Flutamide - Hydroxypropyl-ß-cyclodextrin Complex: Formulation, Physical Characterization, and Absorption Studies using the Caco-2 in vitro Model

Flutamide - Hydroxypropyl-ß-cyclodextrin Complex: Formulation, Physical Characterization, and Absorption Studies using the Caco-2 in vitro Model

Manuscript received March 21st, 2000, Revised May 3rd, 2000; Accepted June 26th, 2000.

Bruce Stevenson
Department of Cell Biology, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Alberta, Canada

Purpose: The objective of this research was to formulate flutamide (FLT) in hydroxypropyl-ß-cyclodextrin (HPßCyD), and to investigate FLT transcellular permeation from the complex using the Caco-2 monolayer in vitro model.

Methods: Classical solubility data were used to derive thermodynamic parameters which, together with Differential Scanning Calorimetry (DSC), ¹H-NMR and ¹⁹F-NMR, were used to characterize and derive stability constants for the FLT-HPßCyD complex. The Caco-2 cell line was used to examine the role of HPßCyD on the passage of FLT across cell monolayers in vitro.

Results: The solubility of FLT in water (1.46 mmol/L) increased almost 170 times (to 243.45 mmol/L) in the presence of 50% (w/v) HPßCyD. Solubility data for FLT in aqueous HPbCyD were used to derive thermodynamic parameters (DG° at 298 K = -3.48, DH° = 2.85, DS° at 298 K = 21.24). The solubility of FLT in HPßCyD increased proportionally with an increase in temperature. The FLT-HPßCyD complex had an A_L-type (DSC) isotherm, consistent with a linear increase in FLT solubility and unchanged stoichiometry. The DSC of free FLT and HPßCyD showed endothermic peaks at 110 °C and 300 °C, respectively. FLT-HPßCyD did not display a free-FLT endothermic response, but exhibited broadening of the endothermic peak in the HPßCyD region. ¹⁹F- and ¹H-NMR chemical shifts of FLT moved upfield as a function of its increased solubility in the presence of HPßCyD. The FLT-HPßCyD stability constant, Ks (1:1) was estimated to be 356 M^-1 and 357 M^-1, from thermodynamic and ¹⁹F NMR data, respectively. The apical-to-basal permeability coefficient (P_eff = 4.75´10^-5 cm.s^-1) for FLT across Caco-2 cell monolayers at 37 °C increased as HPbCyD concentrations were reduced, indicative of transepithelial passage via passive diffusion of available free FLT in solution. Studies in the presence and absence of Ca²⁺ ruled out a significant paracellular transport component.

Conclusions: FLT-HPßCyD is a relatively stable, 1:1 inclusion complex. Formation of this complex substantially increases the water solubility of FLT, but HPßCyD, except in high dilution, reduces transcellular passage of FLT in the Caco-2 cell in vitro model.

Introduction

Flutamide (2-methyl-N-[4-nitro-3-(trifluormethyl)-phenyl] propanamide; FLT; Fig. 1) is a non-steroidal fluorine-containing antiandrogen used in prostate cancer chemotherapy (1). FLT is a prodrug that is rapidly metabolized to hydroxyflutamide, its major, active metabolite (2). The recommended dose (250 mg orally three times daily) is associated with nausea, diarrhea, vomiting and increased appetite (3-5). The low water solubility (40 mg/mL) and poor wettability of FLT may contribute to its low absorption from the commercially available tablets, and preclude intravenous (iv) loading. Formulations that produce higher concentrations of FLT in solution may therefore provide important therapeutic options for patients.

Figure 1. Chemical structure of flutamide (FLT).

Cyclodextrins (CyDs) are "host" molecules that trap a great variety of "guest" molecules having the size of one or two benzene rings, and thereby increase their water solubility without the need for organic co-solvents or surfactants. Beta cyclodextrin (bCyD; 7 glucose residues) has ideal dimensions to complex a range of commonly used drugs. Unfortunately, it also has a particularly high affinity for cholesterol, forming a poorly soluble cholesterol-bCyD complex which may crystallize in the kidneys and cause nephrotoxicity. Hydroxypropyl-ß-cyclodextrin (HPßCyD), a chemical derivative of ßCyD, similarly improves the solubility of many drugs, but it is more hydrophilic than the ßCyD, forms a less stable complex with cholesterol, and is therefore less toxic. (6). This work therefore focuses on increasing the solubility of FLT by forming an FLT-HPßCyD inclusion complex.

The inadequate absorption of FLT through the gastrointestinal tract (GI) may not be due to its inability to permeate the intestinal epithelium, but to its low concentration at the absorption surface. The most commonly used, commercially available, human intestinal cell line for drug absorption studies is Caco-2 (7-9). The Caco-2 cell line differentiates spontaneously to enterocyte-like cells under conventional cell culture conditions. Advantages of the Caco-2 cell line model include rapid evaluation of the transepithelial permeability coefficients (P_eff) and absorption mechanisms under controlled conditions.

The objectives of this study were to solubilize FLT in the presence of HPbCyD, to physicochemically characterize the FLT-HPbCyD inclusion complex, and to study the passage of FLT across Caco-2 monolayers.

materials and methods

FLT, HPßCyD (average molar substitution 0.8), polyethylene glycol (M_n@ 200), dimethylsulfoxide (DMSO), Dulbecco's Modified Eagle's Medium (DMEM), fetal bovine serum (FBS), L-glutamine, transferrin, penicillin-streptomycin, insulin, trypsin and EGTA (ethylene glycol-bis-(b-aminoethyl ether)-N,N,N’,N’–tetraacetic acid) were purchased from Aldrich Chemical Co. PBS (pH 7.4) was prepared in-house (contains 140 mM NaCl, 260 mM KCl, 8.1 mM Na₂HPO₄, 1.47 mM KH₂PO₄). PBS⁺ used in the transport experiment was prepared by adding 225 mL of 2 M CaCl₂ and 200 mL of 1M MgCl₂ to 500 mL of normal PBS. All chemicals and solvents were commercial analytical reagent grade; HPLC grade distilled water was used throughout the study.

FLT was analyzed spectrophotometrically at 300 nm in methanol:water (1:1 v/v) (PHILIPS PU 8740 UV/VIS Scanning Spectrophotometer). The measured e_300nm was 5709 mol-1 cm-1.

Solubility. Excess FLT was added to aqueous solutions of HPbCyD (0 to 333.3 mM) in 20-mL screw-capped vials. After 2 min. of vortex mixing (Vortex Genie^TM) and 5 min. of sonication in a water bath (Branson 2200), samples were shaken in a water bath (Dubnoff Metabolic Shaking Incubator) at 25°C for 3 h. Preliminary experiments showed that 3 h was an appropriate equilibration time. Approximately 0.5 mL of the solution was withdrawn by 1 cc syringe and filtered through a 0.45 mm membrane (Millipore) fixed in a holder (Nucleopore). These phase solubility experiments were repeated at temperatures at 30, 37, 45, 50 °C. Filtrate (0.20 mL) was withdrawn from each sample at each temperature and analyzed by UV after appropriate dilution with methanol: water (1:1 v/v). These solubility data were used to calculate the binding constants for formation of the FLT-HPßCyD complex. The Van’t Hoff equation was used to derive the thermodynamic parameters. (10).

FLT-HPßCyD inclusion complex formation. Excess FLT was added to an aqueous solution of HPßCyD (50% w/v). This mixture was filtered and the filtrate was lyophilized (Freeze Dryer 3, Labconco) to produce the solid inclusion complex as a white powder.

Differential Scanning Calorimetry (DSC). A DSC 120 (SEIKO SII, model SSC/5200) differential scanning calorimeter was used. Free FLT, HPßCyD, the physical mixture of FLT and HPßCyD, and the inclusion complex (FLT-HPßCyD) were sealed in separate aluminum pans, with a sealed empty aluminum pan as reference. The transition temperature of each sample was obtained from its DSC plot.

Nuclear Magnetic Resonance Spectroscopy. ¹⁹F NMR spectra were recorded at 270 MHz on a Bruker AM 300 spectrometer equipped with MacSpect 3 data module. Solutions of FLT (0.5 mM) and HPßCyD (1.7 mM to 33.3 mM) were prepared in D2O with KF (0.2 M) as internal standard. ¹⁹F chemical shifts were compared with a reference sample that contained only FLT and KF in water. Samples containing excess FLT in D2O, or excess FLT in 1% (w/v) HPßCyD in D2O, were equilibrated (2 min vortex mixing, 5 min sonication and 3 h shaking in water bath) and filtered for ¹H NMR analysis at 500 MHz on a Varian spectrometer.

Caco-2 cells. Caco-2 cells were maintained at 37 ^oC in DMEM containing 20% (v/v) fetal bovine serum, 0.058% (w/v) glutamine, 0.001% (w/v) transferrin, 1% (v/v) PBS, and 50 unit/100 mL insulin, in a 5% CO₂, 90 % relative humidity atmosphere. Cells grown in 25 cm² flasks (Corning Costar) and passaged every 5 days at a split ratio of 1:5 reached confluence within 6-7 days after passage. For FLT transport studies, 1.4´10⁶ cells (Spencer hemocytometer) were seeded in each Transwell^Ò insert (24.5 mm diameter, 4.71 cm², pore size 0.4 mm, Corning Costar). The medium in the Transwell^Òwas changed daily. The integrity and permeability of cell monolayers was determined by transepithelial electrical resistance (TEER) measurements (MILLICELL^ÒERS, ENDOHM-24). PBS⁺ (1 mL) was added to the apical chamber of the Transwell and 4 mL of PBS⁺ were added to the basolateral chamber, respectively. TEER values across the Transwell filter, in the absence of cells, were used as reference.

Caco-2 trans-monolayer passage of FLT. Passage of FLT from the apical to the basolateral chamber was studied when the TEER of cell monolayers on the Transwell reached 200 W cm². FLT or FLT-HPbCD solutions (various concentrations in 1.5 mL of PBS⁺ or medium) were carefully added onto the apical surface of the monolayer. During incubation at 37 °C, FLT concentrations in the apical and basolateral chambers were periodically measured by HPLC (Waters 501 HPLC pump, Waters 486 Tunable Absorbance Detector, Waters Radial Pak C18 10 mm reverse-phase column and mBondapakTM C18 Guard-PakTM guard column). To study the role of paracellular passage of FLT, the Caco-2 cell monolayer was pretreated for 45 min with a low calcium medium containing 2.5 mM EGTA to complex free Ca²⁺ ions. Sink conditions were maintained in the basolateral chamber by moving the Transwell^Ò to a fresh PBS⁺ well of a 6-well plate at predetermined intervals. The integrity of the monolayers was checked by measuring TEER at the end of each experiment.

Data analysis. The effective permeability coefficient (Peff) was calculated using the equation (11):

P_eff= V/A*C₀x dc/dt

where dc/dt, the flux across the monolayer (FLT, mM/s), is the initial slope of a plot of the cumulative receiver concentration (basolateral chamber) versus time, V is the volume of the receiver chamber (mL), A is the surface area of the monolayer (here 4.7 cm²) and C₀ is the initial concentration (mM FLT) in apical compartment.

results and discussion

Solubility of FLT at different concentrations of HPßCyD. The solubility of FLT increased approximately 170-fold in the presence of HPßCyD (Table 1). The solubility limit of the complex was not reached within the range of concentrations of HPßCyD used in this study. The stoichiometry and solution stability of the inclusion complex can be determined from the slope and intercept of the phase-solubility plot of FLT solubility as a function of HPßCyD concentration. The DSC isotherm is type A_L, implying a linear increase in solubility with unchanged stoichiometry and a 1:1 FLT:HPßCyD binding ratio (12). This complex is highly water-soluble at room temperature, since no precipitation was observed even at HPßCyD concentrations as high as 0.35 M. The solubility of FLT increased almost 170 times at the highest HPßCyD concentration used. The stability constant Ks for 1:1 FLT-HPßCyD was calculated to be 356 M^-1, using the equation:

Ks (1:1) = Slope / So(1-Slope)

where So is the solubility of the FLT in the absence of HPßCyD (13).

Temperature effect on the aqueous solubility of FLT in HPßCyD. The solubility of FLT in 5% HPßCyD increased proportionally with increases in temperature. Thermodynamic parameters calculated from these data are shown in Table 2. FLT dissolution thermodynamics in aqueous HPßCyD were characterized by a negative DG°, indicative of spontaneous dissolution; and a positive DH°, indicative of endothermic dissolution. It is reported that the driving forces for inclusion complexation between CyD and a guest molecule may include Van der Waals interactions, hydrogen bonding, hydrophobic interactions, release of high-energy water molecules from the cavity of CyD, and release of strain energy in the ring of CyD (14). Usually complex formation with HPßCyD results in a relatively large negative DH°, and a DS° that can be either positive or negative (15). The large DS° (+20 cal.K-1 mole-1) for FLT can be attributed to the transfer of FLT from aqueous medium to a more apolar site, such as the cavity of HPßCyD. This transfer involves breakdown of water structure around FLT, which creates a large positive DS° and a small positive DH°, apparently governed by hydrophobic interactions.

DSC analysis. The DSC isotherms of free FLT and HPßCyD are characterized by sharp endothermic peaks at 110 °C and 300 °C, respectively, and a physical mixture of FLT and HPßCyD exhibits both of these endothermic peaks, although the peak for HPßCyD is only barely discernable. However, the DSC of the FLT-HPßCyD complex showed no endothermic peak for FLT, and the endothermic peak of HPßCyD was appreciably broadened (Figure 2). This was taken as another indication of inclusion complex formation.

NMR studies. DSC is only one of the analytical methods used to confirm the formation of inclusion complexes. Although the DSC test was positive for FLT complexation with HPßCyD, NMR was used for further confirmation. The ¹⁹F NMR chemical shifts of fluorine atoms in FLT, as a function of the concentration of HPßCyD, are shown in Table 3. The observed maximum chemical shift change reached 1.36 ppm, in 33.3 mM HPbCyD. This was taken as further proof of the formation of an inclusion complex between HPbCyD and FLT. The free and complexed forms of FLT gave rise to only one NMR fluorine signal, possibly due to fast exchange between free and bound FLT at equilibrium. The stability constant (Ks) for a 1:1 complex in the presence of a large excess of HPbCyD, was derived from NMR parameters using the Benesi-Hildebrand method (16, 17):

where Ddobs is the ¹⁹F chemical shift difference between free FLT (in the absence of HPbCyD) and the observed value from each sample. [HPbCyD]_total refers to the total concentration of HPbCyD, including free and complexed HPbCyD. Ddc represents the ¹⁹F chemical shift difference between free FLT and FLT in the pure HPbCyD-complex. The negative slope of a plot Ddobs×([HPbCyD] total)-1 against Ddobs will generate Ks. The Ks obtained from this procedure is 357 M^-1 at 298 K, and is in good agreement with the value obtained from the phase solubility test (356 M^-1at 298 K). Since the CF₃ group is on the aromatic ring, and because the F atoms are undergoing a change in chemical shift, one may rationalize that the benzene ring enters the cavity of HPbCyD.

The HPbCD used in the project was a mixture of substitution isomers with an average molar substitution of 0.8. Although the ¹H chemical shift changes of such a mixture of HPbCD's was too complex for analysis, the readily observed chemical shift changes of the FLT aromatic protons further support the model in which the aromatic ring is inserted into the hydrophobic inner cavity of HPbCD. The FLT proton resonances in D₂O occur at 4267, 4000 and 4105 Hz. Decoupling the ¹H doublet at 4000 Hz (aromatic region) of the proton NMR spectrum collapsed it to a singlet at 4105 Hz. In the FLT-HPbCyD complex, the singlet at 4267 Hz shifted upfield to 4242 Hz, while the doublet at 4000 Hz shifted downfield to 4084 Hz. The shift of the doublet at 4105 remained unchanged.

Passage of FLT across Caco-2 cell monolayers. Studies of FLT passage across Caco-2 monolayers at several concentrations of HPbCyD (0.15 to 2.5%) revealed that both the percent passing through the monolayer and the P_eff decreased with the increasing concentrations of HPbCyD (Figure 3). These data imply that the P_eff increases when more free FLT is available, and that trans-monolayer passage is due to passive diffusion of available free FLT. Since dilution will increase FLT release from its HPbCyD inclusion complex, it will increase the diffusion of FLT across the monolayer. There are no literature reports of FLT metabolism by Caco-2 cells, and HPLC analysis for FLT in the apical and basolateral chambers during these studies did not detect any evidence of metabolites.

Haeberlin et al. (18) reported on CyD-associated absorption enhancement. They found that b-, g- cyclodextrin, HPbCyD and DMbCyD did alter oral absorption of modified calcitonin and octreotide in vitro in Caco-2 cell monolayers and in situ in isolated rat jejunal sections, by enhancing permeation of the mucosal membrane. Hovgaard and Brøndsted (19) studied the effects of CyD's as absorption enhancers using the Caco-2 monolayer model. b-CyD (1.8 %), a- and g-CyD (5 %) and HPbCD (5 %) were compared with DMbCyD (2.5% and 5%). They concluded that DMbCyD was the most effective in increasing the permeability of the cytoplasmic membrane in a concentration dependent manner. No significant beneficial effects on transport were seen for the other CyDs relative to the control.

Drugs that are well orally absorbed in vivo have threshold P_eff values around 10^-6 cm/s in the Caco-2 cell line, while drugs that have lower P_eff values are poorly absorbed in animals or humans (20). By extension, current experiments indicate that FLT (P_eff= 4.75´ 10^-5 cm.s^-1) should be well absorbed. If this is correct, then its low bioavailability may be due to the low amount of free drug in solution, which in turn reflects slow dissolution from the formulation.

Epithelial cells are joined through a complex of three separate structures: tight junctions, intermediate junctions and desmosomes. The integrity of these structures, and consequently the paracellular passage of drugs, is dependent on Ca⁺² ion concentration (21). The contribution of paracellular pathways to the overall trans-monolayer passage of FLT was investigated by replacing the PBS in the standard the low Ca²⁺buffer cell culture medium [DMEM] with normal PBS⁺ buffer containing 0.9 mM Ca²⁺. This increases the Ca²⁺ concentration to at least 10 times higher than the normal intracellular concentration (the 100 nM range) (22). After a 45 min FLT absorption experiment with EGTA (2.5 mM) in the buffer to chelate Ca²⁺, the TEER returned to normal values within a 2 h recovery time, indicating that membrane changes due to calcium removal were reversible. It has been reported that exposure of the cells to low Ca²⁺ medium for a time longer than 90 min will result in an irreversible decrease in the TEER (23). In the current experiments, reversible opening of the monolayer tight junctions had no apparent effect on the trans-monolayer movement of FLT (Figure 4). This implies that the trans-cellular passage of FLT is the major contributing pathway, and that the paracellular pathway is of limited importance for FLT absorption.

The Caco-2 monolayers have TEER > 200W·cm², high enough to reduce the transport of hydrophilic compounds to a very low level, so that any small change in paracellular permeability is readily detectable. In general, the opening of the tight junction complex has no effect on the passage of the more lipophilic drugs. These drugs partition rapidly into cell membranes, and their distribution to the intercellular spaces will therefore be limited even after EGTA treatment, since the cell membranes cover a much larger surface area than the intercellular spaces. However, more hydrophilic drugs have lower solubility in the cell membranes, and trans-membrane passage via paracellular channels will be more important for them.

The water solubility of FLT was increased 170-fold in the presence of 50% (w/v) HPßCyD. Thermodynamic parameters derived from FLT solubility in the presence of various concentrations of HPßCyD at several temperatures revealed that the solubility of FLT increased proportionally with an increase in temperature. FLT-HPßCyD inclusion complex formation was characteristic of a very strong hydrophobic interaction. The complexes were characterized by an A_L-type DSC isotherm indicative of a linear increase in FLT solubility with unchanged stoichiometry of the complex. DSC of free FLT and HPßCyD displayed endothermic peaks at 110 °C and around 300 °C, respectively, whereas the DSC of the complex gave no indication of free FLT.

¹⁹F NMR chemical shifts of F atoms in FLT-HPßCyD moved downfield as a function of increased FLT solubility in the presence of HPßCyD. The stability constants calculated based on the ¹⁹F NMR chemical shift were similar to one calculated from the phase solubility test. ¹H NMR further confirmed the inclusion formation, showing aromatic proton chemical shift changes upon the inclusion complex formation.

The apical-to-basal passage of FLT across Caco-2 cell monolayers at 37 °C was transcellular, with little or no evidence for paracellular passage. Apical-to-basal passage decreased with increasing HPbCD concentrations, indicating that transcellular passage is due to passive diffusion of free drug reaching the cell surface.