Effect of Low Isoflurane Concentrations on the Ventilation–Perfusion Distribution in Injured Canine Lungs

After we obtained approval from the Laboratory Animal Care and Use Committee of the University of South Florida, 14 mongrel dogs weighing 20–26 kg (22.2 ± 2.0 kg, mean ± SD) were anesthetized with intravenous sodium pentobarbital, a 12-mg/kg bolus dose, followed by infusion of 20 μg · kg⁻¹· min⁻¹, and paralyzed with intravenous pancuronium bromide (0.1 mg/kg). Animals were placed supine, and their tracheas were intubated with a 9-mm ID cuffed endotracheal tube (Mallinckrodt, Argyle, NY). Dogs then were mechanically ventilated (Servo 900D; Siemens, Schaumburg, IL) with 40% oxygen in nitrogen, a tidal volume of 10 ml/kg, and ventilator frequency adjusted to maintain arterial blood carbon dioxide tension (Paco²) between 35 and 45 mmHg. A catheter was inserted into a femoral artery, and a 7-French, thermistor-tipped, triple-lumen pulmonary artery catheter (93A-131-7F; Baxter Edwards Critical Care, Irvine, CA) was placed through a femoral vein.

Systemic blood pressure, central venous pressure, pulmonary artery pressure, and pulmonary artery occlusion pressure were transduced (Transpac® II; Abbott Critical Care, Chicago, IL) and recorded (TA 2600; Statham Gould, Oxnard, CA). A horizontal plane through the shoulder was taken as zero reference point for blood pressure measurements. Cardiac output (CO) was measured with a thermal dilution technique (Oximetrix® 3; Abbott Critical Care). Ten milliliters of iced 5% dextrose solution was used as the indicator, and three determinations were performed at random moments during the ventilatory cycle and averaged.

Gas flow was measured at the proximal end of the endotracheal tube with a heated pneumotachograph (3791; Hans Rudolph, Kansas City, MO), connected to a differential pressure transducer (DP 1030871; Validyne, Northridge, CA). Tidal volume was derived from the integrated gas flow signal. Expiratory minute ventilation (V̇^E) was measured for control with a calibrated Wright respirometer at the outlet of the gas mixing chamber, which was connected to the expiratory port of the ventilator. Airway pressure was measured at the proximal end of the endotracheal tube with a differential gas-pressure transducer (MP45-871; Validyne). Esophageal pressure was measured with a balloon catheter (Mallinckrodt) connected to a differential pressure transducer (MP45-871, ±35 cm H²O; Validyne) as described by Baydur et al.  19Transpulmonary pressure was obtained by subtracting esophageal pressure from airway pressure. End-inspiratory pressures were measured following a 5-s end-inspiratory occlusion, 20and intrinsic positive end-expiratory pressures following a 5-s end-expiratory occlusion of the airway, as described previously. 21Static lung compliance was obtained by dividing expiratory tidal volume by the difference between end-inspiratory transpulmonary pressure and intrinsic positive end-expiratory pressure. Pulmonary resistance (R^L) was calculated by dividing the difference between maximal and end-inspiratory transpulmonary pressure by the preceding gas flow. 20 

Arterial and mixed venous Po²and Pco²and p  H were determined immediately after sampling with standard blood gas electrodes (Model 1303; Instrumentation Laboratories, Lexington, MA). Oxygen saturation of arterial and mixed venous blood, and total hemoglobin were determined by spectrophotometry with a CO-oximeter (Model 282; Instrumentation Laboratories). Fractions of inspired oxygen (Fio²) and mixed expired oxygen (Feo²), carbon dioxide, and end-tidal isoflurane were determined with a Raman scattering gas analyzer (Rascal® II; Ohmeda, Louisville, CO). 22 

Six inert gases (sulfur hexafluoride, ethane, cyclopropane, enflurane, diethyl ether, and acetone) were dissolved in lactated Ringer's solution and infused into a peripheral vein with a constant rate set at about 0.05% of V̇^Efor at least 40 min. 23,24Arterial and mixed venous blood samples were collected during stable conditions confirmed by constancy of V̇^E, Feo², Feco², and CO. Expired gas samples were collected after the blood samples with a time delay equal to the transit time of the gas through the mixing chamber. 24Inert gases were extracted, their concentrations were measured with a gas chromatograph (HP 5890; Hewlett-Packard, Waltham, MA), and their blood gas partition coefficients were determined. 25To provide adequate separation of the inert gases and isoflurane, temperature programming was used as has been described in detail by Dueck et al . 26 

Arterial to mixed venous (retention) and mixed expired to mixed venous (excretion) concentration ratios of the inert gases were used to obtain retention and excretion solubility curves. 23,24By formal mathematical analysis with enforced smoothing, these relations were transformed into a 50-compartment distribution plot of blood flow and ventilation against V̇^A/Q̇. 23,27Intrapulmonary shunt was defined as the fraction of pulmonary blood flow (Q̇^T) perfusing essentially nonventilated alveoli (V̇^A/Q̇ < 0.005), low V̇^A/Q̇ as the fraction of Q̇^Tperfusing poorly ventilated lung areas (0.005 < V̇^A/Q̇ < 0.1), high V̇^A/Q̇ as the fraction of V̇^Eventilating poorly perfused lung areas (10 < V̇^A/Q̇ < 100), and dead space as fraction of V̇^Eventilating nonperfused lung areas (V̇^A/Q̇ > 100). Mean V̇^A/Q̇ ratio of perfusion (Q̇) and ventilation (V̇), and logarithmic SDs of perfusion (logSD^Q) and ventilation (logSD^V) were derived from the 50-compartment model. Predicted values for Pao²were calculated from the recovered V̇^A/Q̇ distributions as previously described. 24 

Transmural central venous pressure, pulmonary artery pressure (Ppa^tm), and pulmonary artery occlusion pressure (Pao^tm) were derived by subtracting esophageal pressure from the measured vascular pressures. Systemic vascular resistance was calculated as

and pulmonary vascular resistance was calculated as

Oxygen content for arterial blood (Cao²) was calculated as

oxygen delivery (Do²) was calculated as

and oxygen consumption (V̇^O2) was calculated as 28 

Alveolar arterial oxygen difference was calculated as

where P^Bis barometric pressure, P^H2Ois the pressure of the water vapor at body temperature, and R is the respiratory quotient.

Dogs remained supine after instrumentation. Body temperature was kept between 37 and 38°C with a heating pad. Adequate hydration and energy supply (25 kcal · kg⁻¹· d⁻¹) was ensured with an infusion of 5% dextrose and lactated Ringer's solution to achieve an Pao^tmof 8 mmHg. After 90-min stabilization, measurements reflecting the noninjured state were obtained.

Acute lung injury was induced by injection of 0.08 ml/kg purified oleic acid (J. T. Baker Inc., Phillipsburg, NJ) into the right atrial catheter over 15 min. Additional 0.2-ml increments of oleic acid were administered every 30 min until Pao²was less than 75 mmHg. Lung injury was allowed to stabilize for 90 min until measurements were repeated.

The dogs received, in random order, 0, 0.25, and 0.5 vol% end-tidal isoflurane concentrations before and after induction of oleic acid lung injury. Isoflurane was administered with a calibrated vaporizer (isoflurane vaporizer 952; Siemens). Forty minutes of equilibration was allowed for each intervention before measurements. Following each measurement, dogs were ventilated without isoflurane for at least 30 min until V̇^E, Feo², Feco², CO, Ppa^tm, systemic blood pressure, Pao², and Paco²returned to baseline values before or after induction of lung injury (± 5%). To reopen unspecific atelectasis, animals’ lungs were inflated manually to an airway pressure of 30 cm H²O for 10 s after each measurement, before changing the isoflurane concentration.

Results are expressed as mean ± SEM (SE). Data were analyzed using a repeated two-way analysis of variance. When a significant F ratio was obtained, differences between the mean values were isolated with the Scheffé multiple range test. The relation between measured and predicted Pao²was assessed with a linear regression analysis. P < 0.05 was considered statistically significant.

Administration of oleic acid produced a lung injury with severe V̇^A/Q̇ mismatch, which decreased Pao²from 163 ± 8 to 72 ± 4 mmHg during mechanical ventilation with 40% oxygen. Following induction of lung injury, an average of 38 ± 4% of the CO perfused shunt (V̇^A/Q̇ < 0.005), 3.7 ± 1.3% low V̇^A/Q̇ (0.005 < V̇^A/Q̇ < 0.1), and 58 ± 5% normal V̇^A/Q̇ areas (0.1 < V̇^A/Q̇ < 10). Dead space ventilation (V̇^A/Q̇ > 100) comprised 46 ± 4% of the total ventilation, and another 5.4 ± 2.3% was distributed to high V̇^A/Q̇ areas (10 < V̇^A/Q̇ < 100).

The effects of isoflurane and oleic acid–induced lung injury on pulmonary mechanics are shown in table 1. Static lung compliance decreased (P < 0.05) and airway pressures increased (P < 0.05) following induction of lung injury. Inhalation of isoflurane did not affect lung compliance, R^L, or intrinsic positive end-expiratory pressure in the presence or absence of lung injury. Ventilatory rate and V̇^Eremained unchanged throughout the study.

Changes in cardiovascular variables are shown in table 2. Inhalation of 0.5 vol% isoflurane decreased systemic vascular resistance both in presence and absence of lung injury (P < 0.05). CO was lowest during 0.5 vol% isoflurane (P < 0.05). Elevated mean Ppa^tmand pulmonary vascular resistance decreased with 0.25 and 0.5 vol% isoflurane during lung injury (P < 0.05). Heart rate, transmural central venous pressure, and Pao^tmremained unchanged.

Despite inhalation of isoflurane, Pao², alveolar arterial oxygen difference, Do², and Pv̄o²remained unchanged in absence of lung injury. Inhalation of 0.25 and 0.5 vol% isoflurane was associated with a decrease in Pao², Do², and Pv̄o²(P < 0.05) and an increase in alveolar arterial oxygen difference during lung injury (table 3). Arterial p  H, Paco²,and V̇o²remained unchanged during all tested conditions.

Changes reflecting the distribution of ventilation and perfusion are summarized in table 4and figure 1. Patterns ofblood flow and ventilation distributions were unimodal in 11 animals and modestly bimodal in the 3 remaining animals. During lung injury, inhalation of 0.5 vol% isoflurane accounted for a 10 ± 2% increase (P < 0.05) in blood flow to shunt units (V̇^A/Q̇ < 0.005) and an 8 ± 1% decrease (P < 0.05) in the fraction of CO to units with normal V̇^A/Q̇ ratios (0.1 <V̇^A/Q̇ < 10). In the absence of acute lung injury, inhalation of 0.5 vol% isoflurane did not affect pulmonary blood flow distribution. Blood flow distribution curves were centered, with Q̇ ranging from 0.83 ± 0.09 to 1.02 ± 0.16, and their dispersions were increased (logSD^Q> 0.6). Highest logSD^Qwas observed during inhalation of 0.5 vol% isoflurane in acute lung injury (P < 0.05). Ventilation distributions were shifted to the right with V̇ ranging from 1.75 ± 0.16 to 2.90 ± 0.46, while dispersions of the curves were above the upper normal limit (logSD^V> 0.6). Changes in dead space, Q̇, and logSD^Vwere not statistically significant.

Predicted Pao²was close to measured Pao²for all tested conditions (table 5). Mean residual sum of squares indicated acceptably small experimental error for all inert gas measurements.

This study was designed to evaluate the effect of low isoflurane concentrations on pulmonary gas exchange in subjects with acute lung injury. We found that inhalation of 0.25 and 0.5 vol% isoflurane worsened V̇^A/Q̇ mismatch during acute lung injury, as reflected by increases in intrapulmonary shunt and logSD^Q. Inhalation of isoflurane caused systemic vasodilation and reversed pulmonary vasoconstriction induced by lung injury.

Sedation is used in critically ill patients to minimize discomfort and allow effective ventilatory support. 7,29However, sedatives have adverse effects and the potential to prolong duration of ventilatory support and length of intensive care unit stay. 30Faster emergence and weaning from ventilatory support, 4–6,8,10cardiovascular stability, 3,10and bronchodilation 11,31have been suggested an advantages of isoflurane inhalation over midazolam or propofol infusions for prolonged sedation. Ostermann et al.  7recommended further investigations because isoflurane is currently used infrequently for sedation. Although isoflurane in low concentrations has been shown to be useful for sedation of ventilated patients with mild pulmonary dysfunction without causing adverse effects on cardiorespiratory, hepatic, renal, or adrenal function, its effects on gas exchange in acute lung injury are unknown. 4,8,9Furthermore, effects of low isoflurane concentrations on gas exchange may also be important during recovery from general anesthesia.

When comparing our results with those of previous human and animal studies, it should be emphasized that our observations were made in anesthetized dogs with oleic acid–induced lung injury. However, although pathophysiologically different, oleic acid–induced lung injury is a well-established animal model of human acute lung injury. 32The essentially unimodal V̇^A/Q̇ distribution with 38 ± 4% of the pulmonary blood flow perfusing shunt units indicated severe lung injury in our dogs and is consistent with previous observations in the canine oleic acid model. 32,33Distributions of V̇^A/Q̇ ratios observed in dogs with oleic acid–induced lung injury 32,33are essentially comparable to those in humans with acute lung injury, where blood flow is distributed to either shunt or normal V̇^A/Q̇ units. 2The slightly elevated dispersion of the pulmonary blood flow (logSD^Q) indicated mild V̇^A/Q̇ mismatch in our dogs in the absence of lung injury and is consistent with previous observations in anesthetized and mechanically ventilated subjects. 34Anesthesia with sodium pentobarbital in our animals may have had a minor effect on V̇^A/Q̇ matching from inhibition of hypoxic pulmonary vasoconstriction, 35but a constant sodium pentobarbital infusion could not have been responsible for the observed changes in the V̇^A/Q̇ distributions at different end-tidal isoflurane concentrations.

Previous experimental and clinical studies corroborated findings regarding the V̇^A/Q̇ matching during anesthesia with isoflurane. Carlsson et al.  17observed no change in V̇^A/Q̇ matching with 1.0 and 1.5 vol% isoflurane in healthy patients during one-lung ventilation. In contrast, Kellow et al.  18observed a threefold increase intrapulmonary shunt with 1.5–2.0 vol% isoflurane when compared with propofol infusion in patients undergoing thoracic surgery. Our results appear to be in contrast with those of Domino et al. , 12who observed in dogs that 1.4–3.5 vol% isoflurane inhibits hypoxic pulmonary vasoconstriction in a dose-dependent manner during one-lung ventilation. This dose–response curve suggests that low concentrations of isoflurane have little adverse effect on hypoxic pulmonary vasoconstriction and thereby on V̇^A/Q̇ matching. 12However, all previous studies have involved subjects in which 0.5 vol% isoflurane was administered before and after oleic acid–induced lung injury. Therefore, our results reflect the effect of low isoflurane concentrations on V̇^A/Q̇ mismatching in diffuse lung injury.

Inhalation of 0.25 and 0.5 vol% isoflurane during induced lung injury consistently resulted in a marked increase in blood flow to shunt units, without creating increase in low V̇^A/Q̇ areas. Because the normal mean V̇^A/Q̇ of the distribution of pulmonary blood flow and its dispersion (logSD^Q) remained unchanged, some essentially nonventilated lung units became better perfused. These observations support the contention that isoflurane increases intrapulmonary shunting by inhibiting hypoxic pulmonary vasoconstriction. 12–15,18,36In addition, absence of changes in mean V̇^A/Q̇ of the alveolar ventilation distribution, its dispersion (logSD^V), dead space ventilation, and lung compliance indicate that alveolar collapse and decreased ventilation of well-perfused lung regions did not occur. Because in current study pulmonary vascular resistance was not measured at constant pulmonary blood flow, our results do not allow direct conclusions on hypoxic pulmonary vasoconstriction. Our observations are in contrast to a study by Kleinsasser et al. , 37who report that inhalation of 2.0 vol% sevoflurane but not of 1.15 vol% isoflurane increased blood flow to shunt units and low V̇^A/Q̇ areas and decreased arterial blood oxygenation in pigs with pneumoperitoneum-induced deterioration in gas exchange. The mechanism by which sevoflurane but not isoflurane affects pulmonary blood flow distribution could not be explained on the basis of these data. 37The apparent discrepancy to our results may be attributed to the lung injury model used, the degree of the preexisting V̇^A/Q̇ mismatch, and species-related differences. Increase in intraabdominal pressure during pneumoperitoneum will result in alveolar collapse caused by a cephalad shift of the diaphragm most evident in dependent lung areas, whereas the well-ventilated lung units have to be considered healthy. The corresponding intrapulmonary shunt of 9–14% indicates mild pulmonary dysfunction when compared with our dog with an intrapulmonary shunt of 38–41%. 38Intrapulmonary shunting ranging up to 12% in healthy subjects during anesthesia has been found to correlate directly with the formation of nonaerated tissue observed by computed tomography in dependent lung regions adjacent to the diaphragm. 34Experimental 39and clinical 17,38,40,41observations indicate that isoflurane concentrations between 0.5 and 2.0 vol% have only a small effect on venous admixture and gas exchange during one-lung ventilation. In contrast to observations in healthy subjects during one-lung ventilation or during increased intraabdominal pressure, oleic acid causes a direct lesion to the pulmonary vasculature, resulting in a diffuse but not homogeneous lung injury with increased pulmonary vascular permeability, intrapulmonary shunting, and arterial hypoxemia. 42Furthermore, release of endogenous vasodilators, including metabolites of the cyclooxygenase pathway, e.g. , prostacyclin, have been shown to oppose hypoxic pulmonary vasoconstriction in animals with lung injury. 43Isoflurane may attenuate hypoxic pulmonary vasoconstriction by increasing production of vasodilator metabolites of the cyclooxygenase pathway. 14Observations in a dog model showed that isoflurane-induced attenuation of hypoxic pulmonary vasoconstriction is abolished by cyclooxygenase inhibition. 14Recently, isoflurane has been found to induce release of other vasodilator mediators during inflammation. 44Thus, release of endogenous vasodilators may have aggravated the increase in blood flow to shunt units in our dogs with injured lungs.

Small differences between Pao²predicted from the recovered V̇^A/Q̇ distribution and the measured Pao²indicate that alveolar end-capillary equilibration was complete. Thus, observed changes in gas exchange can be attributed to the measured V̇^A/Q̇ mismatch. Corresponding to the increase in V̇^A/Q̇ mismatch, Pao²decreased during 0.25 and 0.5 vol% isoflurane in the presence of acute lung injury. Extrapulmonary factors with potential effects on Pao², such as alveolar ventilation and acid-base status, did not change significantly. 45 

Changes in CO have been reported to positively correlate with the intrapulmonary shunt fraction. 46In our study, isoflurane decreased pulmonary blood flow and directed it preferentially to nonventilated lung units. Consequently, decreased CO was associated with significantly larger intrapulmonary shunting of blood, lower Pao², and markedly reduced Do². Therefore, changes in CO cannot explain the increase in intrapulmonary shunting of blood with isoflurane during lung injury. 46Inhalation of 0.5 vol% isoflurane lowered the elevated mean Ppa^tmand pulmonary vascular resistance during induced lung injury. Isoflurane in concentrations between 0.3 and 1.4 vol% has previously been shown to produce pulmonary vasodilation in subjects with pulmonary hypertension. 47,48Inhalation of isoflurane reduced mean systemic blood pressure and systemic vascular resistance in our dogs, a well-documented effect at these isoflurane concentrations.

Intrapulmonary shunting of blood and reduction in lung compliance results from alveolar collapse during lung injury. Higher intrapulmonary shunt with unchanged lung compliance during inhalation of isoflurane indicates augmented perfusion of essentially nonventilated lung units. An increase in R^Lis a known feature of human acute lung injury; this phenomenon was observed in the current study as well. 49Decrease in R^Lwith isoflurane inhalation during induced bronchoconstriction and in patients with asthma has been reported previously, albeit with some inconsistency. 11,31In contrast, R^Lwas unaffected by inhalation of isoflurane in our dogs, a finding in agreement with a study by Pesenti et al. , 49which demonstrated no change in elevated R^Lduring intravenous β²-agonist infusion in patients with acute lung injury.

Previous reports have recommended inhalation of isoflurane in low concentrations as a sedative agent to facilitate mechanical ventilation in critical ill patients. The results of this study demonstrate that even 0.25 to 0.5 vol% of isoflurane may increase V̇^A/Q̇ mismatching by inhibiting hypoxic pulmonary vasoconstriction in experimental lung injury. Controlled clinical trials and long-term investigations are warranted to evaluate the validity of these results in critically ill patients.