INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It has been shown repeatedly that ventilating experimental animals with a large tidal volume (VT) may damage their lungs and cause ventilator-induced lung injury (VILI) (1). The clinical relevance of this concept was recently clearly illustrated by a trial conducted by the U.S. National Institutes of Health (NIH) (5) that showed a 22% reduction in the mortality of patients with acute respiratory distress syndrome (ARDS) when the mechanical stress on their lungs was lessened by reducing VT during mechanical ventilation. Only 3 yr after the description of ARDS, the visionary paper by Mead and coworkers (6) showed that excessive tissue stress may occur during the mechanical ventilation of ARDS patients. They calculated that there might be considerable tissue stretch around atelectatic areas, even at moderately high airway pressure. This led Mead and coworkers to speculate that "mechanical ventilators, by applying high transpulmonary pressure to the nonuniformly expanded lungs of some patients who would otherwise die of respiratory insufficiency, may cause the hemorrhage and hyaline membranes found in such patients' lungs at death" (6).

Permeability pulmonary edema and hyaline membranes are two characteristic features of VILI (2). Recently, several studies suggested that the release of proinflammatory cytokines by lung cells in response to tissue stretch contributes to VILI (7, 8). Whereas lung cells such as alveolar macrophages and epithelial cells do not release inflammatory cytokines such as tumor necrosis factor (TNF-) in response to mechanical stress (9, 10), the situation may be different when considering the organ as a whole. Tremblay and coworkers (11) found that isolated, unperfused rat lungs ventilated for 2 h with strategies aimed at increasing lung mechanical stress released significant amounts of both proinflammatory (TNF-, interleukin [IL]- 1, IL-6) and antiinflammatory (IL-10) cytokines, as well as the chemokine macrophage inflammatory protein (MIP)-2 into airspaces. TNF- and IL-6 have also been detected in the perfusate of isolated mouse lungs ventilated with a large VT (12). These findings led to the speculation that mechanical ventilation "may play a pivotal role in the initiation and/or propagation of a systemic inflammatory response leading to multiple system organ failure in certain patients" (13). The in vivo demonstration that increased mechanical stress may by itself (i.e., in intact lungs) produce a pulmonary inflammatory response raises this possibility as an unresolved issue.

The studies suggesting that cytokines participate in the development of VILI were done with lungs already damaged by surfactant depletion (8) or exposed to hyperoxia (7). We therefore initiated the present study to corroborate in vivo the observations made on lungs ventilated ex vivo. We ventilated rats with a large VT (an injurious ventilatory strategy) but were unable to detect any TNF- and only trivial amounts of IL-1 either in airspaces, through the use of bronchoalveolar lavage (BAL), or in the systemic circulation, despite the production of a severe VILI. We then reexamined the ex vivo model of Tremblay and coworkers (11) and were unable to reproduce their observations.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental Protocol

Specific pathogen-free male Sprague-Dawley rats (weighing 280 to 320 g; Charles River Laboratories, Saint Aubin-lès-Elbeuf, France) were used in the study. All experiments were conducted according to regulations of the French Ministry of Agriculture.

In Vivo Experiments

Rats were anesthetized by intraperitoneal injection of thiopental (37 mg/kg). A tracheostomy was performed, and each animal was injected with succinylcholine (5 mg/kg) via the dorsal penile vein, after which the animal was ventilated with a rodent volume ventilator (Harvard Apparatus, Ealing, Les Ulis, France). Thiopental was chosen as the anesthetic agent because it is slowly metabolized in rodents. The dose that we used ensures a profound anesthesia for at least 4 h. Two ventilation modalities were used, for 2 h each, as follows: (1) control, with low VT ventilation (7 ml/kg VT and 3 cm H₂O positive end-expiratory pressure [PEEP], 40 breaths/min) (LV group) and (2) an injurious strategy, using a high VT and no PEEP (42 ml/kg VT, zero end-expiratory volume [ZEEP], 40 breaths/min) (HV group). The rats were killed by an intravenous injection of thiopental at the end of the mechanical ventilation period, the thorax was opened, and blood was sampled by cardiac puncture. Simultaneously, three BAL procedures were performed, each with 2 ml of normal saline. The retrieved fluid and the blood were centrifuged (2,000 × g, at 4° C for 10 min), and the supernatant and plasma were stored at 80° C for further processing.

Ex Vivo Lung Model

Isolated rat lungs were ventilated ex vivo as described by Tremblay and coworkers (11). Rats were anesthetized by intraperitoneal injection of acepromazine (2.5 mg/kg) and ketamine (75 mg/kg) and were then injected with 500 µg Salmonella typhosa lipopolysaccharide (LPS) (Sigma Chemical Co., St. Louis, MO) or an equivalent volume of normal saline via the dorsal penile vein. The rats were allowed to breathe normally for 50 min and were then given an intravenous injection of heparin together with a lethal dose of thiopental. A tracheostomy was performed after the rats had stopped breathing, and the thorax was opened by midline sternotomy. Blood was withdrawn by cardiac puncture and was centrifuged, and the resulting plasma was frozen at 80° C for subsequent determination of baseline levels of circulating cytokines. The heart and lungs were removed en bloc and suspended by means of the tracheal catheter in a thermostated, humidified chamber at 37° C. The lungs were slowly inflated with 10 ml of air before being subjected to beginning mechanical ventilation.

The lungs were randomized to one of three ventilatory strategies. They were: (1) maintained in a statically inflated state with 7 cm H₂O continuous positive pressure (Group VT0); (2) mechanically ventilated with room air with a low VT of 7 ml/kg body weight (Group VT7); or (3) mechanically ventilated with a high VT of 42 ml/kg body weight (Group VT42). Mechanical ventilation was done with a volume-cycled ventilator (Harvard). The respiratory rate was set at 40 breaths/min. PEEP was set at 3 cm H₂O in the VT7 group, whereas the VT42 group was ventilated with ZEEP. Mechanical ventilation was continued for 2 h.

BAL was performed at the end of the experiment, using three 2-ml aliquots of normal saline. The aliquots were pooled and centrifuged at 2,000 × g for 10 min, and the supernatants were frozen at 80° C.

Cytokine, Chemokine, and Lung Lavage Protein Determination

The following cytokine and chemokine concentrations were measured by enzyme-linked immunosorbent assay (ELISA) of plasma and BALF supernatant samples: TNF- (Genzyme SA; Cergy-Pontoise, France), IL-1 (R&D Systems, Oxon, UK), and MIP-2 (Biosource International, Camarillo, CA). The total protein concentration in BALF was determined with the Bio-Rad protein assay (Bio-Rad Laboratories, Munich, Germany), with bovine serum albumin as the standard. The accuracy of the Genzyme ELISA was verified with recombinant TNF- (PeproTech, Tebu, France).

Statistical Analysis

Results are expressed as mean ± SEM. Analysis of variance was used for between-group comparisons of protein and cytokine concentrations. A value of p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 38 rats were used (305 ± 3.5 [mean ± SEM] g body weight): four in each in vivo experiment and five in each ex vivo experiment.

Effect of Mechanical Ventilation on Protein Concentrations in BALF

In vivo experiments. The rats were cyanotic after 2 h of ventilation with the HV protocol, and pink edema fluid was flowing through the tracheal cannula. The protein concentration in the BALF was markedly higher in rats ventilated with the HV protocol (9.5 ± 0.80 g/L) than in animals ventilated with the LV protocol (0.2 ± 0.04 g/L) (p < 0.001), indicating severe pulmonary edema in the HV group (Figure 1A).

View larger version (12K): [in this window] [in a new window]   Figure 1.   (A) The protein concentration in bronchoalveolar lavage fluid (BALF) was larger in rats ventilated for 2 h with 42 ml/kg VT and zero end-expiratory pressure (ZEEP) (VT42) than in those ventilated with 7 ml/kg VT and 3 cm H2O positive end-expiratory pressure (PEEP) (VT7). ***p < 0.001 for VT42 versus VT7 group. (B) Protein concentration in BALF of isolated, nonperfused rat lungs ventilated for 2 h in a thermostated chamber (37° C) with humidified air. Solid rectangles indicate rats given lipopolysaccharide (LPS) 50 min before the lungs were removed. VT0: lungs were maintained in a statically inflated condition at 7 cm H2O airway pressure; VT7: lungs were ventilated with 7 ml/kg VT and 3 cm H2O PEEP; VT42: lungs were ventilated with 42 ml/kg VT and ZEEP. LPS treatment did not affect the amount of protein in BALF. The protein concentration in BALF was larger in lungs of the VT42 group. ***p < 0.001 compared with the other groups.

Ex vivo experiments. Protein concentrations in the BALF of lungs left statically inflated and of lungs ventilated with 7 ml/ kg VT ex vivo were similarly low. The protein concentration was markedly higher in lungs ventilated with 42 ml/kg VT (p < 0.0001), whether or not the animals had been given LPS before the lungs were harvested (Figure 1B). The large increase in protein concentration reflected the passage of interstitial fluid into alveolar spaces during HV ventilation.

Effect of Mechanical Ventilation on Cytokine and Chemokine Production In Vivo

No TNF- was detected in the BALF of animals subjected to either 42 ml/kg VT or 7 ml/kg VT (Figure 2). The IL-1 concentrations in the BALF were always below 50 pg/ml but were slightly higher in animals ventilated with 42 ml/kg VT than in those ventilated with 7 ml/kg VT (p < 0.05) (Figure 2). Moderately large ( 300 pg/ml) amounts of MIP-2 were found in the BALF irrespective of the magnitude of VT (p = 0.46) (Figure 2).

View larger version (25K): [in this window] [in a new window]   Figure 2.   Tumor necrosis factor (TNF)-, interleukin (IL)-1, and macrophage inflammatory protein (MIP)-2 concentrations in BALF of rats ventilated for 2 h with 7 ml/kg VT and 3 cm H2O PEEP (VT7) and 42 ml/ kg VT and ZEEP (VT42). TNF- was undectectable whatever the ventilation strategy. IL-1 and MIP-2 were found in small amounts. IL-1 was slightly higher (*p < 0.05) in BALF of rats ventilated with the larger VT. (Note that 100 signifies a value below the detection threshold.)

Effect of Mechanical Ventilation on Cytokine and Chemokine Production in Isolated Lungs

Negligible amounts of TNF- (< 55 pg/ml) were found in the BALF of ex vivo lungs whatever the ventilation strategy (Figure 3). Small amounts of IL-1 were retrieved from lungs maintained in static inflation or ventilated with 7 ml/kg VT (Figure 3). The concentration of IL-1 in BALF was slightly higher in the HV group than in the other groups (p < 0.05), but in no case exceeded 80 pg/ml. The lungs of the HV group released significantly more MIP-2 into the airspaces than did those of the other groups (p < 0.05) (Figure 3). Lungs ventilated with 7 ml/kg VT released the same amounts of IL-1 and MIP-2 into the BALF as did intact animals (see Figures 2 and 3). This was also the case for IL-1 released with 42 ml/kg VT ventilation. However, the BALF MIP-2 concentration was significantly higher in ex vivo preparations than in intact animals ventilated with 42 ml/kg VT (p < 0.02).

View larger version (25K): [in this window] [in a new window]   Figure 3.   TNF-, IL-1, and MIP-2 concentration in BALF of isolated, nonperfused rat lungs maintained for 2 h in a statically inflated state at 7 cm H2O airway pressure (VT0), ventilated with 7 ml/kg VT and 3 cm H2O PEEP (VT7), or ventilated with 42 ml/kg VT and ZEEP (VT42). IL-1 and MIP-2 concentrations were slightly higher (*p < 0.05) in BALF of lungs of the VT42 group. There was no difference for TNF-.

After LPS challenge, negligible amounts (< 30 pg/ml) of TNF- and IL-1 were retrieved in the BALF of lungs maintained in the statically inflated state (Figure 4). Higher levels of TNF- and IL-1 were found in the BALF of ventilated lungs (p < 0.001) (Figure 4). The concentrations of TNF- did not differ with the magnitude of VT, whereas the IL-1 concentration was significantly higher in lungs ventilated with the largest VT (p < 0.01). The BALF MIP-2 concentration was high but did not differ between the study groups (Figure 4).

View larger version (32K): [in this window] [in a new window]   Figure 4.   TNF-, IL-1, and MIP-2 concentrations in BALF of isolated, nonperfused lungs of rats given LPS 50 min before lungs were removed. Lungs were either maintained for 2 h in a statically inflated state at 7 cm H2O airway pressure (VT0), or were ventilated with 7 ml/kg VT and 3 cm H2O PEEP (VT7), or with 42 ml/kg VT and ZEEP (VT42). TNF- and IL-1 concentrations in BALF were larger in ventilated lungs than in statically inflated lungs (***p < 0.001). The TNF- concentration did not differ between the VT7 and VT42 groups, whereas that of IL-1 was slightly higher (p < 0.01) in the VT42 group. There was no difference among the three groups in the BALF MIP-2 concentration.

Effect of Mechanical Ventilation on Plasma Cytokine and Chemokine Concentrations

We did not detect TNF-, IL-1, or MIP-2 in the plasma of rats that were not given LPS, even in those subjected to high-volume ventilation (Figures 5A and 5B). Similarly, in ex vivo experiments and in the absence of LPS pretreatment, no TNF-, IL-1, or MIP-2 was detected in the plasma samples taken before the lungs were removed. The plasma concentrations of TNF-, IL-1, and MIP-2 of animals given LPS were high and did not differ between the study groups (Figure 6).

View larger version (12K): [in this window] [in a new window]   Figure 5.   (A) Plasma TNF-, IL-1, and MIP-2 concentrations were below the detection threshold after 2 h of in vivo mechanical ventilation in rats ventilated with a VT 7 ml/kg. (B) Plasma TNF-, IL-1, and MIP-2 concentrations were below the detection threshold after 2 h of in vivo mechanical ventilation in rats ventilated with 42 ml/kg VT. (Note that 100 signifies a value below the detection threshold.)

View larger version (32K): [in this window] [in a new window]   Figure 6.   Plasma TNF-, IL-1, and MIP-2 concentrations were similarly high in animals given LPS 50 min before their lungs were removed for ventilation ex vivo.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Three main conclusions can be drawn from the present study. First, high-volume mechanical ventilation per se does not cause a significant release of proinflammatory cytokines (TNF- and IL-1) into the airspaces of isolated rat lungs. These results, as discussed later, differ considerably from those of a previous study using the very same experimental protocol (11). They cast doubt on the reproducibility of data obtained with ex vivo lungs. Second, no TNF- was detected in the airspaces or the systemic circulation of animals during the production of severe pulmonary edema by high-VT ventilation. Third, the chemokine MIP-2 is found in vivo in lungs during mechanical ventilation, whatever the magnitude of VT, but our ex vivo data suggest that it is released in larger amounts when lung mechanical stretch is increased. Both our ex vivo and our in vivo data agree with the observation (14) that ventilation strategies that injure lungs do not necessarily result in primary production of proinflammatory cytokines in the lung.

Many studies have been done to describe the alterations in permeability, changes in lung mechanics, and deterioration of oxygenation that occur during the ventilation of normal or diseased lungs with a high VT (15). Several recent studies have suggested that proinflammatory cytokines play a role in the development of VILI. Narimanbekov and Rozycki (7) showed that administration of recombinant IL-1 receptor antagonist reduced the severity of lung injury produced by hyperoxia and ventilation with a 24 cm H₂O peak inspiratory pressure for 8 h in surfactant-depleted rabbits. Similarly, Imai and coworkers (8) reduced the severity of VILI in surfactant-depleted rabbits by giving anti-TNF- antibody.

The impression that proinflammatory cytokines play a role in the development of VILI was reinforced by several studies that showed that these mediators are released by damaged or even intact lungs subjected to "injurious" ventilation strategies (i.e., strategies that result in high tissue stress). Takata and coworkers found large increases in TNF- messenger RNA in the intraalveolar cells of surfactant-depleted rabbits after 1 h of conventional mechanical ventilation with peak inspiratory and end-expiratory pressures of 28 cm H₂O and 5 cm H₂O, respectively (17). Using an ex vivo rat lung model, Tremblay and coworkers found large concentrations of proinflammatory cytokines (TNF-, IL-1, IL-6) and of MIP-2 (11) in the BALF of lungs ventilated with 40 ml/kg VT. Increased concentrations of cytokines were also found in the perfusate of isolated mouse lungs ventilated with a high-volume protocol (12). These observations suggested that proinflammatory cytokines are released by intact lungs submitted to great mechanical stress.

Thus, our initial working hypothesis was that proinflammatory cytokines are also produced in vivo by the lungs of rats subjected to high-VT mechanical ventilation in the absence of PEEP. However, we were unable to detect any TNF- in the BALF of rats ventilated with 42 ml/kg VT for 2 h, which is a very injurious ventilation strategy. It has been shown repeatedly that after such prolonged ventilation with overdistension, small animals such as rats have extremely severe VILI, with profuse permeability edema and hyaline membrane formation. This was confirmed in the present study by the considerably larger BALF protein concentration in these rats than in those ventilated with 7 ml/kg VT. Our inability to find a proinflammatory cytokine response in vivo led us to reconsider the issue of cytokine production during the ventilation of lungs ex vivo. We therefore used exactly the same experimental protocol as Tremblay and coworkers (11) and were surprised to find no TNF- and only trivial amounts of IL-1 in the BALF of lungs ventilated according to the strategy they found most injurious (42 ml/kg VT, ZEEP). The reason for this discrepancy is unclear. It is unlikely that we failed to produce the same degree of lung injury as they did, because the protein concentrations in the BALF in our animals were similar to those they measured. We ruled out a defect in our TNF- ELISA because TNF- was present in the controls pretreated with LPS, and a recombinant TNF- from another supplier was correctly detected. The BAL procedures were performed at the very same time after the beginning of ventilation as in the study by Tremblay and coworkers. We also looked for proinflammatory cytokines in lungs ventilated for 30 and 60 min (data not shown), but found none.

One possible explanation for these discrepancies is that severely ischemic lungs (2 h without circulation) are unstable and yield unpredictable results. Alternatively, Tremblay and coworkers (11) found that the BALF TNF- concentration of lungs ventilated with 40 ml/kg VT did not differ whether or not the rats were given LPS 50 min before the lungs were removed. A possible explanation for this intriguing finding is that the animals that did not receive LPS were already "primed" for some unknown reason. Since Tremblay and coworkers performed no assay for endotoxin contamination, this hypothesis remains purely speculative. Both theirs and our studies used specific pathogen-free Sprague-Dawley rats from the same supplier (Charles River). However, weights of these animals differed in the two studies. We used the animals on the day after their arrival at our institution (body weight: 305 ± 3.5 [mean ± SEM] g), whereas Tremblay and coworkers (11) used older rats (body weight: 407 ± 41 g) that were housed in their institution for 2 wk (A. Slutsky, personal communication).

In the present study, rats were given a smaller dose of endotoxin than that in Tremblay and coworkers' study (11) (500 µg instead of 20 mg). We hypothesized that with large doses of endotoxin, we would have had less chance of observing a significant difference in cytokine release attributable to mechanical ventilation. Despite this smaller dose of LPS, we found significant amounts of TNF- and IL-1 in the BALF of rats injected with LPS. However, the TNF- concentration was not higher in lungs subjected to high-volume ventilation than in lungs ventilated with a low VT, suggesting that the release of this cytokine was due to the priming of lung cells by LPS, and not to high mechanical stress. These observations agree with those in recent studies using different experimental approaches (9, 10, 14). Pugin and coworkers found that human alveolar macrophages subjected to prolonged cyclic pressure-stretching strain (which may be an in vitro analogue of high-volume mechanical ventilation) did not release TNF- (or IL-6), but did release IL-8 (the human equivalent of rodent MIP-2) (9). Similarly, promonocytic human THP-1 cells released neither TNF- nor IL-6 under the same stretch conditions (9). Further, Vlahakis and coworkers found that a human alveolar epithelial cell line (A549) subjected to mechanical stretch did not release TNF- but did release significant amounts of IL-8 (10). These authors did not study the effect of stretch on IL-1 production. Our results suggest that mechanical deformation may affect release of this cytokine. After high VT ventilation (in lungs from animals given and not given LPS), however, we failed to observe the very high IL-1 concentration found in BALF by Tremblay and coworkers (11). Indeed, injurious ventilation in our hands led to the release of smaller amounts of IL-1 than did their control, low-VT ventilation.

Verbrugge and coworkers found that ventilating intact rats with 32 cm H₂O peak inspiratory pressure and ZEEP did not result in significant release of TNF- in the animals' airspaces or in their systemic blood plasma (14). Similarly, no TNF- was found in the BALF of rabbits that had undergone aggressive ventilation (16). Our in vivo results agree with these data and support the fact that high-volume mechanical ventilation alone is not sufficient to trigger the release of a proinflammatory cytokine (such as TNF- or IL-1) by intact lungs. It is worth noting that investigators who have found increased in vivo expression of TNF- in lungs during "injurious" ventilation strategies have always used animals with preinjured ("primed") lungs (8, 17, 18). Even in this setting (preinjured lungs), however, results are conflicting, since Verbrugge and coworkers found no TNF- in BALF of surfactant-depleted rats subjected to an injurious ventilatory strategy (14). In patients with severely inflamed lungs, it has been shown that a high PEEP/low VT ventilatory strategy could modify proinflammatory cytokine concentrations in BALF over time (19).

MIP-2 is a C-X-C chemokine that has chemotactic activity for neutrophils comparable to that of IL-8 in humans. As noted earlier, several investigators have shown that IL-8 is released by lung cells during mechanical stretch (9, 10). Our ex vivo observations are in agreement with these findings. We found more MIP-2 in isolated lungs subjected to ventilation with a 42 ml/kg VT than in lungs left in static inflation or ventilated with a 7 ml/kg VT (Figure 3). This demonstration would probably be more difficult in vivo because MIP-2 is continuously removed by blood flow and lymphatic drainage. The MIP-2 concentration in the BALF of isolated lungs, after 42 ml/kg VT ventilation, was significantly higher than that in the BALF of intact animals similarly ventilated. The release of MIP-2 or IL-8 by lung cells submitted to mechanical deformation may explain the observation of leukocyte recruitment in lungs during mechanical ventilation with large volume excursions (4, 20, 21). These leukocytes, if activated, may contribute to the worsening of VILI and perpetuation of an inflammatory state.

In conclusion, our results suggest that mechanical ventilation with a large VT alone is not sufficient to increase the release of proinflammatory cytokines into airspaces during mechanical ventilation. They support in vitro observations with cultured cells and confirm in vivo findings made with other protocols that showed no release of TNF- by large cell or tissue deformation.