Characterization of Prototype Self-Nanoemulsifying Formulations of Lipophilic Compounds

Characterization of Prototype Self-Nanoemulsifying Formulations of Lipophilic Compounds FLEMMING S. NIELSEN,1 EMILIE GIBAULT,2,3 HELENA LJUSBERG-WAHREN,2,3 LISE ARLETH,4 ¨ LLERTZ1 JAN SKOV PEDERSEN,5 ANETTE MU 1

Department of Pharmaceutics, The Danish University of Pharmaceutical Sciences, Universitetsparken 2, DK-2100 Copenhagen O, Denmark 2

Camurus AB, Ideon, Gamma 2, So¨lveg. 41, SE-223 70 Lund, Sweden

3

Division of Food Technology, Lund University, P.O. Box 124, SE 221 00 Lund, Sweden

4

Department of Natural Sciences, The Royal Veterinary and Agricultural University, Thorvaldsensvej 40, DK-1871 Frederiksberg, Denmark

5

Department of Chemistry, University of Aarhus, Langelandsgade 140, DK-8000 Aarhus C, Denmark

Received 4 June 2005; revised 27 March 2006; accepted 16 April 2006 Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.20673

ABSTRACT: This study describes the evaluation and characterization of a selfnanoemulsifying drug delivery system (SNEDDS) consisting of a nonionic surfactant (Cremophor RH40), a mixture of long chain mono-, di-, and triacylglycerides (Maisine 35-1 and Sesame oil) and ethanol. Compositions containing 10% (w/w) ethanol, 40%–60% (w/w) lipid content, and 30%–50% (w/w) Cremophor RH40 were identified as pharmaceutically relevant, robust, and self-nanoemulsifying when dispersed in aqueous media. The influence of adding three different lipophilic model drug compounds (danazol, halofantrine, and probucol) to the SNEDDS was evaluated. While danazol precipitated from the SNEDDS after dispersion in aqueous media, halofantrine and procubol remained solubilized. Halofantrine- and procubol-loaded SNEDDS were evaluated in both saline and in media simulating fasted and fed-state intestinal fluid (FaSSIF and FeSSIF) using dynamic light scattering and small-angle X-ray scattering (SAXS) techniques. Stable nanoemulsions with droplet sizes in the range of 20–50 nm were formed in all media and with and without drugs. The mean size of the droplets was neither affected significantly by being dispersed into the media simulating gastro intestinal fluid, nor by addition of the drug. ß 2007 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 96:876–892, 2007

Keywords: emulsion/microemulsion; formulation vehicle; surfactants; light scattering (dynamic); self-emulsifying drug delivery system (SEDDS); small-angle X-ray scattering (SAXS)

INTRODUCTION The high-throughput screening approach in drug discovery within the pharmaceutical industry has Correspondence to: A. Mu¨llertz (Telephone: 45 35306440; Fax: 45 35306030; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 96, 876–892 (2007) ß 2007 Wiley-Liss, Inc. and the American Pharmacists Association

876

lead to drug candidates with increasing lipophilicity.1 A typical characteristic for these compounds is low and variable oral bioavailability from solid dosage forms due to their poor water solubility. One increasingly popular approach to overcome this problem, is the use of a selfemulsifying drug delivery system (SEDDS).2–4 The bioavailability enhancing properties of SEDDS compared to solid dosage forms has

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

primarily been attributed to the ability of the vehicles to keep the compound in solution in the gastro intestinal (GI) tract and thereby maintaining a maximal free drug concentration and omitting a rate determining dissolution step.5,6 However, the parameters differentiating the absorption from different SEDDS compositions is not well understood. The most common parameters put forward in order to explain the difference in absorption are rate of dispersion, particle size of the resultant dispersion,7 the rate of digestion for formulations susceptible to digestion8 and possibly also the solubilization capacity of the digested formulation.9 A prerequisite for the use of SEDDS, for oral administration, is that the dose of the compound is soluble in the SEDDS preconcentrate and stay solubilized in the vehicle after dispersion. Solubility in the preconcentrate is the limiting factor for a number of poorly water-soluble compounds, which are also poorly soluble in lipids. Therefore SEDDS, as a rule of thumb, is primarily considered relevant for oral delivery of potent compounds (low dose) and compounds with a log p-value above 4.10 However, a better indication for the applicability of SEDDS is a good solubility of the compound in surfactant and lipid excipients. This is exemplified by cyclosporine, which has low log p value of 3.011 but high-oil solubility. SEDDS can be formulated using a combination of surface-active excipients, lipids and polar cosolvents,4 though SEDDS has also been formulated without the use of a polar cosolvent.10 The typical polar cosolvent is ethanol which is well known, recognized as a safe pharmaceutical excipient and also known to facilitate the selfdispersion of SEDDS.12 Some attempts have been made to categorize SEDDS and predict the behavior of the systems based on type and content of surfactant, lipid phase and cosolvent.10 So far the strategy for the evaluation and characterization of SEDDS has primarily been based on applying basic concepts from equilibrium phase behavior studies of systems mixed with water5 and evaluation of the selfemulsification with respect to rate of emulsification and the particle size and distribution of the resultant emulsion.13 As an example, Khoo et al.14 described a visual grading system where the emulsification rate and resultant emulsion are qualitatively characterized. Emulsification rate and particle size have also been assessed by measurement of turbidity as a function of time.15 The increase in turbidity was used to monitor DOI 10.1002/jps

877

emulsification rate and the final turbidity were correlated to mean particle size.13,16 Particle size measurements of self-emulsifying systems are often performed after dispersion in water or other simple aqueous media.17–19 However, in the GI tract the formulation will encounter a more complex environment containing endogenous surfactants as bile salt (BS) and phospholipid (PL). This may impact the particle size of the resulting SEDDS aggregates. The purpose of the present work was to evaluate and characterize a system known to produce selfnanoemulsifying drug delivery system (SNEDDS) with special emphasis on: (1) the solubility in SNEDDS and solubilization capacity after dispersion; (2) the influence of selected model drug compounds on dispersion properties and particle size of the identified SNEDDS; and (3) investigate whether media simulating the BS/PL rich environment in the gastrointestinal tract would have any effect on the particle size of the identified SNEDDS and if this was dependent on drug load. The selected system, known to produce SNEDDS, consisted of a nonionic surfactant (Cremophor RH40), a mixture of long chain mono-, di-, and triacylglycerides (Maisine 35-1 and Sesame oil) and ethanol. Evaluation of self-emulsifying properties and screening for SNEDDS were evaluated when dispersed in saline using visual inspection and turbidity measurements. Three different lipophilic model drug compounds (danazol, halofantrine, and probucol) were selected on the basis of different physicochemical properties to represent a range of lipophilic compounds. Their solubility was determined in the preconcentrate and precipitation was evaluated after dispersion. Particle size of the identified SNEDDS with and without drug load was evaluated using laser diffraction analysis, dynamic light scattering, and small-angle X-ray scattering (SAXS).

MATERIALS AND METHODS Materials Cremophor1 RH40 (Cr RH40) (polyethoxylated hydrogenated castor oil obtained by ethoxylating hydrogenated castor oil with 40 mol ethylenoxide per mol) was obtained from BASF-BASIS Kemi, Copenhagen, Denmark and Maisine1 35-1 (obtained by partial alcoholysis of maize oil and contains a mixture of monoacylglycerides (MAG), diacylglycerides (DAG), and triacylglycerides (TAG) JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

878

NIELSEN ET AL.

(39.7% MAG, 44.6% DAG, and 14.9% TAG)), was from Gattefosse, Saint-Priest cedex, France. Sesame oil was from Apoteket AB, Sweden and absolute ethanol was from Kemetyl AB, Stockholm, Sweden. Danazol (USP grade) was obtained from Unikem, Copenhagen, Denmark and Halofantrine was kindly provided by GlaxoSmithKline, Uxbridge, UK. Probucol was from Sigma-Aldrich, Brøndby, Denmark and lecithin (Lipoid E 80) was kindly provided by Lipoid GmbH, Ludwigshafen, Germany. Sodium taurocholate (97.6% pure from Biosynth AG, Sweden) was received as a gift from AstraZeneca, Molndal, Sweden. All materials were used as received and all other chemicals used were of analytical grade. Water was freshly prepared by a Milli-Q water purification system from Millipore, Billerica, USA. Lipophilic Model Compounds Danazol, halofantrine, and probucol were selected as model compounds in order to investigate the impact of drug load on dispersion kinetics, particle size of resultant emulsion and possible interaction with bio-relevant media. Chemical structures are shown in Figure 1 and selected physiochemical properties are given in Table 1. Preparation of Preconcentrate Cr RH40, Maisine 35-1, and Sesame oil were melted at elevated temperature (>508C) and a 1:1 mixture of Maisine 35-1 and sesame oil were thoroughly mixed before use. Preconcentrates were prepared by weighing the excipients into a glass vial. The vial was sealed and samples equilibrated over night on a rotating mixer at 408C. Preconcentrates were stored at room temperature until their use in subsequent studies. Assessment of Dispersion Properties of the Vehicle Initial evaluation of self-emulsifying properties was carried out by visual assessment as previously described.14 In brief, the different compositions were categorized on speed of emulsification, turbidity and apparent stability of the resultant emulsion. Visual assessment was performed by dropwise addition of the preconcentrate into the media (0.7% w/v sodium chloride in ultrapure water, hereafter referred to as saline) until the amount was 1% of the media. This was done in a glass beaker at room temperature and the contents were gently stirred magnetically at approxiJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

Figure 1. Chemical structures halofantrine (B), and probucol (C).

of

danazol

(A),

mately 100 rpm. The vehicles were categorized as either forming a homogeneous dispersion within 1 min or within 10 min dispersion in water. Turbidity of the homogenous self-dispersing DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

Table 1.

879

Selected Physicochemical Properties for Model Drug Substances Danazol

Halofantrine

Probucol

337.5 4.5339 NA 1.0540

500.4 8.1b,e 8–937,c 1.841,d

516.8 11e NA 0.002–0.00542

Mw Log P pKa Water solubility [mg/mL]a a

At 378C. For the base specie. For the corresponding acid. d The protonated form (pH 2.5). e Estimated C Log P using ChemDraw Ultra version 8.0.3. b c

vehicles was measured using Aqualytical Turbidimeter (Bie&Berntsen AS, Denmark) and are expressed in formazin nephelolmetric units (FNU). The instrument was calibrated using turbidity standards, 1, 10, 100, and 1000 NTU (nephelometric turbidity units) supplied by the manufacturer. The NTU standard calibrates the following ranges in FNU: 1 NTU standard: 0– 2 FNU, 10 NTU standard: 2–20 FNU, the 100 NTU standard: 20–200 FNU, and 1000 NTU: 200–2000 FNU. Solubility Studies An excess of model drug substance was added to a glass vial containing vehicle and equilibrated on a rotating mixer at 25
18C. To represent a realistic storage temperature 258C was selected. Samples were withdrawn after 24, 48, and 72 h. The content was centrifuged for 10 min at 15000 rpm (Biofuge 15, Fellbach, Germany) and a sample of the supernatant were withdrawn and centrifuged for 10 min at 15000 rpm (Biofuge 15). An aliquot of the supernatant were withdrawn and diluted with absolute ethanol. The sample was diluted to appropriate concentration with mobile phase and analyzed by HPLC.

Analytical Methods Danazol Danazol concentrations were determined using a modified version of a HPLC method previously described.20 Danazol was analyzed by injecting 25 mL on a 150  4.6 mm Phenomenex Luna C18(2) (5 mm) column fitted with a 4  3.0 mm Phenomenex C18 guard column kept at room temperature and detected at 285 nm. The HPLC system consisted of a Hewlett Packard HPLC 1100 series equipped with diode array detector and ChemStation software. The mobile phase DOI 10.1002/jps

consisted of acetonitrile:ultra-pure water 66:34 (v/v) and danazol was eluted isocratic at a typically elution time of 6 min with a flow rate of 1.0 mL/min. Concentrations were calculated from a standard curve covering the concentration range from 0.15–15 mg/mL. Probucol Probucol concentrations were determined using a modified version of a HPLC method previously described.21 Probucol was analyzed by injecting 50 mL on a 125  4.6 mm LiChrospher 100 RP-18 (5 mm) column fitted with a 4  3.0 mm Phenomenex C18 guard column kept at 458C and detected at 242 nm. The HPLC system consisted of a LaChrom HPLC system from Merck-Hitachi equipped with column oven and HSM version 3.1.1 software. The mobile phase consisted of acetonitrile:ultra-pure water 85:15 (v/v) and probucol was eluted isocratic at a typically elution time of 9 min with a flow rate of 1.5 mL/min. Concentrations were calculated from a standard curve covering the concentration range from 0.15–200 mg/mL. Halofantrine Halofantrine concentrations were determined using a modified version of a HPLC method previously described.22 Halofantrine was analyzed by injecting 25 mL on a 150  4.6 mm Phenomenex Luna CN (5 mm) column kept at room temperature and detected at 259 nm. The HPLC system consisted of a Hewlet Packard HPLC 1100 series equipped with diode array detector and ChemStation software. The mobile phase consisted of acetonitrile: 25 mM acetate buffer (adjusted to pH 5.0) 90:10 (v/v) and halofantrine was eluted isocratic at a typically elution time of 6 min with a flow rate of 1.5 mL/ min. Concentrations were calculated from a JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

880

NIELSEN ET AL.

standard curve covering the concentration range from 0.03–3 mg/mL.

Dispersion Test and Precipitation Assessment Dispersion test were performed using a USP dissolution apparatus 2 (paddle method) (Erweka, model DT70, Germany) and simulated gastric fluid without pepsin (SGFPF) as media. The simulated SGFPF was prepared as SGF without pepsin according to USP XXVI. Hard gelatine capsules (size 000 from Capsugelß, Belgium) were filled with 1.0 g of preconcentrate and burdened with a coil of stainless steel wire in order to prevent the capsule from floating. Three capsules were transferred to 300 mL of USP SGFPF media, at 378C and with a rotating speed of 100 rpm. Samples of 10 mL were withdrawn at different times and analyzed using a turbidity meter. After analysis the samples were returned to the dissolution vessel. The dispersion experiments were carried out in duplicates. Precipitation were evaluated by dispersing the preconcentrate loaded with either 1.28% w/w danazol; 3.13% w/w halofantrine, or 7.61% w/w probucol in 0.7% saline. The drug load was equivalent to 80% of the solubility in the preconcentrate. Precipitation of drug from the emulsions was evaluated by visual inspection after 24 h.

Simulated Bio-Relevant Media Formulation interaction with endogenous amphiphiles was studied using simulated fasted and fed state intestinal fluid (FaSSIF and FeSSIF media). FaSSIF and FeSSIF were prepared according to previous descriptions.23 All ingredients were weighed into a beaker with water and the suspension was stirred and heated gently until all ingredients were solubilized. The pH was

Table 2.

Particle Size Analysis The particle size distribution of the resultant emulsions (1%) was determined by laser diffraction analysis and dynamic light scattering analysis. Laser diffraction analysis was performed on a Coulter particle size analyzer (Model LS230) or a Malvern Mastersizer S equipped with a 300RF lens and a Small Volume Sample Preparation Unit running Mastersizer software version 2.19 (Malvern Instruments, UK). The mean particle size was calculated from the volume size distribution. Dynamic light scattering (DLS) analysis was performed on a DynaPro instrument running Dynamics software version 5.26.38 (ProteinSolutions, UK). The mean particle size was calculated on %mass using the regularization histogram. In DLS, the diffusion coefficient of particles in solution is measured. Using the Stokes–Einstein equation, this is related to the hydrodynamic diameter of the particles, that is, to the diameter of a sphere that has the same diffusion properties as the particles. DLS is therefore an indirect way of determining particle size. If the sample contains a monomodal distribution of spherical particles, DLS is generally a very fast and reliable method for determining particle size. However, in the case of more complex samples, it is the experience from our laboratories that the results of a DLS experiment may be ambiguous and depend on model assumptions and the software used to analyze the data (unpublished data).

Composition of Simulated Intestinal Fluids

Component NaTC PL KCl KH2PO4 Acetate pH

adjusted at 37
18C within 0.05 of the intended pH. In order to investigate a potential pH effect on particle size, pH-modified FaSSIF and FeSSIF media were prepared in the same way. Compositions and pH of the applied simulated bio-relevant media are given in Table 2.

FaSSIF

pH-Modified FaSSIF

3 mM 0.75 mM 7.7 g/L 28.7 mM

3 mM 0.75 mM 7.7 g/L 28.7 mM

6.5

5.0

FeSSIF 15 mM 3.75 mM 15.2 g/L 144 mM 5.0

pH-Modified FeSSIF 15 mM 3.75 mM 15.2 g/L 144 mM 6.5

NaTC, sodium taurocholate; PL, phosphatidylcholine. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

Some of the samples were evaluated with SAXS.24 In a SAXS experiment, the Fourier transformation of the electron density fluctuations in a sample is measured. For the nanoemulsions the electron density was low in the oil-cores and high in the surrounding water phase. This was exploited to determine the shape and size of the nanoemulsion droplets. More generally the SAXS is used for investigating structures on the 1–100 nm scale. The SAXS experiments were performed on the Bruker NanoStar instrument at the University of Aarhus:25 The instrument is optimized for solution scattering and a q-range of ˚ is covered with the applied setup 0.009–0.28 1/A (the scattering vector, q, is defined by q ¼ 4p sin y/l, where 2y is the scattering angle and l is the wavelength of the X-rays). Background subtractions and absolute scale calibration were performed according to the standard procedures of the facility.26 The average shape and size of the particles were determined from the scattering data using a model-independent computer program for indirect Fourier transformation (IFT)27 that has previously been developed by Pedersen et al.26 The program provides the pair distance distribution function p(r). This function is a histogram of pairs of distances within a particle weighted by the excess electron density at the points, where ‘‘excess’’ means electron density difference relative to that of the solvent. For monodisperse homogeneous spherical particles, p(r) is bell-shaped and symmetric with maximum value at the radius R of the particle; the function goes to zero at r ¼ 2R. The function deviates from the ideal shape if the particles are not spherical or if they are polydisperse.28,29 A SAXS experiment typically takes longer time to carry out than a DLS experiment. However, the SAXS is a more direct way of determining particle size, shape, and internal structure and generally, the SAXS results for particle diameters are more reliable. By combining DLS and SAXS, the advantages of both techniques could be exploited; DLS was used for making a relatively quick screening of the samples. While SAXS was used to re-investigate a series of samples, that gave us unsystematic results when evaluated by DLS. All particle size determinations in bio-relevant media were performed after equilibrating the dispersion for at least 4 h at 37
18C. The temperature of 378C was selected to mimic temperature in vivo and the 4 h incubation time selected was employed to evaluate the effect at equilibrium as suggested by preliminary studies DOI 10.1002/jps

881

(data not shown). If necessary the samples were diluted with media in order to obtain an appropriate count rate.

RESULTS AND DISCUSSION The evaluated system was composed of the nonionic surfactant Cremophor RH40 (Cr RH40), a lipid phase consisting of a 1:1 w/w mixture of Maisine 35-1 and sesame oil, and ethanol. The system has previously been found to include self-emulsifying compositions giving resultant emulsions with particle size in the nanometer range, designated SNEDDS in the following.14,30 Screening for SNEDDS All samples evaluated were isotropic and homogenous at 258C. Samples containing up to 80% w/w ethanol or up to 90% w/w Cr RH40 were evaluated in the screening. The samples containing high proportions of ethanol and Cr RH40 were primarily included in order allow for a discussion of the different factors on a more mechanistic level. It should be noted that formulations containing larger proportions of ethanol than 10%–15% would not be suitable in gelatine capsules. However, they are still applicable as preconcentrate which are dispersed in a glass of water prior to administration (as Neoral1 Oral Solution). Turbidity was used for screening the potential SNEDDS for optimal dispersion properties. Preliminary studies showed that the SNEDDS were robust with respect to particle size when diluted with saline in the range 0.1%–5% (v/v). A sample concentration of 1% preconcentrate was selected in order to simulate a realistic dilution in vivo upon administration with a glass of water and the dispersion properties and turbidity were assessed by dispersing 1% preconcentrate in water at room temperature. The results are summarized in Figure 2. Generally, an increasing ratio of Cr RH40 to lipid phase increased the time needed for dispersion but decreased turbidity of the resulting dispersion. All compositions below the gray line in Figure 2 gave dispersions with low turbidity (<20 FNU) and long dispersion times (1–10 min). The border between fast dispersing (<1 min) and slow dispersing (1–10 min) compositions coincides with the border (gray line) between compositions resulting in low and intermediate turbidity JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

882

NIELSEN ET AL.

Figure 2. Mapping of the dispersion time and turbidity assessed by dispersion of 1% preconcentrate in saline (0.7% w/v sodium chloride) at room temperature. The lower solid line indicates the border between fast dispersing (<1 min) and slow dispersing (1–10 min) compositions with the latter being below the line. The lower solid line also indicates the border between compositions resulting in low turbidity (<20 FNU) and intermediate turbidity (20–1000 FNU) with the former being below the line. The upper solid line indicates the border between compositions resulting in intermediate turbidity (below the line) and compositions resulting in high turbidity (>1000 FNU) or compositions showing phase separation within 24 h. Particle size and distribution, turbidity and dispersion time for the compositions denoted A, B, C, D, E, and F are given in Table 3. Particle size distribution for the compositions denoted I and IV are given in Figure 3 panel A and panel B, respectively. Dispersion time from capsules for the compositions I, II, and III are given in Table 4.

(20–1000 FNU). The compositions above the black line in Figure 2 gave dispersions with high turbidity (>1000 FNU) and/or dispersions that were unstable when stored for one day. The compositions between the gray and black line were rapidly self-dispersing, gave dispersions with intermediate turbidity and were therefore identified as potential SNEDDS. Table 3 summarizes the turbidity values and corresponding particle sizes for dispersions of compositions with increasing lipid content and constant Cr RH40 to ethanol ratio (1:1) indicated in Figure 2 as compositions A–F. The table indicates that the turbidity increases initially with lipid content, but declines slightly when the particle size of the particles gets into the micrometer region. When the particle sizes are in the nanometer range, turbidity can be related to particle size. This makes turbidity measurements a fast and easy screening tool in the development of SEDDS. Figure 2 shows that the amount of ethanol in the vehicles has the greatest effect on turbidity of the resultant emulsion when the proportion of lipid phase is above approximately 30%–40% w/w. This is illustrated by comparing the particle size distribution of compositions denoted I and IV in Figure 2 displayed in Figure 3 panel A and B, respectively. The compositions of I and IV have the same ratio of Cr RH40 to lipid phase (55:45) but different amounts of ethanol. The sample with 10% w/w ethanol (I) gives a well-defined mono-modal dispersion with submicron particles (turbidity 52 FNU) while the sample with 20% w/w ethanol (IV) displays a very broad particle size distribution (turbidity >1000 FNU). Exploiting this feature it

Table 3. Particle Size Distribution and Turbidity of Aqueous Dispersion (99% Water) of the Preconcentrate Compositions Shown in Figure 2

Sample ID A B C D E F I II III IV

Lipid Phase (%)

Cr RH40 (%)

Ethanol (%)

0 10 20 40 60 80 40 50 60 35

50 44.5 40 30 19.5 10 50 40 30 45

50 45.5 40 30 20.5 10 10 10 10 20

D(0.5)a

Monomodal

<40 nm <40 nm 131 nm 484 nm 516 nm 27.98 mm

— — Yes No No Yes

Turbidity (FNU) 2.2 23 564 >1000 >1000 957 27 62 181 >1000

Dispersion Time >1 <1 <1 <1 <1 <1

min min min min min min

<1 min

a

D(0.5): Mean volume diameter measured using laser diffraction analysis.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

883

Figure 3. Volume % versus particle size diameter (mm) measured using static light scattering of the resultant emulsion in saline generated from two compositions (I and IV in Fig. 2) containing the same ratio (55:45 w/w) of Cr RH40 to lipid phase but 10% w/w (panel A) or 20% w/w (panel B) of ethanol, respectively.

is possible to develop self-emulsifying formulations having the same surfactant to lipid phase ratio but yielding emulsions with very different particle size distributions. We are currently comparing absorption of probucol from a SNEDDS and a SEDDS with equal lipid phase to Cr RH40 ratio in order to evaluate how the particle size distribution of the emulsions and the bioavailability of probucol are related. DOI 10.1002/jps

The observed sensitivity to ethanol, for mixtures of approximately equal contents of lipid phase and surfactant, exemplified by composition I and IV in Figure 3, indicates that the preconcentrate with 10% ethanol has different structure than the one with 20%. A possible explanation for the difference could be that at the lower content of polar solvent a structured L2 phase exists, which easily inverts into a water continuous well-defined JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

884

NIELSEN ET AL.

microemulsion, while the preconcentrate with high-ethanol content is a true solution, forming undefined aggregates upon dilution. Further studies, with other methods like NMR, are needed to fully understand the observation.31

Drug Solubilization Capacity of the Vehicles For the further evaluation of preconcentrate containing a drug, an ethanol concentration of 10% w/w was chosen. This ethanol concentration was selected as a compromise between our wishes to optimize dispersion kinetics and obtain a welldefined mono-modal dispersion with particle sizes in the nanometer range. One important consideration when formulating a self-emulsifying formulation is to avoid precipitation of drug in vivo.10 The obvious prerequisite for this is that the drug is soluble in the preconcentrate and stays solubilized in the resultant dispersion. Therefore the maximal drug load of the three model compounds in preconcentrates containing 10% w/w ethanol and various proportions of lipid phase was determined. The results are summarized in Figure 4. The observed solubility rank order (probucol > halofantrine > danazol)

were, in accordance with the rank order of their lipophilicity, represented by their log p-values. The solubility of probucol in the preconcentrates was around 100 mg/g and almost independent of vehicle composition. With increasing lipid content of the preconcentrate the solubility of halofantrine increased from 40.2 to 57.4 mg/g. The solubility of halofantrine in the preconcentrate is in good accordance with the reported solubility in soybean oil and Maisine of 47.3 and 49.1 mg/g, respectively32 and the solubility in ethanol was 72.5
3.7 mg/g. The solubility of danazol in the preconcentrates decreased from 31.8 to 16.0 mg/g, when the proportion of lipid content increased from 30% to 90%. Danazol has a relative low solubility in sesame oil (3.1
0.3 mg/g) and in Maisine (11.9 mg/g),32 whereas the solubility in ethanol has been determined to 36
1.6 mg/g. The relative low oil solubility and relative high-ethanol solubility compared to the solubility in the different preconcentrates indicate that the solubility of danazol in the preconcentrate is primarily related to the proportion of Cr RH40 and to the content of ethanol. Cr RH40 has a HLB value of approximately 15 whereas the lipid phase has a lowHLB value resulting in mixtures with increasing

Figure 4. Equilibrium solubility of danazol (~), halofantrine (&), and probucol (*) versus proportion of lipid phase in preconcentrates containing a fixed content (10% w/w) of ethanol determined at 25
18C (n  3). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

proportion of Cr RH40 that has an increasing HLB value. Solubility of danazol in Labrafil-Tween systems has previously been shown to increase linearly with increasing HLB of the mixtures.33 The resultant emulsions formed after dispersion of the preconcentrates (SEDDS) potentially possess a drug solubilization capacity that differs from the preconcentrate. Therefore precipitation was evaluated after dispersion (1%) in saline using the same preconcentrate compositions as used in the solubility determination. These were loaded with an amount of drug equivalent to 80% of the solubility in the preconcentrate, which were either 1.28% w/w danazol; 3.13% w/w halofantrine or 7.61% w/w probucol. Halofantrine and probucol did not precipitate after dispersion within the tested time frame (24 h) whereas danazol did. For this reason, was precipitation evaluated from preconcentrates containing a danazol load equivalent to 60% and 40% of the solubility in the preconcentrate, which were 0.96% w/w and 0.64% w/w, respectively. Precipitation occurred for the former but not the latter. This indicates that the ethanol, in the case of danazol, works as an efficient solubilization enhancer in the preconcentrates as discussed earlier. However, being water-soluble, ethanol is anticipated to enter the water phase upon dispersion and reported to redistribute mainly between the water phase and emulsion-water interphase34 resulting in a loss of solvent capacity of the vehicle. This emphasizes the importance of evaluating precipitation upon dispersion, when formulating SEDDS using ethanol as cosolvent, in cases where the compound has limited oil solubility compared to ethanol solubility. In vivo precipitation of drug compound from the formulation after administration is also of concern for formulations when they contain digestible excipients. Estimation of precipitation in vivo is however difficult due to the complex nature of concurrent digestion and absorption processes. Estimation of the in vivo precipitation using an in vitro digestion model is currently being evaluated in our lab. In the following, the impact of drug load of halofantrine and probucol on the dispersion kinetics of three selected SNEDDS (composition I, II, and III as indicated on Fig. 2)

885

and must preferably be fast and not slower than stomach emptying. Dispersion kinetics from capsules containing the preconcentrate was evaluated in simulated gastric fluid, pepsin free (SGFPF) in a USP dissolution apparatus at 100 rpm which Scholz et al.35 has found to simulate hydrodynamic conditions in the upper GI tract. The time needed for dispersion from the capsules was longer compared to what was found during the screening, due to time needed to disintegrate the capsule before dispersion can occur (Tab. 4). Loading of the vehicle with halofantrine and probucol has only minor influence on the time for dispersion. Particle Size—Impact of Dispersion Media and Drug Load Particle size of self-emulsifying systems intended for oral administration is often tested in simple buffer media, but the small intestine contains endogenous surfactants like bile salts (BS) and lecithin (PL), which potentially could influence particle size of the resultant emulsions. Therefore, the particle size was evaluated in media simulating fasted and fed state, FaSSIF and FeSSIF, respectively. The particle sizes of the mixed micelles in FaSSIF and FeSSIF were determined using DLS to 57.2
2.2 nm and 5.8
0.0 nm (n ¼ 3), respectively,. The different particle sizes of the micelles in FaSSIF and FeSSIF is an effect of the high free monomer concentration of the bile salt. Macroscopically, the FaSSIF and FeSSIF media have similar molar ratios of phospholipid and bile salt (BS/PL). However, due to the high free monomer concentration of the bile salt, the FaSSIF micelles have a much lower BS/PL than the FeSSIF micelles. It is well documented in the literature that this induces the formation of PL-rich wormlike micelles and vesicles.36,37 Table 4. Time (Minutes) for 90% Dispersion of Formulation in Hard Gelatine Capsules in USP Paddle Dissolution Containing SGFPF Medium at 100 rpm I

II

III

<10 <10

<5 <5

<3 <2

<6

<4

<3.5

Dispersion Kinetics of Selected SNEDDS

Vehicle Halofantrine loaded (3.13% w/w) Probucol loaded (7.61% w/w)

Dispersion of the preconcentrate is necessary for the proper function of a self-emulsifying system

Composition I, II, and III contains 50:40:10 (Cr RH40:lipid phase:ethanol), 40:50:10, and 30:60:10, respectively.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

886

NIELSEN ET AL.

Preconcentrate compositions with 10% w/w ethanol and containing different ratios of Cr RH40 and lipid phase were tested by dispersing 1% in saline, FaSSIF, and FeSSIF. In saline, stable emulsions could only be obtained when the lipid content was below 60% w/w. Dispersion in FaSSIF, and to greater extent FeSSIF media, had a stabilizing effect on the resultant emulsions from preconcentrates with a lipid content greater than 60% w/w, in terms of a slower phase separation.

The impact of FeSSIF on particle size of dispersions from a preconcentrate with a high-lipid phase to Cr RH40 ratio (Cr RH40:lipid phase:ethanol, 13:77:10), measured using static light scattering, is illustrated in Figure 5. As can be seen, these preconcentrates were quite sensitive to BS/PL media. Stable emulsions were not obtained, but it is possible that compositions borderline to stable resultant emulsions in water will be stable in FaSSIF and FeSSIF media or in the GI-tract,

Figure 5. Volume % versus particle size diameter (mm) measured using static light scattering of the resultant emulsion generated from a lipid-rich vehicle (Cr RH40:lipid phase:ethanol; 13:77:10) dispersed in saline (panel A) and in fed-state simulated intestinal fluid (FeSSIF) (panel B). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

which should be considered during formulation development. The DLS evaluation of the nanoemulsions obtained when dispersing the drug-free preconcentrate (I–III) in saline revealed an increase in particle size from 28.9
1.2 to 41.5
0.2 nm when the lipid content was increased from 40% to 60% w/ w (Tab. 5). This effect can be directly correlated to the composition of the vehicle formulation: A decreased surfactant to lipid ratio of the vehicle, gives rise to a decreased surface to volume ratio of the formed nanoemulsion droplets and therefore an increase of the droplet size. The particle sizes of these nanoemulsion droplets arising from the dispersed blank vehicles were virtually independent of dispersion media and not affected by dispersion in FaSSIF and FeSSIF media, as seen for lipid-rich and less Cr RH40 stabilized compositions. Loading the vehicle with halofantrine and probucol did not affect the particle size of the resultant nanoemulsion when dispersed in saline. However, when dispersed in BS/PL media a slight change in particle size was found for halofantrine, but not probucol, loaded formulations when evaluated using DLS (Tab. 5). The increase in particle size was found after dispersion in FeSSIF but not in FaSSIF and was only significant for composition III, containing the highest lipid load of the tested compositions. Modification of pH in the two media indicated a relation between pH and particle size in FaSSIF

887

and FeSSIF media when studied by DLS. Particle size of the mixed micelles in pH-modified FaSSIF and pH-modified FeSSIF were not different from standard FaSSIF and FeSSIF (54.7
2.7 nm and 5.4
0.5 nm (n ¼ 3)), showing that the mean particle size of the mixed micelles was dependent on BS/PL concentration, but independent of pH in the range from 5 to 6.5. As mentioned in the Materials and Methods, DLS is an indirect way of determining particle size and a DLS analysis does not always give ambiguous results. In order to investigate the possible relation between halofantrine and nanoemulsion droplet size as well as the possible interaction with the BS/PL micelles in FaSSIF and FeSSIF, a series of SAXS measurements were carried out on samples of composition III with and without halofantrine and dispersed in different media (Saline, FaSSIF, FeSSIF, and pH modified FaSSIF and FeSSIF). The SAXS data and the corresponding pair-distance distribution functions, p(r), from pH-modified FeSSIF, from the nanoemulsion droplets of composition III with and without halofantrine and dispersed in saline, and from the same nanoemulsion droplets in pH-modified FeSSIF are shown in Figure 6. In the SAXS data from the FeSSIF medium an oscillation is observed at high q, while at low q the scattering data approximately follows a q2 dependence (Fig. 6 panel A). The high-q behavior is most likely due to a core/shell contrast of the aggregates formed in the aqueous medium. The

Table 5. Particle Size Determined by DLS (mean
SD, n ¼ 3) and [SAXS] of Preconcentrate Composition I, II, and III as Vehicle (Veh), Loaded with 7.61% w/w Probucol (Pro) or Loaded with 3.13% w/w Halofantrine (Hf) in Saline, FaSSIF, FeSSIF, pH-Modified FaSSIF, and pH-Modified FeSSIF Formulation Vehicle Composition Drug I

Veh Hf Pro Veh Hf Pro Veh Hf Pro

II

III

Dispersion Media

Saline

FaSSIF

28.9
1.2 28.0
2.6 26.4
2.2 32.7
0.6 34.6
2.3 34.8
0.5 41.5
0.2 [30.5] 41.7
2.3 [31.5] 45.0
3.4

25.6
1.0 24.9
3.1 25.6
2.0 29.3
0.8 31.7
0.8 31.7
1.1 38.4
1.8 [28.4] 41.7
2.4 [27.9] 42.6
1.5

FeSSIF 23.0
0.3 30.7 NS
2.0 28.4
1.1 31.3
0.9 40.6NS
6.0 33.9
1.3 39.0
3.2 [28.9] 56.2*
3.1 [30.2] 44.1
1.0

pH Modified FaSSIF

pH Modified FeSSIF

ND ND ND ND ND ND ND ND ND ND ND ND 36.3
3.4 [29.4] 39.2
6.1 [31.5] 55.4*
1.8 [29.2] 49.7NS
5.5 [34.1] 46.3
6.4 44.9
5.7

Composition I, II, and III contains 50:40:10 (Cr RH40:lipid phase:ethanol), 40:50:10, and 30:60:10, respectively. *Tested and found significantly different from saline (p < 0.05, Student’s t-test). NS Tested and not found to differ significantly from saline (Student’s t-test). DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

888

NIELSEN ET AL.

Figure 6. Examples of SAXS data (left column) and deduced pair-distance distribution functions (right column) of pH-modified FeSSIF (panels A and B), composition III dispersed in saline (panels C and D), Halofantrine-loaded composition III dispersed in saline (panels E and F), composition III dispersed in pH-modified FeSSIF (panels G and H) and Halofantrine-loaded composition III dispersed in pH-modified FeSSIF (panels I and J). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

hydrocarbon chains of the PL form a hydrophobic core with a lower electron density than water, while the hydrophilic headgroups of the PL in combination with the BS form a hydrophilic layer that surrounds the hydrophobic core and has a higher electron density than water. The resulting oscillating excess electron density profile of the aggregates gives rise to the observed oscillation at small r in the p(r) function, which in turn results in the oscillation of the SAXS data at high q.28,29 It is not clear from the SAXS data what type of aggregates is formed in the FeSSIF medium. The ˚ ) indicates that small high-q part (q > 0.03 1/A micelles are formed. This is also what would be expected from the composition and concentration of the FeSSIF. However, the q2 dependence of the low-q part as well as the nearly bell-shaped pairdistance distribution function (Fig. 6 panel B) indicate that larger aggregates, maybe bilayer vesicles are present in the sample. The SAXS data from composition III in saline are plotted in Figure 6 panel C. The corresponding bell-shaped pair-distance distribution function (Fig. 6 panel D) shows that spherical droplets with a relatively low polydispersity are formed in the sample. From the IFT analysis of the scattering data an average radius of gyration of the droplets of RG ¼ 11.8 nm is found. This corresponds to an average diameter of approximately D ¼ 30.5 pffiffiffiffiffiffiffiffinm (when using the solid sphere result RG ¼ 3=5R, where R is the radius of the droplets). Addition of halofantrine to composition III does not affect this picture (Fig. 6 panel E and F), and spherical droplets with a RG ¼ 12.0 nm corresponding to D ¼ 31.0 nm are found. No effect on the size of the nanoemulsion droplets was observed when the saline aqueous medium was substituted with FeSSIF (Fig. 6 panel G and H). Large droplets of approximately the same size as in saline were observed. Around ˚ an oscillation was observed, which q ¼ 0.1 1/A corresponded largely to the oscillation observed in the pure FeSSIF medium and the scattering data indicate that the BS/PL aggregates and the composition III droplets coexist side by side. However, our data do not allow us to exclude a weak mixing of the two types of aggregates. The same conclusions hold for the Halofantrine loaded nanoemulsion droplets in pH-modified FeSSIF where no significant effect is observed upon loading composition III with Halofantrine (Fig. 6 panel I and J). As seen from Table 5, which compares the particle diameters determined by SAXS and DOI 10.1002/jps

889

DLS, the SAXS results for the droplet sizes are very close to 30 nm for all samples. Only minor variations in particle size are observed. The obtained particle diameters are thus significantly lower and less variable than the hydrodynamic diameters obtained by DLS. The observed difference can in part be assigned to systematic differences between SAXS and DLS: in SAXS, the main electron density difference originates from the oil region of the droplets, consisting of the lipid phase and of the hydrocarbon part of the surfactant. These both have a large negative electron density relative to the water. The surrounding shell consisting of the ethylene oxide (EO) head groups of the Cr RH40 surfactant has positive electron density difference; this, however, is significantly reduced due to the presence of a large amount of water in the EO shell. Therefore, we expect that the EO region is not reflected in our p(r) functions. DLS measures the diffusion constants of the droplets and relate them to the size. Since the water in the EO region follows the diffusion of the particles, the full extent of the EO region is reflected in the size derived from DLS, and the hydrodynamic diameter is therefore larger than the size determined by SAXS. The difference in the diameter found by DLS and SAXS is 11 nm (composition III, Tab. 5), implying that the EO shell has a thickness of about 5 nm. This seems reasonable considering the average of 40 EO groups per Cr RH40 surfactant molecule. As described above, dispersion in the different media had no effect on the particle size of the halofantrine-loaded composition III formulation when using SAXS. This observation is in line with the SAXS results for the vehicle in the equivalent media, but contradicts the observations using DLS. We are unable to provide a precise explanation for the deviating results obtained using DLS and SAXS. However, the fact that the effect was only seen for the most lipid-rich composition indicates that halofantrine is present in the nanoemulsion/water interphase30 and that the concentration of halofantrine in the surface is coupled to a decreased Cr RH40 content. Halofantrine is a base and the corresponding acid has an apparent pKa value of 6.92 in NaTC/PC media38 implying that halofantrine in the nanoemulsion/water interface is primarily protonated at pH 5.0 while partially protonated at pH 6.5. The employed BS, taurocholate, and PC are deprotonated and hence negatively charged at both pH 5.0 and 6.5, as they have pKa values of approximately 1 and 0.8, respectively.39 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

890

NIELSEN ET AL.

One possible explanation to the apparent increase in halofantrine loaded particle size observed in FeSSIF and pH-modified FaSSIF media, both pH 5.0, using DLS is a possible electrostatic interaction between positively charged halofantrine in the surface of the formulations particles and negatively charged amphiphiles (BS and/or PL) or micelles. The interaction could cause a decreased diffusion coefficient of the nanoemulsion particles and hence an increase in apparent particle size. Another possibility is that the interaction causes depletion of BS from the micelles in media and hence alters the form and particle size of the micelles. This explanation is however less plausible as the total concentration of halofantrine only amount to 0.625 mmol per liter compared to the BS concentration of 15 and 3 mM in the FeSSIF and pH-modified FaSSIF media, respectively. Overall, we assign the deviations in the determined particle sizes using DLS and SAXS for the halofantrine-loaded nanoemulsions to the methodological shortcomings of the DLS method discussed in the Particle Size paragraph of the Materials and Methods.

CONCLUSION We have used turbidity measurements for rapid screening of different combinations of excipients in the development of robust SNEDDSs. Drug solubilization capacity of the preconcentrates were found to be compound dependent and upon dispersion precipitated danazol, for which the solubility was governed by ethanol. In the identified SNEDDS, the inclusion of a lipophilic drug compound had no profound effect on dispersion properties and particle size in saline. The droplet size of the identified self-nanoemulsifying compositions was not affected significantly by whether the system was dispersed into the media simulating gastrointestinal fluid, respectively, FaSSIF and FeSSIF, or saline. Inclusion of the drugs procubol and halofantrine did also not affect the particle size significantly. This suggests that SNEDDS can be optimized without drug loading, at least for the three selected model drug compounds and that the endogeneous surfactants in the GI tract only play a minor role for the dispersion properties. However, the importance of studying drug solubility after dispersion were emphasized by the danazol case were the solubiJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

lization capacity of the SNEDDS preconcentrate greatly exceed the solubilization capacity of the dispersed SNEDDS.

ACKNOWLEDGMENTS This work is part of the project ‘‘Explorative Pharmaceutical Formulations’’ and financially supported by VINNOVA in Sweden and VTU— Ministry of Science, Technology and Innovation in Denmark. Linda So¨derberg, Camurus AB are acknowledged for excellent technical assistance.

REFERENCES 1. Lipinski CA. 2000. Drug-like properties and the causes of poor solubility and poor permeability. J Pharmacol Toxicol Methods 44:235–249. 2. Constantinides PP. 1995. Lipid microemulsions for improving drug dissolution and oral absorption: Physical and biopharmaceutical aspects. Pharm Res 12:1561–1572. 3. Humberstone AJ, Charman WN. 1997. Lipid-based vehicles for the oral delivery of poorly water soluble drugs. Adv Drug Deliv Rev 25:103–128. 4. Gershanik T, Benita S. 2000. Self-dispersing lipid formulations for improving oral absorption of lipophilic drugs. Eur J Pharm Biopharm 50:179– 188. 5. Pouton CW. 1997. Formulation of self-emulsifying drug delivery systems. Adv Drug Deliv Rev 25: 47–58. 6. Porter CJ, Charman WN. 2001. In vitro assessment of oral lipid based formulations. Adv Drug Deliv Rev 50:S127–S147. 7. Gursoy RN, Benita S. 2004. Self-emulsifying drug delivery systems (SEDDS) for improved oral delivery of lipophilic drugs. Biomed Pharmacother 58:173–182. 8. de Smidt PC, Campanero MA, Troconiz IF. 2004. Intestinal absorption of penclomedine from lipid vehicles in the conscious rat: Contribution of emulsification versus digestibility. Int J Pharm 270:109–118. 9. Kossena GA, Charman WN, Boyd BJ, Dunstan DE, Porter CJ. 2004. Probing drug solubilization patterns in the gastrointestinal tract after administration of lipid-based delivery systems: A phase diagram approach. J Pharm Sci 93:332–348. 10. Pouton CW. 2000. Lipid formulations for oral administration of drugs: Non-emulsifying, selfemulsifying and ‘self-microemulsifying’ drug delivery systems. Eur J Pharm Sci 11:S93–S98. DOI 10.1002/jps

CHARACTERIZATION OF PROTOTYPE SELF-NANOEMULSIFYING FORMULATIONS

11. el Tayar N, Mark AE, Vallat P, Brunne RM, Testa B, van Gunsteren WF. 1993. Solvent-dependent conformation and hydrogen-bonding capacity of cyclosporin A: Evidence from partition coefficients and molecular dynamics simulations. J Med Chem 36:3757–3764. 12. Garti N, Yaghmur A, Leser ME, Clement V, Watzke HJ. 2001. Improved oil solubilization in oil/water food grade microemulsions in the presence of polyols and ethanol. J Agric Food Chem 49:2552–2562. 13. Pouton CW. 1985. Self-emulsifying drug delivery systems: Assessment of the efficiency of emulsification. Int J Pharm 27:335–348. 14. Khoo SM, Humberstone AJ, Porter CJH, Edwards GA, Charman WN. 1998. Formulation design and bioavailability assessment of lipidic self-emulsifying formulations of halofantrine. Int J Pharm 167: 155–164. 15. Taha EI, Al Saidan S, Samy AM, Khan MA. 2004. Preparation and in vitro characterization of selfnanoemulsified drug delivery system (SNEDDS) of all-trans-retinol acetate. Int J Pharm 285:109–119. 16. Nazzal S, Smalyukh II, Lavrentovich OD, Khan MA. 2002. Preparation and in vitro characterization of a eutectic based semisolid self-nanoemulsified drug delivery system (SNEDDS) of ubiquinone: Mechanism and progress of emulsion formation. Int J Pharm 235:247–265. 17. Kim JY, Ku YS. 2000. Enhanced absorption of indomethacin after oral or rectal administration of a self-emulsifying system containing indomethacin to rats. Int J Pharm 194:81–89. 18. Kommuru TR, Gurley B, Khan MA, Reddy IK. 2001. Self-emulsifying drug delivery systems (SEDDS) of coenzyme Q10: Formulation development and bioavailability assessment. Int J Pharm 212:233–246. 19. Kawakami K, Yoshikawa T, Moroto Y, Kanaoka E, Takahashi K, Nishihara Y, Masuda K. 2002. Microemulsion formulation for enhanced absorption of poorly soluble drugs. I. Prescription design. J Control Release 81:65–74. 20. Boyd BJ, Porter CJ, Charman WN. 2003. Using the polymer partitioning method to probe the thermodynamic activity of poorly water-soluble drugs solubilized in model lipid digestion products. J Pharm Sci 92:1262–1271. 21. Jizomoto H, Kanaoka E, Sugita K, Hirano K. 1993. Gelatin-acacia microcapsules for trapping micro oil droplets containing lipophilic drugs and ready disintegration in the gastrointestinal tract. Pharm Res 10:1115–1122. 22. Abdul-Fattah AM, Bhargava HN. 2002. A new high-performance liquid chromatography (HPLC) method for the analysis of halofantrine (HF) in pharmaceuticals. J Pharm Biomed Anal 29:901– 908.

DOI 10.1002/jps

891

23. Galia E, Nicolaides E, Horter D, Lobenberg R, Reppas C, Dressman JB. 1998. Evaluation of various dissolution media for predicting in vivo performance of class I and II drugs. Pharm Res 15:698–705. 24. Spalla O. 2002. General theorems in small angle scattering. In: Lindner P, Zemb Th. editors. Neutrons, X-rays and Light: Scattering methods applied to soft condensed matter. Amsterdam: Elsevier, pp 43–72. 25. Pedersen JS. 2004. A flux- and background-optimized version of the NanoSTAR small-angle X-ray scattering camera for solution scattering. J Appl Crystallogr 37:369–380. 26. Pedersen JS, Hansen S, Bauer R. 1994. The aggregation behavior of zinc-free insulin studied by small-angle neutron-scattering. Eur Biophys J Biophy 22:379–389. 27. Glatter O. 1977. New method for evaluation of small-angle scattering data. J Appl Crystallogr 10: 415–421. 28. Glatter O. 2002. The inverse scattering problem in small-angle scattering. In: Lindner P, Zemb Th, editors. Neutrons, X-rays and light: Scattering methods applied to soft condensed matter. Amsterdam: Elsevier. pp 73–102. 29. Glatter O. 2002. Fourier transformation and deconvolution. In: Lindner P, Zemb Th, editors. Neutrons, X-rays and light: Scattering methods applied to soft condensed matter; Amsterdam: Elsevier. pp. 103–126. 30. Ljusberg-Wahren H, Seier NF, Brogard M, Troedsson E, Mullertz A. 2005. Enzymatic characterization of lipid-based drug delivery systems. Int J Pharm 298:328–332. 31. Nilsson PG, Lindman B. 1982. Solution structure of nonionic surfactant microemulsions from nuclear magnetic resonance self-diffusion studies. J Phys Chem-US 86:271–279. 32. Kaukonen AM, Boyd BJ, Porter CJ, Charman WN. 2004. Drug solubilization behavior during in vitro digestion of simple triglyceride lipid solution formulations. Pharm Res 21:245–253. 33. Devani M, Ashford M, Craig DQ. 2004. The emulsification and solubilisation properties of polyglycolysed oils in self-emulsifying formulations. J Pharm Pharmacol 56:307–316. 34. de Campo L, Yaghmur A, Garti N, Leser ME, Folmer B, Glatter O. 2004. Five-component foodgrade microemulsions: Structural characterization by SANS. J Colloid Interface Sci 274:251– 267. 35. Scholz A, Kostewicz E, Abrahamsson B, Dressman JB. 2003. Can the USP paddle method be used to represent in-vivo hydrodynamics? J Pharm Pharmacol 55:443–451. 36. Mazer NA, Benedek GB, Carey MC. 1980. Quasielastic light-scattering studies of aqueous biliary

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

892

NIELSEN ET AL.

lipid systems. Mixed micelle formation in bile saltlecithin solutions. Biochemistry 19:601–615. 37. Hjelm RP, Thiyagarajan P, Alkan H. 1988. A smallangle neutron-scattering study of the effects of dilution on particle morphology in mixtures of glycocholate and lecithin. J Appl Crystallogr 21: 858–863. 38. Taillardat-Bertschinger A, Perry CS, Galland A, Prankerd RJ, Charman WN. 2003. Partitioning of halofantrine hydrochloride between water, micellar solutions, and soybean oil: Effects on its apparent ionization constant. J Pharm Sci 92: 2217–2228. 39. Moncelli MR, Becucci L, Guidelli R. 1994. The intrinsic pKa values for phosphatidylcholine, phos-

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 96, NO. 4, APRIL 2007

phatidylethanolamine, and phosphatidylserine in monolayers deposited on mercury electrodes. Biophys J 66:1969–1980. 40. Mithani SD, Bakatselou V, TenHoor CN, Dressman JB. 1996. Estimation of the increase in solubility of drugs as a function of bile salt concentration. Pharm Res 13:163–167. 41. Lim LY, Go ML. 2000. Caffeine and nicotinamide enhances the aqueous solubility of the antimalarial agent halofantrine. Eur J Pharm Sci 10:17–28. 42. Yagi N, Terashima Y, Kenmotsu H, Sekikawa H, Takada M. 1996. Dissolution behavior of probucol from solid dispersion systems of probucol-polyvinylpyrrolidone. Chem Pharm Bull (Tokyo) 44:241– 244.

DOI 10.1002/jps