Increased Susceptibility to Ventilator-associated Lung Injury Persists after Clinical Recovery from Experimental Endotoxemia

This study was approved by the institutional and local Committee on the Care and Use of Animals (Thüringer Landesamt für Lebensmittelsicherheit und Verbraucherschutz, Weimar, Germany). Male Wistar rats weighing 350–380 g were used for this study.

Animals were anesthetized with 1.5% isoflurane in 100% oxygen via  a face–shoulder mask. Tail artery blood (200 μl) was collected to obtain baseline blood gas analysis (ABL 50; Radiometer, Copenhagen, Denmark), blood leukocyte counts, and interleukin 6 (IL-6) concentration. Subsequently, 1 mg endotoxin (lipopolysaccharide, serotype Escherichia coli  O55:B5) dissolved in 1 ml phosphate-buffered saline (PBS) or 1 ml PBS (control group) was injected intravenously in a tail vein. This dose of endotoxin had resulted in transient clinical illness and rapid recovery but no mortality at 48 h in preliminary experiments by our group. Anesthesia was discontinued, and the animals were transferred to their cages with free access to water and food. After 24 h, control and endotoxin-treated animals were randomly assigned to undergo no MV or MV with V^T8, 24, 27, or 30 ml/kg body weight (n = 10/group). A range of high V^Twas chosen to be able to identify gradual pulmonary deterioration (dose–response relation) and to maximize lung injury without causing mortality. Selection of the highest V^Twas based on preliminary experiments demonstrating that severe lung injury and hypoxemia occurred in healthy (PBS-injected) rats at V^T30 ml/kg.

Inhalation anesthesia was repeated at 6 and 24 h after treatment to obtain blood samples for arterial blood gas analysis, leukocyte counts, and IL-6 levels. After drawing the 24-h blood sample, ketamine (70 mg) and midazolam (4 mg) were injected intraperitoneally, resulting in deep anesthesia. The abdomen was opened, and animals were exsanguinated by rapid collection of 8–10 ml blood from the inferior vena cava. Sternotomy was performed, and the lungs were removed en bloc . Further measurements and sample collection were performed identically as in ventilated animals.

Twenty-four hours after treatment, animals were anesthetized using an intraperitoneal injection of ketamine (70 mg) and midazolam (4 mg). Cefuroxime, 12.5 mg, was injected subcutaneously for antibiotic prophylaxis (Cefuroxim-ratiopharm® 250 mg; Ratiopharm GmbH, Ulm, Germany). The right external jugular vein and the right carotid artery were surgically exposed and polyethylene catheters (ID 0.5 mm) were placed in each vessel and secured with ligatures. A venous blood sample was drawn for leukocyte counts. Pressure transducers were connected to each catheter, and central venous and arterial blood pressures were displayed on a monitor continuously (SMU 611; PPG Hellige GmbH, Freiburg, Germany). Infusion of balanced electrolyte solutions at a rate of 10 ml/h (Thomaejonin; Delta Pharma GmbH, Pfullingen Germany) and intravenous anesthetics (18 mg/h ketamine and 0.18 mg/h midazolam) at a rate of 0.6 ml/h was started and maintained throughout the ventilation period. The trachea was exposed and incised, and a blunt cannula (OD 2 mm) was placed in the tracheal lumen. Skin incisions were closed with metal clips.

Mechanical ventilation was started (respiratory rate, 40 breaths/min; inspiratory-to-expiratory ratio, 1:1; fraction of inspired oxygen, 0.4; positive end-expiratory pressure, 4 cm H²O; Animal Ventilator CIV 101; Columbus Instruments, Columbus, OH), and V^Twas adjusted according to group allocation. Carbon dioxide was added to the inspiratory gas if high V^Twas applied, to attain arterial normocapnia at the beginning of the ventilation period. No further corrections of carbon dioxide were made throughout the ventilation period. A pressure transducer was connected to the tracheal cannula, and airway pressures were displayed on a monitor (SMU 611). Pancuronium bromide, 0.2 mg, was injected intravenously for muscle relaxation. Body temperature was maintained at 38°C using a feedback warming system (Theraterm de Luxe, 250W; Osram, Munich, Germany; with HSE temperature regulator; Hugo Sachs Elektronic, March, Germany). Heart rate, arterial and central venous blood pressures, peak airway pressure, and arterial blood gases were recorded every 30 min throughout the ventilation period. After 4 h of MV, animals were killed, and lungs were harvested as described above (see Experiments and Measurements in Nonventilated Animals).

Using blunt clamps, a small part of the lower lobe of the right (in nonventilated animals, also of the left) lung was excluded from the lavage procedure. Lung lavage was performed with 10 ml PBS using 2.5-ml aliquots. The pooled effluent (7.5–8.5 ml) was centrifuged at 3,000 rpm for 10 min. The cell pellet was resuspended in 100 μl PBS, and absolute numbers of neutrophils and macrophages were counted using a hemocytometer and corrected for the total amount of lavage fluid obtained. The supernatant of the lavage fluid was stored at −70C° until cytokine levels and protein concentration were determined.

Protein content in the supernatant of the lung lavage fluid was determined using turbidimetry (Roche Diagnostics, Mannheim, Germany; analyzer: Hitachi 717) with a detection threshold of 60 mg/l.

Total leukocyte counts were determined using a hemocytometer. The percentage of blood neutrophils was obtained from a cell smear (May-Gruenwald stain; 100 leukocytes were counted at a magnification of 500×), and absolute neutrophil numbers (cells/μl) were calculated using the corresponding total leukocyte counts. The supernatant from centrifuged (4,000 rpm for 10 min) blood samples was stored at −70°C until IL-6 assays were performed.

Levels of IL-6 in plasma and of IL-6, interleukin-10 (IL-10), and macrophage inflammatory protein 2 (MIP-2) in the supernatant of lung were assessed by enzyme-linked immunosorbent assay using commercially available kits specific for rats (IL-6: KRC0062; MIP-2: KRC1022; IL-10: KRC0101; BioSource, Solingen, Germany). Measurements were made according to manufacturer guidelines. The detection thresholds were 8 pg/ml for IL-6, 1 pg/ml for MIP-2, and 5 pg/ml for IL-10.

The part of the right lung that had been excluded from the lavage procedure was weighed, desiccated (48°C for 48 h), and weighed again. The wet-to-dry weight (W/D) ratio was calculated and normalized to the ratio in nonventilated control animals.

The nonlavaged part of the left lung that had been obtained from nonventilated animals was fixed in 10% buffered formalin, and hematoxylin and eosin–stained slides were prepared for light microscopy. The average number of neutrophils from 10 randomly selected (excluding conducting airways and blood vessels) high-power fields (magnification: 500×) in each slide was recorded.

Animals were excluded from data analysis if they died before the end of the period of MV. Because mortality was high in animals ventilated with V^T30 ml/kg, this group was stopped after 11 animals (6 endotoxin, 5 controls) had been enrolled.

Data are presented as mean ± SEM. Two-way analysis of variance was used to assess significant effects of treatment (endotoxin vs.  control) and V^Ton variables. A repeated-measures term was used where appropriate and Bonferroni post hoc  correction was made in case of multiple comparisons. A t  test was used to assess differences between two groups. P < 0.05 was considered statistically significant. The statistics software SPSS Version 11 (Cary, NC) was used for data analysis.

Endotoxin-treated animals exhibited symptoms of illness (tachypnea, reduced locomotion, piloerection) for 4–6 h but appeared unaffected at 12 and 24 h after the injection.

Respiratory variables are shown in table 1. Oxygenation at baseline and at 6 and 24 h after treatment was comparable in both groups. In endotoxin-treated animals, arterial carbon dioxide tension was lower at 6 and 24 h after treatment as compared with baseline and as compared with controls. There were no differences in arterial pH between groups at any time point.

Markers of systemic inflammation are shown in table 1. Total leukocyte counts were significantly lower at 6 h and significantly higher at 24 h after endotoxin injection compared with baseline. No significant changes in total leukocyte counts were seen in control animals. Blood neutrophil counts in endotoxin-treated animals increased at 6 and 24 h after treatment. In control animals, blood neutrophil counts were significantly higher at 6 but not at 24 h compared with baseline. IL-6 levels in blood were below the detection threshold before endotoxin treatment in all but one animal, measurable in all animals (mean ± SEM: 638.6 ± 221 pg/ml) at 6 h, and below the detection threshold again in all but one animal at 24 h after endotoxin injection (the same animal that had detectable IL-6 before the injection). IL-6 was not detectable in most of the control animals at any time. Low levels were found in three control animals before PBS injection and did not change at 6 or 24 h after injection.

There were no gross differences in lung histology between groups (fig. 1), but neutrophil numbers in lung tissue (12.9 ± 1 vs.  5.0 ± 1 cells/high-power field; P < 0.001) were significantly higher in endotoxin-treated than in control animals. Lung W/D ratio (fig. 2) and macrophage count and protein content in lung lavage fluid (fig. 3) were not different between endotoxin-treated and control animals, but neutrophil counts in lung lavage fluid were significantly higher in endotoxin-treated animals (fig. 3). MIP-2 was detectable in lung lavage fluid in all control and endotoxin-treated animals, and levels were not different between groups (fig. 4). IL-6 levels in lung lavage were below the detection threshold in six of nine endotoxin-treated animals and in eight of nine control animals and low (< 55 pg/ml) in the remaining four animals. IL-10 in lung lavage fluid was below the detection threshold in seven of nine endotoxin-treated animals and eight of nine control animals, and less than 160 pg/ml in the remaining three animals.

No animal died in the groups with V^T8 ml/kg. One control and two endotoxin-treated animals in the V^T24 ml/kg groups and two endotoxin-treated and two control animals in the V^T27 ml/kg groups became progressively hypotensive and died. In the groups ventilated with V^T30 ml/kg, five of six endotoxin-treated and one of five control animals died. All animals ventilated with V^T30 ml/kg had development of severe pulmonary edema and became progressively hypotensive. Because of the low number of surviving animals ventilated with V^T30 ml/kg, no further parameters and no statistics are reported from these groups.

In endotoxin-treated but not in control animals ventilated with V^T27 ml/kg, arterial oxygen tension decreased significantly at the end of the MV period. Arterial oxygen tension remained unchanged over time in all other groups and did not differ between endotoxin-treated and control animals (fig. 5). Further hemodynamic and respiratory variables at the beginning and at the end of MV are shown in table 2. Heart rate and arterial pressure decreased significantly over time in all groups regardless of V^Tand pretreatment. Peak airway pressure at the beginning and at the end of the ventilation period was comparable in both groups ventilated with V^T8 ml/kg but was significantly higher in endotoxin-treated animals ventilated with 24 or 27 ml/kg as compared with the respective control groups. Arterial carbon dioxide tension was higher and pH was lower in both groups ventilated with V^T8 ml/kg as compared with animals ventilated with 24 or 27 ml/kg, but there were no differences regarding these variables between endotoxin-treated and control groups at the same V^T.

Regardless of pretreatment, most animals ventilated with V^T27 ml/kg had development of pulmonary edema at the end of the ventilation period, with frothy fluid filling the airways and tracheal cannula. During MV, edema fluid was removed if it extended beyond the tracheal cannula into the ventilator tubes (eight animals). Quantification of neutrophils and macrophages in the removed fluid was attempted but was successful only in four endotoxin-treated animals. Cell counts were fourfold to eightfold higher in edema than in lavage fluid.

Lung W/D ratio in both groups of animals ventilated with low V^Twas comparable to that of nonventilated animals (fig. 2). As compared with nonventilated animals, ventilation with V^T24 and 27 ml/kg significantly increased lung W/D ratio in endotoxin-treated animals (fig. 2). In control animals, lung W/D ratio increased significantly with V^T27 ml/kg only (fig. 2).

In endotoxin-treated animals, lung lavage protein content was comparable with that in nonventilated animals after MV with V^T8 ml/kg but increased significantly after MV with 24 and 27 ml/kg (fig. 3A). In control animals, lung lavage protein increased significantly only in animals ventilated with V^T27 ml/kg (fig. 3A).

Neutrophils in lung lavage fluid at any V^Twere significantly higher in endotoxin-treated animals than in control animals (fig. 3B). In endotoxin-treated animals ventilated with 8 or 24 but not with 27 ml/kg, lung lavage neutrophil counts were significantly higher than in nonventilated animals (fig. 3B). No significant differences in lung lavage neutrophil counts were seen in control animals (fig. 3B). Irrespective of pretreatment or V^T, there was no significant change in macrophage counts in lung lavage fluid (fig. 3C).

Interleukin 6 and MIP-2 levels are shown in figure 4.

Interleukin-6 levels in lung lavage fluid at low V^Twere comparable between endotoxin-treated and control animals. High V^Tresulted in increased IL-6 levels in both groups, but more so in endotoxin-treated animals.

Macrophage inflammatory protein-2 levels in lung lavage fluid were significantly higher in endotoxin-treated than in control animals at corresponding V^T. In endotoxin-treated animals, MIP-2 at each V^Twas higher than in nonventilated animals. In control animals, only MIP-2 levels in animals ventilated with V^T24 or 27 ml/kg were higher than in nonventilated animals.

In control animals, IL-10 levels were detectable only in one or two animals at each V^T. In endotoxin-treated animals, IL-10 was detectable in one animal in the 3-ml and in three animals in the 24-ml group. All values were less than 200 pg/ml. No statistical analysis was performed in these groups. IL-10 was detected in all but one animal in the 27-ml/kg group (mean ± SEM: 95.4 ± 29 pg/ml).

Blood neutrophil counts before MV were significantly higher in endotoxin-treated animals than in control animals (fig. 6). There was a significant increase in blood neutrophil counts at the end of MV in all V^Tgroups in control animals, whereas a significant decrease was seen in endotoxin-treated animals ventilated with V^T27 and 30 ml/kg (fig. 6).

We studied the effects of MV with increasing V^Ts on pulmonary inflammation and lung injury in animals after recovery from an endotoxin challenge. By establishing a 24-h delay between the endotoxin injection and the ventilation period, our study represents a two-hit model of lung injury. Our study thus differs substantially from previous studies,9,10,13which exposed animals to artificial ventilation and endotoxin simultaneously, thereby creating a situation where the injurious effects of MV occur during the maximal acute systemic inflammatory response to endotoxin. We established a range of V^Ts, which enabled us to identify not only gradual increases in the severity of lung injury but also subtle differences between lipopolysaccharide-treated and control animals. In addition, we included a group of animals ventilated with low V^Tand positive end-expiratory pressure, thereby simulating ventilator settings close to recommended clinical practice in preinjured lungs.14We found an increased susceptibility to ventilator-associated lung injury despite clinical recovery from a single injection of endotoxin. The detrimental effects of increasing V^Ts on the lungs of endotoxin-pretreated as compared with control animals were characterized by an increase in cytokine release and neutrophil infiltration. This inflammatory injury was accompanied by functional pulmonary impairment and by increased mortality in animals ventilated with high V^T. However, even a low V^Tapplied 24 h after endotoxin exposure induced a detectable pulmonary inflammatory response.

We aimed to expose the animals to an only moderate initial injury. Therefore, the endotoxin dose used in our study was adjusted to result in 100% survival at 48 h. Results from a comprehensive murine study demonstrated that survival at 24 h after a systemic endotoxin challenge always translated to long-term survival.12We saw only transient clinical deterioration in our study, and 24 h after endotoxemia (before MV), blood and lung cytokine levels and oxygenation were comparable between the endotoxin and control groups. In endotoxin-treated animals, arterial carbon dioxide levels at 6 and 24 h after treatment were lower as compared with baseline and as compared with control animals. This suggests that residual effects of endotoxin on pulmonary function were still present; however, this finding was not reflected by overt tachypnea or other clinical symptoms of respiratory impairment. In addition, comparable hemodynamics in endotoxin-treated and control animals during MV support our assumption of constitutional recovery from the endotoxin exposure.

Twenty-four hours after the endotoxin challenge, blood and lung lavage neutrophil counts were significantly higher in the endotoxin group as compared with the control group. In previous studies, maximum neutrophil infiltration in the lung interstitium has been found at 3–6 h after a systemic endotoxin challenge, whereas neutrophil counts in lung lavage fluid peaked at 24 h.11,15Given this time course, our findings likely reflect residual pulmonary infiltration, rather than ongoing neutrophil recruitment to the lungs. In our study, the higher baseline level of activated neutrophils in the circulation and lungs may have predisposed these animals to an enhanced inflammatory response to MV.

During ventilation, blood neutrophil counts in endotoxin-pretreated animals decreased during ventilation (with increasing V^T) while lung lavage neutrophils increased. In control animals, blood neutrophil levels increased during MV, but lung lavage neutrophil counts did not. This suggests that in the endotoxin group, previous neutrophilia and neutrophil activation led to its increased sequestration in the lung, an effect not seen in the control group.

During ventilation, lung lavage cytokine levels (MIP-2, IL-6) increased with increasing V^Twith or without endotoxin pretreatment. The occurrence and relevance of increased pulmonary cytokine levels in the setting of MV alone has been a matter of debate.16Our results regarding IL-6 and MIP-2 within the context of moderate increases in neutrophil counts in control animals support the view of a proinflammatory pulmonary reaction with increasing V^T. Furthermore, measurable IL-10 levels in lung lavage fluid in endotoxin-treated animals ventilated with V^T27 ml/kg suggest a simultaneous antiinflammatory response in this group. The exaggerated increase in lung cytokine levels in the endotoxin group suggest higher production rates due to previous activation of neutrophils and, possibly, macrophages in this group.

Irrespective of pretreatment and V^T, MV did not significantly affect alveolar macrophage counts. Interestingly, the lowest individual counts were seen in animals ventilated with the highest V^T. Consistent with our results, other groups reported decreased alveolar macrophage counts in rats submitted to MV with high V^Tin both ex vivo  and in vivo  experiments.17,18Extraalveolar relocation of macrophages was discussed as a possible reason, but explanations remained speculative.17,19In our study, a technical aspect might partially explain the macrophage numbers: Most of the animals in the V^T27 ml/kg groups developed severe pulmonary edema, and a variable amount of edema fluid (as much as 2 ml in some animals) was lost before lung lavage. Because cell counts were considerably higher in edema than in lavage fluid, this loss may have resulted in underestimation of inflammatory markers in lavage fluid. The fact that neutrophil counts and MIP-2 levels in lung lavage fluid, irrespective of pretreatment, did not increase further with the highest V^Tsupports this hypothesis.

In general, our animal model exemplifies the two-hit theory in the pathogenesis of organ injury. According to this theory, the first intervention (hit) primes the organism for an exaggerated response to a second intervention (hit). Endotoxin is frequently used as a first intervention in these models. In our study, endotoxin and ventilation are the two interventions that were applied in a clinically relevant time-related fashion.

There are a number of similarities between our study (delayed MV after endotoxemia) and previous studies in which endotoxin was mostly immediately followed by MV. In one study, endotoxin alone did not impair oxygenation but caused a significant decrease in blood neutrophil counts and an increase in lung lavage neutrophils in nonventilated animals at 4 h, indicating a pulmonary inflammatory response similar to that in our study.9Endotoxin with immediate ventilation with a moderate V^Tof 10 ml/kg resulted in aggravated lung injury as compared with either MV or endotoxin alone. In contrast to our findings with a V^Tof 8 ml/kg, the V^Tof 10 ml/kg also caused functional pulmonary impairment, suggesting that pulmonary susceptibility to detrimental effects of MV may be higher in the early phase of the endotoxin response, when pulmonary permeability changes are most pronounced. In another study, rabbits received a low dose of endotoxin intravenously while being ventilated with V^T15 ml/kg without positive end-expiratory pressure.13In accord with our findings, combination of endotoxin and MV aggravated the pulmonary release of inflammatory mediators as compared with MV or endotoxin treatment alone.

Our results, in keeping with the two-hit theory of experimental organ injury, are clinically relevant. Transient endotoxemia occurs in various clinical scenarios, including major surgery,3–5trauma,20,21endoscopic procedures,22,23ischemia–reperfusion,24and after enteral refeeding.25Residual subclinical pulmonary alterations may persist after these procedures and may leave these patients at risk for ventilator-associated injury. While clearly recommended in patients with acute respiratory distress syndrome,14recent data indicate that the use of low V^Tin patients with presumably healthy lungs is less well established.26Therefore, these patients are at increased risk of being exposed to a potentially harmful ventilation strategy.

When translating our findings to clinical scenarios, some limitations of our study must be mentioned. First, to obtain measurable pulmonary effects within a short period of time, we applied V^Ts far beyond settings used in clinical practice. One may hypothesize, however, that detrimental pulmonary effects may also be provoked by less excessive V^Tand a longer duration of MV. Second, to provoke clinical effects in rodents, much higher endotoxin doses are required than in human volunteers.1Nevertheless, despite differences in sensitivity, responses to endotoxin exposure seem to be qualitatively similar in different species.27Third, although increased alveolar neutrophil counts have been shown in humans after bronchial endotoxin instillation,28we are missing from clinical studies clear evidence that neutrophil translocation into the lung interstitium and alveolar space occurs after systemic exposure to endotoxin. However, basic pulmonary reactions to systemic endotoxin in humans resemble effects observed in animals,29and increased adhesion and accumulation of neutrophils in the pulmonary vasculature, a process commonly occurring before extravasation, has been demonstrated in human endotoxemia.30Therefore, our findings likely maintain clinical relevance. Last, our method of determining lung W/D ratio does not take in account residual intravascular blood in the lung. Therefore, results may have been affected by different amounts of lung perfusion secondary to different inflammatory states of the lung. However, because animals were exsanguinated before lung harvest (thereby reducing the amount of residual blood in the pulmonary circulation) and because of the collaborative finding of frank edema fluid in the proximal airways of animals ventilated with high V^T, we assume that this limitation was of minor relevance and that our findings regarding lung W/D ratio truly reflect pulmonary edema.

In conclusion, in this experimental study, a transient period of endotoxemia with subsequent clinical recovery before initiation of MV resulted in increased susceptibility to the deleterious effects of increasing V^T. Our data provide indirect evidence that pulmonary neutrophil infiltration before MV predisposes to these pulmonary alterations. Our results further suggest that it may be advisable to avoid high V^Tin subjects with presumably healthy lungs even after clinical recovery from a period of endotoxemia.