is a major cause of lung infections, particularly nosocomial pneumonia (9
). Mortality rates of patients with P. aeruginosa
pneumonia are higher than the mortality rates of patients with pneumonia caused by other pathogens (14
) because P. aeruginosa
pneumonia frequently disseminates to cause bacteremia and sepsis (6
). This rapid dissemination is associated with the generation of acute lung epithelial injury by some strains of P. aeruginosa
). To protect patients who have P. aeruginosa
pneumonia from bacteremia, sepsis, and death, the effective utilization of antibiotics is important. However, most of the strains of P. aeruginosa
are inherently resistant to many antibiotics (10
The hypothesis tested by this study was that acute lung injury may be decreased by a single bolus of antibiotics if the antibiotics decrease the number of bacteria instilled in the lungs of rats. We also sought to determine whether the amount of lung injury was related to the change in endotoxin concentration over the experimental period. In this investigation, we quantitated the acute lung injury caused by the airspace instillation of a virulent strain of P. aeruginosa
, PA103, in anesthetized rats. The effects of different inoculums of the same strain, PA103, on multiple, independent parameters of lung injury were determined. We then investigated the effect of antibiotic treatment on these parameters of bacterially induced lung injury. We chose two antibiotics, aztreonam and imipenem-cilastatin, and compared their effects on the parameters of acute lung injury in the rat model. Both aztreonam (1
) and imipenem (3
) are utilized clinically to treat infections caused by this organism.
The MIC in vitro bacterial sensitivity test is utilized routinely to choose an antibiotic that can effectively kill a bacterial pathogen. However, the concentration of antibiotics at the site of infection, which is affected by various pharmacokinetic factors, including the rate of penetration into the specific tissues and the speed of metabolism of the drugs, may be more important in the efficacy of antibiotics in vivo. Therefore, we utilized this animal model to compare two antibiotics that had different MICs to determine which of the antibiotics was more efficacious in improving the parameters of bacterially induced lung injury.
MATERIALS AND METHODS
Surgical preparation and ventilation.
Male Sprague-Dawley rats (300 to 350 g; Simonsen, Gilroy, Calif.), certified pathogen free, were anesthetized with 30 mg of pentobarbital given intraperitoneally; anesthesia was maintained with 12 mg of intraperitoneally administered pentobarbital every 2 h. All rats remained anesthetized, intubated, and ventilated throughout the entire experiment. The right carotid artery was cannulated with a polyethylene tube (PE50; Clay Adams, Parsippany, N.J.), for monitoring arterial blood pressure and for blood sampling. The right jugular vein was cannulated with a PE50 tube, and the drugs were administered intravenously. Pancuronium bromide (0.3 mg/kg of body weight) was given intravenously every 2 h for neuromuscular blockade. An endotracheal tube (PE240; Clay Adams) was inserted into the trachea via a tracheostomy. The rats were ventilated with a constant-volume animal respirator (Harvard Apparatus, South Natick, Mass.) with an O2 fraction of 1.0 and 3-cm positive end-expiratory pressure. The respiratory rate was adjusted to maintain an arterial PCO2 between 35 and 45 mm Hg. All animal experiments were conducted in compliance with the guidelines of the Animal Care Committee of the University of California at San Francisco.
Culture conditions, bacterial strains, and instillate preparation.
PA103 was instilled to induce lung injury, as it has been shown to significantly increase the measurements of lung injury (20
). The strain was stored as a bacterial stock at −70°C in a 10% sterile skim milk solution. The bacteria were streaked onto a Vogel-Bonner minimal medium plate for 36 h at 37°C, and then the bacteria were cultured in tryptic soy broth for 13 h at 32°C in a shaking incubator.
The instillate consisted of a 1-ml iso-osmolar albumin solution prepared from Ringer’s lactate, 2 mg of Evans blue dye, and 3 μCi of125I-labeled albumin. P. aeruginosa at a concentration of 106, 107, or 108CFU, determined spectrophotometrically, was added to the instillate just prior to airspace instillation. A sample of the instillate was saved for measurement of radioactivity, protein concentration, and quantitative bacterial cultures to assure accurate inoculations.
The MICs of the antibiotics were determined by the standard microdilution technique, according to the methods of the National Committee for Clinical Laboratory Standards (21
). Aztreonam (Squibb Inc., Princeton, N.J.) at 60 mg/kg or imipenem-cilastatin (Merck, Inc., Rahway, N.J.) at 30 mg/kg was administered as an intravenous bolus dose 1 h after the airspace bacterial instillation. Control rats received an equal volume of normal saline (Baxter, Roundlake, Ill.) as a bolus. The investigator was blinded to whether a drug was administered. Due to the length of surgery, only two rats were studied in 1 day to maintain reproducibility of the model; a control rat was always one of the experimental animals.
Serum samples for measurements of endotoxin were obtained just prior to and 4 h after the bacterial instillation. The samples were allowed to clot for 20 min and then were centrifuged and immediately frozen at −70°C until they were analyzed. The serum samples were measured for free endotoxin with theLimulus amebocyte lysate assay (Chromogenic-Limulus Amebocyte; Wittaker M. A. Bioproducts, Walkersville, Md.).
General experimental protocol.
In all experiments, after the surgical preparation, 3 μCi of 131
I-albumin (Merck-Frosst, Kirkland, Quebec, Canada) was injected intravenously. The rats were then placed in a right lateral decubitus position. With the use of 1-ml syringes and PE50 tubes, the instillates were delivered to the right lungs (primarily the right lower lobes) over a 30-min interval. The rats were injected with 500 mg ofd
-galactosamine/kg intraperitoneally during the bacterial instillation in order to sensitize them to the effects of endotoxin (2
). Systemic arterial and airway pressures were measured at 1-h intervals. Blood samples were obtained for131
I-albumin and 125
I-albumin measurement every hour after the instillation for 6 h.
After 6 h, the rats were deeply anesthetized, their abdomens were opened, and they were exsanguinated. Samples of blood, urine, right pleural fluid, and liver were obtained for bacterial cultures and radioactivity counts. The lungs were removed, and any remaining instillate was aspirated with a 3-ml syringe and a PE50 tube passed into the right lower lobes. The total protein and radioactivity of the alveolar samples were measured. The right and left lungs were homogenized separately for water-to-dry-weight ratio measurements and for culture. Measurements of the total protein concentrations of plasma, pleural fluid, and initial and final instillate samples were performed by the Biruet method.
Measurements of lung injury.
Lung injury was quantified in four ways, as previously described (29
). The first method quantified the movement of the alveolar protein tracer,125
I-albumin, from the lung into the blood by measuring the amount of residual 125
I-albumin in the lung as well as the accumulation of the tracer in the plasma over the 6-h course of the experiment. Total 125
I-albumin instilled into the lung was determined by measuring duplicate samples of the instillate for total radioactivity (counts per minute per gram) and multiplying this amount by the total volume instilled into the lung. To calculate the residual125
I-albumin in the lung after 6 h, the average of two 0.5-g samples obtained from the lung homogenate was multiplied by the total volume of the lung homogenate. The radioactivities in the lung homogenates were added to the recovered radioactive counts in the aspirated fluid from the lungs. Circulating plasma125
I-albumin was measured from a sample of plasma obtained at the end of 6 h. The plasma fraction was accounted for by multiplying the counts per milliliter by the plasma volume [body weight × 0.07 (1 hematocrit)].
The second method required the measurement of 131
I-albumin (the vascular protein tracer) in the airspace compartment of the lung at the end of the experiment. This was accomplished by measuring the131
I-albumin counts in the final airspace sample. Plasma131
I-albumin counts were averaged over the course of the experiment, and the 131
I-albumin counts in the airspaces were expressed as a ratio to the plasma counts. This ratio provides a good index of equilibration of the vascular protein tracer into the alveolar compartment, as had been shown previously (30
Third, extravascular lung water was determined for each separate lung by calculating the water-to-dry-weight ratio. The excess water in the experimental lung was calculated with the same equation described previously (30
). Briefly, the volume of excess lung water (ELW; in ml) of the serum-instilled experimental lung was calculated as follows: ELW = [We
− P) − Wc
− P ), where W
are extravascular lung water and blood-free dry weights, respectively, of the experimental lung (e
) and the control lung (c
is the blood-free dry weight of the initial alveolar fluid multiplied by the fraction of 125
I-albumin remaining in the lung.
Finally, the accumulation of the vascular protein tracer into the extravascular space of the lung was calculated by determining the total extravascular count of 131I-albumin in the lung divided by the average counts in the plasma over the 6-h study period and was expressed as total plasma equivalents in milliliters.
All data presented are means ± standard errors. The data were analyzed by one-way analysis of variance followed by Dunnet’s multiple-comparison test. A P value of <0.05 was accepted as statistically significant.
A model that produces a consistent quantity of bacterially-induced lung injury is a useful tool for comparing the effects of various antibiotic treatments on markers of lung injury. In the development of such a model we have shown that the administration of small doses (106 CFU) of P. aeruginosa for longer intervals (4 h) consistently led to moderate quantities of lung injury. The administration of larger doses (108 CFU) of P. aeruginosa for longer intervals produced larger quantities of lung injury with very small standard deviations in the measurements. Therefore, the airspace administration of P. aeruginosa can be utilized to produce a consistent quantity of lung injury that can be increased or decreased by modifying the dose or the interval. These results allowed us to choose a dose of P. aeruginosa, 107 CFU, and an interval, 6 h, to produce a moderate amount of lung injury. While this model has some limitations, such as the short evaluation period of 6 h, it also has advantages, such as reproducible, quantifiable lung injury measurements. Because of these advantages, we elected to use this model to test the hypothesis that acute lung injury may be decreased by a single bolus of antibiotics if the antibiotics decrease the number of bacteria instilled in the lungs of rats. We also sought to determine whether the amount of lung injury was related to the change in serum endotoxin concentration over the experimental period. In fact, we were able to document that one bolus of imipenem-cilastatin, but not aztreonam, significantly improved all parameters of the bacterially induced lung injury as evidenced by the extravascular lung water and total plasma equivalents. The extravascular lung water is an overall measurement of lung edema. Total plasma equivalents is also a measure of lung edema but is more specific for the movement of plasma into the lung. We were not able to correlate these parameters of lung injury with plasma endotoxin concentration.
Imipenem-cilastatin, but not aztreonam, was able to significantly improve lung injury. P. aeruginosa
PA103 is sensitive to aztreonam, as the MIC for it was 0.5 μg/ml; therefore the lack of improved killing compared to untreated controls was not due to antibiotic resistance. In contrast, there was a trend for increased killing by imipenem-cilastatin despite a MIC of imipenem (16 μg/ml) higher than the MIC of aztreonam (0.5 μg/ml). This result suggests that measurement of the in vitro killing of bacteria does not correlate with antibiotic-induced killing in an airspace infection in vivo. The pharmacokinetics of aztreonam and imipenem-cilastatin have been described previously in rats (17
). The penetrations of these antibiotics into the lung tissue are comparable, measuring approximately 50% of that in the serum. The protein binding of aztreonam is nearly 50 to 60%, whereas that of imipenem-cilastatin is only 10 to 20% (7
). In order to account for this difference in protein binding in the model, we elected to give a higher dose of aztreonam (60 mg/kg) than of imipenem-cilastatin (30 mg/kg). Aztreonam displays an effect of inoculum on the MIC at 107
CFU/ml, whereas impenem does not show an inoculum effect on the MIC until 108
). It is possible that the inoculum selected for study, 107
CFU/ml, favored the activity of imipenem-cilastatin. However, the MIC of aztreonam was five times lower that that of imipenem-cilastatin, which would favor the activity of aztreonam.
This model of lung injury also resulted in bacteremia, as evidenced by the positive liver cultures. Administration of imipenem-cilastatin, but not aztreonam, resulted in a reduction of systemic hypotension over the experimental period, which may have contributed to the increased lung injury observed in the aztreonam-treated rats.
It has been demonstrated that antibiotics which target PBP-2, such as imipenem, liberate less endotoxin than antibiotics which bind PBP-3, such as aztreonam (12
). Aztreonam may cause endotoxin release due to filamentation, which appears to be secondary to the properties of the penicillin binding protein. In fact, we previously documented that the airspace instillation of aztreonam along with P. aeruginosa
caused endotoxin release and increased acute lung injury (24
). However, in this model, there were no differences in the serum levels of endotoxin between the groups. Therefore, antibiotic-induced release of serum endotoxin did not appear to explain the differences in the lung injury measurements between the aztreonam-treated rats and the imipenem-cilastatin-treated rats. It is possible that lung endotoxin levels were different in the treatment groups; however, we were unable to measure lung endotoxin levels using this model.
In conclusion, a virulent strain of P. aeruginosa, PA103, can be utilized to create a consistent quantity of lung injury for comparisons of antibiotic effectiveness in airspace infections. Imipenem-cilastatin administration led to significant improvement in all the measured parameters of lung injury. In contrast, aztreonam, which was effective in killing this bacterium in vitro, did not improve any of the parameters of lung injury compared to nontreated animals. These results suggest that the evaluation of antibiotic effects in vivo may be important in airspace infections. Our animal model forPseudomonas pneumonia is useful for evaluation of the comparative effects of antibiotics.