Because maintaining arterial oxygenation (PaO2) during one-lung ventilation (OLV) can be a clinical problem, it is useful to be aware of factors that influence PaO2 in this situation and are under the control of the anesthesiologist. It is unknown whether, among the commonly used volatile anesthetic agents, one is associated with higher PaO2 levels. Clinical studies suggest that isoflurane provides superior PaO2 during OLV than does halothane. These have not been compared to enflurane. The authors studied PaO2 and hemodynamics during OLV with 1 MAC enflurane versus 1 MAC isoflurane.
Twenty-eight adults who had prolonged periods of OLV anesthesia with minimal trauma to the nonventilated lung (thoracoscopic or esophageal surgery) were studied in a cross-over design. Patients were randomized to two groups: Group 1 received 1 MAC enflurane in oxygen from induction until after the first 30 min of OLV, then were switched to 1 MAC isoflurane. In group 2, the order of the anesthetics was reversed.
Isoflurane was associated with higher PaO2 values during OLV (P < 0.0001). Mean PaO2 (+/- SD) after 30 min OLV isoflurane was 231 (+/- 125) mmHg versus 184 (+/- 106) mmHg after 30 min OLV enflurane. The difference in PaO2 between the two anesthetics was most marked in the patients with the highest PaO2 during OLV: PaO2 isoflurane PaO2 enflurane varies; is directly proportional to PaO2 isoflurane (r = 0.65, P < 0.001). There were no other significant differences between anesthetic gases in the measured hemodynamic or respiratory variables. In the subgroup of patients with pulmonary artery catheters (n = 7), PaO2 correlated with cardiac output during OLV for both anesthetics (r = 0.81, P < 0.001).
During OLV, the PaO2 values with 1 MAC isoflurane were greater than those with enflurane. The dependence of PaO2 on cardiac output does not support the hypothesis that an increase in cardiac output will cause a decrease in hypoxic pulmonary vasoconstriction and a decrease in PaO2 during OLV.
Key words: Anesthesia, thoracic. Anesthetics, volatile: enflurane; isoflurane. Anesthetic techniques: one-lung ventilation. Heart: cardiac output. Lung: atelectasis; blood flow; oxygen; shunting. Surgery: thoracic. Ventilation: one-lung ventilation.
HYPOXEMIA during one-lung ventilation (OLV) for thoracotomy or thoracoscopy remains a clinical problem for anesthesiologists. Recent clinical studies have reported that unacceptably low levels of arterial oxygen tension (PaO2) occur in 9-27% of cases. [1,2]This continuing problem has prompted research to understand the mechanisms involved in the etiology of this hypoxemia and to discover anesthetic techniques associated with optimal arterial oxygenation during OLV. In most cases, hypoxemia can be prevented by the use of a high inspired oxygen concentration (FIO2) or by application of continuous positive airway pressure (CPAP) to the nonventilated lung. However, in certain clinical situations, it is inappropriate or ineffective to use these methods.
Volatile anesthetics are popular maintenance agents during thoracic surgery, because they are bronchodilators and are rapidly titratable. In the majority of comparative studies, they have provided arterial oxygenation comparable to that with intravenous anesthetics. It would be useful to know whether, among the commonly used volatile anesthetics, one is associated with higher levels of PaOsub 2 during OLV.
Previously, PaO2during OLV with isoflurane was shown to be superior to that with halothane, but these have not been compared to enflurane. The purpose of this study was to compare Pa sub O2during stable OLV with isoflurane versus enflurane using a randomized crossover design. Hemodynamic changes that might influence the PaO2were measured whenever possible.
With Ethics Committee approval, adult patients were studied. Subjects studied were consecutive patients who met the requirements of the experimental protocol and signed informed consent. A sequential sampling plan (CSS Statistica, StatSoft) was developed using OLV PaO2variance data from a previous study and alpha and beta error probabilities of 0.05 and 0.1, respectively. Study recruitment continued until the cumulative sum of enflurane/isoflurane PaO2differences exceeded the predetermined limits. Potential subjects were consecutive patients scheduled for procedures that involved a long period of intraoperative OLV with minimal lung trauma, thoracoscopic surgery, or esophageal surgery who were anesthetized by one of the authors. Subjects were randomized to one of two study groups by lottery: in group 1, enflurane administration preceded isoflurane; in group 2, isoflurane was followed by enflurane. For all patients, age, height, weight, pulse, blood pressure, arterial blood gases breathing air, and spirometry in the sitting position were recorded on the day before surgery.
Patients were not premedicated. On admission to the operating room, intraarterial and venous cannulas were placed under local anesthesia, and patients were placed on a water-circulating heating mattress. In those patients having thoracotomies for esophageal surgery, a lumbar epidural catheter was placed and tested with 3 ml 2% lidocaine with adrenaline 1/200,000. No further epidural medications were administered until after the study period. Additional monitoring for all patients included electrocardiogram, pulse oximetry, esophageal temperature, urine output, neuromuscular blockade, and peak airway pressure. In the esophageal surgery patients, a flow-directed pulmonary artery (PA) catheter was inserted via the right internal jugular vein after induction of anesthesia and zeroed in the horizontal plane of the vertebral column in the lateral position. End-tidal concentrations of anesthetic gases and carbon dioxide were measured with a Nellcor-N2500 anesthesia safety monitor (Nellcor, Hayward, CA), which was calibrated against commercial standard gas samples.
Anesthesia was induced with thiopental (2-3 mg/kg), fentanyl (3 micro gram/kg), and pancuronium (0.1 mg/kg). In group 1, enflurane in oxygen at a concentration sufficient to produce an end-tidal concentration (FET) of 1 MAC (1.7%) was administered. A motor blockade of 90-95% was maintained with incremental doses of pancuronium. Intravenous crystalloids and fentanyl were titrated to maintain the systolic blood pressure within 15% of preinduction values.
After induction of anesthesia, a left-sided double-lumen endobronchial tube was placed in all patients and initially positioned by auscultation. After turning the patient to the lateral decubitus position, the endobronchial tube position was confirmed and adjusted with fiberoptic bronchoscopy just before the initiation of OLV. Tidal volumes of 10 ml/kg were used during both one- and two-lung ventilation (2LV) with the rate adjusted to maintain the arterial carbon dioxide (Pa sub CO2) between 35 and 45 mmHG. Effective lung isolation was confirmed by the absence of a leak from the nonventilated lumen of the endobronchial tube via a 1-cm underwater seal. This was monitored during thoracotomy or thoracoscopy by direct observation of the collapsed nonventilated lung in the operative hemithorax.
Arterial blood gases were drawn during 2LV immediately before the initiation of OLV and then every 10 min during OLV for at least 30 min or until the PaO2change between consecutive samples was < 20%. All arterial blood gas samples were placed on ice and analyzed within 5 min using a Stat Pro-1 blood gas analyzer (Nova Biomedical, Mississauga, Ontario, Canada). In group 1, enflurane was then discontinued and isoflurane titrated to maintain FETisoflurane 1 MAC (1.2%). Arterial blood gases were drawn every 10 min for at least 30 min or until the FETenflurane was < 0.1%. Isoflurane was maintained at FET1 MAC until the resumption of 2LV. In those patients with PA catheters: thermodilution cardiac output (CO), mixed venous blood gases, pulmonary artery occlusion pressures, and central venous pressures were measured during 2LV immediately before OLV, after 30 min OLV with 1 MAC enflurane, after 30 min OLV with 1 MAC isoflurane, and after resumption of 2LV. Venous admixture (QS/QT) at each of these intervals was calculated from standard formulas, and oxygen consumption from the Fick equation. For group 2 subjects, the protocol was identical except that the order of enflurane and isoflurane administration was reversed (Table 1).
Statistical analysis was performed with the aid of a computer program (CSS Statistics, StatSoft); in all tests the level of significance was set at 5%. Mean preoperative demographic and laboratory variables of group 1 and 2 subjects were compared using Student's t test for independent samples. The effects of anesthetic on arterial blood gas and hemodynamic variables were tested by multivariate repeated measures ANOVA, using one between-subjects factor (group) and two within-subject factors (anesthetic gas and number of lungs ventilated). Comparison of PaO2values at specific times during OLV was by paired Student's t test with Bonferroni correction for multiple comparisons. CO and pulmonary pressures during OLV with enflurane and isoflurane in the subgroup (n = 7) in whom PA catheters were inserted were compared using paired Student's t test. Significant correlations, between OLV PaOsub 2, and CO as well as between within-subject PaO2difference and isoflurane PaO2were tested using least-squares linear regression.
Thirty-one subjects were enrolled in the study. Three patients were excluded from the analysis because reinflation of the nondependent lung was requested by the surgeon before completion of the study protocol. Complete OLV and 2LV data for both anesthetic agents was obtained for 28 subjects, and pulmonary artery pressures, mixed venous samples, and COs were obtained in 7 subjects. Four subjects underwent esophagogastrectomy, and the balance of the procedures were thoracoscopic, including wedge resection, bullectomy, biopsy, and pericardial window. Nineteen were right-sided, and 9 were left-sided. There were no significant differences between groups 1 and 2 with respect to any of the preoperative variables. The values for both groups combined are shown in Table 2. The PaO2changes between 20 and 30 min during OLV with isoflurane or enflurane were 6.0% (range 0-23%) and 5.1% (range 0-21%), respectively; therefore, administration of the initial agent was not prolonged after the first 30 min of OLV. During the study, in no patient was SpO2sufficiently decreased (< 90%) to require reinflation of the nonventilated lung.
There were no significant differences in the measured variables attributable to the order of administration of anesthetic gases (i.e., group 1 vs. group 2); therefore, mean and SD values refer to pooled data from both groups. Significant main effects on PaOsub 2 alone were found due to both anesthetic gas (P < 0.000001) and mode of ventilation (P = 0.000001) as well as the interaction of these two factors (P = 0.000003). PaO2values were significantly higher after 10, 20, and 30 min OLV with 1 MAC isoflurane than during OLV with 1 MAC enflurane (P < 0.003) but did not differ during 2LV (Figure 1). The average differences were 37, 44, and 48 mmHg at 10, 20, and 30 min OLV, respectively. There were no significant differences between anesthetics in the other measured or derived hemodynamic or respiratory variables (Table 3and Table 4). In the PA catheter subgroup, both PaO2and PvO2during OLV with 160 plus/minus 53 mmHg and 51 plus/minus 8.6 mmHg isoflurane, respectively, were significantly (P = 0.002, P = 0.003) higher than during OLV with 125 plus/minus 44 and 47 plus/minus 8./5 enflurane, respectively.
During OLV, the difference in PaO2between isoflurane and enflurane was proportional to the absolute value of isoflurane PaO2(R = 0.65, P = 0.0001; Figure 2). There was a significant correlation between PaO2and CO during OLV for each anesthetic separately (enflurane r1= 0.88, isoflurane r2= 0.79, P r1= r2> 0.69) and combined (r = 0.81); the combined results are shown in Figure 3. Although no difference in CO was found between anesthetics in the PA catheter subgroup, because of the small number of cases (n = 7), the beta error for calculation of a false-negative result (beta = 0.88) exceeds the limits that permit a statistically significant conclusion.
The purpose of this investigation was to determine whether there is a difference between two commonly used volatile anesthetics with regard to arterial oxygenation during OLV. The main result was the finding of a higher PaO2during OLV for 30 min with 1 MAC isoflurane compared to 1 MAC enflurane. In the subgroup of patients with PA catheters, PaO2values during OLV with either enflurane or isoflurane correlated with CO.
The pathophysiology of OLV is not completely understood. Pa sub O2during OLV is known to be influenced by factors such as hypoxic pulmonary vasoconstriction in the nonventilated lung, and the alveolar pressure in the ventilated lung. We have demonstrated that the choice of volatile anesthetic also affects PaO2during OLV.
Primary therapy to prevent arterial oxygen desaturation during OLV is the use of a high FIO2. However, this is not always advisable. An increasing number of drugs, such as bleomycin, amiodarone, and mitomycin-C, have been associated with postoperative pulmonary oxygen toxicity when a high FIO2has been used intraoperatively for thoracic surgery. The next most useful method to prevent or treat hypoxemia during OLV is the application of CPAP to the nonventilated lung. However, CPAP has not solved the problem of hypoxemia during OLV. In patients with obstruction of the bronchus to the nonventilated lung or a bronchopleural fistula, CPAP will not generate adequate alveolar oxygen tension in the nonventilated lung to relieve hypoxemia. CPAP also may interfere with surgical access in the operative hemithorax, particularly during thoracoscopy, which tends to defeat the usefulness of OLV. Other methods of dealing with hypoxemia during OLV include repeated inflation of the nonventilated lung, application of positive and expiratory pressure, or manipulation of the tidal volume in the ventilated lung. These other maneuvers either tend to hinder surgery or are unreliable. Because these methods of maintaining arterial oxygenation in this situation are not completely satisfactory, it is important to be aware of all potential management options that influence the PaO2during OLV. The choice of volatile anesthetic is one of these options.
In comparative human studies, halothane has been shown to be associated with lower mean PaO2levels during OLV than an intravenous anesthetic technique. Both isoflurane [6,18]and enflurane have been shown to provide arterial oxygenation that was not significantly different than that during intravenous anesthesia. Although isoflurane has been demonstrated to cause inhibition of HPV in several animal models, [20,21]the one human study of isoflurane in usual clinical concentrations (1-1.5%) during one-lung hypoxia showed that its effect on HPV was almost immeasurable. The volatile anesthetics have not been compared previously in a randomized crossover fashion.
Patients were studied during prolonged periods of OLV that involved minimal trauma to the nonventilated lung (thorascopic surgery and esophageal surgery). This population was studied in an attempt to minimize the effects of surgical manipulation of the nonventilated lung on the partition of blood flow between the lungs during OLV. It is assumed that surgical manipulation of pulmonary vessels may temporarily inhibit HPV. .
The improved PaO2values during OLV with isoflurane versus enflurane may be attributed to either decreased venous admixture due to decreased perfusion of the nonventilated lung or increased mixed venous oxygen content secondary to increased CO. The significantly higher Pv with barO2during isoflurane OLV in the subgroup with PA catheters suggests the latter is more important. In patients with large shunt fractions, such as those seen during OLV (e.g., 20-30%), an increase in Pv with barO2can have a significant effect on PaO2. .
A greater CO is a possibility with isoflurane compared to enflurane. It has been shown in dogs that isoflurane causes less depression of CO than does enflurane at equivalent MAC levels. By design, the current study was terminated when a significant difference between the two anesthetics with respect to PaO2was demonstrated. The data do not permit any definitive conclusion about differences in CO between the two volatile anesthetics. Other factors besides CO, such as oxygen consumption (V with dotO2), influence the Pv with barO2. However, there is little reason to suspect that changes in VO2played any significant role in the findings, because the subjects remained anesthetized, paralyzed, and at a constant core temperature during the study period, and the sequence of anesthetic administration was randomized.
A positive correlation between PaO2and CO normally is anticipated during anesthesia, and it is expected that this effect of CO on PaO2will be magnified in patients, such as ours, with a large QS/QT. However, it has been suggested based on animal studies that increases in CO during OLV will tend to inhibit HPV because of passive dilatation of the pulmonary vasculature and result in a deterioration of PaO2. From this concept, it has been extrapolated that, during clinical OLV, it may be best to avoid drugs that increase CO. One other human study, which used inotropic agents, showed that augmentation of CO leads to an increase and not a decrease in PaO2during OLV. .
When the relationship between PaO2and CO in the current study is compared to the expected results using a computer lung model for HPV, the results are consistent with a mild degree of inhibition of HPV due to the volatile anesthetics. This partial decrease of HPV does not override the augmentation of PaO2because of increased CO and subsequent increased Pv with barO2when V with dotO2is held constant.*
The differences in PaO2values between OLV with isoflurane versus enflurane for individual patients were most noticeable in patients with high PaO2values. The observed mean difference between isoflurane and enflurane mixed venous oxygen content of 0.51 ml/100 ml, as seen in the PA catheter subgroup, will tend to increase the PaO2in a patient with a high PaO2(e.g., 400 mmHg) more than in one with a low PaO2(e.g., 70 mmHg). When using a high FIO2(e.g., 1.0) during OLV, the small difference in arterial oxygen content changing from enflurane to isoflurane may not be of any clinical benefit in patients who are on the borderline for developing hypoxemia. Also, it is not possible to estimate the effect of alterations of the anesthetic doses on the observed differences in oxygenation between the two anesthetics.
Ventilatory maneuvers probably have more effect on PaOsub 2 during OLV than the choice of anesthetic agent. A study by Capan et al. found that the mean increase in PaO2during OLV when CPAP of 10 cmH2O was applied to the nonventilated lung was 131 mmHg. This compares to a difference in mean PaO2values of 48 mmHg between 30 min of OLV with isoflurane and 30 min of OLV with enflurane in this study. However, CPAP of 10 cmH2O results in a large resting volume in the nonventilated lung, which tends to impede surgery.
In summary, 1 MAC isoflurane anesthesia was associated with greater PaO2values than was 1 MAC enflurane in a randomized crossover study during OLV with an FIO2of 1.0. Arterial oxygenation during OLV showed a positive correlation with CO. Whether these associations are of clinical importance when a lower FIO2or CPAP to the nonventilated lung are used remains to be studied.
The authors thank C. Colligan, for her secretarial assistance.
*Marshall B: Personal communication, 1993.