Oxygen Attenuates Atelectasis-induced Injury in the In Vivo  Rat Lung

Institutional ethics approval (conforming to the guidelines of the Canadian Council for Animal Care) was obtained for all experiments.

Male Sprague-Dawley rats (200–300 g) were used in all experiments. General anesthesia was induced with ketamine and xylazine and was maintained with intravenous ketamine. A surgical tracheotomy was performed. The following ventilation parameters were used: tidal volume (V^T), 8 ml/kg; frequency, 50–60 min⁻¹; PEEP, 1 cm H²O; and 21% O², using a small-animal ventilator (model 683; Harvard Apparatus, MA). Increments of pancuronium were used for muscle relaxation, having confirmed no response to stimulation. Four series of experiments were performed.

To determine the dose–response effects of supplemental oxygen, animals were randomly assigned to one of six groups based on the inspired oxygen concentration (%) as follows: group 1, 21%; group 2, 24%; group 3, 30%; group 4, 40%; group 5, 70%; and group 6, 100%. After randomization, PEEP was discontinued, the assigned inspired oxygen was initiated, and the remainder of the ventilator settings was unchanged for the next 150 min. No recruitment maneuvers or PEEP was used during the remainder of the experiment. Pulmonary alveolar–capillary permeability was assessed by measurement of Evans blue in bronchoalveolar lavage (BAL) fluid after intravenous administration. Evans blue (15 mg/kg) was administered intravenously 20 min after randomization.14At the end of the experiment, the animal was killed, and the heart–lung block was dissected. The right lung was lavaged with a single aliquot of 3 ml (normal saline), which was injected and aspirated a total of three times. The resulting bronchoalveolar fluid was collected, and the volume was recorded. The concentration of Evans blue in the supernatant was then determined spectrophotometrically by measurement of its absorbance at 620 nm. At completion of the experiments, in addition to BAL, plasma samples were also collected, and lung wet/dry weight ratio was calculated.15 

To determine the cardiac and pulmonary vascular effects of supplemental oxygen, six pairs of animals were randomly assigned to ventilation (same parameters as in series 1) with either room air (21% O²) or 100% O². The same experimental protocol was followed as for series 1 except that Evans blue was not administered. A baseline echocardiogram was performed before randomization and repeated, at 30-min intervals for 150 min, by an experienced echocardiographer. Each echocardiogram took approximately 3 min to complete. The echocardiogram was used to determine the following parameters: pulmonary arterial acceleration time, right ventricular end-diastolic area, and left ventricular end-diastolic area. Echocardiography was performed using a Hewlett-Packard (Andover, MA) prototype Sonos 5500 echocardiographic system (M2424A) with a 7.5-MHz transducer. The animals were examined in the supine position with the chest closed. The transducer was gently applied to the left parasternal border to obtain a short-axis view of the left (distal to the mitral valve leaflets) and right ventricles at a frame rate of 113 Hz and depth of 2 cm. Serial two-dimensional images were acquired using short-axis views at baseline and at 30-min intervals until completion of the experiment. From a left parasternal long-axis view, the pulsed Doppler gate (with an angle of insonation below 20°) was carefully placed at the tip of the mitral valve leaflets to obtain diastolic flow patterns (E and A waves) and velocities. Data were obtained at baseline and at 30-min intervals. All images acquired were stored electronically for subsequent analysis.

Bronchoalveolar lavage was performed at the end of the experiment on all animals, and prostanoid concentrations were measured. An additional group (unventilated controls, n = 6) was anesthetized and killed, and BAL was performed for control concentrations of prostanoids. Prostanoids were measured using a Sciex API 4000 tandem mass spectrometry16in the electrospray ionization negative ion mode with TurboIonSpray (MDS Sciex, Ontario, Canada). Samples were spiked with 1.0 ng of a mixture of deuterated analogs of the prostanoids to be measured (Cayman Chemical Co., Ann Arbor, MI), acidified to pH 4 with 1N HCl, and extracted with ethyl acetate. The organic phase was evaporated to dryness and transferred to siliconized mini vials for analysis by mass spectrometry. Quantitation was performed by comparing the deuterium/protium ratio of the individual prostanoids in the sample with standard lines generated from authentic prostanoids. An Agilent high-performance liquid chromatography 1100 was at the front end, equipped with a Zorbax SB-phenyl column (3.0 × 50 mm, 3.5 m spherical size; Chromatographic Specialties Inc., Brockville, Ontario, Canada). The tandem mass spectrometry source temperature was maintained at 500°C, and the ion source voltage at −4,500 V. Compounds were separated on high-performance liquid chromatography with a direct inlet into the mass spectrometry source. The high-performance liquid chromatography solvent followed the program: water/acetonitrile 80/20 at sample injection and maintained for 2 min, 75/25 for 0.5 min, 50/50 by 5 min, 45/55 by 6.2 min, and 0/100 by 11 min, where this was maintained for 1.5 min. The solvent was then recycled to 80/20 for the next run. The flow rate was at 400 ml/min. High-performance liquid chromatography solvents contained 2 μl/l propionic acid.

To illustrate that atelectasis was present in animals ventilated without recruitment, computed tomography scans were performed in animals randomly assigned to one of three groups: ventilation with recruitment (21% O²) or ventilation without recruitment (21% O²or 100% O²). Animals were imaged with use of a General Electric Lightspeed Ultra 8 slice scanner (Milwaukee, WI). The scan protocol included a 9.6-cm field of view, 120 KVp, 100 mAs, and a 512 × 512 matrix. Axial 2-s scans through the lungs were obtained at a slice thickness of 0.65 mm with edge reconstruction. Sections at the level of the division of the right mainstem bronchus were selected.

To provide quantitative data demonstrating that supplemental oxygen increased the degree of atelectasis, we randomly assigned animals to ventilation (same parameters as in series 1) with either room air (21% O²) or 100% O². The same experimental protocol was followed as for series 2. At the end of the experiment, the trachea was occluded at end-expiration, and the functional residual capacity was measured using direct volume displacement.17,18 

Parametric data are expressed as mean ± SD, and nonparametric data are expressed as median ± interquartile range. Two-way analysis of variance was used for parametric statistical analysis, and analysis of variance on ranks (Kruskal-Wallis) was used for nonparametric statistical analysis. For serial measurements over time (i.e. , echocardiographic data), two-way repeated-measures analysis of variance was used, and where data were nonparametric, log transformation was performed before analysis of variance. The Bonferroni correction was applied where multiple comparisons were made with post hoc t  tests. Regression analysis was used to determine dose–response relations. Significance was set at P < 0.05.

Thirty animals were entered for series 1. One animal did not meet the randomization criteria and was excluded from the study. All other 29 animals were eligible. All baseline parameters (weight, hemoglobin, partial pressure of arterial oxygen [Pao²], partial pressure of arterial carbon dioxide [Paco²], bicarbonate, base excess, airway pressure, and lung compliance) were comparable among the six groups (table 1). The survival with 21% O²was four out of seven (57%); all animals survived the experimental protocol in the other groups (P = 0.05; table 1and fig. 1).

The comparisons of the respiratory and acid–base data are presented (table 1). Pulmonary microvascular leakage was maximal in the absence of supplemental oxygen, and the addition of supplemental oxygen resulted in a dose-dependent decrease in protein leakage (fig. 2A). The baseline values of Pao²were similar in all the groups during breathing of 21% O², before changing to the experimental inspired oxygen (table 1). Atelectasis resulted in a decrease in Pao²in the 21% O²group (final vs.  baseline; P < 0.05). There was a dose-dependent increase in Pao²associated with increasing inspired oxygen (fig. 2B).

Increased inspired oxygen was associated with an increased alveolar–arterial oxygenation gradient (AaO²) in a dose-dependent manner in each of the experimental groups. To overcome effects of increasing inspired oxygen on AaO²gradient associated with the nonlinear shape of the oxygen dissociation curve, we also calculated the delta AaDO²(final − baseline AaO²) and demonstrated that increases in inspired O²are associated with progressive increase in delta AaDO²(P < 0.05; fig. 2C).

Twelve animals were anesthetized and were allocated in a pairwise manner to either 21% or 100% O². All animals met the entry criteria and were eligible for the study. Two out of six animals in the 21% O²group did not survive the complete protocol, and the corresponding paired animals in the 100% O²group were terminated at the same time point. Furthermore, six additional animals were anesthetized and killed and were used as controls (not mechanically ventilated).

Pulmonary arterial acceleration time, which reflects impedance to right ventricular ejection, decreased progressively over time during ventilation with 21% O²but not with 100% O²(fig. 3A). The ratio of right ventricular versus  left ventricular diastolic dimension was greater overall in ventilation with 21% versus  100% O²(fig. 3B). Normal control conditions of the left and right ventricle are presented (fig. 4A). An illustrative example of the impairment of right ventricular function, and underfilling of the left ventricle in the 21% O²group, is presented (figs. 4B and C).

The BAL concentrations of important prostanoid metabolites were measured (fig. 5and table 2). The concentrations of thromboxane B²(reflecting the vasoconstrictor thromboxane A²; fig. 5A) and prostaglandin E²(the vasodilatory prostanoid; fig. 5B) were increased in both 21% and 100% O²compared with control animals, but neither was different between 21% versus  100% O²(figs. 5A and B). The concentrations of stable prostanoid, prostaglandin F^2α, was significantly greater with 21% O²versus  both 100% O²and controls (P < 0.05), and there was no difference in the concentrations between 100% versus  controls (fig. 5C). There were no among-group differences in the levels of 8-isoprostane, 6-keto-prostaglandin F¹-α, prostaglandin D², leukotriene B⁴, 12-hydroxyeicosatetraenoic acid, or arachidonic acid (table 2).

Computed tomography demonstrated that atelectasis was not present in animals ventilated using a recruitment strategy (figs. 6A and B). Atelectasis was evident in animals ventilated without recruitment and with 21% O²(figs. 6C and D); however, atelectasis was far more apparent after ventilation without recruitment maneuvers and with 100% O²(figs. 6E and F).

Functional residual capacity was significantly lower in animals ventilated with 100% versus  21% O²(P < 0.05; fig. 7).

This article demonstrates the dose-dependent effect of supplemental oxygen in reducing adverse effects of atelectasis-associated regional lung hypoxia, despite worsening the degree of atelectasis, in anesthetized rats. We have previously demonstrated that ventilation in this model without recruitment leads to atelectasis or a reduced functional residual capacity.9The survival rate in the animals ventilated without recruitment with 21% inspired O²was 57%, which was similar to the survival rate of 59% in our previous study for animals ventilated with 21% O²in the absence of recruitment.9We confirmed that atelectasis without supplemental oxygen leads to reduced survival, impaired oxygenation, and adverse vascular effects but also demonstrated an increase in pulmonary prostanoid concentrations. In the current study, despite worsening atelectasis, addition of supplemental oxygen eliminated mortality, improved systemic oxygen tension and cardiac function, and decreased pulmonary prostaglandin F^2αconcentrations.

Clearly, such dramatic results are not observed in patients who undergo general anesthesia without the use of either PEEP or recruitment maneuvers. Nonetheless, atelectasis is a common occurrence after general anesthesia,2and recent work suggests that prevention or reversal of atelectasis may be associated with a reduced incidence of lung injury and multiorgan failure in high-risk postoperative patients.10The latter report demonstrated that the early use of continuous positive airway pressure reduces the need for intubation in postoperative respiratory insufficiency and lessens the incidence of severe complications.10Our results may be relevant to the clinical setting and prevention of atelectasis, and the addition of supplemental oxygen may improve outcome in high-risk patients.

Atelectasis leads to reduced local alveolar lung tissue oxygen tension.1,3Therefore, impaired oxygenation may be a mediating influence of atelectasis-induced increases in lung permeability. Our previous study demonstrated ultrastructural evidence of microvascular endothelial disruption in nonrecruited animals as a cause for increased lung permeability.9In this study, pulmonary microvascular protein leakage was maximal in the absence of supplemental oxygen.

We assessed pulmonary microvascular protein leakage using Evans blue concentration in BAL fluid. The use of this technique has been used extensively both in the ex vivo  19and the in vivo  lung.20–22The exact passage and distribution of Evans blue dye in atelectatic regions in comparison with expanded lung is probably a complicated process. However, we use liquid for the BAL, which enters and expands collapsed units far more easily than gas.23Although there was no effect of inspired oxygen on lung wet/dry ratio, other authors have found wet/dry ratio to be a less sensitive marker of lung injury than Evans blue dye leakage.24Numerous errors may occur in gravimetric assessment of lung injury, including evaporative losses, impact of ventilation, and failure to reveal any regional heterogeneity.25 

Areas of the lung that become atelectatic comprise alveoli that are nonaerated as well as alveoli that are poorly aerated.26Volumetric computed tomography analysis of the lungs during general anesthesia has demonstrated nonhomogenous distribution of atelectasis.27The morphologic basis for what constitutes atelectasis has undergone significant review recently with the perspective by Hubmayr,28who argues that the dependent lung may be derecruited because it is filled with fluid or foam, and not collapsed as originally thought. Therefore, it is possible that supplemental oxygen can impact areas of atelectasis and correct local hypoxia. However, a cost of using higher concentrations of inspired oxygen was a dose-dependent increase in the alveolar–arterial oxygenation gradient, consistent with the visible worsening of atelectasis on imaging, and an increase in quantitatively measured functional residual capacity. Thus, the lung injury and increased pulmonary vascular resistance are likely as a result of the atelectasis-associated regional lung hypoxia, not by the atelectasis-associated volume loss per se , because supplemental oxygen prevents the injury and corrects the hypoxia but worsens the atelectasis.

Recent data indicating that alveolar hypoxia may result in pulmonary vascular leak via  decreased lung neprilysin expression29or may induce lung inflammation through macrophage recruitment30support this contention. Atelectasis in combination with pulmonary artery occlusion depletes rabbit lung adenosine triphosphate,31a finding that is consistent with a pathogenic role of local oxygen supply.

When alveolar oxygen tension is reduced in regions of the lung, local vasoconstriction of small pulmonary arteries is induced. This phenomenon, hypoxic pulmonary vasoconstriction, results in a dual response of increased pulmonary perfusion pressure and diversion of blood flow from hypoxic regions to normoxic regions.32The mechanism of decreased blood flow to an atelectatic lobe of lung seems to be through hypoxic pulmonary vasoconstriction,33,34regardless of whether the chest is open or closed or whether ventilation is spontaneous or due to positive pressure.

There are a number of mechanisms that could explain the microvascular injury in the atelectatic lung. First, the increased stress and strain that mechanical ventilation provides to the remaining opened alveoli after atelectasis has formed. Second, the amount of lung tissue subjected to intratidal opening and closing might have been higher in the nonrecruited group, despite lung collapse, because these animals were ventilated with a PEEP of 0 cm H²O, in comparison with a PEEP of 2 cm H²O in the “recruited” group.9Third, the injury might be the consequence of pulmonary hypertension and the reduction of the vascular bed available for pulmonary circulation. In this context, more energy may be dissipated through pulmonary blood flow in relatively restricted (i.e. , with less capacitance, higher resistance, or both), thereby causing flow-induced injury. Fourth, the local hypoxia induced by atelectasis formation might induce a direct hypoxic injury on the endothelial wall. In the current context, the first and second mechanisms may be less likely because the increase in atelectasis per se  did not increase injury. However, microvascular endothelial injury may be relevant as supplemental oxygen normalized right ventricular function, and we have previously demonstrated ultrastructural evidence of microvascular endothelial disruption in nonrecruited lungs.9In addition, local hypoxia–induced injury may be important, and although previous studies demonstrating the effects of hypoxia refer to expanded lung, they may be relevant here.30,35In fact, in terms of effects on pulmonary vascular tone, this is important, because Barer et al.  36have demonstrated, using an isolated in vivo  feline lung lobe, that the alveolar and mixed venous oxygen tension, not the volume of the lobe per se , were the major determinants of the local pulmonary vascular resistance.

In addition to impaired oxygenation, atelectasis increases pulmonary vascular resistance.37,38We have previously demonstrated that right ventricular dysfunction and pulmonary hypertension occur in the absence of recruitment9and have demonstrated that the cause of death in this model is cardiogenic shock on the basis of right ventricular failure. There are several mechanisms whereby atelectasis can compromise the pulmonary vasculature. For example, functional residual capacity is related to pulmonary vascular resistance through direct effects on the pulmonary vasculature. Studies in ex vivo  lungs suggest that the relation of lung volume with pulmonary vascular resistance followed a U-shaped curve, with the pulmonary vascular resistance lowest at functional residual capacity.39,40However, such concepts may not translate to the in vivo  setting, and when examined in vivo , it has been found that local oxygen tension, not lung volume, is the key determinant of hypoxic pulmonary vasoconstriction.36Therefore, as long as increasing inspired oxygen tension is observed to increase systemic oxygenation, that implies that the atelectasis is relative (i.e. , not total), and the reduction in pulmonary vascular resistance observed is consistent with previous literature.36Therefore, high levels of alveolar oxygenation seem to supersede accompanying absorption atelectasis in regulation of pulmonary vascular tone as well as vascular leak.

Constriction of small pulmonary arteries and arterioles and focal vascular injury are features of pulmonary hypertension. Thromboxane A²is derived from cyclic endoperoxides by the action of thromboxane synthetase. It is a biologically active metabolite of arachidonic acid produced in the phospholipids of cell membranes. Because thromboxane A²is both a vasoconstrictor and a potent stimulus for platelet aggregation, it may be an important mediator of pulmonary hypertension. Its effects are antagonized by prostacyclin, which is released by vascular endothelial cells.41Thromboxane receptor blockade prevents the pulmonary hypertension and the decline in oxygenation seen in experimental acute lung injury.42 

Prostaglandin endoperoxide is the direct precursor of the prostaglandins and thromboxane. Thromboxane B²is the stable metabolite of thromboxane A²formed through thromboxane synthase. Similarly, 6-keto-prostaglandin F^1αis a stable metabolite of prostacyclin, formed through prostacyclin synthase, whereas prostaglandin E²and prostaglandin F^2aare the stable end products formed through an isomerase and a reductase, respectively. The current study demonstrated an increase in both thromboxane B²and prostaglandin E²in the BAL fluid of derecruited rat lungs regardless of the addition of supplemental oxygen. However, in the presence of 100% inspired oxygen, the concentrations of prostaglandin F^2αwere significantly lower than with 21% oxygen. Prostaglandin F^2αis a cyclooxygenase metabolite of arachidonic acid involving the action of a reductase on prostaglandin endoperoxide, and it is also responsible for potent vasoconstriction.43A few possibilities exist as to why prostaglandin F^2αis decreased in the presence of oxygen but thromboxane B²and prostaglandin E²are not. One hundred percent oxygen may divert the substrate prostaglandin endoperoxide (arachidonic acid is converted to prostaglandin G², which then decomposes to prostaglandin H²) from the pathway leading to the production of prostaglandin F^2α, or it may inhibit the reductase enzyme responsible for the production of prostaglandin F^2α. Ventilation in the presence of either 21% or 100% O²was associated with increased BAL concentrations of thromboxane B², prostaglandin E², and prostaglandin F^2αcompared with nonventilated control animals.

There was no difference in the local pulmonary concentrations of the vasoactive prostanoids thromboxane B²(reflecting constrictor thromboxane A²) or prostaglandin E²(a vasodilatory prostanoid) between 21% and 100% O²conditions. However, because prostaglandin F^2awas increased in the 21% oxygen group and significantly lower in the presence of 100% inspired O², this mediator may be important in atelectasis-induced pulmonary hypertension and injury and warrants further investigation in a future study.

The current study has several limitations that restrict extrapolation to the clinical context. First, we did not have a control group to compare the effects of oxygen during mechanical ventilation with PEEP and recruitment maneuvers. However, we have previously shown that mechanical ventilation with 21% O²in the context of recruitment was not injurious.9Second, the higher inspired oxygen led to an increased alveolar–arterial gradient through absorption atelectasis, which should ultimately increase, not decrease, hypoxic pulmonary vasoconstriction. We terminated the experiment at 2.5 h and did not explore the longer-term consequences of the use of a higher inspired oxygen. Third, echocardiograms were performed at 30-min intervals in all animals in series 2. Although a special pediatric probe was used, additional pressure on the chest wall might alter responses. In addition, we did not directly measure pulmonary artery pressures as we have previously done,9which may have provided additional insight. Finally, although prostaglandin F^2awas identified as a mediator in atelectasis-induced injury, we did not explore the effects of antagonizing this prostanoid or other mediators of hypoxia-induced lung injury, such as nitric oxide.

The current data confirm that mechanical ventilation without recruitment results in increased microvascular protein leakage, which is attenuated, in a dose-dependent manner, by addition of supplementary oxygen. In the absence of supplemental oxygen, the animals develop progressive right heart failure associated with increased pulmonary vascular resistance. The application of 100% oxygen prevents these effects but worsens the computed tomography evidence of atelectasis, suggesting that the microvascular injury is mediated by hypoxemia, not volume loss per se , and possibly by the hypoxia-induced alteration in pulmonary vascular tone. Finally, analysis of local stable metabolites suggests that vasoactive prostanoids do not mediate the pulmonary vascular effects associated with atelectasis in this model.

The authors wish to thank Guila Ben David and Nancy Parfield (Technologists, Department of Radiology, The Hospital for Sick Children, Toronto, Ontario, Canada) for their expertise in obtaining the computed tomography radiographs and Derek Stephens, Ph.D. (Biostatistician, Programme in Population Health Sciences, The Research Institute, The Hospital for Sick Children), for statistical advice.