Optimization of preparation method for ketoprofen-loaded microspheres consisting polymeric blends using simplex lattice mixture design

Materials Science and Engineering C 69 (2016) 598–608

Contents lists available at ScienceDirect

Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec

Optimization of preparation method for ketoprofen-loaded microspheres consisting polymeric blends using simplex lattice mixture design Sanjoy Kumar Das ⁎, Jasmina Khanam, Arunabha Nanda Department of Pharmaceutical Technology, Jadavpur University, Kolkata 700032, India

a r t i c l e

i n f o

Article history: Received 3 May 2016 Received in revised form 30 June 2016 Accepted 4 July 2016 Available online 5 July 2016 Keywords: Optimization Ketoprofen Microspheres Polymeric blend Simplex lattice mixture design

a b s t r a c t In the present investigation, simplex lattice mixture design was applied for formulation development and optimization of a controlled release dosage form of ketoprofen microspheres consisting polymers like ethylcellulose and Eudragit®RL 100; when those were formed by oil-in-oil emulsion solvent evaporation method. The investigation was carried out to observe the effects of polymer amount, stirring speed and emulsifier concentration (% w/w) on percentage yield, average particle size, drug entrapment efficiency and in vitro drug release in 8 h from the microspheres. Analysis of variance (ANOVA) was used to estimate the significance of the models. Based on the desirability function approach numerical optimization was carried out. Optimized formulation (KTF-O) showed close match between actual and predicted responses with desirability factor 0.811. No adverse reaction between drug and polymers were observed on the basis of Fourier transform infrared (FTIR) spectroscopy and Differential scanning calorimetric (DSC) analysis. Scanning electron microscopy (SEM) was carried out to show discreteness of microspheres (149.2 ± 1.25 μm) and their surface conditions during pre and post dissolution operations. The drug release pattern from KTF-O was best explained by Korsmeyer-Peppas and Higuchi models. The batch of optimized microspheres were found with maximum entrapment (~90%), minimum loss (~10%) and prolonged drug release for 8 h (91.25%) which may be considered as favourable criteria of controlled release dosage form. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Ketoprofen (2-(3-Benzoylphenyl) propionic acid), a nonsteroidal anti-inflammatory drug (NSAID) used in chronic disorders such as spondylitis, osteoarthritis and rheumatoid arthritis [1,2]. Its short biological half-life (2 − 3 h) directs frequent administration [3] of doses to relief pains in chronic inflammatory ailments which causes nausea, vomiting as unabsorbed remaining amount of drug between two dosing causes the irritation to gastrointestinal tract (GIT) membrane owing to insolubility of drug [4]. To overcome these associated adverse effects and to avoid problems of short half-life ketoprofen may be considered as a suitable candidate for developing prolonged release dosage form (microspheres), which may meet up therapeutic objective, patient acceptance and drug management [5]. Over the last two decades commonly reported methods of preparing ketoprofen-loaded microspheres include spray drying [6,7], emulsion cross-linking [8], ionotropic gelation [9], complex coacervation [10], melt dispersion [11], quasi emulsion solvent diffusion method [12,13] and emulsion solvent evaporation [14]. Among these methods, emulsion solvent evaporation (oil-in-oil, o/o) has many advantages such as cost effectiveness [15], simplicity, success with poorly aqueous soluble ⁎ Corresponding author. E-mail address: [email protected] (S.K. Das).

http://dx.doi.org/10.1016/j.msec.2016.07.010 0928-4931/© 2016 Elsevier B.V. All rights reserved.

drug, and production of microspheres with relatively high drug loading [16], increase of surface area for better release and requirement of only mild conditions such as ambient temperature and constant stirring [17]. This method had not been much used to prepare ketoprofen-loaded microspheres [14,18–21]. Ethylcellulose polymers are recognized as ‘generally regarded as safe’ (GRAS) and used widely in tablet coatings, controlled-release coatings, microencapsulation, granulation, and in taste masking [22]. Ethylcellulose microspheres suffer from too slow and incomplete drug release as it is not porous type [23]. The coupling of permeable polymer with ethylcellulose can modify drug release profile [24]. Eudragit®RL 100 is the copolymers of acrylic and methacrylic acid esters with a low content in quaternary ammonium groups. The ammonium groups are present as salts and make this polymer permeable [25]. The nature of polymers greatly influences the rate of diffusion of drug molecules from the matrix of microspheres [26,27]. Conventional experimentation is empirical and lengthy and it involves a good deal of efforts as the effect of each variables was evaluated separately. Conventional method based on trial and error often lacks reproducibility, validity and versatility as this is not validated statistically. So its success frequently relies on the statistical knowledge and the working experience of the formulation scientist [28,29]. Software based statistical tool analyzes the effects of the interactions among different variables under investigation [30]. Many design methods are there to optimize the conditions of this method which are selected

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

judiciously [31]. In factorial design levels of the factors are chosen under individual frame/limit. The factor levels can be fixed too within a composition (either % or fraction of the whole) [32]. This is used generally in the cases of composition of materials involved to prepare a product. Therefore, mixture design is found suitable where all fractions of the components must sum to unity and is known as simplex lattice mixture design (SLMD). The advantage of this design is that it requires least number of experimental runs [29,33–35]. The published data on modelling of microencapsulation of ketoprofen embedded in Eudragit®RL 100, a more permeable polymer, along with Ethylcellulose is conspicuously rare to our knowledge. Ketoprofen needs to be formulated as controlled release dosage form in a better way for post-operative patients in order to enhance its potentiality and to negate drawbacks of frequently administered dosage forms. The aim of this study was to design formulation criteria of controlled release dosage form of ketoprofen and the main objective was to examine the effects of independent variables, i.e., amount of polymer, stirring speed and emulsifier's (Span 80) concentration on the responses such as percentage yield, particle size, drug entrapment efficiency and in vitro drug release in 8 h span from microspheres with the help of models and response surfaces generated by simplex lattice mixture design and to optimize the process variables to achieve the desired criteria in final product. Other objectives were to validate models and characterizations of products formed by emulsion solvent evaporation method. 2. Materials and methods 2.1. Materials Ketoprofen (MW = 254.281) was purchased from Yarrow Chem Products (Mumbai, India). Ethocel (Ethylcellulose, viscosity range 18– 22 mPa·s, ethoxyl content 48.0–49.5%, Dow) and Eudragit®RL 100 (ethyl acrylate, methyl methacrylate, trimethylammonioethyl methacrylate polymer, MW approx. 150,000) were kindly donated by Colorcon Asia Pvt. Ltd. (Goa, India) and Evonik India Pvt. Ltd. (Mumbai, India) respectively. Span 80 (MW 428.61, HLB value 4.7; Loba Chemie Pvt. Ltd., India), Acetone (Merck Specialties Pvt. Ltd., India), methanol (Merck Specialties Pvt. Ltd., India), light liquid paraffin (density 823 kg/m3 at 25 °C, viscosity 29.15 mPa·s at 25 °C; measured in our Laboratory) (Ranbaxy Fine Chemical Ltd., India) and petroleum ether (Ranbaxy

599

Fine Chemical Ltd., India) were purchased and used as received without any further purification. All other reagents and solvents used were of analytical grade. 2.2. Experimental design In the present study, the simplex lattice mixture design (SLMD) was used to evaluate the effect of amount of polymer (X1, mg), stirring speed (X2, rpm) and emulsifier (span 80) concentration (X3, % w/w) on the responses: yield (%), average particle size (μm), drug entrapment efficiency (%) and cumulative drug release after 8 h (%) from microspheres. In contrary individual levels (actual amounts) used for each factor here, we used fractions of each level and sum of contributory fractions makes one always, i.e. 0.5 + 0 + 0.5 = 1 [30]. The factors, levels (in terms of coded value), and matrix of the experimental design were outlined in Table 1. These levels (coded) were ascribed to each of three factors. Design was generated by using Design-Expert®7 trial version software (Stat-Ease Inc., Minneapolis, USA). The measured response function (Y) of the mixture model was explained by using following quadratic equation [36]: Y¼

Xq

βx i¼1 i i

þ

X Xq

β xx ib j ij i j

ð1Þ

where, Y is the response variable; βi and βij are the regression coefficients of pure component and interaction components respectively; xi is the variable and xixj represents the interaction between variables. In this study, total 14 runs were conducted of which six pure component blends, four binary blends, and four ternary blends. A coded level of simplex lattice mixture design was shown in Fig. 1. Analysis of variance (ANOVA) was used to estimate the significance of the model and to remove the non-significant terms (p N 0.05). F-test and Lack of Fit values confirm the applicability of the model. The best fit equations for all the responses were obtained after removing the non-significant terms [28]. A numerical optimization technique based on desirability function approach was used to optimize the compositions of formulation. However, optimization of all variables at the same time is not possible at a time because several responses may be applicable to do the same when one response may antagonize other [37]. During numerical optimization software suggests numerous check point solutions within the experimental domain along with optimized formulation. In view

Table 1 Design matrix and observed values of response. Variables Level (code)

Low (0)

High (1)

500 Amount of polymer (mg) (X1) Stirring speed (rpm) (X2) 800 Span 80 Conc. (% w/w) (X3) 1 Polymer used (80% Ethylcellulose and 20% Eudragit®RL 100) but total amount of polymer varies Run order

Factor 1

Factor 2

Factor 3

1500 1200 3

Response 1

Response 2

Response 3

Response 4

Code

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Amount of polymer (X1)

Stirring speed (X2)

Emulsifier concentration (X3)

Yield (Y1), %

Particle size (Y2), μm

DEE (Y3), %

Rel 8 h (Y4), %

0.5 1 0 0 0 1 0.5 0 0.667 0.167 0.333 0.167 0.5 0

0 0 0 0 1 0 0.5 1 0.167 0.667 0.333 0.167 0.5 0.5

0.5 0 1 1 0 0 0 0 0.167 0.167 0.333 0.667 0 0.5

91.64 ± 1.17 93.05 ± 1.79 77.07 ± 2.53 77.97 ± 1.38 79.03 ± 1.45 92.47 ± 1.59 87.96 ± 1.10 76.10 ± 1.87 92.10 ± 1.68 82.12 ± 1.90 85.20 ± 1.55 80.98 ± 1.09 89.44 ± 1.17 74.97 ± 1.31

160 ± 4.01 181.93 ± 2.53 94.73 ± 2.83 110.07 ± 3.41 91.93 ± 3.24 175.33 ± 5.52 150 ± 1.78 79.93 ± 3.16 162 ± 3.86 112.33 ± 2.20 121.87 ± 1.36 116.07 ± 3.44 155.47 ± 1.63 77.27 ± 3.07

89.53 ± 4.00 91.92 ± 5.98 81.21 ± 5.84 81.01 ± 5.51 80.00 ± 6.56 92.93 ± 4.94 88.00 ± 5.74 82.02 ± 5.13 90.42 ± 4.32 83.00 ± 5.21 85.10 ± 4.82 82.05 ± 2.67 89.99 ± 7.35 78.99 ± 6.37

89.67 ± 1.49 86.06 ± 1.27 96.25 ± 1.24 96.09 ± 2.69 96.92 ± 1.34 86.96 ± 1.70 88.98 ± 1.36 96.48 ± 1.97 88.44 ± 1.11 92.95 ± 1.12 89.76 ± 1.00 91.63 ± 1.02 91.08 ± 1.11 95.05 ± 1.26

600

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

Fig. 1. Coded levels of simplex lattice mixture design.

To prepare various batches of microspheres (Table 1) loaded with the drug ketoprofen active and solvents were as follows: drug

(500 mg, fixed), polymer blend (500–1500 mg) (ethylcellulose: Eudragit®RL 100, 80:20), mixed solvent system (MSS) (acetone:methanol, 2:1, v/v) and emulsifier (span 80, 1–3%, w/w). A 50 ml beaker containing 10 ml of MSS was kept on a magnetic stirrer and a mixture of drug and polymer blend (Table 1) was added to it and top of the beaker was covered with parafilm. The content of the beaker was blended thoroughly by the magnetic stirrer for 15 min at room temperature. This mixture was then slowly added, using a dropper into light liquid paraffin (120 ml) in a beaker (internal diameter

Table 2 ANOVA for yield (%) and average particle size (μm).

Table 3 ANOVA for DEE (%) and R8 h (%).

of validating the models, we have chosen randomly few solutions. The check point formulations (CPFs) and optimized formulation batches were prepared in triplicate. Then the predicted values of the responses were compared with that of experimentally obtained values. 2.3. Preparation of microspheres

Source

Sum of squares

d.f.a

Mean square

F value

Probability N F

For yield (%) Model 580.47 5 116.09 111.05 b0.0001 Linear mixture 517.96 2 258.98 247.73 b0.0001 X1 X2 18.58 1 18.58 17.77 0.0029 X1 X3 39.17 1 39.17 37.47 0.0003 X2 X3 7.97 1 7.97 7.62 0.0247 Residual 8.36 8 1.05 Lack of fit 2.40 4 0.60 0.40 0.7998 Pure error 5.96 4 1.49 Corrected total 588.83 13 Other statistics: Std. Dev. = 1.02, Mean = 84.29, CV (%) = 1.21, PRESS = 27.46, R2

Source

Sum of squares

d.f.a

Mean square

F value

Probability N F

For DEE (%) Model 298.44 5 59.69 74.55 b0.0001 Linear Mixture 280.96 2 140.48 175.47 b0.0001 X1 X2 6.37 1 6.37 7.95 0.0225 X1 X3 4.97 1 4.97 6.20 0.0375 X2 X3 7.32 1 7.32 9.14 0.0165 Residual 6.40 8 0.80 Lack of fit 1.85 4 0.46 0.41 0.7970 Pure error 4.55 4 1.14 Corrected total 304.85 13 Other statistics: Std. Dev. = 0.89, Mean = 85.44, CV (%) = 1.05, PRESS = 25.92, R2

= 0.9858, Adjusted R2 = 0.9769, Predicted R2 = 0.9534, Adequate precision =

= 0.9790, Adjusted R2 = 0.9659, Predicted R2 = 0.9150, Adequate precision =

26.975, Corresponding chance of ‘Lack of Fit F-value’ = 79.98%

24.107, Corresponding chance of ‘Lack of Fit F-value’ = 79.70%

For average particle size (μm) Model 16,255.68 5 3251.14 74.32 b0.0001 Linear Mixture 15,163.42 2 7581.71 173.32 b0.0001 X1 X2 534.47 1 534.47 12.22 0.0081 X1 X3 269.38 1 269.38 6.16 0.0380 X2 X3 347.57 1 347.57 7.95 0.0225 Residual 349.95 8 43.74 Lack of fit 123.55 4 30.89 0.55 0.7141 Pure error 226.40 4 56.60 Corrected total 16,605.63 13 Other statistics: Std. Dev. = 6.49, Mean = 127.71, CV (%) = 5.08, PRESS = 1113.47, R2 = 0.9789, Adjusted R2 = 0.9658, Predicted R2 = 0.9306, Adequate precision = 23.517, Corresponding chance of ‘Lack of Fit F-value’ = 71.41% p N 0.05 is considered as significant. a d.f. indicates degree of freedom.

For R8 h (%) Model 177.83 5 35.57 55.27 b0.0001 Linear Mixture 162.57 2 81.29 126.31 b0.0001 X1 X2 4.73 1 4.73 7.35 0.0266 X1 X3 4.99 1 4.99 7.76 0.0237 X2 X3 4.38 1 4.38 6.81 0.0312 Residual 5.15 8 0.64 Lack of Fit 2.43 4 0.61 0.89 0.5423 Pure error 2.72 4 0.68 Corrected total 182.98 13 Other statistics: Std. Dev. = 0.80, Mean = 91.88, CV (%) = 0.87, PRESS = 20.39, R2 = 0.9719, Adjusted R2 = 0.9543, Predicted R2 = 0.8886, Adequate precision = 19.218, Corresponding chance of ‘Lack of Fit F-value’ = 54.23% p b 0.05 is considered as significant. a d.f. indicates degree of freedom.

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

7 cm) containing span 80 and the liquid was stirred (800–1200 rpm) using mechanical stirrer (propeller diameter 3.5 cm) to form oil1-inoil2 emulsion. The discrete droplets (drug-polymer-solvent) constitute dispersed phase. Droplets were hardened and microspheres were collected by filtration and washed with petroleum ether (50 ml, 4 times) and dried under vacuum at room temperature for a period of 12 h and stored in a desiccator for further use. Blank microspheres without drug were prepared following the same procedure as mentioned above. All the batches of microspheres were prepared in triplicate [38].

601

2.4. Solid-state studies Fourier transform infrared (FTIR) analysis was performed in a FTIR spectrophotometer (FTIR-8300, Shimadzu C., Kyoto, Japan) combined with Quick Snap sampling modules. The samples (ketoprofen powder, carriers, physical mixtures, blank microspheres and ketoprofen-loaded microspheres) were combined with infrared-grade potassium bromide and compressed into disc by applying pressure in a hydraulic press. The discs were scanned over a wave number range of 4000–400 cm−1.

Fig. 2. Contour (a) and response surface plots (b) showing the effects on yield (%), particle size (μm), DEE (%) and R8 h (%).

602

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

Table 4 Validation of models with check point formulations (CPFs). Code

Composition coded level

CPF 1

0.274 (774)c

0.678 (1071)

0.048 (1.096)

CPF 2

0.669 (1169)

0.045 (818)

0.287 (1.574)

CPF 3

0.468 (968)

0.319 (928)

0.212 (1.424)

X1

a b c

X2

Responses

Experimental valuesa

Predicted values

% Errorb

Yield (%) Particle size (μm) DEE (%) R8 h (%) Yield (%) Particle size (μm) DEE (%) R8 h (%) Yield (%) Particle size (μm) DEE (%) R8 h (%)

86.37 ± 1.46 122 ± 2.43 83.77 ± 2.78 93.32 ± 1.71 91.47 ± 1.51 163.33 ± 1.42 91.87 ± 3.59 86.57 ± 2.73 87.19 ± 2.91 143.33 ± 2.66 89.27 ± 3.28 87.79 ± 1.68

84.44 125.38 85.53 92.24 92.82 166.57 90.60 87.90 88.45 145.86 87.73 89.29

2.29 −2.70 −2.06 1.17 −1.45 −1.95 1.40 −1.51 −1.42 −1.73 1.76 −1.68

X3

Mean ± S.D.; n = 3. Percentage of error (%) = [(actual value − predicted value) / predicted value] × 100. Actual composition values shown in parenthesis.

Thermal analysis was performed by DSC (Pyris diamond TG/DTA; Perkins Elmer Instruments, Mumbai, India). Samples (2–8 mg) were heated from 30 to 250 °C at a heating rate 10 °C/min under a steady stream of nitrogen gas (flow rate 25 ml/min).

DEEð%Þ ¼

2.5. Characterization of microspheres 2.5.1. Determination of percentage yield, particle size and drug entrapment efficiency Percentage yield of product was measured with respect to initial load of mixture of drug and polymer blend. Following equation was used: Yieldð%Þ ¼

2 h sonication in a bath sonicator (PCI Analytics, Mumbai, India) at 25 °C. The percentage of drug entrapment efficiency (DEE) was calculated using following formula:

weight of microspheres  100 weight of polymer þ weight of drug

ð2Þ

The average particle size of the microspheres was measured by optical microscope (KYOWA Genter microscope, Tokyo). Three replicates were measured for each batch of microspheres. Standard curve was constructed plotting the data of absorbance at λmax of 260 nm, in a range of concentration (1–10 μg/ml) of ketoprofen. Equation of standard curve, Y = 0.066X + 0.004 (R2 = 0.999), was used to estimate amount of ketoprofen throughout this study. Actual drug content in each batch of microspheres was determined by solvent extraction method as described by Pandit et al. [14] with slight modification. Methanol (50 ml) was used to dissolve ketoprofen from 100 mg of microspheres, and dissolution/extraction of drug was enhanced by

actual drug content in microspheres  100 theoretical drug content in microspheres

ð3Þ

2.5.2. In vitro release study of ketoprofen-loaded microspheres The experiments of drug release were performed following standardized method used earlier [14] with little modifications. A batch of microspheres, equivalent to 100 mg of ketoprofen was taken in a small piece of muslin cloth and tied to the paddle of the dissolution apparatus (Electrolab, TDT 06P, India) and then placed in the dissolution medium (900 ml of phosphate buffer, pH 7.4, 37 ± 0.5 °C) and the medium was agitated at the speed of 100 rpm during dissolution study. At the preset time intervals, an aliquot of the medium (5 ml) was withdrawn and replaced with an equal volume of fresh medium at 37 ± 0.5 °C. After filtration withdrawn samples were assayed for drug quantification at λmax, 260 nm in a UV-VIS Spectrophotometer (UV-180, Analab, India). The release studies were carried out three times for obtaining a mean value. The data obtained from in vitro release studies were fitted to zero-order model (Qt = Qo + Kot), first order model (ln Qt = ln Qo + K1 t), Higuchi's model (Qt = ln Qo + KHt1/2), Korsmeyer-Peppas model (Mt/Mα = KPtn), 1/3 and Hixson-Crowell model (Q1/3 t -Qo = KHCt) to understand the drug release mechanism [39]. Where, Qt is the amount of drug released at time t; K0, K1, KH, Kp and KHC are the release rate constants for the respective

Table 5 The criterion for numerical optimization with desirability function approach. Parameters

Goal

Lower limit

Upper limit

Lower weight

Upper weight

Importance

X1: Amount of polymer (mg) X2: Stirring speed (rpm) X3: Span 80 conc. (% w/w) Yield (Y1, %) Particle size (Y2, μm) DEE (Y3, %) R8 h (Y4, %)

In range In range In range Maximize Is in range Maximize Target = 90

0 0 0 74.97 77 78.99 86.06

1 1 1 93.05 182 92.93 96.92

1 1 1 1 1 1 1

1 1 1 1 1 1 1

3 3 3 3 3 3 3

Solutions Code

KTF-O

Variables

Responses

X1

X2

X3

0.492 (992)

0.508 (1003)

0 (1)

Actual composition values shown in parenthesis.

Yield (%) Particle size (μm) DEE (%) R8 h (%)

Desirability Experimental value

Predicted value

% Error

89.81 ± 2.01 149.2 ± 1.25 89.98 ± 2.75 91.25 ± 1.14

88.70 150.96 88.79 90.00

1.36 −1.17 1.34 1.39

0.811

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

603

emulsion solvent evaporation method using span 80 as emulsifying agent. We found mixed solvent system (MSS) of acetone and methanol (2:1, v/v) as effective solvent of dispersed phase (dielectric constant ~ 27 27 at 25 °C) to dissolve both drug and polymers used in this study [40]. Nonpolar liquid paraffin light (dielectric constant ~ 4.7) was found suitable continuous phase with the selected dispersed phase because liquid paraffin shows poor miscibility with dispersed phase [41,42]. In order to stabilize the interface between dispersed (emulsion droplets) and continuous phase a nonionic surface active agent, Span 80 was used to form an emulsion droplet by decreasing the surface tension. In this study we used span 80 from 1 to 3%, w/w. The surface tension of continuous phase (Liquid paraffin light) was measured by tensiometer at 25 °C with or without adding Span 80. Surface tension of liquid paraffin light was 30 dyne/cm at 25 °C. The surface tension of the external phase was decreased from 29.5 dyne/cm to 28 dyne/cm as the concentration of the Span 80 was increased from 1 to 3%, w/w. Fourteen formulations of ‘ketoprofen-loaded microspheres’ was prepared by emulsion solvent evaporation method using simplex lattice mixture design. 3.2. Data analysis, model validation and optimization

Fig. 3. Dissolution profiles of pure ketoprofen, different CPFs and optimized microspheres (KTF-O) in buffer pH 7.2.

model equations; Q0 is the initial amount of the drug in the solution in result of a burst effect. In Korsmeyer-Peppas model, Mt/Mα is the fraction of the drug release at time t and n is the release exponent [40]. 2.5.3. Scanning electron microscopy The surface morphology of the microspheres was analyzed by Scanning Electron Microscope (SEM, JSM-6700F, JEOL Ltd., Japan) at 5 kV. Prior to the analysis a sample of fully dried microspheres was placed on a metal stub using double-sided adhesive tape and were coated with a thin layer of platinum for 60 s under reduced pressure of 2.54 Pa; at a voltage of 10 kV, current of 25 mA using an ion sputtering device (Auto Fine Coater, JFC-1600, JEOL Ltd., Japan). 2.6. Statistical analysis All estimated data were expressed as mean ± standard deviation (S.D.), and each measurement was done in triplicate. Computational work including design matrix, analysis of variance, fitting of the models, model validation and graphical representations (2D and 3D plots) was performed using a statistical package (Design Expert®7 trial version; Stat Ease Inc., Minneapolis, USA). The variances were considered statistically significant at p b 0.05. 3. Results and discussion 3.1. Preparation of microspheres Ketoprofen-loaded microspheres consisting of polymeric blends (ethylcellulose and Eudragit®RL 100) were prepared by oil1-in-oil2

Present experimental prognosis gives ample opportunity to analyze the influence of aptly selected experimental parameters on the performance of experiments. The values for the responses percentage yield (Y1), particle size (Y2), DEE (Y3) and in vitro drug release in 8 h (Y4) of microspheres were analyzed using the Design-Expert®7 trial version software and the mathematical model for each response was generated. The ANOVA results along with statistical parameters for all the responses were shown in Tables 2 and 3. After elimination of non-significant terms following second-order polynomial equations for each response variable are shown in Eqs. (4)–(7) below. Y1 ðYield %Þ ¼ 92:83X1 þ 77:73X2 þ 77:38X3 þ 14:15X1 X2 þ 24:63X1 X3 −11:11X2 X3

ð4Þ

Y2 ðParticle size μmÞ ¼ 178:44X1 þ 87:00X2 þ 102:92X3 þ 75:91X1 X2 þ 64:60X1 X3 −73:38X2 X3

ð5Þ

Y3 ðDEE %Þ ¼ 92:52X1 þ 81:09X2 þ 81:03X3 þ 8:29X1 X2 þ 8:77X1 X3 −10:65X2 X3

ð6Þ

Y4 ðR8 h %Þ ¼ 86:66X1 þ 96:75X2 þ 96:08X3 −7:14X1 X2 −8:79X1 X3 −8:24X2 X3

ð7Þ

where, Y1, Y2, Y3 and Y4 denoted the four responses. X1, X2, and X3 denoted the amount of polymer, stirring speed and span 80 concentration respectively and the interactions X1X2, X1X3, and X2X3 show how the responses change when two factors change simultaneously [43]. The yield values of the microspheres for the 14 runs varied from 74.97–93.05% (Table 1), which provided the opportunity to observe the effectiveness of oil1-in-oil2 emulsion solvent evaporation method. ANOVA for percentage yield (Table 2) showed that the X1X2, X1X3 and X2X3 are the significant model terms. The Model F-value for percentage yield was found 111.05 which implied the model is significant. There is only a 0.01% chance for percentage yield that a ‘Model F-Value’ this large

Table 6 Correlation coefficients for model fitting of the in vitro release data of CPFs, optimized microspheres and MF. Code

CPF 1 CPF 2 CPF 3 KTF-O MF

Zero order

First order

Hixson-Crowell

Higuchi

Korsmeyer-Peppas

Ko (h−1)

r2

K1 (h−1)

r2

KHC (h− 1/3)

r2

KH (h− 1/2)

r2

KKP (h−n)

r2

n

7.4987 7.7074 7.4362 7.8283 12.5289

0.9182 0.9545 0.9268 0.9450 0.9752

−0.133 −0.1008 −0.1045 −0.1186 −0.1899

0.9881 0.998 0.9967 0.9874 0.9569

−0.1527 −0.1704 −0.1605 −0.1674 −0.2811

0.8847 0.9185 0.8818 0.9007 0.9198

33.4415 31.79 31.9604 32.6679 41.4278

0.9853 0.9957 0.9922 0.9956 0.9776

0.3851 0.3061 0.3460 0.3438 0.1993

0.9574 0.9826 0.9970 0.9980 0.9854

0.48 0.58 0.50 0.50 1.08

604

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

Fig. 4. SEM micrographs of run 14, 8, 2 and 7 (a: whole microspheres; b: surface).

could occur due to noise. The Lack of Fit test was not significant (pvalue = 0.7998 N 0.05) relative to pure error, indicating that the model explained the response (percentage yield) very well within the domains of factors. The ‘Lack of Fit F-value’ for percentage yield was 0.40 and was not significant which is relative to the pure error. For percentage yield, there is a 79.98% chance that a ‘Lack of Fit F-value’ this large could occur due to noise. The low value of coefficient of variation (CV) (1.21%) for yield indicates improved precision and reliability of the experiment carried out. Adequate precision (AP) measures the signal to noise ratio. The value of AP for yield N 4 indicates suitability of the model. Eq. (4) clearly indicates that the regression coefficient value of pure component X1 (92.83) was higher compared to regression coefficient values of X2 (77.73) and X3 (77.38), because amount of

polymer had a greater role for the percentage yield of microspheres. Eq. (4) can also be presented in the form of a two dimensional contour plots and three dimensional response surface plots as seen in Fig. 2 (a1, a2, b1, and b2). The yield of the microspheres lowered to 80% because of these runs (run 3, 4, 5, 8, and 14) were prepared with lower amount of polymer. When we compared the mixture components we found that stirring speed and emulsifier concentration had opposite effect on percentage yield. Table 1 shows that the average particle size values for the 14 runs showing a wide variation ranges from a minimum of 77.27 μm to a maximum of 181.93 μm. The data undoubtedly indicates that the average particle size value is strongly dependent on the chosen variables (independent variables) [44]. ANOVA for particle size showed that the linear

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

Fig. 5. SEM micrographs of optimized microspheres (KTF-O).

mixture components, X1X2, X1X3 and X2X3 are the significant model terms. Other statistical terms (Model F-value 74.32, Lack of Fit F-value 0.55, CV 5.08%, Adequate precision 23.982) for Y2 have similar explanation as that of Y1. The regression coefficient values of pure components were 178.44, 87.00, and 102.92 for amount of polymer (X1), stirring speed (X2) and emulsifier concentration (X3) respectively (Eq. (5)). From these coefficient values we found that the amount of polymer (X1) exhibited a major influence on the particle size. Increasing amount of polymer (X1) resulted in the production of larger particles owing to higher internal phase viscosity [17]. In this study we also measured the viscosity of the internal phase using Brookfield viscometer (Model TV-10, USA) with low viscosity spindle (spindle no. M1, cord no. 20) and operated at 60 rpm at room temperature. For run 3, 4, 5, 8 and 14 internal phase viscosity was 13.6 mPa·s and this low viscosity resulted in the production of smaller particles in the range of 77.27 to 110.07 μm. For run 1, 7 and 13 particle size varies from 150 to 160 μm and this happened because of the higher internal phase viscosity (100.7 mPa·s). The internal phase viscosity of run 2 and 6 were 101 mPa·s and this value was slightly higher as measured for run 1, 7 and 13. Particle size obtained for run 2 and 6 in the range of 175.33 to 181.93 μm. The interfacial tension between internal (dispersed) and external (continuous) phase

605

was measured at 5 min by tensiometer at room temperature. The interfacial tension between the two phases was decreased from 10 dyne/cm to 6 dyne/cm as the concentration of the emulsifier (X3) was increased from 1% to 3% w/w. When we compared the mixture components we found that stirring speed (X2) and emulsifier concentration (X3) had opposite effect on average particle size as evidenced from the equation. As reported previously, that the average particle size of the microspheres decreased with increasing stirring speed [40]. Eq. (5) can also be presented in the form of a two dimensional contour plot and three dimensional response surface plot as seen in Fig. 2 (a3, a4, b3 and b4). The drug entrapment efficiency varies from 78.99–92.93% (Table 1). For DEE, linear mixture components X1X2, X1X3 and X2X3 are the significant model terms. Other statistical terms (Model F-value 74.55, Lack of Fit F-value 0.41, CV 1.05%, Adequate precision 24.107) for Y3 have similar explanation as that of Y1. The regression coefficient values of pure components was 92.52, 81.09 and 81.03 for amount of polymer (X1), stirring speed (X2) and emulsifier concentration (X3) respectively. Eq. (6) clearly indicates that amount of polymer had similar major role on the response as stated earlier. As reported previously, the DEE increased with increased amount of polymer in a fixed volume of organic solvent [45]. The drug entrapment efficiencies were higher for run 2 and 6 owing to higher amount of polymer (1500 mg) and higher viscosity of the internal phase, which further led to solidification at a faster rate and minimized the leaching of the drug into the continuous phase [14]. Eq. (6) can also be presented in the form of a two dimensional contour plots and three dimensional response surface plots as seen in Fig. 2 (a5, a6, b5, and b6). The drug release in 8 h of the prepared microspheres varied from 86.06–96.92%. In this case linear mixture components, X1X2, X1X3 and X2X3 are significant model terms. Other statistical terms (Model Fvalue 55.27, Lack of Fit F-value 0.89, CV 0.87%, Adequate precision 19.218) for Y4 have similar explanation as that of Y1. The regression coefficient values of pure components for drug release in 8 h was 86.66, 96.75 and 96.08 for amount of polymer (X1), stirring speed (X2) and emulsifier concentration (X3) respectively (Eq. (7)). Stirring speed during preparation had a prominent effect on the drug release and it was evidenced from the high regression value. Eq. (7) can also be presented in the form of a two dimensional contour plots and three dimensional response surface plots as seen in Fig. 2 (a7, a8, b7 and b8). Fig. 2 revealed an increase in the drug release in 8 h with the increase of stirring speed because higher stirring speed produced smaller particle size of the microspheres which further facilitated drug release owing to increased surface area. To verify the validity of the model [46] three check point formulations were chosen within the factors space and check point formulations were prepared (composition shown in Table 4). Responses for the CPFs based on the calculation using the model equations (Eqs. (4)–(7) and experimentally determined response values were found to be within limits which indicated the validity of the model. The optimization was carried out to derive the optimum combination of polymer amount (X1), stirring speed (X2) and emulsifier concentration (X3) to formulate microspheres with ‘maximum’ percentage yield, particle size ‘in range’, ‘maximum’ DEE (%) and R8 h at 90%. The combination of polymer amount, 992 mg; stirring speed, 1003 rpm and Span 80 concentration, 1% w/w was found to give a desirability value of 0.811 (Table 5). Then the suggested formulation composed with optimal composition was prepared as per actual compositions and further characterized. Optimized formulation was prepared in triplicate. The optimized microspheres (KTF-O) showed percentage yield of 89.21 ± 2.01%, particle size of 149.2 ± 1.25 μm, DEE of 89.98 ± 2.75% and R8 h % of 91.25 ± 1.14 with small percentage errors (b1.39%). These small percentage error values between experimental and predicted responses proving high prognostic ability of the methodology utilized [47].

606

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

Fig. 6. FTIR spectra of (a) Ketoprofen, (b) Ethylcellulose, (c) Eudragit®RL 100, (d) Physical mixture (PM) (1:2), (e) Blank microspheres and (f) Optimized microspheres (KTF-O).

3.3. Release kinetics The drug release profiles of various batches of check point formulations (CPFs), marketed formulation (100 mg ketoprofen SR capsule), pure ketoprofen and optimized microspheres were shown in Fig. 3. It was observed from the in vitro drug release study that CPF 1, CPF 2, CPF 3, and KTF-O were able to sustain the drug release up to 8 h while marketed formulation (100 mg ketoprofen SR capsule) was able to sustain the drug up to 7 h with 98.62 ± 1.24% drug release. At 8 h, the releases were 93.32 ± 1.71%, 86.57 ± 2.73%, 87.79 ± 1.68% and 91.23 ± 1.14% for CPF 1, CPF 2, CPF 3 and KTF-O respectively. Drug was released from the CPFs prepared with low polymeric amount were little faster compared to microsphere preparations with high amount of polymer as the latter exerts more resistance to drug release. After fitting the drug release data of CPFs, marketed formulation (MF) and optimized microspheres in various kinetic models, it was observed that release of ketoprofen from different formulations was

governed by different mechanisms. The release constants, the correlation coefficients (r2) and exponents (n) for CPFs, MF and (KTF-O) were shown in Table 6. Release patterns of ketoprofen from microspheres were fitted well with the First order (r2, 0.9874–0.9980), Higuchi (r2, 0.9853–0.9957) and Korsmeyer-Peppas (r2, 0.9574–0.9980) equations than other equations (zero order, Hixson-Crowell). According to Korsmeyer-Peppas model, release exponents (n) ranging from 0.48 to 0.58 for all checkpoint formulations and optimized microspheres indicated its inclination towards non-Fickian type (0.45 b n b 0.89) and it may be attributed to slight swelling of polymer and its relaxation in physiological pH. Release pattern from MF was fitted well with the Korsmeyer-Peppas (r2, 0.9854; n ~ 1), Higuchi (r2, 0.9776), Zero order (r2, 0.9752) than other equations. According to Korsmeyer-Peppas model, release exponent (n) for MF was 1.08 which followed super case II-type release. The result of curve fitting studies revealed that drug release pattern from the optimized microsphere (KTF-O) could be best explained by Korsmeyer-Peppas (r2 = 0.998) and Higuchi models (r2 = 0.9956) as n equals to 0.5, following non-Fickian type release. 3.4. Surface morphology analysis The SEM photomicrographs of ketoprofen-loaded microspheres were found spherical and discrete with varying surface texture (roughness) because of varied processing variables. The photomicrographs of some spherical samples were shown in Fig. 4, illustrating surface smoothness, slight shrinkage which may appear owing to drying condition, polymer concentration, stirring speed and emulsifier concentration. The photomicrographs in Fig. 5 were of optimized microspheres. Post dissolution microspheres appeared spherically intact and some erosion marks (large pores) were visible on the surface caused by eddy diffusion associated with molecular diffusion as explained in Higuchi and Korsmeyer-Peppas models of molecular transport. 3.5. Solid state studies

Fig. 7. DSC curves of: (a) Ketoprofen, (b) Ethylcellulose, (c) Eudragit®RL 100, (d) PM (1:2), (e) Blank microspheres and (f) Optimized microspheres (KTF-O).

Fig. 6 illustrated FTIR absorption spectra of (a) ketoprofen, (b) ethylcellulose (c) Eudragit RL 100, (d) physical mixture (PM, 1:2), (e) blank microspheres, (f) optimized microspheres (KTF-O). Any chemical incompatibility among ingredients can be detected by absorption

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

spectra. In Fig. 6a, bands at 2975, 2937 cm−1 are indicative of presence of asymmetric and symmetric C\\H stretch of\\CH3; two typical bands at 1696.90 cm−1 and 1653.41 cm−1 are indicated C_O stretching of presence of ketone and carboxylic acid. Other bands at 1595 cm−1 (aromatic, C_C) and 1448 cm−1 (C\\C stretch in ring) were observed. IR spectra in the range of 860–640 cm−1 indicated the presence of aromatic ring. Ethylcellulose (Fig. 6b) has the characteristic peaks at 3479.04 cm− 1 (O\\H stretching), 2978.23 cm−1 (C\\H stretching brand), 2929.02 cm− 1 (CH stretching in CH3) 2876.52 cm−1 (C\\H stretching brand), 1450.54 cm−1 (C_C aromatic stretching), 1380.73 cm− 1 (C\\H bending), and 1054.55 cm− 1 (\\C\\O\\Cstretching vibration). Eudragit RL 100 showed (Fig. 6c) characteristics peaks at 3442.46 cm−1 (OH stretch), and 1739.22 cm−1 (C_O ester vibration). In the IR spectra of the physical mixture (PM) (Fig. 6d), the principal peaks for ketoprofen obtained at 1697.38 cm− 1 and 1655.40 cm−1 which indicated clearly that there was no interaction between ketoprofen and other components. Blank microspheres (Fig. 6e) showed only few characteristic peaks of ethylcellulose and Eudragit RL 100. The characteristics sharp peaks in Fig. 6a appeared shortened in the spectrum of ‘ketoprofen-loaded microspheres’ (Fig. 6f) as ketoprofen molecules were blended homogeneously with that of polymers in presence of solvent which may cause formation of weak physical bonds between the carboxylic acid groups in the ketoprofen and the ester group in Eudragit RL 100 [40]. Similar physical interaction due to hydrogen bonding was reported by Wu and McGinity [48]. This weak interaction owing to weak forces does not cause any compatibility problem between the drug and the polymers [49]. Fig. 7 shows thermograms of ketoprofen, ethylcellulose, Eudragit RL 100, PM, blank, and optimized formulation respectively. DSC measurements showed the same thermal behavior in all drug-loaded microspheres. Glass transition temperature of ethylcellulose (Tg, 139.42 °C), Eudragit®RL 100 (Tg, 62.71 °C), and melting point of ketoprofen (96.77 °C) were illustrated for all pure samples. The thermogram of physical mixture (PM) exhibited the Tg values at 130.57 and 56 °C for ethylcellulose and Eudragit®RL100 respectively with shorter endothermic melting peak of ketoprofen (92.40 °C). The characteristic endothermic peak of ketoprofen in PMs was shifted slightly towards lower temperature and this could be due to the presence of higher amount of polymers. The DSC thermogram of blank microspheres showed the Tg values located practically almost at the same temperatures as in PM. The melting point was not detectable in the thermograms of optimized formulation demonstrating that, during preparation, crystalline ketoprofen was converted either to a molecular dispersion or amorphous state when it was exposed to high temperature during DSC experiment. Similar explanation was furnished earlier [18]. Profiles of FTIR and DSC do not suggest any chemical incompatibility between drug and other ingredients. 4. Conclusions Ketoprofen-loaded microspheres with polymeric blend were prepared by o/o emulsion solvent evaporation method. Four process variables like amount of polymer (mg), stirring speed (rpm) and emulsifier concentration (% w/w) were chosen to investigate their combined effect on the responses such as yield (%), average particle size (μm), drug entrapment efficiency (DEE, %) and in vitro drug release in 8 h (R8h, %) from the microspheres with an aim to optimize the product. Process optimization in laboratory scale was achieved to prepare controlled release drug delivery system of ketoprofen performing least number of experiments (using simplex lattice mixture design). Optimized formulation depicted favourable responses with desirability factor 0.811. The optimized batch of drug loaded microspheres showed yield of 89.21 ± 2.01%; average particle size of 149.2 ± 1.25 μm; drug entrapment efficiency of 89.98 ± 2.75%; and cumulative percentage release of drug at 8 h 91.25 ± 1.14%. Software generated models were reliable and can be extrapolated to higher scale keeping hydrodynamicity

607

of stirred emulsion in similar fashion. The product batches were characterized by number of routine tests to claim its efficacy as dosage form. The SEM photomicrograph of optimized microspheres were found spherical. This designed dosage form was prepared with ease by emulsion solvent evaporation technique, and its long release period proved its efficacy in sustaining its effect as painkiller medicine. Conflict of interest Authors declare that there is no conflict of interest in this work. Acknowledgements The authors are grateful to Evonik India Private Ltd., Mumbai, India and Colorcon, Goa, India for providing the gift sample of Eudragit®RL 100 and ethylcellulose respectively. The authors express gratitude to the Jadavpur University and instrument room, Department of Pharmaceutical Technology, Jadavpur University for viscosity analysis. References [1] J.E.F. Reynolds (Ed.), Martindale, The Extra Pharmacopeia, The Pharmaceutical Press, London, 1996. [2] G.F. Palmieri, G. Bonacucina, P.D. Martino, S. Martelli, Microencapsulation of semisolid ketoprofen/polymer microspheres, Int. J. Pharm. 242 (2002) 175–178. [3] L. Azouz, F. Dahmoune, F. Rezgui, C. G'Sell, Full factorial design optimization of antiinflammatory drug release by PCL-PEG-PCL microspheres, Mater. Sci. Eng. C. 58 (2016) 412–419. [4] S.U. Jan, G.M. Khan, H. Khan, A.U. Rehman, K.A. Khan, S.U. Shah, K.U. Shah, A. Badshah, I. Hussain, Release pattern of three new polymers in ketoprofen controlled-release tablets, Afr. J. Pharm. Pharmacol 6 (2012) 601–607. [5] F. Cui, D. Cun, A. Tao, M. Yang, K. Shi, M. Zhao, Y. Guan, Preparation and characterization of melittin-loaded poly (DL-lactic acid) or poly (DL-lactic-co-glycolic acid) microspheres made by the double emulsion method, J. Control Rel. 107 (2005) 310–319. [6] P. Giunchedi, B. Conti, L. Maggi, U. Conte, Cellulose acetate butyrate and polycaprolactone for ketoprofen spray-dried microsphere preparation, J. Microencapsul. 11 (1994) 381–393. [7] G. Rassu, E. Gavini, G. Spada, P. Giunchedi, S. Marceddu, Ketoprofen spray-dried microspheres based on Eudragit® RS and RL: study of the manufacturing parameters, Drug Dev. Ind. Pharm. 34 (2008) 1178–1187. [8] S.T. Mathew, S.G. Devi, V.V. Prasanth, B. Vinod, Formulation and in vitro–in vivo evaluation of ketoprofen-loaded albumin microspheres for intramuscular administration, J. Microencapsul. 26 (2009) 456–469. [9] I. El-Gibaly, Oral delayed-release system based on Zn-pectinate gel (ZPG) microparticles as an alternative carrier to calcium pectinate beads for colonic drug delivery, Int. J. Pharm. 232 (2002) 199–211. [10] G.F. Palmieri, S. Martelli, D. Lauri, P. Wehrle´, Gelatin-Acacia complex coacervation as a method for ketoprofen microencapsulation, Drug Dev. Ind. Pharm. 22 (1996) 951–957. [11] F. Maestrelli, N. Zerrouk, M. Cirri, N. Mennini, P. Mura, Microspheres for colonic delivery of ketoprofen-hydroxypropyl-β-cyclodextrin complex, Eur. J. Pharm. Sci. 34 (2015) I–II. [12] M.I. Re, B. Biscans, Preparation of microspheres of ketoprofen with acrylic polymers by a quasi-emulsion solvent diffusion method, Powder Technol. 101 (1999) 120–133. [13] A.H. El-Kamal, M.S. Sorkar, S.S. Al-Gamal, V.F. Naggar, Preparation and evaluation of ketoprofen floating oral delivery system, Int. J. Pharm. 220 (2001) 13–21. [14] S.S. Pandit, D.P. Hase, M.M. Banker, A.T. Patil, N.J. Gaikward, Ketoprofen-loaded Eudragit RSPO microspheres: an influence of sodium carbonate on in vitro drug release and surface topology, J. Microencapsul. 26 (2009) 195–201. [15] P. Parida, S.C. Mishra, S. Sahoo, A. Behera, B.P. Nayak, Development and characterization of ethylcellulose based microspheres for sustained release nefidipine, J. Pharm. Anal. (2014), http://dx.doi.org/10.1016/j.jpha.2014.02.001. [16] D. Perumal, Microencapsulation of ibuprofen and Eudragit® RS 100 by the emulsion solvent diffusion technique, Int. J. Pharm. 218 (2001) 1–11. [17] B.K. Kim, S.J. Hwang, J.B. Park, H.J. Park, Preparation and characterization of drugloaded polymethacrylate microspheres by an emulsion solvent evaporation method, J. Microencapsul. 19 (2002) 811–822. [18] F. Gabor, B. Ertl, M. Wirth, R. Mallinger, Ketoprofen-poly(D,L-lactic-co-glycolic acid) microspheres: influence of manufacturing parameters and type of polymer on the release characteristics, J. Microencapsul. 16 (1999) 1–12. [19] T.M. Mateovic-Rojnik, R. Frlan, M. Bogataj, P. Bukovec, A. Mrhar, Effect of preparation temperature in solvent evaporation process on Eudragit RS microsphere properties, Chem. Pharm. Bull. 53 (2005) 143–146. [20] M. Ricci, P. Blasi, S. Giovagnoli, C. Rossi, G. Macchiarulo, G. Luca, G. Basta, R. Calafiore, Ketoprofen controlled release from composite microcapsules for cell encapsulation: effect on post transplant acute inflammation, J. Control. Rel. 107 (2005) 395–407.

608

S.K. Das et al. / Materials Science and Engineering C 69 (2016) 598–608

[21] G.C. Bazzo, A.T.N. Lemos-Senna, Pires, poly(hydroxybutyrate)/chitosan/ketoprofen or piroxicam composite microparticles: preparation and controlled drug release evaluation, Carbohydr. Polym. 77 (2009) 839–844. [22] Ethocel™ overview, 2013. Available online: http://msdssearch.dow.com/ PublishedLiteratureDOWCOM/dh_08e5/0901b803808e5464.pdf?filepath= dowwolff/pdfs/noreg/198-02292.pdf&fromPage=GetDoc (20.03.15). [23] D. Babay, A. Hoffman, S. Benita, Design and release kinetic pattern evaluation of indomethacin microspheres intended for oral administration, Biomaterials 9 (1988) 482–488. [24] J. Bidone, A.P.P. Melo, G.C. Bazzo, F. Carmignan, M.S. Soldi, A.T.N. Pires, E. LemonSenna, Preparation and characterization of ibuprofen-loaded microspheres consisting of poly(3-hydroxybutyrate) and methoxy poly (ethylene glycol)-b-poly (D,L-lactide) blends or poly(3-hydroxybutyrate) and gelatin composites for controlled drug release, Mater. Sci. Eng. C 29 (2009) 588–593. [25] Technical Information Eudragit, 2014. (Available) online: http://eudragit.evonik. com/product/eudragit/Documents/evonik-specification-eudragit-rl-100-rl-po-rs100-rs-po.pdf (2015.03.22). [26] H. Hamishehkar, J. Emami, A.R. Najafabadi, K. Gilani, M. Minaiyan, H. Mahdavi, A. Nokhodchi, The effect of formulation variables on the characteristics of insulin-loaded poly(lactic-co-glycolic acid) microspheres prepared by a single phase oil in oil solvent evaporation method, Colliods Surf B: Biointerfaces 74 (2009) 340–349. [27] D.S. Jones, K.J. Pearce, An investigation of the effects of some process variables on the microencapsulation of propranolol hydrochloride by the solvent evaporation method, Int. J. Pharm. 118 (1995) 199–205. [28] D.C. Montgomery, Design and Analysis of Experiments, fifth ed. Wiley, New York, 2001. [29] D.C. Montgomery, Introduction to Statistical Quality Control, fourth ed. Wiley India (P.) Ltd., New Delhi, 2008. [30] M.T. Yilmaz, O. Yildiz, B. Yurt, O.S. Toper, S. Karaman, A. Basturk, A mixture design study to determine interaction effects of wheat, buckwheat, and rice flours in an aqueous model system, LWT - Food Sci. Technol. 61 (2015) 583–589. [31] H. Bagheri, A.A. Asgharinezhad, H. Ebrahimzadeh, Determination of trace amounts of Cd (II), Ca (II), and Ni (II) in food samples using a novel functionalized magnetic nanosorbent, Food Anal. Methods 9 (2016) 876–888. [32] H. Bagheri, Y. Yamini, M. Safari, H. Asiabi, M. Karimi, A. Heydari, Simultaneous determination of pyrethroids residues in fruit and vegetable samples via supercritical fluid extraction followed by HPLC-UV, J. Supercritical Fluids 107 (2016) 571–580. [33] R. Huisman, H.V. Van Kamp, J.W. Weyland, D.A. Doornbos, G.K. Bolhuis, C.F. Lerk, Development and optimization of pharmaceutical formulations using a simplex lattice design, Pharm Weekbl Sci. 6 (1984) 185–194. [34] K.T. Fang, Y. Wang, Number-theoretic Methods in Statistics, first ed. Chapman and Hall, London, 1994. [35] P.J. Waaler, C. Graffner, B.W. Müller, Optimization of a matrix tablet formulation using a mixture design, Acta Pharm Nord. 4 (1992) 9–16.

[36] K.A. Batista, L.A.M. Bataus, I.T.N. Campos, K.F. Fernandes, Development of culture medium using extruded bean as a nitrogen source for yeast growth, J. Microbiol. Methods 92 (2013) 310–315. [37] R.H. EI-Dahmy, I. Elsayed, A.H. Elshafeey, N.A.A. El-Gawad, O.N. El-Gazayerly, Optimization of long circulating mixed polymeric micelles containing vinpocetine using simple lattice mixture design, in vitro and in vivo characterization, Int. J. Pharm. 477 (2014) 39–46. [38] S.K. Jain, G. Rai, D.K. Saraf, G.P. Agarwal, The preparation and evaluation of albendazole microspheres for colonic delivery, Pharm. Technol. December 66-71 (2004). [39] P. Costa, J.M.S. Lobo, Modeling and comparison of dissolution profile, Eur. J. Pharm. Sci. 13 (2001) 123–133. [40] S.K. Das, K. Yuvaraja, J. Khanam, A. Nanda, Formulation development and statistical optimization of ibuprofen-loaded polymethacrylate microspheres using response surface methodology, Chem. Eng. Res. Des. 96 (2015) 1–14. [41] T. Mateovic, B. Kriznar, M. Bogataj, A. Mrhar, The influence of stirring rate on biopharmaceutical properties of Eudragit RS microspheres, J. Microencapsul. 19 (2002) 29–36. [42] C.T. Sengel, C. Hascicek, N. Gonul, Development and in-vitro evaluation of modified release tablets including ethylcellulose microspheres loaded with diltiazem hydrochloride, J. Microencapsul. 23 (2006) 135–152. [43] J. Cui, B. Yu, Y. Zhu, H. Li, H. Lou, G. Zhai, Enhancement of oral absorption of curcumin by self-microemulsifying drug delivery systems, Int. J. Pharm. 371 (2009) 148–155. [44] M.D. Bhavsar, S.B. Tiwari, M.M. Amiji, Formulation optimization for the nanoparticles-in-microspheres hybrid oral delivery system using factorial design, J. Control. Rel. 110 (2006) 422–430. [45] B.B.C. Youan, T.L. Jacson, L. Dickens, C. Hernandez, G.O. Ababio, Protein release profiles and morphology of biodegradable microcapsules containing an oily core, J. Control. Rel. 76 (2001) 313–326. [46] K. Yuvaraja, S.K. Das, J. Khanam, Process optimization and characterization of carvedilol solid dispersion with hydroxypropyl-β-cyclodextrin and tartaric acid, Korean J. Chem. Eng. (2014), http://dx.doi.org/10.1007/s11814-014-0192-2. [47] M.A. Sallam, A.M. Motawaa, S.M. Mortada, A modern approach for controlled transdermal delivery of diflunisal: optimization and in vivo evaluation, Drug Dev. Ind. Pharm. 39 (2013) 600–610. [48] C. Wu, J.W. McGinity, Influence of ibuprofen as a solid state plasticizer in Eudragit® RS 30D on the physicochemical properties of coated beads, AAPS PharmSciTech 2 (2001) 1–9. [49] D. Yu, G. Williams, X. Wang, J. Yang, X. Li, W. Qian, Y. Li, Polymer-based nanoparticulate solid dispersions prepared by a modified electro spraying process, J. Biomed. Sci. Eng. 4 (2011) 741–749.