Respiratory Mechanics during Xenon Anesthesia in Pigs : Comparison with Nitrous Oxide

The study design was approved by the institutional and federal animal care committee (Tuebingen, Germany). Eighteen pigs (mean body weight 41 ± 4 kg), randomly assigned to receive either N²O or xenon (n = 9 for each group), were anesthetized with pentobarbital sodium (Nembutal, Sanofi-Wintrop, Munich, Germany; 15 mg/kg induction dose, followed by a continuous infusion of 6–12 mg · kg⁻¹· h⁻¹) supplemented every 4 h and before any surgical or noxious stimuli by 0.3-mg intravenous boluses of buphrenorphine (Temgesic; Boehringer Mannheim, Mannheim, Germany) to prevent an increase in heart rate and blood pressure. The adequacy of this anesthetic procedure has been tested in previous experiments. 16To eliminate any possible respiratory effort, the animals were paralyzed using alcuronium dichloride (Alloferin, Hoffmann–LaRoche, Basel, Switzerland; 0.25 mg/kg initial dose followed by a continuous application of 0.25 mg · kg⁻¹· h⁻¹) throughout the study period. Depth of anesthesia was controlled by hemodynamic variables and continuous electroencephalographic monitoring (Neurotrac; Interspec Inc., Conshohocken, PA). The spectral edge frequency was always below 15 Hz, and the median power frequency was 5–10 Hz.

After induction of anesthesia, the pigs were intubated with a cuffed endotracheal tube (8.5 mm ID), and their lungs were mechanically ventilated while in the supine position using a semiclosed circuit anesthesia machine (Cicero; Drägerwerk AG, Lübeck, Germany) with a carbon dioxide absorber placed in the inspiratory limb of the breathing circuit. A central venous catheter and a thermistor-tipped pulmonary artery flotation catheter (model 93A 754 7F; Baxter Healthcare, Irvine, CA) were placed through an external jugular vein into the superior vena cava and into a pulmonary artery, respectively, for drug infusion, monitoring, and data sampling of hemodynamics and gas exchange. A 4-French catheter was inserted into a femoral artery for hemodynamic monitoring and arterial blood gas sampling. Mixed venous and arterial blood samples were analyzed for oxygen tension (P^O2) and carbon dioxide tension (P^CO2), using an IL 1306 blood gas analyzer (Instrumentation Laboratory, Lexington, MA), and for arterial and mixed venous hemoglobin oxygen saturation (Sa^O2and Sv^O2) and total hemoglobin concentration by means of the Co-Oximeter IL 282 (Instrumentation Laboratory), which was precalibrated for pig blood.

The Cicero anesthesia machine, which was modified to provide xenon application, performs mechanical ventilation by means of a motor-driven piston pump, which provides constant inspiratory flow. Ventilator settings throughout the experiment were as follows: tidal volume (V^T)= 12–14 ml/kg (adjusted to achieve a Pa^CO2between 37 and 43 mmHg); frequency (f)= 12/min; inspiratory time (Tⁱ)= 1.5 s; inspiration hold = 1 s; expiratory time (T^e)= 2.5 s; and positive end-expiratory pressure (PEEP)= 5 cm H²O. The inspiratory oxygen fraction (FI^O2) was kept constant at 30%, regardless of the inspiratory mixture actually used (N²O–oxygen, xenon–oxygen, or nitrogen–oxygen, see fig. 1) and was continuously measured in the inspiratory limb of the ventilator circuit by the machine-integrated oxygen monitor (a fuel cell sensor with an accuracy of ± 2%) calibrated before each experiment. Fresh gas supply of the anesthesia circuit was set to one half the minute ventilation throughout the experiment. Density and viscosity of the gas mixtures for the experimental conditions (37°C, 1 bar) were ρ^xe/O2= 3.95, η^xe/O2= 2.4, ρ^N2O/O²= 1.57, η^N2O/O²= 1.67, ρ^N2/O²= 1.13, and η^N2/O²= 1.93. Airway pressure (P^aw) was measured through a port placed between the endotracheal tube and Y piece with a pressure transducer (142PC01G; Honeywell, Plymouth, MN) with a range of 70 cm H²O, a linearity of 0.75%, and a precision of 0.25%. Flow was measured with a heated pneumotachograph (Fleisch No. 2; Fleisch, Lausanne, Switzerland) connected to a differential pressure transducer (PC 500; PasCal, Schwerin, Germany) and calibrated for each gas mixture using a 1-l supersyringe. Expiratory P^CO2was analyzed with the side-stream infrared carbon dioxide analyzer integrated in the anesthesia machine. Gas sampling rate was 200 ml/min, the delay between flow and carbon dioxide signal was corrected using values determined separately for gas mixtures containing 70% nitrogen, N²O, or xenon in 30% oxygen (Delay^N2= 185 ms, Delay^N2O = 175 ms and Delay^Xe= 240 ms. For recording P^aw, Flow, and P^CO2signals on a personal computer we used the DaqBoard 216 A/D converter system (IOtech, Cleveland, OH) and the data acquisition software DASYLab (Datalog, Mönchengladbach, Germany).

Measurements were performed at the end of three sequential 45-min periods of ventilation with either nitrogen–oxygen (periods 1 and 3 designated as control 1 and control 2) or anesthetic gas (xenon–oxygen or N²O–oxygen, depending on the random order of the experiment, period 2, designated as test-gas period) before and during methacholine infusion (fig. 1). At each point of measurement, we recorded the P^aw, Flow, and P^CO2signals for three sequential 2-min periods of continuous measure at a frequency of 100 Hz for subsequent off-line analysis. Additionally, we registered heart rate, mean arterial pressure, central venous pressure, mean pulmonary artery pressure, pulmonary artery occlusion pressure, and cardiac output. Methacholine infusion was started with a dosage of 10 μg · kg⁻¹· min⁻¹, which was increased stepwise (5–10 μg · kg⁻¹· min⁻¹) every 3–5 min, until R^awhad approximately doubled when compared to the value measured with baseline conditions before methacholine infusion. Then, the study protocol was repeated identically and analogous to the previous series of measurements.

Pulmonary mechanics were assessed according to the method described by Bates et al.  17The same technique was also used to separately determine the resistance of the endotracheal tube (R^tube) for each experimental gas mixture over a flow range between 0.1 and 1.0 l/s. For this purpose, the tip of the cuffed endotracheal tube was positioned at the distal end of an artificial trachea, which was connected to a 3-l bag.

The two prerequisites for using the occlusion technique described by Bates et al. , 17namely, continuous inspiratory flow and sudden flow interruption at the end of inspiration, are met during volume-controlled ventilation with the CICERO anesthesia machine. In contrast to previous investigations using this technique, the sudden flow interruption is not obtained by occluding the inspiratory valve, as occurs when using intensive care unit ventilators, but by the sudden arrest of the piston pump. Nevertheless, a square flow pattern is achieved regardless of the interruption mode. According to previous descriptions of this occlusion technique, we determined the peak airway pressure (P^max), and the airway pressures registered both immediately after end-inspiratory flow interruption (P¹) and after an occlusion period of 6 s (P²) from each occluded breath. By subtraction of the pressure component imposed by the resistance of the endotracheal tube (calculated by multiplying R^tubeand inspiratory flow) from P^maxwe obtained an estimation of the peak pressure within the trachea (P^max′). The correction proposed by Bates et al. , 18which consists of extrapolating the pre- and postocclusion pressure signals to the point in time when the inspiratory valve is semiclosed, was used for determining P^maxand P¹, to account for the finite time of flow interruption. For this purpose, the postocclusion pressure decay, which is caused by the stress relaxation of the lung tissue and by gas redistribution, was analyzed with a logarithmic function determined by the least-squares fit technique as described by Marquardt. 19This logarithmic fit was back-extrapolated to the time of flow interruption to obtain P¹. Inspiratory resistance was then determined as R^minand R^maxby the equations

Because R^minequals the sum of R^awand R^tube, R^awwas obtained by subtraction of R^tubefrom R^min. ΔR was then calculated as

The total inspiratory pressure drop over the bronchial tree (ΔP) was calculated for each gas mixture as

and the ΔP ratio of two different gas mixtures x and y was calculated as ΔP^x/ΔP^y.

Static and dynamic compliances (C^stand C^dyn) were calculated by the equations

where PEEPⁱis the intrinsic positive end-expiratory pressure, determined by the end-expiratory occlusion maneuver, i.e. , P^aw10 s after occlusion. Representative original P^awand Flow curves registered during baseline conditions and during methacholine infusion, which further explain these calculations, are shown in figures 2A and B, respectively.

The carbon dioxide expirograms were analyzed for the alveolar carbon dioxide slope (S^CO2), which was calculated according to Fletcher et al.  20and Meyer et al.  21,22by linear fit of the expiratory P^CO2plotted as a function of expired volume plot between 70 and 95% of expiration, which yields the change of P^CO2normalized for mixed expired P^CO2per unit of expired volume. In addition, series dead space was computed using the method described by Fletcher et al.  20The expired volume was calculated by integration of the flow signal.

Dead space ventilation (V^D/V^T) and venous admixture (Q^S/Q^T) were calculated from blood gas data and mixed expired carbon dioxide pressure (PE^CO2) using the standard equations

where Ca^O2and Cv^O2are the arterial and mixed venous oxygen content, respectively, and Cc^O2is the capillary oxygen content derived from the alveolar P^O2(PA^O2) calculated by the simplified alveolar gas equation 23as

where P^Bis the barometric pressure, P^H2Ois water–vapor pressure, FI^O2is the inspiratory oxygen concentration, and RQ is the respiratory quotient assumed to be 0.8. Arterial to alveolar oxygen difference was then calculated as PA^O2− Pa^O2.

All data are presented as mean ± SD. After normal distribution was confirmed by the Kolmogorov-Smirnov test, a two-way analysis of variance of repeated measures of one repeated factor, followed, when significant, by the Tukey test, was used to compare the data obtained within both groups (repeated factor) during the nitrogen–control periods and during the phases of xenon and N²O ventilation, before and after methacholine, respectively, and to further compare the differences between the two groups (not repeated factor) during xenon and N²O inhalation. Statistical significance was assumed at P < 0.05.

The results of our experiments are summarized in figure 3and in tables 1 and 2. In each group, the first animal had to be discarded from statistical analysis because of a lethal complication during induction of bronchoconstriction in one case (N²O group) and because of insufficient effect of methacholine in the second case (xenon group). Comparing the data obtained during the two control periods of each series, i.e. , the periods of nitrogen–oxygen ventilation before and after each test gas period, we found a statistically significant decrease in C^stand C^dynand increases in P^max, P¹, P², and P^meanin the N²O group during methacholine infusion only. There were no statistically significant differences between the two controls for any of the other conditions. The effects of methacholine infusion were validated by comparing the data related to airway pressure and resistance before and during bronchoconstriction during the nitrogen–oxygen control periods: all these values (except for R^tube) significantly increased during methacholine infusion (P < 0.01); C^stand C^dynsimultaneously decreased (P < 0.01). The methacholine dosage necessary for induction of bronchoconstriction was similar in both groups (40.5 ± 42.2 μg · kg⁻¹· min⁻¹in the xenon group and 36.1 ± 10.4 μg · kg⁻¹· min⁻¹in the N²O group). The large SD in the xenon group is caused by two single animals that required a relatively high (143.4 μg · kg⁻¹· min⁻¹) and low (12 μg · kg⁻¹· min⁻¹) dosage, respectively.

Substitution of xenon for nitrogen significantly increased R^min, R^max, R^aw, and R^tubebefore and during methacholine infusion (P < 0.01). These parameters also were increased when comparing the xenon and N²O group with both study conditions (P < 0.01). In contrast, they were not affected when nitrogen was replaced by N²O. During methacholine-induced bronchoconstriction P^maxand P^meanwere significantly higher with xenon–oxygen ventilation when compared to the nitrogen–oxygen controls (P < 0.01). In contrast, there were no statistically significant differences in P^maxor P^meanduring xenon–oxygen and nitrogen–oxygen ventilation before bronchoconstriction or during N²O–oxygen and nitrogen–oxygen ventilation during baseline conditions and during bronchoconstriction. During bronchoconstriction, P^mean, but not P^max, was significantly higher with xenon–oxygen when compared to the N²O group (P < 0.01). In the xenon group, but not in the N²O group, the ΔP ratio was higher during bronchoconstriction when compared to baseline.

Generally, PEEPⁱwas low and did not change significantly when replacing nitrogen with xenon or with N²O, regardless of the experimental condition, but increased during bronchoconstriction when compared to baseline conditions.

Regardless of the inspiratory gas mixture, pulmonary gas exchange was impaired during methacholine infusion, as shown by the lower Pa^O2, the higher Pa^CO2, the Q^S/Q^T, the alveolar–arterial difference in partial pressure of oxygen (PA^O2− Pa^O2), and the total dead space (V^Dtot) when compared to baseline conditions (table 2). However, before, and during, bronchoconstriction, gas exchange parameters did not differ between the xenon–oxygen and N²O–oxygen groups, or between nitrogen–oxygen controls and either test gas. In both groups, S^CO2was higher with bronchoconstriction when compared to baseline conditions but did not differ between control and test–gas ventilation.

No changes in hemodynamic variables were induced by the different gases used in our experiment, whereas metacholine infusion, when compared to the baseline conditions before bronchoconstriction, significantly decreased mean arterial pressure, concomitant with a simultaneous increase in mean pulmonary artery pressure in both groups. There were no statistically significant differences regarding heart rate, central venous pressure, pulmonary artery occlusion pressure between baseline conditions, and bronchoconstriction. During xenon ventilation, but not during N²O ventilation, cardiac output decreased significantly with bronchoconstriction.

The aim of this investigation was to compare pulmonary mechanics and gas exchange in mechanically ventilated pigs during xenon–oxygen and N²O–oxygen anesthesia before and during continuous intravenous methacholine. The main result was that R^awand pressure increased during xenon–oxygen ventilation. In terms of absolute values, these changes were moderate before methacholine infusion, as suggested by the unchanged P^max, despite the increased resistance, but more pronounced during intravenous methacholine.

Our results agree with previous data regarding the effect of physical gas properties on respiratory mechanics. 2,4They also confirm data regarding pulmonary resistance in dogs during 70% xenon and 70% N²O anesthesia, previously published by Zhang et al.  14; however, in our study, the relative change in R^awproduced by xenon was more pronounced (1.54 before and 1.88 during bronchoconstriction compared to 1.35 as measured by Zhang et al.  14). With inspiratory xenon concentrations of 50%, as tested by these authors during methacholine-induced bronchoconstriction, they did not find any significant difference between xenon and nitrogen or N²O. The discrepancy in our data during bronchoconstriction may be caused by the maintenance of the anesthetic concentration at 70% even during this condition. This is crucial for xenon because, due to the particularly high density and viscosity of this noble gas, the effects of a xenon–oxygen mixture on respiratory mechanics are markedly sensitive to changes of the inspired xenon concentration. Moreover, it must be taken into account that baseline R^awvalues measured by Zhang et al.  14in dogs were lower when compared to our data, suggesting larger airway dimensions in this species. Finally, Zhang et al.  14performed their measurements in open-chest animals, a condition that affects the response to methacholine. 15 

According to the theoretical approaches proposed by Pedley et al. , 24,25Olson et al.,  26and Jaffrin and Kesic, 27which were more recently reviewed by Pedley and Drazen, 28both the high viscosity and the density of xenon are probably responsible for the results of our experiment. Briefly, these authors applied laws of fluid mechanics to complex systems of branching tubes such as the bronchial airways. Their model calculations showed how gas flow, despite laminar throughout the bronchial system, is disturbed at branching sites, resulting in blunted flow velocity profiles. A specific distance (called “entrance length”), which depends on airway diameter and the Reynold number (and hence on viscosity and density of the gas mixture), has to be covered by the bulk flow at each generation of the bronchial tree until the typical parabolic velocity profile of fully developed laminar flow is reestablished. Because this is an energy-consuming process, previous predictions of total pressure drop over the bronchial system based on the assumption of overall fully developed laminar flow, 29which only depends on the viscosity of the inspiratory gas mixture, resulted in underestimation of experimental data. 30,31This is also the case for our results insofar as both ratios η^Xe/O2/η^N2/O²(1.25 at 37°C) and η^N2O/O²/η^N2/O²(0.87 at 37°C) were lower than ΔP^Xe/O2/ΔP^N2/O²(1.55 before and 1.90 during bronchoconstriction) and ΔP^N2O/O²/ΔP^N2/O²(1.23 before and 1.16 during bronchoconstriction). In contrast, ΔP ratios calculated according to Pedley et al.,  25,i.e. , ΔP^x/ΔP^y=(ρ^x*η^x)^1/2/(ρ^y*η^y)^1/2, which yields theoretical values for ΔP^Xe/O2/ΔP^N2/O²= 2.08 and for ΔP^N2O/O²/ΔP^N2/O²= 1.1, more closely approximated the measured ΔP^Xe/O2/ΔP^N2/O²and ΔP^N2O/O²/ΔP^N2/O²ratios. However, ΔP^Xe/O2/ΔP^N2/O²before methacholine infusion was overestimated by Pedley's formula. According to Ingram and Pedley 32and Wood et al. , 4the presence of flow conditions equivalent to fully developed laminar flow in the more peripheral sections of the bronchial system (bronchial diameters < 4 mm) partially may have caused this overestimation. These flow conditions are not explained by the theory of Pedley et al.  25and tend to reduce the measured pressure drop over the distal airways when compared to the predicted values, in particular by the use of high-density gases. We can exclude an airway narrowing effect of xenon as a possible further cause of the increased R^awobserved in our experiment when considering the data for ΔR, C^st, and C^dyn: neither of these changed with respect to the different inspiratory gas mixtures as one would expect if airway smooth muscle tone was raised.

The different effect of xenon on R^awand airway pressure is caused by the fact that airway pressure is determined by both a resistive and an elastic component. During xenon anesthesia resistance, consequently, the resistive pressure component only, was increased, but not compliance.

In contrast to respiratory mechanics, xenon and N²O did not affect gas exchange or hemodynamics. Even the methacholine-induced deterioration in gas exchange, as documented by the Pa^O2; Q^s/Q^T; PA^O2− Pa^O2; Pa^CO2; and V^D/V^T, was of similar degree, regardless of the composition of the inspiratory gas. In addition, the S^CO2was unaffected by the different inspiratory gas mixtures, suggesting that the properties of the gas species used in our experiment did not significantly affect the V^A/Q distribution in the lung. The alveolar slope of the expired P^CO2is considered as an indicator of the nonhomogenous V^A/Q distribution. 21Interpreting this test, however, is particularly difficult, because the alveolar P^CO2slope is determined by both the series and the parallel nonhomogeneity of the V^A/Q ratio, 22and these two components cannot be analyzed separately.

Our gas exchange data agree fairly well with previously published investigations of the effects of gas mixtures with different physical properties on V^A/Q matching. In fact, in three of these studies, 7,10,11PA^O2− Pa^O2even decreased with increasing gas density, but the changes were only subtle and probably of no clinical relevance. In contrast, in two other studies, PA^O2− Pa^O2was uninfluenced or even slightly increased with higher gas density. In line with these results, investigations by Schumacker et al.  12and Schulz et al.  33showed only a slight relation between gas density and intrapulmonary gas distribution, gas distribution being more homogeneous in high-density atmospheres.

In summary, airway mechanics in pigs are affected during xenon anesthesia. This response is moderate with baseline conditions but more pronounced during bronchoconstriction. The results of our findings, however, probably assume only minor importance for the general anesthetic use of xenon for the following reasons:(1) during mechanical ventilation, as usual during general anesthesia, the ventilator but not the patient has to overcome the increased inspiratory resistance caused by the high density and viscosity of xenon. (2) The consequent effects on airway pressure are probably of minor relevance because they are negligible in healthy conditions and only moderate during bronchoconstriction. Furthermore, the increment in airway pressure is likely to be less accentuated in more peripheral lung regions and to disappear at the alveolar level because it is caused by the physical properties of xenon but not by an increase in airway muscle tone that would narrow the airway diameter and increase lung compliance. (3) Gas exchange does not deteriorate during xenon anesthesia, and, therefore, xenon is unlikely to impair oxygenation even during bronchoconstriction.