Share this article on:

Attenuation of Acute Lung Injury with Propofol in Endotoxemia

Takao, Yumiko MD; Mikawa, Katsuya MD; Nishina, Kahoru MD; Obara, Hidefumi MD

doi: 10.1213/01.ANE.0000144775.19385.8C
Critical Care and Trauma: Research Report

Endotoxin causes acute lung injury (ALI) through many mediators of inflammatory and immune responses. Propofol is an antiinflammatory and immunosuppressive drug. We conducted this study to evaluate whether propofol attenuates ALI associated with endotoxemia. Thirty-two anesthetized rabbits were randomly divided into four groups (n = 8 each). ALI was induced by IV endotoxin 5 mg/kg over 30 min in 3 groups. In 2 of the ALI groups, IV administration of propofol (2 or 5 mg/kg as a bolus followed by continuous infusion at 4 or 15 mg · kg−1 · h−1) was started 15 min before endotoxin. The other ALI group received soybean-oil emulsion. The nonlung injury control group received infusion of both vehicles. The lungs were mechanically ventilated with 40% oxygen for 6 h after endotoxin. Hemodynamics did not differ among groups. The large dose of propofol attenuated lung leukosequestration, pulmonary edema (as assessed by lung wet/dry weight ratio), and pulmonary hyperpermeability (as assessed by albumin levels in bronchoalveolar lavage fluid) and resulted in better oxygenation, lung mechanics, and histological change. The small dose of propofol failed to do so. Our findings suggest that a large dose of propofol successfully mitigates physiological, biochemical, and histological deterioration in ALI in endotoxemia.

IMPLICATIONS: Propofol successfully attenuated physiological, biochemical, and pathologic changes in acute lung injury associated with endotoxemia in rabbits. However, the effect of propofol on fibroproliferative lung damage and on outcome in humans remains to be elucidated.

Department of Anesthesia & Perioperative Medicine, Faculty of Medical Sciences, Kobe University Graduate School of Medicine, Kobe, Japan

Accepted for publication August 27, 2004.

Address correspondence and reprint requests to Katsuya Mikawa, MD, Department of Anesthesia & Perioperative Medicine, Faculty of Medical Sciences, Kobe University Graduate School of Medicine, Kusunoki-cho 7, Chuo-ku, Kobe 650-0017, Japan. Address e-mail to katzmikawa@yahoo.co.jp.

The IV infusion of endotoxin causes acute lung injury (ALI) (1,2). Numerous humoral and cellular inflammatory/immune mediators are involved in the initiation and perpetuation of endotoxin-induced ALI (3). In particular, neutrophils are key players in the pathogenesis of ALI, releasing lipid and enzyme mediators and oxygen radicals (3–5). Neutrophils are accumulated into the lung and activated in response to several chemotaxins (e.g., interleukin [IL]-8, thromboxane B2 [TxB2], and leukotriene B4 [LTB4]) in endotoxemia (4,5). Platelet activation has the potential for causing tissue injury by the release of lysosomal enzymes and TxB2 (3,6). Propofol, often used to sedate critically ill intensive care unit (ICU) patients who receive respiratory management, possesses multiple antiinflammatory and immunomodulatory effects: it enhances tissue antioxidative capacity (7), inhibits neutrophil function (e.g., superoxide production) (8), and reduces IL-8 release from polymorphonuclear leukocytes (9). Propofol reduces platelet aggregability and decreases the serum level of TxB2 (10,11). In a previous study using rats, propofol attenuated inflammatory cytokine (IL-6, IL-10, and tumor necrosis factor-α) production, hemodynamic derangement, metabolic acidosis, and neutrophil activation in response to endotoxemia (12). Thus, propofol may attenuate physiological, biochemical, and morphological lung damage in endotoxemia through these beneficial effects. We conducted this study to test this hypothesis.

Back to Top | Article Outline

Methods

This study was approved by the animal care review board of Kobe University Graduate School of Medicine. We used 32 male Japanese White rabbits (body weight, 2.0–2.5 kg), which were anesthetized with 60 mg/kg of ketamine given IM and intubated with an endotracheal tube through a tracheotomy. A catheter was inserted into a left ear vein for infusion of fluids. Anesthesia was maintained with infusion of ketamine at 30 mg · kg−1 · h−1 and IM xylazine 8 mg · kg−1 · h−1. An arterial catheter was placed through a cutdown in the right femoral artery to monitor mean arterial blood pressure (MAP) and to take samples for blood gas analysis and peripheral count of leukocytes and platelets. The lungs were ventilated with an infant pressure-controlled and time-cycled ventilator (IV100B; Sechrist, Anaheim, CA) with an inspired oxygen concentration of 40%. The tidal volume was initially set to 10 mL/kg (peak inspiratory pressure, 11–13 cm H2O), and 2 cm H2O of peak expiratory pressure was added. The respiratory rate was controlled to produce an initial Paco2 of 31–37 mm Hg with an inspiratory/expiratory time ratio of 1:2. The constant flow rate of the ventilator was set at 10 L/min. The animals were placed supine on a heating pad under a radiant heat lamp to keep body temperature between 38°C and 40°C at the esophagus. Heart rate was continuously monitored with an electrocardiograph. Central venous pressure was also monitored via a catheter inserted through the right jugular vein. In all groups, lactated Ringer’s solution was administered IV at 10 mL · kg−1 · h−1. The blood withdrawn (1 mL each for blood sampling) was replaced with bolus administration of lactated Ringer’s solution (2 mL).

The animals were randomly divided into four groups (n = 8 per group) by using a sealed-envelope method. ALI was induced by IV infusion of lipopolysaccharide (LPS) from Escherichia coli (055:B5 from the same lot; Sigma, St. Louis, MO) 5 mg/kg over 30 min in three groups. LPS was suspended in saline. In the two ALI groups, IV administration of propofol (Diprivan®; Astra-Zeneca KK, Osaka, Japan) at small or large dose was started 15 min before endotoxin: the small-dose group (group LPS+PROsmall) received 2 mg/kg as a bolus followed by continuous infusion at 4 mg · kg−1 · h−1, and the large-dose group (group LPS+PROlarge) received 5 mg/kg as a bolus followed by continuous infusion at 15 mg · kg−1 · h−1. The other ALI group (group LPS+INT) received an equivalent volume of soybean-oil emulsion (Intralipid 10%®; Otsuka, Tokyo, Japan). The nonlung-injury control group (group saline+INT) received infusion of saline (instead of LPS) and Intralipid® (instead of propofol) because commercial propofol is prepared in Intralipid® solution.

Hemodynamics, lung compliance and resistance, arterial gas analysis values, and peripheral leukocyte and platelet counts were recorded hourly. Arterial blood gases (Pao2, Paco2, and pH) were analyzed with a blood gas analyzer (ABL2; Radiometer, Copenhagen, Denmark), and the number of leukocytes was measured with an automated blood cell counter (Sysmex K-1000; Sysmex, Kobe, Japan). In groups LPS+PROsmall and LPS+PROlarge, blood was taken to determine plasma propofol concentrations immediately before death (6 h after the start of endotoxin), followed by centrifugation at 3000 rpm for 10 min at 4°C. Plasma was stored at −80°C until measurement of propofol levels with high-performance liquid chromatography by using the technique of Plummer (13).

Lung mechanics were measured by the single occlusion and passive expiratory flow-volume technique as described by LeSouef et al. (14), with computer-assisted calculations. A Fleisch 00 pneumotachograph was attached to the endotracheal tube, and occlusions were performed manually at maximal inspiration. The airflow was measured with the pneumotachograph and a differential pressure transducer (Model MP045; Validyne Engineering, Northbridge, CA). Airway pressure was measured at the proximal end of the pneumotachometer with a semiconductor pressure transducer (Model P-300 501G; Copal Electronics, Tokyo, Japan). The volume was determined for each breath by digital integration of airflow with a respiration monitor (Aivision, Tokyo, Japan). Flow-volume and pressure plots of the occluded breaths were presented on a screen. Expirations with a pressure plateau of at least 1.0 s and a linear flow-volume plot after occlusion were used for automatic calculation of the time constant of the respiratory system. Compliance of the respiratory system was then calculated from the volume and pressure differences at occlusions and resistance from the time constant of the respiratory system and the compliance data. The data were recorded as the mean of two occlusions. The compliance and resistance of the total respiratory system were then calculated with a personal computer (PC9801 VM11; NEC, Tokyo, Japan).

At the end of the experiments (6 h after the start of endotoxin or saline infusion), all rabbits were killed by an overdose injection of thiamylal. The heart and the lung were removed en bloc by an observer (KN) blinded to the nature of the experiments. The left upper lobe of each lung was weighed and then dried to constant weight at 60°C for 24 h in an oven. The ratio of lung wet/dry weight (W/D) weight was calculated to assess tissue edema.

Saline (35 mL) with ethylenediaminetetraacetic acid/2Na at 4°C was slowly infused and withdrawn to obtain bronchoalveolar lavage (BAL) fluid from the whole right lung. Indomethacin was added to the BAL fluid to inhibit further metabolism of arachidonic acid to prostaglandins during analysis. The BAL fluid was analyzed for cell counts and cell differentiation. A cytocentrifuged preparation (Cytospin 2; Shandon Southern Products, Pittsburgh, PA) of the BAL fluid was stained with Diff-Quick® (Harleco, Gibbstown, NJ) for cell differentiation. The cells present in the fluid were counted with a Bürker-Türk hemocytometer. The fluid was centrifuged at 250g at 4°C for 10 min to remove the cells. The cell-free supernatant was divided into several aliquots and stored at −80°C until assay. The following substances and mediators in the BAL fluid were measured: albumin concentrations were determined by nephelometry with the immunoglobulin G fraction of goat anti-rabbit albumin (Cappel, Durham, NC); concentrations of LTB4 were measured by enzyme immunoassay (Amersham-Pharmacia, Buckinghamshire, UK); and concentrations of TxB2 and prostacyclin were quantified by radioimmunoassay kit (NEN, Boston, MA, and Amersham-Pharmacia) as 11-dehydro-TxB2 (11-DTxB2) and 6-keto prostaglandin F, the stable metabolites, respectively. Activated complement factor C3 levels were determined by radioimmunoassay kit (Amersham). Concentrations of IL-8 were determined by using a human enzyme immunoassay kit (Amersham) that cross-reacts to rabbit IL-8; recombinant rabbit IL-8 was used for the standard concentration curve (15).

The left lower lobe was fixed by instillation of 10% formaldehyde solution through the left lower bronchus at 20 cm H2O. The specimens were embedded in paraffin wax, stained with hematoxylin and eosin, and examined under a light microscope. Lung injury was scored by a blinded observer (KN) according to the following four items: alveolar congestion, hemorrhage, infiltration or aggregation of neutrophils in the airspace or vessel wall, and thickness of the alveolar wall/hyaline membrane formation. Each item was graded according to a five-point scale: 0 = minimal (little) damage, 1+ = mild damage, 2+ = moderate damage, 3+ = severe damage, and 4+ = maximal damage. Thus, minimum and maximum possible scores were 0 and 16, respectively. Eight microscopic images were obtained from each tissue sample by using ACT-1 (Nikon, Tokyo, Japan), and the area of alveolar space was morphologically determined with image-analysis software (WinRoof®; Mitani, Tokyo, Japan). Alveolar size was expressed as a ratio of the alveolar/parenchymal area.

The crude overall repeated-measures data (i.e., oxygenation, lung mechanics, peripheral leukocytes and platelets, and hemodynamics) were statistically analyzed with repeated-measures analysis of variance (ANOVA) to compare the effects of the drug intervention (intergroup differences) and the changes in those effects over time, as well as to assess interactions among the effects. For several variables, there were major and significant differences within and between groups. However, interaction between the two factors (time and drug intervention) was observed in all of the variables, so a multiple comparison test could not be performed. Because the noninjury group (saline+INT) was heterogeneous in this study and because our interest lay mainly in intergroup differences, we narrowed the repeated-measures data to 6 sampling points (1, 2, 3, 4, 5, and 6 h) for the 3 groups (LPS+INT, LPS+PROsmall, and LPS+PROlarge) and reanalyzed them with repeated-measures ANOVA. However, there was still an interaction between the two factors. To interpret the data, one-way ANOVA of these data (18 categories = 3 groups × 6 time points) followed by the Tukey-Kramer post hoc test was used. Interaction was interpreted as follows: 1) propofol had the attenuating effect, and 2) there was no correlation between the effect of propofol and time. The data of BAL fluid, lung W/D weight ratio, and aerated alveolar area were analyzed by one-way ANOVA followed by the Tukey-Kramer post hoc test. The ALI score (given as a median) was analyzed with the Kruskal-Wallis rank test followed by the Dunnett test. P < 0.05 was deemed statistically significant. The statistical analyses were performed with a commercial software package (StatView 5.0; SAS Institute Inc., Cary, NC).

Back to Top | Article Outline

Results

Plasma propofol concentrations (mean ± sd) at 6-h postendotoxin points in groups LPS+PROsmall and LPS+PROlarge were 2.3 ± 0.8 μg/mL and 8.2 ± 1.9 μg/mL, respectively. Endotoxin gradually reduced Pao2 and compliance and increased resistance (Fig. 1). A large dose of propofol attenuated the deterioration of oxygenation and lung mechanics, although a small dose of propofol failed to do so.

Figure 1

Figure 1

MAP slightly decreased in group LPS+INT. There was no difference in MAP, heart rate, or central venous pressure among the groups (Table 1). Peripheral circulating leukocytes and platelets were gradually decreased after endotoxin infusion (Table 2). Leukopenia and thrombocytopenia were less severe in rabbits receiving a large dose of propofol (Table 2).

Table 1

Table 1

Table 2

Table 2

A large dose of propofol significantly attenuated an increase in the lung W/D weight ratio in endotoxemia (Fig. 2). Recovery percentages of BAL fluid in the groups ranged between 69% and 77%, indicating no differences among groups. A large dose of propofol mitigated an increase in leukocyte counts (monocytes, macrophages, lymphocytes, and neutrophils), the neutrophil/total leukocytes ratio (percentage neutrophils), and albumin concentrations in the BAL fluid in rabbits receiving endotoxin (Table 3). Endotoxin infusion increased IL-8, activated complement factor C3, LTB4, 11-DTxB2, and 6-keto prostaglandin F levels in BAL fluid (Table 3). A large dose of propofol significantly reduced the increase in 11-DTxB2 concentrations and seemed to decrease levels of other mediators, although not significantly.

Figure 2

Figure 2

Table 3

Table 3

Endotoxemia caused extensive morphological lung damage: edema, hemorrhage, thickness of the alveolar wall, and infiltration of inflammatory cells into alveolar spaces and interstitial spaces (Fig. 3A). These morphological changes were less pronounced with a large dose of propofol (Fig. 3C). The ALI score was reduced by a large dose of propofol (Table 2). The aerated alveolar area was larger in the rabbits receiving a large dose of propofol than in those receiving endotoxin and Intralipid (Table 2).

Figure 3

Figure 3

Back to Top | Article Outline

Discussion

In this study, we have shown that IV infusion of endotoxin caused pulmonary edema (as assessed by increased W/D weight ratio) and morphological destruction of alveolar structure, leading to deterioration of gas exchange and lung compliance compared with the noninjury control group, although lung inflammation was less severe in our endotoxemia model (5% neutrophils in BAL fluid). Furthermore, a large dose of propofol attenuated pulmonary edema, in part by inhibiting the hyperpermeability of the pulmonary endothelium (as assessed by increased BAL fluid albumin levels). Hence, the drug successfully mitigated lung inflammation (neutrophil sequestration), thus securing improvement of oxygenation and lung mechanics, and histological changes. The attenuating effect of propofol on physiological alterations is a new finding and may have significant implications for clinical use. The effectiveness is not explained only by the cardiovascular action of propofol, because major hemodynamic changes were not observed in this study.

Inhibition of neutrophil accumulation in the lung is a key strategy to control ALI. A large dose of propofol demonstrated a significant inhibition of release of TxB2, one of the representative chemotaxins, although the drug failed to inhibit other chemotactic mediators (e.g., IL-8). The reduction of TxB2 release by propofol is probably responsible for the mitigation of neutrophil recruitment in the lung. However, the mechanism underlying the efficacy of propofol in attenuating ALI is not fully understood.

Our experimental model has limitations. Lung inflammatory changes associated with endotoxin infusion are thought to be less severe than those induced by intratracheal endotoxin (approximately 50% neutrophils) (16). However, assessment of the degree of ALI induced by intratracheal endotoxin may be more complicated because the agent given via this route is often distributed heterogeneously. Unlike our ALI model, intratracheal endotoxin mimics pneumonia-associated ALI. Our ALI model in endotoxemia does not correlate with acute lung damage in human sepsis, especially septic shock, in which severe hemodynamic derangement often occurs. In our previous preliminary experiments, many rabbits receiving an IV bolus of endotoxin suddenly died of shock without lung inflammation. We believe that endotoxin injection should be used as a model of septic shock, but not of ALI. However, in this study, we may have just missed hemodynamic changes associated with endotoxemia, because the time pattern of hemodynamic variables may vary. Furthermore, although we used endotoxin extracted from E. coli, the pathogenic bacteria are often not known in human sepsis. Mixed infections involving both Gram-negative and Gram-positive bacteria are also common in patients with sepsis (17). We used quite a large dose of propofol in this study because the dose requirement of IV anesthetics seems to be generally larger in animals than in humans (18), and Taniguchi et al. (12) and Gao et al. (19) used a large dose of the drug (10 mg · kg−1 · h−1) in rats. However, the dose used in our experiment is higher than the currently recommended dose for humans. Thus, we are unable to easily extrapolate our findings to the clinical setting.

In addition to inflammatory cells, platelets play a crucial role in the progression of endotoxin-induced ALI by liberating various compounds found in their granules, including TxB2 (3,6). In this process, platelets are activated, aggregated, and trapped as microemboli in the pulmonary capillaries, resulting in peripheral thrombocytopenia. Arachidonic acid metabolites (e.g., TxB2) contribute to the development and progression of lung damage by causing pulmonary hypertension (related to pulmonary vasoconstriction) and vascular hyperpermeability (20). In this study, propofol successfully mitigated peripheral thrombocytopenia and a significant increase in TxB2 levels in BAL fluid, thus suggesting that the drug had an inhibitory effect on the aggregation and activation of platelets. Propofol has been demonstrated to directly reduce pulmonary vascular resistance (21). Although pulmonary arterial pressure was not measured, propofol may have attenuated pulmonary hypertension through these mechanisms.

Peroxynitrite, a powerful oxidant formed by the reaction of superoxide with nitric oxide, is responsible for several types of tissue injury by membrane peroxidation and protein degeneration (22). Gao et al. demonstrated that IV endotoxin produces a large amount of nitric oxide through upregulation of expression for inducible nitric oxide synthase messenger RNA in the lung (19). They also found lung generation of peroxynitrite, which is detected as nitrotyrosine residues (19). Early administration of propofol decreases peroxynitrite generation in the lung tissue and downregulates expression for inducible nitric oxide synthase messenger RNA, and this leads to attenuation of endotoxin-induced lung edema (19). In this study, propofol may have had favorable effects on lung damage in endotoxemia through elimination of peroxynitrite, although nitrotyrosine immunostain was not assessed.

Endotoxemia often causes severe reduction of myocardial contractility and peripheral vascular resistance (23). In this study, infusion of endotoxin over 30 minutes minimized reductions of MAP and resulted in no significant differences among groups, as was previously reported (24,25). We could exclude the influences of left heart failure and severe vasodilation (hemodynamic derangement) in endotoxemia on arterial oxygenation with our endotoxin infusion model. In this study, a large dose of propofol did not enhance hypotension associated with endotoxemia. Hypotensive effects of propofol may have been counterbalanced by a reduction of detrimental mediators, inducing cardiovascular dysfunction observed in endotoxemia.

Propofol is often used to sedate ICU patients receiving mechanical ventilation. It may be possible to give propofol in advance to critically ill patients who are predisposed to acute respiratory distress syndrome (e.g., shock and major trauma). In this context, we believe that this study, which explored the effect of propofol pretreatment on ALI in endotoxemia, has clinical relevance and implications. A potential negative aspect of propofol-induced impairment of neutrophil functions, which has been implicated as an important microbicidal mechanism, includes a possible increase in susceptibility to infection. Our data from this study are unable to provide an obvious solution to this problem, because we did not assess the effect of propofol on the bactericidal system in this setting. Whether propofol would enhance susceptibility to infection deserves further study.

In conclusion, we have shown that a large dose of propofol attenuated physiological, biochemical, and histological changes of ALI in endotoxemia. However, further studies are required to assess the effects of posttreatment with propofol on the late (proliferative and fibrotic) phase and on the acute (infiltrative) aspect of lung damage.

Back to Top | Article Outline

References

1. Ghosh S, Latimer RD, Gray BM, et al. Endotoxin-induced organ injury. Crit Care Med 1993;21:S19–24.
2. Brigham KL, Meyrick B. Endotoxin and lung injury. Am Rev Respir Dis 1986;133:913–27.
3. Schlag G, Redl H. Mediators of injury and inflammation. World J Surg 1996;20:406–10.
4. Aldridge AJ. Role of the neutrophil in septic shock and the adult respiratory distress syndrome. Eur J Surg 2002;168:204–14.
5. Abraham E. Neutrophils and acute lung injury. Crit Care Med 2003;31:S195–9.
6. Spragg RG. Platelets and acute lung injury. In: Artigas A, Lemaire F, Suter PM, Zapol WM, eds. Adult respiratory distress syndrome. New York: Churchill Livingstone, 1992:89–94.
7. Runzer TD, Ansley DM, Godin DV, Chambers GK. Tissue antioxidant capacity during anesthesia: propofol enhances in vivo red cell and tissue antioxidant capacity in a rat model. Anesth Analg 2002;94:89–93.
8. Mikawa K, Akamatsu H, Nishina K, et al. Propofol inhibits human neutrophil functions. Anesth Analg 1998;87:695–700.
9. Galley HF, Dubbels AM, Webster NR. The effect of midazolam and propofol on interleukin-8 from human polymorphonuclear leukocytes. Anesth Analg 1998;86:1289–93.
10. De La Cruz JP, Paez MV, Carmona JA, De La Cuesta FS. Antiplatelet effect of the anaesthetic drug propofol: influence of red blood cells and leucocytes. Br J Pharmacol 1999;128:1538–44.
11. Aoki H, Mizobe T, Nozuchi S, Hiramatsu N. In vivo and in vitro studies of the inhibitory effect of propofol on human platelet aggregation. Anesthesiology 1998;88:362–70.
12. Taniguchi T, Yamamoto K, Ohmoto N, et al. Effects of propofol on hemodynamic and inflammatory responses to endotoxemia in rats. Crit Care Med 2000;28:1101–6.
13. Plummer GF. Improved method for the determination of propofol in blood by high-performance liquid chromatography with fluorescence detection. J Chromatogr 1987;421:171–6.
14. LeSouef PN, Engl SJ, Bryan AC. Total resistance of the respiratory system in preterm infants with and without an endotracheal tube. J Pediatr 1984;104:108–11.
15. Matsukawa A, Yoshimura T, Miyamoto K, et al. Analysis of the inflammatory cytokine network among TNF, IL-1 beta, IL-1 receptor antagonist, and IL-8 in LPS-induced rabbit arthritis. Lab Invest 1997;76:629–38.
16. O’Grady NP, Preas HL, Pugin J, et al. Local inflammatory responses following bronchial endotoxin instillation in humans. Am J Respir Crit Care Med 2001;163:1591–8.
17. Esmon CT. Why do animal models (sometimes) fail to mimic human sepsis? Crit Care Med 2004;32:S219–22.
18. Flecknell PA. Anaesthesia of animals for biomedical research. Br J Anaesth 1993;71:885–94.
19. Gao J, Zeng BX, Zhou LJ, Yuan SY. Protective effects of early treatment with propofol on endotoxin-induced acute lung injury in rats. Br J Anaesth 2004;92:277–9.
20. Samuelsson B, Dahlen SE, Lindgren JA, et al. Leukotrienes and lipoxins: structures, biosynthesis, and biological effects. Science 1987;237:1171–6.
21. Rich GF, Roos CM, Anderson SM, et al. Direct effects of intravenous anesthetics on pulmonary vascular resistance in the isolated rat lung. Anesth Analg 1994;78:961–6.
22. Kam PCA, Govender G. Nitric oxide: basic science and clinical applications. Anaesthesia 1994;49:515–21.
23. Martich GD, Boujoukos AJ, Suffredini AF. Response of man to endotoxin. Immunobiology 1993;187:403–16.
24. Nishina K, Mikawa K, Takao Y, et al. ONO-5046, an elastase inhibitor, attenuates endotoxin-induced acute lung injury in rabbits. Anesth Analg 1997;84:1097–103.
25. Mikawa K, Nishina K, Tamada M, et al. Aminoguanidine attenuates endotoxin-induced acute lung injury in rabbits. Crit Care Med 1998;26:905–11.
© 2005 International Anesthesia Research Society