INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: fluorocarbons; mechanical ventilation; antibiotics; drug administration routes; topical administration; gentamicin; vancomycin; bacterial pneumonia

Pneumonia may result in impaired gas exchange because of impaired diffusion, ventilation-perfusion mismatch, and atelectasis, and may also result in impaired lung mechanics (1).

Partial liquid ventilation (PLV), also known as perfluorocarbon-associated gas exchange, improved oxygenation and pulmonary compliance in several animal models of surfactant deficiency (2, 3), in meconium aspiration syndrome (4, 5), and also in human infants, children, and adults with severe respiratory distress syndrome (6). Pulmonary perfluorocarbon instillation also improved survival in a model of lethal pneumococcal pneumonia in rats (11). Therefore, PLV may also become an option for the treatment of severe respiratory failure caused by bacterial pneumonia.

Intrapulmonary administration of antibiotics during pneumonia may deliver maximum amounts of antibiotics to the site of infection and simultaneously avoid potential systemic side effects. Intrapulmonary administration by inhalation of aerosolized aminoglycosides has been recommended for patients with cystic fibrosis (12). However, this form of intrapulmonary antibiotic administration is limited by insufficient drug delivery to peripheral lung regions and by the inability to quantify the actually administered dose (15).

Several investigators have examined the efficacy of the intrapulmonary administration of vasoactive substances (18- 20), narcotics (21), and antibiotics (22) during partial and tidal liquid ventilation. However, most of these investigators studied the intrapulmonary bolus administration of an aqueous solution during liquid ventilation (18, 20, 22, 23). Because aqueous solutions do not mix well with perfluorocarbons, it is very unlikely that a homogeneous distribution of the agent in the lungs was achieved.

The aim of this study was to examine the pharmacokinetics after intrapulmonary administration of gentamicin and vancomycin using specifically designed emulsions of gentamicin and vancomycin in perfluorodecaline (PFD) in comparison with intravenous administration in rabbits with and without lung disease undergoing PLV. We hypothesized that intrapulmonary administration of a PFD-gentamicin and a PFD-vancomycin emulsion would result in higher pulmonary antibiotic tissue concentrations (PATC) than the intravenous administration of the same dose of gentamicin and vancomycin, respectively.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The experiments were approved by the animal care committee of the State of Baden-Württemberg, Germany, and were performed according to current animal care guidelines. A detailed description of the instrumentation, experimental procedures, determination of PATC, pharmacokinetics, and histologic evaluation is available in the online data supplement.

Instrumentation and Experimental Procedures

Forty adult New Zealand White rabbits were anesthetized, intubated, and mechanically ventilated. A femoral arterial and a pulmonary arterial line were placed for blood sampling, pressure, and cardiac output monitoring. Twenty animals were surfactant depleted by repeated saline lavage (25).

Industrial grade perfluorodecaline (Sigma-Aldrich, Deisenhofen, Germany) was purified and PFD-antibiotic emulsions were prepared using natural bovine surfactant (Alveofact, Boehringer Ingelheim, Biberach, Germany) as an emulsifier (W.R. and R.P.F).

The animals were randomized to one of three groups using sealed opaque envelopes. Animals in the intravenous group received 5 mg/kg gentamicin and simultaneously 15 mg/kg vancomycin in 20 ml normal saline intravenously over 15 min, and 20 ml prewarmed PFD intrapulmonary. Animals in the PFD-gentamicin group received 5 mg/kg gentamicin emulsified in 20 ml of prewarmed PFD intrapulmonary and 20 ml of normal saline intravenously over 15 min. Animals in the PFD-vancomycin group received 15 mg/kg vancomycin emulsified in 20 ml of prewarmed PFD intrapulmonary and 20 ml of normal saline intravenously over 15 min.

Antibiotic plasma concentrations were measured at 15, 30, 45, and 60 min and hourly thereafter. After 5 h, animals were killed. The right lung was removed to determine PATC, and measurements of PATC were corrected for remaining PFD emulsion within the lung. The left lung was fixed in situ by perfusion with a formalin/glutaraldehyde solution while maintained on positive airway pressure. Lung histology was evaluated by a pathologist (M.E.), blinded to the animal's group assignment, according to a modified score (26). Pharmacokinetics at different time intervals and after different routes of administration were calculated.

Data Analysis and Statistical Evaluation

The primary outcome variable was the PATC at 5 h. Secondary outcome variables were plasma antibiotic concentrations, lung histology scores, respiratory rate, minute ventilation, mean airway pressure, dynamic compliance and airway resistance, arterial blood gas analyses, arterial, pulmonary arterial, and central venous blood pressures, heart rate, and cardiac output.

The sample size calculation revealed that at least five animals would be necessary per group to detect a difference of lung tissue concentrations of 4 µg/g between the two modes of antibiotic administration, assuming a standard deviation of 1.6 µg/g (22), an -error of 0.025 (Bonferroni correction for two comparisons), and a power of 0.80. To compensate for potential losses we decided to study 20 animals with and without lung disease.

PATC and histologic scores were compared using the two-sided exact Wilcoxon test. All other variables were analyzed using two-way analysis of variance for repeated measurements as a function of time and group for the time 60-300 min after randomization. The significance level was 0.05, and Bonferroni correction for multiple testing was applied for the primary outcome variable.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Eighteen of 20 animals without lung disease and 15 of 20 surfactant-depleted animals completed the protocol. One of the animals without lung disease died during instrumentation and another one (in the PFD-gentamicin group) died prematurely during the study. In the surfactant-depleted animals, two animals died during the lavage procedure, one died prior to randomization, and two died prematurely during the study period (both in the PFD-vancomycin group). Animals dying prematurely were excluded from the analysis of PATC.

Surfactant Depletion

In the 17 surfactant-depleted animals entering the study, dynamic compliance during synchronized, time-cycled, volume-controlled ventilation decreased from 0.95 ± 0.42 ml/cm H₂O/kg before lavage to 0.53 ± 0.35 ml/cm H₂O/kg before the initiation of PLV, and the Pa_O2 (at a fraction of inspired oxygen [FI_O2] of 1.0) decreased from 469 ± 68 mm Hg to 241 ± 115 mm Hg.

Baseline Data

Before antibiotic administration, parameters of hemodynamics, gas exchange, and lung mechanics were similar between the three treatment groups in both cohorts of animals, except for a higher mean central venous pressure (8.2 ± 0.9 versus 5.9 ± 1.0 mm Hg, p = 0.008) and a trend toward a lower mean arterial blood pressure (46.6 ± 6.2 versus 55.5 ± 6.7 mm Hg, p = 0.07) in the surfactant-depleted rabbits of the PFD-vancomycin group compared with the intravenous group (the complete baseline data are shown in Table E1 in the online data supplement).

Tissue Concentrations

PATC at 5 h were similar in dependent and nondependent regions of the lung after both intravenous and intrapulmonary administration, and therefore mean PATC are presented. PATC at 5 h were significantly higher after intrapulmonary administration of PFD emulsions with gentamicin or vancomycin in both animals without lung disease and in surfactant-depleted animals, if compared with intravenous administration of the same dose (Table 1). PATC after intrapulmonary administration of vancomycin were significantly lower in surfactant- depleted animals compared with the animals without lung disease (p = 0.01). PATC after intrapulmonary administration of gentamicin also tended to be lower in surfactant-depleted animals, and PATC after intravenous administration of both antibiotics tended to be higher in surfactant-depleted animals if compared with animals without lung disease, but these differences did not reach statistical significance.

Pharmacokinetics

After intravenous antibiotic administration, plasma concentrations were well described by a two-compartment model (Table E2 in the online data supplement depicts the appropriate variables describing such a model). Peak plasma concentrations of both antibiotics were reached at the end of infusion (15 min) with an estimated volume of distribution (V_D) of about 0.1 L/kg approximately corresponding to the blood volume of the rabbits. Thereafter, plasma concentrations initially decreased with an elimination constant (k_e) of 1.1-1.8 for gentamicin and a k_e = 0.6-1.5 for vancomycin with both elimination and simultaneous distribution to a final estimated V_D of about 0.41 L/kg for gentamicin and 0.47 L/kg for vancomycin. By about 1 h distribution was completed and thereafter elimination occurred with a k_e = 0.5 for gentamicin and k_e = 0.4 for vancomycin, closely following a first-order elimination kinetic (Figures 1 and 2, Table 2). There were no major differences in pharmacokinetics between animals without lung disease and surfactant-depleted animals after intravenous administration.

View larger version (24K): [in this window] [in a new window]   Figure 1.   Plasma concentrations of gentamicin (A) and vancomycin (B) in animals without lung disease. Solid line = intravenous administration; broken line = pulmonary administration. Values are depicted as mean ± SD.

View larger version (25K): [in this window] [in a new window]   Figure 2.   Plasma concentrations of gentamicin (A) and vancomycin (B) in surfactant-depleted animals. Solid line = intravenous administration; broken line = pulmonary administration. Values are depicted as mean ± SD.

In animals without lung disease, plasma antibiotic concentrations after intrapulmonary administration did not reach the peak values seen early after intravenous administration (Figure 1, Table 2). In contrast, plasma antibiotic levels remained higher with apparently slower elimination at 1-5 h after intrapulmonary administration (Figure 1, Table 2), suggesting that in addition to the two-compartment system with distribution and elimination seen after intravenous administration, simultaneous resorption occurred.

In surfactant-depleted animals, plasma antibiotic concentrations were similar after both intravenous and intrapulmonary administration at least during the last 4 h of the experiment (Figure 2, Table 2). Whereas gentamicin reached similar peak concentrations after intrapulmonary and intravenous administration in the surfactant-depleted animals, vancomycin did not. The elimination coefficients at 1-5 h after intrapulmonary administration in surfactant-depleted animals were similar to those seen after intravenous administration and tended to be higher than those after intrapulmonary administration in animals without lung disease.

The concentrations of gentamicin and vancomycin in the PFD emulsions decreased rapidly within the first hour (Figure 3).

View larger version (20K): [in this window] [in a new window]   Figure 3.   Concentrations of gentamicin and vancomycin in PFD emulsions after intrapulmonary administration in animals with and without lung disease. Values are depicted as percent of initial concentration (% BL). Note that the decline of PFD-vancomycin concentrations was similar in animals with and without lung disease (RM ANOVA: p = 0.87), whereas PFD-gentamicin dropped more rapidly in surfactant-depleted animals compared with animals without lung disease (RM ANOVA: p = 0.022).

Lung Mechanics, Gas Exchange, and Hemodynamics

In animals without lung disease, significant differences were found between the PFD-vancomycin group and the intravenous group in respiratory rate (33 ± 6 versus 28 ± 2 breaths/ min, p = 0.04, PFD-vancomycin group versus the intravenous group) and airway resistance (54 ± 9 versus 74 ± 12 cm H₂O/ L/s, p = 0.006, PFD-vancomycin group versus the intravenous group), but no differences were found between the intravenous and the PFD-gentamicin group. No differences were found for parameters of gas exchange and hemodynamics. In surfactant-depleted animals, no significant differences were found for any parameter of hemodynamics, gas exchange, or lung mechanics (complete data are depicted in Table E3 in the online data supplement).

Lung Histology

Lower lobes and dependent areas were more severely injured than upper lobes and nondependent areas (data available upon request). Because these patterns of injury were equally present in all three treatment groups, sum scores were calculated for each variable in each animal. Surfactant-depleted lungs were more severely injured than lungs without surfactant depletion. There were more inflammatory changes in the intravenous group compared with the PFD-gentamicin group in animals without lung disease and compared with the PFD- vancomycin group in surfactant-depleted animals. In comparison with the intravenous groups, the scores for alveolar edema were higher in the PFD-vancomycin group in animals without lung disease and the PFD-gentamicin group in surfactant- depleted animals. There was a trend toward more overdistension in both PFD groups in healthy animals (Table 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nosocomial pneumonia is a major problem in adult, pediatric, and neonatal intensive care units, occurring in 9-60% of mechanically ventilated patients with a mortality rate of 20-50% (27). Aminoglycosides, frequently administered in patients with pneumonia, penetrate poorly into lung parenchymal tissue and bronchial secretions, as demonstrated in animal models and humans (30). Similarly, vancomycin concentrations in epithelial lining fluid of the lower respiratory tract approximate only a sixth of the plasma concentration after intravenous administration (33) and vancomycin may not even be detectable in bronchial secretions using a typical intravenous dosing regimen (34).

This study demonstrated that PATC were significantly higher at 5 h after a single dose of either vancomycin or gentamicin given by intrapulmonary administration during PLV compared with intravenous administration during PLV. This is in agreement with previous studies using either aqueous solutions of gentamicin during total liquid ventilation (22, 23) or a nanocrystal suspension (gentamicin in perflubron) during PLV (24). Although PATC above most minimal inhibitory concentrations for susceptible pathogens were achieved in most (50-75%) of the intravenously treated animals, higher concentrations are required to ensure bactericidal activity. In vitro studies showed that an aminoglycoside bactericidal effect can be reliably produced only with concentrations up to 25 times the minimum inhibitory concentration (30), suggesting that the increase in lung tissue concentrations achieved by intrapulmonary administration may be clinically important.

In agreement with findings in humans (35, 36), pharmacokinetics of gentamicin and vancomycin following intravenous administration were best described by a two-compartment model (Figures 1 and 2, Table 2) with a faster elimination from plasma early on until the plasma was equilibrated with the final volume of distribution and a slower elimination that closely approximated first-order kinetics during the last 3-4 h of the experimental period. In contrast, the time course of plasma concentrations after intrapulmonary administration in animals without lung disease did not follow this model, suggesting the existence of at least one further compartment, from which antibiotics were continuously absorbed during the first 1-2 h of the experiment. The very rapid decline of the antibiotic concentration in the PFD emulsion (Figure 3) suggests that this third compartment was not the emulsion, but probably rather the lung tissue itself.

The rapid decline of the antibiotic concentration in the PFD emulsion is best explained by the in vitro observation that the particles within the emulsion have a high affinity to any aqueous phase: Adding small amounts of saline or plasma to the emulsion and shaking it vigorously for 10 s results in transfer of more than 90% of the suspended antibiotic dose into the aqueous phase (data not shown). Similar findings were reported by other investigators using a nanocrystal suspension of gentamicin in perflubron (24).

In surfactant-depleted animals, PATC after intrapulmonary administration of vancomycin were lower and there was a trend toward higher PATC after intravenous administration of both antibiotics if compared with animals without lung disease. Moreover, plasma levels after intrapulmonary administration more closely resembled those after intravenous administration in surfactant-depleted animals, suggesting a higher permeability for gentamicin and vancomycin in both directions, that is, into and out of the lungs. Previous studies also demonstrated that the permeability of the "blood-bronchial barrier" (31) for gentamicin was increased after induction of pulmonary inflammation (32), and that PATC of vancomycin were higher after intravenous administration if the alveolar capillary membrane protein permeability was increased (34).

In surfactant-depleted animals, peak vancomycin plasma concentrations after intrapulmonary administration were lower than after intravenous administration. This was in contrast to peak gentamicin plasma concentrations, which were very similar after both modes of administration. Furthermore, the concentrations of vancomycin in the PFD emulsion dropped at a similar rate in both animals with and without lung disease, whereas the concentrations of gentamicin in the PFD emulsions dropped significantly faster in surfactant-depleted animals (Figure 3, p = 0.022). This difference may be explained by the larger molecular weight of vancomycin compared with gentamicin.

Although the high-density PFD emulsions probably would be preferentially directed to dependent lung regions in a gas-filled lung, no significant differences in PATC were found between dependent and nondependent areas in this study. However, lungs had been partially filled with PFD before instillation of the PFD emulsions, an approach that was previously shown to result in a homogeneous drug distribution after administration of a nanocrystal suspension (24).

Although the distribution of inhaled aerosolized antibiotics to peripheral lung areas is diminished if lung function is impaired (16), one may speculate that antibiotics suspended in perfluorocarbon liquids will reach even poorly ventilated areas because of their ability to reduce alveolar surface tension and to recruit and stabilize collapsed alveoli (37). However, future studies are needed to verify this speculation.

No unexpected clinically significant differences in gas exchange, lung mechanics, and hemodynamics were found between the different modes of drug administration. We found trends toward more alveolar edema and more overdistension after intrapulmonary administration, which may be best explained by the large number of comparisons performed. One could speculate that the presence of the osmotically active antibiotics in the alveolar lining fluid may increase alveolar edema. The slight increase in alveolar edema observed is probably not clinically significant because it was not associated with impaired gas exchange and it may resolve once antibiotics are completely absorbed. The trend toward higher scores for overdistension after intrapulmonary administration in animals without lung disease could reflect plugging of the small airways. However, the significance of this finding is unclear because overdistension was not observed after intrapulmonary administration in surfactant-depleted animals. In summary, our data do not provide evidence for relevant short-term toxicity to the lung caused by the high PATC. However, it is noteworthy that 3 of the 26 animals treated with intrapulmonary PFD-antibiotic emulsions died before completion of the study, whereas all 10 animals that received the antibiotics intravenously survived until the end of the study (p = 0.54 by two-sided Fisher's exact test).

These increased PATC were achieved at lower peak plasma levels. Plasma levels at 5 h were similar in both intrapulmonary and intravenously treated surfactant-depleted animals, but tended to be higher after intrapulmonary administration in animals without lung disease. Nevertheless, normal plasma levels of both gentamicin (< 2 mg/L) and vancomycin (< 10 mg/L) would have been reached at 8 or 12 h after intrapulmonary administration, taking into account the 5 h plasma level and the elimination half-life period, suggesting that the increased PATC can be achieved without increasing oto- and nephrotoxicity.

This study is limited by the small number of animals studied. To evaluate the pharmacokinetics and toxicity of intrapulmonary antibiotic administration more precisely more animals need to be studied. Furthermore, PATC should be measured at different time points and the effect of repeated dosing and longer observational periods should be explored.

In conclusion, we have demonstrated that intrapulmonary administration of emulsions of both gentamicin and vancomycin in PFD results in higher PATC than intavenous administration of the same dose. We speculate that this mode of drug delivery may be helpful in the treatment of severe pneumonia with respiratory failure because highest drug levels would be reached at the site of infection, which may be essential to ensure bactericidal effects while limiting the risk of systemic toxicity. Further studies are needed to evaluate the role of PLV and intrapulmonary drug administration in subjects with bacterial pneumonia.