During one-lung ventilation (OLV) for thoracic surgery, the nonventilated lung remains perfused, leading to an increase in shunt fraction and a decrease in oxygenation. Oxygenation during OLV can be improved by increasing blood flow to the ventilated lung or decreasing blood flow to the nonventilated lung. Inhaled nitric oxide (NO) decreases pulmonary resistance to flow in various animal models and in clinical studies (1,2) Theoretically, the administration of inhaled NO during OLV could increase oxygenation by selectively decreasing pulmonary resistance and increasing blood flow to the ventilated lung. A number of studies have investigated the effects of inhaled NO on oxygenation and pulmonary hemodynamics during OLV (1,3–6). These studies found that during OLV, NO decreases pulmonary artery pressure but fails to increase Pao2. However, these studies were limited by small patient numbers, coadministration of volatile anesthetics, supine position, closed chest, or a combination of these factors. Furthermore, these studies investigated the effects of NO on Pao2 but not on desaturation (oxygen saturation of hemoglobin [O2Hb] <90%) during OLV. We therefore conducted a prospective study to evaluate the effects of NO on Pao2 and on the occurrence of desaturation during OLV. We conducted the investigation at three levels of inspired fraction of oxygen (Fio2) because changes in Pao2 are best detected during high Fio2, whereas those in O2Hb are best seen during low levels of Fio2. This is because during high Fio2, even small changes in shunt fraction lead to appreciable changes in Pao2. During low levels of Fio2 or Pao2, however, small changes in Pao2 lead to large changes in O2Hb.
All patients were anesthetized, ventilated, and monitored according to a standard protocol. Three consecutive measurements of oxygenation and hemodynamics were made at 10, 20, and 30 min after thoracotomy in patients in the lateral decubitus position and after establishment of OLV. During OLV, patients were ventilated either with or without NO. Oxygenation and desaturation were evaluated at an Fio2 of 0.3, 0.5, or 1.0.
After obtaining approval of the local ethics committee and informed written consent of patients, we studied ASA physical status I–III patients scheduled either for elective video-assisted thoracoscopic surgery (VATS) or open thoracotomy for lung resection in the lateral decubitus position. Exclusion criteria were ASA physical status IV and V and previous lobectomy or bilobectomy in the medical history. Four categories of patients were identified depending on the scheduled surgical procedure: VATS, metastasectomy, lobectomy, or pneumonectomy. Within the type of surgical procedure, patients were randomized on a 1:1 basis to be treated either with or without 20 ppm NO during OLV. Furthermore, patients were ventilated with an Fio2 of 1.0, 0.5, or 0.3. The levels of Fio2 were not randomized but used in a subsequent manner (Fio2 = 1, first group of 50 patients; Fio2 = 0.5, second group of 52 patients; and Fio2 = 0.3, last group of 50 patients). Preoperative lung perfusion scans were performed in a standard fashion with 99mTc-MAA (macro-aggregated albumin) in 60 of the 152 patients: in 2 patients undergoing VATS, in 20 patients undergoing metastasectomy, in 29 patients undergoing lobectomy, and in 9 patients undergoing pneumonectomy.
Patients were premedicated with 25 mg clorazepate dipotassium PO on the evening before surgery and with 0.5 mg atropine, 7.5–15 mg piritramide, and 2.5 mg droperidol IM 1 h before the scheduled operation. Anesthesia was induced with propofol (1–2 mg/kg) and remifentanil (0.5–1.0 μg/kg). Rocuronium bromide (0.6 mg/kg) was used to facilitate endotracheal intubation. Anesthesia was maintained with continuous infusion of propofol (6–10 mg · kg−1 · h−1) and remifentanil (5–20 μg · kg−1 · h−1). We intubated patients with a left- or right-sided double-lumen endotracheal tube (Broncho-Cath™, Mallinckrodt, Athlone, Ireland), size 39F or 41F for men and 37F or 39F for women. The correct position of the tube was controlled with a fiberoptic bronchoscope after intubation and after positioning the patient in the lateral decubitus position (7). If a thoracic epidural catheter was placed (mostly T6–8 interspaces) for postoperative pain relief, the catheter was tested with 3 mL lidocaine 1% and 0.015 mg epinephrine before the induction of general anesthesia. No further epidural medication was given until the end of the study period. A radial arterial line was inserted in every patient. A central venous catheter was inserted only if the condition of the patient and the operative procedure made it necessary. Intraoperative monitoring included electrocardiography, pulse oximetry, nasopharyngeal temperature, neuromuscular monitoring, measurement of ventilation pressures and ventilation volumes, measurement of end-tidal carbon dioxide concentration (Capnomac; Datex, Helsinki, Finland), and urine output. The sensor of the pulse oximeter (Datex, Helsinki, Finland) was placed either on a finger or the ear lobe, whichever led to better Spo2 readings. To avoid intraoperative hypothermia, an air warming blanket was placed on the lower extremities, and the arms were covered with a blanket.
An intensive care unit respirator (Evita 2; Dräger, Lübeck, Germany) was used to ventilate the patients during the operative procedure. This was because this ventilator was modified to provide NO at appropriate concentrations and complied with patient safety requirements. The ventilation was pressure controlled, with a peak inspiratory pressure of 30 cm H2O and a positive end-expiratory pressure of 5 cm H20 with oxygen or oxygen and air. Ventilation was adjusted to maintain end-tidal CO2 at approximately 30 mm Hg during OLV. NO (20 ppm) was delivered from a gas cylinder (1000 ppm of NO in nitrogen; AGA, Lidingö, Sweden) into the inspiratory limb of the ventilator by using a delivery device (NODOMO; Dräger, Lübeck, Germany). The NODOMO adjusts NO flow proportionally to the flow of the respirator to maintain a stable NO concentration. In addition, NODOMO can reliably measure NO concentrations in the inspiratory limb of the ventilator.
We started OLV immediately before thoracotomy or trocar placement for thoracoscopy. The operated lung was deflated to atmospheric pressure. Lung collapse was verified by view and monitored by continuous capnometry of the operated lung. After a stabilization period of 10 min, arterial blood gases, including O2Hb measured by cooximetry (ABL 625; Radiometer Copenhagen, Denmark), heart rate, mean arterial pressure, Spo2, and ventilatory variables were measured or noted. Measurements were repeated 20 and 30 min after the beginning of OLV. We planned to discontinue the study if surgical occlusion of blood flow to the nonventilated lung or lung lobe took place; however, this was not necessary in any of the patients studied. If at any time Spo2 decreased <91%, OLV was interrupted, the collapsed operated lung was reexpanded, and a continuous positive airway pressure of 5–10 cm H2O with an Fio2 of 1.0 was applied to maintain oxygenation.
Data were analyzed by using analysis of variance, with the factors time (10, 20, and 30 min), treatment (NO versus no NO), and Fio2 (0.3, 0.5, and 1.0). A χ2 test was used for analyzing the frequency of desaturation in the groups of patients studied. Bland and Altmann tests were used to assess differences between Spo2 and O2Hb during the study. We used the Statistical Package for Social Sciences (SPSS; SPSS Inc., Chicago, IL) version 9.0 for statistical analysis. A P < 0.05 was considered significant. Data are presented as mean and sem or as proportions when appropriate.
We included 152 patients in this study. Demographic data, type of surgery, and preexisting lung or cardiovascular disease were not different among the groups studied (Table 1). There were at least 25 patients in each of the six groups (NO versus control at Fio2 = 0.3, 0.5, or 1.0) studied. During OLV, respiratory variables (tidal volume, respiratory frequency, minute volume, and peak and plateau airway pressures) were not different between patients treated with or without NO. Pulse oximetry desaturation, as defined by an Spo2 <91%, occurred in none of the patients ventilated with an Fio2 of 0.5 or 1.0 and in only two patients ventilated with an Fio2 of 0.3.
Mean Pao2, Paco2, Spo2, and O2Hb did not differ significantly between patients ventilated with or without NO at any Fio2 level or at any time point (Tables 2–4). Arterial desaturation as defined by O2Hb <90% occurred in 20 of 76 patients receiving NO versus 19 of 76 control patients (P = 0.5). O2Hb <90% occurred in 40 of 219 time points in patients receiving NO versus 36 of 227 time points in control patients (P = 0.2). In patients receiving Fio2 = 0.5, O2Hb <90% occurred in 16 of 78 time points in patients receiving NO versus 6 of 78 time points in control patients (P = 0.04). In patients receiving Fio2 = 0.3 or 1.0, O2Hb <90% was not significantly different between the two groups. Levels of methemoglobin ranged between 0.0% and 0.9% (mean ± sem, 0.39 ± 0.01) in the Control group and between 0.0% and 1.2% (mean ± sem, 0.50 ± 0.01) in patients ventilated with NO (P < 0.001).
During OLV, mean Pao2, Spo2, and O2Hb increased with increasing Fio2 (Tables 2 and 3). The frequency of O2Hb <90% decreased with increasing Fio2 (Table 5, P < 0.001). O2Hb <90% occurred in 26 of 50 patients (50 of 146 time points) with an Fio2 = 0.3, in 11 of 52 patients (22 of 156 time points) with an Fio2 = 0.5, and in 2 of 50 patients (5 of 150 time points) with a Fio2 = 1.0 (P < 0.01 for Fio2 effect for both patients and time points). At Fio2 = 0.3, there are four missing time points because two patients with an Fio2 = 0.3 received continuous positive airway pressure in accordance with the study protocol.
Mean difference between Spo2 and O2Hb during OLV was 2.9% ± 0.1%; differences ranged between +10 and −4. The Spo2 − O2Hb difference decreased with increasing Fio2 (3.8 ± 0.2, 3.0 ± 0.1, and 2.1 ± 0.2 for Fio2 = 0.3, 0.5, and 1.0, respectively;P < 0.01).
Pao2 was significantly higher during right lung ventilation (left thoracic surgery) than during left lung ventilation (P < 0.01;Table 6). Across all Fio2 levels, O2Hb <90% occurred in 18 (26%) of 69 patients (39 [19%] of 207 time points) during right lung ventilation and in 23 (28%) of 83 patients (49 [20%] of 245 time points) during left lung ventilation (P = 0.8 for both patients and time points). At Fio2 = 1, O2Hb <90% occurred in 5 of 99 time points during left lung ventilation and in 0 of 51 time points during right lung ventilation (P = 0.1). At Fio2 = 1.0, Pao2 was significantly higher in patients undergoing pneumonectomy (390 ± 26 mm Hg) and in those undergoing lobectomy (256 ± 20 mm Hg) than in those undergoing metastasectomy (164 ± 13 mm Hg) or VATS (166 ± 12 mm Hg) (P < 0.05 for pneumonectomy versus all groups and lobectomy versus metastasectomy and VATS). At Fio2 = 1.0, mean perfusion of the ventilated lung was 45% in patients undergoing metastasectomy (mean of 6 perfusion scans), 53% in patients undergoing lobectomy (mean of 10 perfusion scans), and 68% in patients undergoing pneumonectomy (mean of 4 perfusion scans).
We ventilated patients undergoing OLV for lung surgery either with or without 20 ppm NO and found that NO did not improve oxygenation or reduce the occurrence of O2Hb <90%. In contrast, ventilating patients with increasing levels of Fio2 during OLV dose-dependently improved oxygenation and reduced the rate of O2Hb <90%.
Regarding the effects of NO on Pao2 during OLV, our study confirms previous findings in smaller patient groups. In a cross-over study, Wilson et al. (6) alternately administered control or study gas (oxygen 70%, N2O 30% vs oxygen 70%, N2O 30% with 40 ppm NO) every 15 minutes in six patients during OLV. They found that NO did not improve oxygenation, did not decrease pulmonary pressure, and did not decrease shunt fraction during OLV. Moutafis et al. (5) randomized 20 patients in two groups; one group received only oxygen during OLV, and the other group received oxygen and 20 ppm NO. After 30 minutes of OLV, oxygenation was not significantly different in the two groups of patients. In another study, Fradj et al. (3) allocated 60 patients undergoing lung surgery to a Control or NO (20 ppm) group. After randomization, NO was administered only if Pao2 decreased to <70 mm Hg, which occurred in eight patients of the Control group and eight patients of the NO group. Oxygenation improved in two patients in each group. Thus, our study and previous studies strongly suggest that NO inhalation does not improve oxygenation during OLV. Whereas previous studies were hampered by small patient numbers, coadministration of volatile anesthetics (6), supine position (1), closed chest, or a combination of these factors (3), our study avoids those problems and provides information not only on oxygenation but also on occurrence of hypoxemia (O2Hb <90%) during OLV.
In our study, inhaled NO failed to reduce the frequency of O2Hb <90%. Studying the frequency of O2Hb <90% may be interesting because at low Pao2 levels, even small changes in Pao2 may lead to appreciable changes in hemoglobin saturation. Surprisingly, in the group receiving Fio2 = 0.5, there were significantly more O2Hb <90% time points in the NO group as compared with controls. The level of methemoglobin in the NO group does not adequately explain this because mean methemoglobin levels were 0.4 in the Control group and 0.5 in the NO group. Thus, NO does not decrease the rate of O2Hb <90% during OLV.
Previous dose-response studies have found that NO levels >5 ppm are effective in producing a response in the pulmonary circulation. In our study, we administered NO at 20 ppm, we monitored NO in the inspiratory limb of the respirator in all patients, and methemoglobin levels were increased in the NO group as compared with the Control group. We used a respirator and an NO delivery device that was produced for the explicit purpose of NO application. Therefore, the negative finding in our study is not the result of technical failure or of small NO levels in inspiratory air.
That inhaled NO during OLV may increase oxygenation by diverting blood from the nonventilated to the ventilated lung is an attractive theory. Furthermore, inhaled NO during OLV may also increase oxygenation by improving ventilation-perfusion matching within the ventilated lung. Apparently, these mechanisms were either not operative or insufficient to increase oxygenation or reduce the occurrence of O2Hb <90% during OLV. A number of reasons may explain this. During OLV in the lateral decubitus position, approximately 75%–80% of the cardiac output flows through the ventilated, dependent lung. This suggests that pulmonary vessels in the ventilated lung are dilated to accommodate this increased flow, and further dilation with NO is not possible. Studies investigating the effects of NO in patients with lung injury support this hypothesis. In a clinical trial, Benzing et al. (8) studied whether cardiac output may modify the effects of NO on venous admixture in patients with lung injury. They found that in patients with high cardiac output, NO was less effective in decreasing venous admixture than in patients with low cardiac output. Thus, increased flow in the ventilated, dependent lung during OLV may have prevented NO from further increasing flow or improving ventilation-perfusion matching.
We had decided to ventilate the collapsed lung only if Spo2 decreased to <91%. Surprisingly, Spo2 did not decrease <91% in any of the 102 patients ventilated with an Fio2 of 0.5 or 1.0, and in only 2 of 50 patients ventilated with an Fio2 = 0.3. Because Spo2 overestimated O2Hb, we used O2Hb <90% as a measure of hypoxemia for the purpose of this study. Our study shows that during OLV, patients ventilated with increasing Fio2 levels show corresponding decreases in occurrence of O2Hb <90%. This means that increasing Fio2 levels may be an effective means of increasing oxygenation or treating or preventing hypoxemia during OLV. The use of high levels of Fio2 is detrimental in animal models of lung injury (9) and promotes atelectasis during anesthesia (10). Whether the administration of high levels of oxygen during the relatively short period of OLV may be injurious seems unlikely. However, other maneuvers, such as application of continuous positive airway pressure to the nonventilated lungs, may obviate the need for high Fio2 during OLV.
Despite an Fio2 of 1.0, O2Hb <90% occurred at least once in 2 (4%) of 50 patients or in 5 (<4%) of 150 time points. This finding means that hypoxia cannot be completely eliminated when using Fio2 = 1.0 but that it occurs very infrequently. It should be noted that we used O2Hb to define hypoxemia and desaturation. O2Hb is the measured arterial saturation devoid of carboxyhemoglobin and methemoglobin. Previous studies using Sao2 (arterial saturation, which may include carboxyhemoglobin and methemoglobin) and Spo2 have reported hypoxemia rates of up to 10% during OLV (11–13). Lower levels in our study may be due to increased monitoring of double-lumen tube position at our center (7) and the ventilation strategy used during OLV. Because of study circumstances (see Methods), we used an intensive care unit ventilator and a pressure-controlled mode to ventilate patients during OLV. Most other studies reporting on oxygenation have used volume-controlled ventilation. Tugrul et al. (11) have shown that pressure-controlled ventilation may lead to higher levels of Pao2 than volume controlled ventilation.
The two patients with O2Hb <90% during Fio2 = 1.0 had their left lung ventilated during OLV. Because the left lung is smaller than the right lung, oxygenation is usually lower during left lung ventilation as compared with right lung ventilation. The small difference in shunt volume between right and left lung ventilation leads to a difference in Pao2 that increases with increasing Fio2 levels. At Fio2 = 1.0, mean Pao2 was 283 mm Hg during right lung ventilation and 169 mm Hg during left lung ventilation. Furthermore, occurrence of O2Hb <90% tended to be less (P = 0.1) during right versus left lung ventilation. In agreement with our finding, Slinger et al. (13) found the side of surgery to be an independent predictor of Pao2 during OLV; however, they report neither Pao2 values nor desaturation rates.
Our study also shows that oxygenation was better during pneumonectomy and lobectomy as compared with metastasectomy and thoracoscopy. This may be explained, in part, by lung perfusion studies. In our study and at Fio2 = 1.0, the perfusion of the ventilated lung was higher during lobectomy (55%) and pneumonectomy (68%) as compared with metastasectomy (45%). This means that the pathology involved in patients undergoing lobectomy and pneumonectomy may increasingly impair the perfusion of the nonventilated lung, thus decreasing shunt fraction and improving oxygenation. Thus, oxygenation during OLV may also depend on whether the right or left lung is ventilated and whether blood flow to the operated lung is impaired.
In conclusion, our study shows that NO administration is not associated with better oxygenation during OLV. In contrast, increasing Fio2 levels were associated with better oxygenation during OLV.