Rheological studies on solid lipid nanoparticle based carbopol gels of aceclofenac

Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb

Rheological studies on solid lipid nanoparticle based carbopol gels of aceclofenac Viney Chawla a , Shubhini A. Saraf b,∗ a b

Faculty of Pharmacy, Northern India Engineering College, Lucknow 227105, Uttar Pradesh, India Faculty of Pharmacy, Babu Banarasi Das National Institute of Technology and Management, Lucknow 227105, Uttar Pradesh, India

a r t i c l e

i n f o

Article history: Received 24 March 2011 Received in revised form 3 December 2011 Accepted 5 December 2011 Available online 20 December 2011 Keywords: Solid lipid nanoparticles Gels Rapid visco analyzer Aceclofenac Taguchi experimental design Carbopol

a b s t r a c t Solid lipid nanoparticles (SLN) of aceclofenac were prepared using Taguchi experimental design by Trotta method. The prepared SLN were formulated into a gel preparation, using carbopol 940. Gels were evaluated for drug content, bioadhesion and their stability against change of temperature and shear. The viscosity of prepared gels was found to be temperature independent. Rheological behavior of gels with changing shear was rather complex. Viscosity varied inversely with shear but remained almost constant during short spans of time when shear was kept constant. Viscosity of the gels did not change if shear was not varied. In vitro diffusion studies exhibited an immediate release followed by a sustained release. This could help in maintaining the concentration of bioactives such as aceclofenac in desirable levels at sites of inflammation and injury. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Solid lipid nanoparticles (SLN) are particles of submicron size (50–1000 nm) stabilized by a surfactant and made from lipids that remain in a solid state at room/body temperature. SLN have emerged as alternative carriers to colloidal systems, for controlled and targeted delivery. They are made up of biocompatible and biodegradable material, capable of incorporating lipophilic and hydrophilic drugs [1–3]. Gels of pharmaceutical significance have been prepared by using various types of materials. Carbopol is one such commonly used polymer of acrylic acid which can be crosslinked either with polyalkenyl ethers or divinyl glycol. Gels can be produced from primary polymer particles of about 0.2–6.0 ␮m average diameter. The flocculated agglomerates cannot be broken into the ultimate particles when converted in to gel form. Each particle can be viewed as a network structure of polymer chains interconnected via cross-linking [4]. Carbopols readily absorb water, get hydrated and swell. Different grades of carbopol polymers exhibit different rheological properties depending on their particle size, molecular weight between crosslinks (Mc ), distributions of Mc and fraction of the total units which appear as terminal, i.e. free chain ends. The Mc for carbopol 940 has been reported as 1450 monomer units (or 1450 × 72 = 104,400 g/mole) [5]. Carbopols are

∗ Corresponding author. Tel.: +91 522 3911132; fax: +91 522 2815187. E-mail addresses: vchawla [email protected] (V. Chawla), [email protected] (S.A. Saraf). 0927-7765/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.colsurfb.2011.12.006

essentially nontoxic and nonirritant materials with no evidence of their hypersensitivity in human subjects when used topically [6]. Viscoelasticity is a mechanical property of materials that possess a combined behavior of elastic solid and viscous fluids. These materials include melt polymers, food and pharmaceutical semisolid dosage forms such as cream, ointment and gels [7]. Viscoelastic properties of pharmaceutical gels affect their physical appearance, patient or consumer perceptions, their spreadability and flow behavior [8]. Various researchers have tried to explore the rheological properties of carbopol gels through continuous shear [9–11]. This can deform the gel structure and data thus obtained does not really represent the intact gel structure. Rapid visco analyzer (RVA) is a controlled shear rate instrument through which a constant shear rate (rpm) can be applied and the resultant torque (force, shear stress) measured. Torque and displacement are converted to rheological format by means of measuring system constants. Data can be produced in both tabular and graphical format. RVA is a Searle type viscometer, with a stationary bowl and a combined stirring and sensing element suspended concentrically. Nonlaminar or turbulent flow at high speeds prohibits absolute viscosity measurements, an effect which is exacerbated by the mixing-paddle design of the sensor element [12]. In the current work, carbopol gels of aceclofenac loaded SLN dispersion were prepared and analyzed using RVA at both constant shear and temperature with changing rate of shear, in an attempt to obtain a better picture of the rheological behavior of intact gels. Although RVA has been used frequently in food industry [12,13], this study reports for the first time its use in pharmaceutical gels.

294

V. Chawla, S.A. Saraf / Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

Table 1 Real and orthogonal values of L9 design and resulting characterization parameters. Batch

Independent variables Type of lipid [A]

1-B 2-B 3-B 4-B 5-B 6-B 7-B 8-B 9-B

CA (1) CA (1) CA (1) GP (2) GP (2) GP (2) GB (3) GB (3) GB (3)

Characterization parameters Drug:lipid ratio [B] 1:5 (1) 1:7 (2) 1:10 (3) 1:5 (1) 1:7 (2) 1:10 (3) 1:5 (1) 1:7 (2) 1:10 (3)

Surfactant conc. (%) [C]

Sonication time (min) [D]

PS (nm)

1.5 (1) 2.0 (2) 2.5 (3) 2.0 (2) 2.5 (3) 1.5 (1) 2.5 (3) 1.5 (1) 2.0 (2)

2 (1) 5 (2) 8 (3) 8 (3) 2 (1) 5 (2) 5 (2) 8 (3) 2 (1)

191.2 161.8 146.7 56.8 69.3 49.5 149.2 245.0 123.6

± ± ± ± ± ± ± ± ±

ZP (mV) 5.93 4.92 8.95 0.87 2.97 1.30 8.96 12.61 9.47

−10.70 −8.17 −10.00 −11.00 −11.40 12.30 −5.42 −6.75 −8.10

± ± ± ± ± ± ± ± ±

PI 0.18 0.19 0.50 0.18 0.23 0.14 0.05 0.03 0.08

1.000 0.753 0.956 0.949 0.600 0.450 0.852 0.761 0.580

± ± ± ± ± ± ± ± ±

0.001 0.007 0.012 0.009 0.034 0.009 0.009 0.011 0.010

Figures in parentheses indicate the levels of different variables: 1 = low, 2 = medium, and 3 = high; CA = cetyl alcohol, GP = precirol, GB = compritol, PS = average particle size, ZP = zeta potential, and PI = polydispersity.

2. Materials and methods 2.1. Materials The drug, aceclofenac, was procured as gift sample from Arbro Pharmaceuticals, India. Carbopol 940, Cetyl alcohol (CA) and triethanolamine were procured from SD Fine Chemicals, India. Compritol or glyceryl behenate (GB) and precirol or glyceryl palmitostearate (GP) were obtained as gift samples from Colorcon Asia Pvt. Ltd., India. The surfactant, poloxamer 188 (Lutrol F 68) was received ex-gratia from BASF, Germany. 2.2. Experimental design The batches were prepared as per Taguchi experimental design. The L9 orthogonal array was used (Table 1). Orthogonal array is a matrix of numbers arranged in columns and rows. The Taguchi method employs a generic signal to noise (S/N) ratio to quantify variations. These ratios are meant to be used as measures of the effect of noise factors on performance characteristics. S/N ratios take into account both amount of variability in response data and closeness of average response to the target. There are several S/N ratios available depending upon the type of characteristics: smaller the better, as is the case with particle size and polydispersity index (PI); larger the better, as is the case with drug content and zeta potential magnitude. In some cases, a nominal S/N ratio is the best [14].

SLN dispersion. Dispersion was subjected to ultrasonication (Sartorius Labsonic P Ultrasonicator) for 2–8 min and further evaporation under reduced pressure, in a rota evaporator (Buchi), to remove traces of residual organic solvents. When tested, the resultant dispersion was found to be free from organic solvent residues. 2.4. Gels enriched with nanoparticles Gels were prepared using carbopol 940 (1%). For the preparation of gels, glycerol (10%), nanoparticulate dispersion (20%) and distilled water were weighed in a beaker and stirred. Required quantity of gelling agent was dispersed in aqueous phase under continuous stirring. Neutralization was performed using triethanolamine to attain pH 7.0. The gels were stored in air tight amber colored glass jars. 2.5. Evaluation of SLN dispersion and gel 2.5.1. Characterization of SLN dispersion The prepared batches of SLN dispersion were characterized on the basis of particle size, zeta potential and PI. The particle size, zeta potential and PI were measured using Malvern zetasizer (Nano ZS). Hydrodynamic diameters of particles were recorded. Each sample was made to run five times. The reported values are averages of five determinations. 2.5.2. Characterization of prepared gels The prepared gels were evaluated for drug content, bioadhesion and their stability against change of temperature and shear.

2.3. Preparation of SLN Before preparation, the drug, surfactants and solid lipids (CA, GB and GP) were subjected to physical and spectral characterization. An absorption maximum of the drug was determined using Shimadzu Double beam UV 1700 spectrophotometer. Standard curve of the drug was prepared in methanol at concentrations varying from 10 to 100 ␮g/ml. SLN were prepared by Trotta Diffusion method [15]. The lipids were used in ratios (1:5/1:7/1:10) selected by Taguchi design (Table 1). The lipids were dissolved in a mixture of minimum quantities of chloroform and dichloromethane (2 ml each). Drug was dissolved in this mixture. The aqueous phase containing surfactant was transferred to a homogenizer and lipid phase was dispersed drop by drop into the aqueous phase with a homogenization speed of 3000 rpm. After homogenization for 30 min, the resultant emulsion was poured into ice-cold distilled water up to a volume of 50 ml and stirred (2000 rpm) for 3 h to diffuse the organic solvent into external aqueous phase. It was then centrifuged at 12,000 rpm in a refrigerated centrifuge (Sartorius F18K) at 20 ◦ C for 15 min. The solid mass, thus obtained, was dispersed in distilled water to obtain

2.5.2.1. Determination of drug content in gels. Weighed quantity of gels (10 mg) was digested with 10 ml of methanol using a vortex mixer and subsequently filtered through a Whatman filter paper. 1 ml of the filtrate was diluted with 4 ml of methanol, filtered and absorbance measured at 275 nm. Amount of drug was calculated taking into account dilution factor, if any, with the help of equation y = 0.0048x + 0.0016. A comparative account of drug content among different gels is shown in Fig. 2. 2.5.2.2. Bioadhesion studies on gels. The prepared gel batches were tested for bioadhesion potential. Texture analyzer equipment (TAXT2, Stable Micro Systems, England) was used for this purpose. Gel sample was spread as a thin film on the double adhesive tape which was cut as per shape of texture analyzer probe. The probe was lowered on to a platform after selecting suitable test parameters (pre and post speed 2 mm/s; time 5 s) in the software. The force was measured and recorded as an average of three findings in Table 3. 2.5.2.3. Samples testing on RVA. Samples of gel were tested on RVA (Peten Instruments, Australia) for their stability against change of

V. Chawla, S.A. Saraf / Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

295

Table 2 Statistical parameters of particle size for SLN dispersions. Batch

Observed response values for particle size (nm)

1-B 2-B 3-B 4-B 5-B 6-B 7-B 8-B 9-B

197.8 160.3 142.3 57.2 70.6 51.0 152.3 230.5 112.8

186.3 167.3 157.0 55.8 65.9 48.6 156.2 253.5 127.5

189.5 157.8 140.8 57.4 71.4 48.9 139.1 251.0 130.5

temperature and shear. In the first test, the sample was initially dispersed at a high speed of 960 rpm for few seconds to break any air bubbles. Later, shear was kept constant at 400 rpm. The samples were then subjected to variable temperatures between 30 ◦ C and 70 ◦ C such that the sample was kept at 30, 40, 50, 60 and 70 ◦ C temperature each for 10 min. The change over from 30 ◦ C to 40 ◦ C and so on was gradual, at a rate of 1 ◦ C/min. Viscosity was recorded after each hold time of 10 min at the designated temperature. In the second test, the samples were run in a manner where the temperature was increased at a constant rate from 30 ◦ C to 70 ◦ C. The shear was increased in steps, keeping it constant for 2 min at every step, from 60 rpm to 160 rpm, 400 rpm and 600 rpm, respectively. After holding the shear at 600 rpm for 10 s, it was dropped to 60 rpm and the low–high shear cycles were repeated. This ensured evaluation of viscosity at variable shear and temperature.

Average (nm)

S/N ratio

log S

191.2 161.8 146.7 56.8 69.3 49.5 149.2 245.0 123.6

45.62 44.18 43.32 35.09 49.85 33.81 43.47 47.78 41.84

0.77 0.69 0.95 -0.06 0.47 0.12 0.95 1.10 0.98

3. Results and discussion All peaks observed in I.R. spectra of lipids and surfactants were in agreement with those of reference spectra. I.R. spectrum of pure drug aceclofenac, which is 2-[2-[2-(2,6dichloroanilino)phenyl]acetyl]oxyacetic acid, showed characteristic peaks of secondary amine stretch at 3319 cm−1 , carbonyl group at 1716 cm−1 , aryl chloride group at 1055 cm−1 , C-H bending at 750 cm−1 and N-H bending at 1508 cm−1 . When tested on a UV spectrophotometer, the observed max for drug was in accordance with its reported absorption maxima of 275 nm [16]. Further, a concentration dependent increase in absorbance of drug solutions was observed with good correlation (y = 0.0048x + 0.0016, R2 = 0.9956). 3.1. Preparation of SLN

2.5.2.4. In vitro diffusion studies. Diffusion studies were carried out in Franz diffusion cells to determine the percentage of drug released in 3 h time. A semipermeable cellulose based membrane was used having a molecular weight cut off of 12,000 Da. The membrane was soaked overnight in phosphate buffer (pH 7.4, i.e. PBS) and subsequently placed between the donor and the receptor compartments of Franz diffusion cell. The receptor compartment was filled with PBS maintained at 37 ± 1 ◦ C using a magnetic hot plate with stirrer. After equilibration for half an hour, 2 g of gel was introduced in the donor compartment. Aliquots (1 ml) were withdrawn from receptor chamber at regular intervals of 15, 30, 60, 90, 120, 150 and 180 min. Volume of the receiving solution was maintained by replacing the amount withdrawn with an equal volume of PBS. Samples were analyzed spectrophotometrically at a wavelength of 275 nm. Diffusion coefficients were calculated from the slope of cumulative amount of drug released versus square root of time (Fig. 3).

SLN were successfully prepared by Trotta method. Batch 6-B (Table 3) had an average particle size of 49.50 nm with a PI of 0.450. The size of particles and PI values were generally lesser in case of precirol SLN when compared to cetyl alcohol and glyceryl behenate SLN. The S/N ratios were determined for each of the factors, viz., particle size, PI and zeta potential. The estimated effects from Table 2 are plotted in Fig. 4 which favoured a ‘smaller the better’ S/N ratio. In light of these observations, desirable levels of different parameters would be A2B3C2D2. Similarly, for PI desirable levels were found to be A2B3C1D2 and for zeta potential the findings were A2B3C1D1. Thus, considering the overall effect of three factors, A2B3C1D2, was found to be the best combination of formulation variables which corresponded to batch 6-B of Taguchi array. The results predicted from the design corroborated the findings in that, batch 6-B had highest magnitude of zeta potential (−12.30 mV), drug content (83%), smallest size (49.5 nm) and PI (0.45). Thus, statistical design of experiments through Taguchi experimentation techniques proved its worth.

2.6. Statistical analysis

3.2. Preparation of SLN gels

One-way analysis of variance (ANOVA) was used to determine significant difference among different batches. A 5% variation was considered as statistically significant. All values were reported as mean of three findings. The following equations were used to calculate S/N ratio in Taguchi experimental design. For S/N ratio (smaller the better):

The prepared SLN dispersions were formulated into carbopol gels. Carpobol 940 possesses excellent water sorption property. It swells in water up to 1000 times its original volume and 10 times its original diameter to form a gel when exposed to a pH environment above 6.0. Since the pKa of carbopol 940 varies between 6.0 and 0.5, the carboxylate moiety on the polymer backbone ionizes, resulting in repulsion between the native charges, which adds to the swelling of the polymer. The glass transition temperature of carbopol 940 is 105 ◦ C in powder form. However, glass transition temperature decreases significantly as the polymer comes in contact with water. The polymer chains start gyrating and radius of gyration becomes increasingly larger. Macroscopically, this phenomenon manifests itself as swelling. Carbopol 940 was neutralized with triethanolamine to pH 7.0. The added glycerol acts as a plasticizer, which increases the flexibility of polymer chains and therefore the gel elastic behavior increases [8].

n = −10 Log 10[mean of sum of squares of measured data] For S/N ratio (larger the better): n = −10 Log 10[mean of sum of squares of reciprocal of measured data]

S/N ratios for particles size are shown in Table 2.

296

V. Chawla, S.A. Saraf / Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

Table 3 Bioadhesion force (g) of different batches (n = 3). Batch

1-B

2-B

3-B

4-B

5-B

6-B

7-B

8-B

9-B

Bioadhesion force (g)

14.5 ± 0.20

22.2 ± 0.26

14.5 ± 0.19

34.1 ± 0.34

90.1 ± 0.22

85.5 ± 1.69

45.1 ± 0.95

50.2 ± 0.81

29.2 ± 0.17

the drugs is polymorphic transitions in solid lipids [19]. For similar reasons, drug content of glyceryl behenate SLN gels were higher as compared to cetyl alcohol SLN gels. 3.4. Rheological studies

Fig. 1. General structure of carbopol polymers.

Fig. 2. Drug content of solid lipid nanoparticulate gels.

3.4.1. Effect of shear Studies on RVA revealed that viscosity varied inversely with shear. For example, in case of batch 5-B (Fig. 6), value of viscosity was quite high, i.e. 12,000 cp during first 2 min at a shear value of 60 rpm. After 2 min, as shear was raised from 60 to 160 rpm, viscosity decreased to 4460 cp and remained constant during the time shear was kept constant. When shear was further raised from 160 to 400 rpm, viscosity further decreased to 2146 cp. Similarly, the maintained value of 1607 cp at 600 rpm shear value justified the inverse relationship. But as the shear was suddenly dropped at 8th min, viscosity immediately increased to 17,082 cp, suggesting the elasticity of gels. There was minimal change in viscosity during the 2 min when the shear was kept constant as indicated by an almost horizontal line during this period. Thus, at constant shear, the gels were independent of changing shear. The deformation of gel structure cannot be ruled out since the viscosity at 14 min (1366 cp) was much lower than its original viscosity. It can be concluded that if the gels are prepared at constant shear and if they have attain a particular phase and viscosity, they can be easily transported and stored at any temperature provided they are not subjected to any shear changes which can alter their viscosity and thus the stability and structure.

3.3. Drug content in gels When the gels were subjected to drug content determination, it was found that batch 6-B had maximum drug content of 83% (Fig. 1). The comparatively lower drug content in cetyl alcohol SLN gels may be due to the fact that SLN prepared from cetyl alcohol inherently have lower drug content; the lipid being a straight chain alcohol with almost no imperfections to accommodate the drug. Lipids that form highly crystalline particles with a perfect lattice (e.g. monoacid triglycerides) cause drug expulsion. More complex lipids (mixtures of mono, di, and triglycerides containing fatty acids of different chain length) form less perfect crystals with many imperfections. These imperfections provide space to accommodate the drugs. Higher drug payloads may also be achieved by mixing liquid oils with solid lipids [17]. Precirol or glyceryl palmitostearate has some imperfections so higher drug content was seen in gels prepared from its SLN [1,18]. Another factor which may expel

Fig. 3. In vitro diffusion profile of selected batch 6-B.

3.4.2. Effect of temperature When the studies were carried out at changing temperatures, it was observed that viscosity of the gel samples was more or less the same with minor fluctuations. This difference was not very significant and thus it can be said that the samples were temperature independent (Fig. 5). With a 233% increase in temperature from 30 ◦ C to 70 ◦ C, the difference in viscosity was merely 12% (2329–2046 cp). Gels were stable over high as well as low temperatures, a trait which is helpful in ensuring their shelf stability. 3.5. Bioadhesion studies Studies on texture analyzer were carried out to explore bioadhesion potential of prepared gels. Data in Table 3 revealed that batch 1-B had minimum force whereas batch 5-B exhibited maximum force. This force is a direct measure of the ability of gels to adhere

Fig. 4. Taguchi design based response graph of S/N ratio on particle size.

V. Chawla, S.A. Saraf / Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

297

Fig. 5. Effect of changing temperature on the rheological behavior of batch 5-B.

Fig. 6. Effect of changing shear on the rheological behavior of batch 5-B.

to skin or other biological membranes when exposed in vivo. Studies have shown that preparations containing high concentrations of carbopol 940 should be chosen for use in drug delivery systems and mucoadhesive dosage forms [20]. This is due to the fact that they would prolong the contact time and are resistant to fluctuating electrolyte levels in biological fluids. The increase in contact time would result in an increase in drug bioavailability.

Acknowledgments

3.6. In vitro diffusion studies

References

The batches exhibited an immediate release (Fig. 3) followed by a sustained release. This could help in maintaining the concentration of bioactives such as aceclofenac in desirable levels at sites of inflammation and injury.

4. Conclusion Carbopol gels of aceclofenac SLN have been successfully prepared and characterized. Rapid visco analyzer has been used for the first time to characterize pharmaceutical gels. The viscosity of prepared gels was independent of time and shear. Gels were stable over high as well as low temperatures which can be helpful in their long term stability and transportation.

The authors wish to acknowledge the Babu Banarasi Das Group of Educational Institutions, Lucknow, India for providing infrastructure; SDS Instruments, New Delhi for technical help; Colorcon Asia Pvt. Ltd., Goa, India for samples of lipids (compritol and precirol) and BASF, Germany for surfactant, poloxamer 188.

[1] K. Manjunath, S.J. Reddy, V. Venkateshwarlu, Methods Find. Exp. Clin. Pharmacol. 27 (2005) 1. [2] S. Utreja, N.K. Jain, in: N.K. Jain (Ed.), Advances in Controlled and Novel Drug Delivery, 1st ed., CBS Publishers and Distributors, New Delhi, 2001 (Chapter 17). [3] K. Mader, in: V.P. Trochilin (Ed.), Nanoparticulates as Drug Carriers, 1st ed., Imperial College Press, London, 2006 (Chapter 9). [4] D.L. Goode, E.A. Wiltschko, H.M. Ulmer, E.K. Arendt, J. Inst. Brew. 111 (2005) 165. [5] A.T. Florence, P.U. Jani, Drug Saf. 410 (1994) 233. [6] J.O. Carnalli, M.S. Naser, Colloid. Polym. Sci. 270 (1992) 183. [7] R.C. Rowe, in: P.J. Sheskey, M.E. Quinn (Eds.), Carbomer. In Handbook of Pharmaceutical Excipients, 6th ed., Pharmaceutical Press, London, 2009. [8] R.A. Sasutjarit, A. Sirivat, P. Vayumhasuwan, Pharm. Res. 22 (2005) 2134. [9] R. Sareen, S. Kumar, G.D. Gupta, Curr. Drug Deliv. 8 (2011) 407. [10] G. Bonacucina, M. Cespi, M. Misici-Falzi, G.F. Palmeiri, AAPS J. 10 (2008) 84. [11] P.V. Pople, K.K. Singh, AAPS Pharm. Sci. Tech. 7 (2006) 1.

298

V. Chawla, S.A. Saraf / Colloids and Surfaces B: Biointerfaces 92 (2012) 293–298

[12] S.C. Chin, C.W. Cho, H.K. Choi, Drug Dev. Ind. Pharm. 25 (1999) 273. [13] J.S. Higley, S.L. Love, W.J. Price, J.E. Nelson, K.C. Huber, Am. J. Potato Res. 80 (2003) 195. [14] P. Sharma, A. Verma, R.K. Sidhu, O.P. Pandey, J. Mater. Process. Technol. 168 (2005) 147. [15] M. Trotta, F. Debernardi, O. Caputo, Int. J. Pharm. 257 (2003) 153.

[16] Indian Pharmacopoeia, Controller of Publications, Govt of India, New Delhi, 2010. [17] M. Radtke, E.B. Souto, R.H. Muller, Pharm. Tech. Eur. 17 (2005) 45. [18] V. Chawla, S.A. Saraf, J. Am. Oil Chem. Soc. 88 (2011) 119. [19] H. Bunjes, K. Westesen, M.H.J. Koch, Int. J. Pharm. 129 (1996) 159. [20] K. Edsman, J. Carlfors, K. Harju, Int. J. Pharm. 137 (1996) 233.