing phase; III, alveolar phase; CO2max, maximum alveolar CO2 concentration; t, time.

Gas chromatographic analysis. Substances were desorbed from the activated charcoal by means of microwave energy (22), separated by gas chromatography, detected by flame ionization detection, and identified by mass spectrometry. In this study, two substances coming predominantly from the patient [isoflurane (23) and isoprene (24)] and two substances having measurable concentrations in the inspiratory gas [acetone and pentane (24)] were investigated. Substance concentrations were obtained from calibration curves by using flame ionization detection data. Measurements were made at least in duplicate. Time of sampling, ventilator settings, and the patient's diagnoses were recorded. The analytical method has been described in more detail elsewhere (24).

Variation of the measured values, limits of detection, and precision. In a previous study, we performed an extensive testing of the device in the laboratory with respect to precision of CO₂ measurement, speed, and reproducibility of valve switching. Furthermore, we evaluated the relationship between the response time of the intake valve and cyclical pressure changes within the ventilator tubing system. The response time of the valve did not change up to a clinical relevant pressure of 100 mbar. The switching time of the intake valve is 18 ms.

Statistical analysis. Because substance concentrations were not normally distributed, the Friedman test followed by a Student-Newman-Keuls test was applied for multiple comparisons within patient groups. A P value <0.05 was considered significant. Results are given as medians and 25-75% percentiles. CO₂ concentrations determined from sensor recordings were normally distributed, and results are given as means ± SD.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Patients' demographic data, main diagnoses, ventilator type, and mode of mechanical ventilation are listed in Table 1. A reliable identification of the alveolar phase was observed by simultaneously analyzing the CO₂ concentrations measured at the Y piece of the respiratory circuit and the electrical switching impulses of the valve (Figs. 2 and 4). The valve was switched into expiratory position at a mean expiratory CO₂ concentration of 3.3 ± 0.1% (mean ± SD). Switching into the inspiratory position occurred when CO₂ concentration had fallen by 90 ± 3% of the immediately preceding maximum. In patients 6, 7, and 12, estimates of isoflurane arterial concentrations were derived. Ratios of end-tidal to arterial CO₂ partial pressure were 0.9, 1.0, and 0.6, respectively, in these patients. With standard conditions (volume = 22.4 l/mol) assumed, isoflurane blood concentrations were determined as 42.8, 25.7, and 135 nl/ml blood, respectively (blood-gas partition coefficient = 1.4).

The maximum time necessary to sample one liter of alveolar gas was 22 min in a patient who was tachypneic (27 breaths/min).

Tables 2 and 3 summarize the results obtained by means of mixed expiratory sampling and CO₂-controlled sampling. In all patients, isoflurane and isoprene concentrations were higher in the CO₂-controlled samples than in the mixed expired samples (Table 2). Median expired isoflurane and isoprene concentrations were 1.75 (isoflurane) and 2 (isoprene) times higher in the CO₂-controlled samples than in the mixed expired samples (Table 2). Expired acetone and pentane concentrations were not different in the CO₂-controlled and the mixed expired samples (Table 3). Inspired isoprene and isoflurane concentrations were at the limit of detection. Inspired pentane and acetone concentrations were at 10-50% of the expired concentrations.

Limits of detection for n-pentane, isoprene, acetone, and isoflurane were 0.10, 0.20, 0.15, and 0.95 nmol/l, respectively; precision (as relative SD) was 4.05, 2.4, 9.2, and 11.2%, respectively. Calibration plots of all substances were linear in the investigated range. Correlation coefficients were between 0.943 and 0.997. Relative standard deviations of retention times on the gas chromatography column were <0.3% (n = 5).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We found that alveolar gas sampling was reliable and precise with our newly designed CO₂-triggered "alveolar" sampling valve. Alveolar sampling valves were easily controlled via CO₂ concentration. Alveolar gas sampling started at 3.3 ± 0.1 vol% CO₂ (25 ± 0.8 mmHg), which is in the range of mean ± 2 SD of the preset value of 3.5 vol%. By choosing a high value of 3.5% expired CO₂ as the starting point of alveolar sampling, errors due to admixture of dead space gas are minimized. In patients showing a slow rise of expired CO₂ concentration (e.g., in obstructive lung disease) or in patients who have low expired CO₂ concentrations (e.g., during hyperventilation), sampling times are increased due to the late onset of alveolar sampling. If alveolar CO₂ concentration is <3.5%, the device will not collect alveolar gas. In these cases, the threshold would have to be lowered by changing the internal settings of the prototype used for this study. Therefore, an adjustable onset of alveolar sampling is planned for future realizations of the apparatus.

Because no standard analytical method exists in breath analysis against which to compare new findings, it is impossible to compare "tested" values with "known" values. Internal standards can only be used in the analytical process, not in the patient, i.e., it is not possible to produce known substance concentrations in patients' exhaled air. Therefore, we tested the analytical process by using each patient as his or her own control, comparing measurements taken simultaneously with and without the new sampling device in the same patient.

Estimates of arterial tracer gas partial pressures may be derived from the ratio of end-tidal (alveolar) to arterial CO₂ partial pressures. Arterial blood concentrations could then be derived from blood-gas partition coefficients or solubility. These estimates, however, will only be correct for substances that have a similar solubility to that of CO₂ because end-tidal to arterial partial pressure ratios vary with ventilation and pulmonary perfusion according to the solubility of the volatile substances.

Linearity, precision, and limits of detection of the microwave desorption/flame ionization detection method had been evaluated before (15). In a previous laboratory investigation, we did not find any influence of the ventilatory pattern on the performance of our CO₂-controlled valve. However, in the present study, the number of patients was too small to analyze the systematic effects of the disease state on the performance of our setup.

Although the analytical settings were identical, only two of the four volatile substances analyzed in this study were enriched by the CO₂-controlled sampling technique. The different behavior of these exhaled substances is due to different substance origins. As one would expect, alveolar concentrations are higher than mixed expired concentrations only for those substances originating predominantly from the patient, such as isoflurane or isoprene. Isoflurane is exhaled after exposure during anesthesia and could not be detected in the inspiratory gas of the ventilator systems (23). Isoprene is most probably generated along the mevalonic pathway of cholesterol synthesis (26), and concentrations in the gas delivery system were negligible (24). Substances having considerable concentrations in the inspiratory limb, such as acetone or pentane, were not enriched by CO₂-triggered sampling compared with mixed expiratory sampling. Both substances were found in varying concentrations in background air, ventilator tubing, and gas delivery systems (11, 18, 19, 20,). Dead space concentrations of these substances add to alveolar concentrations, resulting in mixed expiratory concentrations equal to or even higher than those measured by CO₂-triggered sampling. As a consequence, the origin of a substance (coming from patient's blood or from the dead space or the delivery system) can be determined by comparing mixed expiratory to alveolar concentrations.

A sampling of the pure inspiratory gas was not possible by means of the valve used for this study. Because of the mode of operation of the valve, the trap mounted in the inspiratory position collected a mixture of inspired and dead space gas. Therefore, the inspiratory gas had to be sampled from the inspiratory part of the circuit without using the valve. In a previous study, we demonstrated that contamination with expiratory gas did not occur at this location in the respiratory tubing (23). High inspiratory pentane and acetone concentrations are due to the fact that these substances occur in large amounts in the ambient air (24). The separation of inspiratory from dead space gas, planned for future uses of the apparatus, could provide more precise information on substance origins and may be used for assessment of pulmonary gas-mixing properties.

During the last decade, numerous trials have been undertaken to establish systematic and reliable relationships between the chemical composition of (human) exhaled air and patients' clinical conditions (8, 12-14, 24, 25). Interpretation of these studies, however, remains difficult because results are conflicting and the variation of measured substance concentrations is considerable. Analyzing mixed expired air collected in gas-sampling bags, Foster et al. (5) observed an increase in isoprene elimination in humans 19 h after exposure to ozone. The underlying mechanisms may be activation of cholesterol synthesis at the onset of repair processes after oxidative damage to fluid linings in the lungs.

Mendis et al. (14) reported an increase in isoprene elimination in patients with an acute myocardial infarction. They proposed a relationship between isoprene elimination and activation of neutrophils. In this case, end-expiratory sampling was used. The decreasing isoprene elimination found in patients with acute respiratory distress syndrome and pneumonia (24) is in contrast to the work cited above. In this study, mixed expiratory samples were taken from the expiratory limb in the respiratory circuit of mechanically ventilated patients. Using a modified end-expiratory sampling method, Philips et al. (21) reported increased pentane and disulfide concentrations in patients with schizophrenia. Pentane is considered a marker of in vivo lipid peroxidation (10). However, concentrations in exhaled gas reported in the literature vary from 10 (1) to 200 pmol/l (4, 11) up to 1.5 nmol/l (24) without a known relationship to patients' specific clinical conditions. Because online control of efficacy of alveolar gas sampling was not previously possible, dead space may have significantly confounded the results and may be the reason for the difficulty in clinical correlation of expired trace gas analysis.

Besides this issue, there is another problem in breath analysis that has not yet been solved. To improve clinical interpretation of the results, the question of how concentrations in the exhaled air can be converted into blood or tissue concentration needs to be addressed. To solve this problem, investigations of substance concentrations in blood and exhaled air must be undertaken, preferably in the controllable setting of an animal model.

In conclusion, the CO₂-triggered valve enables reliable sampling of alveolar gas in mechanically ventilated patients. A distinction of those substances generated from within the patient from those coming out of the delivery system is possible by comparing mixed expired to alveolar substance concentrations. By use of the CO₂-controlled enrichment technique, accuracy of breath analysis is enhanced, and substances occurring in very low concentrations are made accessible to chemical analysis.