Case Scenario: Management of Intraoperative Hypoxemia during One-lung Ventilation

A 50-yr-old man undergoing resection of the upper right lung for neoplasia developed hypoxemia 15 min after the start of OLV. Preoperative examination revealed satisfactory cardiopulmonary status. Preoperative rest and dobutamine stress echocardiography revealed normal left ventricular function without any wall motion abnormalities. Doppler ultrasound examination of the carotid arteries was normal. Preoperative pulmonary function tests were normal without any obstructive disease (forced expired volume in 1 s was 3.7 l and forced expired volume in 1 s/vital capacity was 73%). Blood electrolytes, hemoglobin concentration, and renal function were normal. There was no history of bronchopulmonary infection, and arterial blood gas values in room air were Pao²83 mmHg, Paco²38 mmHg, pH 7.40, and HCO³25 mm. Electrocardiogram showed regular sinus rhythm, and blood pressure on the morning of surgery was 128/65 mmHg. The patient refused epidural anesthesia for management of postoperative pain. Selective bronchial intubation of the left main-stem bronchus was performed with a 39F left-sided double lumen tube (DLT) (Broncho-part; Rüsch, Kermen, Germany). The correct position of the tube was immediately confirmed with a fiberoptic bronchoscopy, and the patient was then placed in the lateral position. Volume-controlled ventilation, including a 7 ml/kg tidal volume (V^T) of predicted body weight under 100% Fio², a 12 cycles/min respiratory rate, and a 5 cm H²O positive end-expiratory pressure (PEEP), was used. End-tidal carbon dioxide and plateau pressure were 31 mmHg and 19 cm H²O, respectively. No intrinsic PEEP was observed. OLV was initiated without any change in ventilator settings. Fifteen minutes later, the patient exhibited profound hypoxemia with a significant decrease in pulse oxymetry from 94% to 88%. Arterial blood gas values were pH 7.41, Pao²52 mmHg (100% Fio²), Paco²40 mmHg, HCO³24.8 mM, and 87% Sao². The patient's hemodynamics and electrocardiogram remained stable. Expiratory flow and expiratory V^Twere unchanged, and no leak was noticed. The correct position of the DLT was immediately confirmed by fiberoptic inspection. End-expiratory flow was not interrupted by the next insufflation and reached zero before the next respiratory cycle, therefore no dynamic hyperinflation or intrinsic PEEP (iPEEP) was observed. Once the surgeon was informed, the nondependent lung (nonventilated lung) was expanded manually by administration of pure oxygen, and a continuous positive airway pressure (CPAP) at 5 cm H²O was subsequently applied. This strategy allowed rapid improvement of the patient's oxygenation, and oxymetry pulse could be maintained above 95% throughout the surgical procedure under OLV. During the manual expansion of the nondependent (nonventilated) lung, hemodynamics remained stable. Because the surgical procedure was a right open thoracotomy, it was not impeded by the application of CPAP to the nondependent lung, which was stopped at the end of OLV. The postoperative course of this patient was uneventful. He did not exhibit any hypoxemia after tracheal extubation.

The first step when a patient suffers from hypoxemia (i.e ., a drop in pulse oxymetry less than 90%) during OLV is to stop surgery, increase the inspired oxygen fraction to 100%, and rapidly restore two-lung ventilation (TLV). That can be performed with a manual reexpansion of the lung using the hand bag with pure oxygen flow and an airway pressure limited valve between 20 and 30 cm H²O; the reexpansion of the deflated lung is visual. Once acceptable oxygenation has been restored, two simultaneous actions must be performed: (1) a reversible cause of hypoxemia must be ruled out before restarting surgery and (2) oxygenation of the patient must be improved to prevent the recurrence of hypoxemia. Because hypoxemia during OLV is mainly related to ventilator or perfusion mismatch, strategies to restore oxygenation should have two objectives: to improve alveolar ventilation and pulmonary perfusion. The first consists of restoring alveolar ventilation of the nondependent lung by applying oxygen inflation with CPAP and of the dependent lung by performing a recruitment maneuver and applying PEEP. Perfusion, on the other hand, involves strategies to limit intrapulmonary shunt: positioning of the patient (strict lateral decubitus position on the side of the dependent lung provides better oxygenation than semilateral or even supine position), surgical lung compression or clamping the pulmonary artery of the nondependent lung, limitation of airways pressure of the ventilated lung including limitation of intrinsic PEEP, and the use of moderate V^T(6–8 ml/kg) or pressure-controlled ventilation. Some pharmacological treatments, such as nitric oxide and almitrine, can also be used to decrease intrapulmonary shunt. An algorithm for the treatment of hypoxemia during OLV is summarized in figure 1.

Because dislodgment of the DLT is a common cause of hypoxemia, the first step is to check that the DLT is in the correct position and to clear secretions or blood from the tube or the respiratory tract. Initially, blinded bronchoaspiration can be performed on both sides of the DLT. It is sometimes sufficient to clear proximal secretions and blood. Clinical examination of the correct position of the DLT with auscultation and inspection of the DLT is necessary but sometimes fails to identify DLT malpositioning. Consequently, correct positioning of the DLT must be systematically confirmed by fiberoptic.1,2Fiberoptic bronchoscopy must be repeated once hypoxemia occurs or in the presence of increased airways pressure. The DLT can be misplaced leading to impaired oxygenation and inadequate lung separation with high airway pressure or significant leak.2Hemodynamic conditions also need to be assessed. Although patients undergoing lung resection are vulnerable to fluid overhydration, the maintenance of cardiac output, mainly by fluid loading, is required for that oxygen delivery to meet patient metabolic needs. As OLV induces increase in right ventricular overload, intraoperative assessment of ventricular function may be useful.3Right ventricular dysfunction may also be related to volume overload or intrinsic myocardial depression.3Despite some limitations in the analysis of right ventricular function, transesophageal echocardiography is probably the best intraoperative monitoring technique to use during OLV.3The two main signs that must be sought are the right ventricular dilatation as well as paradoxical septal motion.4Moreover, maintenance of an adequate level of anesthesia allows limiting hemodynamic disturbances related to a too-deep level of anesthesia.

Although it is frequently applied, increasing Fio²in the ventilated lung is often not sufficient to improve oxygenation because OLV is responsible for an important shunt-like effect of approximately 30%.5Administration of oxygen to the nondependent lung is usually used to treat hypoxemia during OLV, but it may also be used to prevent it. Oxygen may be administered with or without CPAP of 5–10 cm H²O to the nondependent lung. If no CPAP valve is available, the hand bag with airway pressure limited valve can be used with a concentration of 5–10 cm H²O. The oxygen flow and airway pressure limited valve must be adapted to obtain inflation without interfering with surgery. After informing the surgeon, transitory reinflation of the collapsed lung is necessary by applying a higher airway pressure between 20 and 30 cm H²O and then using CPAP to keep the lung at a constant level of inflation; the reexpansion of the lung is visual. Oxygen flow and CPAP are very effective at improving oxygenation during OLV. A CPAP concentration of 5–10 cm H²O is usually sufficient and does not seem to impede open-chest surgery.6,7Only rarely does hypoxemia during OLV persist when CPAP is applied. In this case, surgery should be transitorily stopped, and a switch to TLV should be performed. However, OLV should be maintained when hypoxemia exists preoperatively or when emergency thoracic surgery is required (hemoptysis or postoperative anastomosis leakage).

Conflicting results have been reported concerning the application of PEEP to the dependent lung during OLV.8,9In some patients, the application of PEEP to the dependent (ventilated) lung might be beneficial by restoring functional residual capacity to normal values,10thus preventing atelectasis when its value is titrated along the static compliance curve.11However, PEEP may also paradoxically impair oxygenation by shifting pulmonary blood flow to the nondependent lung, thereby increasing intrapulmonary shunt. It was commonly thought that a similar tidal volume should be used for both TLV and OLV. Nevertheless, use of a large tidal volume during OLV produces end-inspiratory lung volumes close to concentrations that contribute to acute lung injury.12When small V^Tare used, especially during high Fio², more atelectasis may occur, worsening oxygenation.13Therefore, to avoid atelectasis, PEEP should be applied when VT is small. Valenza et al .14showed that oxygen responders to 10 cm H²O PEEP during OLV were patients in whom forced expiratory volume in 1 s was more than 72%, and that these patients were less likely to develop iPEEP. In obstructive patients, the use of PEEP may sometimes interfere with iPEEP and decrease oxygenation.9,11,15No adverse events have been reported in patients in whom a PEEP up to 10 cm H²O was applied. However, when alveolar pressure is increased, the risk of barotrauma must be considered. Morbidly obese patients given anesthesia exhibit more severe impairment of respiratory mechanics and gas exchange than normal subjects.16PEEP improves respiratory function and oxygenation in these patients, and this is significantly correlated with alveolar recruitment.16,17Therefore, the use of PEEP has to be considered a useful tool to improve alveolar ventilation of the lung during OLV, but its effect on dependent lung perfusion, and therefore on oxygenation, has to be monitored by arterial blood gases.

During OLV in the lateral position, atelectasis zones in the dependent lung contribute to ventilation/perfusion mismatch. These disturbances tend to increase shunt in the nondependent lung. Short-term application of unusually high airway pressure can restore near normal lung volumes at end-expiration and improve oxygenation. Tusman et al .18examined the beneficial effects of an alveolar recruitment strategy with pressure-controlled ventilation (PCV) during OLV. Using alveolar recruitment, they reported that values of Pao²were almost as high as during TLV. As Rothen et al .19pointed out, in anesthetized patients without lung injury, peak airway pressure of 40 cm H²O is required to open atelectatic zones. Alveolar recruitment should be performed progressively and transiently with PCV to avoid or reduce hemodynamic adverse effects. Similarly to PEEP, increasing alveolar pressure may be responsible for barotrauma. This has not been described during OLV, but it is important to bear this complication in mind because a pneumothorax in the dependent lung during OLV with a patient in the lateral position is difficult to diagnose and treat rapidly.

PCV has been suggested to reduce peak airway pressure and intrapulmonary shunt, thereby improving oxygenation and limiting the risk of barotrauma.20The benefits of PCV can be explained by the reduction in dynamic airway pressure. Compression of small intraalveolar vessels during inflation of the alveoli produces increased resistance to pulmonary blood flow in the dependent lung and tends to divert blood to the nonventilated alveoli of the nondependent lung. However, the latter is most closely related to the mean airway pressure or, more specifically, the mean alveolar pressure. Our group has recently demonstrated that peak airway pressure is modestly affected by a switch from volume-controlled ventilation to PCV during OLV when it was measured inside the main ventilated bronchus.21Our results suggest that the beneficial effect of PCV during OLV is not mediated by a decrease in peak pressure. To date, the improvement in oxygenation induced by PCV compared with volume-controlled ventilation during OLV remains controversial.22–24 

During OLV, it is possible to modulate hypoxic pulmonary vasoconstriction (HPV) using inhaled nitric oxide or almitrine. Almitrine is a peripheral chemoreceptor agonist that potentiates HPV. Inhaled nitric oxide only accesses the ventilated alveoli, where it causes localized vasodilation and increases perfusion to the adequately ventilated region of the lung. Almitrine, although delivered systematically, enhances constriction primarily under hypoxic conditions, and thus, at low doses, selectively enhances HPV in the nonventilated operative lung. Moutafis et al .25showed that inhaled nitric oxide alone (20 ppm) in patients undergoing OLV did not improve oxygenation. However, when inhaled nitric oxide was associated with intravenous almitrine (16 μg · kg⁻¹· min⁻¹), arterial partial oxygen pressure was significantly increased. Inhaled nitric oxide and PEEP could have a synergistic effect on oxygenation.26However, Michelet et  al .26found that inhaled nitric oxide only improved oxygenation when PEEP was high (i.e ., 15 cm H²O), and it was associated probably with overdistension. Intravenous almitrine can also be used alone.27A continuous infusion of 8 μg · kg⁻¹· min⁻¹limits OLV-induced hypoxemia without causing any hemodynamic disturbances in patients without pulmonary hypertension.27An adverse effect of almitrine is elevation of pulmonary vascular resistance, which, if excessive, could potentially induce right ventricular dysfunction. It is possible to use almitrine as a treatment strategy in patients with hypoxemia during video-assisted thoracoscopy where CPAP and intermittent TLV are impossible to use because they will interfere with surgery. Almitrine is not approved for clinical use in North America because of the toxicity of the drug, which includes peripheral neuropathy.

The influence of inhalational anesthetics versus  intravenous anesthetic agents on HPV remains a large subject of controversy. Animal28,29as well as human30,31studies found that isoflurane is associated with increased intrapulmonary shunt fraction as a result of HPV inhibition during OLV. Conversely, inhalation anesthetics, by affecting cardiac output, may decrease perfusion of the nondependent lung, and thus decrease shunt fraction. This could explain that HPV may be significantly affected by volatile anesthetics without any clinically relevant changes in oxygenation.32Moreover, a recent clinical study suggests that sevoflurane attenuates the effect of OLV on the pulmonary inflammatory response and improves clinical outcome.33 

The surgeon has access to the perfusion of the nondependent lung. Consequently, he/she is able to reduce pulmonary blood flow, and therefore transpulmonary shunt. Ishikawa et al .34showed that repeated compression of the nondependent lung using a retractor during the surgical procedure may significantly improve oxygenation in patients undergoing OLV. These authors did not report whether compression of the pulmonary vessels induced complications. This strategy is almost never used in clinical practice and is certainly not feasible during a video-assisted thoracic procedure. Likewise, during pneumonectomy, the pulmonary artery of the nondependent lung may be rapidly ligated, and this will immediately improve oxygenation by significantly reducing the shunt-like effect.

The potential therapeutic target for improving oxygenation during one-lung ventilation is summarized in figure 2.

During OLV anesthesia in the lateral position, the pulmonary shunt-like effect ranges from 15% to 40% because of total collapse of the nondependent lung and an increase in atelectasic areas in the dependent lung.5The consequence is a profound ventilation-perfusion mismatch resulting in arterial hypoxemia. Pao²/Fio²ratio is often less than 200 and sometimes even less than 100.35Because transpulmonary venous shunting depends on the amount of cardiac output that is not oxygenated, the less the perfusion of the nonventilated lung and the more the perfusion of the ventilated lung, the higher the Pao²during OLV. One of the reasons why the pulmonary shunt-like effect during OLV is less than 50% is HPV. HPV is unique to the pulmonary circulation because most systemic beds dilate in response to hypoxia. The stimulus for HPV is primarily the alveolar partial pressure of oxygen, consistent with the localization of the oxygen sensor in the resistant pulmonary arteries.36The onset of HPV is rapid, occurring within 7 s in isolated lungs. Conversely, it resolves within minutes after normoxic ventilation is restored.37There is considerable variation in the strength of HPV, even at sea level. For example, many patients with previous high-altitude pulmonary edema have unusually strong HPV, which may contribute to their high-altitude pulmonary edema susceptibility.38In terms of genetic variability, there are individuals who are perfectly healthy but have weak HPV. This may explain why young patients without any comorbidity, undergoing thoracoscopy, sometimes develop life-threatening hypoxemia during OLV.

In-hospital mortality is estimated to be approximately 6% after pneumonectomy. In a recent survey, an analysis of the European and French Thoracic Surgery Database indentified the following independent risk factors for perioperative death: male sex, age, functional performance status, dyspnea score, American Society of Anesthesiologist score, and extent of resection.39During the procedure, depending on the definition, the rate of hypoxemia occurring during OLV has been estimated to be between 4% and 10%.35,40,41However, prediction of intraoperative hypoxemia is much less accurate. The individual response to OLV is hard to predict because it relies on a complicated interaction between the patient's genetic background (in terms of HPV response),36,37preoperative function tests,35underlying disease, anesthetic management, and surgical procedure.

As reported previously by Slinger et al .,42low intraoperative Pao²during TLV may predict hypoxemia during OLV. The distribution of perfusion between the two lungs can be measured preoperatively by perfusion scans and might be helpful to predict hypoxemia during OLV.43Ventilation and perfusion, and consequently shunt, will largely be influenced by the position and the size of the tumor. A large, central, compressing tumor will result in more progressive perfusion impairments because the lung is already partially preoperatively excluded from perfusion and, subsequently, a less intraoperative shunt-like effect will occur once the lung with the tumor is not ventilated than a small, peripheral, noncompressing mass. Although other pulmonary function tests could be useful for predicting postoperative complications and outcome, their accuracy as reliable indicators of hypoxemia is unclear.

An inadequate response or failure of HPV will result in severe hypoxemia. The HPV response relies essentially on genetic background, but weak HPV should be expected in patients with chronic obstructive pulmonary disease or cirrhosis.44,45In addition, HPV failure may be related to iatrogenic and anesthetic factors. Inhaled anesthetics, calcium channel blockers and other vasodilators, alkalosis, high pulmonary vascular pressure, and hypothermia decrease HPV and worsen oxygenation during OLV.46Although it is not possible to classify anesthetic management as a predicting factor for hypoxemia before surgery, airway management, fluid strategy, and ventilation remain a challenge in OLV. The importance of the preoperative workup must be emphasized, and a clear protocol for each patient, involving both the anesthesiologist and the thoracic surgeon as a team, should be decided before surgery. As mentioned above, oxygenation during OLV will be higher during left thoracotomy and in the lateral decubitus position. Surgical manipulations may impede expansion of the dependent lung, and mediastinal pressure can result in indirect compression or modification of cardiac output resulting in hypoxemia.

Hypoxemia is a constant threat during OLV and the ventilation strategy is crucial to decrease its incidence. However, as with the management of acute respiratory distress syndrome 10–20 yr ago, new concepts arise. To avoid intraoperative hypoxemia as the primary endpoint of OLV, the same V^Tduring TLV and OLV has been recommended as in the past. However, it is important to remember that ventilation can be harmful, and acute lung injury is now known to be a problem after thoracic surgery.

Large volume ventilation might not be sufficient per se  to induce acute lung injury and volume-induced lung injury in healthy lungs during short periods of anesthesia, but conditions are totally different during OLV and thoracic surgery. Use of high Fio², frequent underlying pulmonary diseases, surgical manipulations, and resections provide a pathophysiologic basis for multiple hit aggression and subsequent susceptibility to acute lung injury and volume-induced lung injury. In both acute respiratory distress syndrome and OLV, clinicians will have to deal with a reduced lung volume and reduced functional residual capacity resulting in a heterogenous lung and an increase in shunt and ventilation/perfusion mismatch. In these conditions, it is understandable that usual ventilatory setting in the past (high V^Tand no positive end-expiratory pressure) might not be adequate. Gama de Abreu et al . demonstrated that OLV with high V^T(10 ml/kg) caused secondary lung injury in the dependent lung.47Licker et al .48found intraoperative ventilatory hyperpressure to be an independent risk factor for acute lung injury after lung resection. Likewise, a retrospective study found that postoperative pneumonectomy respiratory failure was associated with the use of higher intraoperative tidal volume during OLV.49 

The ideal setting during OLV remains unclear. Michelet et al .12reported a decrease inflammatory response and improved lung function after OLV with protective low volume ventilation using a reduced V^Tof 5 ml/kg and 5 cm H²O PEEP. Schilling et al .50found similar results, including a decrease in alveolar inflammation. However, Shultz et al .,51in a recent review of the literature, could not find clear data supporting the use of decrease V^T. The choice depends on whether the focus is on intraoperative oxygenation or on postoperative complications and clinical outcome. Focusing on intraoperative gas exchange, Katz et al .52showed that large tidal volume ventilation produced the highest Pao². Although it is multifactorial, excessive V^Tand inadequate settings may create an insult in susceptible patients, and acute respiratory distress syndrome network results should clearly alert the thoracic anesthesiologist to the potential repercussions of overinflation. It is noteworthy that the normal tidal volume in mammals is 6.3 ml/kg, and protective ventilation seems to be the right approach.

The consequence of a reduction in tidal volume is consistent alveolar derecruitment with the occurrence of atelectasis. In acute respiratory distress syndrome patients, reducing end-expiratory lung volume or pressure can be injurious, even when tidal volume or peak pressures are controlled; conversely, increasing PEEP can be protective.53In fact, the coexistence of atelectatic and open alveoli may result in shear forces that exceed transpulmonary pressure.54Mechanical ventilation of areas of lung that are atelectatic is associated with repetitive collapse and reexpansion with each breath; this cyclic recruitment is also a source of pulmonary inflammation. Therefore, PEEP may play a key role in reducing pulmonary inflammation and maintaining oxygenation once V^Tis reduced. A really important concept is the interaction between V^Tand PEEP during bipulmonary ventilation and especially during OLV. Depending on the lung mechanics and perfusion status for any individual patient, there is a need to adjust ventilation strategies to keep the lung open without impeding perfusion and minimizing alveolar damage. As Karzai et al .1stated in their recent review on hypoxemia and OLV: “Clinically, there are no viable means of determining this best ventilation strategy in an individual patient. Therefore, patient condition, physiologic reasoning, and best available clinical evidence must be used to tailor a ventilatory strategy that best fits the needs of the individual patient.” The clinical postoperative benefits of lung protective strategies using decrease V^Tcombined with a certain concentration of PEEP during OLV are still to be demonstrated. In the meantime, it appears logical that decrease V^Twith limited plateau pressure must be applied with the initiation of OLV. In addition to a reduction of V^T, PEEP is also required and should be adjusted to reduce alveolar derecruitment and maintain oxygenation with values probably between 5 and 15 cm H²O. The interaction between PEEP and iPEEP during volume-controlled ventilation in chronic obstructive pulmonary disease patients is complex. In fact, Caramez et al . observed three specific responses during stepwise PEEP application in these patients during TLV: (1) paradoxic response with an increased isovolume expiratory flow and a decrease in functional residual capacity, plateau pressures, and total PEEP while external PEEP was applied; (2) biphasic response (no deleterious effect on expiratory-flow, functional residual capacity, plateau pressures, or total PEEP until external PEEP exceeded a specific concentration); and (3) classic overinflation (any application of external PEEP resulted in evident lung overinflation).55These three specifics responses in chronic obstructive pulmonary disease patients during TLV will also have different effect on the dependent lung perfusion during OLV. Therefore, any modification of the concentration of PEEP during OLV in chronic obstructive pulmonary disease patients has to be controlled because the effects on thoracopulmonary compliance and oxygenation are unpredictable.11 

In conclusion, hypoxemia may occur in 5–10% of patients during OLV. The prevention and treatment of hypoxemia during OLV involves control of the correct position of the DLT by fiberoptic bronchoscopy and the use of an appropriate ventilation strategy.1High inspired oxygen fraction associated with CPAP can be applied to the nondependent lung, and this strategy is both effective and safe. Alveolar recruitment, higher concentrations of PEEP, and almitrine can also be used, but pure oxygen flow with CPAP with a constant acceptable inflation to the nondependent lung remains the safer strategy.

The authors thank Erwan Floch, Pharm.D., Newmed Publishing Services, Paris, France, for reviewing the article.