Inhaled corticosteroids (ICSs) remain the first-line therapy for the treatment of asthma (Barnes, 2007). Clinical studies have shown the high antiasthmatic efficacy (e.g., prevention and control of symptoms) and reduced systemic side effects of ICS therapy. However, potential systemic side effects induced by orally swallowed or pulmonary absorbed drug still remain a major concern (Barnes, 1995; Derendorf et al., 1995; Tayab and Hochhaus, 2007). Hypothalamic-pituitary-adrenal axis suppression, bone demineralization, and growth retardation are some of the major systemic effects of corticosteroid therapy.

Several factors determine the risk-benefit ratio of inhaled glucocorticoids. Beneficial properties include low oral bioavailability, high systemic clearance, and sufficient receptor binding affinity (Hochhaus, 2004). The intrinsic clearance of newer glucocorticoids is high, and, consequently, hepatic clearance is close to the liver blood flow and independent of plasma protein binding (Hochhaus, 2004). It is generally acknowledged that the pharmacological activity of a drug is dependent on its free concentration at the receptor site (du Souich et al., 1993). Accepting this “free drug hypothesis,” higher plasma protein binding with unchanged hepatic clearance will translate into lower free plasma concentration and lower systemic side effects.

Consequently, newer inhaled glucocorticoids have been developed that exhibit increased plasma protein binding while maintaining high hepatic clearance. An example is the prodrug ciclesonide (CIC), whose active metabolite shows a very pronounced plasma protein binding (Rohatagi et al., 2005b), low free plasma concentrations, and, consequently, low cortisol suppression (Rohatagi et al., 2003, 2005a). Besides plasma protein binding, drug tissue binding should be considered for a full evaluation of the effects of binding phenomena on systemic side effects and desired pulmonary effects. The volume of distribution quantifies drug distribution between plasma and tissues. For glucocorticoids that are able to easily cross cell membranes, a larger volume of distribution, as observed for newer glucocorticoids, indicates that tissue binding is more pronounced than plasma protein binding. Whereas the importance of high plasma protein binding for low systemic side effects is accepted, the effects of higher tissue binding, e.g., in the lung, are controversial. One opinion is that increased tissue binding in the lung is beneficial for inhaled glucocorticoids because higher tissue concentrations will result in a drug depot that keeps the drug in the lung for a longer time (Esmailpour et al., 2000). However, nonspecific binding commonly displays low affinity; high capacity features and dissociation rates from such interactions are generally very fast (dissociation half-life of less than 1 s) (Talbert et al., 2002). This argues against the possibility of maintaining a “slow release” depot under “sink” conditions (reduction in drug concentrations). Other authors have suggested that the increase in tissue and plasma binding for the newer glucocorticoids will not lead to a reduced antiasthmatic effect because the bound steroid would be stripped from these sites by the extremely high affinity receptors (Daley-Yates et al., 2005). A similar hypothesis was formulated by Van Der Graaf et al. (1997) and van Steeg et al. (2005) (PAGE, http://www.page-meeting.org/?abstract=788) to explain results of a pharmacokinetic/pharmacodynamic (PK/PD) analysis for propranolol in which receptor occupancy was better described by total and not free drug concentrations.

This study was interested in evaluating the effect of varying plasma protein and tissue binding on receptor occupancy in the lung. The binding of budesonide (BUD; 88% plasma protein binding in humans) and des-ciclesonide (des-CIC; the active species of the prodrug CIC; 99% bound to plasma proteins in humans) (Rohatagi et al., 2005b) was determined in rat plasma and rat tissues in the in vitro and ex vivo binding experiments. Free concentrations were determined after intravenous administration of the drugs to rats, and the receptor occupancies in the lung were related to the free tissue drug concentrations. In addition, a previously used PK/PD model was used to simulate differences in protein binding on pulmonary selectivity and targeting after inhalation of drugs that differ in protein binding (Hochhaus et al., 1997).

Materials and Methods

Chemicals. BUD, des-CIC, ³H-BUD (specific activity 32.4 Ci mmol^–1) and ³H-des-CIC (specific activity 38.2 Ci mmol^–1) were provided by Astra-Zeneca (Lund, Sweden).

Determination of Binding of BUD and des-CIC in Spiked Rat Plasma and Tissues by Equilibrium Dialysis. The animal protocol was approved by the local Institutional Animal Care and Use Committee at the University of Florida (Gainesville, FL). Sprague-Dawley male rats, weighing 300 to 350 g, were anesthetized by inhalation of isoflurane and then decapitated. Following decapitation, the blood was collected and centrifuged, and the plasma was collected. The brain, kidney, liver, and lung were removed, rinsed in ice-cold phosphate-buffered saline (PBS), blotted, weighed, and homogenized by adding 5 volumes (for lung) or 3 volumes (for the other tissues) of ice-cold PBS.

Freshly prepared plasma and tissue homogenates were spiked with BUD or des-CIC at two concentrations, 8 or 800 nM, respectively. Three rats were used for each drug at each concentration. One milliliter of spiked rat biometric was added to one chamber of the equilibrium dialysis cell (Spectrum Laboratories, Gardena, CA), and 1 ml of PBS was added to the other chamber. The two chambers were separated by cellulose membrane (Spectrum Laboratories) with 14-kDa cutoff value. The equilibrium dialysis cell was subjected to rotation at 30 rpm in a water bath at 37°C until equilibration of the drug concentrations in the two chambers.

After reaching equilibrium, dialysate and remaining biometric sample (tissue homogenate, plasma) in the dialysis chamber were sampled. All the biological samples were precipitated with 3 parts of acetonitrile/1% acetic acid and then diluted with 3 parts of water. Dialysate samples (PBS) were used directly.

Samples were analyzed by liquid chromatography/tandem mass spectroscopy to determine the drug concentrations for BUD (transitions, 431.6/323.2) and des-Cic (transition, 471.1/323.3) using BUD-D8 as internal standard (transition, 439.5/323.4). The instrument consisted of a Shimadzu LC pump (Shimadzu, Columbia, MD), a CTC autosampler (Agilent Technologies, Santa Clara, CA), and a Sciex API4000 (Applied Biosystems/MDS Sciex, Foster City, CA) mass spectrometer working in a positive electrospray mode. A coupled column system (Aquasil 5 μm, 10 × 1 and 30 × 1 mm) was used at a flow rate of 250 μl/min and an injection volume of 10 μl with mobile phase A consisting of water/0.5% acetic acid and mobile phase B of 95% acetonitrile/5% mobile phase A. A step gradient was used (0.0–0.6 min, 100% mobile phase A to column 1; 0.6–1.8 min, 100% mobile phase A to column B; 1.8–2.6 min, 60% mobile phase A and 40% mobile phase B to column B; subsequent washing of columns 1 and 2 with 100% mobile phase B and 100% mobile phase A). The range was linear in a range of 0.03 to 250 nM (PBS) and 50 to 1000 nM (biological matrix) for BUD and 0.02 to 250 nM (PBS) and 50 to 1000 nM (biological matrix). The coefficients of correlation for the calibration curves were at least 0.998. Using a precision and accuracy of greater than 80% as approval criteria, the lower limit of quantification was 50 nM in biometrics and 30 pM in PBS for BUD. Using the same criteria, the lower limit of quantification was 50 nM in biometrics and 20 pM in PBS for des-CIC.

Calculation of Binding. The free fraction of drug was calculated using the ratio of the concentration in PBS and the concentration in the spiked biometric at equilibrium obtained after dialysis according to eq. 1: where f_u is the fraction of unbound drug in the biometric, C_buffer is the measured drug concentration in the buffer, and C_biometric is the measured drug concentration in the biometric after dialysis. The percentage of bound drug was then calculated as (1 – f_u) · 100. Assuming linear binding (see under Results), the fraction of unbound drug in tissues was calculated from the estimates in tissue homogenate according to eq. 2: where f_u,tissue is the free fraction of drug in the tissue, f_u,homogenate is the free fraction of drug in the tissue homogenate, D is the volume dilution factor of the tissue homogenate: 9.2 for brain, 8.5 for kidney, 8.4 for liver, and 12.5 for lung. The dilution factors of tissue homogenate were converted from w/v to v/v based on the densities of different tissues: 0.98 g/ml for brain, 1.07 g/ml for kidney, 1.08 g/ml for liver, and 0.72 g/ml for lung (Stark, 2001). The fraction bound in the tissue was then calculated as 1 – f_u,tissue.

Determination of Binding of BUD and des-CIC in Rat Plasma and Tissues after Dosing by Equilibrium Dialysis. The animal protocol was approved by the local Institutional Animal Care and Use Committee at the University of Florida. Twelve Sprague-Dawley male rats, weighing 300 to 350 g, were included in the study. Six were treated with BUD, and the others were treated with des-CIC. During the experiment, the rat was anesthetized using urethane. After an initial intravenous dose of the drug at 30 μg/kg, the rat was infused with the drug at the rate of 10 μg/(hkg) for 6 h through the femoral vein. At the end of infusion, the rat was decapitated. Following the same procedure as described in previous sections, the rat plasma and tissue homogenate were collected, and the binding of drug was determined.

Determination of Glucocorticoid Receptor Occupancy in Rat Lung after Dosing. Rats used in the above experiments were also used to determine the pulmonary glucocorticoid receptor occupancy. In addition, four placebo rats were included. The lungs (without the trachea) were resected, weighed, and placed on ice. The lung was then further homogenized using a BioHomogenizer (BioSpec Products, Inc., Bartlesville, OK) (5 s, low speed, with 30-s cool-down period between each step) after the addition of 6 volumes of ice-cold incubation buffer (10 mM Tris/HCL, 10 mM sodium molybdate, and 2 mM 1,4-dithioerythritol). The homogenate was incubated with 5% w/v charcoal suspension (in deionized water) for 5 min. The homogenate was then centrifuged for 10 min at 50,000g in a Beckman Coulter (Fullerton, CA) high-speed centrifuge using a JA-21 fixed angle rotor to obtain a clear supernatant.

Aliquots of the supernatant (160 μl) were transferred into microcentrifuge tubes. Either 20 μl of 100 nM ³H-BUD with 20 μl of ethanol was added to determine total binding or 40 μl of mixture solution with 100 nM ³H-BUD and 100 μM unlabeled BUD was added to determine the nonspecific binding. After a 16- to 24-h incubation period at 4°C, the unbound glucocorticoid was removed by addition of 200 μl of 5% suspension of activated charcoal. The mixture was incubated for 5 min on ice and then centrifuged at 10,000 rpm for 5 min in a microcentrifuge (model 235A; Thermo Fisher Scientific, Waltham, MA). The radioactivity (dpm) in 300 μl of supernatant was determined using a liquid scintillation counter (model LS 5000 TD; Beckman Coulter).

Specific bindings for the lung were used directly as a marker for the number of available glucocorticoid binding sites in the lung because the concentration of ³H-BUD (10 nM) was chosen to be high enough to cover 100% of the total binding sites, and it was assumed that ³H-BUD would not replace bound unlabeled corticosteroid to a significant level as a result of the very slow dissociation rate at 4°C. The specific binding for control rats was treated as the reference level or 100% available binding sites. The ratio of the specific binding for drug-treated rats to that for control rats served as the estimate of percentage of available binding sites. The percentage of receptor occupancy by the unlabeled corticosteroid could be easily calculated from this ratio using eq. 3:

PK/PD Simulations. A previously published PK/PD simulation model (Hochhaus et al., 1997) (Fig. 1) evaluating factors relevant for pulmonary selectivity of ICS was extended to incorporate overall tissue (f_u,T), lung (f_u,L), and plasma protein binding (f_u,p). To reduce some complexity, the model assumed that the oral bioavailability of all the ICS was 0%. Steady-state pulmonary and plasma drug-concentration time profiles were derived for the multiple dosing of an ICS considering the dosing interval τ. The free drug concentrations present in the lung, resulting from dissolved drug particles deposited in the lung, were calculated as C_u,pulm (eq. 4). The equation takes into consideration a defined lung volume (V_L), the amount of pulmonary deposited drug (D_L) calculated from the inhalation efficiency (IE), the total delivered dose (TDD; D_L = TDD · IE), the dissolution rate of the particles (k_diss), the removal of solid particle through the mucociliary clearance (α = k_muc + k_diss), and the absorption rate of dissolved drug from the lung tissue into the systemic circulation (k_a).

The plasma concentration-time profile (C_u,p,plasma) was subsequently calculated in eq. 5 considering the volume of distribution [V_d, calculated from the volume of plasma V_p, the volume of tissue the drug is able to enter (V_T) as V_d = V_p + V_t · f_u/f_uT], and the elimination rate (k_e = Cl/V_d) assuming a one-compartment body model.

The model assumed that the free drug present in the plasma is also available to the lung. Thus, the overall free drug concentration in the lung (C_u,lung) was defined as (eq. 6):

Free lung and plasma concentrations were linked to a standard E_max model to predict the receptor occupancy in lung and the systemic circulation. The receptor occupancy was selected as a biomarker because a large body of literature on the glucocorticoid receptor system suggested that the degree of receptor occupancy correlates closely with the extent of the induced biological response in cell systems (Beato et al., 1972; Diamant et al., 1975). A very close correlation exists between the receptor affinity of a wide range of glucocorticoids and their activity at the site of action (e.g., skin blanching, cortisol suppression) (Dahlberg et al., 1984; Druzgala et al., 1991). Therefore, the glucocorticoid receptor occupancy was used as an surrogate marker of the local and systemic effects by linking free ICS concentrations in the lung (C_u,pulm) and the systemic circulation (C_u,P) to a standard E_max model, where 100 represents the maximum receptors bound to the drug (100%) and k_d is the equilibrium dissociation constant between receptor and glucocorticoid (eq. 7 or 8). The following parameters were used: TDD of 500 or 50 μg, IE of 0.231 h^–1, particle, k_diss of 0.231 h^–1, pulmonary absorption rate (k_a = 10 h^–1), k_d of 80 ng/l, systemic clearance (60 l/h), plasma binding (90 or 99% binding, f_u of 0.01 or 0.1), pulmonary tissue binding (90 or 99% binding f_u,L of 0.1 or 0.01), and general tissue binding (99 or 99.9% binding, f_u,T of 0.01 or 0.001). The overall volume of distribution was calculated from the plasma volume (V_p = 3 liters) and the drug-accessible tissue volume (V_T = 39 liters) as V_d = V_p + V_T · f_u/f_ut. The model is available as an Excel (Microsoft, Redmond, WA) spreadsheet (see under Results).

Results

Preliminary experiments evaluating the in vitro binding of BUD and des-CIC to plasma, kidney, liver, lung, and brain at 8 and 800 nM indicated linear binding for both substances (Table 1). The linear binding ensured that in vivo results would be valid even if the infusions would generate somewhat different concentrations for both substances. Preliminary studies also suggested that the infusion time after administration of the loading dose was sufficient to reach steady state. The degree of binding determined in the ex vivo experiments was comparable with the in vitro results (Table 1).

The total drug concentrations in rat plasma and investigated tissues observed after steady state was achieved and are summarized in Table 2 and Fig. 2. The free drug concentrations were calculated from the percentage binding in Table 1 and the total concentrations measured in the specific tissues (Table 2; Fig. 2). Free concentrations in liver, kidney, lung, and plasma were very similar for a given steroid (0.72–0.79 nM for BUD and 0.10–0.16 nM for des-CIC), with the exception of brain tissue that showed significantly lower free levels (0.18 nM for BUD and 0.03 nM for des-CIC) (Table 2). The receptor occupancy in the rat lung after dosing of BUD or des-CIC was 94.4 ± 8.2 (S.D.), and 48.8 ± 11.5 (S.D.), respectively (Fig. 2).

Results of these animal experiments after systemic administration of des-CIC and BUD were supported by PK/PD simulations predicting the pulmonary and systemic receptor occupancy after inhalation. This simulation module (see Supplemental Data) did not attempt to model the two specific drugs but was only interested in evaluating general relationships between plasma protein and tissue binding and pulmonary and systemic receptor occupancy. Simulations predicted that glucocorticoids with increased protein and lung binding will show lower receptor occupancy for both pulmonary and systemic glucocorticoid receptors when given at the same microgram dose (resulting in similar total concentrations in plasma and tissues). Adjusting the dose, e.g., reducing the dose of the glucocorticoid with lower binding according to differences in the binding (1/10 of the dose in the example shown in Fig. 3), will result in different total but the same free levels and consequently in the same pulmonary and systemic receptor occupancy profiles (Fig. 3).

Discussion

This study evaluated the importance of tissue and plasma protein binding for inhaled glucocorticoids as a response to suggestions that a high protein binding will reduce systemic side effects and be beneficial for topical targeting (Esmailpour et al., 2000) or not affect local and systemic effects (Daley-Yates et al., 2005). All the synthetic glucocorticoids lack affinity to the high affinity protein transcortin (Ballard, 1979) but bind to albumin with low affinity. The low affinity but high capacity of these binding sites is characterized by fast binding kinetics, resulting in almost immediate dissociation once the binding equilibrium is disturbed. Because of the fast dissociation, such binding proteins are unlikely to serve as a slow release depot, contrary to the suggestions of some groups (Esmailpour et al., 2000).

This study compared the total and free drug concentrations, as well as the resulting pulmonary glucocorticoid receptor occupancy of des-CIC (99% plasma protein binding; Rohatagi et al., 2005b) and BUD (88% protein binding; Ryrfeldt et al., 1982), when given as an intravenous infusion. des-CIC is the active species of the inactive prodrug CIC. des-CIC was used to eliminate the activation of the prodrug as a process affecting the steady-state concentrations of des-CIC, potentially leading to differences in the steady-state concentrations of BUD and des-CIC. BUD was selected for comparison because it shows a similar receptor binding affinity and, like des-CIC, a hepatic clearance close to the liver blood flow (Ryrfeldt et al., 1982). The animal experiments were performed after intravenous infusion because the dose can be administered with high precision, contrary to most inhalation models. Constant rate infusion of high extraction drugs BUD and des-CIC also ensured that the resulting steady-state concentrations (Cp_ss) would be similar because these only depend on the infusion rate and systemic clearance and not on protein binding (Cp_ss = k_o/CL). Having achieved steady state (where distribution processes are in equilibrium) ensured that potential differences in total tissue concentrations would only depend on plasma protein and tissue binding differences and not on differences in the rate of tissue distribution. It might be argued that results obtained after systemic administration cannot be relevant because pulmonary drug concentrations and binding characteristics after inhalation will differ. Glucocorticoids have to enter pulmonary cells to induce their effect (Barnes, 2007). This is true both for intravenous and inhalation therapy. Because of their lipophilic character, glucocorticoids easily cross cell membranes (Ballard, 1979), and the degree of binding in pulmonary cells will be the same, independent of the site of administration. Likewise, the potential binding within the surfactant will be the same for both forms of administration because the systemically given drug will probably rapidly enter the interstitial lung fluid and be subjected to binding in the bronchoalveolar fluid (Barth et al., 1989). Thus, assuming concentration-independent binding, tissue binding processes relevant for the pharmacologically active free drug concentration should not differ after inhalation and systemic administration. The above suggests that the ratio between total drug having entered pulmonary tissue and the free drug in lung tissues will not depend on the form of administration.

We used an ex vivo method that measures free concentrations through dialysis of the homogenized tissue and the number of free receptors within an established ex vivo receptor binding assay. Other methods, such as microdialysis, have been suggested to measure free tissue levels. However, microdialysis only measures extravascular concentrations and not intracellular free drug concentrations (Schmidt et al., 2008). Therefore, we decided to use an ex vivo dialysis-based method, especially because this method permitted the measurement of drug binding in a large number of tissues.

Preliminary experiments ensured that the plasma protein and tissue binding was linear and that the results (percentage binding) determined in the ex vivo binding studies were not affected by the dilution steps introduced within the tissue homogenization before and during the dialysis procedure. Total steady-state plasma concentrations (Cp_ss) differed slightly between BUD and des-CIC infusions (Table 2). This effect might indicate a somewhat higher systemic clearance of des-CIC, in agreement with the somewhat lower oral bioavailability of des-CIC found in humans (Hochhaus, 2004). For BUD and des-CIC, total concentrations differed across tissues, with the lowest concentrations seen for the brain. Whereas free concentrations differed between BUD and des-CIC, the free concentrations measured in liver, lung, kidney, and plasma were very similar for a given glucocorticoid, indicating a passive diffusion-based distribution and achievement of equilibrium. The lower free concentration observed in the brain is in agreement with previous findings that inhaled glucocorticoids have a high affinity to P-glycoprotein expressed at the blood-brain barrier (Arya et al., 2005). Differences in total concentrations for a given glucocorticoid across different organs suggest that tissues differ in the degree of binding. Overall, tissue binding is more pronounced than plasma protein binding.

Our experiments showed that the free drug concentrations for des-CIC are not only lower in plasma and systemic organs but also in the lung, which translates into lower pulmonary receptor occupancy for des-CIC (Fig. 2). These results indicate that the free drug hypothesis (only free drug can interact with the receptors) is valid for the investigated glucocorticoids because the smaller free des-CIC concentration in the lung translates into reduced receptor occupancy when compared with BUD. These results clarify two issues, brought up by some investigators, within the context of new glucocorticoids with high receptor affinity and pronounced plasma and tissue binding. First, the reduced receptor occupancy for des-CIC does not support the hypothesis of some investigators that the “extremely” high affinity of newer glucocorticoids will strip the drug from low affinity binding sites such as albumin or lung binding components, and, consequently, the degree of receptor occupancy will not depend on tissue binding (Daley-Yates et al., 2005). Second, increased lung tissue binding has never been critically evaluated as a factor that might potentially affect the clinical outcome at a given dose. Accepting that the degree of receptor occupancy is a valid surrogate marker of pulmonary effects and systemic side effects, our results seem to suggest that glucocorticoids with high tissue binding will also exhibit lower free concentrations and a lower receptor occupancy in systemic tissues, but more importantly also in the lung when given at similar doses, than glucocorticoids that exhibit lower plasma and tissue binding. Our results of increased tissue binding of des-CIC in systemic tissues are in agreement with a human microdialysis study comparing the free drug concentrations of BUD and des-CIC in adipose tissue and skeletal muscle after inhalation of high doses of CIC and BUD (European Respiratory Society, http://www.ers-education.org/pages/default.aspx?id=391&idBrowse=2303). In this study, free concentrations of BUD were detectable in the tissues, whereas the unbound concentrations of des-CIC were negligible. Whereas the authors concluded that these results suggest a lower potential of CIC for systemic side effects, our results in rats clearly indicated that this is not only the case for undesired side effects but also for the desired pulmonary effects.

The dual effects of protein binding on both pulmonary and systemic free drug concentrations were confirmed by a PK/PD simulation model that assessed the effect of tissue and plasma protein binding on pulmonary and systemic receptor occupancy. This model has been previously used to assess and identify PK and PD parameters relevant for pulmonary selectivity (Hochhaus et al., 1997). Two hypothetical glucocorticoids, differing only in tissue and plasma protein binding, were compared at the same pulmonary dose. Results were in agreement with the in vivo experiments, indicating lower free drug concentrations in the lung and the systemic circulation (data not shown) for the drug with more pronounced tissue and plasma protein binding. Thus, inhalation of the “strong binder” resulted in a smaller number of occupied receptors in the lung and the systemic circulation (Fig. 3). When the dose of the “weaker binder” was decreased according to the differences in the tissue binding, the resulting receptor occupancy profiles for the weak and strong binder were identical. Thus, differences in the protein and tissue binding can be accounted for by adjusting the dose, or in other words, when the dose of the strong binder is increased to achieve the same pulmonary effects, side effects will be identical to those observed for the weak binder.

In conclusion, the results showed that the free drug hypothesis is valid for newer inhaled glucocorticoids. This suggests for the clinicians that such glucocorticoids will show lower systemic side effects yet will provide lower pharmacologically relevant concentrations in the lung when given at similar doses than established drugs. Therefore, it cannot be assumed that lung selectivity (i.e., the ratio between local and systemic effects) necessarily is improved by a high plasma protein binding.