Plasma drug concentrations in relation to the minimum inhibitory concentration (MIC) for the bacterial pathogen of interest have traditionally been used as a predictive marker of the efficacy of antibacterial agents.[1] In vitro models and clinical trials during the past two decades have dramatically expanded our knowledge of exposure-effect relationships and have established pharmacokinetic-pharmacodynamic indices for many antibacterial agents.[2–4] Hence more attention has been placed on the importance of anti-infective concentrations, not only in plasma but also at the site of infection.[5–7]

Pulmonary epithelial lining fluid (ELF) and alveolar macrophages (AMs) have been advocated as important sites of infection with common extracellular and intracellular respiratory pathogens, respectively. For community-acquired pneumonia, Streptococcus pneumoniae, Moraxella catarrhalis and Haemophilus influenzae are the predominant extracellular pathogens, and Mycoplasma pneumoniae, Chlamydophila (Chlamydia) pneumoniae and Legionella pneumophila are the primary intracellular pathogens. Thus knowledge of the location of the respective pathogens, drug penetration and drug concentrations in different compartments of the lung should assist in selection of appropriate antibacterial therapy and design of dosing regimens to effectively treat lower respiratory tract infections.[8–11]

The aim of this review is to provide a comprehensive summary of the concentrations of antibacterial agents in ELF and the extent of their penetration. A brief overview has been included to describe the common methodologies that are used in collecting samples from ELF, the limitations associated with these methods and interpretation of the data that are obtained. Also discussed are the physicochemical and host-related factors that may influence penetration of anti-infective agents at the site of lung infections, comparison of ELF to plasma (or serum) concentration ratios, and correlation of pharmacokinetic-pharmacodynamic parameters that are specific for each class of antibacterial agents.

1. Epithelial Lining Fluid (ELF)

A wide variety of different methodologies have been used for measuring concentrations of anti-infective agents and drug distribution patterns within the lungs. Commonly employed methods have included whole-tissue homogenates, sputum, respiratory secretions, bronchial mucosa, pleural fluid, bronchoalveolar lavage (BAL), ELF, microdialysis, positron emission tomography and magnetic resonance spectroscopy.[12] It remains unclear which of these techniques is the most appropriate.[11–15]

Historically, anti-infective drug concentrations were measured by obtaining lung tissue during a surgical procedure. Although this is one of the oldest methods for measuring drug concentrations in the lung, whole-tissue concentrations are no longer recommended.[14] In addition, methods that include measurement of antibacterial concentrations in sputum or bronchial secretions have been considered less than optimal.[11,12,16] For example, the shortcomings of using sputum concentrations as an indicator of lung penetration include the possibility of falsely low saliva drug concentrations (indicating poor lung penetration) because of a dilutional effect caused by saliva, or falsely elevated saliva drug concentrations (suggesting high lung penetration) if the antibacterial is extensively distributed into saliva but not into other pulmonary fluids or tissues. The major drawback of drug concentrations reported from whole lung tissue, bronchial tissues and/or secretions is the assumption that anti-infective agents are uniformly distributed across all lung compartments (e.g. extracellular, intracellular). Thus the reported drug concentration represents a mixture from all compartments versus a value for a specific site of activity.

Methods that measure concentrations of anti-infective agents in specific compartments of the lung (e.g. ELF, interstitial fluid, AMs) are currently preferred and are considered to provide further insight into the significance of drug concentrations at the site of pulmonary infections.[5,11–14,16] ELF has been considered the likely site of extracellular respiratory micro-organisms in bacterial pneumonia and acute bacterial exacerbation of chronic bronchitis. However, one must understand that no method or sampling site is perfect, and each has its advantages, potential limitations and methodological issues.[11–16]

1.1 Collection of ELF and Limitations of Methodology

Currently, the combination of bronchoscopy and BAL has become a safe and effective method for obtaining samples from the bronchoalveolar surface of the lower respiratory tract. The procedure involves a fibre-optic bronchoscope being passed and wedged into a subsegment of the middle or lower lung lobe. BAL is performed by instillations of three or four aliquots of sterile 0.9% normal saline solution into the lung lobe, with the fluid being immediately aspirated and placed on ice after each aliquot. The aspirate from the first instillation is usually collected separately and discarded because of significant contamination with cells from the proximal airways. The aspirates recovered from the subsequent instillations are pooled, the volume is measured and recorded, and removal of a sample for other testing (e.g. a total cell count and a differential cell count) is sent to the laboratory. The remaining volume of BAL is immediately centrifuged, the supernatant and cell pellet are separated, and samples from the supernatant for determining urea and drug concentrations are collected and frozen until the assays can be performed. A blood sample to determine urea and drug concentrations is also obtained at the time of the scheduled bronchoscopy and BAL procedure.

The recovered BAL fluid is a mixture of saline, ELF and cellular components of the ELF. In order to determine the anti-infective drug concentration, the ‘apparent’ volume of ELF must be estimated. Urea (or albumin or creatinine) has been commonly used as an endogenous marker to estimate the volume of ELF.[17] Urea concentrations are normally the same in plasma and in ELF, since urea is non-polar and has a low molecular weight that allows it to rapidly diffuse and establish an equilibrium across the capillary-alveolar membrane. By measuring the urea in BAL aspirate and plasma, a ratio can be calculated and the dilution of the apparent volume of ELF can be estimated (equation 1):

$${\rm{ELF\;volume}} = {{{\rm{Volume\;aspirated}} \times {\rm{Urea}}\;({\rm{BAL\;aspirate}})} \over {{\rm{Urea}}\;({\rm{plasma}})}}$$
((Eq. 1))

Using the calculated value of the ELF volume, the drug concentration in ELF (CELF) can be estimated as follows (equation 2):

$${{\rm{C}}_{{\rm{ELF}}}} = {{{\rm{Volume\;of\;BAL}}} \over {{\rm{ELF\;volume}}}} \times {{\rm{C}}_{{\rm{BAL}}}}$$
((Eq. 2))

where CBAL is the drug concentration measured in BAL.

The techniques of bronchoscopy and BAL are associated with a low level of technical complexity.[12] However, our review of the literature indicates that the technique for performing BAL differs among investigators, with the main variations being the dwell time, the aspiration pressure, the volume of fluid injected and the number of BAL aspirates collected.[11,15,16,18] In addition, different assay methodologies and analytical kits are used to measure urea in BAL and plasma. Each of these variables may influence the correction for ELF dilution and the apparent volume of ELF. It has been suggested by some researchers that overestimation of the ELF volume (and thus underestimation of drug concentrations) can range from 100% to 300%.[9,11,13,16] It has also been suggested that lysis of cells (e.g. AMs) in the collected BAL fluid increases the measured concentrations of anti-infective agents in ELF.[12,13,15] Finally, alterations in the ELF volume, cell counts and protein concentrations in BAL have been associated with subjects who smoke and/or have interstitial lung disease.[19,20] Therefore it is recommended that a standardized BAL procedure be established and used consistently to ensure that technical errors are minimized and that distal ELF samples are obtained. The assays for urea in plasma and BAL need to be precise and reliable. Subject recruitment needs to be carefully considered in order to minimize unexpected variability. Accurate understanding of the methodology and strict adherence to the procedures are crucial to minimize known sources of error.

Other shortcomings include the difficulty of obtaining multiple ELF samples from the same subject at several sampling timepoints, because of the nature of the BAL procedure.[12,13] As a result, clinical studies require a larger number of subjects who are randomized to a single BAL sampling time. Several studies from Japan have been able to overcome this issue by using bronchoscopy with bronchoscopic microsampling (BMS) probes.[21–25] This technique obtains samples of ELF from the bronchial space (whereas BAL samples are obtained from the alveolar space) and does not require measurement of urea concentrations. Differences in methodology and sampling sites explain, in part, why anti-infective drug concentrations have not always been similar when BMS and BAL were performed in the same subjects.

1.2 Factors Influencing Penetration of Antimicrobials into ELF and Comparison with Plasma Concentrations

There are many factors that influence antimicrobial penetration into the pulmonary compartments.[8,9,11–13,15,16] Anti-infective agents must cross the blood-bronchus barrier to reach the alveolar ELF and the lung interstitium in order to achieve adequate concentrations. This barrier is composed of an alveolar-capillary membrane. The capillary endothelium is relatively permeable; however, the alveolar membrane is virtually impermeable because of the presence of tight junctions. The major mode of transport between membranes occurs through passive diffusion. Protein binding determines the amount of the unbound drug that is present to diffuse through membranes, explaining why antimicrobials with a high degree of protein binding do not penetrate well into lung tissue. The degree of lipophilicity, as well as administration of the drug in an un-ionized form, determine how well the antimicrobial will penetrate tissue.[8,9,11–13,15,16] Other factors that favour penetration include the pH at the site of infection and other dynamic processes such as bronchial inflammation at the site of infection.[5,8,11]

Concentrations of anti-infective agents in ELF are considered to provide useful and accurate estimates for determining extracellular drug penetration in the lungs. Various methods have been suggested to compare plasma and ELF concentrations of anti-infective agents. Nearly all studies report comparisons of concentrations (e.g. ELF and plasma concentrations) that were simultaneously obtained at both sites. In fact, that is how the data are reported in our tables for the various classes of antibacterial agents. However, since the concentration-time profiles of drugs in plasma and ELF can increase and decrease at different paces from each other (known as system hysteresis), penetration ratios will vary in magnitude with the sampling time(s) chosen. To overcome this issue, it is recommended that research studies determine penetration ratios from estimates of the area under the concentration-time curve (AUC) values of plasma and ELF data.[26,27] Accurate estimates of the AUC can be obtained even when drug concentrations are only sparsely sampled. Where ratios based on AUC estimates have been reported, we have incorporated them into the text of our review.

1.3 Pharmacokinetic-Pharmacodynamic Parameters of Efficacy

Our knowledge and understanding of the pharmacokinetic-pharmacodynamic parameters of anti-infective agents have grown substantially over the past 30 years.[2–4] Application of these parameters has had a major influence on the dose selection, the dosing interval and/or mode of administration, development of in vivo susceptibility breakpoint determinations, assessment of bacterial resistance and the clinical drug development process. This information has had an impact on how anti-infective agents are used at the bedside and developed in research.

Pharmacokinetic-pharmacodynamic parameters that correlate with the efficacy of specific anti-infective agents include (i) the percentage of time during which unbound (free) drug concentrations remain above the MIC of an infecting micro-organism (fT>MIC) [in the case of β-lactams (e.g. penicillins, cephalosporins, carbapenems, monobactams), linezolid or oritavancin];[2–4,7] (ii) the ratio of the maximum unbound concentration (fCmax) to the MIC of an infecting micro-organism (fCmax/MIC) [in the case of aminoglycosides, fluoroquinolones, daptomycin, oritavancin or metronidazole];[2–4,7] and (iii) the ratio of the unbound AUC during a 24-hour time period (fAUC24) to the MIC of an infecting micro-organism (fAUC24/MIC) [in the case of fluoroquinolones, aminoglycosides, azithromycin, clarithromycin, telithromycin, doxycycline, tigecycline, vancomycin, daptomycin, telavancin, linezolid, clindamycin or quinupristin/dalfopristin].[2–4,6,7]

The concepts and categorization of the pharmacokinetic-pharmacodynamic parameters of anti-infective agents have been based on relationships established from plasma or serum concentrations.[2–4] We are unaware of any clinical studies that have evaluated and documented similar correlations with concentrations at the site of infection (e.g. ELF) for community- or hospital-acquired lower respiratory tract infections.[6,7] While many of the reports that are included in this review have described pharmacokinetic-pharmacodynamic parameters by using ELF concentration-time data, caution must be applied in presuming that the magnitude of exposure required in ELF and/or the relationship between parameters and outcomes are the same or even similar to those observed in plasma. Studies are needed in patients with lower respiratory tract infections to confirm and explore the relationships between ELF concentrations, clinical and microbiological outcomes, and pharmacodynamic parameters.

2. β-Lactams

Intrapulmonary penetration studies have been conducted with β-lactam agents since the early 1990s. Oral and intravenous cephalosporins have accounted for the majority of these studies (tables I and II).[28–41] β-lactam–β-lactamase inhibitor combinations (e.g. amoxicillin/clavulanic acid, piperacillin/tazobactam) have been evaluated for the penicillin class. Several carbapenem agents (e.g. meropenem, ertapenem, biapenam) have been studied. No ELF data have been reported for the monobactam aztreonam.

Table I
figure Tab1

Plasma and epithelial lining fluid (ELF) concentrations of oral penicillins and cephalosporins

Full size image
Table II
figure Tab2

Plasma and epithelial lining fluid (ELF) concentrations of parenteral penicillins and cephalosporins

Full size image

2.1 Oral Penicillins and Cephalosporins

Intrapulmonary studies of oral penicillins and cephalosporins have usually been conducted after a single oral dose (table I). The concentrations of amoxicillin and clavulanic acid in ELF were found to be similar to each other (mean values 0.89 and 0.96 µg/mL, respectively); however, these values were less than 20% of the total plasma concentrations.[28] The concentrations of oral cephalosporins in ELF were commonly observed to be less than 1 µg/mL and were often reported as undetectable.[29–35,42] The ratios of ELF concentrations to total plasma concentrations ranged from 0.1 to 0.38, suggesting low to moderate penetration into the ELF. The exception to this occurred following multiple doses of a modified-release formulation of cefaclor 750 mg.[34] The ELF concentrations of cefaclor were similar to the total plasma concentrations (mean values 2.71 and 3.08 µg/mL, respectively, at 4 hours) or exceeded the total plasma concentrations (mean values 2.16 and 0.68 µg/mL, respectively, at 6 hours; and 0.60 and 0.23 µg/mL, respectively, at 12 hours).

Several methodological limitations of the intrapulmonary studies of oral penicillins and cephalosporins should be noted. Sampling times were often restricted in number and were often limited to the first few hours after antimicrobial administration, thus they did not include the entire dosing interval. The ratios of ELF concentrations to plasma concentrations were based on concentrations at these specific sampling times as opposed to a more accurate comparison being made by estimating AUCs for ELF and plasma. These issues probably contributed to the large difference between plasma and ELF concentrations, secondary to system hysteresis (e.g. concentrations in plasma and ELF often increase and decrease at different paces from each other). In addition, the ratios of ELF to plasma concentrations were based on total (versus unbound) plasma drug concentrations. Finally, the lower level of detection for measuring drug concentrations was limited, since two-thirds of the studies used microbiological assays. While a few recent studies have overcome most of these limitations, the ELF concentrations of cefuroxime were still undetectable throughout the entire dosing interval, and the percentages of penetration by ceftibuten and cefdinir into ELF were still less than 40%.[30,32,33]

2.2 Parenteral Penicillins and Cephalosporins

Intrapulmonary studies of parenteral penicillins and cephalosporins have included a variety of subject types, modes of drug administration and differences in BAL collection techniques (table II). In critically ill patients with severe nosocomial pneumonia, three studies have been conducted using microlavage to determine ELF and plasma concentrations of piperacillin-tazobactam, ceftazidime and cefepime.[36,38,39] The mean (± standard deviation [SD]) concentrations of piperacillin and tazobactam following at least 2 days of therapy with an intravenous dosage regimen of 4 and 0.5 g every 8 hours were 13.6 ± 9.4 and 2.1 ± 1.1 µg/mL, respectively, which were approximately 57% and 91%, respectively, of the total plasma concentrations.[36] The same researchers used a similar study design and methods to evaluate the intrapulmonary penetration of ceftazidime and cefepime in critically ill patients receiving continuous intravenous infusions of 4 g/day.[38,39] The mean (± SD) concentrations of ceftazidime and cefepime were 8.2 ± 4.8 and 14.1 ± 2.8 µg/mL, respectively. However, the respective ratios of ELF to total plasma concentrations were 0.21 and 1.04. Unlike those of other parenteral cephalosporins whose ELF to total plasma concentration ratios range between 7% and 36%, the concentrations of cefepime in ELF were similar to or exceeded the total and unbound plasma concentrations (protein binding of cefepime is approximately 20%).[40,43] The discordance in penetration among the parenteral cephalosporins does not seem to be related to differences in structure or protein binding (range 10–20%).

A novel study design was used to evaluate drug exposure in ELF of ceftobiprole for the treatment of pneumonia caused by meticillin (methicllin)-resistant Staphylococcus aureus.[41] Population pharmacokinetic modelling of ELF and plasma concentrations from both animal and human data were used in Monte Carlo simulations to evaluate the probabilities of achieving a desired target attainment based on bacterial killing rates obtained from a preclinical pneumonia model in neutropenic mice.[44] For a dosage regimen of ceftobiprole 500 mg every 8 hours as a 2-hour intravenous infusion, the probabilities of achieving 1-log and 2-log cell kill rates were 85.6% and 79.7%, respectively, over the MIC distribution of 4958 isolates. This study provided a logical approach for evaluating dosage regimens for the treatment of lower respiratory tract infections, based on preclinical infection models, the pharmacokinetic-pharmacodynamic characteristics of the agent being studied and drug exposure information at the site of infection (e.g. the lung). In addition, this study allowed differences in ELF penetration in mice versus man (68.8% vs 25.5%) to be accounted for and dosage selection in humans to be evaluated.[41,44]

2.3 Carbapenems

Single- and multiple-dose studies have evaluated the intrapulmonary penetration of meropenem and ertapenem (table III).[45–48] The mean concentrations of meropenem in ELF and the penetration ratios ranged from 7.07 to 0.59 µg/mL and from 0.51 to 1.04, respectively, in 30 subjects administered a single 1 g dose intravenously over 30 minutes.[45] A similar range of ELF and plasma concentrations was observed following multiple doses of 1 g every 8 hours in healthy adult subjects.[46] In both studies, concentrations of meropenem in ELF at 8 hours were undetectable or extremely low (mean 0.03 µg/mL). While plasma concentrations increased proportionally as doses of meropenem were increased, concentrations in ELF tended to decrease as meropenem doses increased. The ratios of ELF concentrations to total plasma concentrations ranged from 0.49 to 0.80 for the 500 mg dose, from 0.32 to 0.53 for the 1 g dose and from 0.048 to 0.219 for the 2 g dose.[46] It is unclear why this disproportionate percentage change in ELF concentrations occurred. In contrast to these findings, a mean penetration ratio (based on ratios of the AUC) of 0.72% was observed in critically ill patients with ventilator-associated pneumonia receiving 3-hour infusions of meropenem 2 g administered every 8 hours.[49] Subsequently, pharmacokinetic simulations based on the mean parameter vector of these data suggested that ELF concentrations of meropenem are maintained above MIC values of ≤1 µg/mL for nearly 30% of the dosing interval.[50] Similar median penetration ratios were observed with Bayesian parameter estimates of patients with ELF sampling (0.264) and a 9999-subject Monte Carlo simulation (0.254).

Table III
figure Tab3

Plasma and epithelial lining fluid (ELF) concentrations of carbapenems

Full size image

Two studies have evaluated the penetration of ertapenem into ELF.[47,48] In 15 adult patients undergoing thoracotomy, the mean ELF concentrations of ertapenem ranged from 2.11 to 4.06 µg/mL during the first 5 hours after a single perioperative dose of 1 g.[47] The mean ratios of ELF concentrations to total plasma concentrations ranged from 0.062 to 0.094. In comparison, adult patients treated for early-onset ventilator-associated pneumonia had median ELF concentrations of 9.4 and 0.3 µg/mL at 1 and 24 hours, respectively, after multiple doses of ertapenem 1 g once daily.[48] The ratios of ELF concentrations to total plasma concentrations ranged from 0.21 to 0.64 (median 0.32). Since ertapenem is highly protein bound (e.g. 96% at 10 µg/mL and 84% at 300 µg/mL), the penetration ratios in both studies are closer to or greater than 1 if unbound plasma concentrations are considered.

Kikuchi et al.[22] compared ELF concentrations in six healthy adult subjects administered a single dose of biapenem 300 mg as 0.5-hour and 3-hour intravenous infusions. The mean (± SD) ELF concentrations were 3.48 ± 1.20 µg/mL at the end of the 0.5-hour infusion and 1.33 ± 0.26 µg/mL at the end of the 3-hour infusion. The mean penetration ratios of ELF concentrations to total plasma concentrations were 0.20 for both infusion periods. These authors also evaluated a BMS technique, which used a polyester fibre rod probe to repeatedly obtain ELF concentrations at the surface of a bronchus (termed ‘bronchial ELF’). In contrast to the BAL sampling technique, bronchial ELF concentrations were higher with the 3-hour infusion (4.36 ± 2.07 µg/mL) than with the 0.5-hour infusion (2.35 ± 1.06 µg/mL). No explanations were provided for the discrepancies between the two techniques.

3. Macrolides, Azalides and Ketolides

Intrapulmonary studies have been commonly conducted for agents from the antibacterial classes of macrolides, azalides and ketolides (tables IV–VII).[23,24,30,51–68] The majority of reported studies have evaluated clarithromycin, azithromycin and telithromycin. Although the number of studies is small, data are available for older macrolide agents such as erythromycin, roxithromycin and dirithromycin, as well as the investigational ketolide cethromycin.

Table IV
figure Tab4

Plasma and epithelial lining fluid (ELF) concentrations of oral erythromycin, roxithromycin, dirithromycin and modithromycin

Full size image
Table V
figure Tab5

Plasma and epithelial lining fluid (ELF) concentrations of oral clarithromycin

Full size image
Table VI
figure Tab6

Plasma and epithelial lining fluid (ELF) concentrations of azithromycin

Full size image
Table VII
figure Tab7

Plasma and epithelial lining fluid (ELF) concentrations of oral ketolides

Full size image

3.1 Erythromycin, Roxithromycin, Dirithromycin and Modithromycin

Intrapulmonary concentrations of erythromycin were first reported in 1995 from a multiple-dose study in 12 healthy adult subjects.[51] The mean (± SD) plasma concentrations and ELF concentrations at 4 hours after nine doses of oral erythromycin 250 mg every 6 hours were similar: 0.7 ± 0.2 and 0.8 ± 0.1 µg/mL, respectively. Unfortunately, the two other sampling times (8 and 12 hours) in this study were outside the dosing interval, and the concentrations of erythromycin were reported as undetectable (assay sensitivity = 0.05 µg/mL). Patients treated with oral erythromycin 250 mg every 6 hours for 28 days had measurable ELF concentrations at 2, 4 and 8 hours after the last dose administration.[69] The mean 2-hour ELF concentration of erythromycin was 0.97 µg/mL, and a similar 4-hour concentration was observed in the healthy subject study.

In eight patients with mild chronic bronchitis treated with oral roxithromycin 300 mg every 12 hours for 5 days, the ELF concentration (mean ± SD 2.0 ± 1.7 µg/mL) was only 17% of the simultaneously measured plasma concentration (mean ± SD 8.1 ± 4.0 µg/mL).[52] In comparison, the concentration of roxithromycin in AMs was significantly higher (mean ± SD 21.0 ± 10.0 µg/mL). In a study evaluating the immunoregulatory effects of roxithromycin in patients with chronic respiratory tract infections, the ELF concentrations were measurable in only two of seven patients (1.56 and 4.77 µg/mL) who had received 3 months of oral roxithromycin 300 mg every 12 hours.[53] The plasma concentrations in these two patients were 2.55 and 8.78 µg/mL, respectively.

Dirithromycin was one of the first macrolide agents for which ELF concentrations were reported to be significantly higher than concurrent plasma concentrations.[54] In 25 patients treated for acute exacerbation of chronic bronchitis, the mean ELF concentrations during the first 24–72 hours after five oral doses of dirithromycin 500 mg once daily ranged from 1.57 to 2.25 µg/mL. Subsequent mean ELF concentrations were 0.22 µg/mL at 96 hours and 0.15 µg/mL at 120 hours. In comparison, the mean plasma concentrations ranged from 0.44 µg/mL at 24 hours to 0.11 µg/mL at 120 hours. A similar range of ELF concentrations was observed between 2 and 24 hours (approximately 2.37 µg/mL) after the last oral dose of dirithromycin 500 mg once daily in 20 patients treated for acute exacerbation of mild chronic obstructive pulmonary disease.[69]

The investigational bridged bicyclic macrolide (bicyclolide) modithromycin (also known as S-013420 and EDP-420) has recently been studied in healthy Japanese subjects.[55] Following a single oral dose of 400 mg, the mean concentrations of modithromycin in ELF between 2 and 24 hours ranged from 5.8 to 16.7 µg/mL. The mean plasma concentrations during this same time period ranged from 0.315 to 0.646 µg/mL. As with other newer agents, the mean concentrations of modithromycin in AMs (range 61–176 µg/mL) were well above those observed in plasma and ELF.

3.2 Clarithromycin

Two single-dose studies of clarithromycin have suggested that ELF concentrations are measurable only within the first 6 hours after administration of oral doses of 200 and 500 mg (table V).[23,30] Four multiple-dose studies of the immediate-release formulation of clarithromycin have reported plasma and intrapulmonary concentrations following the oral dosage regimen of 500 mg twice daily.[51,56–58] The mean concentrations of clarithromycin in ELF ranged between 20.46 and 32.4 µg/mL at 4 hours and between 15.1 and 48.6 µg/mL at 12 hours. Plasma concentrations were 10- to 60-fold lower and ranged between 2.0 and 3.96 µg/mL at 4 hours and between 0.8 and 1.22 µg/mL at 12 hours. The reported concentrations were fairly consistent between studies. Plasma and ELF concentrations of the active metabolite, 14-hydroxy-clarithromycin, were also reported in two studies. Concentrations of 14-hydroxy-clarithromycin in ELF (range 1.2–15.09 µg/mL) were greater than concurrent plasma concentrations (0.4–6.41 µg/mL), but the difference (1.1- to 7.45-fold) was less than that observed with clarithromycin.

One study evaluated the intrapulmonary disposition of the extended-release formulation of clarithromycin 1000 mg once daily.[59] Following five doses, the mean concentrations in ELF were 11.50 µg/mL at 9 hours and 6.80 µg/mL at 24 hours, and they remained constant throughout the 24-hour dosing interval. The mean plasma concentrations were 2.22 µg/mL at 9 hours and 0.75 µg/mL at 24 hours. The concentrations of clarithromycin in AMs ranged between 43 and 1087 µg/mL (mean 303 µg/mL) and were comparable to values reported with the immediate-release formulation of clarithromycin. Maximum concentration (Cmax) values in plasma, ELF and AMs occurred at 9 hours with the extended-release formulation, which was later than with the immediate-release formulation.

In vitro and in vivo murine pharmacodynamic models have made use of clinically observed plasma and ELF concentrations to evaluate the bactericidal activity and/or postantibacterial effect of clarithromycin against isolates of H. influenzae and/or S. pneumoniae.[70–73] In addition, Monte Carlo simulations have been used to assess the ability of pharmacodynamic parameters (e.g. an AUC/MIC ratio of ≥25) to determine susceptibility breakpoints and target attainment rates against penicillin-susceptible and -resistant strains of S. pneumoniae.[74,75] Overall, these studies support the concept that the higher concentrations of clarithromycin in ELF contribute to the potential efficacy of clarithromycin against S. pneumoniae strains with MIC values up to 8 µg/mL.

3.3 Azithromycin

Six studies (two single-dose studies and four multiple-dose studies) have evaluated the intrapulmonary disposition of oral azithromycin (table VI). Interestingly, the single-dose studies provided minimal information, since most plasma and ELF samples had undetectable concentrations.[30,60] The multiple-dose studies indicated that the mean ELF concentrations ranged between 0.64 and 3.12 µg/mL during the first 24 hours after the last dose of the standard oral dosing regimen of azithromycin.[57,58,61,62] The concentrations of azithromycin in ELF were fairly stable and demonstrated minimal fluctuations throughout the 24-hour dosing interval. The mean plasma concentrations during this same time period ranged from 0.03 to 0.1 µg/mL. No study has reported ELF to plasma concentration ratios. However, the ELF concentrations were approximately 10- to 20-fold higher than the concurrent plasma concentrations.

Single oral doses of azithromycin in the immediate-release and extended-release formulations were compared in 64 patients who had been diagnosed with lung cancer and required open-chest surgery for lung resection.[76] For the immediate-release formulation, a single 500 mg dose resulted in AUC from 0 to 24 hours (AUC24) values in serum and ELF of 3.1 and 2.3 µg · h/mL, respectively. Following a 2 g dose of the extended-release formulation, the AUC24 values in serum and ELF were 10.0 and 17.6 µg · h/mL, respectively. Higher exposure was also observed in AMs and whole lung tissues with the extended-release formulation.

Only one study has reported plasma and ELF concentrations following multiple intravenous doses of azithromycin 500 mg once daily.[63] The plasma concentrations at the time of bronchoscopy were comparable (range 0.09–0.49 µg/mL) to those associated with oral dosing once the differences in the dose and oral bioavailability were taken into account. Concentrations in ELF ranged from 0.79 to 5.86 µg/mL and were slightly higher than the values associated with oral dosing. The estimated AUC24 of azithromycin in ELF (45.8 µg · h/mL) was approximately 5.6-fold higher than the AUC24 in plasma (8.2 µg · h/mL) following intravenous dosing.

The observed azithromycin concentrations in ELF and plasma have been used to investigate in vitro bactericidal activity and pharmacodynamic target attainment rates against clinical isolates of S. pneumoniae and H. influenzae. In ELF, azithromycin demonstrated greater potency, killing rates and postantibacterial effects than clarithromycin against H. influenzae and tended to be adequate for eradication of macrolide-susceptible S. pneumoniae (suggested susceptibility breakpoint ≤2 µg/mL).[73,74,77–80] The probability of achieving target attainment rates was lower with azithromycin than with clarithromycin.[74]

3.4 Ketolides

Several intrapulmonary studies have reported ELF concentrations for oral telithromycin dosage regimens of 600 and 800 mg once daily (table VII).[24,64–68] Like clarithromycin and azithromycin, telithromycin achieves higher and sustained extracellular concentrations (2- to 17-fold) in ELF than in plasma. The mean ELF concentrations following telithromycin 800 mg once daily for 5 days were approximately 5.0 and 0.9 µg/mL at 2 hours and 24 hours, respectively. The total plasma concentrations at these same sampling times averaged 1.9 and 0.1 µg/mL, respectively. The reported mean ratios for ELF to total plasma concentrations at individual sampling times ranged from 2 to 14.9. A mean penetration ratio of 7.95 (with AUC values of 83.73 and 10.53 µg · h/mL for ELF and plasma, respectively) has been reported, based on a population pharmacokinetic analysis and a 10 000-subject Monte Carlo simulation.[81] These ratios would be greater if unbound plasma concentrations are considered (telithromycin plasma protein binding: 60–70%). Intracellular concentrations of telithromycin are even higher in AMs than in ELF and can range from 50- to >500-fold greater than total plasma concentrations.[64–67]

For the investigational ketolide antibacterial cethromycin, the ELF concentrations between 2 and 24 hours ranged from 0.1 to 0.9 µg/mL after the 150 mg once-daily dosage regimen and from 0.1 to 2.7 µg/mL after the 300 mg once-daily dosage regimen (table VII).[68] In comparison, the plasma concentrations were approximately 10-fold lower than the ELF concentrations, whereas the AM concentrations were 10-fold higher than the ELF concentrations.

Several pharmacodynamic studies have evaluated the in vitro activity and target attainment potential of telithromycin against pathogens commonly associated with community-acquired pneumonia (e.g. macrolide-susceptible and -resistant S. pneumoniae, and β-lactamase-positive and -negative H. influenzae).[73,82–85] Using simulated total and unbound serum and ELF concentrations of telithromycin, these studies have provided recommendations for potential pharmacokinetic-pharmacodynamic parameter values of the Cmax/MIC and AUC24/MIC ratios associated with bacteriostatic and bactericidal activity, as well as no regrowth of telithromycin-susceptible bacteria over 24–48 hours.[73,82,83]

4. Fluoroquinolones

Most fluoroquinolones that have been developed during the past 25 years have been studied to determine the degree of lung penetration. Ciprofloxacin, moxifloxacin and levofloxacin are currently the most commonly recommended fluoroquinolones for the treatment of lower respiratory tract infections. With that in mind, our review focuses on intrapulmonary penetration studies of those three fluoroquinolones only. However, a list of references has been provided for those readers who have an interest in similar studies of other fluoroquinolones.[86–100]

4.1 Ciprofloxacin

Several single- and multiple-dose studies have evaluated intrapulmonary penetration of oral ciprofloxacin in healthy adult subjects and patients undergoing diagnostic bronchoscopy (table VIII).[96,98–101] The concentration-time profile following a single oral dose of 500 mg suggests that ciprofloxacin does not achieve adequate ELF concentrations throughout the dosing interval. Although the mean plasma and ELF concentrations were similar (2.33 and 2.13 µg/mL, respectively) at 2.5 hours, subsequent ELF concentrations (range 5–48 hours) were below the lower limit of quantification (LLQ).[99,100] When multiple oral doses of ciprofloxacin 500 mg were administered every 12 hours, the mean concentrations were slightly lower in ELF than in plasma throughout the dosing interval (2.11 vs 1.87 µg/mL at 4 hours and 0.55 vs 0.41 µg/mL at 12 hours).[101] Two studies that measured ELF concentrations of ciprofloxacin between 3 and 6 hours after multiple doses of an oral regimen of 250 mg every 12 hours reported that ELF concentrations (2.0 and 3.0 µg/mL at 3 and 6 hours, respectively) were higher than concurrent plasma concentrations (1.1 and 1.19 µg/mL at 3 and 6 hours, respectively).[96,98]

Table VIII
figure Tab8

Plasma and epithelial lining fluid (ELF) concentrations of oral ciprofloxacin and moxifloxacin

Full size image

4.2 Moxifloxacin

Concentrations of moxifloxacin in ELF have been reported in two studies involving oral administration (table VIII).[62,102] After a single oral dose of moxifloxacin 400 mg, the mean concentrations in ELF ranged from 20.7 µg/mL at 2.2 hours to 3.57 µg/mL at 24.1 hours.[102] The mean ratios of ELF concentrations to total plasma concentrations ranged from 5.19 to 6.95. Similarly, after multiple oral doses of moxifloxacin 400 mg every 24 hours in older adult subjects (mean age 67 ± 10 years), the mean plasma concentrations ranged from 3.23 µg/mL at 4 hours to 0.78 µg/mL at 24 hours and the mean ELF concentrations ranged from 11.66 µg/mL at 4 hours to 5.71 µg/mL at 24 hours.[62] Even if protein binding (approximately 50%) is taken into account, ELF concentrations remain 2.5- to 3.5-fold higher than unbound plasma concentrations throughout a 24-hour dosing interval. A recent study using a Calu-3 lung epithelial cell model suggested that P-glycoprotein-mediated active transport mechanisms influence the secretion of moxifloxacin into the lung and may explain, in part, why higher concentrations are observed in ELF than in plasma.[103]

The intrapulmonary data generated from these two studies have been used in numerous evaluations to compare the pharmacodynamic characteristics of moxifloxacin with those of other antibacterial agents that are commonly used in lower respiratory tract infections.[77,79,104–110] Most investigations have used simulated plasma and ELF concentrations in vivo to link in vitro bactericidal killing activity, eradication rates against various genotypes and/or resistant patterns of S. pneumoniae and S. aureus, prevention of selection or emergence of resistance, and achievement of AUC/MIC target attainment rates. These studies have provided further support for the concept that ELF concentrations contribute to the efficacy of moxifloxacin against pathogens associated with lower respiratory tract infections.

4.3 Levofloxacin

Concentrations of levofloxacin in ELF have been investigated extensively (table IX).[25,62,63,101,111–115] Studies have included single and multiple doses of levofloxacin administered both orally and intravenously at a wide range of dosages (e.g. 100–1000 mg). The individuals who were studied included healthy adult subjects, patients undergoing diagnostic fibre-optic bronchoscopy, critically ill patients with severe community-acquired pneumonia and outpatients with a clinical diagnosis of mild to moderate chronic bronchitis, chronic obstructive pulmonary disease or lower respiratory tract infection.

Table IX
figure Tab9

Plasma and epithelial lining fluid (ELF) concentrations of levofloxacin

Full size image

Eleven of 13 studies used plasma and ELF sampling over a 24-hour interval.[62,63,101,111–115] The mean ratios of ELF concentrations to total plasma concentrations at individual sampling times ranged from 1.0 to 4.9 when the samples were obtained between 4 and 24 hours after dosing. Several studies reported AUC24 values in plasma and in ELF. For the 500 mg dose, the mean AUC24 values ranged from 34.5 to 180 µg · h/mL in ELF, compared with 50.1–86.7 µg · h/mL in plasma.[62,63,112] For the 750 mg dose, the mean AUC24 values in ELF and in plasma were 151.4 and 95.4 µg · h/mL, respectively.[63] A 1000 mg dose resulted in mean AUC24 values of 260–279.1 µg · h/mL in ELF and 103.6–130 µg · h/mL in plasma.[114,115] The ratios of ELF concentrations to plasma concentrations that were estimated from these AUC values ranged from 1.59 to 2.69.

Drusano et al.[116] performed population pharmacokinetic modelling and Monte Carlo simulations based on steady-state concentrations in plasma and ELF after oral doses of levofloxacin 500 and 750 mg were administered to healthy adult subjects. The penetration ratio based on the AUC values in ELF and in plasma, derived from the mean parameter vector, was 1.16. The mean and median ratios were 3.18 (SD 5.71) and 1.43 (95% confidence interval [CI] 0.143, 19.12), respectively, when calculations were based on 1000-subject Monte Carlo simulations. The penetration ratio was greater than 1 in 61% of simulations.

Other investigators have also evaluated pharmacodynamic target attainment rates using estimates of ELF concentrations of levofloxacin in hospitalized patients with community-acquired pneumonia and in a murine pneumonia model of Pseudomonas aeruginosa infection.[117,118]

5. Aminoglycosides

Aminoglycosides such as gentamicin, tobramycin and netilmicin are most often used intravenously or via inhalation for the treatment of serious respiratory infections involving Gram-negative organisms. Aminoglycosides are known for poor lung penetration and varied concentrations in the lung tissue depending on the anatomical site that is sampled. Intrapulmonary penetration and ELF concentrations of aminoglycosides in critically ill patients with lower respiratory tract infections have been evaluated following intravenous and intramuscular administration of single and multiple doses (table X).[119–124]

Table X
figure Tab10

Plasma and epithelial lining fluid (ELF) concentrations of aminoglycosides

Full size image

Sampling of ELF concentrations in aminoglycoside studies has been limited to the first 8 hours after drug administration, and no estimation of exposure throughout the dosing interval has been provided. In general, concentrations of aminoglycosides in ELF are significantly lower than plasma concentrations during the first 1.5 hours after administration. Subsequently, the concentrations in ELF become similar to those in plasma at approximately 2 or 3 hours after administration. For example, the mean concentrations of tobramycin in ELF were 2.7 µg/mL (approximately 12% of the plasma concentration [22.4 µg/mL]) at 0.5 hours following a 30-minute intravenous infusion of 7–10 mg/kg.[122] Similar ELF concentrations (e.g. <3 µg/mL) were observed at 1 hour after a single dose of gentamicin 240 mg (mean ± SD 3.5 ± 0.1 mg/kg) and during the first 2 hours of adjusted doses of tobramycin.[119,121] Although the mean concentrations of netilmicin in ELF were higher at 1 and 1.5 hours (7.5 and 9.6 µg/mL, respectively) after a single dose of 450 mg, these ELF concentrations were only 35% and 62% of the concurrent mean plasma concentrations (21.4 and 15.3 µg/mL, respectively).[124] Once the Cmax values of gentamicin and netilmicin in ELF occurred at 2 hours (4.24 and 14.7 µg/mL, respectively), the ELF to plasma concentration ratios were 0.85 and >1, respectively.[119,124] Subsequently, ELF and plasma concentrations of the aminoglycosides became similar to each other. Thus system hysteresis has greatly influenced the large variability associated with reported ELF to plasma penetration ratios that have been observed in the limited number of intrapulmonary studies.

6. Glycopeptides and Lipoglycopeptides

6.1 Vancomycin

Vancomycin has been used for over 50 years to treat serious Gram-positive infections, including meticllin-resistant S. aureus. A great deal of attention has recently been paid to appropriate dosing of vancomycin, particularly in lower respiratory tract infections.[125,126] However, only a limited number of studies have evaluated ELF concentrations of vancomycin and assessed its plasma and intrapulmonary disposition (table XI).[127–131]

Table XI
figure Tab11

Plasma and epithelial lining fluid (ELF) concentrations of glycopeptides and lipoglycopeptides

Full size image

Lamer et al.[127] investigated plasma and ELF concentrations of vancomycin in 14 critically ill adult patients (mean ± SD Acute Physiology and Chronic Health Evaluation II [APACHE II] score 18.7 ± 6) who were mechanically ventilated and had signs and symptoms of lower respiratory tract infection. The patients received an initial dose of intravenous vancomycin 15 mg/kg, and subsequent doses were adjusted to achieve a trough plasma concentration between 15 and 20 µg/mL. Blood and BAL samples were collected at an average of 6.6 days (range 5–11 days) after starting treatment with a cumulative dose of 9.4 g (range 3–17.5 g). Concentrations of vancomycin in ELF ranged from 0.4 to 8.1 µg/mL, and a significant linear correlation (r = 0.64; p < 0.02) was observed between plasma and ELF concentrations. Patients with lung inflammation (ELF albumin concentration ≥3.4 mg/mL; n = 7) had a significantly higher (p < 0.02) vancomycin ELF to plasma concentration ratio (mean 0.246; range 0.192–0.426) than patients without inflammation (mean 0.14; range 0.023–0.285) and a normal ELF albumin concentration (<3.4 mg/mL; n = 7).

Georges et al.[128] also evaluated vancomycin plasma and ELF concentrations in ten critically ill adult patients (mean ± SD age 65.5 ± 8.4 years) with meticillin-resistant S. aureus pneumonia requiring mechanical ventilation. Plasma and ELF concentrations of vancomycin were measured 24 hours after starting intravenous vancomycin 7.5 mg/kg every 6 hours, infused over 1 hour. Four of ten patients had measurable concentrations of vancomycin in ELF (range 1.38–2.77 µg/mL), and the concomitant trough plasma concentrations ranged from 20.9 to 23.2 µg/mL. In the six patients with undetectable concentrations of vancomycin in ELF, the trough plasma concentrations ranged from 7.7 to 18.1 µg/mL. The authors of both studies[127,128] recommended that trough plasma concentrations of vancomycin should be approximately 20 µg/mL to ensure that adequate ELF concentrations are achieved.

Plasma and intrapulmonary concentrations of vancomycin in ten healthy adult subjects were measured at 4 and 12 hours after the start of the ninth dose of intravenous vancomycin 1000 mg every 12 hours (table XI).[129] In addition, a population pharmacokinetic analysis and Monte Carlo simulations were conducted.[132] The ratio of ELF to total plasma concentrations was 0.5, based upon AUC values for each matrix. If protein binding of vancomycin in healthy adult subjects is assumed to be approximately 50%, ELF and unbound plasma vancomycin concentrations were essentially of the same magnitude.

An in vitro model evaluated the effects of simulated ELF concentrations and penetration ratios on bacterial killing of meticillin-resistant S. aureus and development of resistance.[133] Bactericidal activity could not be demonstrated for dosage regimens that achieved an unbound AUC24/MIC ratio of 350 with a penetration ratio of 1.0. Development of resistance was suppressed when the unbound AUC24/MIC ratio was ≥280 with a penetration ratio of ≥0.8. These findings suggest that current dosage regimens, which recommend trough plasma vancomycin concentrations of 15–20 µg/mL, are not adequate for bactericidal activity, even when the ELF penetration ratio is equal to 1.0. Development of resistance (e.g. heterogeneous) is likely to occur if MIC values are >1 µg/mL, the unbound AUC24/MIC ratio is <280 and/or the ELF penetration ratio is <0.8.

6.2 Teicoplanin

Teicoplanin is another glycopeptide that is commercially available in many countries other than the US. One study compared steady-state serum and ELF concentrations of intravenous teicoplanin in 13 critically ill patients with ventilator-associated pneumonia who received 12 mg/kg twice daily for 2 days followed by 12 mg/kg once daily (table XI).[130] The total and unbound serum concentrations ranged from 8.8 to 29.9 µg/mL and from 2.0 to 5.4 µg/mL, respectively. The concentrations of teicoplanin in ELF ranged from 2.0 to 11.8 µg/mL, and the ratio of ELF to unbound serum concentrations ranged from 0.48 to 3.32 (median 1.46).

6.3 Telavancin

Telavancin is a lipoglycopeptide developed for the treatment of serious Gram-positive infections, including meticillin-resistant S. aureus. Steady-state plasma and intrapulmonary concentrations of telavancin were evaluated in 20 healthy adult subjects who were administrated 10 mg/kg intravenously once daily for 3 days (table XI).[131] The mean concentrations of telavancin in ELF at 8 and 24 hours after the last dose were 3.73 and 0.89 µg/mL, respectively. The mean plasma concentrations at 12 and 24 hours were 22.9 and 7.28 µg/mL, respectively. The plasma protein binding of telavancin is approximately 90%. A subsequent population pharmacokinetic analysis and Monte Carlo simulation study reported mean and median values for the ratio between the ELF AUC and the unbound plasma AUC of 1.01 and 0.73, respectively.[134] The large variability (SD 0.96) of the ELF AUC from time zero to infinity (AUC∝) skewed the mean value towards 1.

6.4 Oritavancin

Oritavancin is an investigational lipoglycopeptide being developed for the treatment of serious Gram-positive infections, including meticillin-resistant S. aureus. Plasma and bronchopulmonary concentrations of oritavancin were collected in 20 healthy adult subjects for up to 168 hours after the fifth intravenous dose of 800 mg every 24 hours.[127] The mean concentrations of oritavancin in ELF (range 3.1 µg/mL at 4 hours to 1.7 µg/mL at 168 hours) were lower than the total plasma concentrations (range 119.6–10.4 µg/mL) [table XI]. If protein binding for oritavancin in healthy adult subjects is assumed to be approximately 90%, the ELF and unbound plasma oritavancin concentrations were similar.

7. Miscellaneous Antibacterial Agents

Several studies have reported intrapulmonary concentrations of agents from other antibacterial classes (table XII).[135–139] Currently, linezolid and tigecycline are available for clinical use and have been approved for the treatment of lower respiratory tract infections. Iclaprim is an investigational agent that is undergoing clinical development.

Table XII
figure Tab12

Plasma and epithelial lining fluid (ELF) concentrations of linezolid, tigecycline and iclaprim

Full size image

7.1 Linezolid

Linezolid is an oral and parenteral antibacterial agent that is representative of the oxazolidinone class. Linezolid is used in the treatment of serious lower respiratory tract infections caused by Gram-positive pathogens such as meticillin-resistant S. aureus. The intrapulmonary disposition of linezolid has been investigated in healthy adult subjects, patients with chronic obstructive pulmonary disease and critically ill patients requiring mechanical ventilation.

Plasma and bronchopulmonary concentrations of linezolid were measured in 25 healthy adult subjects for up to 48 hours after the fifth oral dose of 600 mg every 12 hours.[135] The mean concentrations and systemic exposure of linezolid in ELF (range 64.3 µg/mL at 4 hours to 0.7 µg/mL at 48 hours; AUC24 672 µg · h/mL) exceeded those in plasma (range 15.5–0.2 µg/mL; AUC24 204.2 µg · h/mL) and in AMs (range 2.2 µg/mL to below the LLQ; AUC24 30.0 µg · h/mL) [table XII]. For ELF, the AUC24/MIC ratio was 168 and the T>MIC was 100% for a 12-hour dosing interval at an MIC required to inhibit the growth of 90% of organisms (MIC90) of 4 µg/mL for S. aureus.

Plasma and intrapulmonary concentrations of linezolid were measured between 2.9 and 7.5 hours following the sixth oral dose of 600 mg every 12 hours in ten subjects undergoing diagnostic bronchoscopy and BAL.[136] Concentrations of linezolid in ELF were detectable in nine subjects and ranged from 13.1 to 52.4 µg/mL (table XII). The concomitant plasma and AM concentrations ranged from 8.6 to 19.6 µg/mL and from 0.5 to 23.7 µg/mL, respectively. The mean ± SD ELF to plasma concentration ratio was 8.35 ± 11.69 and quite variable over the ten sampling times.

Plasma and ELF concentrations of linezolid were determined at steady state in 16 critically ill adult patients with ventilator-associated pneumonia.[137] Following 2 days of intravenous therapy with linezolid 600 mg every 12 hours, bronchoalveolar microlavage (with 40 mL of sterile 0.9% normal saline solution) was performed twice in each subject at 1 and 12 hours after the start of the last dose. Concentrations of linezolid in ELF ranged from 4.5 to 25.4 µg/mL (mean 14.4 µg/mL) at 1 hour and from 0.3 to 5.8 µg/mL (mean 2.6 µg/mL) at 12 hours (table XII). The ELF to plasma concentration ratio was approximately 1 at both sampling times, and the mean AUC24 in plasma was 154.6 µg · h/mL.

7.2 Tigecycline

Tigecycline is a parenteral glycylcycline and belongs to the tetracycline class of anti-infective agents. Tigecycline has shown activity against a wide spectrum of pathogens, including Gram-positive, Gram-negative, atypical and anaerobic pathogens.

Steady-state plasma and intrapulmonary concentrations of tigecycline were measured in 30 healthy adult subjects following the last dose of an intravenous regimen, which included a loading dose of 100 mg followed by 50 mg every 12 hours for a total of seven doses.[138] The mean concentrations of tigecycline in serum and ELF were similar during the entire 12-hour dosing interval and ranged from 0.10 to 0.19 µg/mL and from 0.06 to 0.37 µg/mL, respectively (table XII). In contrast, the mean concentrations in AMs were significantly higher (p < 0.05) and ranged from 10.7 to 15.2 µg/mL. A population pharmacokinetic analysis and Monte Carlo simulations were subsequently conducted to determine systemic exposure (based on AUC24 values) and penetration ratios.[140] The median ELF to serum concentration ratio was 1.15 and was associated with wide variability (5th and 95th percentiles 0.561 and 5.23, respectively).

It is worth noting that the protein binding of tigecycline in humans ranges from 71% to 89%, and the data reported above used total serum concentrations from uninfected subjects. Two recent studies in animal models have suggested that (i) a significantly higher penetration ratio is seen when protein binding is accounted for and unbound serum concentrations are used; (ii) ELF to unbound serum concentration ratios are significantly higher in mice with a pulmonary infection (12.9 and 23.3 at doses of 25 and 50 mg/kg, respectively) than in those without a pulmonary infection (6.2 and 8.1 at doses of 25 and 50 mg/kg, respectively); and (iii) the penetration ratio is dose dependent and tends to increase with larger doses.[141,142] Further studies are needed to confirm these observations in patients with lower respiratory tract infections.

7.3 Iclaprim

Iclaprim is a dihydrofolate reductase inhibitor and a member of the diaminopyrimidine class of anti-infective agents. Iclaprim has potent in vitro activity against Gram-positive pathogens, including meticillin-resistant S. aureus and Streptococcus species.

Plasma and intrapulmonary concentrations of iclaprim were measured in 24 healthy male subjects from 1.8 to 7.08 hours after the start of a 1-hour intravenous infusion of 1.6 mg/kg of bodyweight.[139] The concentrations of iclaprim in ELF (range 0.5–25.3 µg/mL) exceeded the concomitant plasma concentrations at all sampling times with the ratios of ELF to total plasma concentrations ranging from 6 to 51.2 (table XII). The MIC90 values for Gram-positive pathogens such as penicillin-susceptible and -intermediate S. pneumoniae (0.06 and 2 µg/mL, respectively) and meticillin-resistant S. aureus (0.12 µg/mL) were exceeded by the concentrations of iclaprim in ELF for up to 7 hours. The concentrations of iclaprim in AMs were similar to or higher than those in ELF.

8. Conclusions

The penetration of antibacterial agents into the ELF of the lung has been extensively evaluated during the past 20 years. This review includes more than 80 studies that have reported ELF concentrations and extracellular penetration ratios of antibacterial agents that are used to treat lower respiratory tract infections. More studies (e.g. >100) would have been included in the review if we had elected to include all reports on fluoroquinolones instead of just those on ciprofloxacin, moxifloxacin and levofloxacin.[25,62,63,86–102,111–115] In spite of this, fluoroquinolones still accounted for approximately 25% of the studies that we reviewed.

A wide variety of penetration ratios and patterns of ELF disposition have been observed for agents from various antibacterial classes. Aminoglycosides, β-lactams and glycopeptides tend to have ELF to total plasma concentration ratios of ≤1. Aminoglycosides tend to have ELF concentrations that are fairly constant and significantly lower than plasma concentrations in the first few hours after drug administration, with subsequent ELF concentrations being similar to or slightly lower than plasma concentrations during the last half of the dosing interval. A wide range of penetration ratios has been reported for β-lactam agents. Even when protein binding and differences in the characteristics of the molecule are taken into account, the discordance in penetration could not be predicted for various agents from the β-lactam classes. Although the number of studies of glycopeptide or lipoglycopeptide agents (e.g. vancomycin, telavancin, oritavancin) is limited, once protein binding is accounted for, the pattern of disposition and the ratios of ELF to unbound plasma concentrations are similar and approach values closer to 1.

In comparison, antibacterial agents such as macrolides, ketolides, fluoroquinolones (e.g. levofloxacin and moxifloxacin) and linezolid have ELF to total plasma concentration ratios of >1. Why agents from these antibacterial classes tend to have higher concentrations in ELF than in plasma remains unknown. Recent studies have suggested that active transporter systems in epithelial cells may be involved and may influence the secretion of these agents in the lung.[103,143–145] On the other hand, technical issues associated with the collection of BAL samples and measurement of ELF concentrations have been raised as other potential explanations for these observations.[12,13,15] Some authors have provided rationales as to why these confounding variables may not explain the higher observed ELF concentrations. Obviously, more research is needed to understand the underlying mechanisms of drug transporters in the lung, and strict adherence to sample collection procedures is required in order to minimize known sources of error.

The majority of studies have been conducted in adults who were healthy or outpatients undergoing an elective diagnostic fibre-optic bronchoscopy. The BAL procedure has usually consisted of 150–200 mL of 0.9% normal saline solution being instilled as three or four aliquots. A few recent studies have targeted older outpatients or patients with a clinical diagnosis of mild to moderate chronic bronchitis, chronic obstructive pulmonary disease or lower respiratory tract infection.[53,54,62,69,112,115] The reported ELF concentrations were similar to or slightly higher than those observed in healthy subjects. An increasing number of studies have provided ELF data in critically ill patients who were in an intensive care unit, receiving mechanical ventilation and being treated for severe pneumonia.[36,39,40,49,113,118,119,121,122,124,126–128,130,137] In these patients, microlavage or minilavage (e.g. 20 mL of 0.9% normal saline solution instilled as one or two aliquots per sampling time) has commonly been used for the BAL procedure.[18] Despite differences in study procedures and patient populations, the data on the antibacterial agents that have been studied so far (e.g. piperacillin/tazobactam, ceftazidime, cefepime, meropenem, levofloxacin, aminoglycosides, vancomycin, teicoplanin, linezolid) have shown that ELF concentrations were similar to those observed in healthy subjects and outpatients. Thus antibacterial concentrations in ELF from non-infected subjects tend to serve as conservative estimates of the likely drug concentrations at extracellular sites of lung infection. However, further studies are still warranted in patients with pulmonary infections to explore the importance of intrapulmonary concentrations with respect to establishing relationships between pharmacokinetic-pharmacodynamic parameters and clinical or microbiological outcomes.

The interpretation of antibacterial concentrations in ELF is greatly influenced by the study design, sample collection and timing, analytical methods and data analysis. Our review of the literature and our work as investigators in these types of studies have provided us with an appreciation of the following key issues that need to be considered when reviewing and/or designing intrapulmonary penetration studies: (i) multiple-dose studies are more likely to provide meaningful information and measurable ELF concentrations; (ii) serial sampling times should be spaced throughout a dosing interval in order to provide accurate estimates of AUCs in plasma and in ELF; (iii) procedures for collection, handling and storage of BAL samples are required, including careful separation of ELF and cell components; (iv) studies should involve experienced personnel using established bronchoscopy and BAL procedures that are individualized to the type of subjects being studied; (v) analytical methodologies must be sensitive and specific for measurement of urea and drug concentrations in plasma and in ELF; and (vi) pharmacokinetic-pharmacodynamic data analysis and mathematical modelling should take into account protein binding, systemic exposure, simulations of drug concentration profiles and evaluation of target attainment rates. Ideally, a logical approach for evaluating dosage regimens — based on preclinical infection models, the pharmacokinetic-pharmacodynamic characteristics of the agent being studied and drug exposure information at the site of infection (e.g. ELF in the lung) — should be incorporated into the study design.[41] All of these considerations are important to the understanding of ELF penetration and appropriate decision-making to guide dosage regimen designs for patients with lower respiratory tract infections.[7]

This review has focused on human studies that have measured ELF concentrations of antibacterial agents. An increasing number of intrapulmonary penetration studies have been conducted in animals. Antibacterial agents that have and have not been studied in humans have been evaluated for applications in veterinary medicine for foals, horses and dogs.[146–155] In addition, animal studies evaluating different modes of administration or formulations of current and investigational antibacterial agents have been reported.[156–160] These studies are also providing further insights into the penetration of antibacterial agents into ELF.