Simultaneous Semimechanistic Population Analyses of Levofloxacin in Plasma, Lung, and Prostate To Describe the Influence of Efflux Transporters on Drug Distribution following Intravenous and Intratracheal Administration

Abstract

Levofloxacin (LEV) is a broad-spectrum fluoroquinolone used to treat pneumonia, urinary tract infections, chronic bacterial bronchitis, and prostatitis. Efflux transporters, primarily P-glycoprotein (P-gp), are involved in LEV's tissue penetration. In the present work, LEV free lung and prostate interstitial space fluid (ISF) concentrations were evaluated by microdialysis in Wistar rats after intravenous (i.v.) and intratracheal (i.t.) administration (7 mg/kg of body weight) with and without coadministration of the P-gp inhibitor tariquidar (TAR; 15 mg/kg administered i.v.). Plasma and tissue concentration/time profiles were evaluated by noncompartmental analysis (NCA) and population pharmacokinetics (popPK) analysis. The NCA showed significant differences in bioavailability (F) for the control group (0.4) and the TAR group (0.86) after i.t. administration. A four-compartment model simultaneously characterized total plasma and free lung (compartment 2) and prostate (compartment 3) ISF concentrations. Statistically significant differences in lung and prostate average ISF concentrations and levels of kidney active secretion in the TAR group from those measured for the control group (LEV alone) were observed. The estimated population means were as follows: volume of the central compartment (V₁), 0.321 liters; total plasma clearance (CL), 0.220 liters/h; TAR plasma clearance (CL_TAR), 0.180 liters/h. The intercompartmental distribution rate constants (K values) were as follows: K₁₂, 8.826 h⁻¹; K₂₁, 7.271 h⁻¹; K₁₃, 0.047 h⁻¹; K₃₁, 7.738 h⁻¹; K₁₄, 0.908 h⁻¹; K₄₁, 0.409 h⁻¹; K₂₁ lung TAR (K_21LTAR), 8.883 h⁻¹; K₃₁ prostate TAR (K_31PTAR), 4.377 h⁻¹. The presence of P-gp considerably impacted the active renal secretion of LEV but had only a minor impact on the efflux from the lung following intratracheal dosing. Our results strongly support the idea of a role of efflux transporters other than P-gp contributing to LEV's tissue penetration into the prostrate.

INTRODUCTION

The constant increase in the emergence of antimicrobial resistance presents an important challenge for health science, government sectors, and society (1). There are two options for addressing this pending health threat: (i) develop new drugs and (ii) optimize use of available antimicrobials. The process of development of new compounds is very slow and expensive. As a consequence, preserving the efficacy of currently available antimicrobials is essential (2, 3). In order to preserve the efficacy of antimicrobials, it is important to consider the pharmacokinetic (PK) mechanism and factors which can affect drug distribution in the body, besides drug effect.

The extracellular tissue space is the site of infection of the majority of bacterial pathogens. Thus, the activity of antimicrobials depends on their ability to cross biological barriers and maintain an effective unbound drug concentration at the interstitial space fluid (ISF) (4, 5). The influx and efflux transporters involved in drug tissue distribution directly affect the drug concentrations at the ISF. Among the efflux transporters, P-glycoprotein (P-gp) is an ABC family member primarily expressed in the apical (AP) membrane of biological barriers (6). P-gp activity can reduce drug penetration into tissues and may result in subtherapeutic unbound target site concentrations that lead to the emergence of bacterial resistance (7, 8).

Levofloxacin (LEV) is a potent broad-spectrum fluoroquinolone approved by the FDA to treat infections of the respiratory tract (pneumonia, sinusitis, and bronchitis) and urinary tract (prostatitis and pyelonephritis) (9, 10). LEV presents a low level of protein binding (24% to 38%) in humans and readily distributes throughout the body, with a mean volume of distribution (V) of 1.1 liters/kg of body weight (11). Biopsy studies have indicated that LEV has a high level of prostate penetration (12, 13) and lung penetration (14). However, measurements from tissue homogenates can overestimate drug concentrations at the target site, especially for fluoroquinolones that accumulate in the intracellular space (15). Although LEV demonstrated good tissue penetration in studies where free ISF concentrations were measured in human lung (16) and rodent prostate (17), the penetration factor (tissue bioavailability [fT_tissue]) determined by the relationship between the area under the concentration-time curve corresponding to free tissue exposure (AUC_free,tissue) and that corresponding to free plasma exposure (AUC_free,plasma) was lower than 100%. Furthermore, P-gp and other membrane transporters are involved in fluoroquinolone tissue distribution (18), and reports in the literature indicate that P-gp is expressed in lung and prostate (19, 20).

The common strategy to confirm whether a drug is a P-gp substrate is the use of P-gp inhibitors, such as tariquidar (TAR), along with the substrate drug. TAR represents an example of the third-generation P-gp inhibitors characterized by higher specificity and a longer-lasting modulation effect than classical inhibitors such as verapamil and PSC-833 (21). Brillault and coworkers (22) investigated the impact of P-gp on fluoroquinolone permeability in vitro using CALU-3 cells (lung epithelial cells). The results suggest that P-gp plays a role in LEV transport through the lung barrier. So far, the impact of efflux transporters on LEV free ISF concentrations has not been evaluated in vivo.

Microdialysis, a semi-invasive sampling technique that selectively collects the free tissue ISF concentrations of drugs, provides a more realistic insight into unbound concentrations in the extracellular tissue space. Additionally, microdialysis permits simultaneous samplings in multiple organs and the establishment of concentration-time tissue profiles for each animal (5, 23, 24). Our research group has been successfully applying microdialysis to analyze the tissue distribution of antifungal compounds (25, 26) and antimicrobial compounds (27, 28). Recently, we determined LEV prostatic penetration by microdialysis (17). The LEV prostate penetration factor (fT_prostate) value was 0.78, indicating that efflux transporters may be involved in this antimicrobial distribution process, reducing tissue levels.

The objective of the present study was to investigate the involvement of P-gp in LEV tissue penetration in lung and prostate of healthy male Wistar rats by performing simultaneous population pharmacokinetics analyses of the drug free concentrations in plasma and ISF following drug intravenous (i.v.) and intratracheal (i.t.) administration.

MATERIALS AND METHODS

(i) Drugs, chemicals, and reagents.

Levofloxacin (purity ≥ 98.0%), ciprofloxacin (internal standard [IS]; purity ≥ 98.0%), and urethane (ethyl carbamate; purity ≥ 99.0%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Tariquidar (purity ≥ 98%) was purchased from ChemScene (New Jersey, USA). Dimethyl sulfoxide (purity ≥ 99.5%) was purchased from Synth (São Paulo, Brazil). High-performance liquid chromatography (HPLC)-grade methanol and acetonitrile were purchased from Tedia (Fairfield, OH, USA). Formic acid was purchased from Fluka Chemie GmbH (Buchs, Switzerland), and triethylamine and phosphoric acid were purchased from Merck (Darmstadt, Germany). Heparin (5,000 IU/ml) was purchased from Cristália Produtos Químicos Farmacêuticos (São Paulo, Brazil). Ultrapure water was obtained with a Millipore Milli-Q system (Bedford, MA, USA). Other chemicals used in the experiment were of analytical-reagent grade and were purchased from commercial sources. The Ringer's solution consisted of 147 mM NaCl, 1.3 mM CaCl₂, and 4 mM KCl.

(ii) Measurement of levofloxacin in rat plasma and microdialysate samples.

LEV concentrations in plasma and microdialysate samples were analyzed by HPLC with fluorescence detection. The HPLC method was previously validated in our research group for matrices of both plasma and microdialysate samples. The chromatographic separation was achieved using a Shimadzu system equipped with an LC-10AD VP pump, an SIL-10AD VP auto injector, an SCL-10A VP system controller, and a DGU-14A degasser. Chromatographic separation was achieved on a reversed-phase C₁₈ column (Waters Atlantis T3; Milford, MA, USA) (150 by 4.6 mm; particle diameter, 5 μm) coupled to a C₁₈ guard cartridge (Phenomenex, Torrance, CA, USA) (4.0 by 3.0 mm; particle diameter, 4 μm). For both matrices (microdialysate and plasma), detection was achieved with a Shimadzu RF 10AXL fluorescence detector set at excitation and emission wavelengths of 292 nm and 494 nm, respectively. Data acquisition and integration were performed using Shimadzu CLASS-VP v. 6.12 software.

Microdialysate sample analyses were performed at a flow rate of 1.0 ml/min using an injection volume of 20 μl. The temperature of the column oven was maintained at 35 ± 1°C during each analysis. The mobile phase consisted of a mixture of 0.4% aqueous triethylamine (pH adjusted to 3.0 ± 0.1 with 85% phosphoric acid):methanol:acetonitrile (75:22.5:2.5 [vol/vol/vol]). The peak area of LEV was used for the quantification of samples.

For plasma sample cleanup, proteins were precipitated by addition of 200 μl of ice-cold methanol containing 0.5% formic acid, 100 μl of plasma sample, and 10 μl of the IS (ciprofloxacin [10 μg/ml]), followed by 5 min of agitation in a shaker and 10 min of centrifugation at 4 ± 1°C and 12,000 rpm. The supernatant was collected and injected into the HPLC system. The peak area ratio of LEV/IS was used for the quantification of samples.

The HPLC-fluorescence method was established to have linearity in the range of 0.01 to 5.0 μg/ml in plasma and 5 to 1,000 μg/ml in microdialysate with determination coefficients (r²) higher than 0.98. The intra-assay accuracy was >85%, and the intra-assay imprecision was <15%, in accordance with FDA guidelines (29) for bioanalytical method validation for both matrices.

(iii) Microdialysis system.

The microdialysis system consisted of a PHD 2000 syringe pump (Harvard Apparatus, Holliston, MA, USA). Gastight syringes (Hamilton Company, Reno, NV, USA) (500 μl) were used to provide the perfusion solution. CMA 20 concentric microdialysis probes (CMA Microdialysis, Kista, Sweden) (4-mm membrane length, 20-kDa cutoff) were used.

(iv) In vitro probe calibration.

In vitro microdialysis calibration was performed in order to determine the relative recovery (RR) in Ringer's solution using two different methods: determination of extraction efficiency (EE [recovery]) and retrodialysis (RR_RD [delivery]). For both methods, the flow rate and concentration selected were 1.5 μl/min and 0.75 μg/ml, respectively. The microdialysis conditions were optimized based on our previous study evaluating the RR of LEV with different flow rates and concentrations (17).

The relative recoveries determined by both methods in vitro were compared by Student's t test (α = 0.05).

(v) In vivo probe calibration.

The animal experiments were approved by the Federal University of Rio Grande do Sul (UFRGS) Ethics in Animal Use Committee (Protocol 211609). Male Wistar rats (weight, 300 to 400 g) were obtained from the university animal facilities (Centro de Reprodução e Experimentação de Animais de Laboratório [CREAL]/UFRGS) and kept under controlled conditions with a 12-h-light/12-h-dark cycle and water and food ad libitum until the commencement of the experiments.

On the day of experiment, the animals were anesthetized with urethane (1.25 g/kg administered intraperitoneally [i.p.]) and immobilized in a supine position. For probe calibration in prostate, the tissue was exposed with minimal dissection after an abdominal incision. The prostate was identified under the bladder, and the microdialysis probe was inserted with the help of a guide cannula as described previously (30). For probe calibration in lung, the animals were intubated by tracheotomy and maintained with mechanical ventilation using a rodent respirator (Harvard Apparatus model 683), a breathing rate of 64 to 68 min⁻¹, and an air volume of 2.5 ml. The lung was exposed, and the microdialysis probe was inserted carefully into the intermediate lobe (28). The in vivo probe calibration was performed simultaneously for both tissue types in the same animal (n = 3). The in vivo recovery for lung and prostate was determined by retrodialysis. After the surgical procedure, probes flushed with Ringer's solution at a flow rate of 1.5 μl/min were allowed to equilibrate for 1 h. After the equilibration period, Ringer's solution was replaced by LEV (0.75 μg/ml)–Ringer's solution and microdialysate samples (45 μl) were collected every 30 min up to 2 h.

(vi) Experiment PK protocol.

To analyze free LEV concentrations in tissue (lung and prostate) and total plasma concentrations, the animals were randomly assigned to two groups, one for each route of administration: One group was used for blood sampling, and the other group was used to determine the drug penetration into both lung and prostate. Microdialysis probes were inserted into tissues and perfused with Ringer's solution at a flow rate of 1.5 μl/min for 1 h to achieve equilibration prior to the start of the experiment. Microdialysate samples (45 μl) were collected each 30 min for up to 12 h post-drug administration.

For the plasma sampling group, the animals were anesthetized with urethane (1.25 g/kg i.p.). The anesthesia was confirmed by the absence of the reflex of paw pinching. The carotid artery was cannulated for blood sampling, and 200 μl was harvested at each time point. A similar volume of sterile heparinized saline solution (with heparin at 25 IU/ml) was administered through the catheter immediately after the blood harvest, to compensate for the withdrawn blood. The catheter was inserted into the artery and maintained under conditions of irrigation with heparinized saline solution (25 IU/ml). At predetermined times before dosing (time zero) and after dosing (0.08, 0.25, 0.5, 1, 2, 4, 6, 8, and 12 h), blood was collected into heparinized tubes and immediately centrifuged at 12.000 rpm and 4 ± 1°C for 10 min. Plasma samples were stored at −80 ± 2°C for up to 30 days until analysis was performed on the basis of a previous stability study (17).

(a) Levofloxacin i.v. administration.

Animals were randomly separated into four groups (n = 6/group): two control groups (a plasma control group and a tissue [lung and prostate] control group), in which animals received 7 mg/kg of LEV as an intravenous (i.v.) bolus via the right femoral vein, and two tariquidar (TAR) groups (plasma TAR group and tissue [lung and prostate] TAR group), in which animals received 15 mg/kg of tariquidar as an i.v. bolus 30 min prior to LEV administration.

A solution of LEV (5 mg/ml) was freshly prepared in 0.9% saline solution for drug dosing via the right femoral vein. The TAR (4 mg/ml) solution was prepared by adding it to a 5% dimethyl sulfoxide–5% glucose solution. TAR was administered via i.v. bolus in the left femoral vein.

(b) Levofloxacin i.t. nebulization.

Animals were randomly distributed in four groups (n = 6 to 7/group). The first group was used for blood sampling and the second, which received intratracheal nebulization of LEV (7 mg/kg), for microdialysate lung sampling; the third and fourth groups, used for blood and lung microdialysis collection, respectively, received 15 mg/kg of TAR i.v. bolus via the left femoral vein 30 min prior to LEV intratracheal (i.t.) administration.

A MicroSprayer (model IA-1B; PennCentury, Inc., Philadelphia, PA, USA), composed of an angled stainless steel tube mounted onto a syringe with an atomizer at the tip, was used for nebulization. The solution of LEV (12 mg/ml) for nebulization was prepared in 0.9% NaCl, corresponding approximately to a volume of 0.2 ml for each animal. Each animal was anesthetized with urethane (1.25 mg/kg i.p.) and immobilized in a supine position. The trachea was exposed, and a small incision was made between the tracheal rings. The animal was inclined at an angle of 45°, and the tube of the MicroSprayer was placed into the trachea to perform the nebulization. After drug nebulization, the animals were immediately connected to a rodent respirator and kept for 30 min in the same inclination. After completion of this period, the animals were returned to the supine position and were maintained under the previously described conditions of mechanical ventilation up to the end of the experiment.

(vii) PK data analysis. (a) Noncompartmental analysis.

Noncompartmental analysis (NCA) of the individual profiles was performed using Excel 2007 software (Microsoft Corporation, Redmond, USA). The pharmacokinetic parameters calculated included the elimination rate constant (λ), the terminal half-life (t_1/2), the extrapolated concentration at t = 0 (C₀), the area under the concentration-time curve (AUC) calculated by the linear trapezoidal rule, the total clearance (CL), the mean residence time (MRT), and the volume of distribution at steady state (V_ss) (31). The ƒT_tissue was calculated as the ratio of the AUC_tissue,free and AUC_plasma,total values corrected for the fraction unbound to plasma proteins (ƒu_plasma). An unbound-fraction value of 55% was used (17). Absolute bioavailability (F) for intratracheal administration was calculated using equation 1 (32):

$$F=\frac{{\text{AUC}}_{\text{Lung}}}{{\text{Dose}}_{\text{Lung}}}\cdot \frac{{\text{Dose}}_{\mathrm{i}\mathrm{.}\mathrm{v}\mathrm{.}}}{{\text{AUC}}_{\mathrm{i}\mathrm{.}\mathrm{v}\mathrm{.}}}$$ (1)

The PK parameters for the tissues data were compared to those for plasma by one-way analysis of variance (ANOVA; α = 0.05) or Student's t test, as appropriate, employing SigmaStat software version 3.5 (Systat, Richmond, CA, USA).

(b) Population PK model.

A semimechanistic population PK model was developed using nonlinear mixed effect modeling software (NONMEM version 7.2; Icon Development Solutions, Dublin, Ireland) to simultaneously describe the total drug concentration in plasma and the free drug concentration in lung and prostate in the presence or absence of TAR. Different structural models were tested for model improvement in terms of fit and significant changes in the objective function (OFV). The addition of a structural or variance parameter was considered statically significant when the OFV dropped by at least 3.84 (P < 0.05 for degree of freedom). The model consisted of a central compartment (total plasma concentrations), a compartment for free lung concentrations, a compartment for free prostate concentrations, and a peripheral compartment representing tissues other than the prostate and lung (Fig. 1). All models were parameterized as a system of differential equations (ADVAN 13 subroutine). Individual PK parameters were assumed to follow a lognormal distribution with mean (θ, typical value) and variance (ω², interindividual variability). Different residual-error models (additive, proportional, and combined) were tested for total plasma concentrations and free lung and prostate concentrations. The first-order conditional estimation with interaction (FOCEI) method was used for model fitting. Model comparison and selection of a final model were based upon numerical comparisons of objective function values (log-likelihood test; α = 0.05), visual evaluation of basic goodness-of-fit plots, and visual predictive checks (VPCs). The robustness of the final model and of its estimated parameter was evaluated using a nonparametric bootstrap with 1,000 replicates. Graphics were created using ggplot2 libraries for R (Version 3.0.1).

FIG 1.

Schematic representation of the PK model used to describe LEV plasma and tissue concentration-time profiles after administration (ADM) of i.v. bolus and intratracheal 7 mg/kg dose to Wistar rats in the presence and absence of the P-gp inhibitor TAR. A, amounts; V, volume (with the numbers in parentheses indicating the respective compartments). The model represents the results seen after i.v. dosing (outer dotted border) and after i.t. dosing (inner dashed border).

RESULTS

(i) In vitro microdialysis probe calibration.

The relative recovery values obtained by the extraction efficiency (EE) and retrodialysis (RR_RD) methods were 21.2 ± 2.5% and 27.3 ± 8.1%, respectively. No statistically significant difference was observed between the two methods (P < 0.05). The results indicate that LEV did not bind to the probe plastic tubing and that the diffusion process at the microdialysis membrane occurred equally in both directions.

(ii) In vivo microdialysis probe calibration.

The in vivo recovery values determined by RR_RD were 9.0 ± 1.7% and 15.5 ± 2.5% for lung and prostate, respectively. For both tissues, the values achieved are in agreement with literature reports (17, 33). The differences between the lung and prostate RR values can be explained on the basis of the physiological characteristics of each tissue. The cell organization in each organ can change the interstitial volume and drug diffusibility in the interstitial space, thus affecting the process of diffusion of drug through the tissue (34, 35).

(iii) NCA.

The PK parameters determined by noncompartmental analysis (NCA) are summarized in Tables 1 and 2 for plasma and tissues after LEV i.v. dosing, respectively.

TABLE 1.

Plasma PK parameters determined by NCA after LEV 7 mg/kg i.v. bolus administration in the presence (TAR group) and absence (control group) of tariquidar^a

PK parameter	Group valuesb
	Control	TAR
t1/2 (h)	4.9 ± 0.6	4.9 ± 0.4
AUC0–∞ (μg · h/ml)	13.0 ± 2.0	15.6 ± 2.6
MRT (h)	5.9 ± 0.8	5.9 ± 0.4
CL (liters/h/kg)	0.55 ± 0.08	0.46 ± 0.08
Vss (liters/kg)	3.23 ± 0.43	2.73 ± 0.56

^aParameter values are shown as averages ± standard deviations (SD).

TABLE 2.

Prostate and lung PK parameters determined by NCA after LEV administration of 7 mg/kg i.v. bolus dose in the presence (TAR group) and absence (control group) of tariquidar^a

PK parameter	Group value(s)
	Control		TAR
	Lung	Prostate	Lung	Prostate
λ (h−1)	0.25 ± 0.06	0.26 ± 0.09	0.16 ± 0.06	0.18 ± 0.05
t1/2 (h)	3.0 ± 0.8	3.0 ± 1.2	4.8 ± 1.6	4.1 ± 1.3
AUC0–∞ (μg · h/ml)	4.9 ± 2.1	4.8 ± 1.0	6.9 ± 2.3	14.1 ± 4.6*
AUMC0–∞ (μg · h2/ml)	16.4 ± 11.2	17.3 ± 8.2	41.5 ± 19.2	83.2 ± 58.2*
MRT (h)	3.2 ± 0.8	3.5 ± 1.3	5.9 ± 1.9	5.4 ± 1.8
ƒTtissue	0.69	0.68	0.81	1.64

^aParameter values are shown as averages ± SD. ƒT, AUC_{0–∞ tissue,free}/AUC_{0–∞ plasma,free}. Plasma ƒu = 0.55. *, significant difference detected by Student's t test (α = 0.05) compared to equivalent control group results.

The values for the plasma PK parameters determined for LEV in the control group were similar to previously reported values (17). The comparative analysis of the control and TAR groups in plasma demonstrated the similarity of the parameters estimated for the two groups. No statistically significant difference was observed in LEV plasma PK parameters when the P-gp inhibitor TAR was administered 30 min prior to LEV administration (α = 0.05). Although P-gp has been reported in the kidneys (8) and fluoroquinolones, including LEV, are mainly excreted renally (11), the inhibition of this efflux transporter did not cause a significant alteration of LEV clearance and half-life.

Table 2 presents the PK parameters of NCA for lung and prostate in the control and TAR groups. The comparative analysis performed for the control group showed that there were no statistically significant differences among all PK parameters calculated for lung and prostate. This similarity in terms of parameters is reflected by the ƒT values for the two tissues, which were similar.

The association of TAR and LEV did not alter the fluoroquinolone lung distribution. No statistically significant differences were detected in the two lung groups, indicating that the P-gp inhibitor did not affect LEV penetration to the lung, as indicated in Table 2. A trend for increased half-life, mean residence time, area under the concentration-time curve from time 0 to infinity (AUC_0–∞), and area under the first moment of the concentration-time curve from time 0 to infinity (AUMC_0–∞) was observed, but this was not statistically significant. The lung penetration factor increased only 17% in the TAR group compared to the control group.

On the other hand, statistically significant differences between prostate control and TAR groups were identified in the AUC_0–∞ and AUMC_0–∞ values. Also, the presence of TAR caused a more than 2-fold increase in LEV prostate penetration (ƒT = 1.64) compared to the control group (ƒT = 0.68).

Table 3 presents the PK parameters of NCA for plasma and lung in the control and TAR groups after intratracheal administration. The results of comparative analysis of plasma in the presence and absence of TAR show a statistically significant difference between the AUC_0–∞ and AUMC_0–∞ values for the two groups. Basically, there was an increase in LEV exposure in the TAR group, resulting in increased bioavailability with the P-gp inhibitor administration. The presence of TAR in plasma increased the intratracheal bioavailability from 40% (control group) to 86% (TAR group). Lung exposure AUC_0–∞ and AUMC_0–∞ in the TAR group were also significantly increased. The presence of TAR (AUC_0–∞ = 5.8 ± 1.5 μg · h/ml) led to an increase of 87% in lung exposure to LEV compared to control group results (AUC_0–∞ = 3.1 ± 1.8 μg · h/ml).

TABLE 3.

Plasma and lung PK parameters calculated by NCA after LEV intratracheal nebulization of 7 mg/kg dose in the presence (TAR group) and absence (control group) of tariquidar^a

PK parameter	Group value(s)
	Control		TAR
	Plasma	Lung	Plasma	Lung
λ (h−1)	0.16 ± 0.01	0.20 ± 0.05	0.18 ± 0.07	0.19 ± 0.06
t1/2 (h)	4.4 ± 0.4	3.6 ± 0.9	4.25 ± 1.6	4.1 ± 1.4
AUC0–∞ (μg · h/ml)	5.3 ± 1.3	3.1 ± 1.8	13.39 ± 4.1*	5.8 ± 1.5*
AUMC0–∞ (μg · h2/ml)	17.0 ± 7.8	14.0 ± 5.8	66.00 ± 45.3*	25.5 ± 6.1*
MRT (h)	3.1 ± 0.8	4.8 ± 1.3	4.5 ± 1.8	4.6 ± 1.3
CL (liters/h/kg)	0.57 ± 0.15	NC	0.56 ± 0.17	NC
F	0.4		0.86
ƒTtissue		1.1		0.8

^aParameter values are shown as averages ± SD. NC, parameter not calculated; F (bioavailability), AUC_{0–∞ intratracheal}/AUC_{0–∞ i.v}; ƒT_tissue, AUC_{0–∞ tissue,free}/AUC_{0–∞ plasma,free}. Plasma ƒu = 0.55. *, significant differences detected by Student's t test (α = 0.05) compared to equivalent control group results.

(iv) Population PK analysis.

The population PK analysis was performed in a stepwise fashion starting with a separate exploratory analysis of plasma and tissue concentrations. The same strategy was applied for the different routes of administration (intravenous and intratracheal), which permitted a deep evaluation of the data and construction of a more rational and physiological approach to the final combined model. A four-compartment model was selected to simultaneously fit the model to experimental data for total plasma, free lung, and prostate concentrations (Fig. 1), respectively. On the basis of NCA, the values of bioavailability (F) for the control group (0.4) and the TAR group (0.86) were used to account for reduced drug exposure following intratracheal drug administration. A combined-residual-error model was used to describe the data for plasma, lung, and prostate. Statistically significant differences were detected with three extra parameters (lung TAR intercompartmental distribution rate constant [K_21LTAR], K₃₁ prostate TAR [K_31PTAR], and CL_TAR) for the TAR groups, and those parameters were incorporated into the final model. The systems of differential equations for the final model are provided in equations 2a to d:

$$\frac{dA\left(1\right)}{dt}=A\left(2\right)\cdot K_{21}+A\left(3\right)\cdot K_{31}+A\left(4\right)\cdot K_{41}-A\left(1\right)\cdot K_{10}-[f\mathrm{u}\cdot A\left(1\right)\cdot K_{12}]-[f\mathrm{u}\cdot A\left(1\right)\cdot K_{13}]-[f\mathrm{u}\cdot A\left(1\right)\cdot K_{14}]$$ (2a)

$$\frac{dA\left(2\right)}{dt}=[f\mathrm{u}\cdot A\left(1\right)\cdot K_{12}]-\left[A\left(2\right)\cdot K_{21}\right]$$ (2b)

$$\frac{dA\left(3\right)}{dt}=[f\mathrm{u}\cdot A\left(1\right)\cdot K_{13}]-\left[A\left(3\right)\cdot K_{31}\right]$$ (2c)

$$\frac{dA\left(4\right)}{dt}=[f\mathrm{u}\cdot A\left(1\right)\cdot K_{14}]-\left[A\left(4\right)\cdot K_{41}\right]$$ (2d)

where A₁, A₂, A₃, and A₄ are the drug amounts in the central, lung, prostate, and peripheral compartments, respectively, K₁₂, K₂₁, K₁₃, K₃₁, K₁₄, and K₄₁ are the first-order distribution rate constants between compartments, and K₁₀ is the first-order elimination rate constant from the central compartment. The relationship CL = K₁₀ · V₁ was considered, where V₁ represents the volume of the central compartment, and ƒu is the free fraction in plasma (ƒu = 0.55) (17).

Individual and population predictions were able to describe total plasma and free lung and free prostate concentrations for the control and TAR groups reasonably well (Fig. 2). Visual inspection of the basic goodness-of-fit plots did not reveal model misspecification (see Fig. S1 and S2 in the supplemental material). The visual predictive checks (VPCs) of the final model (Fig. 3) confirmed that the model is adequate to describe simultaneously the concentrations in all tissues for each group. The median values represented by solid black (observation) and red (simulation) lines are in good agreement, and most of the observations (open blue dots) are within the 95% prediction interval bands. The stability of the final model was confirmed by the results of bootstrapping analysis. Model parameter estimates obtained from a single estimation run were in good overall agreement with the respective bootstrap values for 1,000 replicates (see Table S1).

FIG 2.

Representative examples of individual concentration-versus-time profiles after administration of LEV 7 mg/kg i.v. bolus (i.v.) dose and intratracheal nebulization (i.t.) dose to Wistar rats (control group) and with pretreatment consisting of 15 mg/kg TAR i.v. bolus administration (TAR group) in different compartments (CMT). Red dots represent the observed total plasma concentrations (1 to 6 and 26 to 32, control group; 7 to 12 and 46 to 51, TAR group [for i.v. and i.t. administration, respectively]). Green and blue dots represent the unbound concentrations measured by microdialysis in lung and prostate, respectively (13 to 19 and 33 to 39, control group; 20 to 25 and 40 to 45, TAR group [for i.v. and i.t. administration, respectively]). The solid and dashed lines represent individual and population concentrations predicted by the popPK model, respectively.

FIG 3.

Visual predictive checks of the final population PK model to describe LEV following administration of 7 mg/kg i.v. bolus and intratracheal dose stratified based on the tissues (plasma, lung, and prostate) and groups (control and TAR). The blue open circles (or dots) indicate the observations. The solid black (observation) and red (simulation) lines connect the median values per bin. The dashed black and red lines represent the 5th (lower) and 95th (upper) percentiles of the observation and simulation, respectively. Blue areas (or bands) indicate the 95% predicted intervals (PI) of the 5th and 95th percentiles of the predicted (simulated) value, whereas the red area (band) indicates the PI of the median.

DISCUSSION

In has been reported in the literature that the tissue distribution of fluoroquinolones is influenced by efflux transporters (8, 22, 36). In the present work, we evaluated the distribution kinetics of LEV in plasma, lung, and prostate in the presence and absence of TAR in order to evaluate the impact of the efflux transporter P-gp (TAR group) on free interstitial drug concentrations. Microdialysis was successfully applied to obtain LEV free, interstitial tissue concentrations, which provided a more realistic insight into LEV's target site exposure.

Prior to choosing the final model, a four-compartment model with saturable distribution processes was tested. Saturable processes were tested by Michaelis-Menten kinetics with V_max (maximum transporter velocity) and K_m (Michaelis constant) incorporated in the transportation from tissues to plasma. A similar population pharmacokinetics (popPK) model was used to describe simultaneously LEV concentration-time data in plasma and prostatic tissue (ISF) after administration of a single i.v. bolus (17). Unfortunately, the estimations of V_max and K_m presented high uncertainty and imprecision. The establishment of a saturable concentration range usually requires the administration of two different doses unless the concentrations of a single dose are above the K_m value estimated. None of these conditions was satisfied in the present study. Alternatively, the model using different rate constants for the control and TAR groups was able to describe the experimental data simultaneously. The three extra parameters introduced into the model (K_21LTAR, K_31PTAR, and CL_TAR) showed statistically significant improvements in terms of fit, model stability, and OFV for the TAR group. The presence of the K_21LTAR parameter (Fig. 1) dropped the OFV by more than 100 points (P < 0.05), which represents a significant improvement in terms of goodness of fit for the lung TAR group. It should be noted that the variability of data in lung is higher than in plasma and prostate due to experimental issues. The lung in rat is a relatively big organ, and accessing it in a minimally invasive way is not easy. Also, insertion of the microdialysis probe in the exact same position in each experiment is difficult, representing an uncontrolled bias in the experimental data.

The use of the K_31PTAR parameter (Fig. 1) was intended to model the changes in LEV distribution into the prostate observed by NCA in the presence of TAR. The tissue penetration factor (fT) values determined for prostate for the control and TAR groups were 0.68 and 1.68, respectively (Table 2). The K_31PTAR value (4.377 h⁻¹) estimated by the PK model was around 40% lower than the K₃₁ value (7.793 h⁻¹) determined for the control group, which indicates that the findings from NCA are in good agreement with the popPK modeling results. When the rate constant of the distribution from the prostate to the central compartment (plasma) was lower in the TAR group, more drug stayed in the prostate ISF in this group, increasing the prostate AUC_0–∞ and, consequently, the fT. The difference found between CL (0.22 liters/h) and CL_TAR (0.18 liters/h) can be explained by the LEV elimination mechanism, which involves glomerular filtration and active secretion (≥80% unchanged in urine). Assuming that the P-gp expressed in the kidneys participates in active secretion of LEV, the administration of TAR results in inhibition of the active secretion process and decreased drug elimination. Although we did not evaluate the TAR concentration in rat plasma, TAR shows a persistent (>22 h) inhibitory effect after the removal of the plasma from in vitro incubation and a long duration of action also in vivo with doses in the range of 6 to 12 mg/kg (21). Therefore, we can assume that P-gp was successfully inhibited during the sample collection of our pharmacokinetic study (up to 12 h) at the dose used (15 mg/kg).

For the lungs, NCA did not reveal statistically significant differences between the parameters calculated for the control and TAR groups. After intravenous administration, the drug penetrates into the lung tissue via the basolateral (BL) side of cells. Findings from in vitro studies suggest that the active contribution of P-gp to LEV transportation to the lung (BL to AP direction) is 30% (22). However, the value corresponding to the assumption of variability concerning the in vivo experiment is around 30%. As a consequence, the corresponding differences in tissue exposure may not be significant, as shown in our study. Furthermore, the in vitro study by Brillault and coworkers (22) was performed using Calu-3 cells, a human cell line, whereas the present study was performed in Wistar rats.

The intratracheal model was initially designed as an extension of the intravenous model with an absorption compartment (depot). Nevertheless, the results of popPK modeling estimated a very high (>50 h⁻¹) absorption rate constant (K_A), indicating that the i.t. dose was very rapidly absorbed. Therefore, the final combined model (without an absorption compartment) was selected because it represents a significant improvement in terms of goodness of fit, OFV, and reduction in residual unknown variability (RUV). However, the intratracheal dosing groups were designed to demonstrate that the influence of P-gp on LEV transportation in the lung is more relevant at the AP side of the cells (air/lung interface) than at the BL side (blood/lung interface). The difference of bioavailability (ƒ) between the control group (0.4) and the TAR group (0.86) detected by NCA confirmed the relevance of P-gp expression at the AP side in lung. The P-gp inhibition caused by the presence of TAR represents an increase of 115% in ƒ for i.t. administration. Our results demonstrated that the transportation mediated by efflux transporters is more affected at AP-BL side than at the BL-AP side, which is in accordance with previous in vitro results reported for fluoroquinolones (22, 37).

One has to consider that microdialysis allows sampling the free interstitial concentrations of drugs. However, according to the literature (38), depending on the type of lung infection (pneumonia or endobronchial or other infections), the microorganism can be located in different compartments such as ISF, endothelial lining fluid, or alveolar or phagocytic cells, and different concentrations may be relevant for the effect. In the case of pneumonias, the free ISF concentrations determined in this study are relevant for bacterial killing. Our results indicated that, when the drug is given intravenously, P-gp transporters do not influence free concentrations and should not impact the pharmacologically active ISF concentrations.

The findings from the present study support the idea that LEV penetration into the prostate is impacted by efflux transporters. Although TAR is an expanded-spectrum member of the P-gp inhibitors with improved selectivity with respect to its P-gp modulatory effect, recent in vitro studies demonstrated that its interactions with breast cancer resistance protein (BCRP) and P-gp are dose dependent. At higher concentrations (≥100 nM), TAR can inhibit both P-gp and BCRP (39). Also, the results of this study suggest that in vivo doses of TAR higher than 2 mg/kg i.v. may affect both carriers. The effective TAR dose described in the literature that is necessary to achieve the in vivo P-gp inhibition effect is usually 15 mg/kg (40–42), which was the dose used in the present study. Unfortunately, a prototypical P-gp inhibitor is currently not available and the dosing regimen used in this work does not allow us to conclude whether the observed efflux inhibition is due to P-gp only or whether BCRP transporters are also involved in LEV's prostate penetration.

Conclusions.

The present work evaluated how the efflux transporters affect the LEV tissue distribution into lung and prostate. A population PK model was successfully developed to characterize LEV simultaneously in total plasma and the ISF of lung and prostate in the presence and absence of P-gp inhibitor for i.v. and i.t. administration. The developed model was able to quantify the interindividual variability in the kinetic disposition of LEV in the investigated tissues. Our results support the idea of a P-gp impact on renal active secretion of LEV, but efflux transporters apparently have minimal effect on drug distribution into the lung following intravenous administration.

Following intratracheal administration, efflux transporters play a role in tissue penetration into lung because P-gp is mainly expressed at AP side of the cells. Our results further indicate that efflux transporters also play a substantial role in LEV penetration into the prostate. However, additional studies are required to identify which efflux transporter is most important for the prostate. The developed model allowed us to integrate all experimental data and, thus, to obtain a better understanding of the mechanisms involved in the distribution of LEV into lung and prostate. This model may now be used in conjunction with antimicrobial efficacy data to optimize the LEV dose and thus to improve antimicrobial efficacy while decreasing the emergence of bacterial resistance.