Methods for the preparation of amorphous solid dispersions – A comparative study

Accepted Manuscript Methods for the preparation of amorphous solid dispersions – A comparative study Michal Beneš, Tomáš Pekárek, Josef Beránek, Jaroslav Havlíček, Lukáš Krejčík, Michal Šimek, Marcela Tkadlecová, Pavel Doležal PII:

S1773-2247(16)30410-5

DOI:

10.1016/j.jddst.2017.02.005

Reference:

JDDST 299

To appear in:

Journal of Drug Delivery Science and Technology

Received Date: 17 October 2016 Revised Date:

13 February 2017

Accepted Date: 13 February 2017

Please cite this article as: M. Beneš, T. Pekárek, J. Beránek, J. Havlíček, L. Krejčík, M. Šimek, M. Tkadlecová, P. Doležal, Methods for the preparation of amorphous solid dispersions – A comparative study, Journal of Drug Delivery Science and Technology (2017), doi: 10.1016/j.jddst.2017.02.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Physical and chemical characteristics of the solid dispersion samples (left) and their influance on in-vitro performance (right).

ACCEPTED MANUSCRIPT

Methods for the preparation of amorphous solid dispersions – a

2

comparative study

3

Author names and affiliations: Michal Beneš1 [email protected]

SC

Tomáš Pekárek2 [email protected]

Jaroslav Havlíček2 [email protected] Lukáš Krejčík2 [email protected] Michal Šimek2 [email protected]

M AN U

Josef Beránek2 [email protected]

Marcela Tkadlecová2 [email protected] Pavel Doležal1 [email protected]

TE D

4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

RI PT

1

1

Charles University, Faculty of Pharmacy, Heyrovského 1203, Hradec Králové, 500 05, Czech Republic

29

2

Zentiva, k.s., U Kabelovny 130, Dolní Měcholupy-Prague 10, 102 37, Czech Republic

AC C

30

EP

28

31

Corresponding authors:

32

Michal Beneš, phone number: +420 725 382 782, E-mail: [email protected];

33

Tomáš Pekárek, phone number: +420 267 243 660, E-mail: [email protected]

34 35 1

ACCEPTED MANUSCRIPT

ABSTRACT

37

Solid dispersion is considered one of the most successful strategies for improving the dissolution and

38

absorption of poorly water-soluble APIs. The primary focus of this study was to investigate the

39

effects of methods of solid dispersion preparation on the pharmaceutically important

40

physicochemical properties of a product based on Soluplus® as the polymer carrier and febuxostat as

41

the model BCS II API. The methods of preparation evaluated were based on melt and solvent

42

mechanisms. The samples prepared were evaluated by modulated differential scanning calorimetry,

43

solid state NMR, Raman spectroscopy, XRPD, scanning electron microscopy, and BET analysis.

44

Differences among the solid dispersions obtained by the specific methods and their mechanisms

45

were clearly observed in in-vitro testing. The dissolution behaviour was shown to be influenced not

46

only by particle characteristics such as the size and specific surface area, but also by the interactions

47

between the API and the polymer matrix on the molecular level (e.g. non-covalent interactions). The

48

set of measurements showed that similar methods of preparation do not lead to solid dispersion

49

systems of similar characteristics. Surprisingly, similar dissolution profiles were found in samples

50

prepared by very different methods although their physical-chemical characteristics differed

51

significantly, and vice versa.

52

EP

TE D

M AN U

SC

RI PT

36

KEY WORDS

54

Solid dispersion, solid solution, hot-melt extrusion, spray drying, methods of preparation, in-vitro

55

performance

56

ABBREVIATIONS

57

Amorphous Solid Dispersion (ASD); Active Pharmaceutical Ingredient (API), modulated Differential

58

Scanning Calorimetry (mDSC); Specific Surface Area (SSA); solid state Nuclear Magnetic Resonance

59

(ssNMR); Melting temperature (Tm); Glass transition temperature (Tg); Hot-Melt Extrusion (HME);

AC C

53

2

ACCEPTED MANUSCRIPT

Slow Cooling (SC); Bench Cooling (BC); Fast Cooling (FC); Spray Drying (SD); Solvent Evaporation (SE);

61

Rotary Vacuum Evaporation (RVE); Infrared (IR); Raman Spectroscopy (RS); Scanning Electron

62

Microscopy (SEM); Brunauer, Emmett and Teller (BET); Particle Size Distribution (PSD); X-Ray Powder

63

Diffraction (XRPD); Principal Component (PC); Principal Component Analysis (PCA); Partial Least

64

Squares regression (PLS); polyvinyl-caprolactam-polyvinyl acetate-polyethylene glycol graft

65

copolymer (Soluplus®)

RI PT

60

AC C

EP

TE D

M AN U

SC

66

3

ACCEPTED MANUSCRIPT

67

1. INTRODUCTION Dispersions of poorly water-soluble APIs in solid inert hydrophilic polymer matrices, known as solid

69

dispersions, have been established as an efficient approach enhancing the dissolution rate and hence

70

the oral bioavailability of APIs being formulated this way. [1-6] The enhanced dissolution rate

71

characteristic of solid dispersions can generally be attributed to one of the following mechanisms:

72

eutectic formation, increased surface area of the drug due to precipitation in the carrier, solid

73

solution formation and improved wettability due to intimate contact with a hydrophilic carrier,

74

precipitation as a metastable crystalline form, or a decrease in substance crystallinity. Both the

75

properties of the carrier-API composition and the method of production generally can influence the

76

type of the solid dispersion formed, and thus the subsequent behaviour of the solid dispersion. [7]

M AN U

SC

RI PT

68

77

Further aspects of solid dispersions such as classification, methods of production including many

78

details as well as choice of additives, mutual miscibility, mechanisms of API release, and methods of

79

characterisation have been studied extensively and presented in review articles. [5, 8, 9] Despite the intense study of solid dispersions, only a limited number of products are available on

TE D

80 81

the market. The poor predictability of the physical stability of solid dispersions together with the

82

difficult production at the industrial scale were identified as the main reasons. [1, 3] Solid dispersions are commonly prepared by the solvent or hot-melt (fusion) methods. [4, 6] In

EP

83

solvent-free methods, the crystal lattice structure is destroyed by high temperatures and by

85

mechanical stress. However, imperfect amorphization, that is, the presence of remaining nuclei or

86

small crystals, may lead to crystallization after preparation. On the other hand, in the solvent

87

methods, the lattice structure is destroyed completely during the preparation process, but

88

crystallization may be enhanced by the presence of residual solvents. [10]

AC C

84

89

A very efficient solvent evaporation-based technology is spray drying (SD), since it allows for

90

extremely rapid solvent evaporation, leading to a fast transformation of the API to the crystalized

91

and/or amorphized form dispersed within solid carrier particles. An excellent review on all aspects of

92

SD was published recently. [11] 4

ACCEPTED MANUSCRIPT

93

Frequently, the method of choice for manufacturing solid dispersions is hot-melt extrusion

94

(HME), mainly because of the health, environmental, and subsequent financial issues associated with

95

the lack of the use of solvents. [12, 13]. Further ways of preparation have been described, such as thermal adhesion granulation [12],

97

ball milling [14], electrospinning [15], co-precipitation [16], supercritical fluid methods, spray freezing

98

[2], etc. Solid dispersions prepared by different methods can exhibit differences in physicochemical

99

properties, which might affect product performance, including manufacturability. [17] The

RI PT

96

dissolution behaviour of amorphous solid dispersions (ASDs) is known to be affected by many

101

variables, such as the type of polymer, API-polymer ratio and interactions, aqueous solubility of

102

components, wettability, and physical stability of the solid dispersion. [1, 18]The degree of molecular

103

interactions was proven to be dependent on the actual-API-polymer miscibility. [19]

M AN U

104

SC

100

Significantly lower aqueous solubility of solid dispersions prepared by SD compared to the bead milling method was reported despite the smaller particle size of the SD material. The differences

106

were attributed to higher surface energy which results in the aggregation of particles of the SD solid

107

dispersions, which in turn negatively effects aqueous solubility. [20] It has also been reported that

108

the surface of a SD powder is dominated by less soluble material due to its adsorption into the

109

air/liquid interface before they turn into dry particles. [1] Also, Joe and co-authors reported the solubility of solid dispersions to be dependent on the

EP

110

TE D

105

method used for their preparation, in the following order: solvent-evaporation method > solvent-

112

wetting method > surface-attached method [6]. The authors have attributed the significant

113

differences in the solubility of the samples to the size of particles, the contact area between the

114

hydrophilic carrier and the API contained, and to the crystallinity of the API. Different particle

115

morphologies, crystallinities, and dissolution rates were assigned to different mechanisms of particle

116

formation by spray freezing and supercritical antisolvent precipitation methods in the comparison

117

study. [2, 21] The supercritical anti-solvent precipitation method showed higher crystallinities of

118

samples, which resulted in lower dissolution rates. A direct comparison of the solid state

AC C

111

5

ACCEPTED MANUSCRIPT

characteristics of solid dispersions prepared by the HME and solvent co-precipitation methods is

120

presented in a study by Dong [16]. The authors reported the same spectroscopic, XRPD, true density

121

and water sorption/desorption behaviour, but a larger specific surface area of co-precipitated ASD,

122

which resulted in its faster dissolution. Other similar works presented characteristics of HM-extruded

123

and SD solid dispersions varying in surface area, powder densities and flow characteristics, whereas

124

in terms of the interactions between the drug and the carrier, the dispersions prepared by both

125

methods are rather similar. The work concludes by suggesting that similar product performance can

126

be obtained when the physical characteristics of materials are similar. However, differences in

127

material properties may affect the physical stability of the solid dispersion, since the SD solid

128

dispersion was shown to be physically less stable compared to the HME solid dispersion under

129

accelerated stability conditions. [22, 23]

SC

M AN U

130

RI PT

119

The aim to investigate the influence of the manufacturing process used in the preparation of glass solutions on the physicochemical characteristics of products has been described in another

132

paper [14]. The study indicated that the preparation method is of importance in terms of chemical

133

stability and processing, but that the choice of the manufacturing approach appears to have minimal

134

influence on physical stability.

135

TE D

131

The primary focus of this study is to investigate the effects of the solid dispersion method on the physicochemical properties of the product. The same API-polymer primary mixture was used in the

137

preparation of the solid dispersions in order, to eliminate any possible variation in the properties of

138

the solids due to their chemical composition. Therefore, the differences in the properties of solid

139

dispersions can be attributed to the molecular interactions and physicochemical properties induced

140

by the preparation method used.

142 143

AC C

141

EP

136

2. MATERIALS AND METHODS 2.1. Preparation of solid dispersions

6

ACCEPTED MANUSCRIPT

144

Soluplus ® (BASF, Ludwigshafen, Germany) was used as the hydrophilic carrier for the preparation of solid dispersions by the various methods described below. Febuxostat

146

(Zentiva, k.s., Prague, Czech Republic) was used for the experiments as the model BCS II API

147

and it was in pharmaceutical quality. The mixture (API-polymer ratio 1:2 w/w) was mixed for

148

15 min in a small container in a Turbula T50A blender and used for all the experiments. This

149

mixture was also used as a physical mixture sample. For solvent methods, a mixture of

150

febuxostat:Soluplus:ethanol (1:2:18, w/w) was prepared and then used in each case. The

151

ethanol used in the study was 96% in Ph. Eur. quality. Water for injection (Ph.Eur.) was used

152

in the sample preparation. All material gained using the procedures described below was

153

ground in a grinding mortar, sieved manually using a 250 µm sieve, and kept in closed glass

154

bottles.

156

SC

M AN U

155

RI PT

145

2.1.1. Hot-melt extrusion (HME)

The Three-Tec ZE12 twin-screw mini extruder was used for the preparation of the sample. The material was conveyed into the extruder gravimetrically, using the Three-Tec

158

ZD 9 FB-C-1M-80 twin screw feeder. The following settings were used for production of

159

the solid dispersion used for testing in this study: feeding rate 300 g / h, 100 rpm of the

160

screws; 1.5 mm nozzle diameter; temperature set in five separate zones

161

100/135/135/135/135 °C.

163 164 165

EP

2.1.2. Fusion method, subsequent cooling (Fusion, SC)

AC C

162

TE D

157

Approximately 5 g of the drug-polymer mixture (1:2) was placed between two

aluminium plates and put on the heater annealed to 180 °C. Aluminium plates were pressed together and shuffled rotationally in order to produce shear stress. This

166

procedure took 10 minutes. Subsequent cooling was performed by putting the aluminium

167

plates with the material between them in a tray dryer, which was set to cool down from

168

180 °C at the rate of 30 °C per hour until 30 °C was achieved.

7

ACCEPTED MANUSCRIPT

170 171 172 173

2.1.3.Fusion method, bench cooling (Fusion, BC) Another sample was prepared in the same manner as described in paragraph 2.1.2. Bench cooling was performed at ambient temperature on the laboratory bench. 2.1.4. Fusion method, fast cooling (Fusion, FC)

RI PT

169

Another sample was prepared in the same manner as described in paragraph 2.1.2. Fast cooling was performed using a liquid nitrogen bath into which the aluminium plates

175

with the material between them were immersed.

177

2.1.5. Melt-extrudates exposed to vapours of 3% ethanol (HME+3%EtOH) A funnel was inserted tightly into a flask containing an ethanol:water solution

M AN U

176

SC

174

(100 g of 3% (w/w) ethanol), using a hole in the cork. A fine mesh was placed on top of

179

the funnel, onto which the melt-extrudate was spread in a thin layer (approx. 3 mm). The

180

whole apparatus was put onto a hot plate (set temperature 150 °C) for 30 minutes.

181

Finally, the material was removed from the mesh, ground in a grinding mortar and sieved

182

through a 250 µm sieve. The material was put into the ventilated hood and left for 3 days

183

at approx. 25 °C in order to remove possible residual solvents.

186 187 188 189

190

The same procedure as that described in paragraph 2.1.5 was used, except that a

EP

185

2.1.6.Melt-extrudates exposed to vapours of 15% ethanol (HME+15% EtOH)

15% w/w ethanol:water solution was used instead of the 3% one.

AC C

184

TE D

178

2.1.7. Melt-extrudates exposed to vapours of 70% ethanol (HME+70% EtOH) The same procedure as that described in paragraph 2.1.5 was used, except that a

70% w/w ethanol:water solution was used instead of the 3% one.

2.1.8. Spray drying (SD)

8

ACCEPTED MANUSCRIPT

191

The solutions described in paragraph 2.1 were used for SD in a Büchi Mini spray dryer B-290, with the following settings: aspirator 80%; pump 20 min -1, process nitrogen

193

temperature 90 °C, temperature of the inert loop -15 °C.

194

2.1.9. Solvent evaporation method (SE)

RI PT

192

Approximately 50 g of the solution described in paragraph 2.1 was introduced into

196

a beaker and left on the laboratory bench at ambient (approx. 25 °C) conditions until the

197

solvent evaporated completely (approximately 2 weeks).

198

2.1.10. Solvent evaporation at elevated temperature (SE at 50 °C)

The procedure described in paragraph 2.1.9 was used. The beaker with the

M AN U

199

SC

195

200

solution, however, was put in a tray dryer heated to 50 °C. It took approximately 1 week

201

to achieve a complete removal of the solvent using this process.

202

2.1.11. Rotary vacuum evaporation (RVE)

Approximately 50 g of the solution described in paragraph 2.1 was introduced into

204

a beaker, stirred and heated gently until a clear solution was obtained. The solvent was

205

evaporated using a RVE at 45 °C, 20 mbar. Then, the material was dried for 4 hrs at 40 °C

206

in a tray dryer.

208 209 210 211 212 213 214

EP 2.2. Analytical methods used for characterization of solid dispersions

AC C

207

TE D

203

2.2.1.Solid-state nuclear magnetic resonance (ssNMR) 13C CP-MAS ss NMR spectra were measured on a Bruker 400 WB spectrometer

(Bruker Biospin, GmbH, Germany) in 4 mm rotors with a spinning frequency of 13 kHz. 2.2.2.Infrared spectrometry (IR) ATR (ZnSe)-IR spectra were acquired by an FTIR spectrometer Nicolet Nexus (Thermo, USA) with an accumulation of 12 scans with 4 cm-1 spectral resolution. 9

ACCEPTED MANUSCRIPT

216 217 218 219

2.2.3.Raman spectrometry (RS) FT-Raman spectra were acquired by a FT-Raman spectrometer RFS 100/S (Bruker, Germany) with an accumulation of 64 scans with 4 cm-1 spectral resolution. 2.2.4.Modulated differential scanning calorimetry (mDSC)

RI PT

215

Modulated DSC measurements were performed on a TA Instruments, Discovery

DSC. The samples were weighed in aluminium pans with covers (40 µL) and measured in a

221

nitrogen flow. Investigations were performed at a temperature range from 0 °C to 200 °C

222

with a heating rate of 5 °C/min (amplitude = 0.8 °C and period = 60 s). The temperatures

223

specified in relation to mDSC analyses are the temperatures of the peak maxima (Tpeak)

224

and the onset temperature (Tonset) of peaks for the crystalline form and the glass

225

transition temperature (Tg) for the amorphous form. The enthalpy is given in J/g. Sample

226

weight was approximately 3-5 mg.

M AN U

SC

220

AC C

EP

TE D

227

10

ACCEPTED MANUSCRIPT

228 229

2.2.5.Specific surface area Specific surface area was measured on the Nova 2000e instrument (Quantachrome, USA). Desorption took place for 2 hours at 40 °C under vacuum. Sorption of nitrogen was

231

carried out in the liquid nitrogen temperature with a 13-point BET (Brunauer, Emmett and

232

Teller) calculation.

234

2.2.6.Scanning electron microscopy (SEM)

The morphologies were studied using a scanning electron microscopy (SEM)

SC

233

RI PT

230

Mira/Tescan II LM (Tescan a.s., Czech Republic). A small amount of each sample was

236

fixed on an aluminium stub using a double-sided conductive adhesive carbon tape and

237

sputtered with platinum for 240 s at 15mA (Sputter SC7640; Quorum Technologies Ltd,

238

Ashford, UK). The evaluations of the sample particle size distribution were performed

239

with image analysis software (NIS Elements 4.11, LIM – Laboratory imaging spol. s.r.o.,

240

Czech Republic). To assess the number-weighted particle size distributions,

241

approximately 600 particles were evaluated in every analysis, which was found to be

242

sufficient in order to achieve Gaussian distribution for all materials in the investigation.

243

Particle size distribution is often reported as a parameter based on the maximum

244

particle size for the given percentage (10%, 50% and 90%) of the sample. For this reason,

245

PSDs were compared by percentile d-values, which are known as the lower decile d(0.1),

246

median d(0.5), and upper decile d(0.9).

248 249

TE D

EP

AC C

247

M AN U

235

2.2.7.In-vitro drug release The dissolution rate was determined using a Sotax AT-7 dissolution apparatus (Sotax AG, Aesch, Switzerland) with the thermostat set at 37 °C, and equipped with low volume

250

vessels and mini-paddles. Experiments were performed using 150 mL of a 50 mM

251

acetate buffer at pH 4.5. The agitation rate was kept constant at 125 rpm throughout

252

the experiment. The concentration of the dissolved drug was measured on-line, upon

11

ACCEPTED MANUSCRIPT

filtration through 0.7 µm glass microfiber filter (Whatman, GE Healthcare Life Sciences,

254

Buckinghamshire, United Kingdom), using a UV/Vis spectrophotometer Lambda 25

255

(Perkin Elmer, Waltham, MA, USA) at 263 nm. The level of absorbance was checked at

256

600 nm in order to display a potential passage of particles through the filter.

257

Uncompressed powder was accurately weight and introduced into a dissolution vessel.

258

The amount introduced corresponded to approximately 10 mg of the active substance in

259

all experiments.

262

3.1. Solid-state nuclear magnetic resonance (ssNMR)

SC

261

3. RESULTS

M AN U

260

RI PT

253

Solid state NMR spectra were taken from all the solid dispersions prepared in the study. Besides that, the pure crystalline and pure amorphous drug and also physical

264

mixtures thereof with the polymer at the same ratio as in the solid dispersion samples

265

were measured. Selected NMR spectra are presented in Figure 1. Clearly, only two solid

266

dispersion samples contained a crystalline form of the drug, which were the two prepared

267

by the solvent evaporation (SE) method at ambient (25 °C) and elevated (SE at 50 °C)

268

temperatures. The spectra of these samples corresponded to the spectra of the physical

269

mixture of the crystalline drug with the polymer. The spectra of amorphous solid

270

dispersion samples were similar (but not identical) to the spectra of the physical mixture

271

of the amorphous drug with the polymer. Thus, ssNMR can be used for the identification

273 274 275 276

EP

AC C

272

TE D

263

of amorphous solid dispersions within the set of samples prepared in the study. However,

distinguishing single-phase amorphous solid dispersion and two-phase dispersion of

amorphous drug in the polymer matrix requires a combination with another analytical

technique. 3.2. Infrared spectrometry and Raman spectrometry

12

ACCEPTED MANUSCRIPT

Infrared (IR) spectra and FT-Raman spectra (RS) were taken from all the solid

278

dispersions prepared in the study. Besides that, the pure crystalline and pure amorphous

279

drug and also physical mixtures thereof with the polymer at the same ratio as in solid

280

dispersion samples were measured and evaluated, using principal component analysis

281

(PCA, see below). Interaction between the API and the polymer was observed in each of

282

these samples. The API’s carbonyl group was probably dissociated according to the IR and

283

Raman spectra.

SC

284

RI PT

277

A PCA analysis of IR spectra (800 - 1600 cm-1) was performed. The results are presented in Figure 2. The sample prepared by SE at 50°C is separated from other

286

samples, i.e. this spectrum is the most different one from all the measured spectra along

287

with PC1, which carries 94% of the explained variance. The spectra of samples prepared

288

by all other methods are similar according to the PC1. Nevertheless, slight clustering in

289

the PCA scores graph (Figure 2) is visible in the case of fusion- SC, FC and HME + 70% EtOH

290

(one cluster). Furthermore, clustering along both PCs is found in samples of the physical

291

mixture with an amorphous API (phys-mix amorph), solvent evap. at 25 °C (one cluster),

292

HME+3% EtOH, HM extrusion and RVE, Fusion, BC (another one). There are different

293

spectra of samples prepared by SD and phys-mix crystal along PC2. It can be assumed that

294

PC2 is assigned to the crystallinity of the sample (the lower the value of PC2, the higher

295

the crystallinity), whereas PC1 is connected with API:polymer interaction. According to

297 298

TE D

EP

AC C

296

M AN U

285

the loadings graph for the first component (Figure 3), the most significant differences in

the spectra of individual samples occur in the fingerprint region. There are also bands,

whose intensity is attributed to the crystallinity of the samples (bands at approx. 1010,

299

1110 and 1270 cm-1). This can be seen in the comparison of the sample spectra shown in

300

Figure 4. Bands specific for the intensity of the mutual interaction between the API and

301

the polymer are rather less evident (e.g., the band at 1240 cm-1).

13

ACCEPTED MANUSCRIPT

In parallel, a PCA analysis of the Raman spectra was performed. The results are

303

presented in Figure 5. Similar to the IR spectra, the spectra corresponding to samples

304

prepared by SE at elevated temperature were located in the region of the physical

305

mixtures spectra in the scores graph. Interestingly, the spectra of the sample prepared by

306

SE at 50 °C were very close to the spectra of the physical mixture of the drug in

307

amorphous form with the polymer which may be due to the presence of interactions

308

between the polymer and the API. Carbonyl groups and the level of interactions can be

309

better identified in IR spectra, which is likely the reason why the sample prepared by SE at

310

50°C is isolated from the remaining samples in the PCA analysis of IR spectra. Graph

311

loadings (Figure 6) show the contribution of the spectral interval to the differentiation of

312

Raman spectra along the first component. For instance, bands at 1430 and 1450 cm-1 were

313

identified as the regions with the highest contribution to the spectral differentiation along

314

the 1st component. A different intensity band can be observed in this region in each

315

Raman spectrum shown in Figure 7, which could be attributed to the different level of

316

interaction between the components of the samples.

317

3.3. Modulated differential scanning calorimetry (mDSC)

TE D

M AN U

SC

RI PT

302

In the samples prepared by SE and SE at 50 °C, a crystalline fraction was identified

319

in a thermogram being characterised by the melting point (Tm). Thus, those two samples

320

have the character of a 2-phase crystalline solid dispersion. A single glass transition

322 323 324 325

AC C

321

EP

318

temperature (Tg) was observed in all of the remaining solid dispersion samples, which is a clear characteristic of solid solutions (single-phase amorphous solid dispersions). For the

results, see Table 1.

3.4. Specific surface area (SSA) The SSA was measured in all the solid dispersions prepared in the study and the

326

data are summarized in Table 1. The largest specific surface was identified in the samples

327

prepared by SD and in the samples where the HME material was exposed to the vapours 14

ACCEPTED MANUSCRIPT

328

of a 3% and 70% ethanol solution in water. The smallest specific area was observed in the

329

sample prepared by SE at 50°C. Overall, the results are in the range from 0.4 to 2.1 g/m2.

330

3.5. Scanning electron microscopy (SEM) Several micrographs of each sample were taken and investigated for the particles

332

size and morphology (for the results, see Figure 8). The most deviating particle size was

333

observed in the solid dispersion prepared by SD, which contained the smallest particles

334

prepared in the study. On the contrary, the largest particles were identified in solid

335

dispersion samples prepared by RVE. The particles in the remaining individual samples

336

were very similar in magnitude.

SC

M AN U

337

RI PT

331

An image analysis of the micrographs obtained showed that the solid dispersion prepared by SD had the smallest particles in the study. They were nearly perfectly spherical

339

and were better visible at larger magnification (see Figure 9 A). Fine and regular rod-shaped

340

particles were observed on the surface of the material prepared by SE at 50°C (see Figure 9

341

B). Crystal-like particles were observed on the surface of the material prepared by SE (see

342

Figure 9). Particles of all the remaining samples were of irregular shape.

344

3.6. In-vitro drug release

The dissolution rates of pure crystalline and amorphous APIs, physical mixtures

EP

343

TE D

338

thereof and of all the solid dispersion samples were measured. Summary results can be

346

found in Figure 12. No wettability issues were observed during the measurements. No

347 348 349

AC C

345

subsidiary noise was observed at the spectrum baseline which was considered a sign of the

absence of particles passing through the filter. The dissolution method was set intentionally such as to not achieve sink conditions in order to determine effectively the dissolution rate

350

and the kinetic solubility of the API from different materials. Equilibrium solubility was

351

defined in foregoing research which is 0.0044 ng/mL in pH 4.5. This value corresponds to

352

approximately 7% of the API dissolved in the dissolution apparatus. Apparently, 3-4 fold 15

ACCEPTED MANUSCRIPT

super-saturation was achieved in the fastest materials (approx. 25% of the API dissolved)

354

which were prepared by HME and SD. This effect is predominantly attributed to the

355

method used for the preparation of the material since the same polymer was used for all

356

samples. The slowest dissolution rate was shown by the sample prepared by SE.

357

Interestingly, the dissolution rate of the sample prepared by SE at 50 °C, which contained

358

crystalline particles, was comparable with the dissolution rate of the sample prepared by

359

RVE, which was homogeneously amorphous. Physical mixtures containing both amorphous

360

and crystalline API were faster in dissolution than the pure APIs, which shows a certain

361

solubilisation effect of the polymer.

363

3.7. General observations

SC

M AN U

362

RI PT

353

The present study explored the effect of the method used for the preparation of a solid dispersion on the physicochemical profile of the sample and the sample’s in vitro

365

performance. Ten different methods were included in the investigation. All methods used

366

for the preparation of the solid dispersion introduced to the sample a specific set of

367

physical characteristics (see summary in Table 1) and influenced the chemical arrangements

368

of the components of which the solid dispersion consisted. Similarities of the sample

369

observations and their importance and relations to different material characteristics

370

investigated can be seen in the PCA and PLS biplots (Figures 10 and 11). The PCA and PLS

371

(partial least squares regression) analysis is based on data contained in Table 1. The

373 374

EP

AC C

372

TE D

364

dissolution rate seems to be the most related to the surface area and particle size.

However, other characteristics representing the crystallinity of the material and the extent

of interaction between the two components (e.g., Tg, Infrared and Raman spectra) of the

375

samples are also significant, although to a lower extent than SSA and PSD. Single band

376

characteristics were used in the PCA and PLS analyses for IR spectra (1272 cm-1) and Raman

377

spectra (1432 cm-1) as these were identified to have the highest contribution to the

378

differences between individual samples. 16

ACCEPTED MANUSCRIPT

379

In two cases (SE and SE at 50 °C) crystalline solid dispersions were prepared, while all the remaining samples were solid solutions. Among the methods whereby a solid

381

solution was achieved, 3 groups of samples could be anticipated. Within the groups, similar

382

physical (Surface area, PSD, Tg) and chemical (ssNMR, IR spectra and Raman spectra)

383

characteristics were observed, which led to similar in vitro performance of the samples

384

within the groups.

Those groups were characterized first by the mechanism of creating a solid solution

SC

385

RI PT

380

(melt, solvent and exposure of melt towards vapours of the solvent), which likely resulted

387

in similar chemical arrangements of the components of the sample, but also, second, by

388

similar physical characteristics of the powder material. The exposure of melt extruded

389

material to the vapours of ethanol results in an increase of the glass transition temperature

390

and in a significant decrease of the dissolution rate. This can be interpreted such that the

391

methods with solvent-based mechanism lead to a material with a more advantageous

392

chemical arrangement (higher Tg resulting in better stability), whereas its physical

393

properties can be in favour of better solubility (i.e., SD with smaller particles and greatest

394

dissolution profile) compared to the HME material.

TE D

The in vitro performance of the solid solution samples could then be explained on

EP

395

M AN U

386

the basis of a thorough understanding of the overall set of the physical and chemical

397

properties. It is equally important to understand both the physical and the chemical profiles

398 399 400

AC C

396

of the sample, which was demonstrated on the comparison of samples prepared by SD and HME. Those samples were characterised by different chemical arrangements induced by

the method used, different sets of physical properties, but similar performance in vitro.

401

Interestingly, small changes in the chemical arrangement of solid dispersion samples can

402

cause significant changes in the dissolution behaviour. On the contrary, only significant

403

changes in the physical properties can be responsible for such changes in the dissolution

404

behaviour. 17

ACCEPTED MANUSCRIPT

Thus, the differences in terms of the in vitro performance of the samples prepared

406

by different methods could only be explained on the basis of a thorough understanding of

407

both the physical and chemical characteristics of the samples and no discrete analytical

408

method is able to explain such differences. Also, physical parameters (primarily Tg) can only

409

be used for characterising the samples as long as the same method is used for their

410

preparation. When examining samples prepared by different methods, a thorough

411

understanding of their chemical arrangements is also required.

SC

412

414

4. DISCUSSION

M AN U

413

RI PT

405

In the group of all 10 samples prepared by different methods, only two were identified as

415

crystalline solid dispersions (i.e., SE and SE at 50°C). The remaining samples were identified as

416

amorphous solid solutions. These observations were achieved with all the relevant techniques

417

(ssNMR, IR spectrometry, RS, mDSC) in good mutual agreement.

When a solid solution and its Tg is considered, although the differences in the mDSC analyses of

TE D

418

the samples were not substantial, the ascendant trend of Tg identified was used for dividing the

420

samples into three or four groups corresponding to the character of the mechanism of the method

421

used for their preparation. Thus, all samples prepared by the melt-based method (all fusions

422

regardless of the cooling rate applied to the samples) showed a single glass transition temperature

423

(Tg) at about 51 °C. Similarly, the HME sample exhibited Tg at 57 °C, which is a temperature in

424

between the group of samples prepared by exposing the extrudates to ethanol vapours and the

425

group of samples prepared by fusion methods. Finally, the SD sample and the sample prepared by

426

RVE (both solvent based mechanism) was characterized by a Tg of approx. 62°C, which is the highest

427

Tg value obtained. All samples mentioned in the previous sentence were characterized as solid

428

solutions for having a single Tg. The two remaining samples (prepared by SE and SE at 50 °C) were

AC C

EP

419

18

ACCEPTED MANUSCRIPT

429

characterized by the Tg and Tm (of the crystalline drug) values, which is why those two samples were

430

evaluated as solid dispersions (the drug is not molecularly dispersed in the polymer matrix).

431

In terms of specific surface area, no fundamental differences were identified in the set of samples included in the study. This is particularly surprising especially in the case of the SD sample as it was

433

supposed to have the largest surface area (smallest particles of all the samples). Interestingly, the

434

methods of preparation based on the fusion principle (HME, Fusion – FC, BC, SC) yielded

435

approximately a half of the value of the specific surface of the products of methods based on SE

436

(extrudates exposed to vapours of Ethanol, SD).

SC

437

RI PT

432

As mentioned earlier, the lowest PSD was attributed to the sample prepared by SD, whereas the size of particles in the remaining samples was almost equal (all below 140 µm), with the exception of

439

two samples (SE at 50 °C and SD). Large particles occurred more frequently in those two samples.

440

This fact is interesting in particular in the case of the sample prepared by SE at 50°C, which had the

441

largest particles but the greatest specific surface area at the same time.

Another interesting observation was related to the micrographs and mDSC measurement and to

TE D

442

M AN U

438

the samples prepared by SE and SE at 50 °C. As mentioned earlier, both samples had the character of

444

solid dispersions, whilst it was likely the second phase of the dispersion that was observed upon

445

larger magnification on the surface of individual particles (see Figures 9 B and C). In terms of

446

microphotographs, phase separation seemed to occur to a far greater extent in the case of the

447

sample prepared by SE.

Phase separation observed in the samples prepared by SE and SE at 50 °C was manifested

AC C

448

EP

443

449

significantly in the drug release rates. This is particularly valid for the sample prepared by SE, which

450

was the slowest in terms of dissolution (this may correspond to the observation from the SEM – the

451

second phase separated on the surface of bigger particles is observable to a larger extent). On the

452

contrary, the fastest dissolution profiles were observed in the samples prepared by HME and SD. This

453

is very interesting as there are significant differences between those two samples in terms of Tg,

454

surface area and particle size. Also, those samples differed from each other in terms of the PCA 19

ACCEPTED MANUSCRIPT

455

analyses (IR and Raman spectra), which is a sign of the different chemical arrangements of those

456

samples. Interestingly, the PCA analyses of Raman spectra could be interpreted in the same mechanism-

458

based groups mentioned in terms of mDSC measurements. Thus, RVE and SD are located in the same

459

area, the group of extrudates exposed to the vapours of ethanol lay close to the area where the

460

sample prepared by HME was located, which also corresponds to the very close values of Tg. Next to

461

this, the samples in the group prepared by the fusion method (all cooling rates) are located close to

462

each other.

SC

When the PCA analysis of IR spectra is considered, RVE and SD samples were located in the same

M AN U

463

RI PT

457

464

region. Clearly, samples prepared by SE and SE at 50 °C were located in the same region of the PCA

465

analyses (both IR and RS) as the physical mixtures with crystalline API. This was also understood as a

466

confirmation of the results from mDSC, ssNMR and SEM, and a sign of its being a solid dispersion.

467

Another surprising observation was related to the comparison of the samples prepared by SD and HME. Both samples were located in different regions in the PCA analysis (both IR and RS), which is a

469

sign of a different extent of interaction between the components of the two samples. Furthermore,

470

those samples were characterized differently in terms of their physical properties (Tg, specific

471

surface). In spite of those findings, the SD and HME materials behaved similarly when in-vitro

472

performance is considered. In-vitro performance is an indicator of both the physical properties of the

473

material and of the chemical arrangement of the solid dispersion components.

475

EP

AC C

474

TE D

468

5. CONCLUSIONS

476

Unlike other research papers in the field which deal with different methods of preparation and

477

compare, rather, the physical parameters of products, this study provides clear evidence of the

478

high importance of the chemical arrangement of solid dispersion components – the API and the

479

polymer. Each preparation method, regardless of the mechanism on which it is based (melt or 20

ACCEPTED MANUSCRIPT

solvent), introduces a specific set of physical and chemical characteristics to the material

481

produced, which influences the in-vitro behaviour of the material. Thus, similar dissolution

482

profiles were observed in samples prepared by HME and SD. This is very interesting as there are

483

significant differences between those two samples in terms of Tg, surface area and particle size.

484

Also, those samples differed from each other in terms of their IR and Raman spectra. Differences

485

in the chemical arrangement of the samples prepared by different methods can be detected by

486

conventional analytical techniques (IR/RS ideally in combination with PCA interpretation).

487

However, an overall characterization which covers not only the physical but also the chemical

488

properties of the material is required when the manufacturing technology is changed (e.g., in

489

process scale up, technology transfer), since the process parameters of the method used for the

490

preparation (such as the evaporation rate, cooling rate) can provide similar physical

491

characteristics (e.g. PSD, Tg), but substantially different dissolution behaviour due to changes in

492

the chemical arrangement.

493

Therefore, the need of a more thorough understanding of the possible arrangements of the

494

components in the amorphous form should be of interest to future research in the field of solid

495

dispersions in order to achieve constant product quality.

SC

M AN U

TE D

EP

496

RI PT

480

ACKNOWLEDGEMENTS

498

Special thanks to MKS Umetrics for providing software support.

499

This study was supported by specific research project 260 291 of Charles University in Prague.

500

This study was supported by Zentiva, k.s., the Prague Development Site.

501

AC C

497

21

ACCEPTED MANUSCRIPT

REFERENCES

[1] Y. Lu, N. Tang, R. Lian, J. Qi, W. Wu, Understanding the relationship between wettability and dissolution of solid dispersion, Int. J. Pharm. 465 (2014) 25-31.

RI PT

[2] E. Badens, V. Majerik, G. Horváth, L. Szokonya, N. Bosc, E. Teillaud, G. Charbit, Comparison of solid dispersions produced by supercritical antisolvent and spray-freezing technologies, Int. J. Pharm. 377 (2009) 25-34.

SC

[3] G. Verreck, K. Six, G. Van den Mooter, L. Baert, J. Peeters, M.E. Brewster, Characterization of solid dispersions of itraconazole and hydroxypropylmethylcellulose prepared by melt extrusion part I, Int. J. Pharm. 251 (2003) 165-174.

M AN U

[4] L.W. Chiou, S. Riegelman, Pharmaceutical Applications of Solid Dispersion Systems, J. Pharm. Sci. 60 (1971) 1281-1302. [5] C. Leuner, J. Dressman, Improving drug solubility for oral delivery using solid dispersions, Eur. J. Pharm. and Biopharm. 50 (2000) 47-60. [6] J.H. Joe, W.M. Lee, Y.J. Park, K.H. Joe, D.H. Oh, Y.G. Seo, J.S. Woo, C.S. Yong, H.G. Choi, Effect of the solid-dispersion method on the solubility and crystalline property of tacrolimus, Int. J. Pharm. 395 (2010) 161-166.

TE D

[7] E. Broman, C. Khoo, L.S. Taylor, A comparison of alternative polymer excipients and processing methods for making solid dispersions of a poorly water soluble drug, Int. J. of Pharm. 222 (2001) 139-151. [8] A.T.M. Serajuddin, Solid Dispersion of Poorly Water-Soluble Drugs: Early Promises, Subsequent Problems, and Recent Breakthroughs, J. Pharm. Sci. 88 (1999) 1058-1066.

EP

[9] D.Q.M. Craig, The mechanisms of drug release from solid dispersions in water-soluble polymers, Int. J. Pharm. 231 (2002) 131-144. [10] K. Kawakami, Current Status of Amorphous Formulation and Other Special Dosage Forms as Formulations for Early Clinical Phases, J. Pharm. Sci. 98 (2009) 2875-2885.

AC C

502

[11] A. Paudel, Z.A. Worku, J. Meeus, S. Guns, G. Van den Mooter, Manufacturing of solid dispersions of poorly water soluble drugs by spray drying: Formulation and process considerations, Int. J. Pharm. 453 (2013) 253-284. [12] Y.C. Chen, H.O. Ho, J.D. Chiou, M.T. Sheu, Physical and dissolution characterization of cilostazol solid dispersions prepared by hot melt granulation (HMG) and thermal adhesion granulation (TAG) methods, Int. J. Pharm. 473 (2014) 458-468. [13] M. Maniruzzaman, J.S. Boateng, M.J. Snowden, D. Douroumis, A Review of Hot-Melt Extrusion:

22

ACCEPTED MANUSCRIPT

Process Technology to Pharmaceutical Products, Int. Sch. Res. Notices. 2012. [14] J.E. Patterson, M.B. James, A.H. Forster, R.W. Lancaster, J.M. Butler, T. Rades, Preparation of glass solutions of three poorly water soluble drugs by spray drying, melt extrusion and ball milling, Int. J. Pharm. 336 (2007) 22-34.

RI PT

[15] U. Paaver, J. Heinämäki, I. Laidmäe, A. Lust, J. Kozlova, E. Sillaste, K. Kirsimäe, P. Veski, K. Kogermann, Electrospun nanofibers as a potential controlled-release solid dispersion system for poorly water-soluble drugs, Int. J. Pharm. 479 (2015) 252-260.

SC

[16] Z. Dong, A. Chatterji, H. Sandhu, D.S. Choi, H. Chokshi, N. Shah, Evaluation of solid state properties of solid dispersions prepared by hot-melt extrusion and solvent co-precipitation, Int. J. Pharm. 355 (2008) 141-149.

M AN U

[17] S. Sethia, E. Squillante, Solid dispersion of carbamazepine in PVP K30 by conventional solvent evaporation and supercritical methods, Int. J. Pharm. 272 (2004) 1-10. [18] D.E. Alonzo, Y. Gao, D. Zhou, H. Mo, G.G. Zhang, L.S. Taylor, Dissolution and Precipitation Behavior of Amorphous Solid Dispersions, J. Pharm. Sci. 100 (2011) 3316-3331. [19] M. Maniruzzaman, D.J. Morgan, A.P. Mendham, J. Pang, M.J. Snowden, D. Douroumis, Drug– polymer intermolecular interactions in hot-melt extruded solid dispersions, Int. J. Pharm. 443 (2013) 199-208.

TE D

[20] H. Al-Obaid, M.J. Lawrence, S. Shah, H. Moghul, N. Al-Saden, F. Bari, Effect of drug–polymer interactions on the aqueous solubility of milled solid dispersions, Int. J. Pharm. 446 (2013) 100105.

EP

[21] P. Ke, S. Hasegawa , H. Al-Obaidi, G. Buckton, Investigation of preparation methods on surface/bulk structural relaxation and glass fragility of amorphous solid dispersions, Int. J. Pharm. 422 (2012) 170-178.

AC C

[22] A.M. Agrawal, M.S. Dudhedia, A.D. Patel, M.S. Raikes, Characterization and performance assessment of solid dispersions prepared by hot melt extrusion and spray drying process, Int. J. Pharm. 457 (2013) 71-81. [23] S. Guns , A. Dereymaker, P. Kayaert, V. Mathot, J.A. Martens, G. Van den Mooter, Comparison Between Hot-Melt Extrusion and Spray-Drying for Manufacturing Solid Dispersions of the Graft Copolymer of Ethylene Glycol and Vinylalcohol, Pharm. Res. 28 (2011) 673-682. [24] K. Kogermann, A. Penkina, K. Predbannikova, K. Jeeger, P. Veski, J. Rantanen, K. Naelapää, Dissolution testing of amorphous solid dispersions, Int. J. Pharm. 444 (2013) 40-46. [25] K. Yamashita, T. Nakate, K. Okimoto, A. Ohike, Y. Tokunaga, R. Ibuki, K. Higaki, T. Kimura, Establishment of new preparation method for solid dispersion formulation of tacrolimus, Int. J. Pharm. 267 (2003) 79-91.

23

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

503

24

ACCEPTED MANUSCRIPT

0,75056 0,54328 0,73109 0,84089 0,87537 0,94243 0,87676 0,90726 0,85925 0,87158 0,91396 0,78441 0,81651 0,48074 0,85206

0,55808 0,52912 0,1353 0,18959 0,28904 0,28371 0,3141 0,2356 0,1934 0,24471 0,24467 0,15618 0,25415 0,17812 0,03702

Surface

PSD d(10)

PSD d(50)

PSD dissolved dissolved dissolved dissolved d(90) in 15min in 30min in 45min in 60min

[°C]

[°C]

[m2/g]

[µm]

[µm]

[µm]

[%]

[%]

[%]

[%]

64,7 106 57 51,7 51 51,3 59 59,2 58,9 63,5 44,7 61,2 61,1

210 116,3 135,2 -

0,8 0,8 1 0,7 1,9 1,1 2,1 2,1 1,2 0,4 0,9

21 21 22 27 35 37 71 1,1 23 37 52

48 49 47 54 67 64 106 2,1 50 119 119

101 118 129 152 142 112 229 5,9 136 239 346

1,0 0,9 7,2 11,2 13,1 11,8 9,9 12,7 12,9 8,8 19,1 19,1 4,1 6,9 7,4

1,3 1,2 11,4 25,7 16,9 16,2 13,7 17,1 18,1 13,0 21,8 21,8 7,1 11,5 11,4

0,0 0,6 11,2 27,5 19,2 19,0 16,3 19,6 21,5 15,9 24,4 24,4 9,0 15,1 14,5

0,6 1,1 11,8 28,9 26,8 21,0 21,1 18,8 21,7 24,0 19,0 27,6 11,7 18,3 17,9

SC

(1432,08 cm-1)

Tm

M AN U

A C A C A A A A A A A A A C C A

(1272,79 cm-1)

Tg

TE D

2.1 2.1 2.1.1. 2.1.2. 2.1.3. 2.1.4. 2.1.5. 2.1.6. 2.1.7. 2.1.8. 2.1.9. 2.1.10. 2.1.11.

RS

EP

Soluplus Febuxostat crystal. Febuxostat amorph. phys. mix (crystal. API) phys. mix (amorph. API) HME Fusion, SC Fusion, BC Fusion, FC HME+3%EtOH HME+15%EtO HME+70%EtOH SD SE SE at 50°C RVE

IR

AC C

sample name

method ssNMR description

RI PT

Table 1 Results summary (A-amorphous; C-crystalline)

SC

RI PT

ACCEPTED MANUSCRIPT

M AN U

Figure 1 Spectra from Nuclear Magnetic Resonance; A – physical mixture of amorphous febuxostat and Soluplus; B – physical mixture of crystalline febuxostat and Soluplus; C – amorphous solid dispersion prepared by spray drying; D – crystalline solid dispersion

AC C

EP

TE D

prepared by SE; E – crystalline solid dispersion prepared by SE at 50°C

SC

RI PT

ACCEPTED MANUSCRIPT

-1

AC C

EP

TE D

M AN U

Figure 1 PCA analysis of the IR spectra in the region of 800-1600 cm

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

Figure 3 Loadings graph for the first component from the IR spectra of the samples.

ACCEPTED MANUSCRIPT

1,0

0,9

RI PT

0,7

SC

0,6

AC C

EP

TE D

Figure 4 IR spectra of selected samples

00

50

16

15

50

00

14

50

14

00

wavenumbers (cm-1)

13

50

13

12

00

M AN U 50

12

00

11

50

11

00

10

0

10

0

0

95

90

85

80

0

0,4

00

phys-mix amorph phys-mix crystal SD SE at 50°C HME

0,5

15

intensity

0,8

-1

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

Figure 5 PCA analysis from the region of 800 - 1600 cm of the Raman spectra

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

Figure 6 Loadings graph for the first component from the Raman spectra of the samples.

ACCEPTED MANUSCRIPT

0,35 0,30

RI PT

0,20 0,15 phys-mix amorph phys-mix crystal SE RVE

AC C

EP

TE D

Figure 7 Raman spectra of selected samples

00

00

15

50

15

00

50

wavelength (cm-1)

14

00

13

13

50

M AN U 00

12

50

12

00

11

50

11

00

10

0

0

0

10

95

90

85

80

0

0,00

16

0,05

50

SC

0,10

14

intensity

0,25

ACCEPTED MANUSCRIPT

d (0.1) d (0.5) d (0.9)

300

RI PT

250 200

O RV

SE

at

SE

O

HM

E+

70

%

%

15

Et

Et

O

Et

E+

HM

3%

H

O

H

C Q HM

E+

Fu

sio

n,

n,

BC

SC si o Fu

Fu

si o

n,

HM

E

0

M AN U

50

50

100

°C

SC

150

SD

Number weight size (um)

350

AC C

EP

TE D

Figure 8 Results of d-values of number-weighted PSDs of the solid dispersion used

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

Figure 9 Larger magnification of solid dispersion material: (A) SD; (B) SE at 50°C, (C) SE

AC C

EP

TE D

M AN U

SC

Figure 10 Comprehensive PCA analysis biplot

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

Figure 11 Comprehensive PLS analysis biplot

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

30

20

SC

15

5

0 0

10

20

30

M AN U

10

40

50

60

time (min)

EP

amorphous state

TE D

Figure 12 Drug release profiles of the solid dispersions prepared and of the pure drug in crystalline and

AC C

dissolved (%)

amorphous febuxostat crystalline febuxostat PhysMix -amorph PhysMix -crystal HME Fusion, SC Fusion, BC Fusion, FC HME + 3% EtOH HME + 15%EtOH HME + 70%EtOH SD SE SE at 50°C RVE

RI PT

25