Вы находитесь на странице: 1из 19

bs_bs_banner

Journal of Pharmacy
Research Paper
And Pharmacology

Using Flory–Huggins phase diagrams as a pre-formulation


tool for the production of amorphous solid dispersions:
a comparison between hot-melt extrusion and spray drying
Yiwei Tiana, Vincent Caronb, David S. Jonesa, Anne-Marie Healyb and Gavin P. Andrewsa
a
The Drug Delivery and Biomaterials Group, School of Pharmacy, Medical Biology Centre, Queen’s University, Belfast, UK and bSchool of Pharmacy
and Pharmaceutical Sciences, Main Department, Trinity College Dublin, Dublin, Ireland

Keywords Abstract
amorphous solid dispersion; Flory–Huggins
phase diagram; hot-melt extrusion; physical Objectives Amorphous drug forms provide a useful method of enhancing the dis-
stability; spray drying solution performance of poorly water-soluble drugs; however, they are inherently
unstable. In this article, we have used Flory–Huggins theory to predict drug solubil-
Correspondence
ity and miscibility in polymer candidates, and used this information to compare
Gavin P. Andrews, The Drug Delivery and
Biomaterials Group, School of Pharmacy,
spray drying and melt extrusion as processes to manufacture solid dispersions.
Medical Biology Centre, Queen’s University, Method Solid dispersions were prepared using two different techniques (hot-melt
97 Lisburn Road, Belfast BT9 7BL, Northern extrusion and spray drying), and characterised using a combination of thermal
Ireland, UK. (thermogravimetric analysis and differential scanning calorimetry), spectroscopic
E-mail: g.andrews@qub.ac.uk (Fourier transform infrared spectroscopy (FTIR) and X-ray diffraction methods.
Key findings Spray drying permitted generation of amorphous solid dispersions
Received May 24, 2013
across a wider drug concentration than melt extrusion. Melt extrusion provided
Accepted August 6, 2013
sufficient energy for more intimate mixing to be achieved between drug and
doi: 10.1111/jphp.12141 polymer, which may improve physical stability. It was also confirmed that stronger
drug–polymer interactions might be generated through melt extrusion. Remixing
and dissolution of recrystallised felodipine into the polymeric matrices did occur
during the modulated differential scanning calorimetry analysis, but the comple-
mentary information provided from FTIR confirms that all freshly prepared
spray-dried samples were amorphous with the existence of amorphous drug
domains within high drug-loaded samples.
Conclusion Using temperature–composition phase diagrams to probe the rel-
evance of temperature and drug composition in specific polymer candidates facili-
tates polymer screening for the purpose of formulating solid dispersions.

Introduction
Amorphous drug forms provide a useful method of is gaining significant relevance in the development of
enhancing the dissolution performance of poorly water- molecular dispersions that perform consistently (with
soluble drugs; however, they are inherently unstable. One regard to physical stability) over pharmaceutical relevant
way to overcome this is to disperse amorphous drug mol- time scales.
ecules, at a molecular level, into polymeric matrices,[1,2] that While numerous approaches have been used to study the
is, through the development of drug delivery systems thermodynamic mixing of polymer–polymer blends,[9,10]
wherein the drug (solute) is completely dissolved into the there are fewer articles available with regard to pharmaceu-
polymer (solid solvent).[3,4] In such cases, drug–polymer tical drug–polymer systems.[11,12] Although the measure-
miscibility has been shown to be highly important in ment of drug solubility within a low molecular weight
understanding the physical stability of these systems.[5–7] solvent is reasonably straightforward, it is experimentally
Consequently, the solubility or miscibility of drug within difficult, because of viscosity effects, to measure the
polymeric matrices has been gaining increasing interest.[8] true solubility of crystalline drug in polymeric carriers,
Furthermore, consideration of thermodynamic miscibility especially across a range of pharmaceutical relevant

256 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

temperatures. A method of doing so, published by Tao et al. HME using a co-rotating twin-screw extruder (HAAKE
involved ball milling or cryomilling to maximise the interac- Minilab, Thermo Fisher Scientific, Loughborough, UK).
tion between drug and polymer particles, thus improving the Three different drug loadings were prepared (20%, 50% and
accuracy in determining the dissolution point of drug within 80% w/w, respectively). To achieve an accurate drug to
the polymeric matrix.[13] This method has been further inves- polymer ratio through the extrusion, a ball mill mixer
tigated and utilised to determine important thermodynamic (Retsch, model MM200, Haan, Germany) was used to break
parameters, such as the Flory–Huggins (F–H) interaction down the particle size of the polymer granules, followed by
parameter (χ) and theoretical solubility of the drug in a dry mixing with drug using a mortar and pestle. The pre-
polymer at defined temperatures.[14,15] A more recent publica- pared mixtures were then extruded at a temperature of
tion by our group has reported a thermal method that can be 140°C and a screw speed of 100 rpm. Melt extrudates were
used in combination with F–H theory to construct the full milled and passed through a 250–180 μm sieve, and stored
temperature–composition phase diagram. Moreover, we have in a desiccator over silica gel at a temperature of 20°C. A
shown the temperature dependence of the F–H interaction suitable quantity of the physical mixtures was kept for
parameter, χ, and defined the level of drug saturation within analysis to compare with corresponding extrudates.
specific polymeric matrices across a wide temperature
range.[16] Knowledge of this information certainly advances
Preparation of amorphous solid dispersion
our understanding of the physical stability in respect to ther-
systems via spray drying
modynamic miscibility as a function of temperature and
drug composition. Amorphous FD solid dispersion samples were manufac-
The objective of this current article was to extend this tured by spray drying (Büchi Laboratoriums-Technik AG,
work through construction of temperature–composition Flawil, Switzerland) using HPMCAS, Soluplus and PVPK15.
phase diagrams of an amorphous drug in the presence of dif- Three different drug loadings were prepared (20%, 50% and
ferent polymeric carriers. Once complete, we wanted to use 80% w/w, respectively). In a typical procedure, a total of 1 g
this information as a pre-formulation tool to rank polymers of drug and polymer were dissolved in 250 ml of solvent
in terms of suitability to formulate molecular drug disper- water solution (acetone 95% v/v, water 5% v/v for
sions. Moreover, based upon the obtained thermodynamic HPMCAS system and ethanol 95% v/v, water 5% v/v for
information of drug–polymer system through the phase Soluplus and PVPK15 systems). The spray drying param-
diagram, we might be starting to understand the effects of eters were set as follows: airflow of 670 Nl/h per pump
different processing techniques, namely hot-melt extrusion setting of 30% (9 ml/min), aspirator setting of 100% (−50
(HME) and spray drying, in the production of amorphous to −60 mbar), inlet temperature 85°C, outlet temperature
solid dispersion. The model drug selected in this study was between 56 and 59°C. The inlet temperature was selected to
felodipine (FD). Three polymeric candidates, namely poly- be slightly above the boiling point of the solvent and ensure
vinylpyrrolidone (PVP grade K15 (PVPK15)), polyvinyl the quick drying of the droplets. Before spray drying, the
caprolactam-polyvinyl acetate-polyethylene glycol (Soluplus, dissolved solutions were sonicated for a suitable length of
BASF, Ludwigshafen, Germany) and hydroxypropyl methyl- time to ensure a complete dissolution of FD into the poly-
cellulose acetate succinate (HPMCAS grade HF (HPMCAS- meric solution. The solubility of FD in the solvent mixture
HF, Shin-Etsu, Tokyo, Japan)), were used as model carriers. was checked by measuring 50 mg of FD into amber
ampoules containing 10 ml of the solvent mixture. The
Materials and Methods drug was freely soluble indicating solubility was above
5 mg/ml, which was sufficient for spray drying.
Materials
Felodipine with a purity of 99.9% was a gift from Eli Lilly
Sample preparation for construction of binary
(Kinsale, Ireland). HPMCAS-HF was supplied by Shin-Etsu
phase diagram
Chemical Co. (brand name AQOAT). Polyvinyl caprolactam-
polyvinyl acetate-polyethylene glycol graft copolymer (brand The method of preparing a ball-milled physical mixture of
name Soluplus) was a generous gift from BASF Chemical Co. drug and polymer was adapted from a previous publication
described by our group.[16] Drug and polymer mixtures with
Methods different compositions were firstly mixed using a mortar
and pestle, followed by a ball mill mixer (Retsch, model
Preparation of amorphous solid dispersion
MM200). In a typical procedure, drug and polymer powder
systems via hot-melt extrusion
(sample total 1 g) was milled with one stainless steel ball at
Amorphous solid dispersions of FD with polymeric carriers 20 Hz. A predefined milling time of 2 min was chosen,
(PVPK15, Soluplus or HPMCAS) were manufactured via which was subsequently followed by a 2-min interval. This

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 257
Spray-dried and HME solid dispersions Yiwei Tian et al.

procedure was repeated to a maximum of ten mill-stop scan rate of 1° 2θ/min. Details of machine calibration and
cycles (max 20-min mill time) to maximise the drug– data processing were described in a previous publication.[16]
polymer interaction.
Fourier transform infrared spectroscopy
Thermal analysis
Infrared spectra were recorded on a Jasco-400 FT-IR spec-
A modulated differential scanning calorimetry (MDSC, trometer (Jasco, Easton, MD, USA) incorporating version 2
Q100, TA Instrument, Elstree, UK) was used throughout. of the Jasco Spectra Manager software. Powdered samples
Indium and zinc were used to standardise the cell tempera- with a particle size ranging 90–180 μm were compressed
ture, and dry nitrogen was used as the purge gas. A 5–10 mg into 0.2-g potassium bromide discs at a weight ratio of 1%.
powder sample was packed into an aluminium pan with a Discs were prepared by compression under 8–10 tons for
lid. A pinhole was made in the lid to allow the moisture to 2 min. Physical mixtures at the same drug–polymer compo-
escape. Before conducting the experiments, all ball-milled sitions were also examined to differentiate drug–polymer
samples were dried in a vacuum oven for at least 24 h prior interactions. All experiments were recorded in absorbance
and after the milling. Melting depression experiments were mode of 350–5000/cm using a resolution of 2/cm and accu-
conducted at heating rate of 1°C/min, 20–200°C. The end- mulation of data across 64 scans.
point of the melting endothermic peak was calculated from
the intercept point of the endothermic trace and the post- Statistical analysis
melting baseline.
For MDSC analysis, the effect of drug loading on the Tg
A heat–cool–heat cycle was also used to characterise the
values of the system was statistically analysed using one-way
solid-state properties of freshly prepared amorphous solid
analysis of variance. Individual differences in each treat-
dispersions. The parameters used for these experiments
ment group were identified using Tukey’s post-hoc test,
were as follows: heating range, 0–170°C; heating rate, 5°C/
wherein P < 0.05 denoted significance.
min; period of modulation, 60 s; and amplitude of modula-
tion, 0.8°C. The temperature of the glass transition was
Results and Discussion
measured at the midpoint of the reversing heat flow signal,
and the temperature of recrystallisation or melting event Characterisation of raw polymers
was measured from the total heat flow signal. The non- and physical mixture of
reversing heat flow signal was also used to confirm several felodipine–polymer systems
thermal events, such as recrystallisation and enthalpic
The unprocessed raw polymer Soluplus, PVPK15 and
recovery.
HPMCAS-HF were characterised by MDSC (Figure S1). All
Thermogravimetric analysis (TGA) was used to examine
three polymers were amorphous with distinctive glass tran-
the thermal stability of pharmaceutical ingredients. Experi-
sitions evident at approximately 85 ± 1.5°C for Soluplus,
ments were conducted using a Thermal Advantage Model
120 ± 0.5°C for PVPK15 and 122 ± 1.2°C for HPMCAS-HF.
Q500 TGA (TA Instruments). Samples were heated at a rate
Quench-cooled FD was characterised using the same
of 10°C/min, 20–400°C, and the % mass remaining was
MDSC procedure, and the glass transition temperature was
plotted as a function of temperature. Isothermal studies
recorded at 46 ± 1°C. Physical mixtures of three different
were also conducted for FD and polymeric ingredients by
polymers with crystalline drug FD were also characterised
heating samples up to a particular temperature of interest
by PXRD (Figures S2–S4) The diffractogram of unpro-
(150–160°C in this study), and holding at this temperature
cessed FD exhibits well-defined Bragg peaks, characteristic
for at least 1 h. In all TGA experiments, nitrogen gas was
of crystalline form I,[17] which is the most stable polymorph
used as the purge gas (flow rate 60 ml/min) in the furnace
at room temperature and atmospheric pressure. The PXRD
chamber.
pattern of crystalline drug–polymer physical mixtures was
characterised by a halo contribution underneath the Bragg
Powder X-ray diffraction
peaks, which is from amorphous polymer contained within
Powder X-ray diffraction (PXRD) measurements were per- the mixture; the amorphous polymer and its area increase
formed on individual ingredients, drug–polymer physical as the crystalline drug composition decreases. Fourier trans-
mixtures and solid dispersions prepared by spray drying or form infrared spectroscopy (FTIR) was also used as a com-
HME. The powders were packed onto a standard glass plementary technique to identify the physical and chemical
sample holder using a Rigaku Miniflex II desktop X-ray nature of crystalline FD, amorphous FD, Soluplus, PVPK15
diffractometer (Rigaku, Woodlands, TX, USA) with Bragg– and HPMCAS-HF (Figures S5 and S6).
Brentano geometry. The PXRD patterns were recorded Potential molecular interactions within crystalline form
3–40° on the 2θ scale, using a step width of 0.01° 2θ and I of FD are shown in Figure 1. The 3D structures were

258 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

was observed at 1744/cm for HPMCAS and 1667/cm for


PVPK15, respectively, while for Soluplus peaks at 1739/cm
and 1638/cm are associated with the carbonyl groups in the
vinyl acetate and caprolactam segments.[19] These specified
groups could potentially interact with drug molecules
within the solid dispersion and significantly influence the
performance of the final product.

Construction of the phase diagram of


felodipine in different polymeric matrices
Crystalline FD exists in two polymorphs: form I with a
melting temperature of 140–145°C and enthalpy of fusion
of 31.5 kJ/mol; and form II with melting temperature of
130–135°C and enthalpy of fusion of 26.7 kJ/mol.[17] FD
was supplied as form I (verified from differential scanning
Figure 1 Molecular interactions within felodipine form I, a coopera-
calorimetry (DSC) and PXRD analyses), and there was no
tion of two methyl ester and two ethyl ester groups connected via
amine groups, the hydrogen atom being shared between methyl ester
detectable polymorphism conversion in any of the pro-
and ethyl ester groups. cesses, neither in the ball mill/mixing nor during slow
heating in the DSC. It was noted that the ball-milled
sample presented less intense Bragg peaks which, when
constructed and optimised using ACDLabs 12.0 software compared with the unprocessed crystalline FD, were
(Toronto, Canada). Crystal packing in form I of crystalline broadened. The relationship of Tend and composition of
FD may result in hydrogen bonding among secondary FD in the mixture is presented in Table 1. The Tend is
amine group, methyl carbonyl group and ethyl carbonyl defined as the endset melting point of FD. Determination
group. Using FTIR (Figures S5–S6), the hydrogen bonded of the melting endset temperature of FD in PVPK15 or
N-H group in crystalline form I gives a stretching band at Soluplus was not possible at wt% FD < 0.7. Below this
3370/cm, which is a red-shifted relative to the level, the drug dissolved into the polymer during slow
non-hydrogen bonded N-H group at 3420/cm.[18] In the heating and the dissolution event were very broad, result-
1600–1800 region of FTIR spectra, this hydrogen bonding ing in an inaccurate determination. With the same
also causes a red shift of the C=O stretching band. The dis- assumption that the upper critical solution temperature
tinctive absorbance peak for C=O stretching in crystalline exists in the system of FD–PVPK15, the linear relationship
form I was recorded at 1695/cm. Relative to crystalline FD, of interaction parameter χ vs 1/T was applied (Figure 2a).
the red-shifted N-H and C=O groups within amorphous At increased temperature, there is a significant deviation
FD were observed at 3339/cm and 1684/cm, respectively. from linearity; however, given that the χ value is less than
There were also absorbance peaks recorded at 3420/cm and would be predicted by the linear relationship, the resultant
1698/cm attributable to the non-hydrogen bonded amine phase diagrams would in many cases be an underestima-
group in the dihydropyridine ring and carbonyl group in tion of miscibility.[16] The plots of free energy of mixing
the ethyl ester, respectively. The occurrence of larger red- (ΔGmix/RT) and temperature–composition phase diagram
shifted bands in the amorphous form of FD relative to crys- of FD–PVPK15 were also constructed (Figure 2b and 2c).
talline form I is indicative of enhancement of hydrogen The phase diagrams of systems FD–HPMCAS and
bonding between the nitrogen in the dihydropyridine ring FD–Soluplus constructed from previous publication were
and the oxygen in the carbonyl group of FD. The shared also used in this work (Figure 3a and 3b; permission
hydrogen bonding between N-H and C=O in form I is required to use this figure).[16] It is clear that the level of
replaced with single hydrogen bonding between N-H of the miscibility between FD and PVPK15 is much higher than
dihydropyridine ring and C=O of the methyl ester, whereas FD–HPMCAS and FD–Soluplus. For example, at a tem-
the C=O of ethyl ester group is free of hydrogen bonding in perature of 25°C, 0.1 w/w of FD may be fully dissolved
the amorphous FD (1698/cm) because of stronger shielding into the PVPK15 based on this constructed phase diagram,
relative to methyl ester. For polymer spectra, in 4000– which is very close to the result reported previously.[20]
2500/cm region, Soluplus and HPMCAS-HF both show a Additionally, for the metastable region of this system, a
broad absorbance peak at 3700–3400/cm associated with value of up to 0.25 w/w of FD may be locally stabilised
the O–H stretching. There is no hydrogen donor in PVP. within the polymer PVPK15 (spinodal curve) at 25°C. This
In the 2000–1500/cm region, a single, sharp carbonyl peak information, when coupled with the maximum solubility

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 259
Spray-dried and HME solid dispersions Yiwei Tian et al.

Table 1 Dissolution temperature endpoint measured by DSC for felodipine with PVPK15, Soluplus and HPMCAS systems; n = 3, represents as
average ± standard deviation

w/w felodipine Tend (°C) with PVPK15 Tend (°C) with Soluplus Tend (°C) with HPMCAS

0.95 141.47 ± 0.0 141.55 ± 0.05 141.1 ± 0.1


0.9 140.50 ± 0.12 140.92 ± 0.2 140.7 ± 0.05
0.85 139.25 ± 0.15 140.11 ± 0.5 140.0 ± 0.02
0.8 137.29 ± 0.21 138.43 ± 0.2 139.62 ± 0.3
0.75 134.93 ± 0.05 136.38 ± 0.0 139.01 ± 0.1
0.7 131.99 ± 0.2 135.05 ± 0.2 138.65 ± 0.5
0.65 a a
138.04 ± 0.8
0.6 a a
137.49 ± 1.1
0.55 a a
137.05 ± 0.5

DSC, differential scanning calorimetry; HPMCAS, hydroxypropyl methylcellulose acetate succinate; PVPK15, PVP grade K15. aDetermination is not
possible because of the weakness of the thermal event.

and metastability values for HPMCAS (0.001 and 0.6, Powder X-ray diffraction of amorphous
respectively) and FD–Soluplus (0.02 and 3.69, respec- solid dispersions
tively), provides a means of ranking polymers against mis-
Figure 4a–4c shows the PXRD patterns of drug–polymer
cibility and solubility at different temperatures.
solid dispersions prepared via the two different techniques.
The absence of Bragg peaks in the PXRD profiles indicates
Hot-melt extrusion and spray drying that almost all the samples were X-ray amorphous except
Undoubtedly, the most important factor when manufactur- melt-extruded 80% FD–HPMCAS, which features a small
ing drug delivery platforms through an HME process is Bragg peak recorded at 21.56°2θ, a peak characteristic of
ensuring thermal stability of the active pharmaceutical crystalline FD form I suggesting the existence of crystalline
ingredient and excipients. TGA provides general informa- content within high drug-loaded melt extrudates. In con-
tion relating to the thermal stability of pharmaceutical trast, samples prepared by spray drying were all XRD amor-
ingredients in this regard. In this study, the thermal degra- phous. As expected, these results correlate with earlier work
dation temperatures were measured for each individual wherein we described FD–HPMCAS combination as having
ingredient, and they were 200°C for FD, 250°C for Soluplus, the weakest interaction (positive χ value) indicative of
380–390°C for PVP and 200°C for HPMCAS. unfavourable mixing. However, the results obtained from
The principle of this study is to examine the use of PXRD cannot be used in isolation to define the physical
temperature–composition phase diagram to evaluate the nature of these solids, since an X-ray amorphous state
ability of preparing fully amorphous solid dispersions; (nanocrystalline) is achievable through grinding, and it is
therefore, the effect of process parameters is not included in often observed as a broad halo peak in PXRD.[22] In such
our considerations. For HME, the temperature was chosen systems, it is recommended that further characterisation be
based on the lowest temperature where the mixture could conducted using complementary techniques (c).
be extruded (140°C) without a significant torque output
(20–80 N.cm). This would reduce the effects of shearing Modulated differential scanning calorimetry
stress, and thus reduce the possibility of mechanical degra- analysis on amorphous solid dispersions
dation.[21] A regular rod-shaped melt extrudate with smooth MDSC was used to further study the nature of solid disper-
surface was obtained in all cases. From visual inspection, sions prepared via HME and spray drying. The MDSC
the 20% w/w and 50% w/w FD-loaded extrudates were results for FD–polymer systems are summarised in Table 2
transparent, while, 80% w/w drug-loaded melt extrudates (standard deviations can be found in Table S1), as well as
were opaque. the theoretical glass transition temperature for each binary
To compare HME and spray drying, the same drug– mixture Tg,mix, calculated using the Couchman–Karasz (CK)
polymer combinations were also prepared by spray drying model, as shown in Equation 1 and 2:
at the same drug compositions. TGA studies show that the
moisture/solvent contents in spray-dried samples were w1Tg 1 + Kw2Tg 2
Tg ,mix = (1)
approximately 3–5%, while no moisture content was w1 + Kw2
detected in melt-extruded samples. PXRD, FTIR and MDSC
were used to characterise HME and spray-dried solid dis- where, w1,2 and Tg,1,2 are the weight fraction and glass transi-
persions immediately after preparation. tion temperature of component one and two as binary

260 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

(a) –0.5

–1

–1.5
x

y = 3679.2x – 10.383
–2 R2 = 0.9084

–2.5

–3
0.0024 0.00241 0.00242 0.00243 0.00244 0.00245 0.00246 0.00247 0.00248
1/T per K
(b) 0.6 0°C
25°C
50°C
0.4 80°C
100°C
120°C
0.2

0
ΔGmix/RT

–0.2

–0.4

–0.6

–0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Volume fraction (φ)
(c) 145

125

105
Temperature (°C)

Solubility FD in PVPK15
Spinodal FD in PVPK15
85 System Tg
UCST

65

45

25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Weight fraction of drug

Figure 2 (a) Variation of the interaction parameter χ as a function of temperature. The solid line represents the line of best fit to experimental
data, (b) ΔGmix/RT as function of drug volume fraction for felodipine and PVPK15, and (c) binary temperature–composition phase diagram for
felodipine and PVPK15. UCST, upper critical solution temperature.

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 261
Spray-dried and HME solid dispersions Yiwei Tian et al.

(a) 145

125

Temperature (°C)
105 Solubility FD in HPMCAS
Spinodal FD in HPMCAS
85 System Tg
UCST
65

45

25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Weight fraction of drug

(b) 145

125
Temperature (°C)

105

Solubility FD in Soluplus
85 Spinodal FD in Soluplus
System Tg
65 UCST

45

25
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Weight fraction of drug

Figure 3 Binary temperature–composition phase diagram for (a) felodipine and hydroxypropyl methylcellulose acetate succinate, and (b) felodipine
and Soluplus, adapted from previous publication[16] (permission required). UCST, upper critical solution temperature.

system. The constant K in this equation may be calculated more homogeneous mixture was expected after the first
from the corresponding heat capacity change of the pure heating. For systems containing 20% FD, the solid disper-
components at the glass transition: sions exhibited amorphous features with only one inter-
mediate glass transition within the temperature range
ΔCp1 between the two pure components, irrespective to the type
K= (2)
ΔCp2 of process technique used for manufacture (Figure 5).
There was almost no variation around this Tg as a function
The heat capacity change (ΔCp) for pure amorphous FD of heating cycle, suggesting that the optimum mixing
at the Tg was 0.39 J/g°C, and for Soluplus, PVPK15 and (molecular level) between 20% drug and polymer was
HPMCAS-HF equal to 0.21 J/g°C, 0.35 J/g°C and 0.32 J/ achieved during spray dry or melt extrusion.[23] Further-
g°C, respectively. As previously described in the Methods more, comparing the Tg values of 20% FD spray-dried solid
section, the glass transition temperature was measured dispersion with the values predicted by CK model, only
twice through continued heat–cool–heat cycles; therefore, a FD–PVPK15 prepared by spray drying was higher than that

262 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

(a) predicted by the CK model, while the values of the other


spray-dried samples were lower than CK predictions
Intensity (CPS)

(∼10°C). It suggests that the drug–polymer interactions


within the 20% FD–PVPK15 spray-dried sample may be
stronger than that achievable through ideal mixing, thus
20% FD + PVPK15 S P resulting in a positive experimental deviation.[18] In con-
50% FD + PVPK15 S P
80% FD + PVPK15 S P trast, Tg measured for 20% FD melt-extruded samples was
5 10 15 20 25 30 35 40 in reasonable agreement with the values predicted by the
CK model.
MDSC thermograms for 50% FD–polymer solid disper-
Intensity (CPS)

sions prepared by melt extrusion and spray drying are


shown in Figures 6 and 7. It is clearly shown that samples
containing PVPK15 and Soluplus possess one Tg after melt
20% FD + PVPK15 HME extrusion, while for the 50% FD–HPMCAS melt extrudate,
50% FD + PVPK15 HME
80% FD + PVPK15 HME a main glass transition (70°C) was observed together with
5 10 15 20 25 30 35 40 a subtle endothermic step at 112°C. Considering the Tg of

50%FD–HPMCAS occurred at 75°C (second heating) or
(b)
87°C (CK model), it is very likely that this second endo-
Intensity (CPS)

thermic step belongs to the glass transition of a polymer-


rich phase.[24] This second Tg was absent on the second
heating with the presence of one Tg (75°C). These results
20% FD + HF S P
50% FD + HF S P
suggest that 50% FD–HPMCAS sample prepared by melt
80% FD + HF S P extrusion is at least partially phase-separated, indicated by
5 10 15 20 25 30 35 40 the presence of more than one glass transition event
during the first heating. In contrast, melt extrudates of
Intensity (CPS)

50% FD–PVPK15 and 50% FD–Soluplus exhibited one


Tg.
For 50% FD–polymer spray-dried samples (Figure 7),
20% FD + HF HME several thermal events were observed on the first heating,
50% FD + HF HME
80% FD + HF HME but all of them appear to have a glass transition-like event
5 10 15 20 25 30 35 40 around temperature 50–90°C, which is higher than the

value of pure amorphous FD (45°C). This is indicative of
(c)
an intermixed amorphous drug–polymer solid disper-
Intensity (CPS)

sion (intermediate Tg). However, when the temperature


increases, a broad exothermic plateau-like trace was
observed in all three 50% FD–polymer spray-dried samples,
20% FD + Solu S P
with no distinctive sign of melting events at higher tem-
50% FD + Solu S P
80% FD + Solu S P
peratures (140–145°C). The reverse heat capacity (Cp) in
5 10 15 20 25 30 35 40 the same region was plotted to help understand the nature
of the exothermic event. Relative to the first heat cycle, the
reversing Cp values in the second heating are higher, espe-
Intensity (CPS)

cially in the temperature range where the exothermic


plateau-like traces were observed (Tg–Tm). The reversing Cp
20% FD + Solu HME
of a pure amorphous mixture usually exhibits a significant
50% FD + Solu HME increase as it goes through the glass transition because of
80% FD + Solu HME
the dramatic increase of its enthalpy and entropy in the
5 10 15 20 25 30 35 40 super-cooled liquid state,[25] while for the crystalline form,

the reversing Cp remains constant until the melting tem-
Figure 4 Normalised powder X-ray diffraction patterns of felodipine perature.[26] Therefore, in this case, the lower Cp in the first
solid dispersions containing (a) PVPK15, (b) hydroxypropyl methyl- heating, following the glass transition and the gradual
cellulose acetate succinate grade HF and (c) Soluplus prepared by hot- decrease of the Cp in the temperature range 70–120°C (for
melt extrusion and spray drying; arrow identifies a characteristic Bragg FD–Soluplus and FD–HPMCAS, respectively), suggests that
peak associated with form I of crystalline felodipine. the crystalline FD content within the spray-dried mixture

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 263
Spray-dried and HME solid dispersions Yiwei Tian et al.

Table 2 Glass transition temperatures (midpoint) for felodipine–polymer solid dispersions prepared by spray drying and hot-melt extrusion; values
presented are the average of triplicates (n = 3)

Measured Measured Tg,mix predicted by Measured Tm on


wt% FD Tg cycle1 (°C) Tg cycle2 (°C) CK model (°C) cycle1 (°C)

Spray-dried
Soluplus 20 71.7 70.2 80.2 —
50 53.9 58.9 70.9 —
80 53.9/85.79 48.8 57.6 138.4 (ΔH = 0.1 J/g)
PVPK15 20 109.1 107.8 106.2 —
50 50.5/90.6 81.5 84.4 —
80 43.35/103.9 56.77 61.3 —
HPMCAS-HF 20 90.6 87.5 108.8 —
50 57.9 58.5/121.7 87.1 —
80 49.68 49.07 62.9 139.5 (ΔH = 1.7 J/g)
Melt extruded
Soluplus 20 77.6 80 80.2 —
50 58.6 62.8 70.9 —
80 46.5/74.2 47.6 57.6 135.7 (ΔH = 1.9 J/g)
PVPK15 20 100.5 103.9 106.2 —
50 87.35 89 84.4 —
80 51.5/79.5/99.3 52.5 61.3 134.79 (ΔH = 1.1 J/g)
HPMCAS-HF 20 106.5 107.7 108.8 —
50 70/112.2 75.9 87 —
80 47.71/124.43 50 62.9 136.8 (ΔH = 9.6 J/g)

CK, Couchman–Karasz; FD, felodipine; HPMCAS, hydroxypropyl methylcellulose acetate succinate; PVPK15, PVP grade K15. If two glass transition
events were recorded, both values are presented in the table as Tg1/Tg2; ‘—’ refers to non-detectable.

20% FD + Soluplus_HME.003
20% FD + PVPK15_HME.001
20% FD + HPMCAS_HME.001
20% FD + Soluplus_SP.txt
20% FD + PVPK15_SP.txt
20% FD + HPMCAS_SP.txt

Exo Up
Rev heat flow (W/g)

20 70 120
Temperature (°C) Universal V4.3A
TA Instruments

Figure 5 Modulated differential scanning calorimetry thermograms of melt-extruded (top three) and spray-dried (bottom three) samples contain-
ing 20 w/w% felodipine with hydroxypropyl methylcellulose acetate succinate grade HF (dash dot line), PVPK15 (dash line) and Soluplus (solid line);
arrow identifies the glass transition of each sample, and the trace of second heating ramp sits below the first heating for each sample.

increases as the temperature increases. This phenomenon sample varies depending upon the polymeric carrier used.
may suggest that although an obvious glass transition was Therefore, if certain temperatures are reached (Tg), the
observed on all 50% FD–polymer spray-dried samples, the reorganisation of amorphous FD into small crystals may
level of drug disorder or distribution in the amorphous occur. And the potential of disordered amorphous FD to

264 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

50% FD + Soluplus_HME.001
50% FD + PVPK15_HME.001
50% FD + HPMCAS_HME.004

Exo Up
Rev heat flow (W/g)

100 105 110 115 120 125

20 40 60 80 100 120 140 160


Temperature (°C) Universal V4.3A
TA Instruments

Figure 6 Modulated differential scanning calorimetry thermograms of melt-extruded samples containing 50% w/w felodipine with hydroxypropyl
methylcellulose acetate succinate grade HF (dash dot line), PVPK15 (dash line) and Soluplus (solid line); the trace of second heating sits below the
first heating for each sample; arrow indicates the expansion of second Tg on the first heating of 50 % w/w felodipine–hydroxypropyl methylcellulose
acetate succinate melt extrudate.

recrystallise during heating may be highly polymer depend- the range 134–136°C, with varying values of enthalpy of
ent. In particular, the slow heating within the MDSC would fusion (shown in Table 2). The values for enthalpy of fusion
provide sufficient time and energy for drug molecular collected on the first heating were in the following order:
rearrangement or recrystallisation in spray-dried sample FD–PVPK15 < FD–Soluplus < FD–HPMCAS. Additionally,
FD–HPMCAS and FD–Soluplus systems, characterised by several thermal events were observed in samples 80%
the presence of exothermic events above the glass transition FD–PVPK15 and 80% FD–Soluplus between the Tg of pure
temperature. amorphous FD (45°C) and polymeric carrier (120°C for
It is also very interesting to note the absence of melting fol- PVPK15 and 85°C for Soluplus). This suggests that, beside a
lowing recrystallisation and the gradual increase in the small level of residual crystalline material, the amorphous
reversing Cp, which may suggest dissolution of recrystallised drug distribution within the extrudates is also not molecu-
FD into the polymeric matrix at increased temperature (110– larly homogeneous. This uneven distribution may result in
140°C). The dissolution of crystalline drug into polymeric a poor physical stability as drug-rich domains would easily
matrices below their melting point is in a good agreement reorganise into nuclei and trigger the spontaneous growth
with our theoretical predictions provided by drug– of crystalline drug.
temperature phase diagrams of these materials (Figures 2 and For 80% FD spray-dried solid dispersions containing
3). More specifically, the dissolution temperatures of 1 : 1 Soluplus and HPMCAS (Figure 9a), a distinctive glass tran-
(50% w/w) FD : polymeric matrix varied depending upon sition in the range 40–50°C and a melting event in the range
polymer type, but were approximately 90°C for PVPK15, 135–140°C were observed. The heat of fusion for the com-
120°C for Soluplus and 125°C for HPMCAS, which is in bination of FD–Soluplus (0.06 ± 0.05 J/g) is smaller than
good agreement with the temperature at which the reversing FD–HPMCAS (1.7 J/g). However, most notable from the
Cp starts to increase (Figure 7). It is also noted that a slight MDSC data is that the spray-dried sample containing 80%
variation in the temperature of dissolution was predicted by FD–PVPK15 exhibited a glass transition, an exothermic
theoretical calculations relative to the results from MDSC. recrystallisation and a broad melting event in the first
This may be attributed to the heating rate used in the DSC.[9] MDSC scan (Figure 9b). The behaviour of this particular
A rate of 1°C/min was used for solubility calculations and spray-dried sample during heating was very similar to the
5°C/min for MDSC experiments. spray-dried sample containing 50% FD, wherein similar
MDSC results are shown in Figure 8 for 80% recrystallisation and broad melting were observed
FD–polymer melt extrudates. All samples exhibit a glass (Figure 7). As described previously, such an exothermic,
transition at a temperature 45–50°C and a melting peak in following the glass transition, could be attributed to a less

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 265
Spray-dried and HME solid dispersions Yiwei Tian et al.

50% FD + HPMCAS
50felo50HPMCAS_SP_first.txt
1.8 50felo50HPMCAS_SP_first.txt
50felo50HPMCAS_SP_second.txt 50felo50HPMCAS_SP_second.txt

1.6
ΔCp first and second heating

Exo Up 1.4

Rev Cp (J/(g·°C))
Rev heat flow (W/g)

1.2

1.0

0.8

0.6
10 30 50 70 90 110 130 150 10 30 50 70 90 110 130 150
Temperature (°C) Universal V4.7A TA Temperature (°C) Universal V4.7A TA

50% FD + PVPK15
50% FD + PVP_SP_first.txt 2.0 50% FD + PVP_SP_first.txt
50% FD + PVP_SP_second.txt 50% FD + PVP_SP_second.txt

1.5
Exo Up
Rev Cp (J/(g·°C))

ΔCp first and second heating


Rev heat flow (W/g)

1.0

0.5

0.0
10 30 50 70 90 110 130 150 10 30 50 70 90 110 130 150
Temperature (°C) Universal V4.7A TA Temperature (°C) Universal V4.7A TA

50% FD + Soluplus
50% FD + Soluplus_SP_first.txt
2.0 50% FD + Soluplus_SP_first.txt
50% FD + Soluplus_SP_second.txt 50% FD + Soluplus_SP_second.txt

ΔCp first and second heating


Rev Cp (J/(g·°C))

Exo Up 1.5
Rev heat flow (W/g)

1.0

0.5
10 30 50 70 90 110 130 150 10 30 50 70 90 110 130 150
Temperature (°C) Universal V4.7A TA Temperature (°C) Universal V4.7A TA

Figure 7 The reversing heat flow (left side) and reversing heat capacity (right side) signals of spray-dried samples containing 50% w/w felodipine.
The first heating ramp is represented as a dashed line, and the second heating ramp is represented as a solid line.

266 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

80% FD + Soluplus_HME.003
80% FD + PVPK15_HME.001
80% FD + HPMCAS_HME.001

Tg2

Exo Up
Rev heat flow (W/g)

10 30 50 70 90 110 130 150


Temperature (°C) Universal V4.3A
TA Instruments

Figure 8 Modulated differential scanning calorimetry thermograms of melt-extruded samples containing 80% felodipine with hydroxypropyl
methylcellulose acetate succinate grade HF (dash dot line), PVPK15 (dash line) and Soluplus (solid line); the trace of second heating sits below the
first heating for each sample.

disordered amorphous system generated via spray drying, the polymer type, drug–polymer composition, temperature
which could reorganise and recrystallise during heating. and processing technique.[27] The polymer, due to its highly
With high drug loadings, the subtle exothermic or endo- viscous nature, may retard amorphous relaxation, and thus
thermic event becomes easier to identify, as shown in the increase the kinetic barrier for amorphous to crystallisation
non-reversing heat flow signals in Figure 9b. In addition, it conversion. In addition, polymeric materials have been
also demonstrates that by using spray drying with the same shown to be efficient crystallisation inhibitors (miscible
parameters, a fully amorphous phase of 80% FD–PVPK15 binary system).[28] Several studies have provided clear evi-
may be generated, even though this type of system is not dence for intermolecular interactions between drug and
capable of remaining stable during heating. In contrast, polymer in solid dispersions, confirming that the existence
for the combinations of 80% FD–HPMCAS and 80% of such bonding is necessary to obtain a completely mis-
FD–Soluplus, fully amorphous materials were not achieved cible system at the molecular level. FTIR was used to
using spray dry using the same process parameters. The examine the intermolecular interactions between FD and
MDSC results for 20%, 50% and 80% FD–polymer systems respective polymers in the solid dispersions. Previous
prepared by both melt extrusion and spray drying demon- studies on the raw components of the dispersions had
strate that the behaviour of different polymers and drug shown that the N-H stretching position is very sensitive to
loading in the preparation of amorphous FD solid disper- the type and strength of hydrogen bonding. In particular,
sions is varied, irrespective of the preparation technique. In comparison of amorphous and crystalline FD confirmed
this study, PVPK 15 has shown the greatest potential in weaker hydrogen bonding in crystalline FD. Thereby, to
forming amorphous solid dispersions followed by Soluplus understand how the polymers interact with FD in the solid
and HPMCAS-HF. This ranking is in a close agreement dispersion, it is necessary to compare the differences in the
with the results predicted using the F–H model in that those spectra of pure amorphous FD and FD embedded in
polymers with a more negative drug–polymer interaction polymer solid dispersions. Figure 10a–10c shows IR spectra
parameter (χ) and high miscibility value should be more in the N-H region of the solid dispersions FD–PVPK15,
favourable to form the amorphous polymer/drug disper- FD–Soluplus and FD–HPMCAS, respectively (see Figure S7
sion even at supersaturated drug concentrations. for C=O region). The N-H stretching peak of amorphous
FD was observed at 3339/cm, and the C=O stretching peak
had split into two peaks at 1702 and 1684/cm. For FD
Fourier transform infrared spectroscopy
embedded in PVPK15, irrespective of the preparation tech-
It was anticipated that the physical state of drug within the nique, the N-H stretching vibration (Figure 10a) exhibited a
polymeric amorphous solid dispersions would vary with sharp peak at 3339/cm at an 80% drug loading, which was

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 267
Spray-dried and HME solid dispersions Yiwei Tian et al.

(a) 80% FD + HPMCAS_SP_first.txt


80% FD + HPMCAS_SP_second.txt
80% FD + Soluplus_SP_first.txt
80% FD + Soluplus_SP_second.txt

Exo Up

Rev heat flow (W/g)

138.38°C
0.06430J/g

10 30 50 70 90 110 130 150


Temperature (°C) Universal V4.7A

(b) 0.00 0.00

–0.02 recrystallisation

–0.05

Non-rev heat flow (W/g)


–0.04
Rev heat flow (W/g)

Melting
–0.06
–0.10
–0.08

–0.10
–0.15
Second
–0.12
first

–0.14 –0.20
0 20 40 60 80 100 120 140 160 180
Exo Up Temperature (°C) Universal V4.3A
TA Instruments

Figure 9 (a) Modulated differential scanning calorimetry thermograms of spray-dried samples containing 80% felodipine with hydroxypropyl
methylcellulose acetate succinate grade HF (dash line) and Soluplus (solid line); the trace of second heating sits below the first heating for each
sample; arrow points to the expansion of melting on the first heating of 80% felodipine–Soluplus. (b) Modulated differential scanning calorimetry
thermograms of spray-dried samples containing 80% felodipine with PVPK15, the recrystallisation and melting were clearly shown on the non-
reversing heating flow signal during first heating.

similar to the N-H stretching in amorphous FD (amor- distribution of drug in to drug–drug phases, while the
phous drug–drug interaction). This peak was shifted to population of drug–polymer interaction (3290/cm) was
3290/cm as drug loading decreased to 50 and 20%. This completely overlapped by the absorption peak at 3339/cm
downward shift in the position of N-H stretching peak indi- (drug–drug interactions) when the drug loading increased
cates that, at lower drug loadings, a stronger drug–polymer to 80%, irrespective of the process technique used.
interaction may be formed instead of amorphous drug– A similar result was observed in FD–Soluplus solid dis-
drug interactions. However, it should also be noted that for persion (Figure 10b); the number of N-H stretching peaks
the 50% FD–PVPK15 spray-dried sample, a peak at in solid dispersion varies as the drug concentration
3339/cm was observed, which may be attributed to the increases. One clear peak at 3294/cm was observed, attribut-
existence of amorphous drug–drug interactions within this able to the N-H stretching peak of FD at a concentration
particular sample, suggesting phase separation and uneven of 20% w/w. This again suggests the existence of strong

268 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

(a)
FD–PVPK15

80% FD SP
80% FD HME

50% FD SP

50% FD HME

20% FD SP
20% FD HME
3750 3550 3350 3150 2950 2750 2550
Wave numbers/cm

(b)
FD–Soluplus

80% FD SP
80% FD HME

50% FD SP

50% FD HME

20% FD SP
20% FD HME

3750 3550 3350 3150 2950 2750 2550


Wave numbers/cm

(c)
FD–HPMCAS

80% FD SP

80% FD HME

50% FD SP
50% FD HME

20% FD SP
20% FD HME

3800 3600 3400 3200 3000 2800 2600


Wave numbers/cm

Figure 10 Infrared spectra of solid dispersions showing the N-H stretching region: (a) felodipine–PVPK15 (b) felodipine–Soluplus and (c)
felodipine–HPMCAS; % represent the weight fraction of felodipine in the solid dispersion.

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 269
Spray-dried and HME solid dispersions Yiwei Tian et al.

Table 3 Infrared spectroscopy peak assignments for amorphous polymer solid dispersions, and are associated with ease of
felodipine and felodipine–polymer solid dispersion systems scale-up, the level of disorder of crystalline drug and homo-
Wave geneity of drug distribution within amorphous dispersions
Sample number/cm Assignment is not well understood.[30,31] Quantitative and qualitative
Amorphous 3339 N-H, H-bonded with other drug molecules analysis of the amorphous content, as well as the differen-
felodipine 1701 C=O, non-H-bonded tiation of amorphous and nanocrystalline solid dispersions,
1684 C=O, H-bonded is very important as it significantly influences the physical
HPMCAS–FD 3470 OH peak, H-bonded with drug molecules state and final performance of drug product. Furthermore,
3366 N-H, H-bonded with HPMCAS the effect of residual crystalline content within amorphous
1745 ester C=O from HPMCAS
solid dispersions is not easily resolved; extensive storage
1701 C=O of felodipine, non-H-bonded
1684 C=O, H-bonded with drug molecule
studies are normally required to understand stability.
Soluplus–FD 3455 free OH from Soluplus Herein, we propose a method that provides the thermody-
3339 N-H, H-bonded with other drug molecule namic drug–polymer temperature–composition phase
3294 N-H, strong H-bonding with Soluplus diagram with the aim of probing solubility and miscibility
1749 Unbonded C=O from Soluplus of a crystalline drug in a polymeric carrier. Development
1701 C=O of felodipine, non-H-bonded of such small-scale screening tool potentially allows us to
PVPK15–FD 3290 N-H, strong H-bonding with PVP
compare the level of drug solubility and miscibility in poly-
3339 N-H, H-bonded with other drug molecule
meric materials, and directly compare how different pro-
FD, felodipine; HPMCAS, hydroxypropyl methylcellulose acetate cessing techniques affect amorphous solid dispersions.
succinate; PVP, polyvinylpyrrolidone; PVPK15, PVP grade K15.

drug–polymer interactions formed between drug and The relevance of drug–polymer


polymer. When the drug concentration increased to 50%, temperature–composition phase diagrams
two separate peaks were recorded at 3294/cm and 3339/cm, to the production of amorphous
attributed to drug–polymer and drug–drug interactions, solid dispersion
respectively. The intensity of peak at 3339/cm was further Freshly manufactured drug–polymer solid dispersions at
increased as the drug concentration reached 80%, which 20%, 50% and 80% w/w drug loadings were characterised
confirms that high drug concentrations in the solid disper- using three different techniques: PXRD, MDSC and FTIR.
sion leads to drug–drug/polymer–drug phase separation. Obviously, there are limitations in regard to detection of
It is also noted that the N-H peak at 3339/cm was residual crystalline material within manufactured solid dis-
also observed in the 50% FD–Soluplus, irrespective of persions,[32] but qualitative comparison demonstrates that
the preparation technique unlike the combination using the F–H model to compare the drug–polymer inter-
FD–PVPK15. action is very useful for ranking suitable polymer candi-
In contrast to FD–PVPK15 and FD–Soluplus solid dis- dates. In this study, we constructed the maximum solubility
persions, there was only a subtle change noted in the N-H and miscibility curves for FD–PVPK15, FD–Soluplus and
stretching peak of FD–HPMCAS solid dispersion, indicat- FD–HPMCAS systems. This diagram reveals important
ing that the interaction formed between N-H of FD and the information to aid understanding of the relevance of
acceptor groups of HPMCAS was weaker than the interac- temperature and drug fraction in terms of designing ther-
tion formed in other two solid dispersions (Figure 10c). modynamically stable amorphous drug solid dispersion
However, it was also noted that for FD–HPMCAS, the systems. As shown in Table 4, using this tool, the physical
broad O–H stretching peak shifted downward from status of these three drug–polymer systems processed at dif-
3485/cm to 3470/cm. In this case, we suggest that the O–H ferent temperatures were predicted (80°C for spray drying
of HPMCAS may also act as an H-bond donor to interact and 140°C for HME). Comparing the predictions with
with the C=O of FD. This may replace the interaction experimental results, we may be able to better understand
between N-H of FD and C=O of HPMCAS. Therefore, a both processes and better define an optimum operating
joint interaction between the O–H, C=O of HPMCAS and space for manufacture of stable amorphous systems. F–H
C=O, N-H of FD may be formed in the solid dispersion.[29] theory was also used to reveal drug–polymer interactions
A summary of main peak shifts relating to FD within the wherein PVPK15 was identified as the polymeric material
three polymer solid dispersions is shown in Table 3. capable of providing the strongest interaction with FD, fol-
lowed by Soluplus and then HPMCAS. Following manufac-
General Discussion
ture of solid dispersions, the characterisation results from
Although it is well accepted that melt extrusion and spray PXRD provided the general classification on whether the
drying are widely used to prepare amorphous drug– system was X-ray amorphous or not. Bragg diffraction

270 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

Table 4 The predicted physical status of drug–polymer solid dispersion systems using constructed F–H temperature–composition phase diagrams

FD (w/w%) in PVP FD (w/w%) in HPMCAS FD (w/w%) in Soluplus

Spray dry at 85°C 20 50 80 20 50 80 20 50 80


Physical status √ √ Meta x x x Meta x x
HME at 140°C 20 50 80 20 50 80 20 50 80
Physical status √ √ √ √ √ √ √ √ √

FD, felodipine, HME, hot-melt extrusion; PVP, polyvinylpyrrolidone. ‘√’, ‘Meta’ and ‘x’ indicates the thermodynamic status of systems as stable, meta-
stable and unstable at conditions of processing based on the phase diagrams, respectively.

peaks were only observed in 80% FD–HPMCAS-melt been verified by several groups.[6,34] Additionally, it has also
extrudates, while the rest of the solid dispersions were clas- been demonstrated that HME provides sufficient energy
sified as X-ray amorphous, irrespective of the drug compo- for a high-level drug–polymer interaction to occur within
sition and preparation technique. However, it was evident the process, as the glass transition temperatures of melt
that crystalline content was present in samples containing extruded formulation were significantly higher than those
80% w/w FD solid dispersions, which was classified as X-ray calculated using Gordon–Taylor or CK model.[6] Relative to
amorphous (Figures 4 and 8). Melting events recorded by HME, spray drying is a well-established method for produc-
MDSC for all 80% FD-melt extrudates, as well as 80% ing amorphous solid dispersions. This process is inexpen-
FD–Soluplus and 80% FD–HPMCAS spray-dried samples, sive, fast and easy to scale up in the industry, which is why
confirm the presence of crystalline drug. Because of the it has been used in preparing numerous amorphous solid
complex nature of amorphous solid dispersion systems, dispersions.[35] Typically, the drug and polymer are both dis-
further characterisation using alternative X-ray equipment solved into a common solvent or co-solvent, and the
or spectral data treatment (pairwise distribution function) obtained solution is sprayed into tiny droplets, rapidly dried
may be required to identify information regarding molecu- through hot air and solidified into a solid form. The solvent
lar packing and level of disorder.[33] This suggests that F–H selection, the concentration of the drug and polymer in the
phase diagrams reveal important information relating to solvent solution, and the speed and temperature of the
the solubility and miscibility of drug in different polymeric drying process significantly affect the properties of formed
systems, and also provide a good approximation of the solid dispersions, such as the stability, dissolution rate and
ability of a polymeric carrier to form amorphous solid water uptake.[36]
dispersions. From this study, it was evident that the final physical
status of the binary mixture after manufacture was not only
affected by thermodynamic properties but also the nature
Production of amorphous solid dispersion
of preparation technique. For the systems under considera-
by melt extrusion or spray drying
tion, spray drying provided the greatest potential to gener-
HME is a well-known technique within the food and plastic ate drug–polymer dispersions across a wider concentration
industries, with increasing use over the last century. in which the sample should be thermodynamically unsta-
Recently, it has started to draw attention from the pharma- ble, and thus a high tendency to recrystallise (in the case of
ceutical industry as an emerging non-ambient drug delivery FD–HPMCAS and FD–Soluplus; Table 4). The two tech-
technology capable of providing a viable method to niques are very different in nature; spray drying is a solvent
produce a range of delivery platforms with sustained, modi- solution and drying process, whereas melt extrusion
fied or targeted drug release profiles. In particular, over the involves heating and shear mixing. It is clear that for melt
last decade, HME has been touted as an attractive method extrusion, an acceptable level of flowability is essential for
for producing amorphous solid drug dispersions. material to be conveyed forward within the extrusion
During pharmaceutical HME, the drug is typically pre- barrel, yet not completely liquid since the friction between
mixed with a polymeric carrier and introduced into a material and screw is the driving force for this conveyance.
heated barrel containing rotating twin screws that convey There is a limited window for extrusion in terms of
the mixture through a feed, mixing and melting zone. temperature and drug composition; however, within this
Often, during extrusion, the drug dissolves into the polymer window, the viscosity of the system may prevent intimate
melt and formed using a preselected die or other mixing, and hence the inability to generate fully amorphous
downstream-processing equipment. The details of extru- systems (50% and 80% high drug-loading systems; Table 4).
sion, while important, are beyond the remit of this article. For example, as shown in Figures 2 and 3, the dissolution
After cooling, the drug is dispersed at the micro-, nano- or temperature for FD into polymeric carrier increases as the
even molecular level within the formulation, which has drug concentration increases. However, at high drug con-

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 271
Spray-dried and HME solid dispersions Yiwei Tian et al.

centration, if the temperature exceeds the requirement for drying, while the amorphous drug within the solid disper-
drug dissolution, the viscosity of the mixture may be too low sion obtained from melt extrusion was more homogene-
to be conveyed forward. Therefore, the need for matrix flow ously distributed. As for quantitative analysis on the drug
limits the drug loading that may be manufactured via melt distribution within the solid dispersion, this may be
extrusion. Although the processability, in terms of drug achieved by further studies using higher resolution tech-
loading, is less than that achievable via spray drying, melt niques, such as localised thermal analysis or nanoscale IR
extrusion can optimise the drug–polymer interaction as well spectroscopy.
as amorphous drug distribution within the mixture. These
two factors may be attributed to the advanced design of twin- Importance of drug disorder and
screw extrusion systems. As shown in this study, a clear and distribution within solid dispersions
repeatable glass transition was recorded for most melt-
First, the principal aim of forming a drug into an amor-
extruded samples at drug loadings up to 50% (Figures 5 and
phous form is to achieve a solubility advantage, as proposed
6). Although multiple amorphous phases may be obtained
by the method of Parks and coworkers:[39]
for high drug-loaded extrudates, no recrystallisation was
observed in any of these samples during MDSC analysis. ⎛ σa ⎞
Moreover, no difference was observed in the N-H peak posi- ΔG ≅ − RTln ⎜ Tc ⎟ (3)
⎝ σT ⎠
tion in solid dispersions prepared via melt extrusion;
however, the ratio of drug–drug and drug–polymer interac- where ΔG is the free energy difference between the crystal-
tions appears to be different especially in certain spray-dried line and amorphous states, σTa σTc is the solubility ratio of
samples (Figure 10). For example, there was only one group amorphous to crystalline form, R is the gas constant, and T
of interactions (drug–polymer at 3290/cm) observed in melt- is the temperature of interest. It can be seen that the overall
extruded 50% FD–PVPK15 solid dispersions compared with free energy difference between amorphous and crystalline
two groups of interaction (3290 and 3339/cm) in the spray- form is the reason for higher solubility. On the other hand,
dried sample. this energy difference is also the major thermodynamic
On the other hand, for spray drying, the reason for better driving force for the amorphous to crystalline transforma-
efficacy in obtaining amorphous solid dispersion maybe tion. As a formulation strategy, solid dispersions employ a
due to the fact that it is a rapid drying process that is by long-chain viscous polymer alongside amorphous drugs to
nature a solution of disordered drug and polymer liquid. maintain the solubility advantage. The overall conforma-
The two excipients are already molecularly dissolved within tion of the drug molecules within the mixture will be
the solvent. Therefore, obtaining an amorphous solid dis- increased because of the presence of the polymer; thus, the
persion mainly involves rapidly drying the solvent without entropic barrier for amorphous drug recrystallisation will
inducing any nucleation or crystalline material. The selec- also be significantly altered. Second, when a binary system is
tion of polymeric carrier will affect the level of solvent formed, the drug distribution within the system becomes an
retained in the sample as well as the extent of drug–polymer added factor to consider. To compare to its own status, the
interaction, which would subsequently influence the physi- solid dispersion of amorphous drug evenly distributed
cal stability of the solid dispersion samples.[23] Essentially, if within the amorphous polymeric matrix would exhibit a
the interaction between the residual solvent and each of the large entropy that would be expected to show the best
two components is different, the solvent may induce immis- physical stability. The ranking of drug–polymer systems
cibility. It has also been reported that the residual moisture using F–H theory has shown good agreement to the physi-
will promote immiscibility and lead to extensive phase sepa- cal state of amorphous solid dispersion in this study, result-
ration.[37,38] In this study, the residual solvent in the spray- ing in different levels of drug disorder or drug distribution
dried samples immediately following manufacture (∼3–5%) within the polymeric matrices. This disorder and distribu-
may potentially reduce the level of drug–polymer interac- tion would be crucial to understanding the physical stability
tions, resulting in a less disordered system, for example, of amorphous solid dispersion systems.[40]
amorphous drug–drug domains distributed within the Clearly, more work is needed to understand the effects of
polymeric matrix, which could lead to recrystallisation process parameter in terms of individual preparation tech-
during temperature fluctuation (Figures 7 and 9). Thus, niques at the level of drug disorder and distribution within
two types of interaction were observed in FTIR and the polymeric matrices. In addition, given the complexity of
recrystallisation within MDSC for spray-dried samples con- the physical state and the interactions promoted between
taining 50% and 80% w/w FD systems. The variation in drug and polymer candidates during the processing, more
drug–polymer/drug–drug interaction probed by FTIR efforts are required in developing and validating the F–H
together with the evidence from MDSC may indicate that a theory-based phase diagrams for the purpose of forming a
less uniform drug distribution was obtained through spray stable amorphous drug–polymer solid dispersions.

272 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274
Yiwei Tian et al. Spray-dried and HME solid dispersions

Conclusions However, the residual solvent could potentially induce or


promote drug–polymer phase separation, resulting in local
In this article, F–H theory has been used to predict the drug drug domains within the system. Melt extrusion provides
solubility and miscibility in polymer candidates, and a sufficient energy for more intimate mixing to be achieved
drug–polymer temperature–composition phase diagrams between drug and polymer, which may improve physical
were constructed for individual drug–polymer combina- stability. It was also confirmed that stronger drug–polymer
tions. Drug solubility and miscibility at defined tempera- interactions might be generated through melt extrusion. In
tures have an impact upon the drug distribution within the conclusion, achieving molecular level miscibility and high
polymeric matrices. Given this information, we evaluated levels of drug distribution within the polymer is of critical
polymeric materials and made a direct comparison between importance in the optimisation of the physical stability.
two commonly used solid dispersions processing techniques Using temperature–composition phase diagrams is an
(spray drying and melt extrusion). Spray drying permitted important screening tool to probe the relevance of tempera-
generation of amorphous solid dispersions to be produced ture and drug composition in specific polymer candidates
across a wider drug concentration than melt extrusion. for the purpose of formulating solid dispersions.

References 8. Sun Y et al. Solubilities of crystalline of sulfathiazole/polyvinylpyrrolidone


drugs in polymers: an improved and sulfadimidine/polyvinylpyrrolid-
1. Andrews GP et al. Physicochemical analytical method and comparison of one. Mol Pharm 2011; 8: 532–542.
characterization and drug-release solubilities of indomethacin and 15. Marsac PJ et al. Effect of temperature
properties of celecoxib hot-melt nifedipine in PVP, PVP/VA, and and moisture on the miscibility of
extruded glass solutions. J Pharm PVAc. J Pharm Sci 2010; 99: 4023– amorphous dispersions of felodipine
Pharmacol 2010; 62: 1580–1590. 4031. and poly(vinyl pyrrolidone). J Pharm
2. Karavas E et al. Hydrophilic matrices 9. Koningsveld R, Solc K. Liquid-liquid Sci 2010; 99: 169–185.
as carriers in felodipine solid disper- phase-separation in multicomponent 16. Tian Y et al. Construction of
sion systems. Trends Colloid Interface polymer systems – influence of molar- drug–polymer thermodynamic phase
Sci Xv 2001; 118: 149–152. mass distribution on shadow curve diagrams using Flory–Huggins
3. Karavas E et al. Combining SEM, and phase-volume ratio. Collect Czech interaction theory: identifying the
TEM, and micro-Raman techniques to Chem Commun 1993; 58: 2305– relevance of temperature and drug
differentiate between the amorphous 2320. weight fraction to phase separation
molecular level dispersions and 10. Koningsveld R, MacKnight W. within solid dispersions. Mol Pharm
nanodispersions of a poorly water- Liquid–liquid phase separation in 2013; 10: 236–248.
soluble drug within a polymer matrix. multicomponent polymer systems. 27. 17. Rollinger J, Burger A. Polymorphism
Int J Pharm 2007; 340: 76–83. Determination of the pair-interaction of racemic felodipine and the unusual
4. Andrews GP et al. Physicochemical function for polymer blends. Polym series of solid solutions in the binary
characterization of hot melt extruded Int 1997; 44: 356–364. system of its enantiomers. J Pharm Sci
bicalutamide-polyvinylpyrrolidone 11. Crowley K, Zografi G. The effect of 2001; 90: 949–959.
solid dispersions. J Pharm Sci 2010; low concentrations of molecularly 18. Tajber L et al. Physicochemical evalu-
99: 1322–1335. dispersed poly(vinylpyrrolidone) on ation of PVP-thiazide diuretic interac-
5. Vanhee S et al. Thermodynamic stabil- indomethacin crystallization from the tions in co-spray-dried composites –
ity of immiscible polymer blends. amorphous state. Pharm Res 2003; 20: analysis of glass transition composi-
Macromolecules 2000; 33: 3924–3931. 1417–1422. tion relationships. Eur J Pharm Sci
6. Abu-Diak OA et al. Understanding 12. Marsac PJ et al. A comparison of the 2005; 24: 553–563.
the performance of melt-extruded physical stability of amorphous 19. Rumondor ACF et al. Effects of mois-
poly(ethylene oxide)-bicalutamide felodipine and nifedipine systems. ture on the growth rate of felodipine
solid dispersions: characterisation of Pharm Res 2006; 23: 2306–2316. crystals in the presence and absence of
microstructural properties using 13. Tao J et al. Solubility of small- polymers. Cryst Growth Des 2010; 10:
thermal, spectroscopic and drug molecule crystals in polymers: 747–753.
release methods. J Pharm Sci 2012; d-mannitol in PVP, indomethacin in 20. Marsac PJ et al. Estimation of drug-
101: 200–213. PVP/VA, and nifedipine in PVP/VA. polymer miscibility and solubility in
7. Graeser KA et al. Correlating thermo- Pharm Res 2009; 26: 855–864. amorphous solid dispersions using
dynamic and kinetic parameters with 14. Caron V et al. A comparison of experimentally determined interaction
amorphous stability. Eur J Pharm Sci spray drying and milling in the pro- parameters. Pharm Res 2009; 26: 139–
2009; 37: 492–498. duction of amorphous dispersions 151.

© 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274 273
Spray-dried and HME solid dispersions Yiwei Tian et al.

21. Capone C et al. Thermal and 31. Bikiaris DN. Solid dispersions, part Supporting Information
mechanical degradation during poly- II: new strategies in manufactur-
Additional Supporting Information may
mer extrusion processing. Polym Eng ing methods for dissolution rate
be found in the online version of this
Sci 2007; 47: 1813–1819. enhancement of poorly water-soluble
article at the publisher’s web site:
22. Bates S et al. Analysis of amorphous drugs. Expert Opin Drug Deliv 2011; 8:
and nanocrystalline solids from their 1663–1680. Table S1 The thermal analysis of amor-
X-ray diffraction patterns. Pharm Res 32. Shah B et al. Analytical techniques phous solid dispersion samples prepared
2006; 23: 2333–2349. for quantification of amorphous/ by HME and Spray drying methods (data
23. Patterson JE et al. Melt extrusion crystalline phases in pharmaceutical are shown as average of three samples ±
and spray drying of carbamazepine solids. J Pharm Sci 2006; 95: 1641– standard deviation).
and dipyridamole with polyvinyl- 1665. Figure S1 Modulated differential scan-
pyrrolidone/vinyl acetate copolymers. 33. Chen X et al. Quantifying amorphous ning calorimetry thermograms of unpro-
Drug Dev Ind Pharm 2008; 34: 95– content of lactose using parallel beam cessed polymers Soluplus, PVPK15 and
106. X-ray powder diffraction and whole HPMCAS-HF, and quench-cooled amor-
24. Rumondor ACF, Taylor LS. Effect of pattern fitting. J Pharm Biomed Anal phous felodipine.
polymer hygroscopicity on the phase 2001; 26: 63–72. Figure S2 The powder X-ray diffraction
behavior of amorphous solid disper- 34. Tho I et al. Formation of nano/ patterns of crystalline felodipine–Soluplus
sions in the presence of moisture. Mol micro-dispersions with improved dis- physical mixtures containing 20%, 50%
Pharm 2010; 7: 477–490. solution properties upon dispersion of and 80% wt% of drug. Crystalline and
25. Zhou D et al. Thermodynamics, ritonavir melt extrudate in aqueous amorphous felodipine were also plotted
molecular mobility and crystallization media. Eur J Pharm Sci 2010; 40: for comparison.
kinetics of amorphous griseofulvin. 25–32. Figure S3 The powder X-ray diffraction
Mol Pharm 2008; 5: 927–936. 35. Janssens S et al. Spray drying from patterns of crystalline felodipine–PVPK15
26. Zhou D et al. A calorimetric investiga- complex solvent systems broadens the physical mixtures containing 20%, 50%
tion of thermodynamic and molecular applicability of Kollicoat IR as a and 80% wt% of drug. Crystalline and
mobility contributions to the phy- carrier in the formulation of solid dis- amorphous felodipine were also plotted
sical stability of two pharmaceu- persions. Eur J Pharm Sci 2009; 37: for comparison.
tical glasses. J Pharm Sci 2007; 96: 241–248. Figure S4 The powder X-ray diffraction
71–83. 36. Al-Obaidi H et al. Anomalous proper- patterns of crystalline felodipine–
27. Tang X et al. The effect of temperature ties of spray dried solid dispersions. HPMCAS-HF physical mixtures contain-
on hydrogen bonding in crystalline J Pharm Sci 2009; 98: 4724–4737. ing 20%, 50% and 80% wt% of drug.
and amorphous phases in dihy- 37. Vasanthavada M et al. Phase behavior Crystalline and amorphous felodipine
dropyridine calcium channel blockers. of amorphous molecular dispersions were also plotted for comparison.
Pharm Res 2002; 19: 484–490. I: determination of the degree and Figure S5 The Fourier transform infrared
28. Taylor L, Zografi G. Spectroscopic mechanism of solid solubility. Pharm spectroscopy spectra of amorphous
characterization of interactions Res 2004; 21: 1598–1606. felodipine, crystalline felodipine, polymer
between PVP and indomethacin in 38. Vasanthavada M et al. Phase behavior Soluplus, HPMCAS-HF and PVPK15 at
amorphous molecular dispersions. of amorphous molecular dispersions – the region of 4000–2500/cm.
Pharm Res 1997; 14: 1691–1698. II: role of hydrogen bonding in solid Figure S6 The Fourier transform infrared
29. Jeffrey GA. An Introduction to Hydro- solubility and phase separation kinet- spectroscopy spectra of amorphous
gen Bonding. New York: Oxford Uni- ics. Pharm Res 2005; 22: 440–448. felodipine, crystalline felodipine, polymer
versity Press, 1997. 39. Hancock B, Parks M. What is the true Soluplus, HPMCAS-HF and PVPK15 at
30. Bikiaris D. Solid dispersions, part I: solubility advantage for amorphous the region of 2000–500/cm.
recent evolutions and future opportu- pharmaceuticals? Pharm Res 2000; 17: Figure S7 Infrared spectra of solid disper-
nities in manufacturing methods for 397–404. sion showing the C=O stretching region
dissolution rate enhancement of 40. Furuyama N et al. Evaluation of solid (a) felodipine–PVPK15 (b) felodipine–
poorly water-soluble drugs. Expert dispersions on a molecular level by Soluplus and (c) felodipine–HPMCAS;
Opin Drug Deliv 2011; 8: 1501– the Raman mapping technique. Int J percentages represent the weight fraction
1519. Pharm 2008; 361: 12–18. of felodipine in the solid dispersion.

274 © 2013 Royal Pharmaceutical Society, Journal of Pharmacy and Pharmacology, 66, pp. 256–274

Вам также может понравиться