Aerosolized Antibiotics for Ventilator-associated Pneumonia: Lessons from Experimental Studies

Understanding factors promoting distal lung deposition and pharmacokinetics of aerosolized antibiotics is a prerequisite for safe and efficient clinical use. It requires experimental models relevant to human ventilator-associated pneumonia.12 

The first experimental models of pneumonia were set up in spontaneously breathing rodents or monkeys.13–16Recent reviews have summarized the specificities and advantages and limits of mouse, rats, and rabbit model of experimental pneumonia based on the intratracheal or intranasal inoculation of various bacterial species, including Streptococcus pneumoniae .17,18Parenchymal lung infection is induced by exposure to aerosolized microorganisms, intranasal instillation, or direct endobronchial instillation. Immediate postmortem quantitative culture of homogenized and serially diluted infected lung remains the reference technique for assessing initial bacterial burden and the effect of antimicrobial therapy. Because of the small size of animals, such a technique does not permit to differentiate bronchial from lung infection. Other methods such as detection of nucleic acids by real-time polymerase chain reaction and bioluminescence19seem attractive options. The accumulation of leukocytes within infected lung can be assessed in bronchoalveolar fluid lavage or, more accurately, using histological morphometric analysis. As far as antibiotic nebulization is concerned, the greatest limitation of mice models comes from marked differences in the anatomy of the murine compared with the human respiratory tract: The tracheobronchial tree is considerably smaller, has fewer and less symmetrical branches, and airways terminate abruptly into alveolar ducts without intervening respiratory bronchioles,20thereby facilitating lung deposition of aerosolized antibiotics.

In the early 1980s, an experimental intensive care unit for baboons was initiated by Johanson et al .21to study mechanisms of lung repair after oleic acid–induced lung injury. After a few days of mechanical ventilation, histological evidence of lung superinfection modified the initial investigator’s target, and finally, the model served for studying and understanding the pathophysiology of ventilator-associated pneumonia.22After completion of oleic acid–induced lung injury, anesthetized animals were ventilated during 7–10 days, a period during which they developed ventilator-associated pneumonia. Bacteriological investigations provided compelling evidence that the source of lung parenchymal infection was bacterial colonization of the tracheobronchial tree originating from the oropharynx.23Antimicrobial agents applied topically in the oropharynx and trachea combined with intravenous antibiotics were efficient for preventing pneumonia in a series of anesthetized baboons ventilated for 7–10 days.24 

In the mid 1990s, Marquette et al .25created an experimental model of tracheobronchial stenoses for studying the efficacy of airway stents. In healthy anesthetized piglets, cartilaginous arches were resected and a caustic agent was bronchoscopically applied to create a tracheobronchial stenosis after 4 days of mechanical ventilation. Again, histological evidence of lung infection after a few days of mechanical ventilation interfered with initial investigators’ aims and served for setting up and describing an experimental model of ventilator-associated pneumonia.26In a series of 23 anesthetized healthy piglets ventilated for 4 days in the prone position, histological pneumonia was evidenced in 22, extending to less than 30% of the lung, involving predominantly dependent lung segments and made essentially of foci of pneumonia. After 4 days of mechanical ventilation, ventilator-associated pneumonia was not severe, with a mild impairment in arterial oxygenation, a limited increase in lung weight, and the absence of large lung areas with confluent bronchopneumonia.26Similar lesions were observed after 60 h of mechanical ventilation.27 

Four experimental conditions seem to be of peculiar importance as far as antibiotic nebulization is concerned. First, animal size should be large enough to approach human tracheobronchial tree anatomical complexity. Distances between proximal trachea and alveolar compartments, airway diameters, and the number of bronchial divisions are critical factors affecting lung deposition of aerosolized particles. Therefore, experiments on large animals whose respiratory system is close to human respiratory system seem highly justified. Experiments on small animals, although easier to perform, would tend to underestimate the difficulty for aerosolized antibiotics to reach the deep lung. Second, severity and extension of lung infection should be reproducible enough to produce different histological grades of pneumonia within the same lung and to assess the influence of aeration loss on antibiotic lung deposition. Third, animals should be anesthetized and ventilated in their physiologic prone position for several days in experimental conditions reproducing intensive care unit environment. Fourth, specific ventilator settings used in experimental animals during nebulization phases should be close to those used in ventilated critically ill patients.

In the last 15 yr, a new experimental model derived from Marquette’s model was set up to specifically study lung deposition and efficiency of intravenous and aerosolized antibiotics.28,29The model combines prolonged mechanical ventilation with massive bronchial inoculation of highly concentrated pathogens in anesthetized piglets. It produces severe bronchopneumonia resulting not only from the initial massive bacterial inoculation but also from the aggravating role of mechanical ventilation.30The model opens the possibility of assessing how aeration loss and severity of parenchymal infection influence lung deposition of aerosolized antibiotics. It requires an experimental intensive care unit with the corresponding equipment and the presence on a 24-h period shift of trained physicians and technicians. Such a structure was set up in the late 1990s by the groups of Marquette and Rouby28and in the late 2000s by the groups of Luna31and Torres.29,32Healthy bred domestic Large White-Landrace piglets, aged 3–4 months and weighing 20 ± 2 kg, are anesthetized using propofol and orotracheally intubated in the supine position. Catheters are inserted in the ear vein, femoral artery, and bladder. After anesthetic induction, piglets are placed in the prone position and anesthesia is maintained with a continuous infusion of midazolam, pancuronium, and fentanyl (fig. 1). Animals are mechanically ventilated in a volume-controlled mode for several days and inspired gases humidified using a conventional humidifier. Tidal volume, respiratory rate, positive end-expiratory pressure, and fraction of inspired oxygen are adjusted throughout the experiments to provide PaO²≥ 80 mmHg and PaCO²⩽ 50 mmHg.

Massive lung infection is produced by the inoculation of 40 ml of a suspension containing 10⁶colony-forming units per milliliter (CFU/ml) of different bacterial species—P. aeruginosa, Escherichia coli , Klebsiella pneumoniae,  methicillin-resistant Staphylococcus aureus —in different parts of the respiratory tract using bronchoscopy. As shown in figure 2, the bronchial inoculation is aimed at reproducing the heterogeneous distribution of lung infection characterizing ventilator-associated pneumonia.30,33Superimposed lung infection, essentially caused by Pasteurella multocida and Streptococcus suis  is constantly observed, resulting from prolonged mechanical ventilation.26,30 

At the end of experiments, heart and lungs are exposed through a cervicothoracic incision, and animals are killed either by exsanguination performed through direct cardiac puncture28or by intravenous overdose of potassium chloride.34Multiple subpleural and juxtahilar 3–4 cm³lung specimens are excised from apical and middle lobes and from apical, anterior, and posterior segments of lower lobes. Each specimen is cut into three parts: the first block for quantitative bacterial culture, the second block for histological analysis and quantification of lung aeration,28and the third block for measurement of antibiotic tissue concentrations. The latter measurement can be performed using high-performance liquid chromatography on tissue samples that are cryomixed in nitrogen, weighed, and homogenized in buffer solution.35 

Although reduced by premortem exsanguination, blood contamination of the lung tissue sample may induce an artifactual increase in antibiotic tissue concentration. The overestimation of antibiotic lung tissue concentration should be corrected as follows. Each lung specimen is cryomixed and centrifuged (2,000g ) to obtain a lung supernatant. The amount of hemoglobin (Hb) present in the lung specimen (Q  ^hb) is defined as Q  ^hb= Hb^lungV  ^supern, where V  ^supernis the supernatant volume and Hb^lungis the Hb supernatant concentration. Blood (V  ^{blood lung}) and plasma (V  ^{plasma lung}) volumes present in the supernatant volumes are calculated as V  ^{blood lung}= Q  ^hb/Hb and V  ^{plasma lung}= V  ^{blood lung}/Ht, where Ht is hematocrit. The amount of antibiotic present in V  ^{plasma lung}(Q  ^{Abplasma lung}) is then calculated as Q  ^{Ab plasma lung}= C  ^{Ab plasma}V  ^{plasma lung}, where C  ^Abplasmais the antibiotic concentration measured in V  ^{plasma lung}. Finally, the “effective” antibiotic concentration present in the lung tissue specimen (C  ^Ablung) is calculated as:

C  ^Ablung= (Q  ^Ablung−Q  ^{Abplasma lung})/V  ^supern

The peak plasma level (C  ^max), the time of the C  ^max(T  ^max), and trough plasma concentrations (C  ^min) were obtained by direct observation of the individual kinetic profiles. The lack of exsanguination at the time of death34may lead to marked overestimation of antibiotic lung tissue concentrations in piglets with methicillin-resistant S. aureus  pneumonia treated by continuous intravenous administration of vancomycin.36 

Histological grade of bronchopneumonia can be divided into five categories of increasing severity33: (1) bronchiolitis, defined as the proliferation of leukocytes within the bronchial lumen leading to the formation of purulent plugs and associated with necrosis and disruption of bronchial mucosa; (2) interstitial bronchopneumonia, defined as the presence of scattered neutrophilic infiltrates localized to alveolar septa and terminal bronchioles; (3) focal bronchopneumonia, defined as an intense proliferation of leukocytes localized to terminal bronchioles and surrounding alveoli; (4) confluent bronchopneumonia, defined as an extension of these elementary lesions to one or several adjacent pulmonary lobules; and (5) purulent bronchopneumonia and lung abscess, defined as confluent bronchopneumonia associated with tissue necrosis and disruption of normal lung architecture. Classification of a given pulmonary lobule is based on the worst category observed. Final classification of a lung segment is defined as the most frequently observed lesion in all secondary pulmonary lobules present in the histological sections cut from the tissue block representative of the lung segment.

To quantify lung aeration, an image analyzer computerized system is coupled to a high-resolution color camera and an optical microscope objective. Each histological section is analyzed on a screen of a personal computer connected to the optical microscope and the color camera. Each optical field is analyzed as an automatically delineated rectangular elementary unit with an area of 2.289 mm². Within the elementary unit, aerated lung structures are automatically identified by a color encoding system included in a computerized program specially created for this experimental model. Pulmonary vessels and interlobular septa are visually detected and manually deselected to include as lung aeration air-filled structures only. Lung aeration of the elementary unit, expressed as a percentage, is computed as the area of alveolar and bronchial air-filled structures divided by the difference between 2.289 mm²and the area of pulmonary vessels and interlobular septa. For a given histological section, representative of a lung segment in a given animal, lung aeration is computed as the mean lung aeration of all elementary units present in the histological section.37 

As shown in figure 3, one possible and strong rationale for comparing bactericidal efficiency of aerosolized and intravenous antibiotics is to deliver an equivalent amount of antibiotics to the respiratory system by the tracheal route and the pulmonary circulation.28,38–42Therefore, to determine the appropriate aerosol dose, assessment of extrapulmonary deposition of aerosolized antibiotics, defined as the sum of nebulizer retention and deposition within respiratory circuits (inspiratory limb, Y piece, endotracheal tube, expiratory limb, and expiratory filter), should be performed. Extrapulmonary deposition of antibiotics can be measured by washing out with a known volume of water the different parts of ventilator circuits and measuring the mean retrieved antibiotic concentration. Pulmonary deposition of inhaled antibiotics is then computed as the total amount of antibiotic administered in the nebulizer chamber minus the extrapulmonary deposition. The appropriate dose of aerosolized antibiotics can be defined as follows:

Dose inserted in the nebulizer = intravenous dose + extrapulmonary deposition

Another possible rationale for comparing aerosolized and intravenous antibiotics would be to deliver the highest aerosolized antibiotic dose that does not produce systemic and lung toxicity. Such determination is difficult to assess and requires repetitive systemic dosages before adjusting the optimal dose for a given animal. Because of its complexity, assessment of such a rationale has not yet been reported in experimental studies.

Anatomy of the tracheobronchial tree of piglets differs from the human anatomy in several ways: the origin of segmental bronchi,42number and orientation of bronchial divisions, and bronchial diameter and size. As a consequence, lung deposition of nebulized antibiotics reported in experimental studies may not be automatically extrapolated to humans. Massive and unique bronchial inoculation of high concentrations of bacteria induces severe lung infection involving pulmonary lobes and segments. Such histological pattern may be different from human ventilator-associated pneumonia, characterized by repetitive bacterial seeding of the tracheobronchial tree through microaspiration of oropharyngeal secretions43and disseminated foci of pneumonia.33,44Last, but not least, the frequent infection of lung parenchyma after a few days of mechanical ventilation with microorganisms colonizing piglet’s oropharyngeal cavity such as P. multocida, S. suis , or Bordetella bronchiseptica  26complicates the evaluation of clinical outcome of animals treated with nebulized antibiotics.

In the 1980s, mechanical ventilation was considered as a factor limiting aerosol lung deposition to less than 5% of the dose deposited into the nebulizer.45Currently, the understanding of technical conditions regulating lung deposition46and numerous technological innovations47have increased lung deposition up to 30–60% of the antibiotic dose deposited in the nebulizer.28,38,41,42,48Mechanisms by which the antibiotic dose delivered to the distal lung is lower than the dose of antibiotic deposited in the nebulizer are summarized in figure 4. Six critical factors influence lung deposition of nebulized antibiotics during mechanical ventilation: aerosol particle size, type of nebulizer, physical characteristics of the carrying gas, respiratory settings, bronchopneumonia severity, and lung aeration.

Only aerosolized particles with mass median aerodynamic diameter ranges between 1 and 5 μm can reach distal bronchioles and alveolar space.49Larger particles do not reach distal respiratory tract because they impact artificial and natural airways (ventilator circuits, endotracheal tube, trachea, and proximal bronchi).50,51Most commercially available nebulizers, such as dry powder inhalers, metered-dose inhalers, jet nebulizers, ultrasonic nebulizers, and vibrating plate nebulizers, produce aerosol particles whose mass median aerodynamic diameter is less than 5 μm. One should keep in mind, however, that the smaller the particle size, the longer the time to deliver a standard antibiotic dose.48 

Nebulizers with large enough reservoir are required for delivering sufficient amounts of antibiotics to the respiratory system. Therefore, metered-dose inhalers are more appropriate for aerosol delivery of bronchodilators and corticosteroids than for antibiotics inhalation.52In fact, particles impaction on ventilator circuits and proximal airways resulting from flow turbulences seems to be the main critical factor reducing distal lung deposition.

Jet nebulizers, which generate the aerosol by superimposing a highly turbulent flow to the inspiratory flow coming from the ventilator, are less efficient in terms of lung deposition than ultrasonic nebulizers53,54: they deliver less than 15% of the initial dose because of high residual volume, massive impaction of aerosol particles in ventilator circuits and endotracheal tube, and loss to atmosphere by the expiratory limb.55 

Ultrasonic nebulizers are equipped with a large reservoir and generate aerosol particles through quartz vibrations, a technique which increases the temperature of the antibiotic solution and may alter the chemical structure of antibiotic molecules. Aerosolized particles are entrained into ventilatory circuits and tracheobronchial tree by a low flow independent of the flow coming from the ventilator. They increased the efficiency of aerosol delivery to 30–40%.28,38 

Vibrating plate nebulizers are equipped with a reservoir placed above a domed aperture plate, which has about 1,000 tapered holes that are electroformed in a sheet. They include a ceramic vibrational element with alternating current–induced expansion and contraction move upward and downward the domed aperture plate, causing a micro-pump effect that produces the aerosol.56,57Particle size depends on the diameter of tapered holes, which can be changed by the manufacturer. Although they seem to be slightly inferior to ultrasonic nebulizers,41vibrating plate nebulizers have several potential advantages over ultrasonic nebulizers46,47: retention in the nebulizer is negligible; the temperature in the reservoir does not increase, minimizing the risk of oxygen desaturation; aerosol generation can be synchronized with inspiration48; and the aerosol can be delivered through an intratracheal catheter inserted in a flexible bronchoscope.58They increased the efficiency of aerosol delivery to 40–60%.41,42 

During experiments in ventilated animals, nebulizers are placed before the Y piece, in parallel to the inspiratory limb. During continuous aerosolization, a bolus of antibiotic is delivered; during inspiration, particles are entrained into the tracheobronchial tree from the tubing between the nebulizer and the Y piece, where aerosolized drug has been collected. During aerosol generation synchronized to the inspiratory phase, the nebulizer is positioned between the Y piece and the proximal tip of the endotracheal tube and the bolus effect is not anymore present. For a given dose of antibiotic, synchronizing aerosol generation with the inspiratory phase significantly increases the duration of nebulization.11Voluminous ultrasonic nebulizers are placed on a fixed support, 40–50 cm before the Y piece, whereas less bulky vibrating plate nebulizers can be directly inserted into the inspiratory limb.

Laminar inspiratory flow provides better distal lung deposition of aerosol particles than turbulent flow.59Turbulence, caused by decelerating flow and angular geometry in the conducting airways, causes aerosolized particles to impact the tubing, trachea, and proximal bronchioles and leads to extrapulmonary deposition and decreased antibiotic deposition in the distal lung. In spontaneously breathing animals, inspiratory flow is highly turbulent and most aerosolized particles impact oropharynx and proximal airways, precluding any significant distal lung deposition. In mechanically ventilated animals, specific ventilator settings should be adopted to limit inspiratory turbulence during the nebulization phase11,46,60: volume-controlled mode using a constant inspiratory flow,61minute ventilation ⩽ 6 l/min, respiratory frequency of 12/min, inspiratory to expiratory ratio ⩽ 50%, and end-inspiratory pause representing 20% of the duty cycle. Decelerating flows should be avoided,62and animal’s triggering and discoordination with the ventilator should also be avoided.62End-inspiratory pause representing 20% of the duty cycle should be used to provide enough time for aerosol sedimentation in the alveolar space.63Because these ventilator settings may not meet physiologic requirements in conscious and awake animals with inoculation pneumonia, deep anesthesia and muscle paralysis are required throughout the experiments to provide full adaptation to the ventilator and optimum distal lung deposition of aerosolized antibiotics.

In most mechanically ventilated patients with ventilator-associated pneumonia, the continuous administration of propofol is required during the nebulization period to allow full adaptation to the ventilator using these specific ventilator settings.64Muscle paralysis is not indispensable, and the short duration of action of propofol allows rapid awakening at the end of the nebulization period. Respecting this rule is a key point of efficiency. To guarantee appropriate antibiotic administration, the nurse in charge of the patient should complete a checklist form ensuring that adequate ventilator settings are used and full adaptation of the patient with the ventilator is obtained with or without propofol administration.64 

Heated and humidified inspiratory gas coming from the ventilator increases mass median aerodynamic diameter of aerosol particles,65increases deposition in ventilator circuits,46and markedly reduces distal lung deposition.48,51The filter of the heat and moisture exchanger offers a barrier to aerosol delivery and should be removed during the nebulization period. If the period of nebulization does not exceed 30 min, the aerosol provides a partial humidification of inspired gas coming from the ventilator and no additional humidification is required. For nebulization longer than 30 min, a conventional humidifier should be inserted on the inspiratory limb to avoid damage of tracheal and bronchial mucosa resulting from the prolonged administration of cold and incompletely humidified inspiratory gas. As a result, the aerosol deposition might be reduced by 40%.46 

Replacing nitrogen/oxygen by helium/oxygen (heliox) reduces density of the inspired gas and increases lung deposition of aerosol particles.66,67Heliox reduces flow turbulence and reduces tracheobronchial wall impaction of aerosol particles. In bench experiments, operating the nebulizer with heliox reduces drug output and disposable mass.68Therefore, to maximize lung deposition, it was initially suggested to generate the aerosol with a nitrogen–oxygen mixture and to entrain aerosol particles with heliox. These in vitro  experiments were confirmed in mechanically ventilated piglets with healthy lungs: Lung deposition of ceftazidime increased by 33% when the ventilator was operated with heliox compared with the classical 65%/35% nitrogen–oxygen mixture.40This beneficial effect, however, was not observed in animals with massive inoculation pneumonia, suggesting that bronchiolitis-induced purulent obstructions were predominant over heliox reduction of flow turbulence.40Therefore, there are not enough experimental data to suggest the use of heliox as the carrying gas for nebulizing antibiotics in animals with inoculation pneumonia.

Ventilator tubings and circuit connections by which aerosol particles are delivered to the tracheobronchial tree should be optimized to limit extrapulmonary deposition. Their inner surface should be smooth and their junctions should avoid acute or right angles. As shown in figure 5, using a special Y piece with obtuse angles reduces the number of aerosol particles hitting the walls of the ventilator circuit. The direct connection of the Y piece to the proximal tip of the endotracheal tube may also decrease impaction within respiratory circuits.

Studies performed in anesthetized piglets on prolonged mechanical ventilation have demonstrated that lung deposition of nebulized amikacin is significantly greater in animals with healthy lungs than in animals with inoculation pneumonia.28,38As shown in figure 6, amikacin lung tissue concentrations are homogeneously distributed in healthy animals and heterogeneously distributed in infected animals. The loss of lung aeration, the severity and extension of parenchymal infection, and the injury to the alveolar–capillary barrier are factors that influence the lung deposition of nebulized antibiotics.

By analogy with histological characteristics of ventilator-associated pneumonia,33the obstruction of distal bronchioles by purulent plugs is a limiting factor for alveolar deposition of aerosolized antibiotics in animals with massive bronchial inoculation pneumonia. In anesthetized and mechanically ventilated piglets whose lungs were infected by bronchial inoculation of E. coli , lung aeration loss had opposite effects in animals receiving intravenous amikacin and in animals receiving aerosolized amikacin.39As shown in figure 7, the loss of lung aeration tended to increase amikacin tissue concentrations in the intravenous group, whereas an opposite effect was observed in the aerosol group. Very likely, the increased permeability of the alveolar–capillary barrier resulting from severe lung infection tends to promote intravenous amikacin penetration into the lung, whereas the multiple purulent plugs obstructing distal bronchioles tend to impair lung deposition of nebulized amikacin.

It should be pointed out that despite the increased permeability of the alveolar–capillary barrier, amikacin lung tissue concentrations remained significantly low after the intravenous injection. Conversely, lung tissue concentrations were always found above minimal inhibitory concentrations within consolidated infected lung regions raising the question how aerosolized antibiotics reach infected lung parenchyma despite the lack of any pulmonary aeration. In fact, as shown in figure 2, intraparenchymal pseudocysts and severe bronchiolar distension are frequently observed in condensed lung areas of ventilated animals with inoculation pneumonia.28,30,37Such lesions of air-space enlargement characterizing ventilator-induced lung injury likely represent one of the routes by which aerosolized antibiotics reach condensed infected lung regions devoid of alveolar aeration. From these experimental data, it can be reasonably hypothesized that ventilator settings aimed at recruiting nonaerated lung areas such as positive end-expiratory pressure or recruitment maneuver may help to increase lung deposition of nebulized antibiotics.

The extension and severity of experimental lung infection can be quantified according to the histological classification proposed in the early 1990s for human ventilator-associated pneumonia and described above.33,69 

In anesthetized piglets on prolonged mechanical ventilation for inoculation pneumonia, lung tissue concentrations of various aerosolized antibiotics were markedly higher in pulmonary segments with early stages of lung infection than in segments with confluent pneumonia and lung abscess.28,39,41,42As shown in figure 8, such differences were not observed when antibiotics were intravenously administered. These experimental data clearly support the administration of aerosolized antibiotics at early stages of ventilator-associated pneumonia.

Antibiotic diffusion through respiratory epithelium (bronchial and alveolar) and pulmonary vascular endothelium markedly influences antibiotics’ lung tissue deposition. Bronchiolar epithelium and normal alveolar–capillary barrier offer a high resistance to lung penetration of intravenous antibiotics and to systemic diffusion of nebulized antibiotics.28As any type of acute lung injury, lung infection results in an increased permeability of the alveolar–capillary barrier,70which in turn facilitates the diffusion of nebulized antibiotics into the pulmonary blood stream and increases systemic bioavailability.38In presence of severe experimental lung infection, amikacin and ceftazidime plasma concentrations were found in the same range after nebulization and intravenous administration.38,41In others words, damage of the alveolar–capillary barrier resulting from the infectious process facilitates leakage of nebulized antibiotics toward the systemic compartment, thereby decreasing the tissue concentrations and increasing the risk of systemic toxicity.

In mammals, the aerodigestive tract above the vocal cords is normally heavily colonized by bacteria, whereas the lower airways and pulmonary parenchyma are sterile. Several mechanisms are involved in removing particulate matter and eliminate microbes that have gained access to the bronchial tree: mucociliary clearance, cough, and cellular and humoral immune systems present along the tracheobronchial tree. Tracheal intubation, deep sedation, and mechanical ventilation weaken host defenses in several ways: cough reflex is depressed or abolished, mucociliary clearance is impaired when inflating endotracheal cuff,71and endotracheal tube injures the tracheal epithelial surface and provides a direct conduit for rapid access of bacteria into the lower respiratory tract. In addition, the antibiotic-resistant bacterial biofilm present on the inner surface of the endotracheal tube is fragmented and disseminated into the deep lung during tracheal suctioning or fiberoptic procedures.72The combination of impaired host defenses and continuous bacterial seeding of the lower respiratory tract by pathogens present in the aerodigestive reservoir produces favorable conditions to the development of deep lung infection. As a consequence, antimicrobial therapy directed against ventilator-associated pneumonia should be efficient not only on lung parenchymal infection but also on bacterial reservoir and tracheobronchial inoculum.

In baboons with acute lung injury resulting from intravenous injection of oleic acid, prolonged exposure to 100% oxygen or endobronchial inoculation of P. aeruginosa , antimicrobial agents applied topically in the oropharynx and trachea can prevent bacterial pneumonia complicating the course of prolonged mechanical ventilation.23,24Polymyxin B and/or gentamicin instilled daily in the posterior oropharynx and in the endotracheal tube significantly reduced oropharyngeal colonization by Gram-negative bacilli, bacterial inoculum in the trachea, and ventilator-associated pneumonia caused by Gram-negative bacilli.24It has to be pointed out that intravenous prophylaxis with gentamicin was ineffective in preventing tracheal colonization and ventilator-associated pneumonia, outlining the limited diffusion of intravenous antibiotics through oropharyngeal mucosa and tracheal epithelium.73Clinical studies have also suggested that endotracheal administration of antimicrobial agents, either as direct bolus administration or nebulization, significantly reduce the incidence of ventilator-associated pneumonia,74,75likely via  a marked reduction of tracheobronchial inoculums.76As shown in figure 9, critically ill patients with ventilator-associated tracheobronchitis had a significant decrease in tracheobronchial bacterial burden after being treated by several days of inhaled vancomycin or gentamicin when compared with the sole intravenous administration of these antibiotics during the same period.76Experimental and clinical data strongly suggest that inhaled antibiotics significantly decrease tracheobronchial bacterial inoculum.

One particular form of bronchial inoculum consists of bacteria growing within the biofilm present on the inner surface of endotracheal tubes. During tracheal suctioning, microbial aggregates are detached from the biofilm and disseminated toward the lung parenchyma by the inspiratory flow coming from the ventilator.77This reservoir of infecting microorganisms cannot be prevented or eradicated by intravenous antibiotics that do not reach the endotracheal tube. It can be prevented by the early nebulization of 80 mg of gentamicin every 8 h.78However, the impact of such a preventive therapy on bacterial ecology of intensive care units has not been assessed, and therefore, it cannot be recommended as a routine therapy. Although not evaluated, it is highly likely that nebulized antibiotics cannot eradicate a preexisting biofilm: microorganisms present within the biofilm are likely protected from nebulized antibiotics by accretion of a protective glycocalyx.79The only way to efficiently remove the endotracheal biofilm is to suction ventilated animals with a catheter equipped with a balloon that is inflated during catheter retrieval.80The lack of evaluation of such a procedure on ventilator-associated pneumonia prevention does not permit to make recommendations concerning its use.

Bactericidal efficiency of aerosolized antibiotics requires obtaining appropriate lung tissue concentrations that kill pathogens infecting the lung parenchyma. It is directly assessed in experimental animals by measuring antibiotic lung tissue concentrations and assessing quantitative bacteriology of postmortem lung tissue samples. Bactericidal activity depends on the pharmacokinetic profile of antibiotics. When concentration-dependent antibiotics are administered, the higher the peak tissue concentration, the greater the bactericidal activity. When time-dependent antibiotics are administered, lung tissue concentrations should remain permanently 10 times greater than minimal inhibitory concentrations.

Antibiotic concentrations measured from homogenized lung represent the total amount of antibiotic present in interstitial and cell compartments. Most pathogens and particularly Gram-negative bacilli do not penetrate into cells and remain in the interstitial space where antibiotics exert their bactericidal activity by binding to bacterial cell membrane. Therefore, antibiotic concentrations measured from homogenized lung biopsies tend to underestimate “effective” interstitial concentrations because of a dilution factor caused by intracellular components.81 

More than 30 yr ago, it was reported that aerosolized kanamycin provided higher bacterial killing and survival rates in spontaneously breathing mice and squirrel monkeys with K. pneumoniae  inoculation pneumonia, compared with intramuscular administration.82,83In the same model, the authors demonstrated that aerosolized kanamycin administered before intratracheal instillation of K. pneumoniae  prevented the onset of bronchopneumonia, whereas the intramuscular administration was not protective.84Pharmacokinetic data demonstrated that nebulized kanamycin remained in the lungs longer and at higher concentrations compared with an intramuscular administration. Similar results were later on obtained in spontaneously breathing guinea pigs with P. aeruginosa  inoculation pneumonia: a combination of aerosolized and intramuscular tobramycin achieved higher rates of survival and total eradication of P. aeruginosa  from lungs compared with nebulized or intramuscular tobramycin alone.85 

The first experimental study looking at the antibacterial efficiency of nebulized antibiotics was performed in anesthetized piglets ventilated for a severe E. coli  inoculation pneumonia.28Twenty-four hours after massive bacterial inoculation, ventilated animals were treated either by aerosolized or intravenous amikacin, a concentration-dependent antibiotic. To deliver equivalent doses to the respiratory system (extrapulmonary deposition represented 60% of the dose inserted into the nebulizer chamber), 45 mg/kg were nebulized and 15 mg/kg intravenously administered in a single dose. Animals received a second dose and were killed 1 h after the second dose, and five subpleural specimens were sampled from upper, middle, and lower lobes. After nebulization, amikacin lung tissue peak concentrations were 3–30 folds higher than after intravenous administration. After two doses (25-h treatment), 71% of lung segments were sterile, whereas cultures of lung segments were comparable in nontreated and intravenously treated animals.

Two studies looked at the bactericidal activity of ceftazidime, a time-dependent antibiotic, administered to anesthetized piglets ventilated for P. aeruginosa  inoculation pneumonia.40,41In the first study, 24 h after intrabronchial bacterial inoculation, ventilated animals received either aerosolized or intravenous ceftazidime.40To deliver comparable doses to the respiratory system (extrapulmonary deposition represented 30% of the dose inserted into the nebulizer chamber), 50 mg/kg was nebulized and 33 mg/kg intravenously administered in a single dose and animals were killed 1 h later. Five subpleural specimens sampled from upper, middle, and lower lobes served for assessing ceftazidime tissue concentrations. After nebulization, ceftazidime lung tissue concentrations were 5–30 fold higher than after intravenous administration.40In the second study performed in the same experimental conditions, ceftazidime was administered for 24 h using either repetitive aerosols each 3 h or continuous intravenous administration.41After eight aerosols of 25 mg/kg or a continuous intravenous administration during 24 h, animals were killed and five subpleural specimens were sampled from upper, middle, and lower lobes for assessing ceftazidime tissue concentrations and lung bacterial burden. After nebulization, ceftazidime lung tissue trough concentrations were three to six folds higher than after intravenous administration. As shown in figure 10, after eight ceftazidime aerosols (24-h treatment), 83% of lung segments were sterile, whereas cultures of lung segments were comparable in nontreated and intravenously treated animals.

A fourth study was performed to compare bactericidal activity of aerosolized and intravenous colistin, a concentration- and time-dependent antibiotic for Gram-negative bacteria, in anesthetized piglets ventilated for P. aeruginosa  inoculation pneumonia.42Twenty-four hours after intrabronchial bacterial inoculation, ventilated animals received either intravenous or aerosolized colistin using vibrating plate nebulizers. To deliver comparable amount of colistin to the respiratory system (extrapulmonary deposition represented 40% of the dose inserted into the nebulizer chamber), 16 mg·kg⁻¹·day⁻¹were administered via  two aerosols and 9 mg·kg⁻¹·day⁻¹were intravenously administered in three daily doses. Animals were killed 1 h after the third aerosol and 1 h after the fourth intravenous administration, and five subpleural specimens were sampled from upper, middle, and lower lobes for assessing colistin tissue concentrations and lung bacterial burden. After intravenous administration, colistin remained undetectable in the infected lung parenchyma. After nebulization, colistin lung tissue peak concentrations were one to seven folds higher than minimal inhibitory concentrations. After four colistin aerosols (24-h treatment), 67% of lung segments were sterile, whereas more than 70% of lung segments were massively infected in nontreated and intravenously treated animals.42This result confirms the lack of efficiency of parenteral colistin for treating experimental pneumonia, as previously demonstrated in spontaneously breathing mice whose lungs had been massively infected by intratracheal inoculation of multidrug resistant P. aeruginosa  and A. baumannii .86,87 

These experimental studies, performed in different animal species with inoculation pneumonia caused by various Gram-negative bacteria, clearly suggest that concentration-dependent antibiotics, such as aminoglycosides or colistin, and time-dependent antibiotics, such as cephalosporins, have a higher bactericidal efficiency when administered by nebulization than by the intravenous route.

The alveolar–capillary barrier offers a difficult-to-cross obstacle for drugs reaching the pulmonary circulation. After intravenous administration, many antibiotics have a limited diffusion, resulting in insufficient lung tissue penetration. A recent experimental study performed in ventilated piglets with P. aeruginosa  inoculation pneumonia documented the absence of any lung deposition of intravenously administered colistin.42experimental studies have shown that antibiotic lung tissue concentrations are much lower after intravenous than after aerosol administration.28,38,40–42Acute experimental pneumonia is characterized by pulmonary vasoconstriction and regional thrombosis, two vascular injuries reducing lung perfusion and impairing pulmonary penetration of circulating antibiotics.

Bypassing the alveolar–capillary membrane by reaching alveoli via  the bronchial tree is the basic rationale for delivering aerosolized antibiotics. By analogy with the difficulty of crossing the alveolar–capillary membrane from the pulmonary circulation, it was initially believed that systemic diffusion from the alveolar space of aerosolized antibiotics would be reduced, thereby limiting their toxicity. Unfortunately, for most antibiotics, experimental data did not confirm such a benefit. As shown in figure 11, the intact alveolar–capillary membrane is a limiting factor for the systemic diffusion of antibiotics present in the alveolar space.38Lung infection, however, markedly increases the permeability of the alveolar–capillary membrane and promotes systemic diffusion of aminoglycosides and cephalosporins.28,40,41As a consequence, expecting less systemic toxicity for these antibiotics by using the aerosol route is likely unwarranted. This prediction, however, will have to be confirmed by prospective studies including a large number of patients.

A recent experimental study performed in ventilated piglets with P. aeruginosa  inoculation pneumonia has clearly shown that intravenous colistin does not cross the alveolar–capillary membrane, whereas aerosolized colistin has a limited systemic diffusion.42After colistin nebulization, the high lung deposition was associated with rapid and efficient bacterial killing, although systemic exposure was reduced. After intravenous colistin, the lack of lung deposition was associated with lack of bacterial killing, confirming a recent study performed in mice with pneumonia caused by multidrug resistant P. aeruginosa  where intravenous colistin did not confer any survival protection.87Aerosolized or intravenous colistin, which is negatively charged at body pH, is confronted with alveolar and endothelial membranes also characterized by negative charges.88The alveolar basement membrane is well known to inhibit the systemic passage of anionic molecules present in alveolar space, thus explaining why aerosolized colistin has a limited systemic diffusion and intravenous colistin a weak lung penetration.42,89As a consequence, a dosage exceeding threefold the dose commonly reported in the literature90can be used for treating pneumonia caused by sensitive P. aeruginosa  without increasing the risk of toxicity.

One generally believes that aerosol as adjunctive to intravenous therapy offers an attractive alternative to intravenous or nebulized therapy alone. The hypothesis that a combination of nebulized and intravenous antibiotics could increase lung tissue concentrations and accelerate bacterial killing was previously tested in four experimental piglets whose lungs were infected by a massive bronchial inoculation of E. coli  and who received a combination of nebulized and intravenous amikacin (data were part of a previously published study).28Unfortunately, no additional increase in lung tissue concentrations was measured, whereas increased trough systemic concentrations were observed, increasing the risk of systemic toxicity. In this experimental study, high amikacin concentrations were found in lymphatic vessels, suggesting a large absorption of amikacin into lymphatic vessels of lung interstitial space. It has to be pointed out that intravenous amikacin, in contrast to colistin, diffuses into the alveolar space and that nebulized amikacin diffuses into the systemic circulation in presence of lung infection. As described above, there is solid evidence that lung deposition of intravenous colistin is extremely reduced, if not null. Therefore, it is hazardous to expect that combining nebulized and intravenous colistin might increase colistin tissue concentrations. In contrast, increased systemic concentrations resulting from intravenous administration may increase the risk of renal toxicity. Colistin, unlike aminoglycosides and β-lactams, does not cross the alveolar–capillary barrier of the infected lung parenchyma. Therefore, the rationale for combining nebulized and intravenous colistin as a treatment of ventilator-associated pneumonia seems weak. In fact, there is a strong rationale for treating bacteremic ventilator-associated pneumonia caused by resistant P. aeruginosa  or A. baumannii  with a combination of nebulized and intravenous colistin. Such a combination remains the only therapeutic option. Unfortunately, our model of inoculation pneumonia is extremely rarely associated with positive blood cultures and does not allow the assessment of this specific issue.

Few complications have been described in experimental animals receiving aerosolized antibiotics. Bronchoconstriction and hypoxemia may result from the aerosol nebulization of any drug, particularly in animals with pneumonia. Another serious potential complication of nebulized antibiotics could be the emergence of multiresistant pathogens. Experimental studies did not find any increase in the incidence of resistant pathogens when polymyxin B or colistin were endotracheally administered to prevent ventilator-associated pneumonia.24When aminoglycosides are nebulized into the tracheobronchial tree, significant concentrations are found in the serum.28,38Most often, these concentrations are below minimal inhibitory concentrations of Gram-negative bacteria, a fact which may promote the emergence of aminoglycoside-resistant strains if an extrapulmonary infection is present. As a consequence, it seems reasonable to limit the administration of nebulized antibiotics to isolated and nonbacteremic ventilator-associated pneumonia.

Experimental studies have shown that aerosolized antibiotics for treating ventilator-associated pneumonia differ in several ways from inhaled bronchodilator therapy for treating bronchospasm or severe asthma. Aerosolized antibiotics must penetrate into the distal lung, whereas bronchodilators should reach exclusively the bronchial tree. Because antibiotic tissue concentrations in the infected lung parenchyma should markedly exceed minimal inhibitory concentrations of pathogens, optimizing dosage is a more critical issue for inhaled antibiotic therapy than for inhaled bronchodilator therapy.

As shown in figure 12, it is of critical importance to use adequate nebulizers and appropriate circuit connections and modify ventilator settings during the nebulization period for providing laminar inspiratory flow. The appropriate dose of aerosolized antibiotic should be determined according to extrapulmonary deposition and the interval between two aerosols. Last, but not least, any discoordination with the ventilator should be avoided to prevent inspiratory turbulence and impaction of aerosolized antibiotics on circuit walls and upper airways. If these conditions are strictly respected, then high lung tissue deposition associated with rapid and efficient bacterial killing can be expected. For aerosolized aminoglycosides and cephalosporins, a decrease in systemic exposure leading to reduced toxicity is not proven by experimental studies likely because the permeability of the alveolar–capillary membrane markedly increases with lung parenchymal infection. Aerosolized colistin, however, does not cross the alveolar–capillary membrane even in the presence of severe lung infection, and high doses can be delivered by nebulization without significant systemic exposure and risk of toxicity.

Rubin Luo, M.D., Research Fellow, Mao Zhang, M.D., Ph.D., Medical Director (Department of Emergency Medicine, Second Affiliated Hospital, Zhejiang University, School of Medicine, Hangzhou, China); Marine Lecorre, M.D., Fabio Ferrari, M.D., Ph.D., Corinne Vezinet, M.D., Liliane Bodin, M.D. (Multidisciplinary Intensive Care Unit, Department of Anesthesiology and Critical Care Medicine, La Pitié-Salpêtrière Hospital, Assistance-Publique-Hôpitaux-de-Paris, UPMC Univ Paris 06, France); Charles-Hugo Marquette, M.D., Ph.D. Medical Director of the Department of Respiratory Diseases, Pasteur Hospital, University of Nice-Sophia Antipolis, Nice, France; Olivier Petit-Jean, Ph.D., Director of the Department of Pharmacology, Avicenne Hospital, Assistance Publique-Hôpitaux de Paris, Bobigny, France; Alexandra Aubry, M.D., Ph.D., Jérôme Robert, M.D., Ph.D. (Department of Bacteriology, La Pitié-Salpêtrière Hospital, Assistance-Publique-Hôpitaux-de-Paris, UPMC Univ Paris 06); Noêl Zahr, Pharm.D., Ph.D., Christian Funck-Brentano, M.D., Ph.D., Medical Director, Department of Pharmacology, La Pitié-Salpêtrière Hospital, UPMC Univ Paris 06, France.