**DOI:**10.1128/AAC.01110-07

## ABSTRACT

Telavancin is an investigational bactericidal lipoglycopeptide with a multifunctional mechanism of action, as demonstrated against methicillin-resistant *Staphylococcus aureus*. While the plasma pharmacokinetics have been described, the extent of the penetration of the drug into the lung, measured by the epithelial lining fluid (ELF), remains unknown. Population modeling and Monte Carlo simulation were employed to estimate the penetration of telavancin into ELF. Plasma and ELF pharmacokinetic data were obtained from 20 healthy volunteers, and the pharmacokinetic samples were assayed by a validated liquid chromatography-tandem mass spectrometry technique. Concentration-time profiles in plasma and ELF were simultaneously modeled using a three-compartment model with zero-order infusion and first-order elimination and transfer. The model parameters were identified in a population pharmacokinetic analysis (BigNPAG). Monte Carlo simulation of 9,999 subjects was performed to calculate the ELF/plasma penetration ratios by estimating the area under the concentration-time curve (AUC) for the drug in ELF (AUC_{ELF}) and for the free drug in plasma (free AUC_{plasma}) from zero to infinity after a single dose. After the Bayesian step, the overall fits of the model to the data were good, and plots of predicted versus observed concentrations in plasma and ELF showed slopes and intercepts very close to the ideal values of 1.0 and 0.0, respectively. The median AUC_{ELF}/free AUC_{plasma} penetration ratio was 0.73, and the 25th and 75th percentile value ratios were 0.43 and 1.24, respectively. In uninfected lung tissue, the median AUC_{ELF} is approximately 75% of the free AUC_{plasma}.

*Staphylococcus aureus* is the most frequent cause of health care-associated pneumonia in the United States, leading to considerable morbidity and mortality (10, 26). Treatment of *S. aureus* infections has been complicated by the emergence of *S. aureus* isolates expressing resistance to methicillin (5, 16, 22, 24). Infections caused by methicillin-resistant *S. aureus* (MRSA) account for >50% of all *S. aureus* strains isolated in many institutions. The increasing incidence of cross-resistant MRSA strains is further complicating treatment decisions (5, 16, 22, 24). Against this background of multidrug resistance, vancomycin has emerged as the drug of choice against MRSA. Despite its favorable susceptibility profile against MRSA, several reports within the past 10 years have described MRSA strains with intermediate susceptibility or high-level resistance to vancomycin, and some also question vancomycin's reduced activity against MRSA strains with MICs at the high end of the susceptible range (2, 3, 20, 28, 31).

Telavancin, an investigational bactericidal lipoglycopeptide antibiotic, has been shown to have multifunctional mechanisms of action against MRSA (14, 15, 18). While the plasma pharmacokinetics have been described (29, 30), the ability of telavancin to penetrate and concentrate in the lung, as measured by the epithelial lining fluid (ELF), has not been fully elucidated. This study describes the population pharmacokinetics of telavancin in the plasma and ELF in healthy volunteers through nonparametric population pharmacokinetic modeling and Monte Carlo simulation. These analyses estimate the range of ELF penetration likely to be observed in this population (a surrogate for clinical practice), as measured by the ratio of the area under the concentration-time curve (AUC) for the drug in ELF to the AUC for the free drug in plasma (AUC_{ELF}/free AUC_{plasma} ratio).

(This research was presented in part as a poster at the 43rd Annual Meeting of the Infectious Diseases Society of America, San Francisco, CA, October 2005 [T. P. Lodise, L. Ma, M. Gotfried, S. Barriere, and G. L. Drusano, abstr. 526].)

## MATERIALS AND METHODS

Patient population.Plasma and ELF concentration-time data for telavancin were obtained from a phase I, open-label, single-arm, multiple-dose, single-center study of 20 healthy Caucasian volunteers (12). All subjects received 10 mg of telavancin/kg of body weight intravenously as a 60-min infusion once daily for three consecutive days. Plasma samples for telavancin concentrations were collected preinfusion on days 1 and 3 and at 1, 2, 4, 6, 8, 12, and 24 h after completion of the day 3 infusion. Bronchoscopy samples for telavancin concentrations were collected on day 3 at 4, 8, 12, or 24 h following initiation of the study infusion. Each subject had a single bronchoscopy, and five subjects were nonrandomly assigned to be sampled at each bronchoscopy sampling time. Concentrations in plasma and ELF were assayed using a validated liquid chromatography-tandem mass spectrometry technique, with ELF volume determined by using urea as an endogenous marker (12).

Population pharmacokinetic modeling.All data were analyzed in a population pharmacokinetic model using the big nonparametric adaptive grid (BigNPAG) with adaptive γ program of Leary, Jelliffe, Schumitzky, and Van Guilder (19). The pharmacokinetic model was parameterized as a three-compartment model with zero-order infusion into the central compartment. A three-compartment model with zero-order infusion was selected based on Akaike's information criterion and rule of parsimony (32). Elimination from the central compartment and all intercompartmental distribution processes were modeled as first-order processes.

The general differential equations for the model are as follows:
$$mathtex$$\[dX(1)/dt{=}R(t){-}{\{}[(\mathrm{CL}/V){+}K_{12}{+}K_{13}]{\cdot}X(1){\}}{+}[K_{21}{\cdot}X(2)]{+}[K_{31}{\cdot}X(3)]\]$$mathtex$$
$$mathtex$$\[dX(2)/dt{=}[K_{12}{\cdot}X(1)]{-}[K_{21}{\cdot}X(2)]\]$$mathtex$$
$$mathtex$$\[dX(3)/dt{=}[K_{13}{\cdot}X(1)]{-}[K_{31}{\cdot}X(3)]\]$$mathtex$$ where *X*(1) is the amount of drug in the central compartment (in milligrams), *X*(2) is the amount of drug in the peripheral compartment (in milligrams), *X*(3) is the amount of drug in the ELF compartment (in milligrams), CL is clearance from the central compartment (in liters per hour), *K*_{12}, *K*_{21}, *K*_{13}, and *K*_{31} are first-order intercompartmental transfer rate constants (in hour^{−1}), *V* is a scalar and represents the volume of the central compartment (in liters), and *R*(*t*) is the time-delimited zero-order rate of drug input (piecewise input function) into the central compartment (in milligrams per hour). V_{ELF}, a scalar term that represents the apparent volume of ELF, is not included in the equations.

The inverse of the estimated assay variance was used as the first estimate for weighting in the pharmacokinetic modeling. Weighting was accomplished by making the assumption that total observation variance was proportional to assay variance. Assay variance was determined on a between-day basis. When convergence was attained, Bayesian estimates for each patient were obtained using the BigNPAG “population-of-one” utility. The mean, median, and modal values were employed as measures of central tendency for the population parameter estimates and were evaluated in the Bayesian analysis. Scatter plots were examined for individual patients and for the population as a whole. Goodness of fit was assessed by regression with an observed-predicted plot, coefficients of determination, and log-likelihood values. Predictive performance evaluation was based on weighted mean error and the bias-adjusted weighted mean-squared error.

Monte Carlo simulation.The mean parameter vector and covariance matrix from the population pharmacokinetic model were embedded in Subroutine PRIOR of D'Argenio and Schumitzky's ADAPT II software package (4). The population simulation without process noise option was employed. A Monte Carlo simulation with 9,999 subjects was performed and was used to calculate the mean and median ratios of ELF penetration to plasma penetration by estimating the AUC_{ELF} and free or unbound AUC_{plasma} from zero to infinity (AUC_{ELF,0-∞} and free AUC_{plasma,0-∞}, respectively) after a single simulated 750-mg dose and computing the ratio. Specifically, the AUCs in both ELF and plasma were calculated by integrating the concentration-time profile in each compartment from time zero (start of administration) to hour 2,000 post-start of administration. We integrated the profiles from time zero to hour 2,000 in order to approximate the AUC from zero to infinity (AUC_{0-∞}). Given that the half-life of telavancin is ∼10 h, 2,000 h represents more than 200 telavancin half-lives and captures 99.9% of the cumulative exposure. The AUC_{ELF}/free AUC_{plasma} penetration ratio derived from the mean parameter vector from the population model was also calculated.

Both normal and lognormal distributions were evaluated, and these were determined based on their abilities to recreate the mean parameter vector and corresponding standard deviations from the population model. The plasma pharmacokinetic data were adjusted for 90% protein binding (Theravance, Inc., internal report) to reflect unbound or free drug concentrations in the data analysis. The 90% protein binding value used for the analysis is lower than the published protein binding value of 93%, which is based on an early [^{3}H]telavancin study (29). More recent studies have used ^{14}C-radiolabeled telavancin of higher purity and have revealed protein (human plasma and albumin) binding to be approximately 88%; these data will be published in the near future (Theravance, Inc., internal report). To be conservative, we used 90% for the analysis. The ELF pharmacokinetic data were not adjusted for protein binding, because the protein binding of telavancin in ELF is currently unknown. Systat for Windows (version 10.2) was used for all data transformations.

## RESULTS

The population parameter estimates identified by BigNPAG for the pharmacokinetic model are displayed in Table 1. Using the population mean parameter values as the measure of central tendency, the overall fit of the model to the data was good and the observed-predicted plots for plasma and ELF after the Bayesian step were highly acceptable. For plasma, the *r*^{2} was 0.994 and the observed-predicted plot showed a best-fit regression line of observed = (1.008 × predicted) − 0.008 (Fig. 1A). For ELF, the *r*^{2} was 0.999 and the observed-predicted plot showed a best-fit regression line of observed = (1.0001 × predicted) − 0.001 (Fig. 1B).

A 9,999-subject Monte Carlo simulation was performed, and lognormal distributions were selected based on their abilities to recapitulate the original mean parameter values and corresponding standard deviations. The distributions of AUC_{ELF,0-∞} and free AUC_{plasma,0-∞} are shown in Fig. 2 and 3, respectively. Greater variability in AUCs was observed in ELF than in plasma. The mean (standard deviation) AUC_{ELF,0-∞} and free AUC_{plasma,0-∞} were 74.75 (73.23) mg·h/liter and 74.04 (12.52) mg·h/liter, respectively. The median (25th and 75th percentile values) AUC_{ELF,0-∞} and free AUC_{plasma,0-∞} were 53.74 (30.90 and 92.43) mg·h/liter and 73.14 (65.03 and 82.02) mg·h/liter, respectively. The distribution of AUC_{ELF,0-∞}/free AUC_{plasma,0-∞} penetration ratios is shown in Fig. 4. The mean AUC_{ELF}/free AUC_{plasma} penetration ratio ± standard deviation was 1.01 ± 0.96. The median AUC_{ELF}/free AUC_{plasma} penetration ratio was 0.73, and the 25th and 75th percentile value ratios were 0.43 and 1.24, respectively. The average value for the Monte Carlo simulation is skewed because of outliers, as is evident when one examines the AUC_{ELF,0-∞} distribution (Fig. 2), the median penetration ratio of 0.73, the mean ratio of 1.01, and the large standard deviation of 0.96. The AUC_{ELF}/AUC_{plasma} penetration ratio derived from the mean parameter vector from the population model was 0.66 and further reflects the influence of outliers on the mean penetration ratio from the Monte Carlo simulation.

## DISCUSSION

Telavancin is an investigational lipoglycopeptide antibiotic with bactericidal activity against MRSA (14, 18) and is a potential treatment for hospital-acquired pneumonia. Therefore, it is important to examine the ability of telavancin to concentrate in the ELF relative to plasma. Since optimal treatment depends on delivery of the antibiotic to the site of infection, it is imperative to accurately estimate the drug's ability to penetrate the infected site and achieve sufficient concentrations. For extracellular respiratory tract pathogens such as *S. aureus*, determination of the drug concentration in ELF is the best estimate available for ascertaining the degree of drug exposure against these organisms (6-8).

Population pharmacokinetic modeling and Monte Carlo simulation techniques were used to identify the range of ELF concentration-time profiles (exposures) relative to plasma concentration-time profiles that one would observe in a normal healthy Caucasian. Most often, analysis of ELF penetration data is limited to obtaining ratios of drug concentrations in ELF to drug concentrations determined simultaneously in plasma. Because the drug has to penetrate from plasma to ELF, these ratios will change as a function of time, as observed in this study. This phenomenon, known as system hysteresis, makes examination of single time point penetration ratios suboptimal, because the estimates of drug penetration will strongly depend on the sampling time. Population pharmacokinetic modeling is able to overcome this limitation because of its ability to estimate population pharmacokinetics and their associated dispersions for subjects with minimal sampling times. Once the population pharmacokinetics are estimated, Monte Carlo simulation can be performed to estimate the ability of a drug to penetrate the site of infection and to characterize its AUC at that site.

The mean AUC_{ELF}/free AUC_{plasma} ratios indicate that telavancin penetrates reasonably well into ELF compared to plasma. Specifically, the mean AUC_{ELF}/free AUC_{plasma} penetration ratio ± standard deviation was 1.01 ± 0.96, and the median AUC_{ELF}/free AUC_{plasma} penetration ratio was 0.73 (25th and 75th percentile value ratios, 0.43 and 1.24, respectively). When one considers that the plasma protein binding of telavancin is approximately 90%, the total AUC for the drug in ELF approximates the AUC for the free drug in plasma. We employed estimated concentrations of the free drug in plasma for the penetration analysis because there are considerable data that demonstrate that protein binding has an adverse impact on microbiological outcomes and that only free or unbound drug is microbiologically active (1, 21). It should be noted, however, that there is evidence that telavancin operates, at least in part, through a membrane effect that is less impacted by the degree of protein binding (13). Total drug concentration was studied in ELF because the influence of protein binding in ELF has not yet been studied for telavancin or any other antibiotic, to our knowledge. Furthermore, previous studies suggest that most of the drug recovered in ELF will most likely be active as long as binding to tissue components is not appreciable (23, 25). Further study is needed to delineate the influence of protein binding in ELF.

It is difficult to formally compare ELF penetration between telavancin and vancomycin; we are unaware of any randomized, crossover pharmacokinetic study that has employed population modeling and Monte Carlo simulation to compare the AUC_{ELF}/free AUC_{plasma} ratios of telavancin and vancomycin. The best vancomycin ELF penetration is derived from a study by G. L. Drusano et al. that used population modeling to estimate the penetration of vancomycin into the ELF of healthy subjects (9). In this study, the AUC_{ELF}/free AUC_{plasma} ratio for vancomycin was ca. 0.5. For hospitalized patients, the best vancomycin ELF penetration data come from a study that evaluated the entry of vancomycin into the ELF of critically ill patients (17). In this study, concentrations of vancomycin in plasma and ELF were collected simultaneously at various times for mechanically ventilated patients who received at least 5 days of vancomycin treatment. Assuming that vancomycin is 50% protein bound (27), the mean (standard deviation) and median (25th and 75th percentile values) ELF/free plasma concentration ratios were 0.38 (0.19) and 0.39 (0.26 and 0.48), respectively. In a similar study that evaluated the pulmonary disposition of vancomycin in critically ill patients 24 h after the onset of treatment (just before the next planned infusion of vancomycin), the average ratio of the penetration of vancomycin into ELF to the penetration of free vancomycin into plasma was only 0.14, and 6 of the 10 patients had no detectable vancomycin in their ELF (11). Further study of the AUC_{ELF}/free AUC_{plasma} ratios for both telavancin and vancomycin by population modeling and Monte Carlo simulation techniques is still needed, especially among hospitalized patients, but it may be inferred from these data that telavancin has a higher ELF/free plasma ratio than vancomycin. Of course, given that no randomized, crossover pharmacokinetic trial has been performed to date, overinterpretation of these data should be avoided.

In summary, telavancin is an investigational lipoglycopeptide antibiotic that is being evaluated in phase III clinical trials for hospital-acquired pneumonia. Since antibiotic delivery to the site of infection is imperative for optimal therapy, we used population pharmacokinetic modeling and Monte Carlo simulation to characterize the penetration of telavancin into ELF relative to plasma. Overall, the results indicate that telavancin penetrates reasonably well into ELF compared to plasma, as defined by the mean AUC_{ELF}/free AUC_{plasma} ratio in normal healthy volunteers.

## ACKNOWLEDGMENTS

This study was supported by a grant from Theravance, Inc. T.P.L. was the principal investigator for this grant. Theravance only provided support to complete the project and was not involved in the following: design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation of the manuscript. Theravance reviewed the final manuscript. No other conflicts of interest exist for any of the authors.

## FOOTNOTES

- Received 22 August 2007.
- Returned for modification 8 November 2007.
- Accepted 13 April 2008.
↵▿ Published ahead of print on 21 April 2008.

- American Society for Microbiology