Secondary Logo

Journal Logo

Short-Term Administration of a High Oxygen Concentration Is Not Injurious in an Ex-Vivo Rabbit Model of Ventilator-Induced Lung Injury

Section Editor(s): Takala, JukkaKopterides, Petros MD*†; Kapetanakis, Theodoros MD; Siempos, Ilias I. MD; Magkou, Christina MD; Pelekanou, Aimilia MD§; Tsaganos, Thomas MD§; Giamarellos-Bourboulis, Evangelos MD§; Roussos, Charis MD; Armaganidis, Apostolos MD*

doi: 10.1213/ane.0b013e31818f10f7
Critical Care and Trauma: Research Reports

BACKGROUND: Mechanical ventilation and administration of a high oxygen concentration are simultaneously used in the management of respiratory failure. We conducted this study to evaluate the effect of a high inspired oxygen concentration on ventilator-induced lung injury.

METHODS: Forty sets of isolated/perfused rabbit lungs were randomized for 60 min of pressure-control ventilation at a plateau inspiratory pressure of 25 or 15 cm H2O and positive end-expiratory pressure of 3 cm H2O while receiving 100% or 21% O2. The temperature, pH, and partial pressure of CO2 in the perfusate were maintained the same in all groups (n = 10 for each group). The outcome measures used to assess lung injury included: the change in weight gain and ultrafiltration coefficient, the frequency of vascular failure, the histological lesions and the concentration of tumor necrosis factor-α and malondialdehyde in the bronchoalveolar lavage fluid.

RESULTS: The two groups ventilated at the higher inspiratory pressure/tidal volume experienced greater weight gain and increases in the ultrafiltration coefficient, more frequently suffered vascular failure, and presented higher composite scores of histological damage than the two groups ventilated at the lower inspiratory pressure/tidal volume. Hyperoxia was not found to further increase any of the monitored markers of lung injury. No difference was noticed among the four experimental groups in the alveolar lavage fluid levels of tumor necrosis factor-α or malondialdehyde.

CONCLUSIONS: These findings suggest that short-term administration of a high oxygen concentration is not a major determinant of ventilator-induced lung injury in this experimental model.

IMPLICATIONS: Short-term administration of a high oxygen concentration is not a major determinant of ventilator-induced lung injury in this experimental model. When mechanical ventilation is instituted for a limited time (for example, in the operating room or in the initial phase of trauma and septic shock resuscitation), the main focus should probably be on the avoidance of lung over-distention and not of high oxygen concentration.

From the *Second Critical Care Department, Attiko University Hospital, University of Athens Medical School, Athens, Greece; †Department of Experimental Surgery, Evangelismos Hospital, University of Athens Medical School, Athens, Greece; ‡Department of Histopathology, Mitera Hospital, Athens, Greece; §Fourth Department of Internal Medicine, Attiko University Hospital, University of Athens Medical School, Athens, Greece; and ∥First Critical Care Department-Pulmonary Services, Evangelismos Hospital, University of Athens Medical School, Athens, Greece.

Accepted for publication July 31, 2008.

This study was performed in the Department of Experimental Surgery at Evangelismos Hospital in Athens, Greece, and was supported by the Thorax Foundation.

Address correspondence and reprint requests to Petros Kopterides, MD, 68 Kamaterou St, Kamatero–Athens 13451, Greece. Address e-mail to

A significant number of experimental studies recently summarized in excellent reviews1–6 and, most importantly, two randomized clinical trials7,8 have shown that mechanical ventilation may cause or exacerbate lung injury when it leads to lung over-distention and repetitive alveolar collapse and reexpansion. However, hyperoxia has been a recognized cause of lung injury9–12 via a number of proposed mechanisms (i.e., the generation of reactive oxygen species (ROS)13,14) that are encompassed by the unifying term pulmonary oxygen toxicity.10

Even though the institution of mechanical ventilation and the administration of a high oxygen concentration are interdependent and simultaneously used interventions in the management of respiratory failure, the study of their interaction had received little attention until recently. Two published studies reported increased lung injury in rodent lungs that were exposed in vivo to hyperoxia (defined as exposure to a fraction of inspired oxygen [Fio2] concentration > 90%) and mechanical stretch either sequentially15 or simultaneously.16 In addition, in a large tidal volume (VT) in vivo rabbit model of ventilator-induced lung injury (VILI), hyperoxia (defined as exposure to Fio2 = 50%), was shown to exacerbate lung injury via mechanisms independent of augmented cytokine expression or direct lipid peroxidation.17

Based on the above information, we conducted this study to determine whether the lung injury induced by lung over-distention is exacerbated by the administration of a high inspired oxygen concentration in an ex vivo rabbit model of lung stretch. We also sought to determine evidence of activation of inflammatory or redox mechanisms in this particular experimental model of lung injury.

Back to Top | Article Outline


The study was approved by the Veterinary Directorate of the Perfecture of Athens according to Greek legislation and in conformance with the 160/1991 Council Directive of the European Union.

Back to Top | Article Outline

Animal Instrumentation and Isolated/Perfused Heart-Lung Preparation

Forty white New Zealand male rabbits were used. Animals were housed in single metal cages and had access to tap water and standard balanced rabbit chow ad libitum. Room temperature ranged between 18 and 22°C, relative humidity ranged between 55% and 65% and the light/dark cycle was from 6 am to 6 pm.

Briefly, the animals were anesthetized with an IM injection of 1 mL/kg ketamine (100 mg/mL; Imalgene, Merial, Lyon, France) and 0.5 mL/kg xylazine (20 mg/mL; Rompun, Bayer, Leverkusen, Germany). A 20 G plastic catheter was placed in the jugular vein, a bolus dose of 5000 U of unfractionated heparin was administered, and the rest of the experiment was conducted under continuous IV drip of thiopental at a dose of 5–10 mg · kg−1 · h−1. An endotracheal tube was inserted via a tracheostomy and secured in place with a suture. A midline sternotomy was performed and the heart-lung preparation was excised en bloc and weighed (Kern 770/GS/GJ, Kern & Sohn Gmbh, Albstadt, Germany). Fifty milliliter of blood was collected after laceration of the inferior vena cava and the right subclavian artery and vein and later added to the perfusate. Perfusion cannulas were placed in the pulmonary artery via incision in the free wall of the right ventricle and in the left atrium via incision in the free wall of the left ventricle and dilation of the mitral valve annulus. The heart-lung preparation was suspended from a counterbalanced force transducer placed in series with a balance (BG 025, Mark-10, NY) before initiation of ventilation/perfusion. Ischemic time was defined as the time period between the animal's exsanguination and the initiation of ventilation/perfusion.

Back to Top | Article Outline

Perfusate Description

The perfusate in the circuit consisted of 300 mL of Krebs-Henseleit solution (Sigma-Aldrich Chemie GmbH, Steinheim, Germany) buffered to a pH of approximately 7.4 with the addition of ΝaHCO3 (88.4 mg/mL; Demo, Athens, Greece). Bovine serum albumin 5% and the 50 mL of blood that had been collected earlier were added to the perfusate in order to facilitate the recognition of vascular damage and hemorrhagic lesions by the histologist. The resulting hematocrit of the perfusate was approximately 3%,18 and the white blood cell count was negligible. A cyclooxygenase inhibitor was not added to the perfusate.

Back to Top | Article Outline

Circuit Description and Data Recording

The perfusion circuit, starting from the left atrium, consisted sequentially of: 1) the left atrium cannula, which was connected to 2) the plastic tubing which ended above 3) a reservoir that could be raised or lowered to adjust the effective left atrial pressure. From the reservoir, 4) plastic tubing led to 5) a digital rotary pump (Masterflex 07550; Cole-Parmer Instrument CO, Barrington, IL), which maintained the nonpulsatile flow, 6) a water bath (WB 7, Memmert GmbH + Co.KG, Schwabach, Germany) which kept the perfusate temperature constant at 36°C, and 7) a bubble trap that was placed after the water bath and before the 8) plastic tubing leading to 9) the pulmonary artery cannula and the lungs. In addition, the pulmonary artery cannula housed a multiparameter sensor (Trend Care, 7+™, TCM 7000, Diametrics Medical, Roseville, MN) that allowed continuous intraarterial blood gas monitoring of pH, Pao2, Paco2, and temperature.

The vascular pressure (in the pulmonary artery and the left atrium) and weight signals were amplified (AC bridge Amplifier–ABC module, Raytech Instruments, Vancouver, BC, Canada) and stored on a personal computer via an analog-digital converter (Direc/NEP 201 B Physiologic Recording System, Raytech Instruments, Vancouver, BC, Canada). The digital signals were later analyzed with dedicated software (Direcwin, Data Acquisition Program, Release version 2.18a, Raytech Instruments, Vancouver, BC, Canada).

Back to Top | Article Outline

Initial Hemodynamic and Ventilatory Settings

The heart-lung preparation was connected to the ventilator (T Bird AVS III, Thermo Respiratory Group, Palm Springs, CA) and remained for 5 min under continuous positive airway pressure (CPAP) of 5 cm Η2Ο while flow increased gradually from zero to 100 mL/min. The free end of the left atrial cannula was positioned so that the left atrial pressure was 6 mm Hg, ensuring West zone III conditions (pulmonary arterial pressure > left atrial pressure > alveolar pressure) along the vertical lung axis19 (Fig. 1).

Figure 1

Figure 1

In the next 15 min, flow increased gradually from 100 to 300 mL/min while the preparation was ventilated with pressure-control ventilation (PCV) at a plateau inspiratory pressure of 15 cm Η2Ο, positive end-expiratory pressure (PEEP) (ΡΕΕΡ) of 3 cm Η2Ο, a respiratory rate (RR) of 15/min and an inspiration: expiration (I:E) ratio of 1:2. After a 5-min stabilization period with CPAP at 5 cm Η2Ο, the heart-lung preparations were allocated to the hyperoxic and normoxic group and ventilated for 20 min with PCV 15/3 cm Η2Ο, RR: 15/min, I:E = 1:2, PEEP = 3 cm Η2Ο, and Fio2 = 100% or 21%, respectively.

Back to Top | Article Outline

Baseline Measurement of Capillary Ultrafiltration Coefficient and Capillary Pressure

After achievement of isogravimetric conditions (no weight change during CPAP 5 cm H2O), capillary pressure was measured in triplicate by the simultaneous double occlusion technique, as described elsewhere.20 The free end of the left atrial cannula was then raised by 6 cm while the weight of the heart-lung preparation was continuously recorded. Perfusion of the lung under West zone III conditions was documented by simultaneously recording the increase in pulmonary artery pressure associated with the step increment in left atrial pressure,21 and capillary pressure was measured again in triplicate. The ultrafiltration coefficient was calculated by dividing the rate of weight gain between the 5th and 7th min after the increase of the left atrial pressure by the change in the capillary pressure, and was subsequently normalized to the initial wet lung weight (thus, expressed in g · min−1 · mm Hg−1 per 100 g of lung tissue).22

Back to Top | Article Outline

Group Allocation and Ventilation/Perfusion Experimental Protocol

After left atrial pressure returned to the baseline value of 6 mm Hg and another 5-min stabilization period, each of the 40 heart-lung preparations was randomly allocated to 1 of the 4 experimental groups (n = 10 in each group) and ventilated with PCV for 60 min at 1 of 2 levels of peak inspiratory pressure (PIP) (25 cm H2O versus 15 cm H2O) combined with the administration of a high (100%) or normal (21%) oxygen concentration: group A (high inspiratory pressure– hyperoxemia: HPH), group B (high inspiratory pressure–normoxemia: HPN), group C (low inspiratory pressure–hyperoxemia: LPH), and group D (low inspiratory pressure–normoxemia: LPN). In all groups, PEEP was set at 3 cm H2O, RR was 15/min, and the I:E ratio 1:2. As expected, the two groups (HPH and HPN) ventilated at the higher PIP tended to have a lower Paco2, which was managed with the addition of CO2 in the venous reservoir.

Back to Top | Article Outline

Post Ventilation/Perfusion Protocol Measurements

After the completion of 60 min of ventilation and the second measurement of ultrafiltration coefficient, the flow was stopped, the perfusate was drained, and the catheters were removed from the pulmonary artery and the left atrium.

The weight gain, capillary ultrafiltration coefficient, histological analysis and measurements of tumor necrosis factor (TNF)-α and malondialdehyde (MDA) levels were performed in the same heart-lung preparations and not in different groups of animals.

Back to Top | Article Outline

Weight Gain and Ultrafiltration Coefficient

Weight gain was defined as the weight change between the second min of PCV and the end of the experiment (at the 60th min or earlier). We defined as ΔKf,c as the increase of the ultrafiltration coefficient between the 2 measurements (before and after the institution of the 60-min PCV). Of interest, a number of heart-lung preparations suffered vascular failure before the completion of 60 min of PCV; we defined vascular failure as the sudden occurrence of gross vascular leak or a mean pulmonary artery pressure change higher than 40 mm Hg.23,24 The ultrafiltration coefficient (and ΔKf,c) could not be measured in preparations that developed vascular failure.

Back to Top | Article Outline

Postexperiment Tissue Sampling

After the end of the ventilation/perfusion protocol, the right lung from each of the heart-lung preparations that had previously undergone 1 h of ventilation was lavaged by infusion of 20 mL of normal saline twice. The recovered bronchoalveolar lavage fluid (BALF) was centrifuged (500g for 10 min at 4°C) (Universal 32 R, Hettich Zentrifugen, Tuttlingen, Germany), and the supernatant was placed in plastic vials and kept frozen at −80°C for later determination of TNF-α and MDA.

In addition, the left lung was first fixed by intratracheal insufflation with 10% formalin at a hydrostatic pressure of 15 cm H2O and later cut into slices 5 mm thick in a coronal fashion from apex to base. Samples were embedded in paraffin and stained with hematoxylin and eosin. An experienced pathologist (C.M.) evaluated these samples in a blinded fashion and graded the degree of lung injury using a previously described scoring system with slight modifications.25,26 The following pathological features were determined: i) capillary congestion, ii) intraalveolar and iii) perivascular hemorrhage, iv) interstitial, and v) intraalveolar neutrophil infiltration. Each feature was scored from 0 to 4 based on its severity, and a cumulative total histological score was determined. Therefore, a total score of 0 represented normal histology and a score of 20 represented maximal damage.

Back to Top | Article Outline

Assays for TNF-α and MDA

TNF-α was measured by a bioassay on a L929 fibrosarcoma cell line, as described previously.27,28 Concentrations of TNF-α were estimated by the reduction of the optical density of control wells by unknown samples applying a standard curve generated by standard concentrations. All determinations were performed in quadruplicate. The interday variation of the assay was 13.75% and the lower limit of detection was 11.50 pg/mL.

Lipid peroxidation was estimated by the concentration of MDA, as described previously,27,29 with the thiobarbiturate assay. The pulmonary content of MDA has been previously shown to increase significantly under hyperoxic conditions, indicating enhanced lipid peroxidation.30 All determinations were performed in duplicate. The lower limit of detection was 0.25 μmol/L.

Back to Top | Article Outline

Statistical Analysis

Results are presented as mean ± sd in both figures and tables. One-way analysis of variance was used to determine the statistical significance of between-group differences. When statistical significance was indicated, it was further examined by post hoc analysis (Tukey's modification). A statistical software package (Statistica 6.0; StatSoft, Tulsa, OK) was used, and P < 0.05 was considered to be statistically significant.

Back to Top | Article Outline


Baseline Characteristics and Effectiveness of Protocol Implementation

The ischemic time and the animal or initial lung weights were not different among the four groups. In addition, the initial capillary pressures and ultrafiltration coefficient did not differ significantly among groups (Table 1).

Table 1

Table 1

Immediately after the beginning of the 60-min period of PCV, the mean pulmonary artery pressure and the pH, Paco2, and temperature of the perfusate were comparable among groups, whereas, as dictated by the study protocol, the hyperoxic groups (HPH, LPH) had higher perfusate Pao2 (Table 1).

Back to Top | Article Outline

Overt Vascular Failure

During the institution of the 60 min of PCV, none of the 20 heart-lung preparations ventilated at the lower PIP (and lower end-inspiratory volume) suffered vascular failure. On the contrary, 1 preparation (10%) of the HPH group and 2 preparations (20%) of the HPN group developed overt vascular failure during this phase of the experiment.

Back to Top | Article Outline

Pulmonary Edema Formation

The preparations ventilated at the higher PIP gained more weight at each 20-min time point and at the end of the experiment compared with the preparations ventilated at the lower PIP (n = 10 for each group) (Fig. 2). For example, after 40 min of PCV, the HPH group had gained 5.60 ± 2.80 g and the HPN Group 7.80 ± 8.30 g, whereas the LPH and the LPN groups had gained only 0.05 ± 1.50 and 0.11 ± 0.80 g, respectively (P < 0.05, for the comparisons between HP and LP groups). Weight gain was statistically significantly higher in the HPH and HPN groups even when it was normalized to the initial lung weight or the actual time of mechanical ventilation (Table 2). It should be noted that there was no difference in the weight gain between the HPH and HPN groups or between the LPH and the LPN groups. Therefore, hyperoxia per se had no effect on weight gain.

Figure 2

Figure 2

Table 2

Table 2

Back to Top | Article Outline

Ultrafiltration Coefficient

At the completion of 1 h of mechanical ventilation, the absolute Kf,c values (in g · min−1 · mm Hg−1 per 100 g lung tissue) were higher in the high-pressure/high-volume groups (HPH [n = 9]: 0.15 ± 0.09, HPN [n = 8]: 0.35 ± 0.17, LPH [n = 9]: 0.069 ± 0.03, LPN [n = 10]: 0.096 ± 0.08) (Table 3). The HPH and HPN groups seemed to have experienced the highest absolute and relative increase in ΔKf,c (HPH: 0.074 ± 0.10 or increase approximately 70.0%; HPN: 0.190 ± 0.17 or increase approximately 120.0%), whereas Kf,c of the low-pressure/low-volume groups did increase but less markedly (LPH: 0.004 ± 0.03 or increase approximately 6.0%; LPN: 0.028 ± 0.05 or increase approximately 43.0%). There was no difference in Kf,c between the LPH and the LPN groups at the end of the experiment, whereas it was higher in the HPN group compared with the other three groups.

Table 3

Table 3

Back to Top | Article Outline


The groups ventilated at the higher PIP/higher volume manifested more extensive histological lesions in terms of perivascular hemorrhage (HPH: 1.3 ± 0.4; HPN: 0.5 ± 0.4; LPH: 0.0 ± 0.0; LPN: 0.0 ± 0.0), intraalveolar hemorrhage (HPH: 1.1 ± 0.4; HPN: 1.6 ± 0.6; LPH: 0.0 ± 0.0; LPN: 0.0 ± 0.0) and capillary congestion (HPH: 1.4 ± 0.4; HPN: 1.3 ± 0.4; LPH: 0.6 ± 0.2; LPN: 0.3 ± 0.1). As a result, even though the infiltration of the interstitium (HPH: 0.3 ± 0.1; HPN: 0.4 ± 0.2; LPH: 0.1 ± 0.1; LPN: 0.2 ± 0.0) and the alveolar spaces (HPH: 0.2 ± 0.2; HPN: 0.1 ± 0.1; LPH: 0.1 ± 0.1; LPN: 0.0 ± 0.2) by leukocytes were not different among the groups, the composite histological score was higher in the HP groups (HPH [n = 9]: 4.3 ± 0.7; HPN [n = 8]: 3.9 ± 0.7; LPH [n = 8]: 0.8 ± 0.3; LPN [n = 8]: 0.5 ± 0.3; P < 0.05 for the comparison between the HP and LP groups). It should be noted that hyaline membrane formation and thickening of the basal membrane were not observed in any of the groups.

Back to Top | Article Outline

TNF-α and MDA Levels

The HPH and HPN groups showed numerically higher BALF levels of TNF-α, but the difference was not statistically significant (HPH [n = 8]: 566.8 ± 515.5 pg/mL; HPN [n = 10]: 609.6 ± 430.2; LPH [n = 8]: 477.8 ± 401.2; LPN [n = 10]: 319.3 ± 262.2; P = 0.57 for the comparison among groups). Similarly, there was no difference among groups in the BALF levels of MDA (HPH [n = 8]: 1.11 ± 1.49 μmol/L; HPN [n = 10]: 1.27 ± 1.57; LPH [n = 8]: 1.51 ± 1.17; LPN [n = 10]: 0.77 ± 0.91; P = 0.28 for the comparison among groups).

Back to Top | Article Outline


The main finding of our study in this model of VILI is that mechanical ventilation at a high inspiratory pressure/high Vt induces lung injury that does not seem to be further exacerbated by the concurrent administration of a high oxygen concentration. A second finding is that, under the specific settings of our experiment, hyperoxia does not lead to over-production of TNF-α or enhanced lipid peroxidation in the BAF.

The use of the isolated perfused lung model has been criticized, but provides an excellent way to control simultaneously important physiological variables that affect the development of lung injury, such as vascular flow,18 pH31 and temperature.32 This manipulation of variables would be unethical or impossible to perform in a real-life scenario. In addition, this experimental model provides the opportunity to continuously monitor the two most widely used markers of lung injury, namely the formation of pulmonary edema and the propagation of altered capillary permeability. Nevertheless, some important limitations of the model must be borne in mind to temper the temptation of directly extrapolating any findings obtained from it to the intact animal. For example, the vascular flow we used in our experiment is nonpulsatile and approximately 50% lower than the normal pulsatile cardiac output of a 3-kg rabbit and may have mitigated pulmonary edema formation. This is probably partially offset by the fact that the interruption of the lymphatic system drainage in the isolated lung accentuates edema formation. In addition, the distribution of regional pressures, ventilation and perfusion is different when compared with closed-chest animals. Finally, the perfusate we used contained few leukocytes, and this may be of relevance since neutrophils are thought to play a significant role in the initiation of lung injury.33

It is not easy to assess the propagation of lung injury in a quantitative way. Each index of lung injury has its own limitations. Therefore, we chose to assess lung injury using a combination of various indices of injury: weight gain, change in capillary permeability, histological lesions, and changes in the BALF levels of TNF-α and MDA. It is noteworthy that, while the higher inspiratory pressure/higher Vt groups gained more weight, had a greater increase in the ultrafiltration coefficient, and suffered more extensive structural damage, no difference was found in terms of the BALF levels of TNF-α and MDA. In particular, the important difference in ΔKf,c between the HPH and HPN groups was unexpected; it may be a chance finding, or it may be attributed to the higher ultrafiltration coefficient of the HPN group at baseline. If anything, the HPH group did not experience an increase in capillary permeability.

The experimental data regarding the role of TNF-α in the development of VILI are conflicting.1,34 Discordant results may be obtained if the model involves isolated-perfused lungs versus intact animals, healthy versus preinjured lungs, etc. Although it is known that exposure to acute hyperoxia induces proinflammatory mechanisms,35,36 there are scarce data regarding the role of TNF-α in VILI among studies assessing the interaction of lung over-distention and inspired oxygen concentration. Specifically, in the study by Bailey et al.,15 the BALF TNF-α level, but not interleukin (IL)-6, was higher in the group exposed to hyperoxic conditions, whereas it did not differ among the other groups. In a second study,16 lung over-distention led to increased topical production of macrophage inflammatory protein (MIP)-2, the rodent equivalent of human IL-8, that did not increase further by the administration of O2 > 95%. Unfortunately, the study protocol did not include measurement of BALF TNF-α. In the study by Sinclair et al.,17 even though the groups exposed to hyperoxic conditions suffered more severe lung injury, no TNF-α was detected in their BALF. This contradicts our study in which all groups had increased, but not statistically different, BALF TNF-α levels, but it may be partly explained by the different experimental model, Vt, and methods of cytokine measurement. It is also possible that the preparation/surgical phase of our experiments was injurious and “primed” the lungs,37 even though at least the ischemic time was comparable to those reported in similar studies in the field.22 Moreover, we believe that the main explanations for the lack of BALF TNF-α difference across the four groups were the short (1-h) duration of the experiment and the fact that even the lower inspiratory pressure/lower Vt groups were exposed to injurious ventilation, as they were actually receiving a Vt of approximately 20 mL/kg. Finally, since the higher inspiratory pressure/higher Vt groups showed higher, but without statistical significance, BALF levels of TNF-α, we cannot exclude the possibility that an experiment of longer duration might permit a difference in BALF TNF-α levels to become evident.

Experimental evidence suggests that the production of ROS by lungs exposed to hyperoxic conditions may damage both the alveolar epithelium and endothelium, and that this damage may be ameliorated with the use of antioxidants.30,38 MDA is a by-product of the oxidation of unsaturated fatty acids, and is commonly used as an index of oxidative stress,17,27,29 despite its limitations.39 The lack of an observed association in our study between hyperoxic exposure and MDA production may be explained by the short-term duration of our experimental protocol, which did not allow hyperoxia to become functional. This is in line with the study by Sinclair et al.17 in which hyperoxia was shown to exacerbate the injury imposed by large Vt ventilation without an increase of the MDA levels. Of note, Sinclair et al. measured MDA in lung homogenates, whereas we measured it only in BALF. It has been proposed that “lung injury is apparently caused by lipid peroxidation in plasma rather than by high oxygen pressure in the alveoli.”40 Our study protocol dictated the presence of high O2 concentration on both sides of the alveolar-capillary barrier and still did not show any evidence of lipid peroxidation. Thus, we think that the lack of difference in the BALF MDA levels among groups in our study was valid. Finally, it should be emphasized that according to published data, cyclic mechanical stretch of the pulmonary epithelial cells increases the production of ROS and may contribute to the onset of VILI.41

Our study findings should be placed in the context of other studies investigating the effect of O2 concentration on VILI. In the first study,15 isolated mouse lungs preexposed to hyperoxia and high-stretch ventilation had significantly lower compliance, altered pulmonary surfactant, and higher BALF TNF-α levels. Limitations of that study were the assessment of the effect of the sequential application of hyperoxia and high-stretch ventilation, not of their simultaneous combination, the application of zero PEEP, which increased the possibility for cyclic alveolar opening/closure and the lack of perfusion during the experiments. In the study by Quinn et al.,16 rats were randomized to receive in vivo a Vt of 7 or 20 mL/kg for 2 h in 100% O2 or room air. In rats ventilated with high Vt, the formation of pulmonary edema was significantly increased by hyperoxia, but the effect of oxygen was additive, not synergistic, and it did not increase MIP-2 production or BALF neutrophil counts. The authors suggested that mechanisms independent of the production of MIP-2 “may be involved in hyperoxia-augmented neutrophil alveolar content in VILI.” Limitations of the study were the lack of hemodynamic monitoring during the experiments and the application of zero PEEP. In the most recently published study,17 intact rabbits were ventilated with Vt 25 mL/kg at zero PEEP with either hyperoxia (Fio2 = 50%) or normoxia (Fio2 = 21%). Hyperoxia increased histologic lung injury scores, BALF neutrophils, and growth-related oncogene-a and monocyte chemotactic protein-1 concentrations, whereas BALF IL-8 levels were not different and TNF-α was not detected in any group. In a second set of experiments, a separate group of animals was randomized to ventilation with either hyperoxia or normoxia and normal or high Vt. Even though high Vt ventilation and hyperoxia interacted synergistically to increase permeability to dextran and the BALF cytokines and neutrophils were elevated in both high Vt groups, hyperoxia did not further increase the values of these variables over normoxia, and no difference in lipid peroxidation was seen between groups. Unexpectedly, the combination of high Vt ventilation and hyperoxia resulted in lower BALF IL-8 and monocyte chemotactic protein-1 concentrations in the hyperoxia/high Vt group compared with the normoxia/high Vt group.

This study has a number of limitations. First, it is potentially under-powered. It should be noted that in the design of this study the existence of minimal baseline experimental information regarding the expected magnitude of change of the monitored lung injury variables precluded the performance of formal power analysis. However, in retrospect, the lack of even a trend for worsening of these variables by the institution of hyperoxic ventilation implies that sample size calculation might not be indispensable. Second, our study shows a relatively high variability in the monitored variables of lung injury. Animal studies are prone to the introduction of variability by the inherent variation among animals and the variability induced by the experimenters.42 Every effort was made during the experiments to account for this variability. For example, in order to reduce measurement errors, we made multiple observations of the variable of interest: capillary pressure (necessary for the calculation of ultrafiltration coefficient) was measured in triplicate and TNF-α and MDA levels in quadruplicate and duplicate, respectively. In conclusion, even though other studies in the field have used the same or smaller numbers of animals and have shown similar degrees of variability,17,18,22,23,32 our data and statistical analysis should be cautiously viewed as being of exploratory nature and by no means confirmatory.

In addition, the experimental model we used has many limitations that were previously discussed. Thus, it prevents the direct extrapolation of any results to the intact animal and certainly to patients. It should be borne in mind that our study design focused on a combination of only two levels of administered oxygen concentration and PIP, while vascular flow, temperature, RR and PEEP were held constant. It is possible that the institution of different ventilatory settings might have influenced the results.

Finally, and perhaps more importantly, this study was of a short duration. Even though it should be noted that VILI has been extensively studied in this model and a number of variables, such as vascular flow,18 pH,32 and temperature,33 have been shown to be injurious in the same limited timeframe, it is possible that a longer timeframe might be necessary in order to observe the development of lung injury under hyperoxic conditions. However, the short duration of this experimental work was intentional, as it may provide insight into the role of supplementary oxygen in the initial phase of mechanical ventilation or when mechanical ventilation is instituted for a limited time (for example, in the operating room or in the initial phase of trauma or septic shock resuscitation). Therefore, at least in these settings, our data may suggest that clinicians should mainly focus on the avoidance of well-established injurious factors, such as lung over-distention, and not on the avoidance of high Fio2.

Despite the ample experimental evidence supporting a direct injurious effect of high oxygen concentration, a number of animal and human studies have come to different conclusions. First, it is clear that the harmful pulmonary effects of oxygen vary greatly among different species; the larger the animal, the more resistant it is to pulmonary oxygen toxicity.43 This may be explained by differences in the ability to neutralize ROS, even in the same animal model (intraspecies/interstrain differences).44 In addition, tolerance and resistance to oxygen toxicity have been shown in rats through preexposure to hypoxia, hyperoxia or pretreatment with endotoxin.45 Some researchers believe that pulmonary oxygen toxicity is not mediated by the ill-defined toxic effects of ROS but by the well-documented resorption atelectasis caused in lung regions with low ventilation/perfusion ratios. In the clinical setting, Aboab et al.46 recently showed that “in mechanically ventilated patients with ALI the breathing of pure oxygen leads to derecruitment, which is prevented by high PEEP.” Finally, in a very recently published study, hyperoxia during hyperdynamic porcine septic shock showed that hyperoxia did not affect lung mechanics or pulmonary gas exchange and did not aggravate oxidative stress.47

In conclusion, mechanical ventilation with large Vt and not high inspired oxygen concentration was the major determinant of VILI in this short-term, ex vivo experimental model of lung injury. Since some bench and bedside studies have cast doubt on the long-held belief of pulmonary oxygen toxicity, we suggest that further experimental studies with more clinically relevant designs are warranted in order to better assess the interaction between mechanical ventilation and hyperoxia and the time period for which the administration of a high oxygen concentration might be safe.

Back to Top | Article Outline


The authors thank Zoi Kollia, technician at the Department of Experimental Surgery of Evangelismos Hospital, for the time and technical assistance she provided, and Christina Sotiropoulou for statistical review of the study data.

Back to Top | Article Outline


1. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from experimental studies. Am J Respir Crit Care Med 1998;157:294–323
2. Gattinoni L, Carlesso E, Cadringher P, Valenza F, Vagginelli F, Chiumello D. Physical and biological triggers of ventilator-induced lung injury and its prevention. Eur Respir J Suppl 2003;47:15s–25s
3. Ricard JD, Dreyfuss D, Saumon G. Ventilator-induced lung injury. Eur Respir J Suppl 2003;42:2s–9s
4. Lionetti V, Recchia FA, Ranieri VM. Overview of ventilator-induced lung injury mechanisms. Curr Opin Crit Care 2005;11:82–6
5. Tremblay LN, Slutsky AS. Ventilator-induced lung injury: from the bench to the bedside. Intensive Care Med 2006;32:24–33
6. Pinhu L, Whitehead T, Evans T, Griffiths M. Ventilator-associated lung injury. Lancet 2003;361:332–40
7. Amato MB, Barbas CS, Medeiros DM, Magaldi RB, Schettino GP, Lorenzi-Filho G, Kairalla RA, Deheinzelin D, Munoz C, Oliveira R, Takagaki TY, Carvalho CR. Effect of a protective-ventilation strategy on mortality in the acute respiratory distress syndrome. N Engl J Med 1998;338:347–54
8. The Acute Respiratory Distress Syndrome Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000;342:1301–8
9. Nash G, Blennerhassett J, Pontoppidan H. Pulmonary lesions associated with oxygen therapy and artificial ventilation. N Engl J Med 1967;276:368–74
10. Jackson RM. Pulmonary oxygen toxicity. Chest 1985;88:900–5
11. Yam J, Frank L, Roberts RJ. Oxygen toxicity: comparison of lung biochemical responses in neonatal and adult rats. Pediatr Res 1978;12:115–9
12. Fox RB, Hoidal JR, Brown DM, Repine JE. Pulmonary inflammation due to oxygen toxicity: involvement of chemotactic factors and polymorphonuclear leukocytes. Am Rev Respir Dis 1981;123:521–3
13. Kelly FJ, Lubec G. Hyperoxic injury of immature guinea pig lung is mediated via hydroxyl radicals. Pediatr Res 1995; 38:286–91
14. Buccellato LJ, Tso M, Akinci OI, Chandel NS, Budinger GR. Reactive oxygen species are required for hyperoxia-induced Bax activation and cell death in alveolar epithelial cells. J Biol Chem 2004;279:6753–60
15. Bailey TC, Martin EL, Zhao L, Veldhuizen RA. High oxygen concentrations predispose mouse lungs to the deleterious effects of high stretch ventilation. J Appl Physiol 2003;94:975–82
16. Quinn DA, Moufarrej RK, Volokhov A, Hales CA. Interactions of lung stretch, hyperoxia, and MIP-2 production in ventilator-induced lung injury. J Appl Physiol 2002;93:517–25
17. Sinclair SE, Altemeier WA, Matute-Bello G, Chi EY. Augmented lung injury due to interaction between hyperoxia and mechanical ventilation. Crit Care Med 2004;32:2496–501
18. Broccard AF, Hotchkiss JR, Kuwayama N, Olson DA, Jamal S, Wangensteen DO, Marini JJ. Consequences of vascular flow on lung injury induced by mechanical ventilation. Am J Respir Crit Care Med 1998;157:1935–42
19. West JB, Dollery CT, Naimark A. Distribution of blood flow in isolated lung: relation to vascular and alveolar pressures. J Appl Physiol 1964;19:713–24
20. Townsley MI, Korthuis RJ, Rippe B, Parker JC, Taylor AE. Validation of double vascular occlusion method for Pc,i in lung and skeletal muscle. J Appl Physiol 1986;61:127–32
21. Overholser KA, Lomangino NA, Parker RE, Pou NA, Harris TR. Pulmonary vascular resistance distribution and recruitment of microvascular surface area. J Appl Physiol 1994;77:845–55
22. Hotchkiss JR Jr, Blanch L, Murias G, Adams AB, Olson DA, Wangensteen OD, Leo PH, Marini JJ. Effects of decreased respiratory frequency on ventilator-induced lung injury. Am J Respir Crit Care Med 2000;161:463–8
23. Broccard AF, Vannay C, Feihl F, Schaller MD. Impact of low pulmonary vascular pressure on ventilator-induced lung injury. Crit Care Med 2002;30:2183–90
24. López-Aguilar J, Villagrá A, Bernabé F, Murias G, Piacentini E, Real J, Fernández-Segoviano P, Romero PV, Hotchkiss JR, Blanch L. Massive brain injury enhances lung damage in an isolated lung model of ventilator-induced lung injury. Crit Care Med 2005;33:1077–83
25. Murao Y, Hata M, Ohnishi K, Okuchi K, Nakajima Y, Hiasa Y, Junger WG, Hoyt DB, Ohnishi T. Hypertonic saline resuscitation reduces apoptosis and tissue damage of the small intestine in a mouse model of hemorrhagic shock. Shock 2003;20:23–8
26. Kotanidou A, Loutrari H, Papadomichelakis E, Glynos C, Magkou C, Armaganidis A, Papapetropoulos A, Roussos C, Orfanos SE. Inhaled activated protein C attenuates lung injury induced by aerosolized endotoxin in mice. Vascul Pharmacol 2006;45:134–40
27. Giamarellos-Bourboulis EJ, Adamis T, Laoutaris G, Sabracos L, Koussoulas V, Mouktaroudi M, Perrea D, Karayannacos PE, Giamarellou H. Immunomodulatory clarithromycin treatment of experimental sepsis and acute pyelonephritis caused by multidrug resistant Pseudomonas aeruginosa. Antimicrob Agents Chemother 2004;48:93–9
28. Englelhard D, Pomeranz S, Gallily G, Strauss N, Tuomanen E. Serotype-related differences in inflammatory response to Streptococcus pneumoniae in experimental meningitis. J Infect Dis 1997;175:979–82
29. Agarwal R, Chase SD. Rapid, fluorometric-liquid chromatographic determination of malondialdehyde in biological samples. J Chromatogr 2002;775:121–6
30. Jacobson JM, Michael JR, Jafri MH Jr, Gurtner GH. Antioxidants and antioxidant enzymes protect against pulmonary oxygen toxicity in the rabbit. J Appl Physiol 1990;68:1252–9
31. Sinclair SE, Kregenow DA, Lamm WJ, Starr IR, Chi EY, Hlastala MP. Hypercapnic acidosis is protective in an in vivo model of ventilator-induced lung injury. Am J Respir Crit Care Med 2002;166:403–8
32. Suzuki S, Hotchkiss JR, Takahashi T, Olson D, Adams AB, Marini JJ. Effect of core body temperature on ventilator-induced lung injury. Crit Care Med 2004;32:144–9
33. Kawano T, Mori S, Cybulsky M, Burger R, Ballin A, Cutz E, Bryan AC. Effect of granulocyte depletion in a ventilated surfactant-depleted lung. J Appl Physiol 1987;62:27–33
34. Ricard JD, Dreyfuss D, Saumon G. Production of inflammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001;163:1176–80
35. Horinouchi H, Wang CC, Shepherd KE, Jones R. TNF-alpha gene and protein expression in alveolar macrophages in acute and chronic hyperoxia-induced lung injury. Am J Physiol Lung Cell Mol Physiol 1996;14:548–55
36. Suzuki Y, Nishio K, Takeshita K, Takeuchi O, Watanabe K, Sato N, Naoki K, Kudo H, Aoki T, Yamaguchi K. Effect of steroid on hyperoxia-induced ICAM-1 expression in pulmonary endothelial cells. Am J Physiol Lung Cell Mol Physiol 2000;278: L245–L252
37. Serrick C, Adoumie R, Giaid A, Shennib H. The early release of interleukin-2, tumor necrosis factor-alpha and interferon-gamma after ischemia reperfusion injury in the lung allograft. Transplantation 1994;58:1158–62
38. Nagata K, Iwasaki Y, Yamada T, Yuba T, Kono K, Hosogi S, Ohsugi S, Kuwahara H, Marunaka Y. Overexpression of manganese superoxide dismutase by N-acetylcysteine in hyperoxic lung injury. Respir Med 2007;101:800–7
39. Janero DR. Malondialdehyde and thiobarbituric acid-reactivity as diagnostic indices of lipid peroxidation and peroxidative tissue injury. Free Radic Biol Med 1990;9:515–40
40. Jehle R, Schlame M, Büttner C, Frey B, Sinha P, Rüstow B. Platelet-activating factor (PAF)-acetylhydrolase and PAF-like compounds in the lung: effects of hyperoxia. Biochim Biophys Acta 2001;532:60–6
41. Chapman KE, Sinclair SE, Zhuang D, Hassid A, Desai LP, Waters CM. Cyclic mechanical strain increases reactive oxygen species production in pulmonary epithelial cells. Am J Physiol Lung Cell Mol Physiol 2005;289:L834–L841
42. Howard BR. Control of variability. ILAR J 2002;43:194–201
43. Lumb AB Hyperoxia and oxygen toxicity. In: Lumb AB, ed. Nunn's Applied Respiratory Physiology. Edinburgh: Butterworth-Heinemann, 2000:500–7
44. Hudak BB, Zhang LY, Kleeberger SR. Inter-strain variation in susceptibility to hyperoxic injury of murine airways. Pharmacogenetics 1993;3:135–43
45. Capellier G, Beuret P, Clement G, Depardieu F, Ract C, Regnard J, Robert D, Barale F. Oxygen tolerance in patients with acute respiratory failure. Intensive Care Med 1998;24:422–8
46. Aboab J, Jonson B, Kouatchet A, Taille S, Niklason L, Brochard L. Effect of inspired oxygen fraction on alveolar derecruitment in acute respiratory distress syndrome. Intensive Care Med 2006;32:1979–86
47. Barth E, Bassi G, Maybauer DM, Simon F, Gröger M, Oter S, Speit G, Nguyen CD, Hasel C, Möller P, Wachter U, Vogt JA, Matejovic M, Radermacher P, Calzia E. Effects of ventilation with 100% oxygen during early hyperdynamic porcine fecal peritonitis. Crit Care Med 2008;36:495–503
© 2009 International Anesthesia Research Society