INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Since its introduction into clinical practice more than 40 yr ago, mechanical ventilation has proved to be an important therapeutic tool for the treatment of respiratory failure. However, as our knowledge of how to apply mechanical ventilation has increased it has become evident that a number of ventilatory strategies can produce or worsen lung injury (1). The use of large tidal volumes (VT) (2), high peak airway pressures (3), high inspiratory flows (4), high respiratory rates (RR) (5), and end-expiratory alveolar collapse with cyclic reopening (6) have all been proposed to play a role in the pathogenesis of ventilator-induced lung injury. Despite the development of many different ventilatory strategies to reduce ventilator-induced lung injury, it remains a significant problem in the care of the critically ill (7).

In patients developing acute respiratory distress syndrome (ARDS), the leading cause of death is failure of vital organs as part of the multiple organ dysfunction syndrome (MODS) (8, 9). The mechanisms by which MODS develops in patients with ARDS are not certain, but a reasonable hypothesis is that the inflammatory cells, mediators, and cytokines, which have been implicated in the pathogenesis of ARDS (10, 11), play an important role. During acute lung injury the epithelial and endothelial barrier is damaged and compartmentalization of alveolar cytokines can be lost, with "leak" into the vascular systema process that can initiate or propagate a systemic inflammatory response (12). Within the context of this model, the lung is not only a target organ of an inflammatory response to a noxious stimulus (e.g., infection or trauma) but is also a source of systemic inflammation that could play an active role in the development of MODS.

There is evidence that the mechanical ventilation strategy can affect the inflammatory milieu in the lung. In an animal model of ARDS, conventional mechanical ventilation, as opposed to high-frequency ventilation, increased the concentration of inflammatory mediators (platelet-activating factor, thromboxanes, prostaglandins) in lung lavage (15), and intra-alveolar expression of tumor necrosis factor-alpha (TNF-) (16). Von Bethmann and coworkers, using an isolated perfused mouse lung model, found an increase of TNF- and interleukin-6 (IL-6) in the perfusate after only 30 min of ventilation with large VT (17). Our group recently reported that ventilation strategies associated with zero positive end-expiratory pressure (PEEP) and/or excessive end-inspiratory lung volume increased the concentration of lung lavage cytokines (18). To the extent that mechanical ventilation can lead to an inflammatory response in the lung (18) and increase alveolar capillary permeability, it could initiate or propagate a vicious cycle of inflammation leading to local and systemic tissue injury and thus could play a role in the development of MODS. A corollary of this hypothesis is that it might be possible to abrogate or worsen this systemic inflammatory response by manipulating the particular ventilatory strategy. In the present study we set out to examine the hypothesis that the use of injurious ventilatory strategies (large VT/low PEEP) could lead to an increase in systemic cytokine levels. To address this question we studied the effects of different ventilatory strategies on the local and systemic release of cytokines in a rat model of direct lung injury.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

All animals received care in compliance with the Principles of Laboratory Animal Care formulated by the Institute of Laboratory Resources, the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (1985, NIH Publication No. 86-23), and the Guide to the Care and Use of Experimental Animals formulated by the Canadian Council of Animal Care (1993, CCAC, 2nd ed.).

The studies were performed in male adult Sprague-Dawley rats (Charles River Labs, St. Constant, PQ, Canada), weighing 320 to 400 g. The rats were anesthetized with xylazine (20 mg/kg) and ketamine (50 mg/kg) intramuscularly. A tracheostomy was performed, and a cannula (14-gauge) was inserted into the trachea. To ensure subsequent bilateral distribution of the acid used to induce injury, the cannula was advanced only 0.5 cm. An angiocatheter (24-gauge) was inserted into the right carotid artery to sample blood for gas analysis (Ciba-Corning Model 248 blood gas analyzer; Corning Medical, Medfield, MA) and to measure mean arterial blood pressure (), via a pressure transducer (Pd 23; Gould, Inc., Cleveland, OH) zeroed at the midthorax. Anesthesia was maintained by intermittent intramuscular injections of ketamine (50 mg/kg/h). The animals were ventilated using a volume cycled ventilator (Harvard Rodent Ventilator, Model 683; Harvard Apparatus, South Natick, MA), at the following ventilator settings: VT 12 ml/kg, RR 38 breaths/min, fraction of inspired oxygen (FI_O2) 1. Muscle relaxation was obtained using pancuronium bromide (1 mg/kg/h) administered into the penile vein. Airway pressure was measured by a pressure transducer (HP 45 transducer; Validyne Engineering Corp., Northridge, CA), connected at the tracheostomy cannula, and the signal was recorded on a strip chart recorder (Gould ES 1000 recorder; Gould, Inc., Cleveland, OH).

The rats were stabilized (Pa_O2 > 400 mm Hg), placed in the left lateral decubitus position and then 0.8 ml/kg of HCl (pH 1.5, room temperature) was administered intratracheally followed by a bolus of 5 ml of air. The animals were shaken vigorously to facilitate dispersion of the acid. The rats were then placed in the right lateral decubitus position and 1.7 ml/kg of HCl was instilled, again followed by a bolus of 5 ml of air and vigorous shaking. Once the Pa_O2 decreased below 90 mm Hg, and peak inspiratory (PIP) and mean airway pressure () increased above 28 and 10 cm H₂O, respectively, the animals were randomized to the groups described subsequently.

Experimental Protocol

Five ventilatory strategies were employed to examine the effect of two different VT (high and low) in the presence or absence of PEEP, as well as to examine the effect of a recruitment maneuver. Group 1 (n = 8) was ventilated with high VT and PEEP (HVZP: VT 16 ml/kg, RR 32 breaths/min, FI_O2 1). Group 2 (n = 8) was ventilated with high VT and PEEP (HVP: VT 16 ml/kg, RR 32 breaths/min, PEEP 5 cm H₂O, FI_O2 1). In these two groups a small piece of silicone tubing was added between the y-piece of the ventilator circuit and the tracheostomy cannula to increase dead space to maintain normocapnia. Group 3 (n = 8) was ventilated with low VT and zero PEEP (LVZP: VT 9 ml/kg, RR 44 breaths/min, FI_O2 1). Group 4 (n = 8) was ventilated with low VT and PEEP (LVP: VT 9 ml/kg, RR 44 breaths/min, PEEP 5 cm H₂O, FI_O2 1). The level of PEEP was selected to maintain the same as in Group 1. Group 5 (LVPR; n = 8) was ventilated with the same ventilator settings as Group 4 plus an end-inspiratory hold of 30 s at an airway pressure of 30 cm H₂O performed at the beginning and every hour during the experiment (19), to evaluate the efficacy of a recruitment maneuver. After the recruitment maneuver, the tracheostomy connector was clamped to maintain the lung volume while mechanical ventilation was restarted in the expiratory phase.

At baseline, after acid aspiration injury, and every hour after randomization, , PIP, and were measured, and arterial blood samples (for cytokine assays and blood gases) were taken. In addition, after acid aspiration injury but before randomization, and at the second and fourth hour of the experiment, lung fluid was aspirated from the trachea using a cannula (20-gauge). To ensure maximal aspiration of fluid, during this procedure the animals were placed in the Trendelenburg position. The samples were immediately centrifuged (Eppendorf 5403; DraMed, Mississauga, ON, Canada) for 10 min (3,000 rpm; 4° C), and then the serum and lung fluid supernatants were stored at 70° C.

Lactated Ringer's solution was infused at a rate of 2 ml/h into the penile vein to replace insensible water loss. Pentastarch (Pentaspan DuPont Pharma, Clarkson, ON, Canada) was given each hour to replace the blood removed for sampling and also when needed to treat hypotension (  60 mm Hg). All animals were supine for the duration of the experiment. Four hours after acid aspiration, the animals were exsanguinated by transection of the carotid artery.

Cytokine Assays

Analysis of TNF- and macrophage inflammatory protein-2 (MIP-2) was carried out on the serum and lung fluid supernatants, in a blinded fashion, using a commercially available ELISA kit (Biosource International, Camarillo, CA). The sensitivity of these kits for TNF- and MIP-2 were 4 and 1 pg/ml, respectively, and both were specific for rat. The absorbance of each well was read at 450 nm with a MR600 microplate reader (Thermomax-Molecular Devices; Fisher Scientific Instrument, Nepean, ON, Canada). Background absorbency of blank wells was subtracted from the standards and unknowns prior to determination of sample concentrations.

Respiratory System Pressure Volume (PV) Curves

At the end of the experiment the animals were disconnected from the ventilator to allow the pressure in the lungs to reach atmospheric pressure. The lungs were then inflated twice to an airway pressure of 30 cm H₂O. Respiratory system PV curves were inscribed by manually inflating/deflating the lungs with 1-ml increments of pure O₂, starting at atmospheric pressure and continuing until Paw was 30 cm H₂O. Volumes were held at each step for 6 to 8 s (20).

To examine the effect of the acid aspiration alone on lung mechanics, we measured PV curves in eight additional animals that were not entered into the randomization protocol and were not ventilated for the 4-h study period. These measurements were made in these eight additional rats because we were concerned about potential hemodynamic instability and the development of severe hypotension during this maneuver, with a possible effect on cytokine levels. PV curves were measured as described previously, before induction of injury (i.e., acid installation) and immediately after they fulfilled our injury criteria (i.e., after reaching our end points: Pa_O2 < 90 mm Hg, PIP > 28 cm H₂O and  > 10 cm H₂O). Immediately after measurement of the postinjury PV curve, these animals were killed.

Statistical Analysis

All data are expressed as mean ± SEM. Comparison among groups was performed using two-way analysis of variance with two-way entry (ventilatory strategy, time). Student-Newman-Keuls correction was used for multiple comparisons (21). One-way analysis of variance or one-way analysis of variance on rank was used where appropriate. To evaluate the strength of the association between the levels of cytokines in serum and in lung fluid we carried out a Spearman rank order correction. A p value of < 0.05 was considered to be statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Table 1 summarizes the ventilatory variables for each of the five groups in the study. There were no differences among animals with respect to rat weight (350 ± 2 g). Before acid aspiration there were no significant differences in , Pa_O2, Pa_CO2, PIP, and . Similarly, after acid injury (before the randomization of animals into the five groups) there were no significant differences in these variables. After acid aspiration, decreased from 101 ± 2 to 80 ± 3 mm Hg (p < 0.01), and Pa_O2 from 454 ± 16 to 67 ± 5 mm Hg (p < 0.01), whereas Pa_CO2 increased from 31 ± 2 to 39 ± 3 mm Hg (p < 0.01). PIP increased from 13 ± 0.4 to 30 ± 0.7 cm H₂O (p < 0.01) and from 4 ± 0.2 to 10 ± 0.2 cm H₂O (p < 0.01). All the animals ventilated with low volume and zero PEEP died before the end of the third hour of the experiment.

Hemodynamics and Gas Exchange

After randomization and allocation to the five ventilatory groups, there were no intergroup differences in at any time point, although the trend in in the LVZP group is likely artificially elevated due to dropout of dead animals (Figure 1). The mean volume of Pentastarch infused in all the groups was 3.9 ± 0.1 ml; the only significant difference was between the HVZP versus LVP groups (4.8 ± 0.2 versus 3.5 ± 0.3 ml [p < 0.05]).

View larger version (14K): [in this window] [in a new window]   Figure 1.   Pa versus time for all study groups. Data are represented as mean ± SEM; number of animals in parentheses.  = high volume, zero PEEP (HVZP);  = high volume, PEEP (HVP);  = low volume, zero PEEP (LVZP);  = low volume, PEEP (LVP);  = low volume, PEEP + recruitment (LVPR).

The HVP group had the highest Pa_O2 compared with all the other groups (p < 0.05) (Figure 2, left panel ). Pa_CO2 was slightly higher in the LVP group compared with all the other groups, although this difference was not statistically significant (Figure 2, right panel ).

View larger version (17K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   Figure 2.   (Left panel ) PaO2 versus time for all study groups; *p < 0.05 versus all other groups. (Right panel ) PaCO2 versus time for all study groups. Data are expressed as mean ± SEM; number of animals in parentheses. Symbols as in Figure 1.

Respiratory Mechanics

The PIP level was higher in the HVZP versus the LVPR and LVP groups (p < 0.05); PIP was lower in the LVPR versus the HVP group (p < 0.05) (Figure 3). The HVP group had the highest , whereas the LVZP group had the lowest (p < 0.05) versus all the other conditions (Figure 3, bottom panel ).

View larger version (24K): [in this window] [in a new window]   Figure 3.   Upper portion of graph: PIP versus time for all study groups. p < 0.05 versus LVP/LVPR; p < 0.05 versus HVP. Lower portion of graph: Paw versus time. *p < 0.05 versus all other groups. Data are expressed as mean ± SEM; number of animals in parentheses. Symbols as in Figure 1.

The results of the PV curves are presented in Figure 4. Data were not obtained in the LVZP group because of the early deaths in all animals. The dotted lines in Figure 4 represent the mean curves obtained in the eight additional animals (those not allocated to Groups 1-5) studied specifically for determination of PV curves. These results are also summarized as the change in lung volume from FRC to an airway pressure of 30 cm H₂O (V₃₀) and presented in Figure 5 for the four groups. Acid injury by itself was associated with a decrease in V₃₀ (p < 0.05). After 4 h of ventilation the HVZP and LVP groups had a further reduction in V₃₀ (p < 0.05) and a downward shift of the PV curves compared with the HVP and LVPR groups.

View larger version (25K): [in this window] [in a new window]   Figure 4.   Static compliance curves of total respiratory system referenced to FRC. Dotted lines indicate mean data for the eight animals studied specifically for determination of PV curves after acid aspiration, and are identical in all four panels. Solid lines represent PV curves after 4 h; each panel represents a different ventilatory strategy. Data are expressed as mean ± SEM.

View larger version (17K): [in this window] [in a new window]   Figure 5.   Changes in volumes from FRC to an airway pressure of 30 cm H2O ( Volume). Note that the volumes for the group "Before Acid Injury" and the group "Acid Injury" were obtained from the eight animals studied specifically for determination of PV curves. *p < 0.05 versus all other groups; p < 0.05 versus Acid Injury, HVZP, LVP. Data are expressed as mean ± SEM.

Lung Fluid Production

There were no differences among animals in the volume of lung fluid aspirated at baseline: 0.2 ml. The mean volume of the lung fluid aspirated (second plus fourth hour) was 4 ± 0.2 ml. The HVZP had the highest fluid production, 5 ± 0.3 ml (p < 0.05), compared with the HVP and LVP groups.

Cytokines in Lung Fluid

We found no differences in the concentration of TNF- (8 ± 2 pg/ml) and MIP-2 (313 ± 27 pg/ml) at baseline among animals. There were no differences in lung fluid cytokine concentrations among the ventilatory strategies at either 2 or 4 h (Table 2). We also calculated the total quantity of lung cytokines (picograms) (volume of lung fluid aspirated [milliliters] times the respective concentration of cytokines [picograms per milliliter]). The highest lung levels for both cytokines were in the HVZP group versus all other groups (p < 0.05) at the fourth hour (Figure 6). There was a significant correlation between the total amount of TNF- and MIP-2 in lung fluid (Spearman rank order correlation [r_s] = 0.79, p < 0.01).

View larger version (15K): [in this window] [in a new window]   Figure 6.   Mean absolute quantity of cytokine (calculated as volume of lung fluid aspirated times cytokine concentration). *p < 0.05 versus all other groups at 4 h; p < 0.05 versus time 0; p < 0.05 versus time 0 and 2 h. Data are expressed as mean ± SEM; number of animals in parentheses. Symbols as in Figure 1.

Serum Cytokines

There were no significant differences among animals in mean serum levels of TNF- (11 ± 3 pg/ml) and MIP-2 (3 ± 1 pg/ ml) at baseline. In the HVZP group there was a progressive increase in serum TNF- that was significantly different than all other groups from 1 to 4 h. Serum MIP-2 concentrations were higher at all time points in the HVZP group after 1 h and this reached statistical significance at 3 and 4 h. The HVP group had significantly higher MIP-2 levels at 4 h compared with the LVP and LVPR groups. The highest levels of both cytokines were found in the HVZP group (p < 0.05) at the third and fourth hour (Figure 7). There were significant correlations between the concentrations of TNF- and MIP-2 in serum (r_s = 0.63, p < 0.01); between the total amount of TNF- in lung fluid versus serum concentration (r_s = 0.57, p < 0.01); and between the total amount of MIP-2 in lung fluid versus serum concentrations (r_s = 0.75, p < 0.01).

View larger version (15K): [in this window] [in a new window]   Figure 7.   Mean serum cytokine levels during experiment. *p < 0.05 versus all other groups at same time point; p < 0.05 versus time 0; p < 0.05 versus all time points; §p < 0.05 versus time 0 and 1 h. Data are expressed as mean ± SEM; number of animals in parentheses. Symbols as in Figure 1.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The major findings of this study are that the particular mode of ventilation used in the treatment of acute lung injury can have a significant effect on the deterioration of pulmonary mechanics, the production of cytokines in the lung, and most importantly on the development of elevated serum cytokine levels.

We used acid aspiration because it is an established, extensively investigated model of ARDS. Moreover, this type of injury is a well-known cause of ARDS with an associated mortality rate above 40% (22, 23). Aspiration of acid into the airways of the lung causes a deterioration of gas exchange, pulmonary edema, decrease of lung compliance, and release of proinflammatory cytokines (24). An advantage of the model in assessment of airway/alveolar production of cytokines is the high production of pulmonary fluid, which allows direct aspiration of lung fluid without requiring lung lavage, thus overcoming the problem of how to normalize the data for dilution due to the volume of the lavage fluid. In addition, we did not want to perform too many lung lavages with saline to minimize the risk of hypoxemia, hemodynamic instability, and the potential of an "artifactual" alteration of lung mechanics due to the effect of the saline in the lung.

We studied the effect of two different VT (high and low) in the presence or absence of PEEP, as well as the effect of a recruitment maneuver. The ventilatory settings and the levels of PEEP were selected to maintain similar between the high and low VT groups, so as to hopefully obtain similar (25, 26). We initially wanted to use higher levels of VT and/or PEEP but in pilot studies the animals became unstable, developing systemic hypotension and metabolic acidosis.

There was severe impairment of gas exchange in all ventilatory groups and it is unlikely that the differences in cytokine concentrations could have been caused by differences in Pa_O2, especially since the highest cytokine levels were found in the group with Pa_O2 in the midrange, and Pa_O2 was not significantly different from those with much lower cytokine production (all low VT groups). Similarly these results are unlikely to be due to differences in Pentastarch infusion (27) since these were only different between the HVP and LVP groups; nor to differences in hemodynamics, since was not different among groups.

Effect of PEEP

We found the highest cytokine concentrations in the group ventilated with high VT and zero PEEP (HVZP). This strategy was associated with a decrease in compliance and a shift of the PV curves compared with the group ventilated with the same VT and PEEP (HVP). This shift in the PV curves was similar to our previous ex vivo observations (6) and may be related to inactivation or depletion of pulmonary surfactant (2), or to overdistension and/or recruitment/derecruitment of alveolar units.

The absence of PEEP in the group with low VT (LVZP) had a dramatic effect on the survival with all animals becoming extremely hypotensive before they died. We did not obtain postmortem pathological examination of the animals and can only speculate on the cause of the high mortality. We suggest that low VT without PEEP caused development of regional traction and shear forces in the atelectatic lung leading to irreversible worsening of preexisting lung injury with associated hemodynamic instability (i.e., hypotension), possibly aggravated by an increased pulmonary vascular resistance at the lower in this group (6). The high-volume group with zero PEEP may have recruited sufficient alveolar units to minimize these effects (28).

Effect of Recruitment Maneuver and VT

Although ventilation with small VT and PEEP above the inflection point has been shown to reduce lung injury, it carries with it the risks of hyperinflating the lungs and air leaks (6, 26). In our model we could not use levels of PEEP in the range of the inflection point because of marked hemodynamic impairment, as assessed in pilot studies. Small VT ventilation can lead to a progressive decrease in compliance, which can be reversed by introducing a large breath or a sigh (29). During high-frequency oscillatory ventilation, a recruitment maneuver has been found to decrease lung injury (30). In a uniformly recruited lung, transpulmonary and local distending stress forces are equal. However, during acute lung injury, nonhomogeneity is present, and the pressure tending to open an atelectatic part of lung would no longer be the transpulmonary pressure. Mead and coworkers demonstrated that the stress on an atelectatic region surrounded by a fully expanded lung could be as high as 140 cm H₂O (31). The key factor in reducing this injury is reexpansion of lung parenchyma and maintenance of adequate alveolar volume.

We tried to minimize these regional traction forces by using a recruitment maneuver, i.e., inspiratory hold at 30 cm H₂O of airway pressure for 30 s that was performed at the beginning and every hour during the experiment. This maneuver prevented the reduction of compliance and the shift of the PV curve, compared with the group without recruitment.

We found a significantly better compliance in the group with high VT compared with low VT at the same level of PEEP. We suggest that the use of high VT compared with low VT permitted a sufficiently high opening pressure with intratidal lung recruitment (i.e., recruitment during the inspiratory phase of tidal ventilation only in the groups with high VT) that avoided the cycling opening and collapsing of alveolar units (28).

Implications

It is now well known that mechanical ventilation can initiate or augment lung injury (1). Despite the focus on manipulating ventilatory strategies to overcome this injury, the mortality rate in patients with ARDS remains very high (32). Of note, although the most obvious abnormalities during ARDS are referable to the lung, the major cause of mortality is not hypoxemia but the development of MODS secondary to a systemic inflammatory response (9). One hypothesis to explain this observation in some patients is that mechanical ventilation can begin and/or amplify an inflammatory response in the lung that then propagates a vicious cycle leading to local and systemic tissue injury (12, 18). Using an ex vivo nonperfused rat model of ARDS, we recently demonstrated that different ventilatory strategies produced markedly different levels of lung lavage cytokine concentrations and tissue expression of an immediate early response gene (18). In the present study, the cytokine concentrations in the HVZP group at 4 h were comparable or slightly higher than the other groups. There was, however, an increase in the total quantity of lung cytokines owing to an increase in lung fluid aspirated, but the relative increase (approximately twofold) was relatively modest. This lack of increase in cytokine concentrations in the HVZP group may be due to a number of factors: (1) the degree of lung stretch in the present in vivo model was less than in the previous ex vivo model because we used smaller VT to ensure hemodynamic stability; (2) the extent of recruitment/ derecruitment at low lung volumes would also have been less because at zero end-expiration pressure, the transpulmonary pressure was greater than zero due to the pleural pressure in the alive rat; and/or (3) the "control" group in the present study had acid-induced lung injury, and not healthy lungs as in the previous study.

The increased serum cytokines in the HVZP group could have resulted from an increase in airspace cytokines and/or an increase in alveolar-capillary permeability induced by the ventilatory strategy. Given the comparable values for lung fluid cytokine concentrations, we think that the latter explanation is more likely in this model. A number of investigators have shown that the use of high VT alone (33) or the use of a ventilatory strategy without PEEP at low lung volumes can increase capillary permeability (23) because of reopening and collapse of lung parenchyma likely through the generation of elevated shear forces. An increase in permeability appears to be required for translocation of cytokines from the lung into the circulation (12, 35).

The present study did not examine whether the increased lung cytokines affected lung injury, or whether lung injury caused the increase in cytokines, although both hypotheses are likely correct to some extent. For example, using a model of lung lavage/hypoxia/ventilator-induced lung injury, Narimanbekov and Rozycki showed that rabbits treated with a recombinant IL-1 receptor antagonist had significantly lower concentrations of albumin and elastase, and lower lung neutrophil counts in their lungs although there was no effect on dynamic compliance and oxygenation (36).

These results of the present study provide further evidence for the hypothesis that particular injurious ventilatory strategies may play a role in inducing a cytokine response in the lung that may lead to an increase in serum cytokine levels. We speculate that this increase in cytokines can initiate or promulgate multisystem organ failure (37). If this hypothesis is correct, the development of optimal ventilatory strategies becomes crucial, not only to provide adequate gas exchange and minimize ventilator-induced lung injury, but also to prevent the development of multiple organ failure, a major cause of death in patients with ARDS.