Journal of Pharmaceutical Sciences xxx (2016) 1e11

Contents lists available at ScienceDirect

Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org

Pharmaceutics, Drug Delivery and Pharmaceutical Technology

Solid Phospholipid Dispersions for Oral Delivery of Poorly Soluble Drugs: Investigation into Celecoxib Incorporation and Solubility-In Vitro Permeability Enhancement Sophia Yui Kau Fong, Susana M. Martins, Martin Brandl, Annette Bauer-Brandl* Department of Physics, Chemistry and Pharmacy, University of Southern Denmark, 5230 Odense M, Denmark

a r t i c l e i n f o

a b s t r a c t

Article history: Received 24 September 2015 Revised 10 November 2015 Accepted 1 December 2015

Celecoxib (CXB) is a Biopharmaceutical Classification System class II drug in which its oral bioavailability is limited by poor aqueous solubility. Although a range of formulations aiming to increase the solubility of CXB have been developed, it is not completely understood, whether (1) an increase in CXB solubility leads to a subsequent increase in permeability across intestinal barrier and (2) the presence of bile salts affects the solubility and permeability behavior of CXB formulations. By formulating CXB solid phospholipid (PL) dispersions with various PL-to-drug ratios using freeze drying, the present study illustrated that the enhancement of CXB solubility was not proportionally translated into enhanced permeability; both parameters were highly dependent on the PL-to-drug ratios as well as the dispersion media (i.e., the presence of 3-mM sodium taurocholate). This study highlights the importance of evaluating both, solubility and permeability, and the use of biorelevant medium for testing the candidate-enabling performance of liposomal formulations. Mechanisms at molecular level that may explain the effect of PL formulations on the permeability of CXB are also discussed. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.

Keywords: celecoxib phospholipid solid dispersion solubility permeability amorphous sodium taurocholate

Introduction The effective absorption of an orally administered drug is governed by 2 primary factors: dissolution within the gastrointestinal fluids and permeation across the intestinal epithelial barrier. Recent increases in the use of combinatorial chemistry and highthroughput screening in drug discovery, however, have led to an increasing number of drug leads with poor water solubility but high permeability (Biopharmaceutical Classification System class II).1,2 Celecoxib (CXB), for instance, is one of the Biopharmaceutical Classification System class II drugs, which has a high hydrophobicity (log P 3.68) and low water solubility (99%) was purchased from Selleck Chemicals (TX). Lipoid E80 PL, an egg lecithin containing 80%-85% phosphatidylcholine and 7.0%-9.5% phosphatidylethanolamine according to the specification from manufacturer, was donated from Lipoid GmBH (Ludwigshafen, Germany). Sodium taurocholate hydrate, sodium dihydrate phosphate monohydrate, sodium chloride, sodium hydroxide, Triton X-100, methanol (HPLC grade), formic acid, and tertiary butanol were purchased from Sigma-Aldrich (St. Louis). Preparation of CXB-PL Solid Dispersion by Freeze Drying The thermodynamic solubilities of CXB and PL in t-butanol were first experimentally determined, which were found to be 15.0 ± 3.5 mg/mL (CXB) and >500 mg/mL (PL), respectively. Stock solutions of CXB (10 mg/mL) and PL (50-250 mg/mL) were prepared in t-butanol at concentrations below their thermodynamic solubilities. No crystals were detected in the prepared stock solutions by optical microscope which indicated that CXB and PL were well dissolved in t-butanol at the selected concentrations. Formulations with different PL-to-drug ratios, ranging from 2.5:1 to 250:1 (m/m), were prepared by diluting the stock solutions using water in t-butanol (7:93, m/m; Table 1). The prepared formulations were frozen at 80 C for 24 h before placing in the precooled freeze

Table 1 Compositions of the Studied Formulations Formulation Code

CXB crystalline CXB FDa 2.5: 1 10:1 50:1 100:1 250:1 a

Composition (Mass, wt/wt) PL

CXB

0 0 2.5 10 50 100 250

1 1 1 1 1 1 1

FD: freeze-dried CXB.

dryer at 60 C. Freeze drying was performed with Christ Gamma 2-16 LSC Freeze Dryer (Martin Christ GmbH, UK) according to the following program: a main drying phase with shelf temperature at 25 C and pressure of 0.1 mBar for 24 h, followed by a final drying phase with shelf temperature at 25 C and pressure of 0.01 mbar for 4 h. The vials were sealed and stored at room temperature in a desiccator above calcium chloride until analysis. Structure Analysis Differentiation Scanning Calorimetry Differential scanning calorimetry (DSC) analysis was performed using a PerkinElmer® DSC 8500 calorimeter (MA). Accurately weighed samples of crystalline CXB, freeze-dried CXB, and PL-CXB (2.5:1) formulation were hermetically sealed in aluminum pans and heated at a rate of 10 K/min from 30 C to 180 C under a nitrogen atmosphere. A similar empty aluminum pan was used as the reference. Thermograms were analyzed using Pyris® software. The instrument was calibrated using indium standard (purity >99.9%, melting point 156.6 C, enthalpy of fusion 28.5 J/g) purchased from PerkinElmer® (MA) and performed according to manufacturer’s instruction. Powder X-Ray Diffraction Powder X-ray diffraction (PXRD) analysis was carried out by an X-ray diffractometer (MiniFlex 600; Rigaku®, Japan) with Cu Ka radiation (l ¼ 1.5418 Å). The powder samples of crystalline CXB, freeze-dried CXB, and PL-CXB (2.5:1) formulation were measured within an angular range of 14 -27 2q, with a step size of 0.02 under the following conditions: current 10 mA, voltage 30 kV, and scanning speed 10 2q/min. The percentages of crystallinity of the freeze-dried CXB and the PL-CXB formulation were calculated from the following equation22:

Percentages of crystallinity ð%Þ ¼

Ic  100% Ix

where Ic and Ix are the intensities of X-rays scattered from the crystalline regions (area under the most sharp peak in the X-ray diffractograms of CXB) in the freeze-dried samples and the crystalline CXB sample, respectively. Drug Solubilization Studies Preparation of Liposome Dispersions Liposome dispersions were formed by hydrating the freezedried samples with 2 media, respectively: (1) PBS buffer (pH 6.5) and (2) PBS buffer containing 3 mM of sodium taurocholate (pH 6.5). Such liposome dispersions typically contain a mixture of different types of vesicles with varying size and lamellarity, in which the large multilamellar vesicles are the most abundant ones.

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

During solubilization studies,23 that is, when excess, precipitated drug is to be removed from the liposome dispersion by ultracentrifugation (see the subsequent section), rather small and homogeneous liposomes are needed. To reduce both, the size and heterogeneity of mechanically dispersed liposomes, the lipid dispersions were probe sonicated at 50 W in an ice bath for 3 cycles of 5 min each using a Labsonic U probe sonicator (B. Braun Biotech International, Leverkusen, Germany). After sonication, the liposome dispersion (the micelles in case of taurocholate medium) was refrigerated at 4 C overnight to equilibrate. Ultracentrifugation For the separation of liposomes (or micelles) from precipitated drug, the dispersions were centrifuged for 25 min at 100,000 g at 10 C, using an Optima L8-70M ultracentrifuge, equipped with an SW 60 Ti rotor (Beckman Instruments, Inc., Brea, CA). The liposomes-micelles (supernatant parts) were diluted with PBS in methanol (40:60, wt/wt) before analysis. Phosphatidylcholine Quantification The content of phosphatidylcholine in the supernatant was determined to quantify the liposome recovery under centrifugation using an enzymatic assay described by Grohganz et al.24 with a commercial test kit (Phospholipids B; Wako Chemicals USA Inc., VA). Briefly, 50 mL of the liposome sample (dissolved in PBS containing 5% Triton [wt/wt]) and 250 mL of phospholipid B reagent solution were allowed to react for 30-45 min while being incubated in the titer plate reader at 37 C. UV absorbance was measured at l ¼ 492 nm against a blank consisting of 50 mL of PBS buffer (with 5% Triton) and 250 mL of phospholipid B reagent solution. Both samples and blank were measured in triplicates. Lipid content in the supernatant relative to the total content was expressed as % PL recovery. Permeability Studies Franz diffusion cells (SES GmbH-Analysesysteme, Bechenheim, Germany) with dialysis cellulose membranes (MWCO 12-14 kDa; Sigma-Aldrich) placed between the donor and acceptor compartments were used as a measure of permeability. The effective diffusional area was 1 cm2, and the compartment capacities of the upper and lower chambers were 2 and 8 mL, respectively. A reverse dialysis setup was used so that the lower chamber was used as the donor compartment. Before the permeation experiment, crystalline CXB (2 mg/mL) and the PL formulations (in amounts equivalent to 444 mg/mL of CXB) were freshly dispersed in 2 media, respectively: PBS buffer (pH 6.5) or PBS buffer containing 3 mM of sodium taurocholate (pH 6.5). The concentrations of CXB were chosen according to their respective solubility limits in PBS buffer (pH 6.5). At time 0 (t ¼ 0), 8 mL of the prepared dispersions were loaded onto the donor compartment while the acceptor compartment was filled with 1.8 mL of PBS buffer (pH 6.5). Samples of 100 mL were withdrawn at 30-min intervals over 6 h, and the last sampling time point was t ¼ 24 h. The withdrawn volume was replaced by an equal amount of fresh buffer at each respective time point to ensure sink conditions. All experiments were performed at 25 C and in triplicates. The amount of CXB in the acceptor compartment was quantified by the analytical method described in the following. The cumulative amount of CXB that had permeated through the barrier was plotted against the time, giving the cumulative flux. When the flux reached steady state (i.e., the slope was linear), the steady state flux (Jss), defined as the amount of permeant crossing the membrane at a constant rate, was calculated according to the following equation25:

Jss ¼

3

dm 1  dt A

in which dm is the cumulative amount of CXB permeated by the time dt, and A is the effective diffusional area of the membrane. Steady state conditions of the fluxes (r2 0.99) were typically achieved after 1 h. Quantification of CXB by HPLC-UV Detection HPLC-UV detection (2487 Dual Absorbance; Waters, MA) was used for the quantification of CXB in all samples. Chromatographic separation was achieved by a reverse-phased Acclaim® C18 column (150 mm  4.6 mm i.d., 3-mm particle size; Thermo Fisher) equipped with a Acclaim® guard filter (5 mm, Thermo Fisher). Isocratic elution with mobile phase consisting of 0.1% formic acid:methanol (16:84) at a flow rate of 1 mL/min was used, and the oven temperature was set at 30 C. The UV detection wavelength of CXB was set at 254 nm, and the injection volume was 20 mL. Working solutions for calibration curve was prepared by serial dilution of a CXB stock solution (1 mg/mL in methanol) with PBS in methanol (40:60, wt/wt). To avoid the bias to the lower concentration range of the calibration curve by the high concentrations, the calibration curves were separated into 2 ranges, 0.1-2 mg/mL and 1-200 mg/mL, and both ranges showed a good linearity (r2 > 0.999). The detection limit and quantification limit of CXB by the developed assay were calculated by the following equations:

Detection limit ¼

3:3s S

Quantification limit ¼

10 s S

in which s is the SD of the response and S is the slope of the calibration curve. It was found that the detection and quantification limits of CXB were 9.2 and 27.9 ng/mL, respectively. Therefore, the assay method should be sensitive enough for the quantification of CXB in the present studies. Statistical Analysis Data are presented as mean ± SD. Statistical analyses were performed using Student t-test (for comparing the results between the 2 dispersion media), 1-way ANOVA with Tukey post hoc (for comparing the results among the different formulations with the same testing media), or Pearson coefficient (for correlating apparent solubility and permeability) by SPSS® Statistics 16.0 (SPSS Inc.). A p < 0.05 was set as the criterion of significance. Results Structure Analysis by DSC and PXRD DSC analyses were performed to evaluate the physical state of CXB in the formulation. DSC thermograms of pure drug (crystalline CXB), freeze-dried CXB, and CXB-PL solid dispersion (PL:drug 2.5:1) are shown in Figure 1a (i), (ii), and (iii), respectively. Crystalline CXB showed 1 sharp endotherm at 163.7 C, corresponding to its melting point (Fig. 1a (i)). Thermograms of the freeze-dried CXB and CXB-PL formulation (PL:drug 2.5:1) also displayed this melting endotherm (at 162 C-163 C) but with reduced enthalpies. In addition, an exothermic peak at 95 C-110 C was identified (Fig. 1a (ii and iii)). This indicates that freeze-dried CXB is (at least in part) in

4

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

in conjunction with DSC. As shown in Figure 1b (i), the diffraction pattern of crystalline CXB powder revealed several sharp highintensity peaks, with the most distinctive one at diffraction angle 2q of 16.07 (area under the peak ¼ 20,738 counts), which is consistent with previous report.6 On the other hand, the diffractogram of the freeze-dried CXB revealed a noticeable decrease in the number and intensities of peaks when compared with the corresponding crystalline CXB (Fig. 1b (ii)), indicating that CXB was changed from crystalline to partially amorphous form during freeze drying. At the same angle, the area under the peak reduced to 2489 counts; and the calculated percentage of crystallinity of freezedried CXB was 12%. In the presence of PL, as shown in Figure 1b (iii), a halo diffuse appearance with no distinct X-ray scattering was observed in the region 14 -27 2q, suggesting CXB was molecularly dispersed within the PLs in the formulation, thereby creating an amorphous solid dispersion. These findings confirmed the preceding DSC results that the CXB-PL formulations are fully amorphous. Maximum Solubilization of CXB in Liposomes-Micelles

Figure 1. Structure analysis by (a) DSC and (b) PXRD of (i) crystalline CXB, (ii) freezedried CXB, and (iii) PL-CXB solid dispersion (PL:CXB 2.5:1) by freeze drying.

amorphous state because this exothermic peak is related to the recrystallization of the amorphous CXB. Obviously, the amorphous samples first crystallized and then melted above 160 C which corresponds to the melting point of their crystalline forms. In a preliminary study, physical mixtures of CXB and PL (1: 2.5, without freeze drying) did not exhibit this exothermic peak (data not shown). Therefore, the previously mentioned observations provide evidence that freeze drying generates an amorphous solid state of CXB dispersed in PL. Compared with freeze-dried CXB, the PL formulation shows a higher temperature of the exothermic event (110 C vs. 95 C) and a broader peak. The former observation provides a hint that the amorphous state of CXB is stabilized by PL by an interaction between both components, whereas the latter observation could be explained by the dilution effect of CXB in the PL matrix. Furthermore, PL did not show any peaks within the range of the used temperature scan (data not shown); PL thus had minimum impact on the studied peaks of CXB. It is interesting to note that a second exothermic event with small enthalpy change (DH 2.4 J/g) occurred at 123 C for the freeze-dried CXB samples (Fig. 1a (ii)). This may indicate the presence of small amounts of residual moisture in the freeze-dried samples despite the long freeze drying process. In an attempt to reach a definite conclusion regarding the physical state and to calculate the percentage of crystallinity of freeze-dried CXB and PL formulation, PXRD analysis was conducted

Drug Incorporation Capacity of Formulations in PBS Buffer pH 6.5 Freeze-dried formulations with increasing PL-to-drug ratio (from 1:2.5 to 1:250) were dispersed in PBS buffer pH 6.5 to form liposomes. Each liposomal preparation was prepared in triplicates, and the content of CXB remaining in the supernatant after ultracentrifugation (under conditions, where the liposomes stay in the supernatant) was quantified within the same day (i.e., the apparent solubility); the results are given in Table 2. All liposome formulations were found to contain significantly increased amounts of CXB in the supernatant compared with crystalline CXB in PBS buffer (p < 0.05). The overall solubilities of crystalline CXB and freezedried CXB in PBS buffer at pH 6.5 (25 C) were 2.0 ± 0.7 and 13.5 ± 2.2 mg/mL, respectively. When PL was present, considerable amounts of CXB were solubilized by the liposomes, with apparent solubilities increased by 35- to 218-fold (as compared with crystalline drug). Because a minor proportion of liposomes may get trapped in the pellet after ultracentrifugation, the amount of phosphatidylcholine in the supernatant was quantified in comparison to the phosphatidylcholine content before ultracentrifugation to obtain the percentage of PL recovery (Table 2). PL recovery in the supernatant in general was quite high (>85%) and varied only modestly, both within experiments and between the various formulations. Nevertheless, to compensate for variable PL recoveries, each CXB concentration value was adjusted by the corresponding PL recovery value; such that CXB incorporation was normalized by the PL content after subtracting the water-dissolved fraction of the drug. That is to say, the amount of CXB incorporated in liposome was calculated by the following equation:

Incorporated CXB in liposome ðmg=gÞ 
¼ CXB½supernatant  CXB½watersoluble fraction . % PL recovery  100 in which CXB[supernatant] is the concentration of CXB quantified from the supernatant after ultracentrifugation and CXB[water-soluble fraction] is the concentration of crystalline CXB dissolved in buffer (i.e., 2.0 mg/mL). The calculated liposome incorporated CXB values were plotted against the log10 [PL:CXB mass ratios] used in the formulations (Fig. 2, -). A sigmoidal relationship between Log10 PL content and CXB incorporated into the liposomes was obtained, with no further significant increase in CXB-incorporation capacity observed when increasing PL:CXB mass ratio from 50 to 100 (incorporation

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

5

Table 2 Recovery of CXB and PL in Liposome Dispersion and the Incorporated CXB in Liposomes-Mixed Micelles in PBS pH 6.5 of Different Formulations (n ¼ 3) Formulation

PBS pH 6.5 CXB crystalline CXB FD 2.5:1 10:1 50:1 100:1 250:1 a b

PL:CXB (by Mass Ratio)

CXB Concentration in Supernatant, mg/mL (Mean ± SD)

0 0 2.5 10 50 100 250

2.0 13.5 69.8 276.3 435.2 377.2 393.5

± ± ± ± ± ± ±

0.7 2.2 11.5b 51.5b 8.1b 44.4b 13.4b

% PL Recovery (Mean ± SD)

Incorporated CXB in Liposomes/Mixed Micelles, mg/g (Mean ± SD)a

e e 102.8 94.4 85.1 92.6 95.8

e e 66.2 263.1 399.2 381.9 372.8

± ± ± ± ±

0.6 15.9 16.7 0.4 6.2

± ± ± ± ±

11.2 43.2 63.5 34.0 9.4

Calculated by the equation of (CXB[supernatant] e CXB[water-soluble fraction])/% PL recovery. p < .05 compared to crystalline CXB (analyzed by 1-way ANOVA with Tukey post hoc test).

capacity of 399.2 ± 63.5 vs. 381.9 ± 34.0 mg/g, p > 0.05). Thus, the maximum incorporation of CXB occurred at PL-to-drug ratio of 50 (399.2 ± 63.5 mg/g), at which the apparent solubility of CXB increased by 218 times as compared with that of crystalline CXB. Drug Solubilization Capacity of Formulations in PBS Buffer Containing Sodium Taurocholate In a subsequent step, the drug solubilization capacity of CXB in liposomes or mixed micelles was evaluated in a buffer medium containing 3 mM of sodium taurocholate at pH 6.5. The apparent solubilities of both, crystalline CXB and freeze-dried CXB, significantly increased to 13.2 ± 2.3 and 35.3 ± 8.3 mg/mL, respectively (p < 0.05, as compared with their corresponding solubilities in PBS buffer). Liposomal CXB dispersed in sodium taurocholate medium further increased the apparent solubilities of CXB. Except for the formulation containing PL:CXB mass ratio of 2.5:1, all other studied formulations (PL:CXB mass ratios of 10-250) significantly increased the apparent solubilities of CXB by 25- to 32-fold compared with that of crystalline CXB in the same dispersion medium (p < 0.05, Table 3). Similar to the previously mentioned description, CXB solubilized in liposome-micelles (microgram/gram) was calculated by the equation of (CXB[supernatant] e CXB[media-soluble fraction])/%PL recovery  100, in which CXB[supernatant] is the concentration of CXB

Figure 2. CXB-solubilization capacity in liposomes/mixed micelles as a function of log10 [PL:drug mass ratio] of CXB in the formulation (n ¼ 3). ¶Significantly increased compared to corresponding PL:CXB ratio tested with PBS pH 6.5 (p < 0.05).

quantified from the supernatant after ultracentrifugation and CXB[media-soluble fraction] is the concentration of crystalline CXB dissolved in buffer containing sodium taurocholate (i.e., 13.2 mg/mL). The percentage of PL recoveries (which were again >80%) and the calculated amount of CXB solubilized in liposomes-micelles are presented in Table 3. As shown in Figure 2 (open circle), a sigmoidal relationship between Log10 PL content and solubilized CXB was obtained, with no further significant increase was observed when increasing PL:CXB mass ratio from 10 to 50 (solubilization capacity of 344.5 ± 27.2 mg/g vs. 401.2 ± 54.1 mg/g, p > 0.05), that is, the optimum PL-to-drug ratio for maximum incorporation of CXB was 10:1. When comparing the solubilization capacities of CXB between the 2 studied dispersion media, a similar trend was observed, with statistical significance only detected at isolated PL-to-drug ratios: at PL-to-drug ratios of 10:1 and 250:1, the solubilization capacities of CXB were higher in the presence of sodium taurocholate (p < 0.05; Fig. 2 and Table 3). Otherwise, the solubilization capacities of CXB in the presence or absence of taurocholate were (more or less) comparable. Permeation of Formulations Permeation Studies in PBS Buffer pH 6.5 The cumulative amounts of CXB transported across the dialysis membrane from the studied formulations over 24 h are presented in Figure 3a. In all cases, the cumulated amount in the acceptor phase initially increased linearly over time, but cumulative flux curves flattened out at t ¼ 24 h compared with the initial 6 h of experiment, indicating the concentrations of the free drugs decreased over time and redistribution occurred. With crystalline CXB serving as a reference, the liposomal CXB at PL-to-drug ratios of 2.5:1 and 10:1 exhibited relatively higher permeations. However, a further increase in the PL content in the formulation (50:1 and 250:1) resulted in lower flux values of CXB. Steady state flux (Jss) was calculated based on the linear range of the cumulative transport versus time curve. Typically such linear range lied between 1 and 4 h, so that a total of 7 points were used for the calculation of Jss. The Jss values, along with the cumulative amount of CXB transported at the end of the experiment (i.e., t ¼ 24 h), are presented in Figure 4a. The Jss of crystalline CXB at its solubility-limit concentration across the dialysis membrane was 5.94 ± 0.76  105 mg/cm2/s. Taking this value as a reference, a relatively higher Jss was only observed for the formulation with PL-to-drug ratio of 10:1 (Jss of 11.9 ± 0.75  105 mg/cm2/s). On the other hand, a further increase in PL-to-drug ratio resulted in relatively lower Jss (p < 0.05, PL:drug 250:1, with Jss of 1.17 ± 0.02  10-5 mg/cm2/s). In summary, the formulations of CXB transported across dialysis membrane in descending order were 10:1 > 2.5:1> 50:1 > 250:1.

6

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

Table 3 Recovery of CXB and PL in Liposome Dispersion and the Incorporated CXB in Liposomes-Mixed Micelles in PBS-Taurocholate pH 6.5 of Different Formulations (n ¼ 3) Formulation

PBS/taurocholate pH 6.5 CXB crystalline CXB FD 2.5:1 10:1 50:1 100:1 250:1 a b c

PL: CXB (by Mass Ratio)

CXB Concentration in Supernatant, mg/mL (Mean ± SD)

0 0 2.5 10 50 100 250

13.2 35.3 55.3 335.6 428.6 388.6 415.4

± ± ± ± ± ± ±

2.3b 8.3b 4.1 74.7c 4.5c 103.4c 7.3c

% PL Recovery (Mean ± SD)

Incorporated CXB in Liposomes/Mixed Micelles, mg/g (Mean ± SD)a

e e 81.5 81.3 104.9 88.6 92.9

e e 53.1 344.5 401.2 420.0 433.5

± ± ± ± ±

18.0 0.2 15.6 16.5 6.5

± ± ± ± ±

5.1 27.2b 54.1 38.4 19.4b

Calculated by the equation of (CXB [supernatant] e CXB [water-soluble fraction])/% PL recovery. p < .05 compared to the corresponding formulation tested with PBS pH 6.5 (analyzed by Student t-test). p < .05 compared to crystalline CXB (analyzed by 1-way ANOVA with Tukey post hoc test).

Permeation Studies in PBS/Taurocholate Buffer pH 6.5 In a subsequent step, the effect of various proliposome formulations on the permeability of CXB across dialysis membrane was evaluated in a buffered medium containing 3 mM of sodium

taurocholate at pH 6.5. Similar to the trend observed in buffer pH 6.5, formulations with lower PL-to-drug ratios (2.5:1 and 10:1) demonstrated a relatively higher cumulative transport of CXB as compared to crystalline CXB, whereas those with higher PL-to-drug

Figure 3. Cumulative amount of CXB transported across dialysis membrane from different formulations over time in (a) PBS buffer (pH 6.5) and (b) PBS buffer containing sodium taurocholate (pH 6.5). Each data point represents mean ± SD of 3 samples.

Figure 4. Cumulative amount of CXB transported after 24 h and the Jss across dialysis membrane from different formulations in (a) PBS buffer (pH 6.5) and (b) PBS buffer containing 3 mM of sodium taurocholate (pH 6.5; n ¼ 3). *Significantly increased/ decreased compared to crystalline CXB (p < 0.05). ¶Significantly increased/decreased compared to corresponding PL:CXB ratio tested with PBS pH 6.5 (p < 0.05).

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

ratios (50:1) reduced the cumulative transport of CXB (Fig. 3b). Again, all cumulative flux curves flattened out at t ¼ 24 compared with the initial 6 h of experiment. The Jss value of crystalline CXB in PBS buffer containing sodium taurocholate, which served as the reference value for comparison among the PL formulations on permeability performances, was found to be 5.03 ± 0.12  105 mg/cm2/s. As shown in Figure 4b, formulation with PL-to-drug ratio of 2.5:1 had a relatively higher Jss and cumulative amount after 24 h (p < 0.05 compared with crystalline CXB, increased by ~2.5-fold). On the other hand, a further increase in PL-to-drug ratios did not increase (10:1 and 50:1) or even had a significantly lower Jss (p < 0.05, PL:drug 250:1, with Jss of 1.20 ± 0.12  105 mg/cm2/s). In descending order, the permeability of CXB in the presence of sodium taurocholate was 2.5:1> 10:1 > 50:1 > 250:1. The fluxes of CXB in different formulations were compared between the 2 studied media. The Jss of crystalline CXB across the dialysis membrane was virtually the same in PBS buffer (5.94 ± 0.76  105 mg/cm2/s) and in PBS buffer containing sodium taurocholate (5.03 ± 0.12  105 mg/cm2/s). Furthermore, the general trend of enhancement/reduction in permeability performance of CXB was similar in both testing mediadthe lower PL-to-drug ratio had an enhancement effect, whereas the higher PL-to-drug ratio had the opposite effect on CXB permeability. Although the general

7

rank between the 2 testing media is in agreement, significant difference was noted at the formulations with PL-to-drug ratios of 2.5:1 and 10:1 (Fig. 4b, p < 0.05). In the presence of sodium taurocholate, the Jss of CXB was significantly higher at 2.5:1 (14.6 ± 0.28  105 vs. 6.35 ± 0.49  104 mg/cm2/s compared with that in pure PBS buffer), whereas it was significantly lower at 10:1 (7.28 ± 1.85 vs. 11.9 ± 0.75  105 mg/cm2/s).

Correlation Between Apparent Solubility and Permeability The overall apparent solubility versus permeability as a function of PL-to-drug ratio in PBS buffer (pH 6.5) and PBS buffer containing 3 mM of sodium taurocholate are presented in Figures 5a and 5c, respectively. In an attempt to investigate the relationship between the apparent solubility and the permeability performance of the PL formulations, the Jss values from various studied PL-to-drug ratios were plotted versus the corresponding apparent solubility. As shown in Figure 5b, there is no correlation between the 2 parameters in the dispersion medium of PBS buffer pH 6.5 (R ¼ 0.449, p > 0.05). On the other hand, a negatively linear correlation was found between the 2 parameters in the dispersion medium containing 3 mM of sodium taurocholate (Fig. 5d). With a Pearson

Figure 5. Relationship between apparent solubility and permeability of CXB as a function of PL:CXB ratio (aþc) and direct correlation plot (bþd) in 2 dispersion media: (aþb) PBS buffer (pH 6.5) and (cþd) PBS buffer containing 3 mM of sodium taurocholate (pH 6.5; n ¼ 3).

8

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

coefficient of 0.965, such correlation was statistically significant (p < 0.05). Discussion CXB solid dispersions that consisted of drug and PL were prepared by freeze drying. CXB alone was shown to transform (largely) from its crystalline form to an amorphous state by the freeze drying process. In presence of PL, the amorphous state of CXB is further stabilized by an interaction between both components, achieving an amorphous solid dispersion. On redispersion in PBS or PBS containing bile salt, the apparent solubility, drug solubilization, and the permeability performance of the formulation have been investigated in the present study, and it was found that all these parameters highly depend on the PL-to-drug ratios as well as the dispersion medium. Using the method described by Sætern et al.,23 the apparent solubility enhancement and the maximum drug incorporation capacity of CXB by the PL formulation were identified. This method has been validated for complete separation of free and liposomal drug and thus is considered suitable method for determining the drug incorporation capacity in liposomes. Freeze drying of CXB alone induced a moderate (about 7-fold) increase in solubility in PBS buffer. This increase in solubility of CXB is assumed to be due to the amorphous state of the drug on freeze drying, which induces supersaturation. Meanwhile, across all the studied PL-to-drug ratios, the apparent solubilities of CXB in PBS buffer were significantly increased by 35- to >200-fold as compared with that of crystalline CXB, indicating that considerable amounts of CXB have been solubilized by liposomes. In addition to the solubilizing effect by the liposomes, the amorphous state of CXB in the formulation is expected also to play a role in enhancing the apparent solubility of CXB. These results support our hypothesis that the use of an amorphous solid dispersion could enhance the solubility of poorly water-soluble drugs. Nevertheless, the solubility enhancement reached a threshold limit in which by increasing the PL amount, the apparent solubility of CXB was not increased any further. This threshold corresponded to the maximum drug incorporation capacity of CXB in liposomes, and it appeared to be close to 400 mg/g, which was in the same order of magnitude as that reported in the literature for incorporation of camptothecin, a lipophilic anticancer drug,23 and 3-hydroxy-quinolinone derivatives,26 using the same buffer medium and experimental approach. The optimum PL-to-drug ratio reported here is also in the same order of magnitude of ratios used in other studies preparing CXB-loaded liposomes.10,11 Having proved the multiple-fold enhancement in CXB apparent solubility by the PL formulation, the next question was whether the permeation properties of CXB were enhanced accordingly. Because effective oral drug absorption depends not only on solubility but also on permeability, both parameters are necessary to be evaluated to judge whether the formulation is really “candidate enabling.” In the present study, a Franz diffusion cells with dialysis cellulose membranes (MWCO 12-14 kDa) was used for the permeability study. A similar approach investigating the relationship between lipid composition and passive permeability of compounds from liposome formulations has been reported.27 The molecular weight cutoff of the membrane chosen only allowed the passage of the “truly” dissolved fraction of CXB but not the passage of supramolecular assemblies, that is, liposomal- or micellular-bound drug; the membrane pore diameter was equivalent to 2.4 nm,28 which is well below the expected diameters of liposomes (20 nm)29 and PL/taurocholate mixed micelles (4.4 nm).30 The cumulative amount in the acceptor chamber over time and the Jss of CXB from formulations with various PL-to-drug ratios were investigated.

Jss values instead of the commonly presented apparent permeability values were calculated because the calculation of the former is independent of the initial concentration of “dissolved” drug. In cases where the drug partly is molecularly dissolved and partly in solubilized state, for instance the current case of CXB in several studied formulations, apparent permeability coefficients could be misleading because it is not clear what is the relevant donor concentration of “dissolved” drug.31 When taking the permeability values of crystalline CXB as a reference, only the formulation with PL-to-drug ratio of 10:1 demonstrated a higher permeation, which indicates that this is the “optimum” formulation. It should be noted that if a higher loading concentration of crystalline CXB was used, the Jss of CXB (reference) would be theoretically higher; and it follows that all PL formulations (including or excluding the 10:1 formulation) would have lower permeabilities compared with the unformulated drug, despite the demonstrated increase in apparent solubility. Furthermore, there is an inverse relationship between the PL amount and the permeability of the PL formulationdthe formulations with lower PL-to-drug ratios had a positive effect, whereas those with higher PL-to-drug ratios had the opposite effect on CXB permeability across the membrane. Similar observation has been reported by Varshosaz et al.32 in their permeation studies of metoprolol-lecithin nanoemulsion organogel across cellulose membranes. This observation has generally and qualitatively been attributed to the change in thermodynamic activity and the decrease in free fraction of the drug available for membrane permeation with the increasing amount of solubilizing agents.33-35 More recently, Dahan and Miller36 developed a mathematical mass transport model to explain this opposing effect of solubilitypermeability interplay by taking the unstirred water layer (change in effective thickness) and the free drug fraction availability into consideration. Although both explanations are in no doubt relevant, in our opinion, they may be oversimplified because the different physical states of drug and the kinetics that occur in the gastrointestinal tract may be overlooked.21 In the present case, for example, CXB was hypothesized to be present in multiple forms in the donor compartments: in solid state as suspension (amorphous or crystalline form) and in various “dissolved” states (liposome-bound solubilized state and molecularly dissolved state, in which the latter eventually being metastable supersaturated state), and each state is interlinked via different kinetic processes (Fig. 6a). From our experimental setup, only the CXB in molecularly dissolved state is expected to diffuse across the dialysis membrane (the “absorbed” fraction). We hypothesize that the following mechanisms may explain the effect of the PL formulation on the permeability of CXB (Fig. 6a): 1. Solid CXB is (at least partly) in amorphous form on freeze drying, which induces (transient) supersaturation, that is, an increased concentration of molecularly dissolved drug. 2. CXB is solubilized by liposomes, and thereby, the apparently dissolved fraction of CXB is increased. 3. The observed permeability enhancement of CXB at low PL:CXB mass ratios is due to supersaturation. 4. For higher PL:CXB mass ratios, this supersaturation-driven enhanced permeability effect is overcompensated by increasing partitioning of CXB to liposomes; here, the release from supermolecular assemblies is comparably slow and permeation rate limiting. For the first mechanism, supersaturation has been proven using an equilibrium dialysis method we previously described.37 Briefly, a dialysis kit with 6-8 kDa cutoff (Mini Pur-A-Lyzer) was filled with PBS buffer pH 6.5 and placed in a glass vial containing the dispersed formulation (in excess) and different media (including PBS buffer

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

9

Figure 6. Schematic overview of different states of CXB on dispersion of solid PL dispersions in aqueous media: (a) PBS buffer (pH 6.5) and (b) PBS buffer containing 3-mM sodium taurocholate (pH 6.5) that may occur in the gastrointestinal tract and the kinetic processes that can influence its permeability across barrier.

10

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

pH 6.5 and fasted state stimulated intestinal fluid, respectively). Under equilibrium conditions at 25 C, the molecularly dissolved concentration of CXB was found to increase by 3- to 11-fold at the PL-to-drug ratio of 10:1 compared with that of crystalline CXB,38 which suggested true supersaturation being achieved. The third and fourth proposed mechanisms would explain the inverse relationship between the amount of PLs and the permeability performance of CXB. The kinetics between the release-replenish of CXB from/into liposome, which depends on the affinity of CXB to PL, is hypothesized to play a major role in affecting the amount of molecularly dissolved CXB available for permeation. At lower PL-todrug ratio, in which the drug incorporation capacity is limited, supersaturation is prevalent, and the release of molecularly dissolved CXB is thought to be fast. On the other hand, higher PL-todrug ratios tend to favor the incorporation of CXB in liposomes and resulting in a slower release rate of molecularly dissolved CXB. It has previously been suggested that physiological lipids found in the gastrointestinal tract may function as precipitation inhibitors to maintain high levels of supersaturation.39 It is thus possible that in this case, the presence of PL inhibits recrystallization and reduces the precipitation of CXB. To test this hypothesis, we have conducted a kinetic analysis of crystal nucleation in PBS buffer to evaluate the potential of PLs in inhibiting CXB crystallization. To our surprise, the induction time for nucleation (i.e., the time lag for the observable crystal to appear measured by light blockage of UV detection) was not significantly different between the crystalline CXB and the formulation at PL:CXB 100:1 (data not shown). Therefore, the proposed mechanism of PLs inhibiting the recrystallization of amorphous CXB should be ruled out. Given that permeation of drugs from enabling formulations depends on various physical states of the drug, the complex interplay between different states as illustrated in the case of CXB should be considered. Although it is known that the apparent solubility of poorly soluble drugs is significantly influenced by the presence of endogenous surfactants such as bile salts,40 their influence on absorption is controversially discussed. A recent study by Vogtherr et al.41 highlighted the key role of taurocholate in forming taurocholate-drug micelles that affects drug solubilization in biomimetic media. Therefore, the addition of sodium taurocholate to buffer provided greater biorelevance for the understanding of the solubility and permeability enhancement caused by PL formulations and thereby the relationship between the 2 parameters. From our present study it appears that the presence of bile salts in the medium significantly enhances the apparent solubility of CXB, both if the drug is in crystalline or amorphous (freeze dried) state. This is attributed to the association of CXB with taurocholate micelles. In contrast, in case of solid PL dispersions, taurocholate does not significantly affect the amount of solubilized drug as compared with the liposomal dispersion in PBS, indicating that the supramolecular assemblies formed in the presence of taurocholate such as mixed micelles have similar solubilization capacities as the liposomes formed in buffer. With the addition of sodium taurocholate, the proposed schematic overview over different states of CXB during its permeability across intestinal barrier is illustrated in Figure 6b. Being a surfactant, sodium taurocholate dispersed in the aqueous buffer medium is assumed to form taurocholate micelles and interact with PLs to form taurocholate-PL mixed micelles. From our simplest case of crystalline CXB (in which no PL was present), its significant increase in apparent solubility did not induce a subsequent increase in flux after the addition of sodium taurocholate (PBS buffer vs. PBS with taurocholate). This result is in agreement with our previous work that found the flux of a poorly water-soluble drug ABT-102 across the Caco-2 barrier in buffer not to be significantly different compared with that in fasted stated

stimulated intestinal fluid (in which sodium taurocholate is the major component).37 Interestingly, significantly higher flux was observed for the 2.5:1 formulation when taurocholate was added as compared with the dispersion in buffer. These observations suggest that the presence of taurocholate in solid PL dispersions with a low PL content exerts a beneficial impact on the permeation performance, the exact mechanism of which remains to be elucidated. Because taurocholate is known to interact with PL up to a certain taurocholate-/PL-ratio to form taurocholate-PL mixed micelles, the observed permeability enhancement of CXB may be a consequence of CXB interaction with taurocholate-PL mixed micelles. Recently, it has been reported that drug mobility within taurocholate-PL mixed micelles is significantly different from that within pure taurocholate micelles.41 With the increasing PL concentrations in the medium, it may occur that mixed micelles with a different composition are formed or that liposomes are formed besides mixed micelles. Obviously, it would be interesting to try to reveal in future investigations, whether there coexist different supramolecular assemblies in such setting. Furthermore, the relationship between apparent solubility and permeability of CXB also depends on the presence of taurocholate. Although there is no significant correlation between the 2 parameters in PBS buffer, a significant and negatively correlated relationship is found between the apparent solubility and permeability of CXB in buffer containing taurocholate. Using a poorly water-soluble model drug hydrocortisone and in a dispersion medium of neutral PBS, di Cagno and Luppi42 also demonstrated that there is no direct association between apparent solubility and permeability. On the other hand, to our knowledge, the present study is the first experiment that investigated and demonstrated an inverse relationship between apparent solubility and permeability of liposomal formulation in the presence of bile salt. This may have implication to the screening approach for the “optimum” liposomal formulation. Although most of the existing studies select the “optimum” candidate-enabling formulations based on solubility performance (enhancement), our findings imply that this may not be the best approach because the extent of solubility enhancement achieved by the formulation may have the opposite effect on permeability. To provide a more generalized conclusion, more model drugs with poor water solubility and other biomimetic media such as fasted/fed state stimulated intestinal fluid will need to be tested in future studies. However, the present example of CXB-PL formulation highlights the importance of evaluating both apparent solubility and permeability and also using biomimetic medium for testing the candidate-enabling performance of liposomal formulations. Conclusion CXB amorphous solid PL dispersions were developed with various PL-to-drug ratios. On redispersion in PBS buffer or PBS buffer containing taurocholate, the drug solubilization capacity, apparent solubility, and the permeability performance of the formulations have been investigated. No clear correlation was found between the enhancement of CXB solubility and permeability; both parameters were highly dependent on the dispersion media as well as the PL-to-drug ratios. This study highlights the importance of evaluating both apparent solubility and permeability and the use of biomimetic medium for testing the candidate-enabling performance of liposomal formulations. Acknowledgments This work was supported by Phospholipid Research Centre, Heidelberg, Germany (research grant to S.Y.K.F.).

S.Y.K. Fong et al. / Journal of Pharmaceutical Sciences xxx (2016) 1e11

References 1. Lipinski CA, Lombardo F, Dominy BW, Feeney PJ. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev. 2012;64(Supplement):4-17. 2. Gribbon P, Andreas S. High-throughput drug discovery: what can we expect from HTS? Drug Discov Today. 2005;10:17-22. 3. Paulson SK, Vaughn MB, Jessen SM, et al. Pharmacokinetics of celecoxib after oral administration in dogs and humans: effect of food and site of absorption. J Pharmacol Exp Ther. 2001;297:638-645. 4. Fouad EA, EL-Badry M, Mahrous GM, Alanazi FK, Neau SH, Alsarra IA. The use of spray-drying to enhance celecoxib solubility. Drug Dev Ind Pharm. 2011;37: 1463-1472. 5. Lee H, Lee J. Dissolution enhancement of celecoxib via polymer-induced crystallization. J Cryst Growth. 2013;374:37-42. 6. Gupta VR, Mutalik S, Patel MM, Jani GK. Spherical crystals of celecoxib to improve solubility, dissolution rate and micromeritic properties. Acta Pharma (Zagreb). 2007;57:173-184. 7. Morgen M, Bloom C, Beyerinck R, et al. Polymeric nanoparticles for increased oral bioavailability and rapid absorption uing celecoxib as a model of a lowsolubility, high-permeability drug. Pharm Res. 2012;29:427-440. 8. Abu-Diak OA, Jones DS, Andrews GP. An investigation into the dissolution properties of celecoxib melt extrudates: understanding the role of polymer type and concentration in stabilizing supersaturated drug concentrations. Mol Pharm. 2011;8:1362-1371. 9. Reddy MN, Rehana T, Ramakrishna S, Chowdary KPR, Diwan PV. b-cyclodextrin complexes of celecoxib: molecular-modeling, characterization, and dissolution studies. AAPS PharmSci. 2004;6:68-76. 10. Deniz A, Sade A, Severcan F, Keskin D, Tezcaner A, Banerjee S. Celecoxib-loaded liposomes: effect of cholesterol on encapsulation and in vitro release characteristics. Biosci Rep. 2010;30:365-373. 11. Begum MY, Abbuku K, Sudhakar M. Celecoxib loaded liposomes: development, characterization and in vitro evaluation. Int J Pharm Sci Res. 2012;3: 154-161. 12. Song W, Yeom D, Lee D, et al. In situ intestinal permeability and in vivo oral bioavailability of celecoxib in supersaturating self-emulsifying drug delivery system. Arch Pharm Res. 2014;37:626-635. 13. Dahan A, Miller JM, Hoffman A, Amidon GE, Amidon GL. The solubilityepermeability interplay in using cyclodextrins as pharmaceutical solubilizers: Mechanistic modeling and application to progesterone. J Pharm Sci. 2010;99:2739-2749. 14. Fischer SM, Brandl M, Fricker G. Effect of the non-ionic surfactant Poloxamer 188 on passive permeability of poorly soluble drugs across Caco-2 cell monolayers. Eur J Pharm Biopharm. 2011;79:416-422. 15. Fischer SM, Flaten GE, Hagesæther E, Fricker G, Brandl M. In-vitro permeability of poorly water soluble drugs in the phospholipid vesicle-based permeation assay: the influence of nonionic surfactants. J Pharm Pharmacol. 2011;63:10221030. 16. Frank KJ, Westedt U, Rosenblatt KM, et al. What is the mechanism behind increased permeation rate of a poorly soluble drug from aqueous dispersions of an amorphous solid dispersion? J Pharm Sci. 2014;103:1779-1786. 17. Frank KJ, Westedt U, Rosenblatt KM, et al. The amorphous solid dispersion of the poorly soluble ABT-102 forms nano/microparticulate structures in aqueous medium: impact on solubility. Int J Nanomedicine. 2012;7:5757-5768. 18. Frank KJ, Rosenblatt KM, Westedt U, et al. Amorphous solid dispersion enhances permeation of poorly soluble ABT-102: true supersaturation vs. apparent solubility enhancement. Int J Pharm. 2012;437:288-293. 19. Basavaraj S, Betageri GV. Can formulation and drug delivery reduce attrition during drug discovery and development-review of feasibility, benefits and challenges. Acta Pharm Sin B. 2014;4:3-17. 20. Fong SYK, Brandl M, Bauer-Brandl A. Phospholipid-based solid drug formulations for oral bioavailbility enhancement: a meta-analysis. Eur J Pharm Sci. 2015;80:89-110.

11

21. Buckley ST, Frank KJ, Fricker G, Brandl M. Biopharmaceutical classification of poorly soluble drugs with respect to “enabling formulations”. Eur J Pharm Sci. 2013;50:8-16. 22. de Villiers MM, Wurster DE, Van der Watt JG, Ketkar A. X-Ray powder diffraction determination of the relative amount of crystalline acetaminophen in solid dispersions with polyvinylpyrrolidone. Int J Pharm. 1998;163:219-224. 23. Sætern AM, Flaten GE, Brandl M. A method to determine the incorporation capacity of camptothecin in liposomes. AAPS PharmSciTech. 2004;5:30-37. 24. Grohganz H, Ziroli V, Massing U, Brandl M. Quantification of various phosphatidylcholines in liposomes by enzymatic assay. AAPS PharmSciTech. 2003;4: 500-505. 25. Brandl M, Eide Flaten G, Bauer-Brandl A, Begley TP. Passive diffusion across membranes. In: Begley TP, ed. Wiley Encyclopedia of Chemical Biology. Hoboken, NJ: John Wiley & Sons, Inc; 2007.  26. Di Cagno M, Styskala J, Hlava c J, Brandl M, Bauer-Brandl A, Skalko-Basnet N. Liposomal solubilization of new 3-hydroxy-quinolinone derivatives with promising anticancer activity: a screening method to identify maximum incorporation capacity. J Liposome Res. 2011;21:272-278. 27. Fo co A, Gasperlin M, Kristl J. Investigation of liposomes as carriers of sodium ascorbyl phosphate for cutaneous photoprotection. Int J Pharm. 2005;291:21-29. 28. Venkateswarlu V, Manjunath K. Preparation, characterization and in vitro release kinetics of clozapine solid lipid nanoparticles. J Control Release. 2004;95:627-638. 29. Watwe RM, Bellare JR. Manufacture of liposomesda review. Current Sci. 1995;68:715-724. 30. Nichols JW, Ozarowski J. Sizing of lecithin-bile salt mixed micelles by sizeexclusion high-performance liquid chromatography. Biochemistry. 1990;29: 4600-4606. 31. Kanzer J, Tho I, Flaten GE, et al. In-vitro permeability screening of melt extrudate formulations containing poorly water-soluble drug compounds using the phospholipid vesicle-based barrier. J Pharm Pharmcol. 2010;62:1591-1598. 32. Varshosaz J, Andalib S, Tabbakhian M, Ebrahimzadeh N. Development of lecithin nanoemulsion based organogels for permeation enhancement of metoprolol through rat skin. J Nanomaterials. 2013;2013:1-10. 33. Nerurkar MM, Ho NFH, Burton PS, Vidmar TJ, Borchardt RT. Mechanistic roles of neutral surfactants on concurrent polarized and passive membrane transport of a model peptide in Caco-2 cells. J Pharm Sci. 1997;86:813-821. 34. Chiu YY, Higaki K, Neudeck BL, Barnett JL, Welage LS, Amidon GL. Human jejunal permeability of cyclosporin A: influence of surfactants on P-glycoprotein efflux in Caco-2 cells. Pharm Res. 2003;20:749-756. 35. Katneni K, Charman SA, Porter CJH. Permeability assessment of poorly watersoluble compounds under solubilizing conditions: the reciprocal permeability approach. J Pharm Sci. 2006;95:2170-2185. 36. Dahan A, Miller J. The solubilityepermeability interplay and its implications in formulation design and development for poorly soluble drugs. AAPS J. 2012;14: 244-251. 37. Frank KJ, Westedt U, Rosenblatt KM, et al. Impact of FaSSIF on the solubility and dissolution-/permeation rate of a poorly water-soluble compound. Eur J Pharm Sci. 2012;47:16-20. 38. Fong SYK, Ibisogly A, Bauer-Brandl A. Solubility enhancement of BCS Class II drug by solid phospholipid dispersions: spray drying versus freeze-drying. Int J Pharm. 2015. http://dx.doi.org/10.1016/j.ijpharm.2015.10.029. 39. Newman A, Knipp G, Zografi G. Assessing the performance of amorphous solid dispersions. Journal of pharmaceutical sciences. 2012;101:1355-1377. 40. Kleberg K, Jacobsen J, Müllertz A. Characterising the behaviour of poorly water soluble drugs in the intestine: application of biorelevant media for solubility, dissolution and transport studies. J Pharm Pharmacol. 2010;62:1656-1668. 41. Vogtherr M, Marx A, Mieden AC, Saal C. Investigation of solubilising effects of bile salts on an active pharmaceutical ingredient with unusual pH dependent solubility by NMR spectroscopy. Eur J Pharm Biopharm. 2015;92:32-41. 42. di Cagno M, Luppi B. Drug “supersaturation” states induced by polymeric micelles and liposomes: a mechanistic investigation into permeability enhancements. Eur J Pharm Sci. 2013;48:775-780.