Introduction

In the past two decades we have substantially increased our knowledge about many pathophysiological aspects of acute lung injury (ALI) and the acute respiratory distress syndrome (ARDS) [1]. Several authors [2, 3, 4] have suggested that the reported decreases in mortality observed in some clinical trials are a result of differences in patient selection, severity of the underlying disease, or therapeutic changes. Mechanical ventilation (MV) is one of the most important aspects of supportive care in the treatment of patients with ALI and ARDS. However, by using different modes of MV the clinician is often applying levels and patterns of pressure, tidal volumes (VT), and concentration of inspired oxygen well beyond the levels that normal lungs experience [5]. High airway pressures, high VT, and low levels of positive end-expiratory pressure (PEEP) are commonly applied in the treatment of patients with ALI/ARDS [6]. Evidence from experimental studies suggests that the application of high airway pressures and resultant lung overdistension during cyclic mechanical ventilation causes lung injury [7, 8, 9]. Referred to as ventilator-induced lung injury, this condition resembles ALI and ARDS, and it is difficult to identify in humans because its appearance may overlap the underlying disease. Emphasis on this new entity has prompted a number of investigators to suggest that the cost of maintaining normal physiological variables may be too high [10], indicating that ALI/ARDS may in part be a product of our therapy rather than the progression of the underlying disease.

Few experimental studies [11, 12, 13, 14] have explored the effect of different ventilatory patterns on the pulmonary and systemic inflammatory responses. Although those studies have shown that certain modes of MV can lead to induction and release of proinflammatory cytokines from the lungs, the reported models of lung or systemic injury differ from conditions in clinical practice. In a series of experiments we explored the effects of PEEP during 3 h of mechanical ventilation with low or high VT in the setting of an in vivo, clinically relevant model of sepsis-induced ALI. Our hypothesis was that even in the presence of a systemic inflammatory response caused by sepsis the application of PEEP would modify the expression and synthesis of proinflammatory mediators during ventilation with low and injurious tidal volumes.

Methods and materials

Animal preparation

Experimental protocol was approved by the Hospital NS de Candelaria Research Committee and the Animal Care Committee, University La Laguna, Tenerife. We studied male Sprague-Dawley rats weighing 300–350 gm (CRIFFA, Barcelona, Spain) anesthetized by intraperitoneal injection of 50 mg/kg ketamine hydrochloride and 2 mg/kg xylazine. Sepsis was induced by cecal ligation and puncture (CLP), a well-characterized animal model of bacterial peritonitis which mimics many features of the human septic syndrome [15, 16]. All septic animals develop ALI by 18 h after CLP [16]. Histopathological features of the sepsis-induced ALI are atelectasis, pulmonary edema, and acute inflammatory infiltrates. At 18 h after CLP the peritoneal cavity was reopened under anesthesia in surviving animals. The cecum was then excised distal to the ligature and removed. We washed the peritoneal cavity with 20 ml warm, normal saline, and gently squeezed several times. After closing the abdomen each animal received 10 ml normal saline subcutaneously for fluid resuscitation throughout the experiment. Then we performed a cervical tracheotomy using a thin-walled 14-G Teflon catheter that was secured by a ligature around the trachea. Animals were paralyzed with 1 mg/kg pancuronium bromide and connected to a time-cycled, volume-limited rodent ventilator (Ugo Basile, Varese, Italy). In a separate group of five septic animals a pressure-volume curve of the respiratory system was constructed to determine the value of the pressure at the lower inflection point (Pinf). The curve was determined manually by a step-wise injection of 0.5 ml air into the trachea to a total of 5 ml while monitoring the airway pressure. The Pinf in these five representative animals of our model of sepsis-induced ALI was between 6 and 7 cmH2O. This value was used to determine the set PEEP level of 8 cmH2O (above the inflection point) in all animals studied.

Experimental protocol

Rats were randomly divided into six groups. Two groups of animals were anesthetized, observed and not ventilated for 3 h: control, healthy animals (n=20); control, septic animals (n=20). Four groups of animals were ventilated for 3 h with 100% oxygen: group 1 (n=20): septic animals, low VT, zero PEEP; group 2 (n=21): septic animals, high VT, zero PEEP; group 3 (n=22): septic animals, low VT, PEEP=8 cmH2O; group 4 (n=22): septic animals, high VT, PEEP=8 cmH2O. We used a low VT of 6 ml/kg and a high VT of 20 ml/kg, a value that, although double than those used clinically in ALI/ARDS patients, produces regional lung stretch that is likely comparable to that experienced by some patients with ALI/ARDS in nondependent areas of the lung even when relatively small VT are used [17, 18]. In our own preliminary experiments we found that the minimal level of overdistension in rats that causes an easily identifiable injury is 20 ml/kg. We did not measure arterial blood pressure during the study since during the preliminary experiments to set-up the conditions of the present study we did not find significant differences among the ventilatory groups in the values of mean blood pressure measured through a catheter (ED 0.96 mm; Intramedic, Clay Admas, Parsippany, N.J., USA) inserted into the right carotid artery of the rats. Respiratory rate was adjusted taking into account the delivered tidal volume and animal body weight to maintain constant minute ventilation in all groups. Respiratory rate and tidal volume were maintained constant during the 3-h period of MV under anesthesia with ketamine/xylacine and muscle paralysis with pancuronium bromide, with animals lying supine on a restraining board inclined 15° from the horizontal. Peak inspiratory pressures and rectal temperature were recorded.

Outcomes and pathological examination

At the end of the 3-h ventilation period blood samples for gas analysis were obtained from the left ventricle in all surviving animals. A midline thoracotomy/laparotomy was performed on the first six surviving rats from each group and 2 cc blood was withdrawn by cardiac puncture from a beating heart for cytokine analysis, and the abdominal vessels were transected. After the animals were killed, the heart and lung were removed from the thorax en bloc. Then the lungs were isolated from the heart, the trachea was cannulated, and the right lung was fixed by intratracheal instillation of 3 ml 10% neutral buffered formalin. After fixation the lungs were floated in 10% formalin for at least 24 h. The lungs were serially sliced from apex to base. These specimens were processed in the usual manner, embedded in parafin, cut at 10 µm thickness, stained with hematoxylin-eosin and with the Masson-Goldner trichrome technique, and examined under light microscopy. Random sections of the right lung from each animal were examined by two pathologists (F.V., L.D.F.), blinded as to the experimental history of the lung specimens, with particular reference to alveolar and interstitial damage defined as either cellular inflammatory infiltrates, pulmonary edema, disorganization of lung parenchyma, alveolar rupture, and/or hemorrhage. A semiquantitative morphometric analysis of lung injury was performed by scoring from 0 to 4 (none, light, moderate, severe, very severe) for each parameter. A total lung injury score was obtained by adding the individual scores in every animal and averaging the total values in each group.

RNA extraction and cytokine gene expression in the lungs

In the same rats that we used above for histology the left lung was excised, washed with saline, frozen in liquid nitrogen, and stored at −70°C for subsequent RNA extraction (RNA isolation kit, Boehringer-Mannheim, Mannheim, Germany). Total RNA from each sample were reverse transcribed according to manufacturer's instructions (Reverse Transcription System, Promega, Wis., USA). Sample to sample cDNA normalization was carried out by β-actin amplification. We measured gene expression levels of the proinflammatory cytokines tumor necrosis factor (TNF) α, interleukin (IL) 1β and IL-6. The reverse-transcribed cDNA (5 µl) was added to 50 µl of final reaction volume of polymerase chain reaction (PCR) mix containing: PCR buffer (10 mM Tris-HCl pH 8.3, 50 mM KCl), 10 mM of each deoxyribonucleoside triphosphate, 1 mM of MgCl2, 40 pmol of each primer (Table 1), and 0.6 U Bio-Taq DNA polymerase (Bioline). Hot-start PCR amplification was performed in a GeneAmp PCR system 2400 thermal cycler (Perkin-Elmer) following a program with a first step of 94°C for 1 min, n cycles (Table 1) of 94°C for 1 min, specific annealing temperature (Table 1) for 1.5 min and 72°C for 2 min, and a final step of 72°C for 7 min. The amount of cDNA used and the number of cycles of amplification were adjusted to remain within the linear region of amplification to allow semiquantification. PCR products were identified by electrophoresis on a 1.5% agarose gel stained with ethidium bromide and photographed under UV light. The resulting images were captured and densitometry was performed (Quantity One, GS-710 Bio-Rad) normalizing the intensity (0 for negative control; 1 for the highest intensity). The reverse transcriptase PCR analysis was repeated to confirm reproducibility using the samples collected from six individual rats in each experimental group.

Table 1. Specific primers' sequences, PCR cycles and annealing temperature for determining TNF-α, IL-1β, IL-6 and β-actin gene expression (S sense, A antisense)
Full size table

Cytokine ELISAs

At the end of every experiment, blood was collected by cardiac puncture, left to clot and centrifuged for 15 min at 3,000 rpm. Sera were divided into aliquot portions and frozen at −80°C. TNF-α, IL-1β, and IL-6 protein concentrations in serum were measured by enzyme-linked immunosorbent assay (ELISA) in dilutions that allowed interpolation from a simultaneously run standard curve. Levels of TNF-α, IL-1β, and IL-6 were measured with a commercially available ELISA (Cytoscreen, Biosource International, Camarillo, Calif., USA). Results were analyzed by spectrophotometry at 450 nm using an ELISA microplate reader (ELx800 NB Universal Microplate Reader, Bio-Tek Instruments, Winooski, Vt., USA). The threshold sensitivity was less than 8 pg/ml for IL-6 and less than 4 pg/ml for TNF-α and IL-1β.

Data analysis

Statistical analysis was performed with Fisher's exact test and paired and unpaired t tests, as appropriate. Values are expressed as group mean ±SD. Comparisons involving all four groups of septic animals were performed with one-way analysis of variance. If a difference was found, the t test were used. Values derived from densitometry (absorbance units) of cytokine gene expression were expressed as group median and tested with the Kruskall-Wallis and the Mann-Whitney U tests. Data from ELISA were tested for normality (Kolmogorov-Smirnov test with Lilliefors's correction) and analyzed by one-way analysis of variance, followed by the Student-Newman-Keuls all pairwise multiple range test. Data management was performed using SPSS (version 10, SPSS, Chicago, Ill., USA). Differences were regarded as significant at p<0.05.

Results

Outcome and physiopathological evaluations

Although all septic lungs were similar in appearance after 18 h of CLP, they markedly differed after 3 h of MV, depending on the randomization assignment. Three hours of ventilation with high VT was associated with more damage than low VT (Fig. 1). Animals ventilated with high VT + PEEP above Pinf showed the most severe lung injury score whereas animals ventilated with low VT + PEEP above Pinf had less evidence of lung damage (p<0.01; Fig. 2). Peak inspiratory pressures exceeded 30 cmH2O only in those animals ventilated with high VT plus PEEP above Pinf (Table 2). CLP-induced sepsis resulted in respiratory failure with hypoxemia (65±7 mmHg), hypercapnia (55±5 mmHg) and metabolic acidosis (pH 7.18±0.04). With the application of MV all groups had an elevated PaO2, a normal PaCO2 and remained with an acidotic pH at the end of the experiment. Animals ventilated with high VT had the highest values of PaO2 (p<0.05; Table 3). MV decreased the risk of death in most groups compared to nonventilated animals, but the application of PEEP above Pinf in the low VT group had the greatest beneficial effect on mortality (25% vs. 0%, p<0.05) whereas it increased mortality in the high VT group (19% vs. 50%, p=0.03; Table 2).

Fig. 1.
figure 1

Histopathological features of septic lungs after 3 h of mechanical ventilation. Top left Low VT without PEEP; perivascular edema, moderate inflammatory infiltrates, and atelectasis; hematoxylin and eosin, ×100. Top right High VT without PEEP; perivascular edema, moderate acute inflammatory infiltrates, and derangement of lung architecture; hematoxylin and eosin, ×400. Bottom left Low VT with PEEP; decreased edema, occasional cellular infiltrates, relatively normal alveolar size; hematoxylin and eosin, ×200. Bottom right High VT with PEEP above Pinf; intense hemorrhage and edema surrounding arterial vessels, pulmonary infiltrates, disorganization of lung architecture, alveolar rupture and alveolar hemorrhage; Masson-Goldner, ×200

Full size image
Fig. 2.
figure 2

Lung injury score in all experimental groups of septic, mechanically ventilated animals. Animals ventilated with high VT showed the showed the most severe lung damage (*p<0.01). The application of PEEP above the inflection point in low VT ventilated animals was associated with lowest lung damage (*p<0.01). n Normal, healthy, anesthetized, nonventilated; s septic, anesthetized, nonventilated, 3 h after cecum removal; shv septic, high VT and zero PEEP; slv septic, low VT and zero PEEP; shvp septic, high VT plus 8 cmH2O of PEEP; slvp septic, low VT plus 8 cmH2O of PEEP

Full size image
Table 2. Number of animals, mortality rate, ventilator rate and peak inspiratory pressures in experimental animals 18 h after sepsis within 3 h of mechanical ventilation using different ventilatory strategies. Mechanical ventilation decreased the risk of death in groups I, II, and III as compared to septic, control animals (Pins peak inspiratory pressure, V T tidal volume, P inf inflection point on the inspiratory limb of the pressure-volume curve of the respiratory system)
Full size table
Table 3. Arterial blood gas measurements from all septic, surviving animals ventilated at different tidal volumes and PEEP, measured at the end of the 3-h ventilation period (V T tidal volume)
Full size table

Cytokine gene expression in the lungs and Cytokine ELISAs

MV modified TNF-α, IL-1β, and IL-6 expression in the lungs of septic animals (Fig. 3A, B). Although all ventilated animals had their cecum removed, IL-1β and IL-6 mRNA expression in the lung increased after 3 h of MV in all groups except in those lungs ventilated with low VT plus PEEP (p<0.05). During low VT mechanical ventilation PEEP decreased mRNA expression of TNF-α, IL-1β, and IL-6 to values similar to those found in normal lungs. Plasma levels of TNF-α in nonventilated, septic rats were 25 times higher than those in control animals (4.1±1.1 vs. 94±18 pg/ml, p<0.001). Three hours of MV with PEEPflex was associated with a significant reduction in plasma levels of TNF-α, irrespective of VT (p=0.04, p=0.002, p=0.016; Fig. 4). The highest levels of circulating TNF-α were found in animals ventilated with high VT and no PEEP (135±23 pg/ml) whereas the lowest levels were found in animals ventilated with low VT plus PEEP (18±6 pg/ml, p=0.002). As well, MV increased plasma levels of IL-6 in all groups of animals (p<0.05), except in those ventilated with low VT and PEEP in which MV caused a reduction in circulating IL-6 (495±77 pg/ml, p=0.01; Fig. 4). Plasma levels of IL-1 protein were undetectable in all groups of animals.

Fig. 3. a
figure 3

Representative gel of TNF-α, IL-1β, and IL-6 mRNA expression in septic lungs after 3 h of mechanical ventilation. n Normal, healthy, anesthetized, nonventilated; s septic, anesthetized, nonventilated, 3 h after cecum removal; shv septic, high VT and zero PEEP; slv septic, low VT and zero PEEP; shvp septic, high VT plus 8 cmH2O of PEEP; slvp septic, low VT plus 8 cmH2O of PEEP; (−)c negative control, no ARN; M markers. b Densitometric analysis (absorbance units) of TNF-α, IL-1β, and IL-6 gene expression in the lungs of septic animals showing the rise or fall after 3 h of mechanical ventilation at low or high VT with or without PEEP. Bars Median of six rats per group. Septic animals ventilated with low VT plus PEEP had the lowest expression of cytokines. *p<0.05, **p<0.01 (Kruskall-Wallis and Mann-Whitney U tests)

Full size image
Fig. 4.
figure 4

Effects of anesthesia, sepsis, and mechanical ventilation for 3 h with low or high VT (with or without PEEP) on plasma TNF-α and IL-6 concentrations. n Normal, healthy, anesthetized, nonventilated; s septic, anesthetized, nonventilated, 3 h after cecum removal; shv septic, high VT and zero PEEP; slv septic, low VT and zero PEEP; shvp septic, high VT plus 8 cmH2O of PEEP; slvp septic, low VT plus 8 cmH2O of PEEP

Full size image

Discussion

To our knowledge, this is the first report in which several strategies of MV have been fully evaluated in a model of CLP sepsis-induced ALI. Our most remarkable findings were that the application of PEEP in septic animals ventilated with low VT (a) not only attenuated sepsis-induced lung damage but also reduced the systemic and local inflammatory responses associated with sepsis, and (b) prevented animals from dying. Our experimental model of ALI is clinically relevant for several reasons: (a) it is an in vivo model; (b) sepsis is the main cause of human ALI/ARDS; (c) sepsis and its clinical consequences are major contributors to mortality in most ICUs; and (d) the CLP model develops slowly and mimics the human septic syndrome, as seen in patients with peritonitis due to perforated abdominal viscera or in surgical patients with concomitant intra-abdominal infection [5, 16]. Within 18 h after CLP rats have a peak in serum concentrations of endotoxin, a persistent increase in plasma levels of TNF-α, and blood cultures that grow numerous enteric micro-organisms [16].

Several relevant conclusions can be drawn from the present study. First, during sepsis the mode of MV affects outcome. As in previous reports, our data strongly demonstrated that lung overdistension can produce lung damage and worsen lung injury [7, 8, 9, 11, 12, 13, 17, 18, 19, 20]. However, most of those studies were performed under experimental conditions that differ from the clinical setting such as: the use of ex vivo lung models [12, 13, 20], the use of lung injury models induced by lung lavage [20], aspiration [18], drugs or endotoxin injection [12], damaging normal or injured lungs by the application of VT at 40 ml/kg or higher and/or very high inspiratory pressures [7, 8, 9, 12, 19, 21] or examining the effects of MV few minutes after the induction of lung damage [12, 14, 18, 21, 22]. In addition, the number of animals in many studies has been insufficient to support any statistically valid conclusion. Furthermore, some of them have shown contradictory results [12, 21]. Second, in contrast to the Webb and Tierney [7] study, we did not find that the use of high levels of PEEP protected lungs from the high-volume iatrogenic lung injury despite the fact that TNF-α gene expression and plasma levels were slightly reduced. However, there are marked differences between those experiments and our study. First, Webb and Tierney [7] ventilated healthy rat lungs for only 1 h; second, they subjected healthy lungs to peak pressures of 45 cmH2O, a level of pressure that exceeds the pressure reached in any ventilatory strategy currently in use for ventilating healthy lungs; third, although those levels of inspiratory pressure have been used by other investigators [7, 12, 21], the VT required to reach such pressures is at least 40 ml/kg, an inspiratory volume that kills animals within 1 h of ventilation; and forth, although PEEP protected against alveolar flooding and supported gas exchange, there was no evidence of whether it protected against lung injury. Since in those experiments the ventilatory rate was unchanged, lungs were hyperventilated with such high minute ventilation that animals rapidly died from barotrauma, alveolar edema, and shock. In all our ventilated animals we maintained minute ventilation by manipulating ventilator rate. Therefore animals in the high-VT group were ventilated with a lower ventilatory rate than those ventilated with low VT, a condition that never has been explored before. The elevated PaO2 during ventilation with high VT underscores the fact that achieving higher PaO2 during MV does not always mean better outcome. In fact, higher PaO2 may be achieved at the expense of lung injury. In our study the application of high PEEP produced further pulmonary overinflation and caused airspace enlargement and hemorrhage. Rouby et al. [23] found that 86% of lungs in a series of patients with ALI who died after being mechanically ventilated had airspace enlargement that was associated with high peak airway pressures and large VT. Historically patients with ALI and ARDS have been ventilated with VT of 10–15 ml/kg and inspiratory pressures have been allowed to increase above 45 cmH2O. In a recent study Frank et al. [22] showed that a reduction in tidal volume from 12 to 6 ml/kg in a rat model of acid-induced lung injury diminished the degree of pulmonary edema, despite the fact that none of the tested strategies by those authors caused lung injury. In a hallmark paper Suter et al. [24] treated 15 patients with acute respiratory failure with volume-controlled ventilation using VT between 13–15 ml/kg, and until very recently this was the standard of care for ventilatory management of critically ill patients [5]. Although most clinical trials have used VT of 9–10 ml/kg, others have subjected patients to high VT similar to that in our experiments [25]. Lung volume in mammals is directly related to body weight; as a general pattern, all mammals are scaled similarly and have a tidal volume of about 6.3 ml/kg body weight [26]. However, this information never found a place in clinical practice until very recently [27, 28].

The precise mechanism linking sepsis and ALI is still a mystery. Inflammation in the lung involves a delicate balance between pro- and anti-inflammatory mediators, with the cytokines dictating the ultimate response to initial injury [29]. In ALI/ARDS the extreme tissue inflammation may result in severe irreversible lung injury. Recent studies [11, 12, 14, 30] have demonstrated that MV induces an up-regulation of local cytokines that may contribute to ventilator-induced lung injury. Some of those reports [12, 14, 30] speculated that since the pulmonary endothelial barrier is damaged during ALI, lung cytokines released into the circulation can initiate or propagate a systemic inflammatory response and play an active role in the development of multiple system organ dysfunction. Using an isolated rat lung model, Tremblay et al. [12] examined the effects on lung gene expression and lung lavage levels of several cytokines under the effects of 2 h of four different ventilatory strategies in the absence of or 50 min after endotoxin injection. They selected three different VT values (7, 15, 40 ml/kg) and three levels of PEEP (0, 3, 10 cmH2O), but they studied only four combinations among the nine possibilities. Despite the lack of results from those five missing groups, Tremblay et al. [12] found that the concentrations of lung cytokines were significantly higher in lungs ventilated with 0 PEEP plus VT of 40 ml/kg, and were very low in lungs ventilated with 7 ml/kg plus 3 cmH2O of PEEP. It is possible that the pattern of cytokines seen by Tremblay et al. was similar in the endotoxin-injected group than in the control group because 50 min could not be a sufficient time for the developing of a fully sepsis-induced ALI syndrome after the injection of endotoxin. It is difficult to compare their findings with those of our study since we used a different animal model, different volumes, different levels of PEEP, and different combinations of pressure and volume. However, we also found that in septic animals the combination of high VT with zero PEEP had a synergistic effect on cytokine gene expression. Chimello et al. [14] used an in vivo rat model of hydrochloric acid instillation-induced lung injury to examine the hypothesis that MV with a relatively high VT (16 ml/kg) and/or low PEEP would increase the release of TNF-α and macrophage inflammatory protein 2 into the systemic circulation. Rats were mechanically ventilated for 4 h immediately after the intratracheal instillation of the acid. They found that all animals ventilated with low VT plus zero PEEP died before the end of the experiment. Since these authors did not perform a postmortem pathological examination, it is plausible that this fatal outcome was due to the fact that in that model the aspiration of acid into the airways causes severe pulmonary edema and a rapid deterioration in alveolar gas exchange.

In our study only one out of four septic animals ventilated with a similar ventilatory strategy died, and none of the MV strategies was associated with inevitable fatal outcome. However, as in our study, Chimello et al. [14] found that the increased production of cytokines in the lungs was followed by elevated serum cytokine levels, and the highest serum levels were found in the high-VT/0 PEEP group. Although we found that the application of PEEP, independently of VT, decreased the serum concentration of TNF-α, the serum concentration of IL-6 was not attenuated by PEEP in the high VT group. Ranieri et al. [29] randomized 37 patients with ARDS to either MV with conventional VT plus PEEP or low VT plus high PEEP, and measured pulmonary and systemic concentrations of proinflammatory mediators on the second day after randomization. They reported that serum concentrations of cytokines were significantly lower in the low-VT/high PEEP group. However, the sample size and the study design were not intended to assess whether those changes were associated with improvements in outcome. In our study we have now demonstrated that modulation of inflammatory responses by MV has in impact on animal outcome, suggesting that ventilator-induced lung injury can translate into ventilator-induced death [31].

These three studies [12, 14, 30] have been challenged recently by Ricard et al. [21] who found that MV-induced severe lung injury but did not lead to the release of significant amounts of TNF-α or IL-1β by the lung. They found that high VT ventilation per se did not cause a significant release of either cytokine into the airspaces of isolated rat lungs, and they were unable to detect these cytokines in the airspaces or the systemic circulation of animals after 2 h of in vivo MV with low (7 ml/kg) or high (42 ml/kg) VT. Although the study design of Ricard et al. [21] is different from ours, we were unable to reproduce their observations. By contrast, our findings corroborate that cytokines not only participate in the development of ventilator-induced lung injury in previously damaged lungs but also contribute to the propagation of a systemic inflammatory response leading to worsening outcome. In summary, our study demonstrates a direct effect of MV on the immune response in the lung and in the whole septic animal. Additional therapeutic interventions directed towards alterations in cytokine phenotipic profiles may prove beneficial in the treatment of ALI and ARDS.