Neutrophil Stimulation with Granulocyte Colony-stimulating Factor Worsens Ventilator-induced Lung Injury and Mortality in Rats

The experimental protocol for this study was approved by the local Animal Protection Committee (Regierungsprasidium Weimar, Germany). Animals were kept in groups of three or four and were allowed food and water before beginning the study.

After treatment with G-CSF, fucoidin, or placebo (control), anesthetized and paralyzed male Wistar rats weighing between 350 and 400 g were mechanically ventilated for 4 h with either low (10 ml/kg) or high (40 ml/kg) tidal volumes. Immediately before initiation of mechanical ventilation, animals had central venous, carotid arterial, and tracheostomy catheters placed. The following measures were obtained during mechanical ventilation: systolic arterial pressure, heart rate, arterial blood gases, and mean airway pressures at baseline and 30-min intervals; complete blood counts at the beginning (0 h) and immediately before the completion of mechanical ventilation (4 h); and serum IL-6 and TNF concentrations at 4 h only. After measures at 4 h, animals were killed, and single lung lavage was performed for cell, protein, IL-6, and TNF determinations while the other lung was removed for histologic assessment of lung edema and alveolar neutrophils. Animals that died during the 4-h period had blood removed for complete blood counts, and lung lavage and histology were performed immediately at the time of death.

In initial experiments, assignment cards blindly selected for each experiment were used to randomly assign animals to a combination of treatment (placebo, fucoidin, or G-CSF) and tidal volume (10 or 40 ml/kg). Placebo, fucoidin, and G-CSF, respectively, were administered to 19, 9, and 21 animals with low tidal volumes and 20, 20, and 22 with high tidal volumes. Overall, fewer animals were assigned to treatment with fucoidin and low tidal volumes because of limited availability of the study agent.

Both G-CSF and fucoidin were diluted with phosphate-buffered saline (PBS). Animals were given G-CSF (100 μg · kg⁻¹· day⁻¹subcutaneously; Amgen, Munich, Germany) 48 and 24 h before mechanical ventilation; fucoidin (4 mg followed by 6 mg · kg⁻¹· h⁻¹intraperitoneally; Sigma, Deisenhofen, Germany) started 2 h before and continued until the completion of mechanical ventilation; or, as placebo (control), PBS. Doses of G-CSF and fucoidin chosen for study were based on previously published studies and pilot experiments in rats.16,19Control animals received PBS subcutaneously and intraperitoneally at time points simulating the scheduled administration of G-CSF and fucoidin, respectively. Animals receiving G-CSF were treated with intraperitoneal PBS at time points to simulate the scheduled fucoidin administration, whereas animals receiving fucoidin were treated with subcutaneous PBS to simulate the scheduled G-CSF administration. Therefore, all animals received equivalent volumes of drug or diluent based on a similar schedule. Although an aseptic technique was used for all procedures, to further prevent early infection related to skin or airway contamination, all animals received a single dose of cefuroxime (12.5 mg intramuscularly) immediately before mechanical ventilation.

Animals were initially anesthetized with intraperitoneal ketamine (20 mg/kg; Ayarost GmbH and Co., Twistringen, Germany) and midazolam (4 mg/kg; Hoffman-LaRoche AG, Grenzach-Whylen, Germany). Their necks were shaved and disinfected, and a 2-cm skin incision was made parallel to the trachea. The right external jugular and carotid artery were cannulated with plastic catheters (Primed Medizintechnik GmbH, Halberstedt, Germany). The trachea was exposed and cannulated with a blunt metal 16-gauge needle positioned with its tip above the tracheal bifurcation. The needle was secured, and the neck incision was sutured closed. Ventilation was started to ensure that no air leaks were present.

After catheter placement and initiation of mechanical ventilation, animals received intravenous loading doses followed by continuous infusions of ketamine (50 mg · kg⁻¹· h⁻¹) and midazolam (0.5 mg · kg⁻¹· h⁻¹). Pancuronium was administered hourly (0.1 mg · kg⁻¹· h⁻¹; Curamed Pharma GmbH, Karlsruhe, Germany).

Using a volume-controlled small-animal ventilator (Technical & Scientific Equipments GmbH, Bad Homburg, Germany), rats were ventilated for 4 h with either low (10 ml/kg) or high (40 ml/kg) tidal volumes, all at a rate of 40 breaths/min, with a positive end expiratory pressure of 5 cm H²O. These tidal volumes were chosen based on pilot studies showing that for the 4-h period of mechanical ventilation to be investigated, the low one, which generates peak inspiratory pressures of 9–10 mmHg (12–14 cm H²O), was well tolerated, whereas the high one, which produces peak inspiratory pressures of 29–31 mmHg (41–43 cm H²O), was associated with lethality. For the purpose of changing only one variable related to minute ventilation, tidal volumes but not respiratory rates were altered between study groups. However, to achieve comparable partial alveolar and arterial oxygen and carbon dioxide pressures at baseline, animals receiving low tidal volume were ventilated with a fractional inspired oxygen concentration of 0.37 and a fractional inspired carbon dioxide concentration of 0, whereas animals receiving high tidal volume were ventilated with a fractional inspired oxygen concentration of 0.30 and an fractional inspired carbon dioxide concentration of 0.05 to produce an arterial carbon dioxide concentration of 40 mmHg.

After placement, catheters were connected to pressure transducers for arterial pressure and heart rate determinations (SMU 611; Hellige, Freiburg, Germany). Mean airway pressures were transduced and measured (SMU 611). Arterial blood was removed for blood gas determination (ABL 50; Radiometer, Copenhagen, Denmark), and venous blood was removed for complete blood counts and differentials and serum TNF-α and IL-6 determinations. Lung lavage fluid was centrifuged, and the cell pellet was resuspended in PBS. The number of cells in PBS was counted with a hemocytometer, and a differential count was determined on a smear using Wright stain. Lung lavage protein concentrations (Clinical System LX20; Beckman Synchron, Krefeld, Germany), and cytokine concentrations were determined on the supernatant. For histologic assessment of lung edema and neutrophils, an upper lobe bronchus was cannulated, and the lobe was infused and fixed with formaldehyde (4%) (Mallinkrodt Specialty Chemicals, Paris, KY), dehydrated, embedded in paraffin, and sectioned at 6 μm. Lung tissue sections were then analyzed by an observer blinded to the study groups as previously described.19Lung edema was scored on a scale of 0 (no edema evident) to 4 (maximum edema). Edema was characterized primarily by the presence of dilated alveolar spaces filled with amorphous and lightly acidophilic material (pink fluid), congestion of the alveolar walls, and dissociation of perivascular tissue generally associated with hemorrhagic foci. Alveolar neutrophils were enumerated by randomly examining 10 fields/slide at a magnification of ×480 (oil immersion). Fields with large vessels were not evaluated. To measure rat serum and lung lavage TNF-α, we incubated 96-well plates with a 0.5% solution of rabbit anti-rat TNF-α serum. After 24 h of incubation at 4°C and washing, standard or test sera or lavage fluid was added, and the plates were incubated for 4 h at room temperature. After washing, a 1-μg/ml concentrated biotin-conjugated anti-μ rat antibody (Pharmingen, San Diego, CA) was added. After repeated washing, 0.5 μg/ml concentrated streptavidin–peroxidase conjugate solution (Dianova, Hamburg, Germany) was added. After an additional 30 min, the wells were washed, and the peroxidase substrate 3,3′,5,5′-tetramethylbenzidine (Sigma, Deisenhofen, Germany) was added. The reaction was stopped after 10–15 min with 1 m H²SO⁴, and the absorption of the solution was read at 450 nm. Serum and lavage IL-6 concentrations were performed using a commercial assay (Biosource, Camarrilo, CA).

For the high-ventilation group, survival was analyzed using a Wilcoxon model with two one-degree-of-freedom tests (fucoidin vs.  control and G-CSF vs.  control). All other laboratory parameters were analyzed with an analysis of variance (ANOVA) taking into account time of measurement, level of mechanical ventilation, and treatment. Where either the drug main effect or interaction terms involving drug were significant, a Tukey test was performed. For mean airway pressure and arterial oxygen and carbon dioxide pressures, the effect of treatment did not differ over the 4 h of study (P = not significant [NS] for all ANOVA), so the mean value across time for each animal was analyzed.20Measures from all animals were included in analysis up until the time of their death. Numbers of animals used for individual measures are shown on figures or in tables. Changes in oxygenation with high tidal volumes were also analyzed over the first 2 h of mechanical ventilation, when most animals were still present for analysis. A P  value 0.05 or less was considered significant. In figures, P  values are only shown for changes from control that reached significance.

All animals survived during mechanical ventilation with low tidal volumes (10 ml/kg) (19 control, 9 fucoidin, and 21 G-CSF animals). In contrast, mechanical ventilation with high tidal volumes (40 ml/kg) caused lethality in each of the three treatment groups (fig. 1). In controls with high tidal volumes, there were 3 nonsurvivors and 17 survivors (P = 0.09 compared with controls receiving low tidal volumes). Compared with these high-tidal-volume controls, fucoidin did not alter survival significantly (5 nonsurvivors and 15 survivors; P = 0.51 compared with controls, Wilcoxon), but G-CSF significantly worsened it (18 nonsurvivors and 4 survivors; P = 0.001 compared with controls, Wilcoxon).

In controls, circulating neutrophils at the beginning and end of mechanical ventilation did not differ comparing low and high tidal volumes (P = NS, ANOVA; fig. 2). Compared with controls, fucoidin with both low and high tidal volumes increased circulating neutrophils significantly at the end (P = 0.002 and P < 0.0001, respectively, ANOVA) but not the beginning (P = NS for both, ANOVA) of ventilation. In contrast, compared with controls, G-CSF with both low and high tidal volumes increased circulating neutrophils significantly at the beginning (P < 0.0001 for both, ANOVA) but not the end (P = NS for both) of ventilation.

At the end of mechanical ventilation in controls, lung lavage and histology neutrophils were increased with high compared with low tidal volumes (P = 0.007 and P = 0.01 for lavage and histology neutrophils, respectively, ANOVA; figs. 3A and B). Compared with controls, fucoidin decreased lung lavage and histology neutrophils with both low and high tidal volumes, but these differences were only significant for lavage neutrophils with high tidal volumes (P = 0.003, ANOVA). In contrast, compared with controls, G-CSF increased lung lavage and histology neutrophils with both tidal volumes, but these differences were only significant at low tidal volumes (P = 0.004 and P = 0.01 for lung lavage and histology neutrophils, respectively, ANOVA).

In all treatment groups (control, fucoidin, and G-CSF) from 0 to 4 h, systolic arterial blood pressure (data not shown) and heart rate (data not shown) with low and high tidal volumes and arterial oxygen and mean airway pressures with high tidal volumes decreased (P ≤ 0.05 for all for the effect of time on changes in these parameters, ANOVA; table 1). During this time period in all treatment groups, compared with low tidal volumes, high ones were associated with increased mean airway pressures (P = 0.0001) and decreased arterial oxygen (P = 0.04), arterial carbon dioxide (P = 0.002), systolic arterial blood pressure (P = 0.0001), and heart rate (P = 0.0001, ANOVA for all).

Compared with controls, fucoidin did alter any cardiopulmonary parameter significantly with either low or high tidal volumes (P = NS for all, ANOVA). Compared with controls, from 0 to 4 h, G-CSF with low tidal volumes did not alter airway pressures (P = NS, ANOVA). In contrast, compared with controls, from 0 to 4 h, G-CSF with high tidal volumes was associated with significant increases in airway pressure (P < 0.0001 for the effect of G-CSF averaged across time, ANOVA; table 1and fig. 4). Compared with controls, G-CSF from 0 to 4 h decreased arterial oxygen with low and high tidal volumes, but this was only significant with the former (P < 0.0001 averaged across time, ANOVA) and not the latter (P = NS, ANOVA) tidal volume (fig. 4A). However, from 0 to 2 h, when most animals receiving high tidal volumes were still present for study, compared with controls, G-CSF was associated with significant reductions in arterial oxygen with both tidal volumes (P = 0.02 for each averaged across time, ANOVA; fig. 5). Compared with controls, G-CSF from 0 to 4 h did not significantly alter systolic arterial blood pressure, heart rate, or arterial carbon dioxide with either tidal volume (P = NS for all, ANOVA).

At the end of mechanical ventilation in controls, compared with low tidal volumes, high ones were associated with significant increases in alveolar edema on lung histology (P < 0.0001, ANOVA; fig. 4B). Compared with controls, fucoidin did alter alveolar edema with either low or high tidal volumes (P = NS for both, ANOVA). Compared with controls, however, G-CSF was associated with significant increases in alveolar edema with both low and high tidal volumes (P < 0.0001 and P = 0.0006 respectively, ANOVA).

At the end of mechanical ventilation in controls, compared with low tidal volumes, high ones were associated with significant increases in lung lavage protein and IL-6 concentrations (P = 0.02 and P < 0.0001 respectively, ANOVA), but other parameters did not differ significantly (table 2). Compared with controls, fucoidin with high tidal volumes significantly reduced lung lavage IL-6 concentrations (P < 0.0001, ANOVA). In contrast, compared with controls, G-CSF significantly increased IL-6 concentrations in blood with low and high tidal volumes (P = 0.01 and P = 0.02 respectively, ANOVA) and in lung lavage with high tidal volumes (P = 0.007, ANOVA). Other protein and cytokine measures did not differ significantly comparing either tidal volume or treatments (P = NS for all, ANOVA). Other laboratory parameters did not differ comparing either high or low tidal volumes or with either treatment (P = NS for all, ANOVA; table 2).

Different from some previous studies that used lung lavage with surfactant depletion in models of VILI, the current one investigated the effects of neutrophil inhibition in a model of VILI related to increases in tidal volumes alone.13,14,21In this study, fucoidin seemed to inhibit lung neutrophil–endothelial adhesion and extravascular migration as evidenced by increases in circulating neutrophils over the time of mechanical ventilation in association with reductions in lung lavage neutrophils and IL-6 concentrations. Despite these changes, fucoidin did not significantly alter lung injury or prevent lethality associated with increased tidal volumes. Absence of a beneficial survival effect with fucoidin is unlikely related to insufficient sample size because this treatment actually increased mortality rates compared with controls. Analysis showed that more than 400 animals would be required divided between placebo and fucoidin groups to demonstrate a significant reversal in the current survival trend with the latter treatment. Similar to the current findings, we observed before that leukocyte integrin adhesion molecule inhibition reduced lung neutrophil recruitment but not mortality rates with lethal hyperoxia.19One possible but not conclusive interpretation of these findings with fucoidin is that when unstimulated, selectin-mediated neutrophil recruitment may not have a primary role in lung injury in this model. Alternatively, non–selectin-mediated neutrophil events or mediators independent of the neutrophil may contribute to this injury and the reduced survival rates associated with high tidal volumes in this rat model.21–26We cannot exclude the possibility, however, that longer treatment with fucoidin or use of higher doses may have resulted in beneficial effects on either lung injury or survival rates.

In contrast to the negligible effect fucoidin had on outcome, G-CSF treatment sufficient to increase lung neutrophils and both lung and serum IL-6 concentrations worsened survival rates with high tidal volumes. G-CSF also increased alveolar edema on histology, reduced lung compliance as manifested by increased mean airway pressures, and decreased arterial oxygenation. These effects of G-CSF may relate in part to the sequestration of neutrophils in the pulmonary vasculature that has been shown to occur with unstimulated neutrophils during positive-pressure ventilation in other studies.27,28In the presence of G-CSF, however, more neutrophils than in controls may have been sequestered and then recruited into the alveolar space either because of their increased numbers or level of activity. Marked decreases in circulating neutrophils in G-CSF–treated animals over the course of mechanical ventilation in combination with increases in alveolar neutrophils on tissue sections and in lung lavage are consistent with excessive recruitment of circulating cells. The fact that G-CSF worsened survival, mean airway pressures, and oxygenation with high but not low tidal volumes suggests that high tidal volumes or associated mediators in combination with G-CSF worked synergistically to cause injury. Such an interaction between increased pressure or stretch and leukocyte activity, although not described for the neutrophil, has been demonstrated with endotoxin-activated macrophages in vitro .22However, stimuli other than mechanical ones have been shown to work synergistically with G-CSF to increase the activity and release of inflammatory mediators by neutrophils.29–33Although G-CSF has been reported to cause systemic capillary leak syndrome and shock, there was no evidence that it aggravated the hypotension associated with high tidal volumes in this rat model.34,35 

The absence of increases in serum TNF or IL-6 concentrations in control animals receiving high tidal volumes despite decreased systemic hemodynamic function and survival rates in the current study is in contrast to some but not other investigations.36–38However, high tidal volumes were associated with significant increases in alveolar concentrations of IL-6, a cytokine that has been associated with inflammatory tissue injury and VILI in patients and animal models.10Lavage TNF-α was increased as well with high tidal volumes, but not significantly. With longer observation, it is possible that pulmonary production of inflammatory of cytokines such as IL-6 or TNF-α might have become evident in the systemic circulation and contributed to hemodynamic instability or nonpulmonary organ injury. To what extent increased lung IL-6 concentrations were the cause as opposed to the result of lung neutrophil recruitment and injury with high tidal volumes is not known. Consistent with the former possibility, reductions in lavage neutrophils with fucoidin were also associated with reductions in IL-6, whereas the opposite was evident with G-CSF. Consistent with the latter, however, in other studies, inhibition of CXC chemokines, another class of inflammatory cytokine, decreased lung neutrophil recruitment and VILI in mice, suggesting that such cytokine production has a direct role in this injury.39Whether similar chemokines are increased with high tidal volumes in this rat model requires further study. Increased alveolar chemokine production with high tidal volumes in combination with increases in circulating neutrophils could provide a basis for the worsened outcome with G-CSF. Although G-CSF has been reported to both decrease and increase inflammatory cytokines under differing conditions, in the current study, it seemed primarily to have a proinflammatory effect as reflected by the increases in IL-6 it produced.40,41Reductions in IL-6 with fucoidin in the current study are consistent with this agent’s inhibition of other antiinflammatory cytokines in previous studies.42 

Worsened outcome with G-CSF in this model is concerning given its proposed application in nonneutropenic patients with or at risk of pneumonia and sepsis.18,43These patients frequently have development of acute lung injury necessitating mechanical ventilation. Even when tidal volumes are reduced with severe injury, remaining normal lung units may undergo excessive inflation. The findings from our study raise the possibility that G-CSF therapy in patients with advanced lung injury requiring mechanical ventilation could worsen pulmonary dysfunction or even outcome. This danger may be greater with high oxygen concentrations based on studies showing that G-CSF may worsen hyperoxic lung injury.19However, G-CSF does have beneficial host defense effects with pneumonia and other types of infection.18,44,45In patients with acute lung injury related to bacterial infection or who are at risk of infection, the beneficial effects of G-CSF on microbial clearance may outweigh any potential adverse effects related to mechanical ventilation. Two large clinical trials have now evaluated the use of G-CSF in patients with severe pneumonia or sepsis, some of whom would have required mechanical ventilation.46–48Although the incidence of ARDS was increased in each study in patients receiving G-CSF, these increases were not noted to be significant, nor was G-CSF found to be harmful overall in either study.

In summary, although neutrophil inhibition by fucoidin had no measurable effect on reduced survival rates during mechanical ventilation with high tidal volumes in this rat model, neutrophil stimulation with G-CSF worsened them. Extrapolation of these findings clinically suggests that in critically ill patients at risk of VILI, neutrophil activation, either with immunostimulatory agents such as G-CSF or by concurrent conditions such as sepsis, may aggravate that injury. Of note, the increases in circulating neutrophil number noted with G-CSF treatment in the current study are similar in magnitude to the increases noted previously in rats challenged with intrabronchial bacteria.49Whether inhibiting activated neutrophils will be beneficial in ventilated patients at risk of VILI and exposed to such stimuli as G-CSF or sepsis requires further study.