Formulation Optimization and Characterization of Orally in ...

166 Vol. 65, No. 2

© 2017 The Pharmaceutical Society of Japan

Chem. Pharm. Bull. 65, 166–177 (2017)

Regular Article

Formulation Optimization and in Vitro Characterization of Orally Disintegrating Films Using a Factorial Design and Mathematical Modeling for Drug Release

Yeongbin Lee,a Prakash Thapa,b Seong Hoon Jeong,*,b Mi Hee Woo,c and Du Hyung Choi*,a

a Department of Pharmaceutical Engineering, Inje University; Gyeongnam 621–749, Republic of Korea: b College of Pharmacy, Dongguk University-Seoul; Gyeonggi 410–820, Republic of Korea: and c College of Pharmacy, Catholic University of Daegu; Gyeongsan 38430, Republic of Korea.Received September 21, 2016; accepted November 15, 2016; advance publication released online November 30, 2016

Even though experimental designs are becoming popular especially for conventional dosage forms, lim-ited studies have been performed to optimize formulations of orally disintegrating films (ODFs). This study aimed to evaluate sildenafil citrate-loaded ODFs for a controlled release with hydroxypropyl methylcellulose as a film-forming polymer. A factorial design was utilized for optimization with three control factors: etha-nol ratio, plasticizer ratio, and the type of plasticizer. Tensile strength, disintegration time, water contact angle, and thickness were chosen as responses. For optimization, water contact angle, disintegration time, and thickness were minimized, while the tensile strength was maximized. Based on the conditions, optimal formulations were achieved for each type of plasticizer. Evaluation of desirability indicated that the response values were close to the target. When these optimal formulations were validated, the optimal solutions and target values were similar with small biases. The formulations were characterized using scanning electron microscopy, differential scanning calorimetry, Fourier transform infrared spectroscopy, surface pH, in vitro dissolution, and drug release simulation with a mathematical modeling. After the drug was homogenously dispersed throughout the film, the crystalline form of the drug provided strong hydrogen bonds between the drug and the film components. Moreover, it showed a controlled drug release profiles that were well matched with simulated results. This study suggests that the optimized films may present a better alternative to con-ventional tablets for the treatment of male erectile dysfunction.

Key words orally disintegrating film; hydroxypropyl methylcellulose; factorial design; mathematical model-ing; optimization; controlled release

Orally disintegrating films (ODFs) have recently become one of the most popular forms of drug administration due to patient convenience and compliance. The main advantage of the dosage form is its fast disintegration; it can be placed on the tongue without the need for water. This usually results in enhanced bioavailability with faster onset of action com-pared to conventional oral dosage forms.1,2) ODFs are flexible; therefore, they do not require any kinds of special package for protection during transportation and storage, as compared to orally disintegrating tablets (ODTs).3,4) In addition, the ODTs need to be prepared at higher compression for preventing breakage during delivery. The increase in hardness leads to a decreased disintegration time, even with increased amount of disintegrants.5) ODFs also have the advantage of their simple manufacturing process as compared to other dosage forms.6)

Generally, active pharmaceutical ingredients (APIs), poly-mers, and plasticizers are incorporated into the films with areas ranging from 5 to 20 cm2. API is incorporated in the matrix of hydrophilic polymer. Plasticizers reduce the glass transition temperature of polymers increasing flexibility, workability, and spreadability of films.7,8) Numerous types of hydrophilic polymers such as hydroxypropyl methylcellulose (HPMC), pullulan, maltodextrin, polyvinyl alcohol (PVA), and PVA–polyethylene glycol (PEG)-graft-copolymer have been widely studied due to their high chemical stability, biocom-patibility, and low toxicity.9–17) Generally, the polymer con-centration used in preparing ODFs is around 45% w/w; it can increase up to 60–65% (w/w) in order to attain films with the

desired quality attributes and characteristics. The concentra-tion of plasticizer usually ranges from 0 to 20% (w/w). PEG, glycerol, diethyl phthalate, triethyl citrate, tributyl citrate, and propylene glycol (PG) are common plasticizers.18)

There are critical quality attributes (CQAs) of ODFs that should be carefully considered during their development. These properties are inherent to the formulation but are also significantly influenced by the manufacturing process.19) Physical strength is one of the most evident CQAs. The prod-uct should have suitable mechanical properties for easy manu-facturing, packaging, and handling and minimal damage.20,21) The physical appearance of the films is another relevant CQA. The thickness, size, and shape of the film should be carefully studied and selected depending on the strength and applica-tion. In addition, the target drug release profile should be based on the target product profile and defined early in the development. The most reliable tests available for this evalua-tion are the disintegration time and the dissolution profiles.18) Stability, residual water content, organoleptic characteristics, and dose uniformity should also be considered during the development.

A thorough and validated statistical experimental design is becoming essential in the development of pharmaceutical formulations, as it requires less experimentation and provides reasonable estimates of the relative significance of the differ-ent processes and formulation variables. Even though a couple of studies have reported the development of films with optimal properties based on different statistical designs,22–25) a facto-

* To whom correspondence should be addressed. e-mail: [email protected]; [email protected]

Vol. 65, No. 2 (2017) 167Chem. Pharm. Bull.

rial design can be beneficial to simultaneously elucidate the effect of the individual variables and their interactions at sev-eral levels with minimum experimentation.

Sildenafil citrate (SC) is used, as a model drug, to treat erectile dysfunction by inhibiting guanosine 5′-cyclic mono-phosphate (cGMP)-specific phosphodiesterase type 5 (PDE5), an enzyme that promotes degradation of cGMP.26,27) As the dosage forms of SC are mainly tablets, patients may feel un-comfortable to swallow them. Moreover, the bioavailability of SC may vary depending on the dosage form. In this study, SC-loaded ODFs using HPMC was developed and a factorial design was used to obtain the optimized formulations. The three factors were selected as independent variables with three levels. The tensile strength, disintegration time, water contact angle, and thickness were selected as dependent variables. In addition, for the consideration of CQAs during the develop-ment process, the optimized formulations were validated and characterized by numerous physical properties, in vitro dis-solution properties, and morphological properties.

ExperimentalMaterials SC was supplied by DongKoo Bio&Pharma

Co., Ltd. (Hwasung, South Korea). Hydroxypropyl methylcel-lulose (HPMC 400) was obtained from Shin-Etsu Chemical Co., Ltd. (Tokyo, Japan). PG and PEG 300 were purchased

from BASF AG (Ludwigshafen, Germany). Glycerin, car-boxymethyl cellulose sodium (CMC-Na), and absolute ethanol (HPLC grade) were purchased from Sigma-Aldrich Co. (St. Louis, MO, U.S.A.). All other reagents were of analytical or HPLC grade and were used as received.

Preparation of Drug-Loaded and Blank ODFs Table 1 shows the detailed formulation compositions of ODFs. The films were prepared by a solvent-casting method.28,29) An ac-curately weighed HPMC (500 mg) was dispersed in desired amounts of water and ethanol while stirring until the solu-tion became transparent. Subsequently, a desired amount of plasticizer and CMC-Na (5 mg) was added into the solution and blended continuously; the solution was used as the blank ODF solution. Then, accurately weighed drug (90 mg) was uniformly blended in the blank solution while stirring for 1 h. The solution was sealed to prevent evaporation. The prepared solution was carefully poured onto a teflon plate (6×6 cm) to form a uniform liquid layer, and then dried in an oven at 60°C for 24 h. The resulting films were carefully removed and cut into strips of dimensions 2×2 and 2×6 cm2. The loaded drug amount in the film was 2.5 mg/cm2. The film strips were stored in a sealed plastic bag until use. The preparation of blank ODFs was the same as that for drug loaded ODFs, ex-cept for the drug.

Experimental Design A factorial design was used to

Table 1. Designed Experimental Formulations of Blank ODFs with Three Input Independent Variables and One Fixed Variable (%, w/w) and Their Results on the Various Physical Properties

Run order

Independent variables Dependent variables

x1 x2 x3 y1 y2 y3 y4

Ethanol ratio (%, w/w of water)

Plasticizer ratio (%, w/w of HPMC)

Plasticizer type

Tensile strength (MPa) Mean±S.D.

Disintegration time (s) Mean±S.D.

Water contact angle (°) Mean±S.D.

Thickness (mm) Mean±S.D.

1 50 10 PEG300 6.03±0.013 256±3.2 29.62±1.7 0.296±0.0152 10 10 PEG300 6.19±0.015 295.5±3.6 33.24±1.4 0.334±0.0143 30 1 PEG300 16.30±0.023 131.5±1.8 34.88±1.1 0.058±0.0134 10 1 PG 15.47±0.033 126±2.1 34.61±2.1 0.05±0.0175 10 1 PEG300 15.03±0.025 102.5±1.5 33.77±2.8 0.06±0.0226 30 10 PG 13.04±0.033 357.5±2.5 37.82±2.1 0.074±0.0247 10 19 PEG300 3.50±0.022 335±4.5 33.18±3.1 0.53±0.0548 50 1 Glycerin 13.66±0.045 101±1.6 40.51±1.7 0.07±0.0269 50 10 PG 1.66±0.055 301±2.2 18.88±1.4 0.284±0.066

10 10 19 Glycerin 5.01±0.053 212±1.8 30.45±1.9 0.532±0.05811 50 19 PG 1.79±0.011 294±1.4 12.71±1.1 0.364±0.02612 30 19 PEG300 3.21±0.032 327±3.3 29.86±1.2 0.522±0.04313 50 19 PEG300 3.28±0.018 304±4.2 26.69±1.5 0.518±0.03514 10 19 PG 1.21±0.008 158±2.6 19.40±2.4 0.424±0.01615 50 19 Glycerin 4.65±0.015 251.5±1.9 36.18±1.9 0.592±0.01316 10 10 Glycerin 7.28±0.022 112.5±1.4 30.48±2.6 0.292±0.03617 10 1 Glycerin 14.99±0.037 116.5±1.6 33.98±2.5 0.056±0.05418 30 1 PG 15.87±0.029 140±2.1 35.53±2.1 0.06±0.04719 30 19 Glycerin 5.43±0.010 172±2.0 28.66±2.2 0.548±0.06420 50 10 Glycerin 6.90±0.013 163.5±3.3 31.02±3.1 0.29±0.07521 10 10 PG 1.69±0.009 266±4.5 30.55±3.3 0.252±0.04722 50 1 PEG300 14.11±0.036 129±2.0 30.98±2.8 0.066±0.06823 30 1 Glycerin 11.79±0.029 117±1.8 40.42±2.9 0.068±0.05624 30 10 Glycerin 6.75±0.011 195±1.9 27.17±1.9 0.28±0.04625 30 19 PG 2.20±0.010 214.5±1.6 24.57±1.7 0.286±0.02126 30 10 PEG300 5.50±0.014 247±2.6 34.07±1.6 0.276±0.03227 50 1 PG 13.88±0.013 110±1.1 29.09±1.3 0.044±0.022

168 Vol. 65, No. 2 (2017)Chem. Pharm. Bull.

optimize formulations. Three factors were chosen as indepen-dent variables (ethanol ratio (x1), plasticizer ratio (x2), and type of plasticizer (x3)). The effect of the factors on the mechanical properties of ODF was evaluated at three levels. The experi-mental matrix for the trials is presented in Table 1. The tensile strength (y1), disintegration time (y2), water contact angle (y3), and thickness (y4) were selected as dependent variables that were processed by the Design-Expert® software (version 10; Stat-Ease, Inc., MN, U.S.A.). The best-fit mathematical model was selected on the basis of the comparison of several statisti-cal parameters, including the multiple correlation coefficient (R2), adjusted multiple correlation coefficient (adjusted R2), and the predicted residual sum of square (PRESS). PRESS indicates how well the model fits the data, and should be low in the chosen model relative to the other models.30) The 3-D response surfaces were generated using the software. The op-timized formulation was identified by means of the simultane-ous optimization method. In this method, an individual desir-ability function (di) is generated by converting each response, which can be varied across the range 0<di<1. Consequently, the design variables are selected to maximize overall desir-ability, as shown in Eq. 1,

1/1 2 3 4( ) m

mD d d d d d= (1)

where D is overall desirability, di is individual desirability, and m is the number of responses. The goal was to improve the formulation to attain the maximum tensile strength (with a lower limit ≥1.2 and upper limit ≤16.3, the minimum water contact angle (with a lower limit ≥12.7 and upper limit ≤40.5), the minimum thickness (with a lower limit ≥0.04 and upper limit ≤0.59), and the minimum disintegration time (with a lower limit ≥10.1 and upper limit ≤35.7).

Measurement of Tensile Strength Tensile strength of films was evaluated using a Texture Analyzer™ (Stable Micro Systems Ltd., Surrey GU7 1YL, U.K.) equipment with a 5 kg load cell. ODF was cut into narrow strips (2×6 cm2) and then placed between two clamps (TA-96B, Stable Micro Systems Ltd.) positioned at a distance of 10 mm. Tensile strength was determined in tension mode. During the measurement, the lower clamp was fixed whereas the strip was pulled by the upper clamp at a rate of 1 mm/s. The force at the moment of breakage was recorded and tensile strength was calculated by using the following Eq. 2.

2

Tensile strength (TS)Force at break (N)

Initical cross sectional area of the sample (mm )−=

(2)

Measurements were obtained in triplicate and the results were expressed as the mean value.

Measurement of Disintegration Time In vitro disintegra-tion of the ODFs was evaluated in a glass beaker with 100 mL distilled water with magnetic stirring (300 rpm) at 37±5°C. The disintegration time of 2×2 cm2 samples was measured and expressed as the mean value of five experiments.

Measurement of Water Contact Angle Water contact angle measurements were performed using a previously-reported method.31) The strip samples (2×2 cm2) were fixed on a glass slide and drops of distilled water (50 µL) were placed on the film at three different points. Images were captured by

a digital camera and analyzed by Surftens software (version 4.5; OEG GmbH, Frankfurt, Germany), which fitted the drop profile and determined the contact angle.

Measurement of Film Thickness The thickness of the films was measured with a micrometer caliper with a preci-sion of 0.001 mm (Mitutoyo, Japan). The measurement was taken at three different points on the sample. The average of three readings was taken as mean thickness.

Scanning Electron Microscopy (SEM) A scanning elec-tron microscope (Model S-4700, Hitachi, Japan) was used to characterize the morphology and transverse section of the optimized ODFs. The samples were coated with gold under vacuum in an argon atmosphere prior to observation. The photomicrographs of the samples were taken at an acceleration voltage of 15 kV at different magnifications.

Differential Scanning Calorimetry (DSC) DSC mea-surements were performed using a DSC (Model Q2000, TA Instruments, U.S.A.). Samples (2 mg) were accurately weighed and sealed in aluminum pans. An empty aluminum pan was used as a reference. The samples were heated at a scanning rate of 10°C/min from 10 to 250°C under a dry nitrogen gas.

Fourier Transform-Infrared (FT-IR) Spectroscopy The spectra of the samples, including the SC, the blank film, and the optimized ODFs, were characterized using an FT-IR spec-trometer (Model Nicolet™ iS™ 5 FT-IR spectrometer, Thermo Scientific™, U.S.A.) equipped with a diamond crystal. For each sample, 32 scans were collected at a resolution of 4 cm−1 over the wavenumber region 4000–650 cm−1.

Surface pH Measurement Generally, the pH value of ODF should be neutral to avoid the potential of oral mucosa irritation.32) The surface pH of optimized ODFs was evalu-ated according to the method described by Bottenberg et al.33) An agar plate was prepared by dissolving 2% agar in warm phosphate buffer (pH 6.8) with stirring, and then pouring the solution into a Petri dish until it gelled at ambient tempera-ture. The film samples were placed on the agar plate to swell. The pH value was estimated by touching the electrode to the surface of the sample for 5 min to allow equilibration. The av-erage of three readings was taken as the mean pH value.

In Vitro Dissolution Study In vitro drug release tests were carried out according to the USP 39 Apparatus 2 guide-lines (paddle method) (Varian 705 DS; Varian, Cary, NC, U.S.A.) with 100 mL of dissolution medium maintained at 37±0.5°C and a paddle speed of 50 rpm. Each sample was fixed on the glass disk with a cyanoacrylate adhesive (Sigma-Aldrich Co.). The disk was placed on the bottom of the dis-solution vessel. The dissolution medium used was simulated intestinal fluid (pH 6.8, 50 mM phosphate buffer) without en-zymes. Samples were taken at predetermined time intervals and analyzed for drug content using an HPLC system.

Mathematical Model for Drug Release from the Opti-mized ODFs A “shrinking core” model was modified to describe the drug release profiles from the optimized ODFs.34) Based on the Fig. 1, the following assumptions were applied to develop the model: (a) drug diffusion from the edges of film is ignored on the model, (b) solubility of drug in the film is appreciable compared with the rest of the components, (c) there exists a surface layer near the film surface, (d) dissolu-tion yields a drug-free solid phase composed of the polymer on the outer layer of the film, (e) as dissolution proceeds, the drug-free solid phase moves towards the center of the film.

Vol. 65, No. 2 (2017) 169Chem. Pharm. Bull.

The dissolution of drug consists of the following three steps: (i) Diffusion of drug molecules from the solid–liquid interface through the drug-free portion of the film to the solid–liquid interface. (ii) Diffusion of drug molecules from the solid–liq-uid interface through the surface layer to the outer boundary of the surface layer. (iii) Diffusion of drug molecules from the outer boundary of the surface layer to bulk liquid phase. (f) The concentration of drug on the solid–solid interface is at saturation. (g) The diffusion rate of drug is much greater than the rate of moving of the solid–solid interface towards the center of the film (pseudo-steady state). On the basis of these assumptions, a drug balance over the drug-free portion of the film and the surface layer region yields, respectively:

1 c 0,

C CD l l lt l l

∂ ∂ ∂ < < ∂ ∂ ∂ =

(3a)

2 0 0,

C CD l l l δt l l

∂ ∂ ∂ < < ∂ ∂ ∂ = +

(3b)

where C denotes the concentration of drug and t is time. D1 and D2 are the effective diffusivity of solute in the solute-free solid phase and the diffusivity of solute in the surface layer, respectively. l0 and lc are the initial distance from the center of the film and the initial core distance from the center of the film, respectively. δ is the thickness of the surface layer, and l is the distance measured from the center of the film. The diffusion rate of solute is much greater than the rate of mov-ing of the solid–solid interface towards the center of the film (pseudo-steady state). In this case, Eqs. 3a and b reduce to, respectively,

1 c 00 ,CD l l l

l l∂ ∂ < < ∂ ∂

= (4a)

2 0 00 ,CD l l l δ

l l∂ ∂ < < ∂ ∂

= + (4b)

Integrating Eqs. 4a and b, subject to the conditions C=Ce at l=lc, C=Cs at l=l0, and C=Cb, at l=l0+δ, where Ce denotes the concentration of drug at the dissolving front, Cs at the solid–liquid interface, and Cb at the bulk solution, we obtain:

e sc 0

c 0,

C C Cl l l

t l l∂ −

< <∂ −

= (5a)

s b0 0

c 0,

( )C C C

l l l δt l l δ

∂ −< <

∂ −= +

+ (5b)

At the solid–solid interface, we have

0 0

1 2d dd dl l l l

C CSD SD

l l= =

=

(6)

where S denotes the surface area of the film. By referring to Eqs. 5a and b, this expression becomes

e s s b

1 2c 0 c 0( )C C C C

D Dl l l l δ

− −− −

=+

(7)

Solving this expression for Cs, we obtain

e b c 0

sc 0

( )( )

ηδC C l lC

ηδ l l− −− −

= (8)

where η=D2/D1. At pseudo-steady state, we have

1d d

2 constantd dm C

SDt t= = (9)

where m denotes the number of moles of drug contained in the film. Integrating Eq. 9 with respect to l, subject to the bound-ary conditions C=Ce at l=lc and C=Cs at l=l0, yields

c 0 1 e sd

( ) 2 ( )dm

l l SD C Ct− −= (10)

Substituting Eq. 8 into this expression and noting that m=2Slcρε, where ρ is the density of the original film compos-ite, ε is the drug solid fraction, we have

c b e c 0

c 01 c 0

d ( )( )( )

d ( )ρε l C C l l

l lD t ηδ l l

− −−

− −= (11)

In addition, at any time, there is a simple relationship between the drug’s concentration in bulk solution Cb and lc that is de-scribed by following equation:

b 0 c 02

( )Sρε

C l l CV −= + (12)

where V denotes the volume of the bulk liquid phase. Param-eters used in these equations were either based on the initial film properties or obtained by fitting the experimental results.

HPLC Analysis The drug content was analyzed by an Agilent 1100 Series HPLC system (Agilent Technolo-gies, Santa Clara, CA, U.S.A.) at a wavelength of 292 nm. Samples were collected in Eppendorf tubes and centrifuged for 1 min, and then 20 µL of supernatant of each sample was injected into the HPLC system. A Phenomenex-C18 5-µm (4.6×150 mm; Agilent Technologies) column was used and maintained at about 30°C. Acetonitrile–water–triethylamine at a volume ratio of 32 : 68 : 0.5 adjusted to pH 3.1 was used as the mobile phase with its flow rate of 1.0 mL/min.

Fig. 1. A Schematic Representation of the Optimized ODFs during the Dissolution Test

170 Vol. 65, No. 2 (2017)Chem. Pharm. Bull.

Results and DiscussionExperimental Design A 33 full factorial design was used

to evaluate the main effect of the independent variables on the dependent ones. The experimental matrix for the 27 trials is presented in Table 1. The physical evaluations suggested that all the prepared films had a non-sticky, clear, and uniform surface. The mathematical response surface models were gen-erated by applying coded values of factor levels. The random behavior of the residuals was studied with the residual analy-sis, and normal probability plots were prepared for residual errors of the response variables (Fig. 2). The plots appeared to be normally distributed and resembled straight lines; out-lier points were not observed. In a mathematical model, R2 represents the coefficient of determination and the universal variation in the data accounted by the model. Adjusted R2 pro-vides an estimation of amount of variation around the mean explained by the model, and is adjusted for a specific number of parameters. Predicted R2 represents the amount of varia-tion in new data explained or predicted by the mathematical model. The values of the coefficients x1, x2, and x3 are related to the effects of these variables on the response. A positive coefficient represents a synergistic effect, whereas a negative coefficient indicates an antagonistic effect on the response; moreover, a greater coefficient indicates that the independent variable has a stronger effect on the response.30,35,36)

Tensile Strength (y1) Tensile strength may be one of the most important parameters for developing a film because it

significantly influences dosage form stability and in vivo ef-ficacy.17,37) It was evaluated with a texture analyzer. The re-duced quadratic mathematical model equation in coded terms for tensile strength is provided in Eq. 13.

2

1 2 26.12 5.60 2.85y x x−= + (13)

As shown in Table 1, the tensile strength in the different film formulations ranged from 1.21 and 16.30 N. As shown in Table 2, a p value <0.05 for any factors in ANOVA represents a significant effect of the corresponding factors on the tensile strength (y1). It can indicate that the terms x2 and x2

2 have a significant effect on the tensile strength (p<0.05). The actual model R2, predicted R2, and adjusted R2 are 0.8337, 0.7896, and 0.8199, respectively. It can be inferred that their closeness is suggestive of goodness of fit in the data. The high value of adequate precision (14.885) denotes adequate model discrimi-nation. The effect of independent variables on tensile strength is shown in the response surface mapping graph (Fig. 3). The lower plasticizer ratio levels increased the tensile strength of film. The ethanol ratio and the type of plasticizer may not have a significant effect on the tensile strength. Generally, the primary role of plasticizers as low molecular weight non-vola-tile additives is to improve the flexibility and processability of polymers by lowering the second order transition temperature (glass transition temperature, Tg).25) When incorporated into a polymeric material, a plasticizer improves the flexibility of

Fig. 2. Residual Plots for (a) Tensile Strength, (b) Disintegration Time, (c) Water Contact Angle, and (d) Film Thickness

Vol. 65, No. 2 (2017) 171Chem. Pharm. Bull.

the polymer by increasing the intermolecular separation of the polymer molecules. This results in a reduction in elastic modulus, tensile strength, polymer melt viscosity, and Tg. The polymer flexibility is improved and lower thermal process-ing temperatures can be used.38,39) Based on the results, the

strongest tensile strength of film was attained at the highest concentration of plasticizer, regardless of type of plasticizer and ethanol ratio.

Disintegration Time (y2) The disintegration time of film is one of the most important quality attributes for evaluating

Table 2. ANOVA of the Responses of Full Factorial Experimental Design

Responses Source Sum of squares DF* Mean square F Value p Value

y1

Model 613.52 2 306.76 60.17 <0.0001x2 564.82 1 564.82 110.79 <0.0001x2

2 48.69 1 48.69 9.55 0.0050Residual 122.36 24 5.10Cor total 735.88 26

y2

Model 1.282E+005 4 32051.34 13.16 <0.0001x2 79268.35 1 79268.35 32.54 <0.0001x3 28656.24 2 14328.12 5.88 0.0090x2

2 20280.78 1 20280.78 8.32 0.0086Residual 53594.87 22 2436.13Cor total 1.818E+005 26

y3

Model 478.24 3 159.41 6.55 0.0023x2 288.49 1 288.49 11.86 0.0022x3 189.76 2 94.88 3.90 0.0348

Residual 559.48 23 24.33Cor total 1037.73 26

y4

Model 0.85 3 0.28 92.53 <0.0001x2 0.80 1 0.80 259.79 <0.0001x3 0.055 2 0.027 8.90 0.0014

Residual 0.070 23 3.062E-003Cor total 0.92 26

* Degrees of freedom.

Fig. 3. Response Surface Plots Correlating a Dependent Variable (y1: Tensile Strength) with Independent Variables (x1: Ethanol Ratio and x2: Plasti-cizer Ratio)

Independent variable (x3: type of plasticizer) was presented with (a) glycerin, (b) PG, and (c) PEG 300. The black circle points and gray circle points represent the design points above predicted value and below predicted value, respectively.

172 Vol. 65, No. 2 (2017)Chem. Pharm. Bull.

the films. The U.S. Food and Drug Administration (FDA) recommends the United States Pharmacopeia (USP) method for disintegration testing for ODTs. In addition, other methods that can be correlated with or are demonstrated to provide re-sults equivalent to the USP method can also be used to deter-mine disintegration time.17) In this study, 100 mL dissolution medium was used, which is not a real simulation of oral dis-integration. However, this condition may be suitable to screen for the optimal composition of ODF with a relatively short disintegration time.17) The reduced linear mathematical model in coded terms for disintegration time is provided in Eq. 14.

2 2 321.22 7.13 5.22y x x− −= (14)

As shown in Table 1, the disintegration times of the differ-ent film formulations ranged from 10.1 to 35.7 s. As shown in Table 2, p value <0.05 for any factors in ANOVA represents a significant effect of the corresponding factors on the dis-integration time (y2). It can indicate that the terms x2 and x3 have a significant effect on the disintegration time (p<0.05). The actual model R2, predicted R2, and adjusted R2 are 0.7127, 0.5982, and 0.6752, respectively. It can be inferred that their closeness is suggestive of goodness of fit in the data. The value of adequate precision (12.840) denotes adequate model discrimination to navigate the design space. The effect of independent variables on disintegration time is shown on the response surface mapping graph in Fig. 4. Lowering the plas-ticizer ratio levels increased the disintegration time and the type of plasticizer had a significant effect on the disintegration time. The ethanol ratio may not be so significant. This indi-

cated that the hydrophilic plasticizers could cause an increase of water diffusion into the polymer; thus, the disintegration time of the films changed with the change in plasticizer ratio. Plasticizers are arranged among and between polymer chains, disrupting hydrogen bonding and increasing the free volume. Thus, water permeability of the films containing plasticizer are increased and disintegration time is decreased.40) In par-ticular, the glycerin-loaded films tend to have a shorter disin-tegration time, because the low molecular weight glycerin has a large number of molecules.41)

Water Contact Angle (y3) In general, the intrinsic water contact angle of a hydrophilic material is smaller than 90°, while a hydrophobic material has an intrinsic contact angle larger than 90°.39) Based on the above, affinity of the film sur-face to solvent could be evaluated by the contact angle, which may show the rate of water sorption into the film.37) The re-duced linear mathematical model equation in coded terms for water contact angle is provided in Eq. 15.

3 2 330.68 4.00 2.53y x x−= + (15)

As shown in Table 1, the contact angle in the different film formulations ranged from 12.7149 to 40.5146°. A p value <0.05 for any factor in ANOVA represents a significant effect of the corresponding factors on the contact angle (y3) (Table 2). It indicates that the terms x2 and x3 may have a significant effect on the contact angle (p<0.05). The actual model R2, predicted R2, and adjusted R2 are 0.9218, 0.8162, and 0.9025, respectively. It can be inferred that their closeness is sug-gestive of goodness of fit to the data. The value of adequate

Fig. 4. Response Surface Plots Correlating a Dependent Variable (y2: Disintegration Time) with Independent Variables (x1: Ethanol Ratio and x2: Plasticizer Ratio)

Independent variable (x3: type of plasticizer) was showed (a) glycerin, (b) PG, and (c) PEG 300. The black circle points and gray circle points represent the design points above predicted value and below predicted value, respectively.

Vol. 65, No. 2 (2017) 173Chem. Pharm. Bull.

Fig. 5. Response Surface Plots Correlating a Dependent Variable (y3: Water Contact Angle) with Independent Variables (x1: Ethanol Ratio and x2: Plasticizer Ratio)

Independent variable (x3: type of plasticizer) was showed (a) glycerin, (b) PG, and (c) PEG 300. The black circle points and gray circle points represent the design points above predicted value and below predicted value, respectively.

Fig. 6. Response Surface Plots Correlating a Dependent Variable (y4: Thickness) with Independent Variables (x1: Ethanol Ratio and x2: Plasticizer Ratio)

Independent variable (x3: type of plasticizer) was showed (a) glycerin, (b) PG, and (c) PEG 300. The black circle points and gray circle points represent the design points above predicted value and below predicted value, respectively.

174 Vol. 65, No. 2 (2017)Chem. Pharm. Bull.

precision (9.479) denotes adequate model discrimination. The effect of independent variables on contact angle is shown in the response surface mapping graph in Fig. 5. Decreasing the plasticizer ratio levels increased the contact angle. The type of plasticizer had an effect on the contact angle while the ethanol ratio may not be so significant.

Thickness of the Film (y4) The thickness may be help-ful in determining the uniformity of the film. Moreover, this may be directly related to the accuracy of the dose.25,38) The reduced linear mathematical model equation in coded terms for tensile strength is provided in Eq. 16.

4 2 30.27 0.21 0.035y x x= + + (16)

As shown in Table 1, the thickness of the different film for-mulations ranged from 0.044 to 0.592 mm. As shown in Table 2, a p value <0.05 for any factor in ANOVA represents a significant effect of the corresponding factors on the thickness (y4). It can indicate that the terms x2 and x3 have a significant effect on the thickness of film (p<0.05). The actual model R2, predicted R2, and adjusted R2 are 0.9235, 0.8952, and 0.9135, respectively. The high value of adequate precision (24.384) denotes adequate model discrimination. The effect of indepen-dent variables on thickness of film is shown in the response surface mapping graph in Fig. 6. Decreasing the plasticizer ratio levels decreased the thickness of film. The type of plas-ticizer had a significant effect on the film thickness, while the ethanol ratio may not have. In addition, the uniformity of the films could be determined by the value of the standard devia-tion of the measured thickness.

Optimization of Drug Loaded ODFs Generally, the purpose of optimizing pharmaceutical formulations is to determine ideal values of variables for the production of a formulation with desirable characteristics.42) After generating statistically validated model equations to interpret the relation-ship between independent and response variables, the formu-lation is optimized to give the most desirable response values: the contact angle, disintegration time, and thickness were minimized, while the tensile strength was maximized. These properties may present a high release property of the drug from the film with sufficient tensile strength. Based on these

conditions, the four responses were then combined to generate an overall optimum region. The optimal settings are presented in Table 3. Ethanol ratio was 30% at all optimized formula-tions. Plasticizer ratio of glycerin, PG, and PEG 300 was 3.01, 3.14, and 6.35%, respectively. Values of desirability (0.746, 0.646, and 0.696) closer to 1 indicate that the response values are accordingly closer to the target values. These optimal settings were validated with the target values of dependent variables. Table 3 provides biases and relative bias percentages between the optimal solutions and the target values. Based on the results, it can be suggested that optimal solutions and tar-get values were similar with small differences.

In Vitro Characterization of the Optimized ODFs The interactions between the drug and components in ODFs were evaluated with FT-IR. Figure 7 shows the FT-IR spectra of SC, blank ODF, and drug-loaded ODF. The drug had ab-sorption bands in the region of 500–1800 cm−1; the absorp-tion bands in the region of 2200–3500 cm−1 may represent a molecular vibration mode. The bridging of the two different ring systems (pyrazolopyrimidine and piperazine) could be identified with the peak at 1245 cm−1. In addition, there was an

Fig. 7. FT-IR Spectra of (a) SC, (b) Blank ODF, (c) Optimized ODF with Glycerin, (d) Optimized ODF with PG, and (e) Optimized ODF with PEG 300

Table 3. Optimal Solutions, Target Values, and Experimental Results for the Responses

Optimal solutions Dependent variables

x1 x2 x3 y1 y2 y3 y4

EtOH ratio (%)

Plasticizer ratio (%)

Type of plasticiser

Tensile strength (MPa)

Disintegration time (s)

Water contact angle (°)

Thickness (mm)

Desirability

30 3.01 Glycerin 12.43 23.44 37.77 0.15

0.746Target values 12.17 21.53 36.31 0.14

Absolute viases 0.26 1.91 1.46 0.01Relative biases (%) 2.14 8.87 4.02 7.14

30 3.14 PG 11.54 25.43 28.88 0.04

0.646Target values 12.04 27.28 30.07 0.04

Absolute viases 0.50 1.85 1.19 0Relative biases (%) 4.15 6.78 3.96 0

30 6.35 PEG300 8.65 27.85 32.96 0.19

0.696Target values 8.85 28.68 33.43 0.21

Absolute viases 0.20 0.83 0.47 0.02Relative biases (%) 2.26 2.89 1.41 9.52

Vol. 65, No. 2 (2017) 175Chem. Pharm. Bull.

absorption band of the drug at 1686 cm−1 that may represent the carbonyl group of the drug.17) For the optimized ODFs, the corresponding peaks at 1245 and 1686 cm−1 were shifted and absent, respectively. This may be due to the disconnection of the attached ring systems and the existence of hydrogen bond between the carbonyl group of the drug and the hydroxyl group of plasticizer.

The DSC thermograms of blank ODF, physical mixture of SC and blank ODF, optimized ODFs, and SC are shown in Fig. 8. The thermograms showed a strong endothermic peak of SC at about 195°C, which corresponds to the melting point of the drug.43) An obvious endothermic peak was not detected in the optimized ODFs, while a distinct peak appeared in the thermograms of the physical mixture. This may be due to the interactions occurring between SC and the film components. The SEM of optimized ODFs was performed to understand

the surface morphology (Fig. 9). The optimized ODFs showed non-porosity and uniformity in the SEM images of transverse sections and at the film surface. Moreover, the drug particles could be observed in the SEM. The surface pH of optimized ODFs were found to be within the range of 6.5–7.0, which is an almost neutral pH. This may suggest that the optimized ODFs may not irritae the oral mucosa.

In Vitro Dissolution of the Optimized ODFs In order to obtain the dissolution profiles from the optimized ODFs, an in vitro drug release test was carried out. The model used in this study assumes two processes (Phase I and Phase II) during the drug release. Phase I is the hydration of the polymer. During the Phase I, water penetrates, hydrates, and relaxes the poly-mer, which is considered as the dominant factor in the dissolu-tion model.44) Effective drug surface area, which is the portion in actual contact with dissolution medium, increases over time. Phase II starts once the water has completely penetrated through the film matrix and assumes a steady state, while the diffusion of the drug continues. During the Phase II, the ef-fective drug surface area decreases. Based on the experimen-tal data, the dominant step for our dissolution system might be the second step, as it is when the majority of the drug content was released. As shown in Fig. 10, the model fits the second phase of our dissolution system very well. The drug of the op-timized ODFs was dissolved in the dissolution medium even though the drug is poorly water soluble. This may mean that the drug dispersions prepared with polymers could significant-ly enhance the solubility and dissolution rate of poorly-soluble drugs.45,46) The profiles were sharply increased for initial 10 min that represents Phase I in this study. After the initial 10 min, the drug was gradually released from the film sys-tems, which represents the Phase II. Generally, water-soluble excipients such as disintegrating agent in the films allowing water to penetrate into the inner matrix of the film and thus causing fasting disintegration of the films. The disintegrated films increased the contact area of the drug with the medium, which might cause sharp increase of drug release rate for ini-

Fig. 8. DSC Thermograms of Blank ODF, Physical Mixture of SC and Blank ODF, Optimized ODFs, and SC

1-2, 2-2, and 3-2 represent optimized ODFs with glycerin, PG, and PEG 300, respectively.

Fig. 9. SEM Images of (a), (b), and (c) Surface Images; (d), (e), and (f) Transverse Sections of the Optimized ODF with Glycerin, PG, and PEG 300, Respectively

176 Vol. 65, No. 2 (2017)Chem. Pharm. Bull.

tial 10 min. After the initial disintegration, the disintegrated films were continuously hydrated slowly and swelled, causing a thick gel layer. Once a viscous gel layer had formed, it could serve as a barrier to media penetration and decreasing the rate of diffusion of the fluid into the matrix. This might influence on the drug release during the Phase II.

ConclusionIn this study, SC-loaded ODFs were prepared, and their

physical properties were evaluated. A full factorial experi-mental design was used to obtain the optimal formulations in terms of ethanol ratio, plasticizer ratio, and the type of plasticizer. The quantitative effects of different levels of for-mulation variables on the response variables were accurately predicted using polynomial equations, with high linearity observed between the predicted and actual values of the re-sponse variables. The optimized ODFs were obtained using a numeric optimization technique and were characterized using various physical evaluations that should be carefully considered during the development process. The drug was ho-mogenously dispersed throughout the film, and the crystalline form of the drug was observed. The SC-loaded ODFs may be an alternative to conventional tablets for the treatment of male erectile dysfunction. Moreover, in vitro characterization may reflect CQAs during the formulation development.

Acknowledgment This work was supported by the 2015 Inje University research Grant.

Conflict of Interest The authors declare no conflict of interest.

References 1) Panda B., Dey N., Rao M., Int. J. Pharm. Sci. Nanotechnol., 5,

1666–1674 (2012). 2) Takeuchi H., Yamakawa R., Nishimatsu T., Takeuchi Y., Hayakawa

K., Maruyama N., J. Drug Delivery Sci. Technol., 23, 471–475 (2013).

3) Kathpalia H., Gupte A., Curr. Drug Deliv., 10, 667–684 (2013). 4) Siddiqui M. N., Garg G., Sharma P. K., Adv. Biol. Res., 5, 291–303

(2011). 5) Fukami J., Ozawa A., Yoshihashi Y., Yonemochi E., Terada K.,

Chem. Pharm. Bull., 53, 1536–1539 (2005). 6) Bhyan B., Jangra S., Kaur M., Singh H., Int. J. Pharm. Sci. Rev.

Res., 9, 9–15 (2011). 7) Irfan M., Rabel S., Bukhtar Q., Qadir M. I., Jabeen F., Khan A.,

Saudi Pharm. J., 24, 537–546 (2016). 8) Arya A., Chandra A., Sharma V., Pathak K., International Journal

of Chem. Tech. Research, 2, 576–583 (2010). 9) Arya A., Sharma V., Pathak K., Pharm. Dev. Technol., 18, 1329–

1338 (2013).10) Scott R. A., Park K., Panitch A., Eur. J. Pharm. Biopharm., 84,

125–131 (2013).11) Yellanki S., Jagtap S., Masareddy R., J. Young Pharm., 3, 181–188

(2011).12) Sievens-Figueroa L., Bhakay A., Jerez-Rozo J. I., Pandya N., Roma-

ñach R. J., Michniak-Kohn B., Iqbal Z., Bilgili E., Davé R. N., Int. J. Pharm., 423, 496–508 (2012).

13) Mishra R., Amin A., Indian J. Pharm. Educ. Res., 45, 71–77 (2011).14) Kunte S., Tandale P., J. Pharm. Bio. Allied Sci., 2, 325 (2010).15) Patel R., Naik S., Patel J., Baria A., Arch. Pharm. Sci. Res., 1,

212–217 (2009).16) Sakata Y., Otsuka M., Int. J. Pharm., 374, 33–38 (2009).17) Xu L.-L., Shi L.-L., Cao Q.-R., Xu W.-J., Cao Y., Zhu X.-Y., Cui

J.-H., Int. J. Pharm., 473, 398–406 (2014).18) Borges A. F., Silva C., Coelho J. F. J., Simões S., J. Control. Re-

lease, 206, 1–19 (2015).19) Borges J. G., De Carvalho R. A., J. Pharm. Sci., 104, 1431–1439

(2015).20) Preis M., Woertz C., Kleinebudde P., Breitkreutz J., Expert Opin.

Drug Deliv., 10, 1303–1317 (2013).21) Preis M., Knop K., Breitkreutz J., Int. J. Pharm., 461, 22–29 (2014).22) Deepthi A., Reddy B. V., Navaneetha K., Am. J. Adv. Drug Deliv-

ery, 2, 153–163 (2014).23) Pandey G. S., Kumar R., Sharma R., Singh Y., Teotia U., Am. J.

PharmTech Res., 3, 1–17 (2013).24) Panchal M. S., Patel H., Bagada A., Vadalia K., Int. J. Pharm. Res.

Allied Sci., 1, 60–72 (2012).25) Dinge A., Nagarsenker M., AAPS PharmSciTech, 9, 349–356 (2008).26) Boolell M., Gepi-Attee S., Gingell J., Allen M., Br. J. Urol., 78,

257–261 (1996).27) Liu C.-C., Huang S.-P., Wu W.-J., Lee Y.-C., Wang C.-J., Urol. Sci.,

21, 113–117 (2010).28) Cilurzo F., Cupone I. E., Minghetti P., Buratti S., Selmin F., Genna-

ri C. G., Montanari L., AAPS PharmSciTech, 11, 1511–1517 (2010).29) Cilurzo F., Cupone I. E., Minghetti P., Selmin F., Montanari L., Eur.

J. Pharm. Biopharm., 70, 895–900 (2008).30) Huang Y.-B., Tsai Y.-H., Yang W.-C., Chang J.-S., Wu P.-C.,

Takayama K., Eur. J. Pharm. Biopharm., 58, 607–614 (2004).31) Isotton F., Bernardo G., Baldasso C., Rosa L., Zeni M., Ind. Crops

Prod., 76, 717–724 (2015).32) Rathbone M. J., Drummond B. K., Tucker I. G., Adv. Drug Deliv.

Rev., 13, 1–22 (1994).33) Bottenberg P., Cleymaet R., de Muynck C., Remon J. P., Coomans

D., Michotte Y., Slop D., J. Pharm. Pharmacol., 43, 457–464 (1991).34) Hsu W.-L., Lin M.-J., Hsu J.-P., World Acad. Sci. Eng. Technol., 53,

913–918 (2009).35) Huang Y.-B., Tsai Y.-H., Lee S.-H., Chang J.-S., Wu P.-C., Int. J.

Pharm., 289, 87–95 (2005).36) Fouad S. A., Basalious E. B., El-Nabarawi M. A., Tayel S. A., Int. J.

Pharm., 453, 569–578 (2013).37) Sharma R., Kamboj S., Singh G., Rana V., Eur. J. Pharm. Sci., 84,

55–69 (2016).38) Chen M., Tirol G., Schmitt R., Chien C., Dualeh A., 2006 AAPS

Fig. 10. Profiles of the Drug Released from the Optimized ODFs and SC (Pure Drug)

The solid lines represent the results of the simulated fits of the experimental results that are shown as dots. The triangle mark, circle mark, and square mark represent the optimized ODF with glycerin, PG, and PEG 300, respectively. The diamond mark represents SC (pure drug).

Vol. 65, No. 2 (2017) 177Chem. Pharm. Bull.

Annual meetings-posters and papers, p. 2006 (2006).39) Ma Y., Cao X., Feng X., Ma Y., Zou H., Polymer, 48, 7455–7460

(2007).40) Snejdrova E., Dittrich M., “Pharmaceutically Used Plasticizers,

Recent Advances in Plasticizers,” ed. by Lugman M., InTech: ‹http://www.intechopen.com/books/recent-advances-in-plasticizers/pharmaceutically-used-plasticizers-›, 2012.

41) Bourtoom T., Journal of Science and Technology, 30, 149 (2008).42) Basalious E. B., Shawky N., Badr-Eldin S. M., Int. J. Pharm., 391,

203–211 (2010).43) Melnikov P., Corbi P. P., Cuin A., Cavicchioli M., Guimarães W. R.,

J. Pharm. Sci., 92, 2140–2143 (2003).44) Siepmann J., Peppas N., Pharm. Res., 17, 1290–1298 (2000).45) Janssens S., Anné M., Rombaut P., Van den Mooter G., Eur. J.

Pharm. Sci., 37, 241–248 (2009).46) Janssens S., de Armas H. N., Remon J. P., Van den Mooter G., Eur.

J. Pharm. Sci., 30, 288–294 (2007).