- Email: [email protected]
Multivariate Optimization of Formulation Variables Influencing Flurbiprofen Proniosomes Characteristics
Multivariate Optimization of Formulation Variables Influencing Flurbiprofen Proniosomes Characteristics AHMED S. ZIDAN,1,2 MAHMOUD MOKHTAR1 1
Department of Pharmaceutics and Industrial Pharmacy, Faculty of Pharmacy, Zagazig University, Zagazig, Egypt
2
Division of Product Quality and Research, Center for Drug Evaluation and Research, Food and Drug Administration, Maryland, USA
Received 9 August 2010; revised 10 November 2010; accepted 25 November 2010 Published online 21 January 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.22453 ABSTRACT: Flurbiprofen was formulated as a proniosomal transdermal gel with high drug loading (55.4%, w/w), using a series of nonionic surfactant and cholesterol. A two-factor, threelevel randomized full factorial strategy was developed to optimize simultaneously the effect of surfactant fatty acid side chain length and the amount of cholesterol on the properties of the proniosomes, namely drug permeation characteristics such as steady-state transdermal flux (SSTF), permeability coefficient (PC), and drug entrapment efficiency. Graphical and mathematical analysis of the results allowed the identification and quantification of the formulation variables that showed significant effects on the selected responses. Polynomial equations fitted to the data were used to predict the responses in the optimal region. For maximizing the selected responses using a generalized desirability function, an optimum formulation was found to have a maximum side chain length and minimum cholesterol content. Optimized formulation showed highest entrapment of 39.45%, percentages drug permeated through cellulose ester membrane of 3.1 and 28.93 after 0.5 and 8 h, respectively, and SSTF and PC of 152 :g/cm2 h and 0.263 cm/h, respectively, through rabbit skin. These results demonstrated the efficacy of statistical experimental design to unveil the critical formulation interactions and variability affecting the performance of proniosomal formulations. © 2011 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 100:2212–2221, 2011 Keywords: membrane transport; factorial design; gels; percutaneous; surfactants
INTRODUCTION The use of nonionic surfactant vesicles (niosomes) as drug carrier systems has distinct advantages over conventional dosage.1 They can increase the drug efficacy, reduce drug side effects, increase the drug solubility, and develop an effective topical delivery.2 Niosomes have also been studied due to several advantages over liposomes such as lower costs, skin penetration-enhancing properties, and higher chemical stability.3 However, because of the physical instability (aggregation, fusion, and sedimentation of vesicles) on storage, it is more favorable to prepare them as proformulations (proniosomes) that could be hydrated easily using hot water before use or topically applied for transdermal diffusion of drugs.4,5 Correspondence to: Ahmed S. Zidan (Telephone: +202-3795220 (cell), +301-796-0023 (work); Fax: +240-644-3925; E-mail: [email protected], [email protected]) Journal of Pharmaceutical Sciences, Vol. 100, 2212–2221 (2011) © 2011 Wiley-Liss, Inc. and the American Pharmacists Association
2212
Proniosomes offer a versatile vesicle delivery concept and may be a promising carrier for drugs, especially due to their simple production and facile scale-up.6 Hence, it is important to optimize the factors that control the entrapment efficiency (EE) and drug release from proniosomal formulations when used for transdermal, oral, ocular, or injection preparations. Mohammed et al.7 reported that the lipid composition of the vesicles was the most influential factor. The ratio between the nonionic surfactant and cholesterol could affect both the release characteristics and the EE of the incorporated drugs.5 It had been reported that the inclusion of an optimum ratio of surfactant/lecithin in the vesicles played a more important role than cholesterol in modulating drug permeation.6 However, Gregoriadis8 stated that cholesterol was often incorporated within niosomal bilayers to enhance drug loading and retention within the aqueous compartment. Cholesterol could increase the packing densities of phospholipid molecules and complex with the surfactant forming vesicles; accordingly, it had the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
Table 1. Run F1 F2 F3 F4 F5 F6 F7 F8 F9
2213
A 32 Full Factorial Design Layout∗ of Flurbiprofen Proniosomal Patches X1 a
X2 (mM)
Q0.5h (±SD)
Q8h (±SD)
SSTF (:g/cm2 h) (±SD)
PC (cm h−1 ) (±SD)
EE% (±SD)
Span 20 Span 20 Span 20 Span 40 Span 40 Span 40 Span 60 Span 60 Span 60
0.1 0.3 0.5 0.1 0.3 0.5 0.1 0.3 0.5
2.37 ± 0.18 4.19 ± 0.66 2.61 ± 0.45 3.38 ± 1.00 4.62 ± 0.29 2.9 ± 0.58 4.4 ± 0.64 5.78 ± 0.33 3.78 ± 0.30
28.08 ± 0.25 28.77 ± 2.51 22.35 ± 3.09 23.8 ± 1.23 22.56 ± 1.48 16.35 ± 2.34 22.19 ± 0.59 20.67 ± 2.19 15.3 ± 1.46
150.14 ± 52.27 150.39 ± 13.11 116.82 ± 16.15 80.16 ± 10.97 67.17 ± 4.42 48.67 ± 6.97 68.13 ± 17.59 43.93 ± 4.64 32.52 ± 3.1
0.261 ± 0.091 0.26 ± 0.02 0.2 ± 0.03 0.139 ± 0.19 0.12 ± 0.008 0.09 ± 0.01 0.118 ± 0.03 0.076 ± 0.01 0.056 ± 0.005
42.4 ± 2.95 33.9 ± 0.99 34.22 ± 1.15 44.18 ± 2.25 33.06 ± 0.67 31.80 ± 0.75 55.37 ± 0.74 32.82 ± 1.02 34.01 ± 1.49
∗
Data are presented as ±SD; experiments were carried out in triplicate. Amount of total lipid was fixed at 0.9 mM. X1 , type of span (represents span chain length); X2 , cholesterol amount; Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively a
ability to reduce the bilayer permeability to small hydrophilic solutes and ions.9 Moreover, incorporation of cholesterol within vesicles bilayers led to an increase in the hydrophobicity of the interfacial regions of the membranes and could influence the EE of poorly water-soluble drugs.10 The effect of cholesterol incorporation within vesicle membranes on the drug release and EE could differ according to the formulation parameters to form either proniosomes or niosomes. Higher amounts of cholesterol might compete with poorly soluble drugs for packing spaces within the bilayers; hence, excluding the drug as the amphiphiles assemble into vesicles and the EE of the drug decreased.7 On the contrary, Vora et al.5 reported a decreased permeability of levonorgestrel (a poorly water-soluble drug) across rat skin by increasing the cholesterol contents within proniosomes. Flurbiprofen is a derivative of phenylalkanoic acid, a nonsteroidal anti-inflammatory agent; related to ibuprofen in structure; and used in the treatment of rheumatoid arthritis, osteoarthritis, and other rheumatic disorders. The intrinsic solubility of flurbiprofen is 5.0 × 10−5 M. So, it is an ideal model of poorly water-soluble drugs that can be used to study the effect of different factors on drug permeability and encapsulation efficiency using proniosomes and niosomal systems. Different transdermal systems had been reported using flurbiprofen as a model drug to enhance the permeation of poorly water-soluble drugs.11 The objective of this study was to understand and to optimize the effect of cholesterol content and nonionic surfactant chain length (considering spans 20, 40, and 60) on the drug permeation characteristics and EE of the poorly soluble drug flurbiprofen (log P = 4.1) from proniosomal formulations and the formed niosomes after hydration using phosphate buffer (pH 7.4), respectively. A two-factor, three-level randomized (32 ) full factorial design was applied to explore the most influencing factor as well as the interaction among the factors for their effects on the investigated responses. DOI 10.1002/jps
MATERIALS AND METHODS Flurbiprofen was obtained from Egyptian International Pharmaceutical Industries Co., Cairo, Egypt. Sorbitan monolaurate (span 20), sorbitan monopalmitate (span 40), sorbitan monostearate (span 60), sodium azide, and cholesterol (chol > 99%) were purchased from Sigma Chemical Co., St. Louis, Missouri All other chemicals and solvents were of analytical grade and were supplied from El-Nasr Pharmaceutical Chemicals Company, Cairo, Egypt. Preparation of Proniosomes Proniosomes were prepared according to the method reported by Vora et al.5 with some modifications. In glass vials, specified amounts of the surface active agent were mixed with the appropriate amount of cholesterol to make 0.9 mM total lipids (Table 1). Absolute ethanol (about 400 mg) was added to the surfactant/cholesterol mixtures then vials were tightly sealed and warmed in water bath (55–60◦ C) for 5 min while shaking until complete dissolution of cholesterol. To each of the formed transparent solutions, about 0.16 mL hot distilled water (55–60◦ C) was added while warming in the water bath for 3–5 min till a clear or translucent solution was obtained. The mixtures were allowed to cool down to room temperature and were observed for the formation of transparent solution, translucent, or white creamy proniosomal gels. Flurbiprofen (50 mg) was added to the surfactant/cholesterol mixture and dissolved in ethanol while warming in water bath. The obtained formulations were kept in the same closed glass vials in dark for further characterization. Niosomes were prepared by hydration of the proniosomal gels. For this purpose, about 7 mL of Sorensen’s phosphate buffer (pH 7.4) were added into each vial followed by heating for 10 min at a temperature of 60–70◦ C in a water bath while vortexing. The final volume was adjusted to 10 mL by the same buffer. Proniosomal systems of span 20 containing 10% cholesterol JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2214
ZIDAN AND MOKHTAR
formed niosomes upon addition of phosphate buffer and shaking for only 1 min at room temperature. Microscopic Examination The proniosomal gel containing the drug was spread as a thin layer on a glass slide with one drop of water to be examined using light microscope with varied magnification powers (10× and 40×) for niosomal vesicles formation and the presence of drug crystals. EE Determination Entrapment efficiency was determined for each formulation by an exhaustive dialysis technique as described by Udupa et al.15 Unentrapped free drug was removed from the niosomal dispersion by placing 1 mL of the dispersion into a glass cylinder to which a cellulose ester dialysis membrane [molecular weight cutoff (MWCO) 15,000 Da; Spectrum Laboratories, Los Angeles, California] was fitted to one side. Subsequent exhaustive dialysis cycles of 1 h each against 100 mL of 50 mM phosphate buffer (pH 7.4) were performed. The dialysis of free flurbiprofen was completed after about six replacements of buffer solution where no further flurbiprofen could be detected in the solution. The drug content analysis was determined by an in-house developed and validated ultraviolet spectrophotometric analysis (Schimadzu UV-1201, Catalogue Number 206-62409, Schimadzu Corporation, Koyoto, Japan) at 247 nm, using corresponding samples from a blank formulation as a reference. Amount of entrapped drug was obtained by subtracting amount of unentrapped drug from the total drug incorporated.2 EE% =
Actual drug loading × 100 Theoritical drug loading
(1)
Preparation of Rabbit Skin The protocol for the animal study (protocol number 12-2009) was approved by the Animal Study Committee, Zagazig University, Zagazig, Egypt. Albino male rabbits (2–2.5 kg) were used for the permeation study. Hair was removed carefully from the abdominal skin by an electric animal hair clipper without damaging the skin surface. After sacrificing the rabbits by excess chloroform inhalation, the abdominal skin was separated. The skin was stored at −20◦ C until used within 3 days. It was reported by Sebastiani et al.12 that storing the animal skin at −20◦ C could maintain its metabolic activity. Prior to the permeation study, the skin was hydrated by 50 mM phosphate buffer (pH 7.4) containing 0.02% sodium azide as a preservative at 4◦ C over night, and the adipose tissue layer of the skin was removed by rubbing with a cotton swab.13 Permeation Studies On the basis of the preliminary investigations, two permeation studies were performed using either cellulose ester dialysis membrane (MWCO 10 KDa; Spectrum Laboratories) or rabbit skin. Using cellulose ester membranes, the percentages of the drug permeated after 0.5 and 8 h were investigated. On the contrary, the steady-state transdermal flux (SSTF) and permeability coefficient (PC) were calculated from the drug permeation data through rabbit skin. The permeation of flurbiprofen from different proniosomal preparations was performed using a standard dialysis procedure with some modification.14 The experiment included setting up backing membrane reservoir with the formulation (donor compartment) to USP apparatus I having the receptor media. A schematic illustration of the experiment setup is shown in Figure 1.
Figure 1. (a) Niosomal vesicles (batch 1) photomicrographs of Span 60/cholesterol (9:1) containing flurbiprofen as seen under microscope (40×) with no precipitated crystals and (b) schematic illustration of the permeation experiment setup.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
Holders of circular stainless steel cylinders of 2.9 cm inner diameter and 1 mm edge height were used as backing membrane reservoirs for the prepared proniosomal gels. The membrane was mounted at one end of the cylinder, whereas the other end was fitted to another glass cylinder. For skin membranes, skin was mounted with the stratum corneum facing the donor (proniosomal gel loaded circular holders) and the dermal layer facing the receptor compartment. After membrane mounting, the warm proniosomal suspension equivalent to 50 mg/mL of the entrapped drug was poured into the cylinder and immediately sealed to avoid drying then allowed to cool to room temperature. These holders provided uniform spread of the gel through a fixed surface area of 6.61 cm2 exposed to the membrane during the permeation procedures. The backing membrane reservoir was then attached to a hollow glass cylinder (15 cm length and 2.9 cm internal diameter) to be used as the donor compartment. Hundred milliliter of 50 mM phosphate buffer (pH 7.4) containing 0.02% (w/v) sodium azide as preservative and maintained at 37◦ C was used as the receptor medium. Proniosomal suspensions equivalent to 50 mg/mL of the drug were added to the membrane holders and the donor cell was fixed to the shaft of apparatus I of the USP dissolution system. At a stirring rate of 100 rpm, 1 mL aliquots were withdrawn from the receptor compartment at 0.5, 1, 2, 3, 4, 5, 6, and 8 h time intervals and replaced with the same volume of fresh buffer. Samples were analyzed spectrophotometrically at 247 nm using corresponding samples collected from the permeation of placebo systems as references. Each experiment was carried out in triplicate. In vitro permeation rate characteristics such as SSTF and PC for transport of flurbiprofen across rabbit skin were estimated according to the following equations.5 SSTF = PC =
Amount of permeated drug Time × area of permeation membrane
2215
for drug permeation through rabbit skin after 8 h were the dependant variables investigated. Contour and response surface plots were used to visualize the effect of the independent variables. The rank of the investigated variables according to their effects on the responses was elucidated by constructing Pareto charts. Finally, to optimize the levels of the independent variables, a generalized desirability function was applied to maximize the drug entrapment and permeation parameters.
RESULTS AND DISCUSSION Proniosomes could be formulated using a simple procedure with high dissolution power for poorly soluble drugs. Niosomes could be prepared from the formed proniosomes by simple hydration of the gels by warm buffer, where they could be used immediately by different routes of administration. Figure 1 shows the optical microscope image of niosomes formed by simple hydration from proniosomes with no evidence of drug crystals, which indicated complete solubility of the drug in the niosomal formulation. Multivariate Analysis The purpose of proniosomal patch formulation was to produce a gel of a maximum flurbiprofen EE and highest drug flux when applied to rabbit skin. To systematically investigate the different factors affecting the variability controlling the manufacturing of niosomal flurbiprofen transdermal patches, a 32 full factorial design was employed. According to the preliminary experiments, the surfactant chain length, namely Spans 20, 40, and 60, incorporated (X1 ) and the amount of cholesterol (X2 ) were selected as the independent variables. The following equation shows the statistical model incorporating interactive and polynomial terms that were used to evaluate the responses:
(2)
Y = A + Y1 X 1 + Y2 X 2 + Y3 X 1 X 2 + Y4 X 21 + Y5 X 22 (4)
Amount of permeated drug (3) Time × saturated solubility (SS) of drug in receptor fluid
where Y is the dependent variable to be predicted, A is the arithmetic mean response of nine runs; and Y1 and Y2 are the estimated coefficients for the individual effects, X1 and X2 , respectively. The main effects (X1 and X2 ) represented the average result of changing one factor at a time from its low to high value. The interaction term (X1 X2 ) showed how the response changes when two factors were simultaneously changed. The polynomial terms (X1 2 and X2 2 ) were included to investigate colinearity of the independent variables. The dependant variables for the nine batches (F1 to F9) showed a remarkable variability by changing the different combinations of the independent factors. For example, percentage drug permeated through cellulose ester membrane after 0.5 h showed variability from 2.37% to 5.78% at the
Experimental Design A 32 full factorial design was applied to investigate the individual, joint, and polynomial effects of two variables and to optimize the formulation using JMP 7.1 software (SAS, SAS Institute, Cary, North Carolina). The side chain length of the employed nonionic surfactant (X1 ) and the molar ratios of the added amount of cholesterol (X2 ) during proniosomes fabrication were the independent variables (Table 1). The levels of the independent variables were selected from the preliminary experimentation. The drug EE, percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, and SSTF and PC DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2216
ZIDAN AND MOKHTAR
Table 2. Response Q0.5 p value Q8 p value SSTF p value PC p value EE% p value
Summary of the Regression Analysis Results A
y1
y2
y3
y4
y5
3.33 0.0002 34.7 0.0001 184.31 0.0003 0.32 0.017 38.22 0.002
0.04 0.0006 −0.18 0.001 −2.27 0.001 −0.004 0.0003 0.097 0.197
−0.72 0.07 −16.73 0.0012 −83.68 0.014 −0.144 0.028 −34.93 0.01
−0.05 0.04 −0.07 0.4505 −0.143 0.894 −0.0003 0.0017 −0.82 0.107
0.001 0.09 0.0049 0.0247 0.071 0.0148 0.0001 0.00002 0.006 0.318
−40.58 0.0004 −66.38 0.0112 −110.63 0.486 −0.196 0.243 176.75 0.041
Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively.
lowest and the highest chain length and cholesterol content (batches F1 and F8), respectively. On the contrary, the percentages permeated through cellulose ester membrane after 8 h, SSTF and PC for drug permeation through rabbit skin, and EE fluctuated from 15.3%, 32.5 :g/cm2 h, 0.056 cm/h, and 31.8% to 28.77%, 150.14 :g/cm2 h, 0.261 cm/h, and 55.37%, respectively. The mathematical relationships in the form of factor coefficients and their corresponding p values for the measured variables are listed in Table 2. Coefficients with one factor represented the effect of that particular factor, whereas the coefficients with more than one factor represented the interaction between those factors. The confidences that the regression equations would explain the observed values better than the mean for the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE were 98.6%, 97.8%, 96.8%, 96.8%, and 86.5%, respectively. Results in Table 2 showed that surfactant chain length (X1 ) had a significant effect on percentages drug permeated after 0.5 and 8 h, SSTF, and PC. On the contrary, the molar ratio of cholesterol (X2 ) was only insignificant factor for its effect on the percentage drug permeated after 0.5 h. This result could be ascribed to the spontaneous transformation of proniosomes into niosomal vesicles upon hydration; hence, water permeation from the receptor compartment to the skin, drug release from reconstituted vesicles, and the permeation of the dissolved drug occurred very rapidly.16 In addition, the values of X1 coefficients were much smaller than the corresponding coefficients for X2 . This led to the fact that cholesterol was the main retardation factor for drug release from the formed niosomes. Regarding X1 X2 , its effect was significant to the percentage of drug permeated after 0.5 h and nonsignificant to the percentage of drug permeated after 8 h. The polynomial factors (X1 2 and X2 2 ) showed significant influences on the percentage drug permeated after 8 h whereas X1 X2 was insignificant for its influence on this response. The permeation parameters JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
across rabbit skin were significantly affected by the polynomial term of the side chain length and insignificantly affected by the polynomial term of the cholesterol content. The coefficients of X2 2 showed extreme high values than that of X1 2 , which indicated that cholesterol at high concentration retard flurbiprofen release due to increased integrity of the formed niosomal vesicular lamella, and also the increased lipophilicity of lipid membranes to inhibit lipophilic drugs from release.10,19 . In addition, at higher cholesterol concentrations, the effect of surfactant chain length on drug permeation was minimized. The same observation was obtained when considering the SSTF and PC after 8 h. Cholesterol (X2 and X2 2 ) greatly retarding the SSTF and PC more than X1 and X1 X2 ; however, X1 2 showed significant positive effect on flurbiprofen permeation. This result could be explained by the effect of the surfactant on disordering the niosomal membrane and increased permeability.20,21 Taking into consideration the effects of X1 and X2 on EE of flurbiprofen in the formed niosomes, X2 significantly retarded the EE of flurbiprofen due to the thermodynamic molecular configuration of either cholesterol or drug molecules within the packing space in the lipids bilayers; hence, excluding the drug as amphiphiles to assemble into vesicles.22 On the contrary, X2 2 showed a relatively higher positive effect on EE. Bernsdorff et al.10 and Kirby et al.23 explained this observation by the cholesterol contribution to increase the bilayers hydrophobicity and stability of the niosomal membranes, which, in turn, led to efficiently trapping the hydrophobic drug. Analysis of Variance All the quantile–quantile relationships of plotting the measured parameters against the predicted ones yielded linear correlations, with R2 more than >0.95 indicating the validity of the method to predict the investigated dependant variables within the selected design space (Fig. 2). Analysis of variance was applied for estimating the significance of the model at 95% confidence level. A regression model was considered significant if the p value, expressed as DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
2217
Figure 2. Quantile–quantile plots for predicting the dependant variables. Table 3. Summary of Analysis of Variance Analysis for testing the Model in Portions
Q0.5 Q8 SSTF PC EE%
DF
SS
MS
Prob > F
R2
5 5 5 5 5
9.5 163.26 15,726 0.047 470.93
1.9 32.65 3145.2 0.0094 94.18
0.0012 0.0025 0.0043 0.0043 0.0368
0.9949 0.9918 0.9882 0.9881 0.9495
Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively; DF, degree of freedom; SS, sum of squares; MS, mean of squares
“Prob > F” value, was less than 0.05. In addition, graphical analysis of responses was carried out to allow pointing out the important factors for the considered responses and the optimum factor level to be selected. In particular, the active factors were those where a level change determined a response variation that was statistically different from the variation due to the experimental error.17,18 Prob > F values of 0.0012, 0.0025, 0.0043, 0.0043, and 0.0368 for the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE, respectively, were obtained to indicate the significant effects of the independent factors on the selected parameters (Table 3). Response Surface and Contour Plots Contour and response surface plots were further applied to explore the effects of the independent factors DOI 10.1002/jps
on the responses. The effects of the incorporated surfactant chain length and the amount of cholesterol on the investigated responses are given in Figure 3. At low and high fatty acid chain lengths, the EE increased by increasing the amount incorporated of cholesterol. On the contrary, the synergistic effect of fatty acid chain length on EE was more pronounced at lower cholesterol level rather than higher cholesterol level. This could be explained by two opposing mechanisms. The first mechanism assumed that increasing cholesterol level within niosomal lamella led to an increase in the bilayer hydrophobicity and stability10,19 and a decrease in drug permeability.23 These results led to efficiently trapping the hydrophobic drug within the bilayers as vesicles formed. On the contrary, the second mechanism proposed that higher amounts of cholesterol might compete with the drug for packing space within the bilayers; hence, decreasing the entrapment capacity of the drug within the vesicles.7 The effects of the independent factors were more pronounced on the amount of the drug permeated after 0.5 h rather than 8 h. The effect of cholesterol on the burst permeation of the drug at either low or high fatty acid chain length had a biphasic characteristic. At both short and long fatty acid side chains, the percentage of the drug permeated after 0.5 h increased then decreased by increasing the amount incorporated of cholesterol from 0.1 to 0.5 mM. This behavior was not the case for the effect of fatty acid chain length on the burst drug permeation. Increasing the fatty acid chain length at either low or high cholesterol levels led to an increase in the burst JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2218
ZIDAN AND MOKHTAR
Figure 3. Three-dimensional response surface and counter plots showing the effect of the surfactant chain length incorporated (X1 ) and the amount of cholesterol (X2 ) on the dependant variables of flurbiprofen proniosomes.
effect. This result could be ascribed to two factors: (i) the presence of free drug adsorbed at the niosomal surface and (ii) the increased steric geometrical configuration of the amphiphilic molecules as the side chain increased, which interfered with flurbiprofen packing into the niosomal membranes, hence initial burst of the drug occurred. The effects of X1 and X2 on the SSTF and PC were more or less similar. At a cholesterol level of 0.1 mM, SSTF and PC decreased from 151.8 to 62.8 :g/cm2 h and from 0.263 to 0.11 cm/ h by increasing the fatty acid chain length from span 20 to 60, respectively. On the contrary, SSTF and PC decreased from 153.8 to 120.18 :g/cm2 h and from 0.263 to 0.209 cm/h by increasing the amount incorporated of cholesterol from 0.1 to 0.5 mM at low fatty acid chain length, respectively. This result could be due to the fact that increasing the surfactant chain length was accompanied by an increase in the transition temperature of the surfactant; hence, at 37◦ C, the surfactant vesicles were in the ordered gel state, which decreased the drug leakage and permeation.24 On the contrary, addition of cholesterol appeared to disrupt the ordered array of hydrocarbon chains in the gel phase.25 Below the transition temperature of a surfactant, cholesterol disrupted the membrane packing, whereas above transition temperature cholesterol assisted the niosomal membranes packing and less leakiness.26
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
Interaction Plots The main effects of the independent variables on the investigated responses were further elucidated using interaction plots (Fig. 4). At high level of cholesterol content, the effect of changing the fatty acid chain length of the nonionic surfactant on the EE was negligible. On the contrary, minimal effect of changing the cholesterol level was observed on the initial burst effect at low fatty acid chain length. These observations could demonstrate that the presence of cholesterol was the most influential factor to increase niosomal bilayer cohesion. The interaction of cholesterol with the surfactant could result in a dramatic increase in the area of expansion modulus and strength of the bilayer membrane.27 However, owing to spontaneous and rapid niosomal formation from proniosomes and as flurbiprofen was added in excess of what could be entrapped, burst release was observed at both low and high cholesterol concentrations.
DESIRABILITY FUNCTION The concept of desirability scale arose when there was a need to combine the magnitude of several characteristics to a dimensionless scale (say, d). Generally, this desirability function was so constructed that any property (characteristic) value of a product was mapped between zero and one called
DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
2219
Figure 4. Interaction plots showing the quadratic effects of interactions between the independent variables on the investigated responses.
“individualized desirability.” Desirability scale value of one meant the optimum property level of the product or service, whereas zero desirability indicated an unacceptable product. Any value in between zero and one gave an opportunity to improve the product quality. Harrington28 first defined the desirability function, a slight modification of which was done by Gatza and McMillan29 for which the desirability scale could be negative for negative input values. The individual desirabilities were then combined using the geometric mean, which gave the overall generalized desirability D as follows: D = [d1 (Y1 ) × d2 (Y2 ) × · · · × dk(Yk)]1/k
(5)
where k denoting the number of responses. Notice that if any response Yi was completely undesirable [di (Yi ) = 0), then the overall desirability was zero. In practice, fitted response values “i” were used in place of the Yi . Taking into consideration the effect of the independent variables on the studied parameters, the levels of these factors were determined using the generalized desirability function to maximize all the investigated responses (Fig. 5). The predicted values of the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE were 3.06%, 28.9%, 152.07 :g/cm2 h, 0.263 cm/h, and 39.4%, respectively, at X1 and X2 levels of span 20 and 0.14 mM, respectively. As a confirmation process, a fresh formulation was prepared DOI 10.1002/jps
with the optimized values of the independent variables that yielded the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE of 2.49%, 27.4%, 147.23 :g/cm2 h, 0.285 cm/h, and 40.8%, respectively. The observed values were not significantly different (p < 0.05) from the predicted values with standardized residuals for each response not exceeding 4% of the nominal value demonstrating the feasibility of quality by design approach to understand the performance of flurbiprofen proniosomal formulations.
CONCLUSION The present study showed the suitability of proniosomes as carriers for transdermal delivery of poorly water-soluble drugs such as flurbiprofen. As demonstrated by PCs and EE, proniosomes offered systems with high drug loading and skin permeation. The fatty acid side chain length was the most important factor controlling the permeation and EE characteristics. In addition, a maximum permeation value and a minimum EE were obtained at shorter fatty acid chain length of spans and minimum amount of cholesterol. Finally, through the rigorous analysis of the two independent variables and its coeffects on the investigated responses, this study demonstrated the potential of quality by design in developing proniosomal gels as drug carrier systems. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2220
ZIDAN AND MOKHTAR
Figure 5. Generalized desirability profiler to maximize all the investigated responses.
DISCLAIMER The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any Agency determination or policy.
REFERENCES 1. Schreier H, Bouwstra J. 1994. Liposomes and niosomes as topical drug carriers: Dermal and transdermal drug delivery. J Control Release 30:1–15. 2. Aggarwal D, Kaur IP. 2005. Improved pharmacodynamics of timolol maleate from a mucoadhesive niosomal ophthalmic drug delivery system. Int J Pharm 290:155–159 3. Manconi M, Sinico C, Valenti D, Loy G, Fadda AM. 2002. Niosomes as carriers for tretinoin. I. Preparation and properties. Int J Pharm 234:237–248. 4. Hu C, Rhodes DG. 2000. Proniosomes: A novel drug carrier preparation. Int J Pharm 206(1–2):110–122. 5. Vora B, Khopade AJ, Jain NK. 1998. Proniosome based transdermal delivery of levonorgestrel for effective contraception. J Control Release 54:149–165. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
6. Alsarra IA, Bosela AA, Ahmed SM, Mahrous GM. 2005. Proniosomes as a drug carrier for transdermal delivery of ketorolac. Eur J Pharm Biopharm 59(3):585–490. 7. Mohammed AR, Weston N, Coombes AGA, Fitzgerald M, Perrie Y. 2004. Liposome formulation of poorly water soluble drugs: Optimization of drug loading and ESEM analysis of stability. Int J Pharm 285:23–34. 8. Gregoriadis G. 1993. Liposome technology, 2nd ed. Boca Raton, Florida:CRC Press. 9. Gregoriadis G, Perrie Y. 2010. Liposomes. In Encyclopedia of Life Sciences (ELS); Chichester: John Wiley & Sons, Ltd. . . 10. Bernsdorff C, Wolff A, Winter R, Gratton E. 1997. Effect of hydrostatic pressure on water penetration and rotational dynamics in phospholipids–cholesterol bilayers. Biophys J 72:1264–1277. 11. Charoo NA, Shamsher AA, Kohli K, Pillai K, Rahman Z. 2008. Improvement in bioavailability of transdermally applied flurbiprofen using tulsi (Ocimum sanctum) and turpentine oil. Colloids Surf B Biointerfaces 65(2):300–307. 12. Sebastiani P, Nicoli S, Santi P. 2005. Effect of lactic acid and iontophoresis on drug permeation across rabbit ear skin. Int J Pharm 292:119–126. 13. Larrucea E, Arellano A, Santoyo S, Ygarrua P. 2001. Interaction of tenoxicam with cyclodextrins and its influence on the in vitro percutaneous penetration of the drug. Drug Dev Ind Pharm 27(3):25l-260. 14. Mura P, Faucci MT, Bramanti G, Corti P. 2000. Evaluation of transcutol as a clonazepam transdermal permeation enhancer from hydrophilic gel formulations. Eur J Pharm Sci 9:365–372. 15. Udupa N, Chandraprakash KS, Umadevi P, Pillai GK. 1993. Formulation and evaluation of methotrexate niosomes. Drug Dev Ind Pharm 19:1331–1342. 16. Hwang BY, Jung BH, Chung SJ, Lee MH, Shim CK. 1997. In vitro skin permeation of nicotine from proliposomes. J Contr Release 49:177–184. 17. Furlanetto S, Maestrelli F, Orlandini S, Mura P. 2003. Optimization of dissolution test precision for a ketoprofen oral extended-release product. J Pharm Biomed Anal 32:159– 165. 18. Furlanetto S, Cirri M, Maestrelli F, Corti G, Mura P. 2006. Study of formulation variables influencing the drug release rate from matrix tablet by experimental design. Eur J Pharm Biopharm 62:77–84. 19. Gregoriadis G, Davis C. 1979. Stability of liposomes in vivo and in vitro is promoted by their cholesterol content in the presence of blood cells. Biochem Biophys Res Commun 90:1287–1293. 20. Yoshioka T Sternberg B, Florence AT. 1994. Preparation and properties of vesicles (niosomes) of sorbitan monoesters (Span 20, 40, 60, and 80) and sorbitan triester (Span 85) Int J Pharm 105:1–6. 21. Hao Y, Zhao F, Li N, Yang Y, Li K. 2002. Studies on a high encapsulation of cholchicine by niosome system. Int J Pharm 244:73–80. 22. Mokhtar M, Sammour OA, Hammad MA, Megrab NA. 2008. Effect of some formulation parameters on flurbiprofen encapsulation and release rates of niosomes prepared from proniosomes. Int J Pharm 361:104–111. 23. Kirby C, Clarke J, Gregoriadis G. 1980. Effect of the cholesterol content of small unilamellar liposomes on their stability in vivo and in vitro. Biochem J 186:591–598. 24. Uchegbu IF, Vyas SP. 1998. Non-ionic surfactant based vesicles (niosomes) in drug delivery. Int J Pharm 172:33–70. 25. Ladbrooke BD, Williams RM, Chapman D. 1968. Studies on lecithin–cholesterol–water interactions by differential scanning calorimetry and X-ray diffraction. Biochim Biophys Acta 150:333–340. 26. Papahadjopoulos D, Jacobson K, Nir S, Isac T. 1973. Phase transition in phospholipid vesicles: Fluorescence polarization DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
and permeability measurements concerning the effect of temperature and cholesterol. Biochim Biophys Acta 311:330– 334. 27. David N, Rashmi SN. 1990. Elastic deformation and failure of lipid bilayer membranes containing cholesterol. Biophys J 58:997–1009.
DOI 10.1002/jps
2221
28. Harrington, Jr. EC. 1965. “The Desirability Function.” Industrial Quality Control, Volume 21, No. 10, pp. 494-498. 29. Gatza PE, McMillan RC. 1972. The use of experimental design and computerized data analysis in elastomer development studies, Division of Rubber Chemistry, American Chemical Society Fall Meeting, Cincinnati, Ohio, pp 3–6.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
Department of Pharmaceutics and Industrial Pharmacy, Faculty of Pharmacy, Zagazig University, Zagazig, Egypt
2
Division of Product Quality and Research, Center for Drug Evaluation and Research, Food and Drug Administration, Maryland, USA
Received 9 August 2010; revised 10 November 2010; accepted 25 November 2010 Published online 21 January 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.22453 ABSTRACT: Flurbiprofen was formulated as a proniosomal transdermal gel with high drug loading (55.4%, w/w), using a series of nonionic surfactant and cholesterol. A two-factor, threelevel randomized full factorial strategy was developed to optimize simultaneously the effect of surfactant fatty acid side chain length and the amount of cholesterol on the properties of the proniosomes, namely drug permeation characteristics such as steady-state transdermal flux (SSTF), permeability coefficient (PC), and drug entrapment efficiency. Graphical and mathematical analysis of the results allowed the identification and quantification of the formulation variables that showed significant effects on the selected responses. Polynomial equations fitted to the data were used to predict the responses in the optimal region. For maximizing the selected responses using a generalized desirability function, an optimum formulation was found to have a maximum side chain length and minimum cholesterol content. Optimized formulation showed highest entrapment of 39.45%, percentages drug permeated through cellulose ester membrane of 3.1 and 28.93 after 0.5 and 8 h, respectively, and SSTF and PC of 152 :g/cm2 h and 0.263 cm/h, respectively, through rabbit skin. These results demonstrated the efficacy of statistical experimental design to unveil the critical formulation interactions and variability affecting the performance of proniosomal formulations. © 2011 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 100:2212–2221, 2011 Keywords: membrane transport; factorial design; gels; percutaneous; surfactants
INTRODUCTION The use of nonionic surfactant vesicles (niosomes) as drug carrier systems has distinct advantages over conventional dosage.1 They can increase the drug efficacy, reduce drug side effects, increase the drug solubility, and develop an effective topical delivery.2 Niosomes have also been studied due to several advantages over liposomes such as lower costs, skin penetration-enhancing properties, and higher chemical stability.3 However, because of the physical instability (aggregation, fusion, and sedimentation of vesicles) on storage, it is more favorable to prepare them as proformulations (proniosomes) that could be hydrated easily using hot water before use or topically applied for transdermal diffusion of drugs.4,5 Correspondence to: Ahmed S. Zidan (Telephone: +202-3795220 (cell), +301-796-0023 (work); Fax: +240-644-3925; E-mail: [email protected], [email protected]) Journal of Pharmaceutical Sciences, Vol. 100, 2212–2221 (2011) © 2011 Wiley-Liss, Inc. and the American Pharmacists Association
2212
Proniosomes offer a versatile vesicle delivery concept and may be a promising carrier for drugs, especially due to their simple production and facile scale-up.6 Hence, it is important to optimize the factors that control the entrapment efficiency (EE) and drug release from proniosomal formulations when used for transdermal, oral, ocular, or injection preparations. Mohammed et al.7 reported that the lipid composition of the vesicles was the most influential factor. The ratio between the nonionic surfactant and cholesterol could affect both the release characteristics and the EE of the incorporated drugs.5 It had been reported that the inclusion of an optimum ratio of surfactant/lecithin in the vesicles played a more important role than cholesterol in modulating drug permeation.6 However, Gregoriadis8 stated that cholesterol was often incorporated within niosomal bilayers to enhance drug loading and retention within the aqueous compartment. Cholesterol could increase the packing densities of phospholipid molecules and complex with the surfactant forming vesicles; accordingly, it had the
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
Table 1. Run F1 F2 F3 F4 F5 F6 F7 F8 F9
2213
A 32 Full Factorial Design Layout∗ of Flurbiprofen Proniosomal Patches X1 a
X2 (mM)
Q0.5h (±SD)
Q8h (±SD)
SSTF (:g/cm2 h) (±SD)
PC (cm h−1 ) (±SD)
EE% (±SD)
Span 20 Span 20 Span 20 Span 40 Span 40 Span 40 Span 60 Span 60 Span 60
0.1 0.3 0.5 0.1 0.3 0.5 0.1 0.3 0.5
2.37 ± 0.18 4.19 ± 0.66 2.61 ± 0.45 3.38 ± 1.00 4.62 ± 0.29 2.9 ± 0.58 4.4 ± 0.64 5.78 ± 0.33 3.78 ± 0.30
28.08 ± 0.25 28.77 ± 2.51 22.35 ± 3.09 23.8 ± 1.23 22.56 ± 1.48 16.35 ± 2.34 22.19 ± 0.59 20.67 ± 2.19 15.3 ± 1.46
150.14 ± 52.27 150.39 ± 13.11 116.82 ± 16.15 80.16 ± 10.97 67.17 ± 4.42 48.67 ± 6.97 68.13 ± 17.59 43.93 ± 4.64 32.52 ± 3.1
0.261 ± 0.091 0.26 ± 0.02 0.2 ± 0.03 0.139 ± 0.19 0.12 ± 0.008 0.09 ± 0.01 0.118 ± 0.03 0.076 ± 0.01 0.056 ± 0.005
42.4 ± 2.95 33.9 ± 0.99 34.22 ± 1.15 44.18 ± 2.25 33.06 ± 0.67 31.80 ± 0.75 55.37 ± 0.74 32.82 ± 1.02 34.01 ± 1.49
∗
Data are presented as ±SD; experiments were carried out in triplicate. Amount of total lipid was fixed at 0.9 mM. X1 , type of span (represents span chain length); X2 , cholesterol amount; Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively a
ability to reduce the bilayer permeability to small hydrophilic solutes and ions.9 Moreover, incorporation of cholesterol within vesicles bilayers led to an increase in the hydrophobicity of the interfacial regions of the membranes and could influence the EE of poorly water-soluble drugs.10 The effect of cholesterol incorporation within vesicle membranes on the drug release and EE could differ according to the formulation parameters to form either proniosomes or niosomes. Higher amounts of cholesterol might compete with poorly soluble drugs for packing spaces within the bilayers; hence, excluding the drug as the amphiphiles assemble into vesicles and the EE of the drug decreased.7 On the contrary, Vora et al.5 reported a decreased permeability of levonorgestrel (a poorly water-soluble drug) across rat skin by increasing the cholesterol contents within proniosomes. Flurbiprofen is a derivative of phenylalkanoic acid, a nonsteroidal anti-inflammatory agent; related to ibuprofen in structure; and used in the treatment of rheumatoid arthritis, osteoarthritis, and other rheumatic disorders. The intrinsic solubility of flurbiprofen is 5.0 × 10−5 M. So, it is an ideal model of poorly water-soluble drugs that can be used to study the effect of different factors on drug permeability and encapsulation efficiency using proniosomes and niosomal systems. Different transdermal systems had been reported using flurbiprofen as a model drug to enhance the permeation of poorly water-soluble drugs.11 The objective of this study was to understand and to optimize the effect of cholesterol content and nonionic surfactant chain length (considering spans 20, 40, and 60) on the drug permeation characteristics and EE of the poorly soluble drug flurbiprofen (log P = 4.1) from proniosomal formulations and the formed niosomes after hydration using phosphate buffer (pH 7.4), respectively. A two-factor, three-level randomized (32 ) full factorial design was applied to explore the most influencing factor as well as the interaction among the factors for their effects on the investigated responses. DOI 10.1002/jps
MATERIALS AND METHODS Flurbiprofen was obtained from Egyptian International Pharmaceutical Industries Co., Cairo, Egypt. Sorbitan monolaurate (span 20), sorbitan monopalmitate (span 40), sorbitan monostearate (span 60), sodium azide, and cholesterol (chol > 99%) were purchased from Sigma Chemical Co., St. Louis, Missouri All other chemicals and solvents were of analytical grade and were supplied from El-Nasr Pharmaceutical Chemicals Company, Cairo, Egypt. Preparation of Proniosomes Proniosomes were prepared according to the method reported by Vora et al.5 with some modifications. In glass vials, specified amounts of the surface active agent were mixed with the appropriate amount of cholesterol to make 0.9 mM total lipids (Table 1). Absolute ethanol (about 400 mg) was added to the surfactant/cholesterol mixtures then vials were tightly sealed and warmed in water bath (55–60◦ C) for 5 min while shaking until complete dissolution of cholesterol. To each of the formed transparent solutions, about 0.16 mL hot distilled water (55–60◦ C) was added while warming in the water bath for 3–5 min till a clear or translucent solution was obtained. The mixtures were allowed to cool down to room temperature and were observed for the formation of transparent solution, translucent, or white creamy proniosomal gels. Flurbiprofen (50 mg) was added to the surfactant/cholesterol mixture and dissolved in ethanol while warming in water bath. The obtained formulations were kept in the same closed glass vials in dark for further characterization. Niosomes were prepared by hydration of the proniosomal gels. For this purpose, about 7 mL of Sorensen’s phosphate buffer (pH 7.4) were added into each vial followed by heating for 10 min at a temperature of 60–70◦ C in a water bath while vortexing. The final volume was adjusted to 10 mL by the same buffer. Proniosomal systems of span 20 containing 10% cholesterol JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2214
ZIDAN AND MOKHTAR
formed niosomes upon addition of phosphate buffer and shaking for only 1 min at room temperature. Microscopic Examination The proniosomal gel containing the drug was spread as a thin layer on a glass slide with one drop of water to be examined using light microscope with varied magnification powers (10× and 40×) for niosomal vesicles formation and the presence of drug crystals. EE Determination Entrapment efficiency was determined for each formulation by an exhaustive dialysis technique as described by Udupa et al.15 Unentrapped free drug was removed from the niosomal dispersion by placing 1 mL of the dispersion into a glass cylinder to which a cellulose ester dialysis membrane [molecular weight cutoff (MWCO) 15,000 Da; Spectrum Laboratories, Los Angeles, California] was fitted to one side. Subsequent exhaustive dialysis cycles of 1 h each against 100 mL of 50 mM phosphate buffer (pH 7.4) were performed. The dialysis of free flurbiprofen was completed after about six replacements of buffer solution where no further flurbiprofen could be detected in the solution. The drug content analysis was determined by an in-house developed and validated ultraviolet spectrophotometric analysis (Schimadzu UV-1201, Catalogue Number 206-62409, Schimadzu Corporation, Koyoto, Japan) at 247 nm, using corresponding samples from a blank formulation as a reference. Amount of entrapped drug was obtained by subtracting amount of unentrapped drug from the total drug incorporated.2 EE% =
Actual drug loading × 100 Theoritical drug loading
(1)
Preparation of Rabbit Skin The protocol for the animal study (protocol number 12-2009) was approved by the Animal Study Committee, Zagazig University, Zagazig, Egypt. Albino male rabbits (2–2.5 kg) were used for the permeation study. Hair was removed carefully from the abdominal skin by an electric animal hair clipper without damaging the skin surface. After sacrificing the rabbits by excess chloroform inhalation, the abdominal skin was separated. The skin was stored at −20◦ C until used within 3 days. It was reported by Sebastiani et al.12 that storing the animal skin at −20◦ C could maintain its metabolic activity. Prior to the permeation study, the skin was hydrated by 50 mM phosphate buffer (pH 7.4) containing 0.02% sodium azide as a preservative at 4◦ C over night, and the adipose tissue layer of the skin was removed by rubbing with a cotton swab.13 Permeation Studies On the basis of the preliminary investigations, two permeation studies were performed using either cellulose ester dialysis membrane (MWCO 10 KDa; Spectrum Laboratories) or rabbit skin. Using cellulose ester membranes, the percentages of the drug permeated after 0.5 and 8 h were investigated. On the contrary, the steady-state transdermal flux (SSTF) and permeability coefficient (PC) were calculated from the drug permeation data through rabbit skin. The permeation of flurbiprofen from different proniosomal preparations was performed using a standard dialysis procedure with some modification.14 The experiment included setting up backing membrane reservoir with the formulation (donor compartment) to USP apparatus I having the receptor media. A schematic illustration of the experiment setup is shown in Figure 1.
Figure 1. (a) Niosomal vesicles (batch 1) photomicrographs of Span 60/cholesterol (9:1) containing flurbiprofen as seen under microscope (40×) with no precipitated crystals and (b) schematic illustration of the permeation experiment setup.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
Holders of circular stainless steel cylinders of 2.9 cm inner diameter and 1 mm edge height were used as backing membrane reservoirs for the prepared proniosomal gels. The membrane was mounted at one end of the cylinder, whereas the other end was fitted to another glass cylinder. For skin membranes, skin was mounted with the stratum corneum facing the donor (proniosomal gel loaded circular holders) and the dermal layer facing the receptor compartment. After membrane mounting, the warm proniosomal suspension equivalent to 50 mg/mL of the entrapped drug was poured into the cylinder and immediately sealed to avoid drying then allowed to cool to room temperature. These holders provided uniform spread of the gel through a fixed surface area of 6.61 cm2 exposed to the membrane during the permeation procedures. The backing membrane reservoir was then attached to a hollow glass cylinder (15 cm length and 2.9 cm internal diameter) to be used as the donor compartment. Hundred milliliter of 50 mM phosphate buffer (pH 7.4) containing 0.02% (w/v) sodium azide as preservative and maintained at 37◦ C was used as the receptor medium. Proniosomal suspensions equivalent to 50 mg/mL of the drug were added to the membrane holders and the donor cell was fixed to the shaft of apparatus I of the USP dissolution system. At a stirring rate of 100 rpm, 1 mL aliquots were withdrawn from the receptor compartment at 0.5, 1, 2, 3, 4, 5, 6, and 8 h time intervals and replaced with the same volume of fresh buffer. Samples were analyzed spectrophotometrically at 247 nm using corresponding samples collected from the permeation of placebo systems as references. Each experiment was carried out in triplicate. In vitro permeation rate characteristics such as SSTF and PC for transport of flurbiprofen across rabbit skin were estimated according to the following equations.5 SSTF = PC =
Amount of permeated drug Time × area of permeation membrane
2215
for drug permeation through rabbit skin after 8 h were the dependant variables investigated. Contour and response surface plots were used to visualize the effect of the independent variables. The rank of the investigated variables according to their effects on the responses was elucidated by constructing Pareto charts. Finally, to optimize the levels of the independent variables, a generalized desirability function was applied to maximize the drug entrapment and permeation parameters.
RESULTS AND DISCUSSION Proniosomes could be formulated using a simple procedure with high dissolution power for poorly soluble drugs. Niosomes could be prepared from the formed proniosomes by simple hydration of the gels by warm buffer, where they could be used immediately by different routes of administration. Figure 1 shows the optical microscope image of niosomes formed by simple hydration from proniosomes with no evidence of drug crystals, which indicated complete solubility of the drug in the niosomal formulation. Multivariate Analysis The purpose of proniosomal patch formulation was to produce a gel of a maximum flurbiprofen EE and highest drug flux when applied to rabbit skin. To systematically investigate the different factors affecting the variability controlling the manufacturing of niosomal flurbiprofen transdermal patches, a 32 full factorial design was employed. According to the preliminary experiments, the surfactant chain length, namely Spans 20, 40, and 60, incorporated (X1 ) and the amount of cholesterol (X2 ) were selected as the independent variables. The following equation shows the statistical model incorporating interactive and polynomial terms that were used to evaluate the responses:
(2)
Y = A + Y1 X 1 + Y2 X 2 + Y3 X 1 X 2 + Y4 X 21 + Y5 X 22 (4)
Amount of permeated drug (3) Time × saturated solubility (SS) of drug in receptor fluid
where Y is the dependent variable to be predicted, A is the arithmetic mean response of nine runs; and Y1 and Y2 are the estimated coefficients for the individual effects, X1 and X2 , respectively. The main effects (X1 and X2 ) represented the average result of changing one factor at a time from its low to high value. The interaction term (X1 X2 ) showed how the response changes when two factors were simultaneously changed. The polynomial terms (X1 2 and X2 2 ) were included to investigate colinearity of the independent variables. The dependant variables for the nine batches (F1 to F9) showed a remarkable variability by changing the different combinations of the independent factors. For example, percentage drug permeated through cellulose ester membrane after 0.5 h showed variability from 2.37% to 5.78% at the
Experimental Design A 32 full factorial design was applied to investigate the individual, joint, and polynomial effects of two variables and to optimize the formulation using JMP 7.1 software (SAS, SAS Institute, Cary, North Carolina). The side chain length of the employed nonionic surfactant (X1 ) and the molar ratios of the added amount of cholesterol (X2 ) during proniosomes fabrication were the independent variables (Table 1). The levels of the independent variables were selected from the preliminary experimentation. The drug EE, percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, and SSTF and PC DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2216
ZIDAN AND MOKHTAR
Table 2. Response Q0.5 p value Q8 p value SSTF p value PC p value EE% p value
Summary of the Regression Analysis Results A
y1
y2
y3
y4
y5
3.33 0.0002 34.7 0.0001 184.31 0.0003 0.32 0.017 38.22 0.002
0.04 0.0006 −0.18 0.001 −2.27 0.001 −0.004 0.0003 0.097 0.197
−0.72 0.07 −16.73 0.0012 −83.68 0.014 −0.144 0.028 −34.93 0.01
−0.05 0.04 −0.07 0.4505 −0.143 0.894 −0.0003 0.0017 −0.82 0.107
0.001 0.09 0.0049 0.0247 0.071 0.0148 0.0001 0.00002 0.006 0.318
−40.58 0.0004 −66.38 0.0112 −110.63 0.486 −0.196 0.243 176.75 0.041
Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively.
lowest and the highest chain length and cholesterol content (batches F1 and F8), respectively. On the contrary, the percentages permeated through cellulose ester membrane after 8 h, SSTF and PC for drug permeation through rabbit skin, and EE fluctuated from 15.3%, 32.5 :g/cm2 h, 0.056 cm/h, and 31.8% to 28.77%, 150.14 :g/cm2 h, 0.261 cm/h, and 55.37%, respectively. The mathematical relationships in the form of factor coefficients and their corresponding p values for the measured variables are listed in Table 2. Coefficients with one factor represented the effect of that particular factor, whereas the coefficients with more than one factor represented the interaction between those factors. The confidences that the regression equations would explain the observed values better than the mean for the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE were 98.6%, 97.8%, 96.8%, 96.8%, and 86.5%, respectively. Results in Table 2 showed that surfactant chain length (X1 ) had a significant effect on percentages drug permeated after 0.5 and 8 h, SSTF, and PC. On the contrary, the molar ratio of cholesterol (X2 ) was only insignificant factor for its effect on the percentage drug permeated after 0.5 h. This result could be ascribed to the spontaneous transformation of proniosomes into niosomal vesicles upon hydration; hence, water permeation from the receptor compartment to the skin, drug release from reconstituted vesicles, and the permeation of the dissolved drug occurred very rapidly.16 In addition, the values of X1 coefficients were much smaller than the corresponding coefficients for X2 . This led to the fact that cholesterol was the main retardation factor for drug release from the formed niosomes. Regarding X1 X2 , its effect was significant to the percentage of drug permeated after 0.5 h and nonsignificant to the percentage of drug permeated after 8 h. The polynomial factors (X1 2 and X2 2 ) showed significant influences on the percentage drug permeated after 8 h whereas X1 X2 was insignificant for its influence on this response. The permeation parameters JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
across rabbit skin were significantly affected by the polynomial term of the side chain length and insignificantly affected by the polynomial term of the cholesterol content. The coefficients of X2 2 showed extreme high values than that of X1 2 , which indicated that cholesterol at high concentration retard flurbiprofen release due to increased integrity of the formed niosomal vesicular lamella, and also the increased lipophilicity of lipid membranes to inhibit lipophilic drugs from release.10,19 . In addition, at higher cholesterol concentrations, the effect of surfactant chain length on drug permeation was minimized. The same observation was obtained when considering the SSTF and PC after 8 h. Cholesterol (X2 and X2 2 ) greatly retarding the SSTF and PC more than X1 and X1 X2 ; however, X1 2 showed significant positive effect on flurbiprofen permeation. This result could be explained by the effect of the surfactant on disordering the niosomal membrane and increased permeability.20,21 Taking into consideration the effects of X1 and X2 on EE of flurbiprofen in the formed niosomes, X2 significantly retarded the EE of flurbiprofen due to the thermodynamic molecular configuration of either cholesterol or drug molecules within the packing space in the lipids bilayers; hence, excluding the drug as amphiphiles to assemble into vesicles.22 On the contrary, X2 2 showed a relatively higher positive effect on EE. Bernsdorff et al.10 and Kirby et al.23 explained this observation by the cholesterol contribution to increase the bilayers hydrophobicity and stability of the niosomal membranes, which, in turn, led to efficiently trapping the hydrophobic drug. Analysis of Variance All the quantile–quantile relationships of plotting the measured parameters against the predicted ones yielded linear correlations, with R2 more than >0.95 indicating the validity of the method to predict the investigated dependant variables within the selected design space (Fig. 2). Analysis of variance was applied for estimating the significance of the model at 95% confidence level. A regression model was considered significant if the p value, expressed as DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
2217
Figure 2. Quantile–quantile plots for predicting the dependant variables. Table 3. Summary of Analysis of Variance Analysis for testing the Model in Portions
Q0.5 Q8 SSTF PC EE%
DF
SS
MS
Prob > F
R2
5 5 5 5 5
9.5 163.26 15,726 0.047 470.93
1.9 32.65 3145.2 0.0094 94.18
0.0012 0.0025 0.0043 0.0043 0.0368
0.9949 0.9918 0.9882 0.9881 0.9495
Q0.5 and Q8 , percentage of drug permeated through cellulose ester membrane after 0.5 and 8 h, respectively; SSTF and PC, steady-state transdermal flux and permeability coefficient for drug permeation through rabbit skin after 8 h, respectively; DF, degree of freedom; SS, sum of squares; MS, mean of squares
“Prob > F” value, was less than 0.05. In addition, graphical analysis of responses was carried out to allow pointing out the important factors for the considered responses and the optimum factor level to be selected. In particular, the active factors were those where a level change determined a response variation that was statistically different from the variation due to the experimental error.17,18 Prob > F values of 0.0012, 0.0025, 0.0043, 0.0043, and 0.0368 for the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE, respectively, were obtained to indicate the significant effects of the independent factors on the selected parameters (Table 3). Response Surface and Contour Plots Contour and response surface plots were further applied to explore the effects of the independent factors DOI 10.1002/jps
on the responses. The effects of the incorporated surfactant chain length and the amount of cholesterol on the investigated responses are given in Figure 3. At low and high fatty acid chain lengths, the EE increased by increasing the amount incorporated of cholesterol. On the contrary, the synergistic effect of fatty acid chain length on EE was more pronounced at lower cholesterol level rather than higher cholesterol level. This could be explained by two opposing mechanisms. The first mechanism assumed that increasing cholesterol level within niosomal lamella led to an increase in the bilayer hydrophobicity and stability10,19 and a decrease in drug permeability.23 These results led to efficiently trapping the hydrophobic drug within the bilayers as vesicles formed. On the contrary, the second mechanism proposed that higher amounts of cholesterol might compete with the drug for packing space within the bilayers; hence, decreasing the entrapment capacity of the drug within the vesicles.7 The effects of the independent factors were more pronounced on the amount of the drug permeated after 0.5 h rather than 8 h. The effect of cholesterol on the burst permeation of the drug at either low or high fatty acid chain length had a biphasic characteristic. At both short and long fatty acid side chains, the percentage of the drug permeated after 0.5 h increased then decreased by increasing the amount incorporated of cholesterol from 0.1 to 0.5 mM. This behavior was not the case for the effect of fatty acid chain length on the burst drug permeation. Increasing the fatty acid chain length at either low or high cholesterol levels led to an increase in the burst JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2218
ZIDAN AND MOKHTAR
Figure 3. Three-dimensional response surface and counter plots showing the effect of the surfactant chain length incorporated (X1 ) and the amount of cholesterol (X2 ) on the dependant variables of flurbiprofen proniosomes.
effect. This result could be ascribed to two factors: (i) the presence of free drug adsorbed at the niosomal surface and (ii) the increased steric geometrical configuration of the amphiphilic molecules as the side chain increased, which interfered with flurbiprofen packing into the niosomal membranes, hence initial burst of the drug occurred. The effects of X1 and X2 on the SSTF and PC were more or less similar. At a cholesterol level of 0.1 mM, SSTF and PC decreased from 151.8 to 62.8 :g/cm2 h and from 0.263 to 0.11 cm/ h by increasing the fatty acid chain length from span 20 to 60, respectively. On the contrary, SSTF and PC decreased from 153.8 to 120.18 :g/cm2 h and from 0.263 to 0.209 cm/h by increasing the amount incorporated of cholesterol from 0.1 to 0.5 mM at low fatty acid chain length, respectively. This result could be due to the fact that increasing the surfactant chain length was accompanied by an increase in the transition temperature of the surfactant; hence, at 37◦ C, the surfactant vesicles were in the ordered gel state, which decreased the drug leakage and permeation.24 On the contrary, addition of cholesterol appeared to disrupt the ordered array of hydrocarbon chains in the gel phase.25 Below the transition temperature of a surfactant, cholesterol disrupted the membrane packing, whereas above transition temperature cholesterol assisted the niosomal membranes packing and less leakiness.26
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
Interaction Plots The main effects of the independent variables on the investigated responses were further elucidated using interaction plots (Fig. 4). At high level of cholesterol content, the effect of changing the fatty acid chain length of the nonionic surfactant on the EE was negligible. On the contrary, minimal effect of changing the cholesterol level was observed on the initial burst effect at low fatty acid chain length. These observations could demonstrate that the presence of cholesterol was the most influential factor to increase niosomal bilayer cohesion. The interaction of cholesterol with the surfactant could result in a dramatic increase in the area of expansion modulus and strength of the bilayer membrane.27 However, owing to spontaneous and rapid niosomal formation from proniosomes and as flurbiprofen was added in excess of what could be entrapped, burst release was observed at both low and high cholesterol concentrations.
DESIRABILITY FUNCTION The concept of desirability scale arose when there was a need to combine the magnitude of several characteristics to a dimensionless scale (say, d). Generally, this desirability function was so constructed that any property (characteristic) value of a product was mapped between zero and one called
DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
2219
Figure 4. Interaction plots showing the quadratic effects of interactions between the independent variables on the investigated responses.
“individualized desirability.” Desirability scale value of one meant the optimum property level of the product or service, whereas zero desirability indicated an unacceptable product. Any value in between zero and one gave an opportunity to improve the product quality. Harrington28 first defined the desirability function, a slight modification of which was done by Gatza and McMillan29 for which the desirability scale could be negative for negative input values. The individual desirabilities were then combined using the geometric mean, which gave the overall generalized desirability D as follows: D = [d1 (Y1 ) × d2 (Y2 ) × · · · × dk(Yk)]1/k
(5)
where k denoting the number of responses. Notice that if any response Yi was completely undesirable [di (Yi ) = 0), then the overall desirability was zero. In practice, fitted response values “i” were used in place of the Yi . Taking into consideration the effect of the independent variables on the studied parameters, the levels of these factors were determined using the generalized desirability function to maximize all the investigated responses (Fig. 5). The predicted values of the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE were 3.06%, 28.9%, 152.07 :g/cm2 h, 0.263 cm/h, and 39.4%, respectively, at X1 and X2 levels of span 20 and 0.14 mM, respectively. As a confirmation process, a fresh formulation was prepared DOI 10.1002/jps
with the optimized values of the independent variables that yielded the percentages of the drug permeated through cellulose ester membranes after 0.5 and 8 h, SSTF and PC for drug permeation through rabbit skin after 8 h, and EE of 2.49%, 27.4%, 147.23 :g/cm2 h, 0.285 cm/h, and 40.8%, respectively. The observed values were not significantly different (p < 0.05) from the predicted values with standardized residuals for each response not exceeding 4% of the nominal value demonstrating the feasibility of quality by design approach to understand the performance of flurbiprofen proniosomal formulations.
CONCLUSION The present study showed the suitability of proniosomes as carriers for transdermal delivery of poorly water-soluble drugs such as flurbiprofen. As demonstrated by PCs and EE, proniosomes offered systems with high drug loading and skin permeation. The fatty acid side chain length was the most important factor controlling the permeation and EE characteristics. In addition, a maximum permeation value and a minimum EE were obtained at shorter fatty acid chain length of spans and minimum amount of cholesterol. Finally, through the rigorous analysis of the two independent variables and its coeffects on the investigated responses, this study demonstrated the potential of quality by design in developing proniosomal gels as drug carrier systems. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
2220
ZIDAN AND MOKHTAR
Figure 5. Generalized desirability profiler to maximize all the investigated responses.
DISCLAIMER The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any Agency determination or policy.
REFERENCES 1. Schreier H, Bouwstra J. 1994. Liposomes and niosomes as topical drug carriers: Dermal and transdermal drug delivery. J Control Release 30:1–15. 2. Aggarwal D, Kaur IP. 2005. Improved pharmacodynamics of timolol maleate from a mucoadhesive niosomal ophthalmic drug delivery system. Int J Pharm 290:155–159 3. Manconi M, Sinico C, Valenti D, Loy G, Fadda AM. 2002. Niosomes as carriers for tretinoin. I. Preparation and properties. Int J Pharm 234:237–248. 4. Hu C, Rhodes DG. 2000. Proniosomes: A novel drug carrier preparation. Int J Pharm 206(1–2):110–122. 5. Vora B, Khopade AJ, Jain NK. 1998. Proniosome based transdermal delivery of levonorgestrel for effective contraception. J Control Release 54:149–165. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011
6. Alsarra IA, Bosela AA, Ahmed SM, Mahrous GM. 2005. Proniosomes as a drug carrier for transdermal delivery of ketorolac. Eur J Pharm Biopharm 59(3):585–490. 7. Mohammed AR, Weston N, Coombes AGA, Fitzgerald M, Perrie Y. 2004. Liposome formulation of poorly water soluble drugs: Optimization of drug loading and ESEM analysis of stability. Int J Pharm 285:23–34. 8. Gregoriadis G. 1993. Liposome technology, 2nd ed. Boca Raton, Florida:CRC Press. 9. Gregoriadis G, Perrie Y. 2010. Liposomes. In Encyclopedia of Life Sciences (ELS); Chichester: John Wiley & Sons, Ltd. . . 10. Bernsdorff C, Wolff A, Winter R, Gratton E. 1997. Effect of hydrostatic pressure on water penetration and rotational dynamics in phospholipids–cholesterol bilayers. Biophys J 72:1264–1277. 11. Charoo NA, Shamsher AA, Kohli K, Pillai K, Rahman Z. 2008. Improvement in bioavailability of transdermally applied flurbiprofen using tulsi (Ocimum sanctum) and turpentine oil. Colloids Surf B Biointerfaces 65(2):300–307. 12. Sebastiani P, Nicoli S, Santi P. 2005. Effect of lactic acid and iontophoresis on drug permeation across rabbit ear skin. Int J Pharm 292:119–126. 13. Larrucea E, Arellano A, Santoyo S, Ygarrua P. 2001. Interaction of tenoxicam with cyclodextrins and its influence on the in vitro percutaneous penetration of the drug. Drug Dev Ind Pharm 27(3):25l-260. 14. Mura P, Faucci MT, Bramanti G, Corti P. 2000. Evaluation of transcutol as a clonazepam transdermal permeation enhancer from hydrophilic gel formulations. Eur J Pharm Sci 9:365–372. 15. Udupa N, Chandraprakash KS, Umadevi P, Pillai GK. 1993. Formulation and evaluation of methotrexate niosomes. Drug Dev Ind Pharm 19:1331–1342. 16. Hwang BY, Jung BH, Chung SJ, Lee MH, Shim CK. 1997. In vitro skin permeation of nicotine from proliposomes. J Contr Release 49:177–184. 17. Furlanetto S, Maestrelli F, Orlandini S, Mura P. 2003. Optimization of dissolution test precision for a ketoprofen oral extended-release product. J Pharm Biomed Anal 32:159– 165. 18. Furlanetto S, Cirri M, Maestrelli F, Corti G, Mura P. 2006. Study of formulation variables influencing the drug release rate from matrix tablet by experimental design. Eur J Pharm Biopharm 62:77–84. 19. Gregoriadis G, Davis C. 1979. Stability of liposomes in vivo and in vitro is promoted by their cholesterol content in the presence of blood cells. Biochem Biophys Res Commun 90:1287–1293. 20. Yoshioka T Sternberg B, Florence AT. 1994. Preparation and properties of vesicles (niosomes) of sorbitan monoesters (Span 20, 40, 60, and 80) and sorbitan triester (Span 85) Int J Pharm 105:1–6. 21. Hao Y, Zhao F, Li N, Yang Y, Li K. 2002. Studies on a high encapsulation of cholchicine by niosome system. Int J Pharm 244:73–80. 22. Mokhtar M, Sammour OA, Hammad MA, Megrab NA. 2008. Effect of some formulation parameters on flurbiprofen encapsulation and release rates of niosomes prepared from proniosomes. Int J Pharm 361:104–111. 23. Kirby C, Clarke J, Gregoriadis G. 1980. Effect of the cholesterol content of small unilamellar liposomes on their stability in vivo and in vitro. Biochem J 186:591–598. 24. Uchegbu IF, Vyas SP. 1998. Non-ionic surfactant based vesicles (niosomes) in drug delivery. Int J Pharm 172:33–70. 25. Ladbrooke BD, Williams RM, Chapman D. 1968. Studies on lecithin–cholesterol–water interactions by differential scanning calorimetry and X-ray diffraction. Biochim Biophys Acta 150:333–340. 26. Papahadjopoulos D, Jacobson K, Nir S, Isac T. 1973. Phase transition in phospholipid vesicles: Fluorescence polarization DOI 10.1002/jps
MULTIVARIATE OPTIMIZATION OF FLURBIPROFEN PRONIOSOMES
and permeability measurements concerning the effect of temperature and cholesterol. Biochim Biophys Acta 311:330– 334. 27. David N, Rashmi SN. 1990. Elastic deformation and failure of lipid bilayer membranes containing cholesterol. Biophys J 58:997–1009.
DOI 10.1002/jps
2221
28. Harrington, Jr. EC. 1965. “The Desirability Function.” Industrial Quality Control, Volume 21, No. 10, pp. 494-498. 29. Gatza PE, McMillan RC. 1972. The use of experimental design and computerized data analysis in elastomer development studies, Division of Rubber Chemistry, American Chemical Society Fall Meeting, Cincinnati, Ohio, pp 3–6.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 6, MAY 2011